MOLECULAR AND CELLULAR BIOLOGY, July 2010, p. 3620–3634
Copyright © 2010, American Society for Microbiology. All Rights Reserved.
Vol. 30, No. 14
miR-31 Functions as a Negative Regulator of Lymphatic Vascular
Lineage-Specific Differentiation In Vitro and Vascular
Development In Vivo?†
Deena M. Leslie Pedrioli,1‡ Terhi Karpanen,2Vasilios Dabouras,1Giorgia Jurisic,1
Glenn van de Hoek,2Jay W. Shin,1§ Daniela Marino,1Roland E. Ka ¨lin,1¶
Sebastian Leidel,3Paolo Cinelli,4Stefan Schulte-Merker,2
Andre ´ W. Bra ¨ndli,1? and Michael Detmar1*
Institute of Pharmaceutical Sciences, Swiss Federal Institute of Technology, ETH Zu ¨rich, Zu ¨rich, Switzerland1; Hubrecht Institute-KNAW and
University Medical Center, Utrecht, Netherlands2; Institute of Biochemistry, Swiss Federal Institute of Technology, ETH Zu ¨rich,
Zu ¨rich, Switzerland3; and Institute of Laboratory Animal Science, University of Zu ¨rich, Zu ¨rich, Switzerland4
Received 15 February 2010/Returned for modification 2 April 2010/Accepted 6 May 2010
The lymphatic vascular system maintains tissue fluid homeostasis, helps mediate afferent immune re-
sponses, and promotes cancer metastasis. To address the role microRNAs (miRNAs) play in the development
and function of the lymphatic vascular system, we defined the in vitro miRNA expression profiles of primary
human lymphatic endothelial cells (LECs) and blood vascular endothelial cells (BVECs) and identified four
BVEC signature and two LEC signature miRNAs. Their vascular lineage-specific expression patterns were
confirmed in vivo by quantitative real-time PCR and in situ hybridization. Functional characterization of the
BVEC signature miRNA miR-31 identified a novel BVEC-specific posttranscriptional regulatory mechanism
that inhibits the expression of lymphatic lineage-specific transcripts in vitro. We demonstrate that suppression
of lymphatic differentiation is partially mediated via direct repression of PROX1, a transcription factor that
functions as a master regulator of lymphatic lineage-specific differentiation. Finally, in vivo studies of Xenopus
and zebrafish demonstrated that gain of miR-31 function impaired venous sprouting and lymphatic vascular
development, thus highlighting the importance of miR-31 as a negative regulator of lymphatic development.
Collectively, our findings identify miR-31 is a potent regulator of vascular lineage-specific differentiation and
development in vertebrates.
Vertebrates have developed two parallel but structurally and
functionally distinct vascular systems: the blood and lymphatic
vascular systems (1, 7). The lymphatic vascular system controls
tissue fluid homeostasis, absorbs lipids and fat-soluble vitamins
from the intestine, and mediates afferent immune responses by
transporting lymphocytes and antigen-presenting cells to re-
gional lymph nodes (1, 7). In addition, malignant cancers can
induce lymphatic vessel activation and growth (lymphangio-
genesis) within primary tumors and draining lymph nodes,
which enhances cancer metastasis to draining lymph nodes and
beyond (1, 22). These findings have fueled a surge in studies
aimed at defining the molecular characteristics and functional
activities of lymphatic vessels and identifying molecules that
Genomic and proteomic studies have identified novel mo-
lecular markers and growth factors for lymphatic vessels (2, 23,
48, 52). Mouse genetic models have characterized the tran-
scription factors PROX1 and SOX18 as master regulators of
lymphatic vascular development and differentiation in vivo (12,
56, 65). These studies indicate that SOX18 expression in a
subset of cardinal vein endothelial cells initiates lymphatic
vascular development by inducing PROX1 expression (12).
The resulting lymphatic vascular progenitor cells bud off and
migrate away from the cardinal vein and form primitive lymph
sacs, which subsequently develop into functional lymphatics
(12, 56). PROX1 and SOX18 expression in cultured blood
vascular endothelial cells (BVECs) triggers these cells to adopt
lymphatic lineage-specific molecular and phenotypic charac-
teristics (12, 26, 48). Conversely, PROX1 knockdown in lym-
phatic endothelial cells (LECs) inhibits the expression of LEC
signature genes and triggers BVEC signature gene expression
(44; J. W. Shin et al., unpublished data). Despite these ad-
vances, a detailed understanding of the mechanisms control-
ling lymphatic vascular development and cell type-specific dif-
ferentiation remains elusive.
A potentially crucial aspect of lymphatic vascular biology has
remained unexplored to date, i.e., the role of microRNA
(miRNA)-guided posttranscriptional regulation. miRNAs are
genomically encoded 19- to 24-nucleotide (nt) noncoding
RNAs that regulate the flow of genetic information by limiting
* Corresponding author. Mailing address: Institute of Pharmaceuti-
cal Sciences, Swiss Federal Institute of Technology, ETH Zu ¨rich,
Wolfgang-Pauli-Str. 10, HCI H303, CH-8093 Zu ¨rich, Switzerland.
Phone: 41-44-633-7361. Fax: 41-44-633-1364. E-mail: michael.detmar
‡ Present address: Division of Molecular Medicine, College of Life
Sciences, University of Dundee, Dundee, Scotland.
§ Present address: Omics Science Center (OSC), RIKEN Yoko-
hama Institute, Yokohama, Japan.
¶ Present address: Department of Neuropathology, Charite ´-Univer-
sita ¨tsmedizin Berlin, Berlin, Germany.
? Present address: Walter Brendel Centre of Experimental Medi-
cine, Ludwig Maximilians University Munich, Munich, Germany.
† Supplemental material for this article may be found at http://mcb
?Published ahead of print on 17 May 2010.
protein synthesis (10). This regulation is brought about when
mature miRNAs, loaded in the RNA-induced silencing com-
plex, base pair with semicomplementary sites within the 3?
untranslated region (UTR) of target mRNAs. Once base
paired with its target, the miRNA represses translation and/or
induces mRNA degradation (10). Consequently, miRNAs act
as novel and potent regulators of the genome. This notion is
underscored by recent studies defining critical roles for
miRNAs in embryonic development, cell proliferation, cell
cycle progression, differentiation, and apoptosis, as well as
their contribution to the etiology of several diseases (10, 46,
Interestingly, functional roles for miRNAs in blood vascular
development have recently been defined. Downregulation of
the miRNA processing enzymes Dicer and Drosha has been
reported to impair angiogenesis (11, 59). Moreover, a few
miRNAs have been shown to affect human umbilical vein en-
dothelial cell (HUVEC) migration and proliferation in vitro,
regulate nitric oxide synthase expression, promote tumor an-
giogenesis, control vascular inflammation, and directly contrib-
ute to numerous vascular phenotypes (11, 59).
In the study presented here, we identified and addressed the
functional relevance of vascular lineage-specific miRNAs. We
first defined the miRNA expression profiles of primary human
LECs and BVECs and consequently identified four BVEC and
two LEC signature miRNAs. Their vascular lineage-specific
expression was confirmed in normal tissues by quantitative
real-time PCR (qRT-PCR) analysis of ex vivo-isolated murine
LECs and BVECs and by in situ hybridization (ISH). Interest-
ingly, our findings have further classified the widely expressed
(38) metastasis-associated (63, 64) miRNA miR-31 as a BVEC
signature miRNA. In vitro functional analysis of miR-31 dem-
onstrated that this miRNA inhibits lymphatic lineage-specific
differentiation in BVECs by repressing lymphatic lineage-spe-
cific transcript levels. These effects are, in part, due to direct
posttranscriptional repression of Prox1, a master regulator of
lymphatic development. Finally, in vivo gain-of-function stud-
ies with Xenopus and zebrafish embryos established that over-
expression of miR-31 impaired lymphatic development and
reduced venous sprouting. Taken together, these findings in-
dicate that miR-31 plays a pivotal role in regulating lineage-
specific differentiation within the developing vasculature of
MATERIALS AND METHODS
Cell culture. Primary human dermal microvascular LECs and BVECs were
isolated from neonatal human foreskins and cultured as previously described
(23). LECs and HUVECs were purchased from Cambrex (Verviers, Belgium).
IMR91 human dermal fibroblasts (hdFBs) were obtained from the National
Institute on Aging, Bethesda, MD. The immortalized human epidermal kerati-
nocyte line HaCaT was provided by Norbert Fusenig, German Cancer Research
Center, Heidelberg, Germany (4). Cells, except hdFBs, were propagated in
supplemented endothelial cell basal medium (EBM; Cambrex) as previously
described (23). hdFBs were propagated in Dulbecco’s modified Eagle medium
supplemented as described above and transferred 12 h prior to total RNA
isolation to EBM supplemented as described above. Primary cells were used at
In vitro miRNA expression profiling. The TaqMan microRNA Assays Human
Panel Early Access kit (Applied Biosystems, Foster City, CA), containing 157
individual human TaqMan microRNA assays, was used for qRT-PCR miRNA
expression profiling (5). Total RNA was isolated from biological replicates of 80
to 90% confluent 10-cm tissue culture dishes using the mirVana miRNA isola-
tion kit (Ambion, Austin, TX). Reverse transcription reactions were performed
using 2 ng of total RNA and the microRNA Reverse Transcription kit (Applied
Biosystems). miRNA expression levels of technical duplicates were determined
using a 7900HT Fast Real-Time PCR System (Applied Biosystems), and com-
parative threshold cycle (CT) values were acquired after 40 cycles using SDS 2.2
software (Applied Biosystems). TaqMan microRNA assays (Applied Biosys-
tems) for hsa-miR-31, hsa-miR-137, hsa-miR-99a, hsa-miR-125b, hsa-miR-95,
hsa-miR-326, and human RNU48 were used to confirm lineage-specific expres-
For analysis, detection thresholds were set to 0.04 U of fluorescence intensity,
and when a miRNA CTvalue was undetermined in both technical replicates, a CT
value of 41 was assigned. Data sets were normalized relative to let-7a and miR-16
using the formula Ave CTNORM? Ave CTmiRNA? (Ave CTlet-7a/miR-16? 24),
where Ave CTlet-7a/miR-16is the combined average CTvalue for let-7a and miR-16
from each 96-well plate. RNU48 or sno234 was used to normalize the individual
TaqMan microRNA assay data sets using the formula Ave CTNORM? Ave
CTmiRNA? (Ave CTRNU48 or sno234? 25), where Ave CTRNU48 or sno234is the
mean RNU48 or sno234 CTvalue (n ? 3). Relative abundances of LECs and
BVECs were calculated from log2ratios. P values were calculated using a two-
tailed Student t test.
Fluorescence-activated cell sorting (FACS) isolation of endothelial cells from
mouse colons. Experiments with mice were approved by the Kantonales Veteri-
na ¨ramt Zu ¨rich. Colons were excised from sacrificed female FVB mice (12 to 16
weeks old, n ? 8; Charles River, Sulzbach, Germany), opened longitudinally,
washed in cold phosphate-buffered saline (PBS), and placed in 1 mM dithiothre-
itol. Mucus was gently removed by scraping. Small tissue pieces were digested
with 8 mg/ml collagenase IV (Invitrogen, Carlsbad, CA)–0.5 mg/ml DNase I
(Roche, Rotkreuz, Switzerland)–5 mM CaCl2in PBS at 37°C for 15 min. After
passing through a 70-?m cell strainer (BD Biosciences, Franklin Lakes, NJ), the
resulting cell suspensions were centrifuged at 500 ? g for 10 min and resus-
pended in 2% fetal bovine serum-supplemented PBS containing 1 mM EDTA.
The antibodies used for FACS sorting were allophycocyanin-conjugated rat
anti-mouse CD31 (BD Biosciences Pharmingen, San Diego, CA), fluorescein
isothiocyanate-conjugated rat anti-mouse CD45.2 (BD Biosciences), hamster
anti-mouse podoplanin (clone 8.1.1; Developmental Studies Hybridoma Bank,
Iowa City, IA), and phycoerythrin-conjugated anti-hamster (CALTAG/Invitro-
gen) and isotype control antibodies. FACS sorting was performed using a
FACSAria and the FACSDiva software (BD Biosciences). Cells were lysed by
sorting directly into RLT Plus lysis buffer (Qiagen, Hilden, Germany) con-
taining ?-mercaptoethanol. Total RNA was extracted from BVECs (CD45?
CD31?podoplanin?) and LECs (CD45?CD31?podoplanin?) using the
RNeasy Plus Micro kit (Qiagen, Hilden, Germany). For miRNA expression analy-
ses, 6 ng of total RNA and TaqMan miRNA assays for mmu-miR-31, mmu-miR-
326, hsa-miR-137, hsa-miR-99a, hsa-miR-125b, and mouse sno234 were used.
ISH and immunofluorescence staining. miR-31 ISH and Lyve-1/CD31 immu-
nofluorescence staining were performed with 20-?m serial frozen colon sections
obtained from female FVB mice. ISH for mouse miR-31 was performed using
digoxigenin (DIG)-labeled locked nucleic acid (LNA)-modified detection probes
(mmu-miR-31 [catalog no. 39153-00], hsa/mmu/rno-U6 [positive control, catalog
no. 99002-00], sense miR-159 [negative control, catalog no. 99003-00]; Exiqon,
Vedbæk, Denmark) and the formaldehyde-EDC fixation miRNA ISH protocol
(47). Briefly, the LNA-modified detection probes were labeled with DIG using
the DIG Oligonucleotide Tailing kit (Roche, Basel, Switzerland) according to
the manufacturer’s instructions. Tissue sections were fixed in 4% formaldehyde–
Tris-buffered saline for 10 min and then in EDC solution (47) for 1.5 h. The
sections were acetylated in 1% triethanolamine–0.25% acetic anhydride, washed,
and prehybridized in hybridization buffer (47) for 1 h at 53°C. The colon tissue
sections were hybridized with 4 ?M DIG-labeled detection probes overnight at
56°C for miR-31 and 53°C for the controls. Following posthybridization washing
and blocking, the slides were probed with alkaline phosphatase-conjugated anti-
DIG Fab fragments (Roche). They were then washed in TNT buffer (100 mM
Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% Tween 20) and in AP buffer (100 mM
Tris-HCl [pH 9.5], 100 mM NaCl, 50 mM MgCl2). Color development was
performed with developer solution (AP buffer with 0.175 mg/ml 5-bromo-4-
chloro-3-indolylphosphate [BCIP], 0.45 mg/ml Nitro Blue Tetrazolium, and 2
mM levamisol). All incubations and washing steps were performed at room
temperature unless otherwise indicated.
Immunofluorescence staining was performed as previously described (25, 35),
using a rabbit polyclonal antibody against mouse Lyve-1 (AngioBio, Del Mar,
CA), a monoclonal rat antibody against mouse CD31 (BD Biosciences), and
corresponding secondary antibodies labeled with Alexa Fluor 488 or Alexa Fluor
594 (Molecular Probes). Sections were examined on an Axioskop2 microscope
(Carl Zeiss, Feldbach, Switzerland), and images were captured at magnifications
VOL. 30, 2010 miR-31 CONTROLS VASCULAR DEVELOPMENT3621
of ?2.5 (Plan-Neofluar 2.5x, 0.075 numerical aperture) and ?20 (Plan-Neofluar
20x, 0.50 Ph2) with an AxioCam MRm digital camera (Zeiss). Bright-field and
fluorescent channel image acquisition was accomplished using Axio Vision 4.4
software (Zeiss). Adobe Photoshop CS3 (Adobe Systems, San Jose, CA) was
used to adjust image brightness.
Microarray analyses. All transfections were carried out using the Basic
Nucleofector kit for primary mammalian endothelial cells (Amaxa AG, Cologne,
Germany). Five hundred thousand LECs were transfected with 2 ?M pre-
miR-31 or pre-miR-Neg molecules (30) in biological duplicate, and total RNA
was isolated using the mirVana isolation kit at 48 h posttransfection. The tran-
scriptome profiles of these cells were defined using the Applied Biosystems
Human Genome Survey Microarray v2.0 as previously described (54). Briefly,
DIG-UTP-labeled cRNA was generated from 1.5 ?g of total RNA using the
NanoAmp RT-IVT Labeling kit (Applied Biosystems). Twenty micrograms of
cRNA was fragmented and hybridized to the microarrays using the Applied
Biosystems Chemiluminescence Detection kit. Signal detection, image acquisi-
tion, and initial analyses were performed using the Applied Biosystems 1700
Chemiluminescent Microarray Analyzer.
Raw data were normalized using Quantile normalization available from
R/Bioconductor (14). Present calls were defined based on average signal-to-noise
ratios of ?3 and quality (error) values of ?5,000 (54). Feature signal intensities
were converted to log2values. miR-31-repressed genes were identified based on
present calls in both pre-miR-Neg arrays with log2(Pre31/PreNeg) values of
??0.59 and P values of ?0.05, while miR-31-induced genes were present in both
pre-miR-31 arrays and had log2(Pre31/PreNeg) values of ?0.59 and P values of
?0.05. P values were calculated using empirical Bayes statistics for differential
mRNA qRT-PCR analyses. To confirm the microarray data, the mRNA expres-
sion levels of selected candidate miR-31-regulated LEC and BVEC signature genes
were analyzed in triplicate by qRT-PCR using dually labeled TaqMan Gene Ex-
pressionAssaysfor TIMP3 (Assay
(Hs01044146_m1), HOXD10 (Hs00157974_m1), EDNRB (Hs00240752_m1),
PROX1 (Hs00160463_m1), NRCAM (Hs00170554_m1), SELE (Hs00950401_m1),
ICAM1 (Hs99999152_m1), MMP1 (Hs00899658_m1), RGS4 (Hs00194501_m1),
NRG1 (Hs00247620_m1), and LOC554202 (Hs01007340_m1) (all from Applied
Biosystems). The probe and primers for LYVE-1 were as previously described (23).
Twenty-five nanograms of cDNA generated using the High Capacity cDNA Archive
kit (Applied Biosystems) was used. Each reaction was normalized to ?-actin expres-
Detailed analysis of PROX1 mRNA and protein levels. To further character-
ized miR-31 regulation of PROX1, 500,000 LECs were transfected with 2 or 4
?M pre-miR-31 (n ? 4) or pre-miR-Neg (n ? 4) molecules or 4 ?M anti-miR-31
(n ? 2) or anti-miR-Neg (n ? 2) molecules (6). Total RNA and whole-cell
protein lysates were isolated using the mirVana PARIS kit at 48 h posttransfec-
tion. qRT-PCR analysis of PROX1 mRNA was performed as described in Re-
Northern blot analyses were performed with 1 ?g total RNA. PROX1 mRNA
was detected using purified PROX1 3? UTR [?-32P]ATP end-labeled probes
generated from a NotI-linearized human PROX1 3? UTR plasmid (YH1551;
provided by Young Kwon Hong, University of Southern California, Los Ange-
les). The membrane was then stripped, and ?-actin was detected using
[?-32P]ATP end-labeled human ?-actin oligonucleotides (5?-GTGAGGATCTT
For Western blotting, 25-?g protein lysate samples were resolved by sodium
dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellu-
lose membranes. Membranes were probed with rabbit polyclonal anti-human
PROX1 (RELIATech, Braunschweig, Germany) and mouse monoclonal anti-
human ?-actin antibodies (Sigma-Aldrich) detected using horseradish peroxi-
dase-conjugated secondary antibodies and standard chemiluminescence (54).
QuantityOne software (Bio-Rad) was used to semiquantitatively analyze the
amounts of PROX1 protein relative to ?-actin in each sample. Pre-miR-31 and
pre-miR-Neg PROX1 signal volume averages and standard deviations were
calculated from these normalized values (n ? 4). P values were calculated using
a two-tailed Student t test.
DNA constructs. pMIR-Luci/miR31BS and psiCHECK-2/miR31BS contain
perfect-match hsa-miR-31 binding sites (miR31BS). pMIR-Luci/miR31BS was
generated by cloning annealed miR31BS sense (5?-AGCTTGCTGAGCGGCA
AGATGCTGGCATAGCTGA-3?) and antisense (5?-CTAGTCAGCTATGCC
AGCATCTTGCCGCTCAGCA-3?) oligonucleotides into the HindIII and SpeI
restriction sites of the pMIR-REPORT Luciferase vector (6). psiCHECK-2/
miR31BS was constructed by ligating annealed miR31BS/F and miR31BS/R
oligonucleotides (see Table S1 in the supplemental material) into the XhoI and
NotI sites of psiCHECK-2 (Promega, Du ¨bendorf, Switzerland). PROX1 3? UTR
and coding sequence (CDS) luciferase reporter vectors (see Table S1 in the
supplemental material) were constructed by PCR amplification from YH1551
(for the oligonucleotides used, see Table S1 in the supplemental material).
PROX1 3? UTR amplicons were ligated into the XhoI and NotI sites of psi-
CHECK-2, and the PROX1-CDS amplicon was ligated into the PmeI and NotI
sites of psiCHECK-2.
Luciferase reporter assays. For miR-31 overexpression optimization, 500,000
LECs were cotransfected with 0.7 ?g of pMIR-Luci/miR31BS, 0.7 ?g of pMIR-
REPORT ?-galactosidase (?-Gal) control, and 0.02, 0.2, 1, 2, or 4 ?M human
miR-31 precursor (pre-miR-31) or pre-miR-Neg negative-control molecules
(Applied Biosystems). Luciferase and ?-Gal activities were monitored 48 h after
transfection using the Dual-Light Luciferase and ?-Gal Reporter Gene Assay
System (Applied Biosystems). ?-Gal RLUs (relative light units) were used to
normalize luciferase RLUs.
For the PROX1 3? UTR tethering and PROX1 3? UTR miR-31 binding site
mutagenesis assays, 500,000 LECs or HUVECs were cotransfected in triplicate
with 4 ?M miR-31 precursor or inhibitor (6) or the corresponding negative
controls and 0.7 ?g of the psiCHECK-2 constructs containing the miR-31 bind-
ing site (miR31BS), the PROX1 3? UTR fragments (PROX1 FL-F6), the PROX1
CDS, or the PROX1 3? UTR miR-31 binding site mutants (PROX1 FL-mut or
PROX1 F6-mut) (see Table S1 in the supplemental material). Firefly and Renilla
luciferase activities were monitored 48 h after transfection using the Dual-
Luciferase Reporter Assay System (Promega). PROX1 3? UTR/CDS-Renilla
luciferase RLUs were normalized to firefly luciferase RLUs. Both the dual-light
and dual-luciferase assays were performed in triplicate with 20 ?l of cell lysate.
Xenopus microinjection and whole-mount ISH. Xenopus studies were con-
ducted under protocols approved by the Veterinary Office of the Canton of
Zu ¨rich, Switzerland. Xenopus laevis eggs were obtained by hormone-induced
laying, fertilized in vitro, and prepared for microinjection as previously described
(21). Two-cell-stage embryos were unilaterally microinjected with pre-miR-31 or
pre-miR-Neg molecules (10 to 100 ng/blastomere) and 0.2 ng ?-Gal RNA (lin-
eage tracer). The antisense VEGFC morpholino oligomer (MO) (5?-GTAACG
CTCCCTCCAGCAAGTACAT-3?) was purchased from Gene Tools (Philo-
math, OR), and 5 to 10 ng was unilaterally injected into two-cell-stage embryos.
When uninjected embryos reached the embryonic stage indicated, the injected
embryos were fixed and processed for ISH. Whole-mount ISH, ?-Gal staining,
and bleaching of Xenopus embryos were carried out as previously described (29).
DIG-labeled probes were transcribed from linearized plasmids encoding Xeno-
pus pecam1 (29), prox1 (GenBank accession no. BU903551), and vegfr3 (28).
Images were acquired digitally using AxioVision 4.5 (Zeiss) software and an
AxioCam color camera (Zeiss) mounted on a Zeiss Stereo Lumar V12 stereo-
Zebrafish microinjection. Transgenic TG(fli1a:gfp)y1(37) and plcg1t26480ze-
brafish lines were maintained at the Hubrecht Institute. Zebrafish experiments
were approved by the Animal Experimentation Committee (DEC) of the Royal
Netherlands Academy of Arts and Sciences. The plcg1t26480allele is a W1024X
mutation in the plcg1 gene (GenBank accession number AY163168).
MO were ordered from Gene Tools (Philomath, OR). One-cell-stage
TG(fli1a:gfp)y1; plcg1t26480mutant embryos were injected at 40 ng/embryo with
an MO targeting dre-miR-31 (5?-TTAACAGCTATGCCAACATCTTGCC-3?)
or at 25 ng/embryo with an unrelated control MO (5?-GCATTGACTCTGTAA
AACAGACAAT-3?). For miR-31 overexpression, TG(fli1a:gfp)y1; plcg1t26480
mutant embryos were injected with 170 or 340 pg of human pre-miR-31 precur-
sor or pre-miR-Neg control molecules (Applied Biosystems). Venous sprouts
were quantified at 48 h postfertilization (hpf), and statistical significance was
analyzed using the Student t test. For imaging, embryos were mounted in 0.8%
low-melting-point agarose in a dish with a coverslip replacing the bottom. Im-
aging was performed with a Leica SP2 confocal microscope (Leica Microsystems)
using a 20? objective.
Microarray data accession number. The microarray data obtained in this
study are accessible at http://www.ncbi.nlm.nih.gov/geo/ under accession no.
BVEC versus LEC lineage-specific miRNA expression. Us-
ing a TaqMan-based qRT-PCR profiling platform (5), we de-
fined the in vitro expression profiles of 157 human miRNAs in
primary LECs and BVECs, as well as two nonendothelial cell
types (HaCaT keratinocytes and hdFBs) (see Table S2 in the
supplemental material). Following data normalization, expres-
3622LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.
sion profiles for each cell type were defined by setting the CT
value present call cutoff at 34 (see Table S2 in the supplemen-
tal material). Most of the miRNAs analyzed were expressed at
comparable levels in both LECs and BVECs (see Fig. S1 in the
supplemental material). Nevertheless, based on 2-fold or
greater differential expression, 16 candidate LEC and 30 can-
didate BVEC signature miRNAs were identified (Table 1).
Using a P value cutoff of ?0.05, two LEC signature miRNAs
(miR-95 and miR-326) and four BVEC signature miRNAs
(miR-137, miR-31, miR-125b, and miR-99a) were identified
(Table 1). Individual TaqMan miRNA assays confirmed the
vascular lineage specificity of these miRNAs. miR-95 and
miR-326 expression levels were, on average, 46-fold and 7-fold
higher in LECs than in BVECs, respectively (Fig. 1A). Con-
versely, miR-137 expression was 124-fold higher in BVECs
than in LECs, miR-31 expression was 48-fold higher in
BVECs, and miR-125b and miR99a were 3-fold more abun-
dant in BVECs (Fig. 1B).
LEC and BVEC signature miRNA expression patterns are
maintained in vivo. We next isolated BVECs and LECs from
the colons of eight adult mice by FACS sorting using the
leukocyte marker CD45, the panendothelial marker CD31,
TABLE 1. Candidate endothelial cell signature miRNAsa
Fold changeP value
LEC signature miRNAs
BVEC signature miRNAs
aSixteen candidate LEC signature miRNAs and 30 BVEC signature miRNAs were identified among the 157 human miRNAs profile in primary human LECs and
BVECs based on a ?2-fold difference in relative expression between endothelial cell types. let-7a/miR-16 normalized average LEC (LEC CTAve) and BVEC (BVEC
CTAve) CTvalues, the corresponding relative differences (?CTLEC/BVEC), and absolute fold changes are shown. P values were calculated using a two-tailed Student t
VOL. 30, 2010miR-31 CONTROLS VASCULAR DEVELOPMENT 3623
and the LEC marker podoplanin to differentiate among leu-
kocytes (CD45?CD31?podoplanin?and CD45?CD31?po-
doplanin?), BVECs (CD45?CD31?podoplanin?), and LECs
(CD45?CD31?podoplanin?) (18). We obtained 1,500 to
25,000 LECs and 2,500 to 55,000 BVECs, from which total
RNA was extracted and used for ex vivo qRT-PCR miRNA
profiling. Based on 1.5-fold or greater differential expression,
miR-31 was indeed more strongly expressed by BVECs than by
LECs in six of the eight mice analyzed (Fig. 2A). The degrees
of miR-31 differential expression between mouse BVECs and
LECs in vivo were less pronounced than those observed be-
tween in vitro-cultured human endothelial cells. Nevertheless,
statistical analysis of these in vivo data confirmed that the
differences in miR-31 expression between mouse BVECs and
LECs were statistically significant in six out of the eight mice
studied (Fig. 2A). The LEC signature expression pattern of
miR-326 and the BVEC signature classifications of miR-125b
and miR-99a were also confirmed, while no major changes
were found for miR-137 (see Fig. S2 in the supplemental ma-
terial). miR-95 is not present in mice.
Low-magnification (?2.5) microscopic analysis of adult
mouse colon tissue sections probed for miR-31 expression by
ISH revealed strong miR-31 staining throughout the adult
mouse colon (Fig. 2B and D). Importantly, immunofluores-
cence staining of serial sections for the panvascular marker
CD31 and the lymphatic marker LYVE-1 revealed that
miR-31 preferentially colocalized with blood vessels (CD31?
LYVE1?) present in the submucosa and mesenteric attach-
ments of the colon, as well as the lamina propria (Fig. 2B and
FIG. 2. miR-31 is preferentially expressed by BVECs in vivo.
(A) Mouse BVECs (CD45?CD31?podoplanin?) and LECs (CD45?
CD31?podoplanin?) were isolated from adult mouse colons (n ? 8)
by FACS sorting. TaqMan qRT-PCR analysis of mmu-miR-31 dem-
onstrated that miR-31 was at least 1.5-fold more abundant in BVECs
from six of the eight mice. Data were normalized using sno234 and are
shown as relative abundances ? the standard deviations.*, P ? 0.05;
**, P ? 0.01;***, P ? 0.001; ns, not statistically significant. (B to I)
miR-31 (B, D, and F) and negative-control (G; sense miR-159) ISH
and double-immunofluorescence analysis (C, E, H, and I) of serial
sections of adult mouse colon tissues imaged at low magnification
(?2.5; B to E) and high magnification (?20; F to I). Low magnification
of ISH of adult mouse colon tissue revealed strong miR-31 staining in
the majority of the cells (B and D), while double-immunofluorescence
analysis of serial sections (C and E) and comparison of miR-31- and
sense miR-159-probed sections at high magnification (F to I) demon-
strated that miR-31 preferentially colocalized with blood vessels
(CD31?LYVE1?) present in the submucosa and mesenteric attach-
ments of the colon (arrows), as well as the lamina propria (arrow-
heads). In contrast, lymphatic vessels (CD31?LYVE1?) displayed
weak or no miR-31 signals (asterisks). Panels F and G are high-
magnification images of the boxed regions in panels D and E, respec-
tively. Corresponding blood and lymphatic vessels in panels F to I are
numbered. Scale bars are 200 ?m (B to E) and 50 ?m (F to I).
FIG. 1. Identification of vascular lineage-specific miRNAs. qRT-
PCR using individual TaqMan miRNA assays confirmed the lymphatic
lineage-specific expression of miR-95 and miR-326 (A) and the
BVEC-specific expression of miR-137, miR-31, miR-125b, and
miR-99a (B). Data were normalized using RNU48 and are shown as
mean relative abundances ? the standard deviations (n ? 4/group).*,
P ? 0.05;***, P ? 0.01.
3624 LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.
D, arrows and arrowheads). In contrast, miR-31 expression
was weak or absent in lymphatic vessels (Fig. 2B and D, aster-
isks). Independent CD31 and LYVE1 immunofluorescence
staining and negative-control (sense miR-159) ISH of serial
sections of adult colon tissues (Fig. 2H and I), coupled with
high-magnification (?20) image analysis of the miR-31 ISH
and the CD31/LYVE1 immunofluorescently labeled serial sec-
tions demonstrated no preferential staining of either the blood
and lymphatic vessels in the sense miR-159-probed sections
(Fig. 2H and I). Both vessel types developed equivalent back-
ground level signals following ISH processing. In contrast,
miR-31 molecules were preferentially associated with colonic
blood vessels (CD31?LYVE1?; arrows), while the lymphatic
vessels (CD31?LYVE1?) developed no miR-31 signals (as-
terisks) (Fig. 2F and G). Taken together, these in vitro and in
vivo data demonstrate that miR-31 transcripts are rare or ab-
sent in LECs but enriched in BVECs.
miR-31 gene synteny is evolutionarily conserved. miR-31 is
encoded within intron 1 of an uncharacterized gene,
LOC554202 (gene ID 554202). This led us to question if
miR-31 lineage specificity results from the preferential expres-
sion of LOC554202 in BVECs. Indeed, TaqMan-based qRT-
PCR analysis demonstrated that LOC554202 transcripts were
54-fold more abundant in BVECs than in LECs (??CT? 5.78;
see Fig. S3A in the supplemental material), which is compa-
rable to the degree of differential expression defined for
miR-31 in BVECs versus LECs (??CT? 5.58). Interestingly,
further genome and proteome bioinformatic analyses revealed
that the gene is not conserved in vertebrates and the putative
gene product has no homology to proteins of known function.
Thus, LOC554202 might primarily function as a conduit for
miR-31 posttranscriptional regulatory activities in BVECs. Se-
quence alignments of human LOC554202 with 23 eutherian
mammals (Enredo, Pecan, Ortheus) using Ensembl alignment
tools (http://www.ensembl.org/index.html) revealed conserved
synteny for miR-31 with six of the species queried. Moreover,
13 of the eutherian mammals queried are predicted to contain
novel miRNA genes at positions aligned with LOC554202
(data not shown). Further alignment analysis of these 20
miRNA genes demonstrated that these 13 novel miRNA genes
are, in fact, miR-31 orthologs (see Fig. S3B in the supplemen-
tal material). This highly conserved synteny suggests that the
genomic, transcriptional, and epigenetic factors regulating
miR-31 expression have remained conserved during mamma-
Identification of LEC and BVEC signature genes regulated
by miR-31. The strong differential expression of miR-31 in vitro
and in vivo prompted us to further characterize this BVEC
signature miRNA by defining the transcriptome profile
changes in LECs after miR-31 overexpression. Using a lucif-
erase reporter construct containing a miR-31 binding site
(pMIR-Luci/miR31BS) and qRT-PCR, we found that high lev-
els of miR-31 gain of function could be achieved when LECs
were transfected with 2 ?M pre-miR-31 precursor (30) (see
Fig. S4 in the supplemental material). Using ?1.5-fold differ-
ential expression and ?0.05 P value thresholds, gene microar-
ray analyses of 2 ?M pre-miR-31- or pre-miR-Neg-transfected
LECs identified 548 miR-31-repressed and 335 miR-31-in-
duced genes (see Tables S3 and S4 in the supplemental mate-
rial). Comparing the miR-31 targets predicted by TargetScan (17,
39), miRanda (27), miRBase (16), and PicTar (31), 7.2% of the
miR-31-repressed genes were predicted targets of miR-31.
In silico biological process analysis by the PANTHER clas-
sification system (61, 62) was used to assess whether the ob-
served miR-31-mediated reprogramming of LECs specifically
affected lymphatic, blood vascular, and/or endothelial biologi-
cal functions. Intriguingly, genes involved in cell communica-
tion, signal transduction, cell adhesion, apoptosis, and numer-
ous signaling pathways were overrepresented among the
miR-31-repressed genes compared to the expected number of
genes (see Table S5 in the supplemental material). While these
biological processes are common to both endothelial cell types,
this enrichment suggests a dramatic reorganization of the LEC
surface characteristics, as well as cell signaling network activity,
following miR-31 overexpression. The specific effects miR-31
overexpression had on genes involved in vascular lineage-spe-
cific differentiation were then identified by comparing the miR-
31-regulated genes to the LEC (344 genes) and BVEC (479
genes) signature genes previously identified in vitro (23, 32, 48;
Shin et al., unpublished). Interestingly, twice as many LEC
signature molecule-encoding transcripts (9.6%) as BVEC sig-
nature molecule-encoding transcripts (4.8%) were reduced fol-
lowing miR-31 overexpression (Fig. 3A and B; see Table S3 in
the supplemental material). Also, approximately four times as
many BVEC signature genes (4.6%) as LEC signature genes
(1.1%) were induced/stabilized following miR-31 overexpres-
sion (Fig. 3A and B; see Table S4 in the supplemental mate-
rial). Corroborating the microarray data, qRT-PCR experi-
ments confirmed that four of the LEC signature genes
(EDNRB, PROX1, PPP1R9A, and HOXD10) and two of the
BVEC signature genes (ICAM1 and SELE) tested were sig-
nificantly less abundant in the pre-miR-31 samples than in the
pre-miR-Neg control (Fig. 3C and D). Furthermore, we also
validated the upregulation of the miR-31-induced BVEC sig-
nature genes MMP1 and RGS4 (Fig. 3E).
miR-31 inhibits PROX1 protein translation. Among the
miR-31-repressed target genes was PROX1, an essential lym-
phatic lineage-specific transcription factor (1, 7). qRT-PCR
analysis confirmed that transfection of LECs with 4 ?M pre-
miR-31 resulted in a ?60% reduction in PROX1 transcripts
(Fig. 4A), which was further verified by Northern blotting (see
Fig. S5 in the supplemental material). Importantly, immuno-
blotting revealed a consistent decrease in PROX1 protein lev-
els of ?40% following miR-31 overexpression (Fig. 4B and C).
Similar but less consistent results were observed after trans-
fection with 2 ?M pre-miR-31 (see Fig. S6 in the supplemental
material). Conversely, loss of miR-31 function in BVECs via
transfection of HUVECs with miR-31 inhibitor molecules
(anti-miR-31) resulted in a 1.7- to 3.1-fold increase in PROX1
mRNA levels (Fig. 4D). PROX1 protein remained undetect-
able in these samples (data not shown).
PROX1 is a direct target of miR-31 posttranscriptional reg-
ulation. PROX1 was not predicted to be a target of miR-31 by
TargetScan (17, 39), miRanda (27), miRBase (16), or PicTar
(31), but human cells express two isoforms of PROX1 mRNA,
a 7.9-kb isoform that contains a 5.4-kb 3? UTR and a 3.1-kb
isoform that has a much shorter 602-bp 3? UTR (57). Only the
short 3? UTR has thus far been used for miRNA binding site
predictions. In agreement with previous studies (25, 26, 49),
our LECs expressed the longer 7.9-kb isoform of PROX1 (see
VOL. 30, 2010miR-31 CONTROLS VASCULAR DEVELOPMENT 3625
FIG. 3. miR-31 overexpression in LECs modulates the expression of BVEC and LEC signature genes. (A and B) miR-31 was overexpressed in LECs via
transfection with 2 ?M pre-miR-31 precursor (n ? 2) or pre-miR-Neg (n ? 2). Microarray analysis after 48 h demonstrated that miR-31 overexpression
repressed the expression of 33 LEC signature genes (A) and 23 BVEC signature genes (B). In addition, 4 LEC signature genes and 22 BVEC signature genes
were induced. (C to E) TaqMan-based qRT-PCR analyses confirmed statistically significant miR-31-mediated repression of four of the LEC signature genes
studied (C) and of two of the BVEC signature genes studied (D), as well as statistically significant induction of two of the BVEC signature genes studied (E).
Data were normalized using ?-actin and are shown as relative abundances ? the standard deviations.*, P ? 0.05;**, P ? 0.01;***, P ? 0.001.
Fig. S5 in the supplemental material). Therefore, standard
nucleic acid sequence alignment techniques (SIM alignment
tools) and independent Targetscan 5.0 analyses were used to
identify potential miR-31 binding sites in the 5.4-kb 3? UTR.
SIM alignment identified five potential miR-31 recognition
sites, and the Targetscan search identified one of these sites, nt
949 to 971, as a 7mer-m8 binding site (Fig. 5A; see Table S6 in
the supplemental material).
To test the functional relevance of these candidate miR-31
binding sites, luciferase reporter genes containing a full-length
PROX1 3? UTR (PROX1 FL), six PROX1 3? UTR fragments
(PROX1 F1 to F6) (Fig. 5A), or a PROX1 CDS were con-
structed. The activities of these chimeras were monitored fol-
lowing miR-31 gain of function in LECs (Fig. 5B) and loss of
function in HUVECs (Fig. 5C). Confirming pre-miR-31 over-
expression and anti-miR-31 knockdown activity, miR-31 bind-
ing site (miR31BS) luciferase reporter gene activity decreased
or increased by ?40% after cotransfection with pre-miR-31 or
anti-miR-31 molecules, respectively (Fig. 5B and C). The lu-
ciferase activities of the PROX1 FL and PROX1 F2 reporter
genes, which contain the 7mer-m8 site, decreased significantly
(?35% and ?45%, respectively) following miR-31 overexpres-
sion (Fig. 5B). Conversely, their activities increased ?1.4-fold
and 3-fold, respectively, after miR-31 inhibition (Fig. 5C).
While PROX1 3? UTR F1 and F4 reporter gene activities
increased after miR-31 knockdown, reciprocal responses fol-
lowing overexpression were not observed. Together, these
findings confirm the direct posttranscriptional regulation of
PROX1 by miR-31 and suggested that this regulation is medi-
ated via nt 949 to 971 of the 5.4-kb 3? UTR.
To validate this, the seed sequence (nt 964 to 971) and 3?
compensatory site interacting nucleotides (nt 954 to 960) of the
PROX1 3? UTR were mutated to match the miR-31 sequence
in both the PROX1 FL and PROX1 F2 luciferase reporter
plasmids, thus eliminating the predicted PROX1–miR-31 in-
teraction. While the luciferase activities of the wild-type
PROX1 FL and PROX1 F2 constructs decreased or increased
after miR-31 overexpression or knockdown, respectively (Fig.
5D), the activities of the mutant full-length and F2 fragment
luciferase reporter genes did not change significantly under
either condition, thus mapping a bona fide, biologically active
miR-31 binding site to nt 949 to 971 of the PROX1 3? UTR.
miR-31 overexpression inhibits PROX1 target genes. Nu-
merous studies have demonstrated that PROX1 transcriptional
activities help dictate the molecular characteristics and func-
tional activities of vascular endothelial cells (12, 26, 44, 48;
Shin et al., unpublished). Therefore, our characterization of
PROX1 as a target of miR-31 posttranscriptional regulation
suggested that miR-31 repression of PROX1 should specifi-
cally, albeit indirectly, alter the expression PROX1 target
genes. Comparing the miR-31-regulated genes identified
above with a PROX1 loss-of-function data set generated fol-
lowing lentiviral inhibition of PROX1 in LECs (Shin et al.,
unpublished) revealed that approximately 20% of the miR-31-
repressed and -induced genes were similarly repressed or in-
duced, respectively, following Prox1 knockdown (see Table S3
and S4 in the supplemental material). Intriguingly, ?50% of
the miR-31-repressed LEC signature genes (see Table S3 in
the supplemental material) and 36% of the miR-31-induced
BVEC signature genes (see Table S4 in the supplemental ma-
terial) were similarly differentially expressed following PROX1
depletion from LECs.
miR-31 regulates vascular development in vivo. During em-
bryogenesis, PROX1 is expressed in a subpopulation of cardi-
nal vein endothelial cells that give rise to the mammalian
lymphatic vascular system (1, 7). The BVEC-specific expres-
sion of miR-31, together with its ability to posttranscriptionally
repress numerous BVEC and LEC signature genes (see Tables
S3 and S4 in the supplemental material), including Prox1,
FIG. 4. miR-31 gain and loss of function modulate PROX1 mRNA
and protein levels. (A to C) Transfection of LECs with 4 ?M pre-
miR-31 precursor (n ? 4) or pre-miR-Neg (n ? 4). (A) qRT-PCR
analysis revealed a ?60% decrease in PROX1 transcripts after 48 h.
(B) Immunoblotting of cell lysates with anti-PROX1 and anti-?-actin
antibodies (loading control) demonstrated decreased PROX1 protein
levels following miR-31 gain of function. The values to the left are
molecular sizes in kilodaltons. (C) Semiquantitative analysis of
PROX1 immunoblot signal intensities, relative to ?-actin, defined a
?40% reduction of PROX1 protein following miR-31 overexpression.
Data are shown as relative abundances ? the standard deviations (n ?
4/group). (D) qRT-PCR analysis showed that transfection of HUVECs
with 4 ?M anti-miR-31 inhibitor (n ? 2) or anti-miR-Neg (n ? 2)
induced a ?1.69-fold increase in PROX1 mRNA levels after 48 h. Data
were normalized using ?-actin and are shown as mean relative abun-
dances ? the standard deviations.***, P ? 0.001.
VOL. 30, 2010 miR-31 CONTROLS VASCULAR DEVELOPMENT3627
3628 LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.
suggested that this miRNA might play a role in vascular de-
velopment. As many of the BVEC and LEC signature genes
targeted by miR-31 also play major roles in Xenopus vascular
development (9, 20, 45), we reasoned that ectopic expression
of miR-31 in early Xenopus embryos might interfere with lym-
phatic vascular development. To investigate this, two-cell-stage
Xenopus embryos were unilaterally microinjected with human
pre-miR-31 or pre-miR-Neg molecules. Lymphatic and blood
vascular system development was then monitored in stage 39
embryos using whole-mount ISH for specific lymphatic and
blood vascular marker genes (28, 29, 45).
No gross developmental defects or externally visible pheno-
types were observed following pre-miR-Neg or pre-miR-31
microinjection into Xenopus embryos (Fig. 6). Moreover, prox1
and vegfr3 marker gene analysis demonstrated that lymphatic
vascular development progressed normally in 95% and 79% of
the pre-miR-Neg control embryos, respectively (Table 2). The
embryos had well-defined and clearly visible lymph hearts,
lymph vessels, and punctate patches of LECs in their tails (Fig.
6A and B). In contrast, a dose-dependent increase in lymphatic
vascular defects was observed in pre-miR-31-injected embryos.
Specifically, vefgr3 ISH demonstrated that the percentage of
embryos with lymphatic vascular defects, as monitored by the
loss of vegfr3-positive lymphatics sprouting from the lymph
hearts, progressively increased from 6.1 to 76.2% as the
amount of pre-miR-31 molecules increased from 1 to 50 ng
(Table 2). Generally, lymph hearts were present in these em-
bryos but appeared smaller and less well defined than in con-
trol embryos (Fig. 6A and B) or the uninjected side of the
pre-miR-31 embryos (data not shown). Moreover, lymphangio-
genesis, scored by the presence of vegfr3-expressing lymphatic
vessels sprouting from the lymph heart, was either strongly
reduced or absent in the presence of excess miR-31 (Fig. 6A
and B). These phenotypes were similar to those observed fol-
lowing morpholino inhibition of vegfc, where lymphangiogen-
esis was disrupted in the lymph heart region of 67 to 100%
(n ? 3; total number of embryos analyzed ? 71) of the injected
Xenopus embryos (see Fig. S7 in the supplemental material).
pecam1 expression was used to monitor blood vascular sys-
tem development in pre-miR-Neg- and pre-miR-31-microin-
jected Xenopus embryos. In control embryos, all of the major
blood vascular structures, such as the posterior cardinal veins
and the dorsal aorta, were clearly visible, and angiogenic
sprouting of intersomitic veins occurred normally in 73% of
the pre-miR-Neg-injected embryos (Fig. 6C and Table 2). By
comparison, the percentage of embryos displaying unilateral
intersomitic vein growth and/or guidance defects progressively
increased from 0% to 76% with increasing amounts of pre-
miR-31 injected (Fig. 6C and Table 2).
Gain-of-function phenotypes can occasionally be attributed
to the off-target effects associated with nonphysiological ex-
pression levels of a small interfering RNA or miRNA (41, 51).
We therefore sought to confirm the Xenopus miR-31 overex-
pression phenotypes in zebrafish, another highly relevant ver-
tebrate model organism for studying blood and lymphatic vas-
cular development (40). To facilitate the quantification of
venous sprouting from the posterior cardinal vein (24, 33, 67), we
used phospholipase C gamma 1 (plcg1) mutant embryos in the
TG(fli1a:gfp)y1background (36) [TG(fli1a:gfp)y1; plcg1t26480],
in which the venous sprouts contributing to both the blood and
lymphatic vasculature are clearly visible. Venous sprouting was
quantified at 48 hpf following the injection of 170 or 340 pg of
either human pre-miR-31 precursor or pre-miR-Neg control
molecules (Table 3 and Fig. 7A). In agreement with our
Xenopus studies, a significant dose-dependent reduction in
venous sprouting was observed in pre-miR-31-injected em-
bryos compared to embryos injected with 340 pg of pre-miR-
Neg molecules or to uninjected controls (Table 3 and Fig. 7).
The increase in venous sprouting observed in pre-miR-Neg-
injected embryos (Fig. 7A and C) appears to be a stress
response, which we have also observed in a number of un-
related control injections (data not shown). Conversely and
importantly, miR-31 overexpression led to a highly signifi-
cant reduction in venous sprouting and no other develop-
mental defects were observed in these embryos (Table 3 and
Fig. 7A and D).
Venous sprouting and lymphangiogenesis were also moni-
tored in zebrafish embryos following the injection of increasing
concentrations of morpholino oligonucleotides targeting both
mature and precursor dre-miR-31. Significant vascular pheno-
types could not be specifically attributed to loss-of miR-31
activity in these embryos (data not shown). Taken together,
our miR-31 gain-of-function studies with Xenopus and ze-
brafish embryos indicate that appropriate expression levels of
miR-31 during vertebrate embryogenesis are required for nor-
mal lymphatic and blood vascular development.
In the study presented here, we first defined the in vitro
expression profiles of 157 human miRNAs in primary human
LECs and BVECs using a TaqMan-based qRT-PCR profiling
FIG. 5. Direct posttranscriptional regulation of PROX1 by miR-31. (A) Schematic representation of the full-length PROX1 3? UTR and the
consecutive fragments present in the PROX1 3? UTR-luciferase reporter constructs (PROX1 FL to F6). Candidate miR-31 binding sites identified
using standard nucleic acid alignment techniques (gray) and TargetScan 5.0 (black) are indicated. The PROX1–miR-31 base pairing is shown for
the 7mer-m8 TargetScan-predicted binding site. (B to D) LECs or HUVECs were cotransfected with the plasmids containing a miR-31 binding
site (miR31BS), the PROX1 3? UTR (FL to F6), the PROX1 CDS, or PROX1 3? UTR miR-31 binding site mutant forms (FL-mut or F2-mut) and
the pre-miR-31 precursor, anti-miR-31 inhibitor, or negative-control molecules. (B) The activities of luciferase constructs containing miR31BS,
PROX1 FL, and PROX1 F2 decreased significantly following miR-31 overexpression (pre-miR-31; n ? 3) in LECs compared to those of negative
controls (pre-miR-Neg; n ? 3). (C) miR31BS and PROX1 FL reporter gene activities increased significantly following miR-31 inhibition
(anti-miR-31; n ? 3) in HUVECs compared to those of negative controls (anti-miR-Neg; n ? 3). miR-31 loss of function also enhanced the
activities of PROX1 F1 and PROX1 F2 constructs, but these differences were not statistically significant. (D) miR-31 mutant binding site full-length
and F2 PROX1 3? UTR (PROX1 FL-mut and PROX1 F2-mut) luciferase activities showed no major change following miR-31 overexpression
(pre-miR-31, n ? 3) or knockdown (anti-miR-31, n ? 3). Data were normalized to firefly luciferase activities and are shown as mean relative
abundances ? the standard errors. ns, not significant;*, P ? 0.05;**, P ? 0.01;***, P ? 0.001.
VOL. 30, 2010miR-31 CONTROLS VASCULAR DEVELOPMENT 3629
platform, whose increased sensitivity facilitated the detection
of at least twice as many miRNAs in HUVECs as previously
reported (19, 34, 50, 58, 66). We also found that one of the
most highly expressed HUVEC miRNAs, miR-126 (19, 34, 50,
58, 66), was ?600 times more abundant in both endothelial cell
types than in either keratinocytes or fibroblasts. Comparative
analysis identified four BVEC and two LEC signature
miRNAs. Of the four BVEC signature miRNAs, three were
previously reported as highly expressed in HUVECs (19, 34,
50, 58, 66), and a very recent study has demonstrated that
tumor necrosis factor treatment augments miR-31 expression
in HUVECs (60). Moreover, our miRNA profiling study has
further classified the widely expressed (38), metastasis-associ-
ated (63, 64) miRNA miR-31 as a BVEC signature miRNA.
Finally, in agreement with their LEC-specific expression, nei-
ther miR-95 nor miR-326 was detected in the previous studies.
miR-125b, and miR-99a in adult mouse tissues confirmed that
FIG. 6. miR-31 overexpression in Xenopus embryos impairs lymphatic vessel sprouting from the lymph heart. Two-cell-stage Xenopus embryos
were coinjected with pre-miR-31 precursor (50 ng) or pre-miR-Neg (100 ng) control molecules and 0.2 ng of ?-Gal mRNA (lineage tracer) and
raised to stage 39. (A and B) Lymphatic vascular system development was monitored by whole-mount ISH for prox1 (A) and vegfr3 (B).
Pre-miR-31-injected embryos exhibited marked defects in lymphatic vascular development, the most striking of which was impaired lymphatic
vessel sprouting (arrowheads) from the lymph heart (arrow). (C) pecam1 ISH showed that intersomitic vein growth (arrowheads) was unilaterally
misguided or delayed in embryos with elevated levels of miR-31.
3630 LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.
their vascular lineage-specific expression patterns were main-
tained in vivo. The degrees of lineage-specific expression dif-
ferences in vivo were, however, usually less pronounced and
more variable than those observed in vitro. This is likely due to
the mixed populations of BVECs and LECs isolated from the
multiple vessel types present in the colon tissue (capillaries,
postcapillary venules, lymphatic capillaries, lymphatic collect-
ing vessels, etc.), which likely exhibit different gene expression
patterns. Moreover, their relative contributions to the isolated
total RNA might vary, thus contributing to larger variability in
miRNA expression patterns. In addition, the ex vivo miRNA
expression profiling studies were technically challenging as the
whole process took more than 2 h and only a few thousand
endothelial cells could be isolated by high-speed cell sorting.
Consequently, the smaller amounts of isolated total RNA,
reduced RNA quality, and possible gene expression changes
incurred during the 2-h isolation procedures likely contributed
to the observed differences in in vivo and in vitro miR-31
expression, as well as to the observed interindividual variability
in miR-31 expression. Surprisingly, we were unable to confirm
FIG. 7. miR-31 overexpression decreases venous sprouting in ze-
brafish embryos. (A) plcg1t26480mutant embryos in the TG(fli1a:gfp)y1
background were injected with 340 pg of negative-control pre-miR-
Neg (n ? 71) or 340 (n ? 46) or 170 (n ? 24) pg of human pre-miR-31
precursor. Venous sprouts were quantified at 48 hpf. UIC, uninjected
control (n ? 120). Data are shown as averages ? standard deviations
relative to the uninjected control of the same clutch.***, P ? 0.001.
(B to D) Lateral views of plcg1t26480mutant embryos in the TG(fli1a:
gfp)y1background. Dorsal aorta formation and sprouting of interseg-
mental arteries are suppressed in mutant embryos, and therefore only
venous sprouts (examples indicated by arrows) and parachordal lym-
phangioblasts (asterisks) are visible in the dorsal trunk. Examples of an
uninjected control plcg1t26480mutant embryo (B) and plcg1t26480mu-
tant embryos injected with 340 pg of negative-control pre-miR-Neg
(C) or 340 pg of human pre-miR-31 precursor (D) at day 2 postfertil-
ization are shown. PCV, posterior cardinal vein.
TABLE 2. Analysis of Xenopus embryos injected with
pre-miR-31 and pre-miR-Nega
Treatment (% survival
at stage 39-40)
and ISH marker
1 ng pre-miR-Neg (84)
10 ng pre-miR-Neg (84)
100 ng pre-miR-Neg (79)
1 ng pre-miR-31 (74)
10 ng pre-miR-31 (57)
25 ng pre-miR-31 (66)
50 ng pre-miR-31 (52)
aLymphatic and blood vascular development was monitored by whole-mount
ISH of embryos microinjected with pre-miR-31 and pre-miR-Neg. The number
and frequency of lymphatic and blood vascular developmental defects, as scored
based on the presence of several lymphatic and blood vascular structures, are
TABLE 3. Analysis of zebrafish embryos injected with
pre-miR-31 and pre-miR-Nega
SDnP value % of UIC
aVenous sprouting was monitored by fluorescence confocal microscopy in
TG(fli1a:gfp)y1; plcg1t26480mutant zebrafish embryos injected with pre-miR-31
and pre-miR-Neg. The number of venous sprouts was normalized to the unin-
fected control (UIC) and is represented as the relative number of venous sprouts.
VOL. 30, 2010 miR-31 CONTROLS VASCULAR DEVELOPMENT3631
the differential expression patterns of miR-137 in vivo. This is
likely because miR-137 expression levels were very low in the
adult tissues analyzed here, as indicated by the late qRT-PCR
detection (CT, ?35) and high standard deviations between
technical replicates. ISH analysis of chicken embryos revealed
that miR-137 is expressed in blood vessels and cardinal veins at
stage 25 of embryonic development (8), demonstrating that
miR-137 expression is associated with the developing blood
The identification of vascular lineage-specific miRNAs sug-
gested that they might regulate fundamental and lineage-spe-
cific endothelial cell functions and/or differentiation processes.
Indeed, overexpression of the BVEC-specific miRNA miR-31
in LECs induced the preferential degradation of LEC signa-
ture genes, including those for the well-characterized lym-
phatic transcription factors PROX1 and FOXC2. As these
lymphatic lineage-specific molecules act as molecular switches,
their preferential suppression suggests that BVEC-specific
posttranscriptional regulatory mechanisms help maintain
BVEC phenotypes by suppressing lymphatic lineage-specific
transcription programs. This concept was supported by our
findings that ectopic overexpression of miR-31 in LECs pref-
erentially repressed LEC signature gene expression and in-
duced BVEC signature gene expression. In this respect, our
identification and validation of PROX1 as a direct miR-31
target are intriguing, as BVEC-specific posttranscriptional reg-
ulation of PROX1 could, at least in part, explain these in vitro
miR-31-mediated reprogramming events on the molecular
level. Indeed, previous studies have demonstrated that PROX1
overexpression in BVECs induces the expression of lymphatic
vascular markers and suppresses blood vascular markers (26,
48), whereas PROX1 knockdown in LECs inhibits LEC signa-
ture gene expression and triggers BVEC signature gene ex-
pression (44; Shin et al., unpublished). Moreover, the overlaps
between the miR-31-regulated genes identified here and a
PROX1 loss-of-function data set further indicate that tran-
scriptional reprogramming events observed following miR-31
overexpression in LECs were, in part, mediated by miR-31
repression of Prox1. Additional experiments are required to
determine which of the miR-31-regulated candidate Prox1 tar-
get genes may also be direct targets of miR-31.
While PROX1 was not a predicted target gene of miR-31
(16, 17, 27, 31, 39), our manual miR-31 site prediction analyses
of the 5.4-kb PROX1 3? UTR and subsequent luciferase 3?
UTR tethering assays identified a bona fide miR-31 binding
site between nt 949 and 971 of the PROX1 3? UTR. Interest-
ingly, similar manual miR-31 prediction analyses of the chim-
panzee, mouse, rat, chicken, Xenopus, and zebrafish PROX1 3?
UTRs revealed that this site is evolutionarily conserved in
vertebrates and identified additional, potentially functional,
miR-31 binding sites (see Table S6 in the supplemental mate-
rial). Taken together, our transcriptome profiling and bio-
chemical studies have revealed a novel, highly conserved,
BVEC-specific posttranscriptional regulatory mechanism that
suppresses PROX1 expression in the blood vasculature.
Our findings also suggested that miR-31 expression in the
developing blood vascular endothelium could regulate the ac-
quisition of lymphatic lineage-specific characteristics and, thus,
vascular development in vivo. Multiple miR-31 loss-of-function
studies using morpholino oligonucleotides were performed
with both wild-type and plcg1 mutant zebrafish embryos. Sta-
tistically significant vascular phenotype differences were not
observed in zebrafish embryos injected with low-to-moderate
amounts (?10 ng) of MO (data not shown). This suggests that
the miR31-mediated regulation of vascular development iden-
tified here is redundant. This is not surprising, since miRNAs
frequently function cooperatively (3, 15, 17), which in turn
complicates the attribution of specific functions to individual
miRNAs (53). In contrast, miRNA gain-of-function experi-
ments have proven very informative and have defined impor-
tant biological functions of several miRNAs (42, 43, 53). For
example, overexpression studies with Xenopus embryos have
demonstrated that miR-15 and miR-16 restrict the size of Spe-
mann’s organizer in vivo by targeting the nodal type II receptor
acrvr2a (42). We therefore carried out miR-31 overexpression
studies with Xenopus and zebrafish embryos to determine the
effect of miR-31 on cells and tissues that normally do not
express miR-31, such as the lymphatic vasculature.
Our gain-of-function experiments clearly demonstrated that
miR-31 expression is incompatible with normal lymphatic vas-
cular development in Xenopus and, to a lesser extent, zebrafish
embryos. The analysis of Xenopus embryos suggests that some
aspects of lymphatic vascular development, such as specifica-
tion of lymph hearts and LECs in the tail, are unaffected by
miR-31 overexpression. Lymphangiogenesis and the develop-
ment of an extensive lymphatic vasculature in the embryonic
trunk are, however, clearly reduced and/or disrupted. Further-
more, we demonstrated that these observed lymphatic defects
were reminiscent of those observed following MO-mediated
inhibition of vegfc. These phenotypic similarities indicate that
miR-31 overexpression interferes with an early step in lym-
phatic development. The identification of evolutionarily con-
served miR-31 binding sites in PROX1 3? UTRs (see Table S6
in the supplemental material) suggests that miR-31 overex-
pression may directly target and interfere with PROX1 tran-
scripts in vivo. Moreover, the abnormal or disrupted inter-
somitic vein sprouting seen in Xenopus and zebrafish embryos
(data not shown) following miR-31 overexpression implies that
miR-31 also regulates BVEC responsiveness to the environ-
mental stimuli directing blood vascular growth and maturation.
Interestingly, several genes involved in the Slit/Robo, netrin,
and ephrin signaling pathways (see Table S3 in the supplemen-
tal material), which provide crucial guidance cues during blood
vascular development (1), were repressed following miR-31
gain of function in vitro. In vivo posttranscriptional regulation
of any one of these molecules by miR-31 could contribute to
the observed blood vascular maturation defects. Taken to-
gether, our results indicate that appropriate expression of
miR-31 during vertebrate embryogenesis is required for both
lymphatic vascular development and blood vascular growth
and maturation. Interestingly, our in vivo studies also correlate
well with a recent study demonstrating that miR-31 controls
the invasive capacity of breast cancer cells (63, 64). Collec-
tively, these studies suggest roles for miR-31 in the regulation
of cell migratory behavior during normal embryonic develop-
ment and under pathological conditions in the adult body.
On the basis of our in vitro studies, we postulate that PROX1
transcripts represent one of the key targets of miR-31. This
repression would prevent inappropriate and/or premature
transcriptional activation of lymphatic differentiation in the
3632 LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.
developing blood vasculature. While this notion is an attractive
model, it is, however, important to stress that miR-31 targets
several other LEC signature genes. It is therefore unlikely that
posttranscriptional repression of PROX1 by miR-31 is solely
responsible for the vascular developmental defects observed in
Xenopus and zebrafish embryos overexpressing miR-31. For
example, miR-31-mediated repression of FOXC2, a transcrip-
tion factor that is required for specification of the lymphatic
capillaries versus collecting lymphatic vessels at later stages of
embryogenesis (1, 7), may also contribute to the vascular de-
fects observed. Another miR-31 candidate target is RAMP2, a
calcitonin receptor-like receptor-associated receptor activity-
modifying protein that triggers lymphangiogenesis in response
to adrenomedullin signaling (13). Finally, other LEC signature
molecules subject to miR-31 regulation, whose lymphatic lin-
eage-specific functions have not yet been characterized, could
also enhance the effects miR-31 has on lymphatic and blood
The miRNAs profiled in the present study represent approx-
imately 25% of the known human miRNAs. Thus, more com-
prehensive and global miRNA profiling studies may result in
the identification of additional endothelial lineage-specific
miRNAs. In summary, we have defined the first vascular lin-
eage-specific miRNAs and identified with miR-31 a novel
miRNA-mediated regulatory mechanism that inhibits LEC
phenotype acquisition in vitro and vascular development in
vivo. From a therapeutic perspective, it remains to be investi-
gated whether the ectopic expression of miR-31 might also
inhibit malignant tumor-associated (lymph)angiogenesis, thus
preventing tumor growth and cancer metastasis.
This work was supported by National Institutes of Health grant
CA69184; Swiss National Science Foundation grant 3100A0-108207;
Austrian Science Foundation grant S9408-B11; Cancer League Zu ¨rich,
Commission of the European Communities, grant LSHC-CT-2005-
518178 (M.D.); Swiss National Science Foundation grant 3100A0-
101964 (A.W.B.); Netherlands Organization for Scientific Research
(NWO) Venigrant (T.K.); and EMBO long-term fellowships ALTF
1104-2007 (D.M.L.P.) and ALTF 52-2007 (T.K.).
We thank Young Kwon Hong for the PROX1 3? UTR plasmid
(YH1551); Salvatore Oliviero, Jay W. Shin, and Ahmad Salameh for
sharing the PROX1 lentivirus knockdown microarray data set; Patrick
Pedrioli for bioinformatic assistance and critical reading of the manu-
script; and Jana Zielinski, Cornelius Fischer, and Jeannette Scholl for
expert technical assistance. We also thank the Tu ¨bingen 2000 Screen
Consortium for identifying the plcg1t26480allele.
D.M.L.P. and T.K. designed and performed research experiments,
analyzed the data, and wrote the manuscript. V.D., G.J., G.V.D.H.,
and R.E.K. designed and performed research experiments and ana-
lyzed the data. D.M., J.W.S., S.L., and P.C. performed research exper-
iments and analyzed the data. M.D., A.W.B., and S.S.-M. designed
research experiments, analyzed the data, and wrote the manuscript.
1. Adams, R. H., and K. Alitalo. 2007. Molecular regulation of angiogenesis and
lymphangiogenesis. Nat. Rev. Mol. Cell Biol. 8:464–478.
2. Amatschek, S., E. Kriehuber, W. Bauer, B. Reininger, P. Meraner, A. Wolpl,
N. Schweifer, C. Haslinger, G. Stingl, and D. Maurer. 2007. Blood and
lymphatic endothelial cell-specific differentiation programs are stringently
controlled by the tissue environment. Blood 109:4777–4785.
3. Antonov, A. V., S. Dietmann, P. Wong, D. Lutter, and H. W. Mewes. 2009.
GeneSet2miRNA: finding the signature of cooperative miRNA activities in
the gene lists. Nucleic Acids Res. 37:W323–W328.
4. Boukamp, P., R. T. Petrussevska, D. Breitkreutz, J. Hornung, A. Markham,
and N. E. Fusenig. 1988. Normal keratinization in a spontaneously immor-
talized aneuploid human keratinocyte cell line. J. Cell Biol. 106:761–771.
5. Chen, C., D. A. Ridzon, A. J. Broomer, Z. Zhou, D. H. Lee, J. T. Nguyen, M.
Barbisin, N. L. Xu, V. R. Mahuvakar, M. R. Andersen, K. Q. Lao, K. J.
Livak, and K. J. Guegler. 2005. Real-time quantification of microRNAs by
stem-loop RT-PCR. Nucleic Acids Res. 33:e179.
6. Cheng, A. M., M. W. Byrom, J. Shelton, and L. P. Ford. 2005. Antisense
inhibition of human miRNAs and indications for an involvement of miRNA
in cell growth and apoptosis. Nucleic Acids Res. 33:1290–1297.
7. Cueni, L. N., and M. Detmar. 2006. New insights into the molecular control
of the lymphatic vascular system and its role in disease. J. Invest. Dermatol.
8. Darnell, D. K., S. Kaur, S. Stanislaw, S. Davey, J. H. Konieczka, T. A.
Yatskievych, and P. B. Antin. 2007. GEISHA: an in situ hybridization gene
expression resource for the chicken embryo. Cytogenet. Genome Res. 117:
9. De Val, S., N. C. Chi, S. M. Meadows, S. Minovitsky, J. P. Anderson, I. S.
Harris, M. L. Ehlers, P. Agarwal, A. Visel, S. M. Xu, L. A. Pennacchio, I.
Dubchak, P. A. Krieg, D. Y. Stainier, and B. L. Black. 2008. Combinatorial
regulation of endothelial gene expression by ets and forkhead transcription
factors. Cell 135:1053–1064.
10. Filipowicz, W., S. N. Bhattacharyya, and N. Sonenberg. 2008. Mechanisms of
post-transcriptional regulation by microRNAs: are the answers in sight? Nat.
Rev. Genet. 9:102–114.
11. Fish, J. E., and D. Srivastava. 2009. MicroRNAs: opening a new vein in
angiogenesis research. Sci. Signal. 2:pe1.
12. Franc ¸ois, M., A. Caprini, B. Hosking, F. Orsenigo, D. Wilhelm, C. Browne,
K. Paavonen, T. Karnezis, R. Shayan, M. Downes, T. Davidson, D. Tutt,
K. S. Cheah, S. A. Stacker, G. E. Muscat, M. G. Achen, E. Dejana, and P.
Koopman. 2008. Sox18 induces development of the lymphatic vasculature in
mice. Nature 456:643–647.
13. Fritz-Six, K. L., W. P. Dunworth, M. Li, and K. M. Caron. 2008. Ad-
renomedullin signaling is necessary for murine lymphatic vascular develop-
ment. J. Clin. Invest. 118:40–50.
14. Gentleman, R. C., V. J. Carey, D. M. Bates, B. Bolstad, M. Dettling, S.
Dudoit, B. Ellis, L. Gautier, Y. Ge, J. Gentry, K. Hornik, T. Hothorn, W.
Huber, S. Iacus, R. Irizarry, F. Leisch, C. Li, M. Maechler, A. J. Rossini, G.
Sawitzki, C. Smith, G. Smyth, L. Tierney, J. Y. Yang, and J. Zhang. 2004.
Bioconductor: open software development for computational biology and
bioinformatics. Genome Biol. 5:R80.
15. Georges, S. A., M. C. Biery, S. Y. Kim, J. M. Schelter, J. Guo, A. N.
Chang, A. L. Jackson, M. O. Carleton, P. S. Linsley, M. A. Cleary, and
B. N. Chau. 2008. Coordinated regulation of cell cycle transcripts by
p53-inducible microRNAs, miR-192 and miR-215. Cancer Res. 68:10105–
16. Griffiths-Jones, S., H. K. Saini, S. van Dongen, and A. J. Enright. 2008.
miRBase: tools for microRNA genomics. Nucleic Acids Res. 36:D154–D158.
17. Grimson, A., K. K. Farh, W. K. Johnston, P. Garrett-Engele, L. P. Lim, and
D. P. Bartel. 2007. MicroRNA targeting specificity in mammals: determi-
nants beyond seed pairing. Mol. Cell 27:91–105.
18. Halin, C., N. E. Tobler, B. Vigl, L. F. Brown, and M. Detmar. 2007. VEGF-A
produced by chronically inflamed tissue induces lymphangiogenesis in drain-
ing lymph nodes. Blood 110:3158–3167.
19. Harris, T. A., M. Yamakuchi, M. Ferlito, J. T. Mendell, and C. J. Lowen-
stein. 2008. MicroRNA-126 regulates endothelial expression of vascular cell
adhesion molecule 1. Proc. Natl. Acad. Sci. U. S. A. 105:1516–1521.
20. Helbling, P. M., D. M. Saulnier, and A. W. Brandli. 2000. The receptor
tyrosine kinase EphB4 and ephrin-B ligands restrict angiogenic growth of
embryonic veins in Xenopus laevis. Development 127:269–278.
21. Helbling, P. M., C. T. Tran, and A. W. Brandli. 1998. Requirement for EphA
receptor signaling in the segregation of Xenopus third and fourth arch neural
crest cells. Mech. Dev. 78:63–79.
22. Hirakawa, S., L. F. Brown, S. Kodama, K. Paavonen, K. Alitalo, and M.
Detmar. 2007. VEGF-C-induced lymphangiogenesis in sentinel lymph nodes
promotes tumor metastasis to distant sites. Blood 109:1010–1017.
23. Hirakawa, S., Y. K. Hong, N. Harvey, V. Schacht, K. Matsuda, T. Liber-
mann, and M. Detmar. 2003. Identification of vascular lineage-specific genes
by transcriptional profiling of isolated blood vascular and lymphatic endo-
thelial cells. Am. J. Pathol. 162:575–586.
24. Hogan, B. M., F. L. Bos, J. Bussmann, M. Witte, N. C. Chi, H. J. Duckers,
and S. Schulte-Merker. 2009. Ccbe1 is required for embryonic lymphangio-
genesis and venous sprouting. Nat. Genet. 41:396–398.
25. Hong, Y. K., K. Foreman, J. W. Shin, S. Hirakawa, C. L. Curry, D. R. Sage,
T. Libermann, B. J. Dezube, J. D. Fingeroth, and M. Detmar. 2004. Lym-
phatic reprogramming of blood vascular endothelium by Kaposi sarcoma-
associated herpesvirus. Nat. Genet. 36:683–685.
26. Hong, Y. K., N. Harvey, Y. H. Noh, V. Schacht, S. Hirakawa, M. Detmar, and
G. Oliver. 2002. Prox1 is a master control gene in the program specifying
lymphatic endothelial cell fate. Dev. Dyn. 225:351–357.
27. John, B., A. J. Enright, A. Aravin, T. Tuschl, C. Sander, and D. S. Marks.
2004. Human MicroRNA targets. PLoS Biol. 2:e363.
28. Ka ¨lin, R. E., N. E. Banziger-Tobler, M. Detmar, and A. W. Brandli. 2009. An
in vivo chemical library screen in Xenopus tadpoles reveals novel pathways
involved in angiogenesis and lymphangiogenesis. Blood 114:1110–1122.
VOL. 30, 2010 miR-31 CONTROLS VASCULAR DEVELOPMENT3633
29. Ka ¨lin, R. E., M. P. Kretz, A. M. Meyer, A. Kispert, F. L. Heppner, and A. W. Download full-text
Brandli. 2007. Paracrine and autocrine mechanisms of apelin signaling gov-
ern embryonic and tumor angiogenesis. Dev. Biol. 305:599–614.
30. Kiriakidou, M., P. T. Nelson, A. Kouranov, P. Fitziev, C. Bouyioukos, Z.
Mourelatos, and A. Hatzigeorgiou. 2004. A combined computational-exper-
imental approach predicts human microRNA targets. Genes Dev. 18:1165–
31. Krek, A., D. Grun, M. N. Poy, R. Wolf, L. Rosenberg, E. J. Epstein, P.
MacMenamin, I. da Piedade, K. C. Gunsalus, M. Stoffel, and N. Rajewsky.
2005. Combinatorial microRNA target predictions. Nat. Genet. 37:495–500.
32. Kriehuber, E., S. Breiteneder-Geleff, M. Groeger, A. Soleiman, S. F. Schop-
pmann, G. Stingl, D. Kerjaschki, and D. Maurer. 2001. Isolation and char-
acterization of dermal lymphatic and blood endothelial cells reveal stable
and functionally specialized cell lineages. J. Exp. Med. 194:797–808.
33. Ku ¨chler, A. M., E. Gjini, J. Peterson-Maduro, B. Cancilla, H. Wolburg, and
S. Schulte-Merker. 2006. Development of the zebrafish lymphatic system
requires VEGFC signaling. Curr. Biol. 16:1244–1248.
34. Kuehbacher, A., C. Urbich, A. M. Zeiher, and S. Dimmeler. 2007. Role of
Dicer and Drosha for endothelial microRNA expression and angiogenesis.
Circ. Res. 101:59–68.
35. Kunstfeld, R., S. Hirakawa, Y. K. Hong, V. Schacht, B. Lange-Asschenfeldt,
P. Velasco, C. Lin, E. Fiebiger, X. Wei, Y. Wu, D. Hicklin, P. Bohlen, and M.
Detmar. 2004. Induction of cutaneous delayed-type hypersensitivity reac-
tions in VEGF-A transgenic mice results in chronic skin inflammation asso-
ciated with persistent lymphatic hyperplasia. Blood 104:1048–1057.
36. Lawson, N. D., J. W. Mugford, B. A. Diamond, and B. M. Weinstein. 2003.
Phospholipase C gamma-1 is required downstream of vascular endothelial
growth factor during arterial development. Genes Dev. 17:1346–1351.
37. Lawson, N. D., and B. M. Weinstein. 2002. In vivo imaging of embryonic
vascular development using transgenic zebrafish. Dev. Biol. 248:307–318.
38. Lee, E. J., M. Baek, Y. Gusev, D. J. Brackett, G. J. Nuovo, and T. D.
Schmittgen. 2008. Systematic evaluation of microRNA processing patterns
in tissues, cell lines, and tumors. RNA 14:35–42.
39. Lewis, B. P., C. B. Burge, and D. P. Bartel. 2005. Conserved seed pairing,
often flanked by adenosines, indicates that thousands of human genes are
microRNA targets. Cell 120:15–20.
40. Ma ¨kinen, T., C. Norrmen, and T. V. Petrova. 2007. Molecular mechanisms of
lymphatic vascular development. Cell. Mol. Life Sci. 64:1915–1929.
41. Marquez, R. T., and A. P. McCaffrey. 2008. Advances in microRNAs: im-
plications for gene therapists. Hum. Gene Ther. 19:27–38.
42. Martello, G., L. Zacchigna, M. Inui, M. Montagner, M. Adorno, A. Mamidi,
L. Morsut, S. Soligo, U. Tran, S. Dupont, M. Cordenonsi, O. Wessely, and S.
Piccolo. 2007. MicroRNA control of nodal signalling. Nature 449:183–188.
43. Mayoral, R. J., M. E. Pipkin, M. Pachkov, E. van Nimwegen, A. Rao, and S.
Monticelli. 2009. MicroRNA-221-222 regulate the cell cycle in mast cells.
J. Immunol. 182:433–445.
44. Mishima, K., T. Watabe, A. Saito, Y. Yoshimatsu, N. Imaizumi, S. Masui, M.
Hirashima, T. Morisada, Y. Oike, M. Araie, H. Niwa, H. Kubo, T. Suda, and
K. Miyazono. 2007. Prox1 induces lymphatic endothelial differentiation via
integrin alpha9 and other signaling cascades. Mol. Biol. Cell 18:1421–1429.
45. Ny, A., M. Koch, M. Schneider, E. Neven, R. T. Tong, S. Maity, C. Fischer,
S. Plaisance, D. Lambrechts, C. Heligon, S. Terclavers, M. Ciesiolka, R.
Ka ¨lin, W. Y. Man, I. Senn, S. Wyns, F. Lupu, A. Brandli, K. Vleminckx, D.
Collen, M. Dewerchin, E. M. Conway, L. Moons, R. K. Jain, and P. Carme-
liet. 2005. A genetic Xenopus laevis tadpole model to study lymphangiogen-
esis. Nat. Med. 11:998–1004.
46. Osada, H., and T. Takahashi. 2007. MicroRNAs in biological processes and
carcinogenesis. Carcinogenesis 28:2–12.
47. Pena, J. T., C. Sohn-Lee, S. H. Rouhanifard, J. Ludwig, M. Hafner, A.
Mihailovic, C. Lim, D. Holoch, P. Berninger, M. Zavolan, and T. Tuschl.
2009. miRNA in situ hybridization in formaldehyde and EDC-fixed tissues.
Nat. Methods 6:139–141.
48. Petrova, T. V., T. Ma ¨kinen, T. P. Makela, J. Saarela, I. Virtanen, R. E.
Ferrell, D. N. Finegold, D. Kerjaschki, S. Yla-Herttuala, and K. Alitalo.
2002. Lymphatic endothelial reprogramming of vascular endothelial cells by
the Prox-1 homeobox transcription factor. EMBO J. 21:4593–4599.
49. Petrova, T. V., A. Nykanen, C. Norrmen, K. I. Ivanov, L. C. Andersson, C.
Haglund, P. Puolakkainen, F. Wempe, H. von Melchner, G. Gradwohl, S.
Vanharanta, L. A. Aaltonen, J. Saharinen, M. Gentile, A. Clarke, J. Taipale,
G. Oliver, and K. Alitalo. 2008. Transcription factor PROX1 induces colon
cancer progression by promoting the transition from benign to highly dys-
plastic phenotype. Cancer Cell 13:407–419.
50. Poliseno, L., A. Tuccoli, L. Mariani, M. Evangelista, L. Citti, K. Woods, A.
Mercatanti, S. Hammond, and G. Rainaldi. 2006. MicroRNAs modulate the
angiogenic properties of HUVECs. Blood 108:3068–3071.
51. Rao, D. D., N. Senzer, M. A. Cleary, and J. Nemunaitis. 2009. Comparative
assessment of siRNA and shRNA off target effects: what is slowing clinical
development. Cancer Gene Ther. 16:807–809.
52. Roesli, C., V. Mumprecht, D. Neri, and M. Detmar. 2008. Identification of
the surface-accessible, lineage-specific vascular proteome by two-dimen-
sional peptide mapping. FASEB J. 22:1933–1944.
53. Shan, S. W., D. Y. Lee, Z. Deng, T. Shatseva, Z. Jeyapalan, W. W. Du, Y.
Zhang, J. W. Xuan, S. P. Yee, V. Siragam, and B. B. Yang. 2009. MicroRNA
MiR-17 retards tissue growth and represses fibronectin expression. Nat. Cell
54. Shin, J. W., R. Huggenberger, and M. Detmar. 2008. Transcriptional profil-
ing of VEGF-A and VEGF-C target genes in lymphatic endothelium reveals
endothelial-specific molecule-1 as a novel mediator of lymphangiogenesis.
55. Smyth, G. K. 2004. Linear models and empirical Bayes methods for assessing
differential expression in microarray experiments. Stat. Appl. Genet. Mol.
56. Srinivasan, R. S., M. E. Dillard, O. V. Lagutin, F. J. Lin, S. Tsai, M. J. Tsai,
I. M. Samokhvalov, and G. Oliver. 2007. Lineage tracing demonstrates the
venous origin of the mammalian lymphatic vasculature. Genes Dev. 21:2422–
57. Steffensen, K. R., E. Holter, A. Bavner, M. Nilsson, M. Pelto-Huikko, S.
Tomarev, and E. Treuter. 2004. Functional conservation of interactions
between a homeodomain cofactor and a mammalian FTZ-F1 homologue.
EMBO Rep. 5:613–619.
58. Sua ´rez, Y., C. Fernandez-Hernando, J. S. Pober, and W. C. Sessa. 2007.
Dicer dependent microRNAs regulate gene expression and functions in
human endothelial cells. Circ. Res. 100:1164–1173.
59. Sua ´rez, Y., and W. C. Sessa. 2009. MicroRNAs as novel regulators of an-
giogenesis. Circ. Res. 104:442–454.
60. Sua ´rez, Y., C. Wang, T. D. Manes, and J. S. Pober. 2010. Cutting edge:
TNF-induced microRNAs regulate TNF-induced expression of E-selectin
and intercellular adhesion molecule-1 on human endothelial cells: feedback
control of inflammation. J. Immunol. 184:21–25.
61. Thomas, P. D., M. J. Campbell, A. Kejariwal, H. Mi, B. Karlak, R. Daver-
man, K. Diemer, A. Muruganujan, and A. Narechania. 2003. PANTHER: a
library of protein families and subfamilies indexed by function. Genome Res.
62. Thomas, P. D., A. Kejariwal, M. J. Campbell, H. Mi, K. Diemer, N. Guo, I.
Ladunga, B. Ulitsky-Lazareva, A. Muruganujan, S. Rabkin, J. A. Vander-
griff, and O. Doremieux. 2003. PANTHER: a browsable database of gene
products organized by biological function, using curated protein family and
subfamily classification. Nucleic Acids Res. 31:334–341.
63. Valastyan, S., N. Benaich, A. Chang, F. Reinhardt, and R. A. Weinberg. 2009.
Concomitant suppression of three target genes can explain the impact of a
microRNA on metastasis. Genes Dev. 23:2592–2597.
64. Valastyan, S., F. Reinhardt, N. Benaich, D. Calogrias, A. M. Szasz, Z. C.
Wang, J. E. Brock, A. L. Richardson, and R. A. Weinberg. 2009. A pleiotro-
pically acting microRNA, miR-31, inhibits breast cancer metastasis. Cell
65. Wigle, J. T., N. Harvey, M. Detmar, I. Lagutina, G. Grosveld, M. D. Gunn,
D. G. Jackson, and G. Oliver. 2002. An essential role for Prox1 in the
induction of the lymphatic endothelial cell phenotype. EMBO J. 21:1505–
66. Yang, W. J., D. D. Yang, S. Na, G. E. Sandusky, Q. Zhang, and G. Zhao.
2005. Dicer is required for embryonic angiogenesis during mouse develop-
ment. J. Biol. Chem. 280:9330–9335.
67. Yaniv, K., S. Isogai, D. Castranova, L. Dye, J. Hitomi, and B. M. Weinstein.
2006. Live imaging of lymphatic development in the zebrafish. Nat. Med.
68. Zhao, Y., and D. Srivastava. 2007. A developmental view of microRNA
function. Trends Biochem. Sci. 32:189–197.
3634LESLIE PEDRIOLI ET AL.MOL. CELL. BIOL.