|
|
|
|
|
|
|
||||
| Home Help Feedback Subscriptions Archive Search Table of Contents | ||||
First published online August 18, 2003
doi: 10.1242/10.1242/dev.00635
1 Cancer and Developmental Biology Laboratory, Division of Basic Science,
National Cancer Institute, Frederick, MD 21702, USA
2 Regulation of Cell Growth Laboratory, Division of Basic Science, National
Cancer Institute, Frederick, MD 21702, USA
3 Department of Biochemistry and Molecular Biophysics, Medical College of
Virginia Campus, Virginia Commonwealth University, Richmond, VA
23298,USA
4 Department of Cell and Developmental Biology, University of North Carolina at
Chapel Hill, Chapel Hill, NC 27599-7090, USA
Author for correspondence (e-mail:
stewartc{at}ncifcrdc.gov)
Accepted 29 May 2003
 |
SUMMARY |
|---|
 TOP SUMMARY IntroductionMaterials and methodsResultsDiscussionREFERENCES |
|---|
Key words: Lipid phosphate phosphatase, Vasculogenesis, Wnt, Axis duplication
|
Introduction |
|---|
|
TOPSUMMARYIntroductionMaterials and methodsResultsDiscussionREFERENCES |
|---|
;
Liu et al., 1999).
Bioactive phospholipids are synthesized and degraded by a complex set of
metabolic pathways. In adult mammals they are present at nano- to micromolar
concentrations in serum with their principal source being activated platelets
and cells stimulated by growth factors and cytokines
(Yatomi et al., 1997). The
phospholipids, S1P and LPA, act on cells by binding to the S1P and LPA
receptors [formerly the Edg receptors
(Chun et al., 2002)], G
protein coupled transmembrane receptors
(Fukushima and Chun, 2001;
Hla, 2001). Activation of
these receptors enhances adhesion, migration and morphogenesis of capillary
endothelial cells, as well as inhibiting T-cell apoptosis. However, the
mitogenic effects of phospholipids on other cell types may also be mediated by
their action as intracellular second messengers, although they may also be
acting through other, as yet unidentified, receptors.
Given their effects on cells, recent evidence has revealed that these lipid
mediators are important to embryogenesis, particularly in guiding cell
migration. This was first shown by the demonstration that the products of the
Drosophila genes wunen and wunen2, are lipid
phosphate phosphatases regulating germ cell migration during development
(Starz-Gaiano et al., 2001;
Zhang et al., 1997). In
zebrafish, the mutation miles apart results in a failure of the heart
primordia to migrate to the midline and subsequently fuse. The altered gene
encodes a protein with high homology to the lysosphingolipid receptor
S1P2 (Kupperman et al.,
2000). In mice, mutations in some of the S1P receptors, affect
development, with S1P1 deficiency resulting in mid-gestational
hemorrhage and lethality due to an inability of vascular endothelial smooth
muscle precursor cells to surround the blood vessels. LPA1
deficiency caused a high frequency of perinatal lethality, postnatal growth
defects and a low incidence of hematomas
(Contos et al., 2000;
Liu et al., 2000). Null
mutations in other S1P and LPA receptors including S1P2 and
S1P3 and LPA2 had little overt effect, indicating
considerable redundancy between the receptors
(Contos et al., 2002;
Ishii et al., 2002).
Nevertheless, these results suggests that a strict regulation of the levels of
lipid phosphates during development are required to properly control cellular
responses to these molecules.
The lipid phosphate phosphatases (LPPs) are a group of enzymes involved
both in lipid phosphate biosynthesis and maintaining the balance between
bioactive phosphorylated and dephosphorylated forms
(Sciorra and Morris, 2002).
The LPPs are glycoproteins with a channel-like structure containing six
putative transmembrane domains (Kanoh et
al., 1997) and were first characterized from their ability to
dephosphorylate phosphatidic acid (PA) to produce diacylglycerol (DAG). Since
PA and DAG act as potent signaling molecules, LPPs play a key role in signal
transduction in addition to regulating lipid biosynthesis. Two classes of
mammalian LPPs have been identified. The type 1 (LPP1) is a cytoplasmic,
Mg2+-dependent enzyme, sensitive to N-ethylmaleimide and is
required for glycerolipid biosynthesis. The type 2 LPPs, are membrane bound
enzymes, Mg2+-independent and N-ethylmaleimide-insensitive and are
involved in signal transduction mediated by phospholipase D
(Sciorra and Morris, 2002). In
humans, at least three genes coding for type 2 LPP enzymes have been
identified (LPP1, LPP2 and LPP3)
(Kai et al., 1997;
Roberts et al., 1998). In
addition to PA, all LPPs hydrolyze LPA, ceramide-1-phosphate (C-1-P) and S1P
(Kai et al., 1997;
Roberts et al., 1998). Of the
three LPPs, LPP3 has unique characteristics: it localizes to both the
plasma membrane and intracellular organelles depending on cell type
(Sciorra and Morris, 1999) and
its transcription is stimulated by epidermal growth factor
(Kai et al., 1997).
LPP3 corresponds to the previously identified gene product of
Dri42 from rat (Barila et al.,
1996) that is upregulated during intestinal epithelial
differentiation.
Apart from the role of wunen in regulating Drosophila
germ cell migration, and left-right asymmetry in the gut
(Ligoxygakis et al., 2001),
little is known about the role of these signal modulators in development.
Because of the increasing evidence of phospholipids influencing both cellular
morphology and locomotion we investigated the expression and functions of the
murine LPP (Ppap - Mouse Genome Informatics) homologues in
development. We report that LPP3 exhibits a highly tissue-specific
and dynamic pattern of expression during post-implantation development in
contrast to both LPP1 and LPP2, which are more uniformly and
widely expressed. To analyze the role of LPP3 in development we
introduced a deletion into the LPP3 locus. This mutation revealed
that the enzyme is essential for development of the allantoic and yolk sac
vasculature, as well as the chorioallantoic placenta. In addition a subset of
embryos exhibited profound alterations in axis formation that were similar to
mutations associated with axin deficiency
(Zeng et al., 1997). The
latter phenotype is associated with alterations to the Wnt signaling pathways.
We show that transcriptional activity of the TCF co-factor ß-catenin is
regulated by a previously unidentified function of LPP3 that is independent of
its lipid phosphatase activity.
|
Materials and methods |
|---|
|
TOPSUMMARYIntroductionMaterials and methodsResultsDiscussionREFERENCES |
|---|
Genotyping of mouse embryos by PCR
The extra-embryonic membranes were analyzed by PCR (lysis buffer 50 mM KCl,
10 mM Tris pH 8.3, 2 mM MgCl2, 0.01% gelatin, 0.45% NP-40, 0.45%
Tween 20, 100 µg/ml of proteinase K). Embryos were genotyped using a set of
three oligonucleotides distinguishing the wild type and mutant alleles. Wild
type: 5'-gccttctacacgggattgtcac-3'; mutant PGK forward:
5'-cagaaagcgaaggaacaaagctg-3'; common reverse:
5'-ttgtgctcacagagaagaggattc-3'. The PCR products obtained after 30
cycles yielded fragments with the following sizes: wild type allele 302 bp and
mutant allele 500 bp.
Derivation of homozygous mutant ES cells, embryoid bodies and
production of ES cell-derived embryos
ES cell lines were established from blastocysts of heterozygous
intercrosses as described previously
(Abbondanzo et al., 1993). The
genotype of each clone was verified by Southern blot hybridization and lines
with a normal diploid karyotype were identified. Embryoid bodies were prepared
from ES cell lines essentially as described
(Robertson, 1987). ROSA26
blastocysts (Zambrowicz et al.,
1997) were injected with homozygous mutant
LPP3StwO3 ES cells. Embryos were recovered from
equivalent 9.5 to 10.5 days of gestation. ß-galactosidase staining was
performed to reveal the contribution of wild-type and mutant cells to the
conceptuses.
Whole-mount in situ hybridization
Mouse embryos were fixed and processed for in situ hybridization as
described previously (Hogan, 1994). Antisense RNA probes for brachyury,
Shh, Twist, flk1, hex, wnt3, LPP1 (I.M.A.G.E. Consortium Id clone 732307)
and LPP2 (I.M.A.G.E. Consortium Id clone 766619) were utilized. Frog
embryos whole-mount in situ hybridization was done essentially as described by
Harlan (Harlan, 1991) using
riboprobes for Xpax6, Xotx2, Xkrox20.
ß-galactosidase staining
Embryos were stained as described previously (Hogan, 1994). Embryos were
immersed for 15-30 minutes in fixative solution (0.2% glutaraldehyde, 5 mM
EGTA, 2 mM MgCl2 in PBS), washed 3 times 30 minutes with detergent
rinse (2 mM MgCl2, 0.02% NP-40, 0.01% sodium deoxycholate in PBS)
and incubated overnight at 37°C in staining solution (2 mM
MgCl2, 0.01% sodium deoxycholate, 0.02% NP-40, 5 mM potassium
ferricyanide, 5 mM potassium ferrocyanide, X-gal 1 mg/ml). When necessary
embryos were embedded in paraffin, sectioned at 7 µm and counterstained
with Fast Red.
Histology
Tissues were fixed with 4% paraformaldehyde in PBS overnight, ethanol
dehydrated and embedded in wax. 7 µm thick sections were stained with
Hematoxylin and Eosin. For high resolution microscopy embryos were immersed in
Karnovsky's fixative solution (3% glutaraldehyde, 1% paraformaldehyde in 0.1 M
sodium cacodilate buffer, pH 7.4), postfixed with 1% OsO4, ethanol
dehydrated and embedded in epon 812. Semithin sections (1 µm) were stained
with Toluidine Blue.
Immunohistochemistry
Detection of endothelia in allantois cultures
After culture, explants were fixed for 15 minutes with 4% paraformaldehyde
in PBS and blocked with PBSMT (2% skim milk, 0.5% Tween in PBS), samples were
incubated overnight with 10 µg/ml of anti-mouse PECAM1 antibody
(Pharmingen, MEC13.3) at 4°C followed by 6 washes for 1 hour each in PBSMT
and overnight incubation with 1:100 anti-rat HRP-coupled antibody at 4°C.
After an additional round of washes color reaction was performed in the
presence of DAB and H2O2.
PECAM1 whole-mount immunohistochemistry
Embryos were treated essentially as described previously
(Schlaeger et al., 1995).
Antibody was used at the same concentration as for immunohistochemistry.
Frog embryo injections
PCR generated full-length mRNAs of murine LPP3 and LPP3
with a deletion of amino acids 187-219, containing the third outer loop, were
cloned into the pCS+ vector. Xenopus embryos were microinjected with
mRNA derived from NotI-linearized constructs transcribed with the
mMessage mMachine kit (Ambion). 2.5 ng of LPP3 RNA were coinjected
with 1.15 and 0.46 pg of Xwnt3a and Xwnt8 RNA, respectively,
into the two ventral blastomeres of four-cell embryos.
Western blot
Cells or tissues were lysed with 50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40,
1x complete protease inhibitor cocktail (Roche). 50 µg of protein
were run in denaturing acrylamide gels and transferred to PVDF membranes.
Antibodies were to phospho-pan PKC (Cell Signaling, 9371), Anti-active
ß-catenin (anti-ABC) (Upstate Biotechnology) LPP3
(Sciorra and Morris, 1999),
actin (Santa Cruz), phospho-GSK3-ß (Ser-9) (Cell Signaling, 9336S), GSK-3
(Transduction Laboratories, G22320). Protein was detected using the ECL-plus
system (Amersham).
Cell culture
MEFs were derived from wild type and wild
type
LPP3-/- E13 chimeras. Embryos were isolated,
heart and livers removed and then rinsed with PBS, minced, and digested in 2
ml DMEM containing 100 µg/ml DNaseI and collagenase IV (Sigma) for 30
minutes at 37°C. The cells were pelleted and resuspended in DMEM-10% FBS
and the mutant cells isolated by culturing in the presence of 500 µg/ml
G418. The purity of mutant cells was confirmed by PCR genotyping.
Transient transfection and luciferase reporter assays
HEK293 cells were from the American Type Culture Collection. W9.5 and
LPP3-/- ES cell lines (described above) were grown under
feeder-free conditions (Abbondanzo et al.,
1993). 1x105 cells were co-transfected using
Fugene 6 (Roche) with 2 µg of Renilla luciferase internal standard pRLCMV
(Promega), and/or TOPFlash or FOPFlash firefly luciferase reporter plasmid
(Upstate Biotechnology) with ß-catenin (human full-length cDNA in
pcDNA3.1; courtesy of T. Yamaguchi) or mutant ß-catenin (dominant active,
containing deleted Gsk3ß recognition sites in pcDNA3.1; courtesy of T.
Yamaguchi) and full-length or mutant LPP3cDNA in pIRES-hrGFP-1a.
pcDNA3.1 (In Vitrogen) plasmid was added to standardize DNA quantities. After
48 hours, luciferase activity was determined using the Dual-Luciferase assay
system (Promega) as described by the manufacturer.
Measurement of phosphatidic acid levels
Wild-type and LPP3-/- cells (2x106)
were seeded in 100 mm plates. One day later, the cells were incubated with 5
µCi/ml of [3H]palmitic acid for 24 hours. Lipids were extracted
essentially as previously described with a minor modification
(Zhang et al., 1991). Briefly,
cells were scraped in 2x600 µl methanol/HCl (200:2). The extracts
were sonicated and 600 µl of chloroform was added followed by 500 µl of
H2O, and phases separated by addition of 600 µl of 2 M KCl and
600 µl of chloroform, followed by vortexing and centrifugation. An aliquot
of the organic phase containing 10x106 cpm was spotted on
silica gel plates. The plates were developed in the organic phase of a mixture
of ethyl acetate/2,2,4-trimethylpentane/acetic acid/H2O
(13:2:3:10). The products were revealed by autoradiography and identified by
co-migration with standards. The individual phospholipids were scraped from
TLC plates and radioactivity quantified by liquid scintillation.
Measurement of monoacylglycerol and diacylglycerol levels
Diacylglycerol and monoacylglyerol in cellular extracts were measured by
the diacylglycerol kinase enzymatic method
(Olivera et al., 1997).
Briefly, aliquots (10-50 nmol of total phospholipid) of the chloroform phases
from cellular lipid extracts were resuspended in 40 µl of 7.5% (w/v)
octyl-ß-D-glucopyranoside/5 mM cardiolipin in 1 mM DETPAC/10 mM imidazole
(pH 6.6) and solubilized by freeze-thawing and subsequent sonication. The
enzymatic reaction was started by the addition of 20 µl DTT (20 mM), 10
µl E. coli diacylglycerol kinase (0.88 U/ml), 20 µl
[
-32P]ATP (10-20 µCi, 10 mM) and 100 µl reaction
buffer (100 mM imidazole (pH 6.6), 100 mM NaCl, 25 mM MgCl2, and 2
mM EGTA). After incubation for 1 hour at room temperature, lipids were
extracted with 1 ml chloroform/methanol/HCl (100:100:1, v/v) and 0.17 ml of 1
M KCl. Labeled phosphatidic acid and lysophosphatidic acid were resolved by
TLC with chloroform/acetone/methanol/acetic acid/water (10:4:3:2:1, v/v) and
quantified with a Molecular Dynamics Storm phosphorimager. Known amounts of
diacylglycerol and monoacylglycerol standards were included with each
assay.
Analysis of labeled phospholipids released by MEFs
Wild-type and LPP3-/- cells (2x106)
were seeded on 100 mm plate. One day later, the cells were incubated with 40
mCi/ml of [32P]Pi for 24 hours. Lipids in the extracellular medium
were extracted (Bligh and Dyer,
1959). Briefly, 2 ml of medium were extracted with 5.4 ml of
methanol/CHCl3/HCl (100/200/2). Then, 2.4 ml of 2 M KCl and 2.4 ml
of CHCl3 were added. After phase separation, the organic layer was
dried under nitrogen and dissolved in CHCl3/methanol (3:1). Lipids
were separated by TLC using CHCl3/acetone/methanol/acetic
acid/water (5/4/3/2/1). Labeled lipids were detected and quantified using a
PhosphoImager (Image Quant software, Molecular Dynamics) and data expressed as
fold increase normalized to the total amount of phospholipids.
[32P]LPA was identified by co-migration with unlabeled LPA.
|
Results |
|---|
|
TOPSUMMARYIntroductionMaterials and methodsResultsDiscussionREFERENCES |
|---|
|
day 0.5)
(Fig. 1C,D). The position of
the extra embryonic ectoderm in these stages was confirmed by sectioning the
stained embryos (data not shown). By E7.5, LPP3 starts to be strongly
expressed in the anterior visceral endoderm, as well as in the extra-embryonic
membranes (Fig. 1E,F). By E8.0
expression extends to a highly localized region around the node and appears at
the tip of the allantois (Fig.
1G,H). At E8.5 LPP3 is predominantly expressed in the
allantois, the developing gut, the pericardio-peritoneal canal and somites
(Fig. 1I). LPP3 in
E9.5 embryos persists in the umbilical cord, and is also found in the
chorionic region, probably because of the contribution by the allantois to the
chorio-allantoic placenta (before fusion, the chorion exhibits low levels of
LPP3) (Fig. 1J). In
later mid-gestation embryos LPP3 is present at high levels in the
apical ectodermal ridge and mesenchyme of the limb buds
(Fig. 1K), in the peripheral
nervous system, cranial nerves, and mammary gland primordia (not shown). In
adults significant levels of LPP3 mRNA were localized to the lung,
cerebellum and heart atrium (data not shown), revealing a dynamic and changing
pattern of expression throughout the life cycle of the mouse.
Derivation of mice lacking LPP3 activity
To establish whether LPP3 is required during development, we
inactivated the gene by targeted mutagenesis. A replacement-type targeting
vector was used in which a PGKneo cassette, flanked by loxP
sites (floxed), was introduced so that after recombination, the exon coding
for amino acids 214-268 of the protein, containing part of the presumptive
catalytic domain (Fig. 2A), was
eliminated. Deletion of this region removes part of the fourth, fifth and
sixth transmembrane domains, as well as the second internal loop and third
outer loop of the protein. ES clones heterozygous for the mutated allele were
derived (13/149) (Fig. 2B) and
2 independent lines of mice (12.2B4 and 43.1F3) were established by blastocyst
injection. Both lines showed the same phenotypic characteristics (see below).
This allele is referred to as LPP3StwO3.
|
), protein immunoprecipitated from cultured heterozygous EBs
produced DAG from PA with a mean activity of 21.5 pmol/minute. In contrast
LPP3 from homozygous EBs was unable to do so (mean activity 0.6 pmol/minute)
showing that the LPP3StwO3 mutation in the
homozygous state abolished LPP3 phosphatase activity. Subsequently, we
utilized wild-type and LPP3 homozygous mutant mouse embryonic
fibroblasts (MEFs) to determine changes in other intracellular phospholipids.
Homozygous LPP3StwO3 MEFs had an almost 50%
reduction in the levels of DAG compared to wild types, whereas the levels of
monoacylglycerol (MAG), were not significantly changed
(Table 1). Because the amount
of PA is less than 1% of the total phospholipids, it is difficult to
accurately measure small changes in mass. To determine changes in PA levels,
cells were labeled with [3H]palmitic acid to enhance the
sensitivity of detection. This revealed that PA was a minor component compared
to other phospholipids, including phosphatidylcholine and
phosphatidylethanolamine. Two-dimensional TLC analysis did not reveal any
significant changes in the major phospholipids present in the mutant MEFs
compared to wild-type cells with intracellular LPA and S1P being undetectable
in both (data not shown). There was a small but significant increase of PA in
the LPP3-/- cells compared to wild type
(Table 1). In addition, there
was an increase in an unidentified phospholipid that migrated slightly faster
than PA (Table 1). These data
indicate that the lack of LPP3 activity results in, as expected, a decreased
level of DAG and concomitant increase in PA. Since DAG induces the
phosphorylation of several isoforms of PKC, resulting in their activation, we
determined the levels of phosphorylated PKC. Homozygous null EBs showed
approximately 50% lower phospho-pan PKC than their heterozygous counterparts
(Fig. 2F).
|
Together these results reveal that the targeted mutation deleting LPP3, resulted in an increase of both extracellular LPA and intracellular PA, as well as a significant reduction in the intracellular levels of DAG, concomitant with a reduction in the levels of activated PKC.
Absence of LPP3 results in embryonic lethality and is associated with
two post-implantation phenotypes
Heterozygous LPP3StwO3 mice were viable,
fertile and indistinguishable from their wild-type littermates. Matings of
heterozygous LPP3StwO3 on either a mixed
129/SvJxC57BL/6J or a pure 129/SvJ background produced animals only of
wild-type and heterozygous genotype (Table
2), indicating that the mutation is an embryonic lethal. Embryos
from LPP3StwO3 heterozygous intercrosses were
analyzed at 7-10.5 days of gestation and genotyped by PCR
(Fig. 2C). In these, a normal
Mendelian distribution of genotypes was observed
(Table 2). No live homozygous
embryo was recovered beyond E10.5 indicating embryonic lethality prior to or
around this time. Homozygous embryos collected between E7-10.5 were, in
general, developmentally delayed when compared to wild-type or heterozygous
littermates. Moreover, 30% of the homozygous mutant embryos were highly
abnormal revealing a gastrulation defect (see below).
|
|
|
Disruption of allantois morphogenesis leads to defective placenta
formation
To gain further insight into the placental phenotype, we analyzed the
developmental potential of LPP3 null cells in chimeric combination
with wild-type embryos. Homozygous LPP3StwO3 ES
cells were injected into wild-type ROSA26 blastocysts constitutively
expressing the lacZ reporter gene. Under these conditions,
extra-embryonic tissues, including the chorion, will be of wild-type genotype
with the mutant ES cells contributing to both the embryonic and
extra-embryonic mesoderm. In the presence of wild-type extra-embryonic tissue,
allantoises from embryos entirely derived from the null ES cells
(n=3) were able to contact the wild-type chorion
(Fig. 5A). Histological
analysis of the chorio-allantoic region revealed that despite contact between
both structures, the allantois remained as a compact mass of tissue that did
not extensively invade the chorionic plate
(Fig. 5C). The vascular
adhesion molecule 1 (VCAM1) and its receptor, 
4 integrin, are both
required for chorio-allantoic fusion. Embryos lacking VCAM1 in the allantois
or 4 integrin in the chorion fail to establish a proper
chorio-allantoic connection (Gurtner et
al., 1995; Yang et al.,
1995). A comparison between heterozygous and homozygous mutant
embryos, at equivalent stages of development (5-6 somite pairs), revealed no
significant differences in the expression or tissue distribution of VCAM1 and
4 integrin between embryos of both genotypes (data not shown). This
indicated that additional and as yet unidentified cellular components are
involved in chorio-allantoic fusion, and are affected by LPP3 deficiency. In
the chimeras, with contributions from both wild-type and mutant cells to the
embryo proper, chorio-allantoic placental and vascular development was much
improved, however the majority of allantoic endothelial cells were of
wild-type origin (Fig. 5B,D).
These observations reveal that LPP3 is required, both in the chorion
to promote allantois extension and fusion with the chorion, and also in the
allantois to enable it to undergo proper vasculogenesis.
|
) (data
not shown).
LPP3-deficient embryos show defective gastrulation
In the mixed 129/S1xC57BL/6J background, approximately 30% of
LPP3 homozygous mutant embryos recovered between E7.5-E9.5 were
highly abnormal (Table 2). The
embryos exhibited a wide variety of abnormalities including a short
anterior-posterior axis, anterior truncation, embryonic development outside
the yolk sac membranes and frequent duplication of axial structures
(Fig. 6A-B). Histological
analysis of those with axial abnormalities revealed a duplicated notochord,
with an extra row of somites between the notochords with partial to complete
duplication of the neural tube (Fig.
6C,D). The expression of sonic hedgehog (Shh), a
marker for axial mesoderm (notochord) and Twist, a marker of paraxial
mesoderm (somites), confirmed the histological findings
(Fig. 6E and F, respectively).
By using a probe to brachyury (a marker for posterior and axial
mesoderm) LPP3-/- embryos often showed a constriction
between the embryonic and extra-embryonic tissues (arrowhead), with shortening
of the primitive streak (Fig.
6G,H). In addition, axis duplication was detected in embryos as
early as E6.5 (Fig. 6I,J). In
some of the embryos, the extended area of brachyury expression
revealed a broadening of the primitive streak, node and axial mesoderm
domains. Frequently, mesoderm-like projections were observed growing out these
areas (data not shown).
|
;
Staal et al., 2002).
The effect of LPP3 on Wnt signaling was analyzed by measuring
ß-catenin-mediated TCF transcription in the ES cells null for
LPP3. In contrast to wild-type ES cells, the LPP3 null cells
exhibited a 10- to 15-fold increase in luciferase activity following
transfection with the TCF-luciferase reporter plasmid (TOPFlash), a
TCF/ß-catenin responsive reporter gene
(van de Wetering et al.,
1997). The increased TOPflash activity was inhibited by about 50%
following transfection of an LPP3 expression cassette into the null
ES cells (Fig. 7A).
|
To determine whether the inhibitory effect of LPP3 on
ß-catenin-mediated TCF transcription was dependent on the lipid
phosphatase activity of the LPP3s, we used two different LPP3 constructs in
which the phosphatase catalytic site had either been ablated by deleting 32
amino acids or mutated by changing serine 197
threonine (A.J.M.,
unpublished data). Both forms clearly lacked lipid phosphatase activity. The
results presented in Fig. 7D
show that LPP3, null for phosphatase activity, was as effective as the
wild-type LPP3 at inhibiting ß-catenin-mediated TCF transcription.
LPP3 deficiency affects anterior development and
Wnt target gene expression
Increased activation of the Wnt/ß-catenin signaling pathway induces
axis duplication in the mouse (Popperl et
al., 1997; Zeng et al.,
1997). We therefore investigated whether LPP3 loss of
function maybe over-stimulating Wnt signaling in embryos. We analyzed the
kinetics of expression of 3 Wnt regulated genes in differentiating
heterozygous and homozygous LPP3-deficient EBs. Expression of bone
morphogenetic protein (Bmp4; Fig.
8) and nodal (data not shown) was not affected, however
brachyury expression was both elevated and prolonged in homozygous
EBs compared with the heterozygotes (Fig.
8) consistent with the increased expression observed in some
LPP3-/- embryos.
|
). AVE formation was affected in about 30% of the
LPP3 null embryos, which was revealed by abnormal expression of two
markers of the AVE, specifically Hex and dkk-1. In normal
embryos dkk-1, an extracellular antagonist of Wnt signaling, is
expressed in a horseshoe-like pattern extending from the proximal to the
lateral-distal AVE in wild-type and heterozygous E7.5 embryos
(Glinka et al., 1998;
Perea-Gomez et al., 2001;
Zakin et al., 2000). Two out
of 6 LPP3 null embryos at equivalent stages showed reduction and
abnormal expression of dkk-1 transcripts
(Fig. 6K). Although reduced
levels of dkk-1 expression were observed, the region of expression
was more distal, with loss of expression occurring in the most proximal AVE.
Furthermore an accumulation of Hex-positive cells in the distal tip
was also observed in a subset of E7.0 null embryos. In the same null embryos
Wnt3 expression was detected in both the proximal anterior and
posterior embryonic ectoderm. In normal E7.5 embryos Wnt3 expression
becomes restricted to the posterior embryo and primitive streak
(Fig. 6L). Together, these
results suggest that in the absence of LPP3, components regulating
Wnt/ß-catenin responses are both quantitatively and spatially altered,
consistent with the abnormalities in axis patterning.
Murine LPP3 has a ventralizing effect on Xenopus
embryos
The loss of function of LPP3 resulting in the duplication of axial
structures (dorsalization) in the mouse embryo suggested that increased
LPP3 expression was having a ventralizing activity in vivo. To test
this, murine LPP3 mRNA was injected into the 2 dorsal blastomeres of
4-cell stage Xenopus embryos and subsequent larval development was
analyzed. Synthesis and glycosylation of LPP3 protein was confirmed by western
blot analysis of extracts from injected larvae
(Fig. 9A,B). As controls,
embryos were either injected with mLPP3 into their ventral blastomeres
(Fig. 9C) or their dorsal
blastomeres were injected with a mutated version of mLPP3 (lacking the same
amino acids used to generate the phosphatase-deficient mutation). 80-100% of
stage 22-24 embryos injected with 1-2.5 ng of LPP3 mRNA in their
dorsal or ventral blastomeres showed a phenotype consisting of the transient
formation of a 'blister' in the ventral region of the developing larvae. In
contrast, larvae injected with the mutated mLPP3 were unaffected.
These data suggested that mLPP3 injection into frog embryos causes an
ionic imbalance resulting in fluid accumulation under the skin that produces
the blistered phenotype. However, as this phenotype appeared independent of
the site of injection, it was not considered relevant for the analysis of LPP3
participation in axial patterning.
|
),
indicating a mild, but consistent, ventralizing activity of LPP3. The
ventralizing activity of LPP3 was also analyzed by its ability to inhibit
secondary axis formation induced by Xwnt3a and Xwnt8.
Full-length mLPP3 RNA co-injected with Xwnt3a or
Xwnt8 inhibited secondary axis formation, although the effect was not
complete, with the majority of the embryos retaining a weak dorsalized
phenotype (Fig. 9G-J and
Table 4). Furthermore, LPP3
lacking phosphatase activity was also able to reduce the dorsalizing activity
of Xwnt8, although not as effectively as full-length LPP3
(Table 4). These results
strongly support a role for LPP3 in axis patterning through the regulation of
Wnt signaling pathways.
|
|
|
Discussion |
|---|
|
TOPSUMMARYIntroductionMaterials and methodsResultsDiscussionREFERENCES |
|---|
)
we show that embryos lacking LPP3, exhibit profound defects in the
development of the extra-embryonic vasculature and in embryonic axis
formation. The effect of LPP3 deficiency on vascularization and
placental formation was fully penetrant on both a pure 129/S1 and mixed
129/S1xC57BL/6J backgrounds, whereas axis duplication was more prevalent
on the mixed background implying the existence of genetic modifier(s)
affecting the phenotypic outcome.
Development of the vascular system commences in the yolk sac and
subsequently in the embryo proper by the formation of angioblasts in the
cephalic and paraxial mesoderm. Angioblasts migrate and coalesce to form the
primitive vascular plexus. Subsequent development of the system is by
angiogenesis in which new vessels are formed by budding or sprouting from the
vascular primordia. Thereafter maturation of the vascular system requires that
vascular smooth muscle cells and pericytes are recruited to, and migrate
around, the endothelial blood vessels to form the arteries, veins and
capillaries (Carmeliet,
2000)
Many different peptide ligands and their receptors, including VEGF and its
receptors, Flk1 and Flt1 (Ferrara et al.,
1996; Fong et al.,
1995; Shalaby et al.,
1997; Shalaby et al.,
1995), angiopoietin 1 and Tie 2
(Puri et al., 1995;
Suri et al., 1996)
PDGF-ßB, the PDGF receptor ß
(Hellstrom et al., 1999),
together with TGFß1 (Dickson et al.,
1995; Yang et al.,
1999) are essential to the establishment, elaboration and
maintenance of the cardiovascular system. Here we provide the first evidence
that LPP3, an enzyme previously known to regulate bioactive phospholipids
and/or their products is essential to some of the earliest stages in vascular
development in the murine embryo. Embryos that were null for LPP3
consistently failed to establish an intact yolk sac and allantoic vasculature,
with the allantois being unable to form an effective union with the chorion
and the yolk sac endothelial cells failing to organize a vascular plexus. The
allantois, which is derived by outgrowth of the extra-embryonic mesoderm from
the proximal epiblast, did not extend towards and invade the chorion, but
remained as a compact mass of tissue at the posterior of the embryo. An
analysis of chimeras, in which the chorion was wild type and the entire embryo
null for LPP3 resulted in improved development of the mutant
allantois with limited invasion of the chorion, although vascularization and
the extent of invasion was still reduced compared to wild-type embryos. This
revealed that chorionic expression of LPP3 is required for proper
growth and extension of the allantois. It is possible, by analogy with the
role of wunen in regulating germ cell migration, that chorionic LPP3
may be regulating the production of some factor that guides the growth and
extension of the allantois towards the chorion. However, endothelial cells of
the yolk sac and allantois both express LPP3, and the inability of the null
endothelial cells to organize and form vascular cords would appear to be cell
autonomous and this may also affect outgrowth of the allantois. In chimeras
made between wild-type embryos and LPP3 null ES cells, where there
was significant contribution of both genotypes to all embryonic tissues, it
was noticeable that only a few LPP3 null cells contributed to the
umbilical vasculature, despite the allantois invading the chorion. The
molecular basis for morphogenetic failure of the allantois remains obscure as
neither the chorion nor allantois showed alterations in the levels or patterns
of expression of the adhesive proteins VCAM1 and 4 integrin, both of
which are essential to chorioallantoic fusion
(Gurtner et al., 1995;
Yang et al., 1995). Failure of
allantois outgrowth was also associated with an inability of the allantoic
endothelial cells to organize and form the umbilical cord, suggesting that
vasculogenesis in the allantois may be essential for its extension towards the
chorion. This was strikingly demonstrated in the allantoic explants in which
the wild-type endothelial cells organized into a network of cords, whereas
those from null embryos, or wild-type embryos treated with the LPP inhibitor,
propranolol, remained as a compact mass of PECAM1-positive cells.
These results are consistent with previous observations that lipid
signaling pathways are essential to vasculogenesis and development of the
cardiovascular system in later stage embryos. Treatment of cultured vascular
endothelial cells with S1P induces adherens junction assembly and vascular
morphogenesis (Lee et al.,
1999) and these changes are mediated by binding of S1P to the
S1P1 and S1P3 receptors. Likewise loss-of-function
mutations in some of the S1P receptors are associated with failure in
cardiovascular development in midgestation embryos. Also in the zebrafish, the
mutation miles apart (mil), which affects a gene homologous
to S1P5 (a receptor for S1P), results in
defective migration of cardiac precursor cells to the midline and failure of
heart organogenesis (Kupperman et al.,
2000). A loss-of-function mutation in the murine S1P1
receptor results in hemorrhage and embryonic lethality at E12-14 as a
consequence of incomplete vascular maturation, due to the failure of pericytes
to respond to PDGF-induced cell migration and to surrounding the blood vessels
(Hobson et al., 2001). Such
mutations have implicated lipid signaling pathways in the maturation of the
cardiovascular system, particularly with regard to phospholipids regulating
cell migration and adhesive interactions, similar to the effects of
phospholipids on germ-cell migration in the Drosophila mutant
wunen (Starz-Gaiano et al.,
2001; Zhang et al.,
1997). In contrast our results with LPP3 null mice
suggest that lipid signaling pathways maybe required at an even earlier stage,
specifically during morphogenesis of the umbilical and yolk sac
vasculatures.
The role of LPP3 in body axis patterning
Our results also revealed that LPP3 influences axial patterning.
Axis duplication in the mouse can be induced by node transplantation or by
systemic administration of drugs affecting cytoskeletal organization during
early gastrulation (Beddington,
1994; Kaufman and O'Shea,
1978). Similarly, activation of the Wnt signaling pathway induces
axis duplication and embryonic dorsalization, e.g. by ectopic expression of
Cwnt8C (Popperl et al.,
1997). Interaction of the Wnt1 class of ligands (Wnt1, 3a, 8 and
8B) with the appropriate frizzled receptor inactivates the GSK-3-axin-APC
complex with the subsequent stabilization of cytoplasmic ß-catenin
leading to the activation/repression of target genes. This is known as the
canonical Wnt/ß-catenin signaling pathway and when over stimulated,
results in secondary axis formation in Xenopus embryos and
transformation of mammary epithelial cells in the mouse
(Miller et al., 1999).
The severe phenotype observed in 30% of LPP3-/- embryos
is remarkably similar to mutations at the fused locus
(Gluecksohn-Schoenheimer,
1949; Zeng et al.,
1997). As with the LPP3StwO3
mutation, fused mutations display variable expressivity and
incomplete penetrance. Homozygotes for 4 alleles of the locus Fu
(Caspari and David, 1940;
Gluecksohn-Schoenheimer, 1949;
Jacobs-Cohen et al., 1984;
Perry et al., 1995) die around
8-10.5 days of gestation, showing a wide spectrum of abnormalities between
embryos, including developmental delay, duplication of embryonic structures
and development of parts of the embryo outside the amnion and yolk sac. The
product of the fused locus, axin
(Zeng et al., 1997), inhibits
the signal transduction cascade activated by Wnts, by forming a complex with
the ß-cateninphosphorylating form of GSK-3. In the absence of axin, GSK-3
is released from the ß-catenin phosphorylating complex, resulting in the
stabilization of ß-catenin with activation of the canonical Wnt signaling
response(s). The striking similarity between these phenotypes and the
LPP3 null phenotype strongly suggests that loss of LPP3 may
upregulate a canonical Wnt signaling response, with LPP3 functioning as a Wnt
signaling antagonist in vivo.
Supporting this notion are four observations. First, LPP3
expression inhibits TCF transcriptional activity mediated by both endogenous
and exogenous ß-catenin in HEK293 cells, probably by regulating the
availability of the dephosphorylated and stable version of ß-catenin.
Secondly, consistent with the inhibitory action of LPP3 on
ß-catenin, loss of LPP3 resulted in a significant and marked
increase in endogenous ß-catenin activity in ES cells. Thirdly, loss of
LPP3 results in an increased and sustained expression of the Wnt target gene
brachyury in EBs (Yamaguchi et
al., 1999), consistent with the expanded areas of expression in
the primitive streak, node and axial mesoderm in the mutant embryos. The same
alteration in expression of brachyury occurs in mouse embryos either
misexpressing Cwnt8C (Popperl et
al., 1997) or following ectopic transplantation of the node,
resulting in axis duplication with an extra row of somites between the two
axes (Beddington, 1994).
Lastly, the expression of the extracellular antagonist to Wnt signaling,
dkk1 (Glinka et al.,
1998) overlaps with LPP3 in the anterior visceral endoderm (AVE).
In some of the LPP3 null embryos dkk1 expression was reduced
and altered in the AVE. Such alterations may also contribute to deregulated
Wnt3 expression in the anterior embryonic ectoderm and anterior gene
expression in the AVE leading to axis duplication
(Perea-Gomez et al.,
2002).
Additional evidence supporting the role of LPP3 in axis patterning
was derived from the expression of LPP3 in Xenopus embryos.
Ventralization or the reduction of axial structures is induced when some
antagonists to the Wnt signaling pathway are injected into the dorsal
blastomeres of Xenopus embryos
(Cadigan and Nusse, 1997;
Tago et al., 2000). Ectopic
expression of murine LPP3 in the dorsal blastomeres of
Xenopus embryos caused a mild but clear ventralizing effect. In
addition axis duplication, induced by injection of Xwnt8 or
3a mRNA, was mildly but consistently inhibited by co-injection of
LPP3 mRNA, directly demonstrating that LPP3 affects axis
patterning. Together the evidence strongly suggests that axis duplication in
LPP3-deficient embryos arises as a result of increased activation of
the canonical Wnt pathway, and an increase in ß-catenin-mediated TCF
transcription.
How LPP3 regulates ß-catenin-mediated TCF transcription remains to be established. The surprising result from testing the LPP3 forms lacking phosphatase activity, was that they were equally effective as wild-type LPP3 at inhibiting TCF/ß-catenin transcription in HEK293 cells. This revealed that LPP3 contains an additional, as yet undefined functional activity, which inhibits TCF/ß-catenin activity.
Conclusions
Our results revealed that LPP3 is a multifunctional protein essential for
different aspects of embryo development. In addition to its known lipid
phosphatase activity, LPP3 may also regulate ß-catenin activation by
some, as yet undefined mechanism. However, given the multifunctional roles of
LPP3 we propose that the phosphatase activity of LPP3 in regulating bioactive
phospholipids levels may be critical to the vascular phenotype. The
phosphatase activity may also indirectly regulate Wnt signaling pathways.
Endothelial cell migration and/or cell adhesion are affected by changes in
phospholipid levels particularly LPA. LPA promotes or inhibits cell migration,
adhesion and cytoskeletal reorganization depending on the cell type and
concentration of the lipid (Panetti et
al., 2001). Our results showed that, with loss of LPP3,
extracellular levels of LPA were increased by the LPP3 null cells and also
intracellular levels of DAG were reduced, resulting in reduced protein kinase
C (PKC) activation. Activated PKC is required for the morphogenesis of the
vasculature (Tang et al.,
1997; Xia et al.,
1996). Furthermore, propanolol, an inhibitor of LPP3 phosphatase
activity, blocked capillary morphogenesis in the allantois explants.
However, we cannot exclude the possibility that the vasculogenesis
phenotype is also influenced by LPP3 affecting Wnt signaling, as several Wnt
signaling mutants are known to affect vascular, allantois and placental
development (Galceran et al.,
1999; Ishikawa et al.,
2001; Parr et al.,
2001). A third possibility emerged in that a putative RDG-mediated
adhesion function has been recently described for human LPP3 which may affect
endothelial cell adhesion (Humtsoe et al.,
2003), although such an RDG sequence has not been found in the
mouse LPP3.
Certain aspects of the gastrulation phenotype may als
