extraembryonic proteases regulate nodal signalling during gastrulation

brief communications

NATURE CELL BIOLOGY VOL 4 DECEMBER 2002 www.nature.com/naturecellbiology 981

Extraembryonic proteases regulateNodal signalling during gastrulation

Séverine Beck, J. Ann Le Good, Marcela Guzman, Nadav Ben Haim, Karine Roy, Friedrich Beermann andDaniel B. Constam*

Developmental Biology Group, Swiss Institute for Experimental Cancer Research (ISREC), Chemin des Boveresses 155, CH-1066 Epalinges, Switzerland*e-mail: [email protected]

Published online: 25 November 2002; DOI: 10.1038/ncb890

During gastrulation, a cascade of inductive tissue inter-actions converts pre-existing polarity in the mammalianembryo into antero-posterior pattern. This process is trig-gered by Nodal, a protein related to transforming growthfactor-β (TFG-β) that is expressed in the epiblast and vis-ceral endoderm, and its co-receptor Cripto, which isinduced downstream of Nodal. Here we show that theproprotein convertases Spc1 and Spc4 (also known asFurin and Pace4, respectively) are expressed in adjacentextraembryonic ectoderm. They stimulate Nodal matura-tion after its secretion and are required in vivo for Nodalsignalling. Embryo explants deprived of extraembryonicectoderm phenocopy Spc1−/−; Spc4−/− double mutants inthat endogenous Nodal fails to induce Cripto. But recom-binant mature Nodal, unlike uncleaved precursor, canefficiently rescue Cripto expression. Cripto is alsoexpressed in explants treated with bone morphogeneticprotein 4 (BMP4). This indicates that Nodal may induceCripto through both a signalling pathway in the embryoand induction of Bmp4 in the extraembryonic ectoderm.A lack of Spc1 and Spc4 affects both pathways becausethese proteases also stimulate induction of Bmp4.

In the mammalian embryo, the anterior–posterior (A–P) patternbecomes evident with the formation of the primitive streak ascells in the so-called ‘epiblast’ ingress on the prospective posterior

side to become mesoderm and definitive endoderm1,2. Posterior cellfates marked by the expression of Brachyury and Fgf8 are specifiedby the secreted proteins Nodal and Wnt3 (refs 3, 4). Wnt3 isinduced downstream of Nodal and in turn maintains Nodal expres-sion during gastrulation3,4. Nodal and Wnt3 also specify mesendo-dermal cell fates marked by the expression of Gsc, Cer-l and Foxa2(refs 3, 4). Lineage tracing experiments in early gastrulae (E6.5)suggest that these populations emanate from the anterior primitivestreak5,6. Earlier fate maps in chick and functional analysis of avianWnt proteins indicate, however, that cells in the anterior primitivestreak originate at the posterior margin of the blastodisc inresponse to Wnt signals7,8. The corresponding territory in the pre-streak mouse embryo is localized in the proximal epiblast regionmarked by the expression of Wnt3.

In Cripto mutants and in embryos lacking the transcription fac-tor Otx2 (ref. 9), this region also expresses the mesendoderm mark-er Gsc, presumably because mesendoderm progenitors fail toingress into the primitive streak10. At the onset of gastrulation,Cripto is expressed transiently in a proximal–distal gradient andpotentiates Nodal signalling10–12. It binds to Nodal and its signallingreceptors, and thus is thought to act as a Nodal co-receptor13. Theseobservations suggest that a graded Nodal signal may be instrumen-tal in patterning the embryo. Besides regulating posterior cell fates,Nodal also specifies anterior visceral endoderm (AVE) and triggersits proximal movement along one side of the epiblast3. Eventually,

these extraembryonic cells secrete mDkk1, Cer-l and Lefty-1 pro-teins, which antagonize Wnt and Nodal signalling to specify anteri-or cell fates14.

To determine functions of Spc1 and Spc4 that may be masked insingle mutants, we derived Spc1−/−;Spc4−/− embryos. By embryonicday E6.5, these double mutants had cavitated. The epiblast andextraembryonic ectoderm (ExE) were shortened, although theoverall size of the conceptus was normal, owing to an extensionoverlying the distal tip that was presumably composed of extraem-bryonic endoderm. At E7.5, the double mutants were reduced insize by 30–50% and failed to form an amnion, allantois or chorion(Fig. 1a, b). The proximal epiblast region formed a disorganizedprimitive streak that failed to elongate (Fig. 1b, right). Thus, cellsexpressing the mesoderm marker Lhx1 were scarce, and nascentmesoderm positive for FGF8 accumulated ectopically between theproximal epiblast and adjacent ExE, or in the amniotic cavity, orboth (Fig. 1c–f).

Expression of Foxa2 marked prospective axial mesoderm anddefinitive endoderm precursors in the anterior primitive streak(Fig. 1g), but was barely detectable and confined to the proximalepiblast in double mutants (Fig. 1h). Absence of the anterior streakwas also highlighted by the lack of expression of Cer-l, an inde-pendent marker of early definitive endoderm (data not shown).Similarly, expression of Gsc in the embryo proper was lost. We con-clude that Spc1 and Spc4 are necessary to allocate mesendodermfates to the anterior primitive streak.

Before definitive endoderm formation, Cer-l transcripts markthe AVE15. Notably, Cer-l expression was already downregulated atthis early stage in double mutants (Fig. 1i, j). Likewise, expressionof Lefty-1 and the homeobox transcription factors Hesx-1 and Hexin the AVE was abolished (Fig. 1k–p), suggesting that this tissueprematurely looses its molecular identity or fails to form altogeth-er. Gsc and Lhx1, which were upregulated in the AVE of wild-typeembryos, remained expressed diffusely throughout the visceralendoderm, thus confirming a lack of AVE pattern (Fig. 1q–t). Innormal embryos, anterior movement of the AVE is driven by thetranscription factor Otx2 (refs 16, 17). In the visceral endoderm ofdouble mutants, Otx2 messenger RNA was lost (Fig. 1u, v).Expression of Otx2 in the epiblast was attenuated and failed to seg-regate from the Nodal expression domain (Fig. 1w–z), as would beexpected if AVE function is abolished. These observations suggestthat Spc1 and Spc4 control AVE formation and the conversion ofproximal–distal asymmetry into an orthogonal A–P axis.

Because AVE and axis formation are orchestrated by Nodal, weexamined whether Spc1 and Spc4 regulate Nodal signalling.Whole-mount in situ hybridization confirmed that the doublemutants expressed Nodal (Fig. 1z), although initial mRNA levelswere markedly reduced (Fig. 2a–d). This is reminiscent of homozy-gous Nodal mutants, which fail to activate a NodallacZ reporter alleleat this early stage3 but later express lacZ if they develop beyond acrucial stage that relies on Nodal auto-induction (Fig. 2r, n = 4/8).

© 2002 Nature Publishing Group


NATURE CELL BIOLOGY VOL 4 DECEMBER 2002 www.nature.com/naturecellbiology982

In addition, Brachyury, Lefty-2 and Eomesodermin expression isalso lost, both in double mutants (Fig. 2e–j) and in Nodal-deficientembryos3,12. Likewise, expression of Wnt3, Bmp4 and the primitiveectoderm marker Pou5f1 was markedly reduced (Fig. 2k–n, s, t).Immunoblot analysis confirmed that quantities of Bmp4 proteinwere also reduced (Fig. 2u). In addition, double mutants failed toupregulate Cripto (Fig. 2o, p). Because Nodal depends on Cripto asa co-factor to position the A–P axis10, this shows that Spc1 and Spc4function in the Nodal pathway.

Expression of Cripto is itself dependent on Nodal signalling3. Itis possible that Cripto failed to be induced, owing to impairedNodal maturation. To test this directly, we analysed pooled lysatesof up to 26 embryos by immunoblotting using an antiserum thatreacts with unprocessed and mature Nodal18. But as this assay wasnot sensitive enough to detect any form of Nodal in vivo, we testedthe effect of Spc1 and Spc4 on the maturation of Flag-tagged Nodalpro-domain in transfected cells. This amino-terminal fragment ofNodal accumulates in conditioned medium in a time-dependentmanner but is not detected in whole-cell lysates18. We thereforeexamined whether Nodal cleavage involves soluble forms of Spc1 orSpc4 that act after Nodal secretion, and we compared their activi-ties to those of the related proteases Spc7 and two splice variants ofSpc6 that are expressed during later stages19. As expected, soluble

forms of Spc1 and Spc4 were detected in conditioned medium oftransfected cells by immunoblotting. We also observed Spc activityusing the fluorogenic substrate Boc-Arg-Val-Arg-Arg-AMC, andthis activity was blocked by the Spc inhibitor decanoyl-Arg-Val-Arg-Arg-chloromethylketone (decRVRRcmk). Likewise, whenincubated with conditioned medium from cells separately trans-fected with Nodal, soluble Spc activities promoted Nodal precursorcleavage but left a mutated cleavage site intact (Fig. 3a, b).Analogous results were obtained using the related Nodal antagonistLefty-1 as a substrate (Fig. 3c, d), suggesting that extracellularcleavage may account for the maturation of several members of theTGF-β family.

To assess whether Spc1 and Spc4 are required to activate the Nodalprecursor, we derived mutant embryonic stem (ES) cells. Whereascompound heterozygous and single mutant ES cells converted signifi-cant amounts of Nodal, no cleavage was observed after incubationwith double mutant cells (Fig. 3e). Thus, endogenous Spc1 and Spc4are necessary and sufficient to promote Nodal maturation. In addition,we confirmed that proteolytic processing potentiates Nodal signallingin a luciferase reporter assay. Note, however, that mutant Nodal devoidof an Spc cleavage site retained significant activity (Fig. 3f). Thus,uncleaved Nodal may also weakly signal in Spc1; Spc4 double mutantsafter overcoming the initial delay in its onset of expression (Fig. 2a–d).

i

Cer-l

g

Foxa2

h'

g'

nk m p

Lefty-1 Hesx-1 Hex

t t'

u

sq r

s'q' rLhx1 Gsc

Otx2

Otx2 Nodal

v x z

u' v'

Lhx1

*acep

m f'e'

fe

Fgf8c' d'

*ve

mexe

c d

ec

exe

epep

ch

ec

b 2'

m

ch

al

am

ep

ve

ec

ec

ep

ve*

ec

exe

ep

ep

ch

ec

pe

a1

a2 a2'

b 2b 1

E6.5 E7.5

E7.5E6.5

h j

l o

w y

Figure 1 Spc1 and Spc4 regulate mesendoderm and AVE formation.a, b, Unlike normal litter mates (a1), by E6.5 Spc1−/−; Spc4−/− double mutants tran-siently form an abnormally thickened protrusion (arrow) at the distal tip (b1), and byE7.5 are reduced in size (b2). Sagittal sections show that wild-type mesoderm cells(m) migrate between epiblast and extraembryonic visceral endoderm (ve) (a2′′),whereas double mutants form a rudimentary primitive streak (asterisk) that bulgesinto the amniotic cavity and fails to elongate (b2′′). In the extraembryonic ectoderm(exe), an ectoplacental cone (ec) forms, but the chorion (ch) is absent. c–f, Somemesoderm expressing Lhx1 (c, d, top) and Fgf8 (e, f, top) forms in double mutants,but it fails to move distally (d′′, f′′) and bulges into the amniotic cavity (d, f, bottom,asterisks). g, h, At E7.5, Foxa2 transcripts mark mesendoderm populations in theanterior primitive streak (g). In double mutants, Foxa2 is detected only in the proxi-mal epiblast region at very low levels (h). Transverse sections (g′′, h′′) were obtained

at the levels indicated by arrowheads. i–p, Expression of Cer-l (i, j), Lefty-1 (k, l),Hesx-1 (m, n) and Hex (o, p) observed in the AVE at E6.5 (i, k, m, o) is inhibited indouble mutants (j, l, n, p). q–t, Lhx1 (q, r) and Gsc (s, t) expression in the AVE isupregulated in control embryos (q, s), but not in double mutants (r, t). Sagittal sec-tioning showed that, similar to Lhx1, Gsc remains expressed broadly at low levelsthroughout the visceral endoderm (t, right). u–z, Expression of Otx2 at the onset ofgastrulation is widespread in normal epiblast and visceral endoderm (u) but later isinhibited posteriorly (w). Otx2 expression is absent from the visceral endoderm(arrowhead) and reduced in the epiblast of E6.5 double mutants (v), where itremains widespread until E7.5 (x). y–z, Under the influence of a functional AVE,Nodal mRNA normally is confined to the posterior region (y), whereas it remainsbroadly distributed in double mutants (z). In control embryos, anterior is on the left,and posterior on the right side.




Next, we asked whether Nodal is also cleaved after secretion inthe embryo. In keeping with this idea, Spc4 mRNA was detectedspecifically in the ExE (Fig. 4a). Similarly, Spc1 expression waslocalized to the ExE, from where it seemed to spread at reduced lev-els by three or four cell diameters into the adjacent epiblast andproximal visceral endoderm (Fig. 4b, c). Analysis of chimaericembryos confirmed that Spc1 and Spc4 expression is essential inextraembryonic lineages because wild-type ES cells carrying theRosa26lacZ allele failed to rescue gastrulation when injected intodouble mutant host blastocysts (data not shown). This suggeststhat Spc1 and Spc4 pattern the embryo in a non-cell-autonomousmanner.

To test further whether Spc1 and Spc4 are required to stimulateNodal cleavage at the onset of gastrulation, we analysed Nodal sig-nalling in embryo explants cultured in the presence or absence ofExE. We found that endogenous Nodal failed to induce Cripto afterremoval of the ExE (Fig. 4d–g). This is likely to be due to a lack ofNodal maturation, because mature recombinant Nodal (unlike theprecursor form) efficiently rescued Cripto expression in suchexplants (Fig. 4h–k). We conclude that Nodal cleavage is necessaryand sufficient to rescue expression of Cripto.

Notably, Cripto was also induced by recombinant human BMP4(Fig. 4l), raising the possibility that endogenous BMP4 may con-tribute to the regulation of Cripto. But a proportion of Bmp4−/−

embryos develops beyond gastrulation without overt anterior pat-terning defects20, which suggests that Cripto is expressed sufficient-ly to position the AVE. In addition, BMP4 cleavage was not inhib-ited in Spc1−/−;Spc4−/− double mutants. Instead, Bmp4 mRNA andprotein expression in the ExE was attenuated or lost (Fig. 3s–u),showing that a signal upstream of Bmp4 is missing. Thus, the pri-mary defect in double mutants is not simply a lack of Bmp4.

The only known positive regulator of Bmp4 expression in theExE is Nodal3,4. By culturing ExE explants, we confirmed that a sig-nal from the epiblast is required beyond day E5.5 to maintainexpression of Bmp4 (Fig. 4m, n = 8/8). In addition, Bmp4 wasinduced in ExE explants treated with cleavage-deficient Nodal (Fig.4n, n = 9/11). Thus, Spc1 and Spc4 may enhance Bmp4 expressionby stimulating auto-induction of Nodal or by potentiating Nodalactivity, or both. Overall, these results suggest that Nodal inducesCripto both directly through a signalling pathway in the epiblast andindirectly through induction of Bmp4 in the ExE. This mightexplain why peak levels of Cripto mRNA are expressed in proximity

to the ExE10,11. Alternatively, a gradient of Cripto expression may beestablished also in the absence of Bmp4, simply owing to localizedNodal cleavage near the source of Spc activities.

Similar to Bmp4, Fgf8 normally is induced downstream ofNodal3. In mock-treated epiblast explants, Fgf8 was only very weak-ly induced during a culture period of 20 h (Fig. 4p, n = 3/9). Bycontrast, it was robustly expressed by that time in whole embryocultures (Fig. 4o, n = 9/9) and in epiblasts treated with recombi-nant Nodal precursor (Fig. 4q, n = 4/8). Thus, the presence ofuncleaved precursor may account for weak induction of someNodal target genes in double mutants.

By analysing double mutants and chimaeric embryos composedof mutant and wild-type cells, we have shown here that extraem-bryonic Spc1 and Spc4 function redundantly to pattern the epiblastand visceral endoderm. Similar to Nodal mutants, embryos lackingthese proteases fail to form an AVE and to induce the Nodal co-receptor encoded by Cripto. We therefore analysed the expression ofa comprehensive set of additional molecular markers induced byNodal in the epiblast (Nodal, Pou5f1, Wnt3), prospective mesoderm(Fgf8, Lhx1, Brachyury, Lefty-2, Eomes) or mesendoderm (Foxa2,Gsc, Cer-l). With the exception of Fgf8, expression of these markersis absent or reduced to levels comparable to those observed inNodal mutants. Adding to the similarity with Nodal mutants,upregulation of Nodal is itself delayed. A probable explanation isthat Nodal is virtually inactive owing to inhibition of its own pro-teolytic maturation. In keeping with this view, the activity of Nodalin luciferase reporter assays is reduced by 85–90% when the Spccleavage motif Arg-Gln-Arg-Arg (RQRR) is replaced by an unrelat-ed sequence.

At the onset of gastrulation, however, Nodal is expressed in the epi-blast and visceral endoderm, whereas Spc1 and Spc4 transcripts areconfined to the ExE. Thus, a direct interaction of Nodal precursorprotein with Spc1 and Spc4 would have to occur after secretion. Byusing a cell-free assay, we confirmed that soluble forms of Spc1 andSpc4 produced in transfected COS-1 cells can cleave recombinantNodal. In addition, we found that Nodal is cleaved on incubation withES cells, and that the extent of processing correlates with the numberof functional Spc1 and Spc4 alleles present. These findings provide thefirst direct evidence that Nodal is susceptible to cleavage after secre-tion, and that its maturation is dependent on Spc1 and Spc4.

Our results are consistent with a model in which a local sourceof Spc1 and Spc4 in the ExE activates Nodal in adjacent tissue to

BrachyNodale ga c

b f hd

Nodal lacZ (E6.5)

+/lacZ

Bmp4

q r

tsBmp4

γ-tub

1 2u

lacZ/lacZ

E5.5 E7.5 E7.5E6.5 E6.5

CriptoEomes Pou5f1 Wnt3m oi k

j n pl

E6.5 E6.5 E6.5 E6.25

Lefty-2

Figure 2 Genes downstream of Nodal are downregulated in Spc1−−/−−;;Spc4−−/−−

embryos. a–p, As compared with control litter mates (top) double mutants (bot-tom) express reduced amounts of the genes indicated at the top. Nodal mRNAitself is detected in all embryos examined between E5.5 and E6.5 (n = 30),although expression is delayed and markedly attenuated in double mutants (b, d).q, r, At E5.5, Nodal also fails to be upregulated in homozgyous Nodal mutants car-rying a NodallacZ reporter allele, because positive feedback signalling is inhibited3.By contrast, expression is detectable on day E6.5 in heterozygotes (q) and in

homozgotes (r, n = 4/8), arguing that induction can occur independently of Nodalprotein. s–u, Expression of Bmp4 mRNA (s, t) and protein (u) is also reduced indouble mutants (lane 2, n = 26) as compared with control litter mates (lane 1,n = 19). γ-Tubulin was used as a control for protein loading. Densitometric analysisdid not indicate a significant shift in the relative amounts of precursor (relativemolecular mass 54,000; Mr 54K), partially processed (29K) and mature (23K)forms of Bmp4.



NATURE CELL BIOLOGY VOL 4 DECEMBER 2002 www.nature.com/naturecellbiology984

specify AVE and to induce expression of Cripto. Spc1 and Spc4 alsostimulate induction of Bmp4, which signals back to the epiblast toamplify expression of Cripto in proximity to the ExE (Fig. 4r). Incombination, these effects are likely to establish a proximal–distalgradient of Nodal signalling marked by the asymmetric expressionof Cripto, which in turn might coordinate cell fate decisions andcell movements in the embryo or visceral endoderm, or both. Spc1and Spc4 represent the first determinants in the ExE shown to acti-vate Nodal signalling. Thus in mammals, the ExE seems to functionequivalently to the Nieuwkoop centre that specifies dorsal–ventralpolarity in Xenopus laevis embryos. Localized proteolytic activitiesin the Nieuwkoop centre have been speculated to be responsible forlocal maturation of the TGF-β family member Vg-1 (ref. 21), buttheir identification has remained elusive, perhaps because Xenopushomologues of Spc1 and Spc4 also function redundantly and maybe difficult to inhibit simultaneously.

MethodsMouse strains and ES cellsMice cis heterozygous for null alleles of Spc1 (ref. 22) and Spc4 (ref. 23) were maintained on a mixed

C57Bl/6 × 129 SvEv/SvJ genetic background at the ISREC mouse facility in individually ventilated

cages. Owing to the close proximity of the Spc1 and Spc4 loci on chromosome 7, only 9% of Spc1+/−;

Spc4+/− double heterozygotes obtained from Spc1+/−; Spc4−/− matings transmitted the mutant alleles in

cis. When backcrossed with Spc4 homozygotes, their offspring yielded viable Spc1+/−; Spc4−/− male pups,

which were mated with cis heterozygous females to obtain Spc1−/−; Spc4−/− compound homozygotes.

We maintained R26.1 ES cells carrying the Rosa26lacZ allele24 on STO fibroblast feeder cells and used

them at passage numbers 14–17. Double mutant ES cells were derived from blastocyst outgrowths cul-

tured in DMEM medium containing 10% fetal bovine serum (FBS), 5% knockout serum replacement

(Invitrogen, Groningen, The Netherlands) and 30% ES-cell-conditioned medium. Stably transfected

double mutant ES cells were selected in 0.5 µg ml−1 puromycin after electroporation of the murine

Spc1 coding sequence inserted between a human EF1α promoter fragment and an IRESpac cassette25.

The probes for genotyping by Southern hybridization have been described22,26. We genotyped embryos

and mice carrying a NodallacZ reporter allele by polymerase chain reaction (PCR) after X-gal staining27.

Histology and in situ hybridizationsFor histology, paraffin-embedded embryos were sectioned at 7 µm and stained with haematoxylin and

eosin. The probes for RNA whole-mount in situ hybridization have been described3. LacZ expression

was visualized by X-Gal staining overnight after embryos were fixed for 30–45 min on ice. Paraffin-

sections (8 µm) of whole-mount stained embryos were photographed without counterstaining using

Nomarsky optics.

Nodal, Lefty-1 and Bmp4 cleavage assaysNodal, Spc1, Spc4 and Spc7 expression constructs, and transient transfections of COS-1 cells, were done

as described18. Lefty-1 expression vector was obtained from H. Hanada (Osaka, Japan) Spc6A and Spc6B

expression vectors were a gift from K. Nakayama (Tsukuba, Japan). The medium from transfected COS-1

NodalRQRR'HHL

a

b

c

d

e

+/– –/– +/– –/– –/– Spc1+/– +/+ –/– –/– Spc4M

ock

Nod

al

Spc1 Spc 1 4 6A 6B

RQRR'HHL

Mock

f RLU

**

0 10 20 30 40

– 7

NodalSQAG'HLE

Lefty-1

Lefty-1+Inhibitor

–/–

– – – – +

SQAG'HLE p=0.0023

Figure 3 Maturation occurs after secretion and potentiates Nodal activity.Nodal activity is potentiated by Spc-mediated cleavage. a–b, Conditioned mediumof COS-1 cells expressing Flag-tagged Nodal with a wild-type (a) or mutant (b)cleavage site was immunoblotted with antibodies against Flag after incubation for4 h with the soluble Spc activities indicated at the top. c–d, Similar to Nodal, incu-bation of secreted Lefty-1 with conditioned medium of cells transfected with Spc1,Spc4 or Spc6, but not Spc7, results in substantial cleavage (c), which is blockedby the Spc inhibitor decRVRRcmk (d). e, Conditioned medium of COS-1 cells con-taining Flag–Nodal precursor (left) was incubated for 7 h with mutant ES cellsbefore analysis of Nodal cleavage by anti-Flag immunoblotting. The ES cells carriedtwo, three or four copies of the Spc1 and Spc4 null alleles (top). As a control, cellswere transfected with Spc1 (last lane). f, As compared with wild-type Nodal(RR′HHL), cleavage-deficient Nodal (SQAG′HLE) induces reporter activity lesspotently in luciferase assays, although it is significantly more active than empty vec-tor alone (mock).

Mock

Mock

BMP4

i j k l

Bmp4m

n

Fgf8

Mock

o p

q

h

e

ex

ve

Cripto

f g

Spc1

a b cec ec

ec

exe exeexe

E5.5 E5.5 E6.5

Spc4

Cripto

1 2 3d e

r

NodalCripto

AVE

Signaling

Spc1/4Bmp4

NodallacZ

Mock

Figure 4 Regulation of Nodal signalling by extra-embryonic Spc1 and Spc4.a–c, Spc4 (a) and Spc1 (b) are expressed in ExE at the earliest stages examined(E5.5). By E6.5, Spc1 expression spreads to visceral endoderm and proximal epi-blast (c). d–g, Expression of the NodallacZ reporter allele (d, f) and Cripto mRNA (e,g) in whole embryos (d, e) and epiblast explants (f, g) cultured for 20 h after isola-tion on day E5.5. Nodal expression is retained in the absence of ExE (f) but fails toinduce Cripto (g). h, Epiblasts (e) separated from visceral endoderm (ve) and ExE(ex) were cultured in the conditioned medium of 293T cells expressing Nodal withthe wild-type cleavage site (RQRR′HHLP, lane 1) or engineered sites that are eithercompletely processed (RQRR′HELP, lane 2) or resistant18 to Spc-mediated cleavage(SQAG′HELP, lane 3). i–l, Cripto expression in epiblasts treated with supernatant ofuntransfected cells (i) or conditioned medium containing only uncleaved (j) orprocessed (k) Nodal. Arrows indicate two weakly positive explants (j). Cripto is alsoinduced by recombinant BMP4 (l). m–q, Bmp4 and Fgf8 are downregulated in iso-lated ExE (m) and epiblast explants (p), respectively. Cleavage-deficient Nodal pre-cursor induces Bmp4 in ExE (n) and restores Fgf8 expression in some (n = 4/8)cultured epiblasts (q) similar to that observed in whole embryo cultures (o). r,Regulation of Nodal signalling by extraembryonic Spc1 and Spc4. Spc activitiesprovided by ExE (red) cleave Nodal precursor (white arrow) to stimulate Nodal auto-induction and Bmp4 expression (yellow arrow), and to establish a proximal–distalgradient of Nodal signalling marked by the asymmetric expression of Cripto (blue).This model predicts that epiblast cells in proximity to the ExE give rise to mesendo-derm progenitors ( ) in response to peak quantities of Nodal, whereas cellsreceiving lower signals become mesoderm ( ) or ectoderm. In the overlying vis-ceral endoderm (pink), Nodal specifies AVE as a source of feedback inhibitors,such as Lefty-1 and Cer-l, that reinforce a graded Nodal signal.




cells was conditioned 24 h after transfection for 24–48 h in OptiMEM I (Invitrogen) containing 0.25%

knockout serum replacement (Invitrogen), and then cleared of debris by centrifugation. The Spc

inhibitor decRVRRcmk and fluorogenic substrate Boc-Arg-Val-Arg-Arg-AMC were obtained from Alexis

(Switzerland) and the mouse monoclonal antibody against BMP4 was from R&D Systems (Abingdon,

UK). Polyclonal antibody against Lefty-1 was from Santa Cruz Biotechnology, Santa Cruz, CA.

Embryo explant culturesWhole embryos, epiblast and ExE explants from NMRI mice were dissected on day E5.5 and cultured

for 20 h in ES cell medium (DMEM containing 15% FBS, 1% (v/v) glutamine and 100 µg ml−1 gen-

tamycin sulphate) in Millipore filter inserts (pore size 12 µm) underlaid with γ-irradiated STO fibrob-

lasts expressing LIF (leukaemia inhibitory factor). For factor treatment, epiblasts were freed from vis-

ceral endoderm using trypsin and pancreatin, and cultured in OptiMEM I containing 15% (v/v)

knockout-serum replacement factors, 1% glutamine and gentamycine sulphate. Recombinant Nodal

was produced in stably transfected 293T cells. We monitored protein expression as described18 by

immunoblot analysis of 100 µl conditioned medium (OptiMEM I). Supernatants of the parental

(mock) or transfected cell lines were concentrated 20-fold and rediluted to their original concentration

in fresh embryo culture medium. Human BMP4 (R&D systems) was applied at 50 ng ml−1.

RECEIVED 4 OCTOBER 2002, REVISED 24 OCTOBER 2002, ACCEPTED 24 OCTOBER 2002,PUBLISHED 25 NOVEMBER 2002.

1. Beddington, R. S. & Robertson, E. J. Cell 96, 195–209 (1999).

2. Weber, R. J., Pedersen, R. A., Wianny, F., Evans, M. J. & Zernicka-Goetz, M. Development

126, 5591–5598 (1999).

3. Brennan, J. et al. Nature 411, 965–969 (2001).

4. Liu, P. et al. Nature Genet. 22, 361–365 (1999).

5. Lawson, K. A. & Pedersen, R. A. Development 101, 627–652 (1987).

6. Kinder, S. J. et al. Development 128, 3623–3634 (2001).

7. Wei, Y. & Mikawa, T. Development 127, 87–96 (2000).

8. Skromne, I. & Stern, C. D. Development 128, 2915–2927 (2001).

9. Ang, S. L. et al. Development 122, 243–252 (1996).

10. Ding, J. et al. Nature 395, 702–707 (1998).

11. Dono, R. et al. Development 118, 1157–1168 (1993).

12. Saijoh, Y. et al. Mol. Cell 5, 35–47 (2000).

13. Yeo, C.-Y. & Whitman, M. Mol. Cell 7, 949–957 (2001).

14. de Souza, F. S. & Niehrs, C. Cell Tissue Res. 300, 207–217 (2000).

15. Belo, J. A. et al. Mech. Dev. 68, 45–57 (1997).

16. Kimura, C. et al. Dev. Biol. 225, 304–321 (2000).

17. Perea-Gomez, A. et al. Development 128, 753–765 (2001).

18. Constam, D. B. & Robertson, E. J. J. Cell Biol. 144, 139–149 (1999).

19. Constam, D. B., Calfon, M. & Robertson, E. J. J. Cell Biol. 134, 181–191 (1996).

20. Winnier, G., Blessing, M., Labosky, P. A. & Hogan, B. L. M. Genes Dev. 9, 2105–2116 (1995).

21. Thomsen, G. H. & Melton, D. A. Cell 74, 433–441 (1993).

22. Roebroek, A. J. M. et al. Development 125, 4863–4876 (1998).

23. Constam, D. B. & Robertson, E. J. Genes Dev. 14, 1146–1155 (2000).

24. Varlet, I., Collignon, J. & Robertson, E. J. Development 124, 1033–1044 (1997).

25. Hobbs, S., Jitrapakdee, S. & Wallace, J. C. Biochem. Biophys. Res. Commun. 252, 368–372

(1998).

26. Constam, D. B. & Robertson, E. J. Development 127, 245–254 (2000).

27. Collignon, J., Varlet, I. & Robertson, E. J. Nature 381, 155–158 (1996).

ACKNOWLEDGEMENTS

We thank L. Robertson and A. Grapin-Botton for comments on the manuscript; L. Bikoff

for antiserum to Nodal; J. Christian for testing antibodies to BMP4; J. Rossant and A. Russ

for Eomes cDNA; E. Säuberli and G. Badic Benedetto for histology; and P.-A. Christinet

and his staff for animal care. This work was supported by a long term fellowship from the

Human Frontier Science Program (J.A.L.), and a grant from the Swiss National Science

Foundation (D.B.C.).

Correspondence and requests for materials should be addressed to D.B.C.

COMPETING FINANCIAL INTERESTS

The authors declare that they have no competing financial interests.


extraembryonic proteases regulate nodal signalling during gastrulation

Documents