Dorsoventral polarity of the somitic mesoderm is established by competitive signals originating from adjacent tissues. The ventrally located notochord provides the ventralizing signals to specify the sclerotome, while the dorsally located surface ectoderm and dorsal neural tube provide the dorsalizing signals to specify the dermomyotome. Noggin and SHH-N have been implicated as the ventralizing signals produced by the notochord. Members of the WNT family of proteins, on the other hand, have been implicated as the dorsalizing signals derived from the ectoderm and dorsal neural tube. When presomitic explants are confronted with cells secreting SHH-N and WNT1 simultaneously, competition to specify the sclerotome and dermomyotome domains within the naive mesoderm can be observed. Here, using these explant cultures, we provide evidence that SHH-N competes with WNT1, not only by upregulating its own receptor Ptc1, but also by upregulating Sfrp2 (Secreted frizzled-related protein 2), which encodes a potential WNT antagonist. Among the four known Sfrps, Sfrp2 is the only member expressed in the sclerotome and upregulated by SHH-N recombinant protein. We further show that SFRP2-expressing cells can reduce the dermomyotome-inducing activity of WNT1 and WNT4, but not that of WNT3a. Together, our results support the model that SHH-N at least in part employs SFRP2 to reduce WNT1/4 activity in the somitic mesoderm.

Somites are segmented mesodermal structures situated on both sides of the neural tube. Somites eventually give rise to the vertebrae, ribs, muscle and dermis. Each of these tissues comes from distinct groups of cells within a somite (reviewed by Keynes and Stern, 1988). One of the initial specification events is the formation of sclerotome and dermomyotome. Sclerotome, precursor to the vertebrae and ribs, is located ventrally, whereas the dermomyotome, precursor to the muscle and dermis, is located dorsally. The dermomyotome is further divided into the dermotome, the dermal precursor, the medial myotome lip, the epaxial muscle precursor, and the lateral myotome lip, the hypaxial muscle precursor (reviewed by Christ and Ordahl, 1995).

Dorsoventral (D/V) polarity of the somites is presumed to be established through competitive actions of signals derived from the dorsally situated surface ectoderm/dorsal neural tube and the ventrally situated notochord/floorplate (Fan and Tessier-Lavigne, 1994). The molecular identities of the ventral signals are attributed to Noggin and SHH-N (Fan and Tessier-Lavigne, 1994; Johnson et al., 1994; Fan et al., 1995; McMahon et al., 1998). It appears that Noggin functions at the initiation step and SHH-N functions at the maintenance/ augmentation step for sclerotome development (McMahon et al., 1998; Chiang et al., 1996). In vitro studies provide further evidence that Noggin and SHH-N act synergistically to antagonize the BMP2/4 signal(s), thereby allowing the sclerotome program to proceed (McMahon et al., 1998). Dermomyotome, on the other hand, appears to receive input from an array of WNT proteins expressed in the surface ectoderm (WNT4 and WNT6) and dorsal neural tube (WNT1 and WNT3a) (Fan et al., 1997; Capdevila et al., 1998; Marcelle et al., 1997). Although the lack of purified functional recombinant WNTs questions their direct roles, the fact that a retroviral vector carrying an activated form of β-catenin can cause the expansion of the dermomyotomal marker Pax3 expression domain in developing chick somites is consistent with the involvement of WNT signaling (Capdevila et al., 1998). Since both SHH-N and WNTs are secreted molecules and can function over more than the length of a somite in vitro (Fan and Tessier-Lavigne, 1994; Fan et al., 1997), the D/V polarity is presumably maintained by the balanced activities of the two signals, which likely display opposing concentration gradients across the somitic field (Fan et al., 1997).

One interpretation of the observed mutual exclusion between the sclerotome and dermomyotome fates is that it is mediated through competitive ‘intra’-cellular signaling solicited by SHH-N and WNTs. In this scenario, each somitic cell receives both signals, and the stronger signal dominates and determines the cell’s fate. Evidence has accumulated that SHH-N utilizes the PTC/SMO receptor complex-mediated signaling pathway (Stone et al., 1996), and that WNT proteins employ the Frizzled (FRZ) receptor-mediated signaling pathway (Wang et al., 1996; Bhanot et al., 1996; He et al., 1997; reviewed by Wardoz and Nusse, 1998). While the signaling properties of either are not yet completely defined (Slusarski et al., 1997), it is intriguing to note that SMO and FRZ are both seven-pass membrane receptors (Stone et al., 1996; Wang et al., 1996). How the two signaling pathways intersect and negate each other within a somitic cell remains to be determined. Alternatively, it is also possible that this competitive action is mediated via a third factor(s). Recent description of the Hip gene suggested that there is at least one SHH-N inhibitor expressed in the somite (Chuang and McMahon, 1999). However, since Hip is expressed in the ventralmost portion of the sclerotome and is regulated by SHH to refine the range of SHH’s own action, it is not a candidate ‘competition mediator’ utilized by WNTs.

On the other hand SFRP3, a member of the SFRP (Secreted Frizzled-Related-Protein) family, has been shown to function as an inhibitor of WNT proteins via direct binding (Rattner et al., 1997; Finch et al., 1997; Leyns et al., 1997; Mayr et al., 1997; Wang et al., 1997a,b) through its cysteine-rich domain (CRD), which is also found in the FRZ proteins (Lin et al., 1997; reviewed by Wardoz and Nusse, 1998). To investigate whether any of the four known SFRPs helps SHH-N to antagonize WNTs, we first examined their expression in the developing sclerotome. While Sfrp1, Sfrp2 and Sfrp3 are expressed in the presomitic mesoderm, only the Sfrp2 transcripts are upregulated in the sclerotome. We then demonstrated that Sfrp2 expression in the paraxial mesoderm is upregulated by the notochord and SHH-N. Finally, we showed that cells transfected with a myc-tagged Sfrp2 cDNA expression vector produced secreted recombinant protein and functioned to counteract the dermomyotome-inducing activity of WNT1 and WNT4.

Cell culture

COS, RHW1 and RatB1a cells were cultured in 10% fetal bovine serum (Hyclone) /DME (Gibco-BRL) with penicillin/streptomycin supplement (Gibco-BRL). COS cells were transfected with vector alone or Sfrp2myc cDNA expression plasmids (gifts from Dr J. Nathans) by lipofectamine (Gibco-BRL). Cell aggregates of these cell lines were used as the source of recombinant WNT1 and SFRP2. Parental 3T3 cells and WNT7A-, WNT4- and WNT3A-expressing 3T3 cells were cultured in the same medium in the presence of 300

μg/ml of G418 (Kispert et al., 1998). Purified recombinant BMP4, SHH-N and Noggin were supplemented in the explant culture medium where indicated (Fan et al., 1995; McMahon et al., 1998).

SFRP2 recombinant protein assay

Approximately 1×105 COS cells in a 6-well dish were used for each transfection (using 1 μg of control or expression vector) using the lipofectamine method (Gibco-BRL). 24 hours after transfection, the regular medium was replaced with 1 ml of OPTI-MEMI (Gibco-BRL) for an additional 40 hours. The conditioned medium was collected, and the cells were washed 3× with PBS and then extracted with 0.5 ml of RIPA buffer (Harlow and Lane, 1988). 20 μl of the conditioned supernatant and 10 μl of the RIPA extracts were used for SDS-PAGE/western analysis using the 9E10 antibody, followed by alkaline-phosphatase conjugated goat anti-mouse antibody and BCIP/NBT color development (Gibco-BRL).

Explant culture

Presomitic mesodermal (psm) explants were isolated from gestation day (E)9.5 mouse embryos and cultured in collagen gels as described (Fan and Tessier-Lavigne, 1994.). Cocultured explants with cell clumps and recombinant proteins are specified in the corresponding text and figure legends.

In situ hybridization

Collagen gel embedded explants were lightly treated with collagenase (2 units/ml; Sigma) in culture medium without serum for approximately 10 minutes at room temperature or until the top layer of collagen was just about to detach. The collagenase solution was then replaced by 4% paraformaldehyde in PBS for overnight fixation. Standard DIG whole-mount in situ hybridization was performed with these gel/explant pieces using great care. Anti-sense Pax3 probe was produced as previously described (Fan and Tessier-Lavigne, 1994) using DIG-UTP for labeling. T7 polymerase (Promega)-transcribed Sfrp1-4 anti-sense DIG-labeled probes were used to characterize their expression patterns in E9.5 embryos by whole-mount in situ hybridization. For section in situ, E9.5 embryos were fixed by Carnoy’s, embedded in paraffin wax and sectioned at 8 μm thickness. [35S]UTP-incorporated Sfrp1-4 anti-sense transcripts (plasmids used to synthesize the Sfrp1-4 probes were kindly provided by Dr J. Nathans) were used for hybridization. To assess regulated Sfrp2 expression in the psm explants (referring to Fig. 2C), cryostat section (12 μm) radioactive in situ hybridization was performed using a [35S]UTP-labeled anti-sense Sfrp2 probe. The radioactive section in situ hybridization protocol is described in Frohman et al. (1990). The slides were dipped in NTB-2 emulsion and developed after 8-10 days of exposure. The photography is presented as double-exposure images of a darkfield image with red filter and a brightfield image with blue filter under a Zeiss Axioskop. To quantitate the difference of expression levels between psm and somites, darkfield black-and-white digital images were taken and the density of the exposed silver granules (as white signals; density = pixels of granule-positive area/pixels of area of interest) were quantified and compared (granule density in psm/granule density in sclerotome) by the NIH Image program. Background silver granule densities measured from nearby tissue-negative areas of the same pixel numbers were subtracted prior to comparison.

RT-PCR

Total RNA was extracted from each explant by RNAsol, purified and resuspended in 10 μl of H2O. 250-500 ng of total RNA was usually isolated from each explant culture. Typically, 3 μl of each RNA sample was reverse transcribed (RT) using the random priming protocol recommended by the manufacturer (Gibco-BRL) in 15 μl reactions; when RT reactions were performed at different volumes, the ratio of RNA and other components was kept the same. 1 μl of the RT products was then used for PCR in the presence of 10 μCi [γ-32P]dCTP (3000 Ci/mmol) and 0.12 mM dNTPs in 25 μl reaction mixtures. 5 μl samples of each reaction mixture were analyzed on a 6% polyacrylamide gel as previously described (Fan et al., 1995). The primers used for RT-PCR are listed below:

Sfrp1: 5′ primer, CGAGATGCTCAAATGTGACAAGT; 3′ primer, CCCATGATGAGAAGTTGTGGCT; product, 360 bp.

Sfrp2: 5′ primer, AGACATGCTGGAGTGCGACCGTT; 3′ primer, CACTGGGAAACTAGCATTGCAGC; PCR product, 465 bp.

Sfrp3: 5′ primer, ATTCTCATCAAGTACCGCCAC; 3′ primer, TGCCTTTAGAATTTCCTTCAC; product, 270 bp.

Sfrp4: 5′ primer, TCATGAAGATGTATAACCACAG; 3′ primer, GCCACTCATAACACATGATTAG; product, 453 bp.

Ptc1: 5′ primer, AAGGCAGCTAATCTCGAGACC; 3′ primer, GCCAACTTCGGCTTTATTCAG; PCR product, 515 bp.

Primers for Pax1 and β-actin are as described (Fan et al., 1995). The four mouse Sfrp cDNA sequences are described in Rattner et al. (1997). All PCR primers were tested to prime and amplify corresponding genes using plasmid DNA. To choose the mid-exponential amplification brackets for quantitation and comparison (referring to Fig. 2B), the PCR products of different cycles were resolved on 6% polyacrylamide gels, the product bands were excised (after exposure) and counted in scintillation fluid using open channel with 10% gain to discriminate non-specific counts (according to the Packard Tri-Carb liquid scintillation spectrometer manual). Background counts of corresponding product positions from PCR samples performed without RT reactions were subtracted. The final radioactive counts of the RT-PCR products and the number of cycles used were plotted in Fig. 2B for Sfrp2. Under the same experimental conditions, Ptc1 and Pax1 can be detected above background around 17 and 20 cycles of PCR (the concentration threshold, ct) and gradually plateaus after cycles 25 and 26 (data not shown). Thus, for these three genes, 23 cycles of PCR was chosen as the standard for the comparative studies. Similar results were obtained when Sfrp1 and Sfrp3 transcripts were assessed (data not shown). However, Sfrp4 message was not detected within this range of PCR cycles. When comparing these scintillation counting results to the densitometry of the autorads quantified by the NIH image program, we found the relative PCR product ratios obtained by these two methods were virtually identical. Most of the quantitative statements below are based on the densitometry data because of its convenience. Although the autorads presented here were exposed for 2 hours, appropriate exposure times were routinely chosen to ensure that the densitometry analyses were performed within the non-saturating linear range of the X-ray film. The quantity of each RNA sample differed slightly due to the variability of RNA extraction from small tissues and the different growth rates of explants that have undergone different treatments. The fold of induction was therefore ‘normalized’ as the density and/or the radioactive counts of (Sfrp2-induced/β-actin-induced)/(Sfrp2-uninduced/β-actin uninduced).

Sfrps are expressed in the mouse somitic mesoderm

To test whether any of the known Sfrps are involved in somite patterning, we first examined their expression in the somitic mesoderm. In situ RNA expression patterns of some of the Sfrps had been previously described but had not been specifically focused on early somitic development (Hoang et al., 1998; Leimeister et al., 1998; Lescher et al., 1998). We used tissue section in situ hybridization with [35S]UTP-labeled anti-sense probes to examine their somitic expression pattern in E9.5 mouse embryos. E9.5 embryos contain paraxial mesoderm at stages ranging from an undifferentiated state (the presomitic mesoderm, psm) to fully differentiated sclerotome and dermomyotome (Fig. 1A). Except for Sfrp4, which does not appear to be expressed in either psm or somites (Fig. 1B), Sfrp1, Sfrp2 and Sfrp3 are all expressed in the psm at a comparable level. As somites form and progress, Sfrp1 is immediately downregulated and retains transient and low levels of expression in the lateral tip of the dermomyotome (not shown). In contrast, Sfrp2 becomes upregulated in the sclerotome; however, its expression in the dermomyotome remains at a similar level to that in the psm. When the exposed silver granules were quantified (see Materials and Methods), we observed a 4.18±0.53-fold increase (N=3) of Sfrp2 expression in the sclerotome versus psm. Sfrp3, on the other hand, is downregulated but not absent in the newly formed somites. At the 6th or 7th somites anterior to the psm, Sfrp3 expression is no longer detectable. Consistent with the published results (Hoang et al., 1998), Sfrp3 expression is also detected transiently in cells between the somites and the neural tube, which are most likely the neural crest cells (not shown). The above results are also confirmed by whole-mount in situ hybridization using DIG-labeled probes (not shown).

Fig. 1.

In situ hybridization indicates that Sfrp2 is expressed in the sclerotome. [35S]UTP-labeled Sfrp1, Sfrp2, Sfrp3 and Sfrp4 anti-sense RNA probes were used to evaluate their spatial expression patterns within the developing somitic mesoderm of E9.5 mice. (A) A drawing of an E9.5 mouse. The two lines across the caudal region of the embryo indicate the approximate axial levels of sections presented in (B). psm, presomitic mesoderm; som, somites. (B) In situ hybridization data of Sfrp1-4. The probes used are indicated on the left and the levels of the paraxial mesoderm are labeled on the top. To accurately assess the levels of the paraxial mesoderm sections, all transverse sections (8 μm) of each embryo were collected and hybridized to one probe. The presomitic sections presented are within 100 μm of the last recognizable forming somite, and the somite level is at the 5th or 6th somite from the psm. The variable curvature of the E9.5 body axis precludes perfect transverse sections through both the psm and the somite levels of a given embryo, as shown by some of the elongated/distorted sections. Sfrp1, Sfrp2 and Sfrp3 (but not Sfrp4) transcripts were detected in the psm. Photographs were taken as double exposures of a darkfield/red filter image and a phase /blue filter image. Sfrp2 is expressed at a higher level in the sclerotome than the psm as shown by the increased density of the pinkish silver granules. Sfrp1 expression in the lateral dermomyotome and Sfrp3 expression in the somites, although above background, cannot be clearly seen under the same exposure time that was used to show Sfrp2 expression. S, Sclerotome; dm, dermomyotome; nt, neural tube; mn, mesonepheric mesoderm; e, endoderm; lp, lateral plate mesoderm. Bar, 100 μm.

Fig. 1.

In situ hybridization indicates that Sfrp2 is expressed in the sclerotome. [35S]UTP-labeled Sfrp1, Sfrp2, Sfrp3 and Sfrp4 anti-sense RNA probes were used to evaluate their spatial expression patterns within the developing somitic mesoderm of E9.5 mice. (A) A drawing of an E9.5 mouse. The two lines across the caudal region of the embryo indicate the approximate axial levels of sections presented in (B). psm, presomitic mesoderm; som, somites. (B) In situ hybridization data of Sfrp1-4. The probes used are indicated on the left and the levels of the paraxial mesoderm are labeled on the top. To accurately assess the levels of the paraxial mesoderm sections, all transverse sections (8 μm) of each embryo were collected and hybridized to one probe. The presomitic sections presented are within 100 μm of the last recognizable forming somite, and the somite level is at the 5th or 6th somite from the psm. The variable curvature of the E9.5 body axis precludes perfect transverse sections through both the psm and the somite levels of a given embryo, as shown by some of the elongated/distorted sections. Sfrp1, Sfrp2 and Sfrp3 (but not Sfrp4) transcripts were detected in the psm. Photographs were taken as double exposures of a darkfield/red filter image and a phase /blue filter image. Sfrp2 is expressed at a higher level in the sclerotome than the psm as shown by the increased density of the pinkish silver granules. Sfrp1 expression in the lateral dermomyotome and Sfrp3 expression in the somites, although above background, cannot be clearly seen under the same exposure time that was used to show Sfrp2 expression. S, Sclerotome; dm, dermomyotome; nt, neural tube; mn, mesonepheric mesoderm; e, endoderm; lp, lateral plate mesoderm. Bar, 100 μm.

Sfrp2 is a downstream target gene induced by the notochord and SHH-N in the somitic mesoderm

The expression patterns of Sfrp1-3 in the somitic mesoderm suggest that they may all be involved in some aspects of somite development. Since Sfrp2 is the only member whose expression is upregulated in the sclerotome, it is the most likely candidate to be utilized by SHH-N to antagonize WNT proteins. To further explore this possibility, we tested whether its expression could be regulated by recombinant SHH-N or the endogenous SHH-producing tissue notochord. Psm explants isolated from E9.5 embryos were cultured for 24 hours, either alone or in the presence of the notochord or recombinant SHH-N (Fan and Tessier-Lavigne, 1995). Total RNA of these explants was prepared and RT-PCR performed to monitor the level of gene expression (Fig. 2A). To better quantify the induction by PCR, Sfrp1, Sfrp2, Sfrp3, Pax1 and Ptc1 were tested for their exponential amplification bracket to be between 19-25 cycles using total RNA isolated from the psm and the five most caudal somites of E9.5 embryos (not shown). 23 cycles of PCR were used for all of the following experiments to standardize the comparison. β-actin message level was assessed as a control to normalize the comparison (see Materials and Methods). Although the levels of β-actin appeared to be slightly upregulated in the SHH-N or notochord treated samples (Fig. 2A), these results correlate directly with the fact that more RNA was isolated from these explants, presumably due to the growth-promoting activity of the notochord and SHH-N (Fan et al., 1995).

Consistent with its endogenous expression pattern, Sfrp2 transcripts were detected in the freshly isolated psm explant (Fig. 2A; N=4). After 24 hours of culture, its expression was maintained or slightly reduced in the explants (after normalization to the β-actin control). When cocultured with the notochord or 500 ng/ml of SHH-N, psm explants were induced to express higher levels of Sfrp2, strongly suggesting that its role is to mediate the antagonism between SHH-N and WNTs. When normalized to the β-actin control, induction of Sfrp2 by SHH-N was five-to eightfold, even with this saturating concentration of SHH-N (N=4; and see below). This observation is consistent with the result obtained using the same amount of harvested sample RNA for RT-PCR to compare the Sfrp2 expression level of induced and non-induced explants. The incorporated radioactive counts of the mid-exponential amplification cycles (22-24) indicated that the induction is about sixfold (N=2; Fig. 2B). Furthermore, explants treated with SHH-N (500 ng/ml) did indeed express higher levels of Sfrp2 than the untreated samples, as confirmed by radioactive in situ hybridization (N=4; Fig. 2C). The induction was assessed to be six-to ninefold by comparing the silver granule density over the explant sections. When the cell number of each section was taken into consideration, approximately five-to sevenfold of induction was obtained. Since similar conclusions were obtained from these independent measurements, we conclude that Sfrp2 expression in the psm can be upregulated by SHH-N in vitro.

Sfrp1 expression was slightly upregulated in cultured psm compared to the freshly isolated psm (Fig. 2A). Surprisingly, Sfrp1 expression was also upregulated by the notochord and to a lesser extent by SHH-N. At present, we do not understand this regulated expression in vitro. Since Sfrp1 transcripts were not detected in the sclerotome, we excluded its role as an antagonistic factor to WNTs in the sclerotome. Unlike Sfrp1 and Sfrp2, Sfrp3 expression was not maintained in the psm cultured alone. Its expression was also not detected in psm cocultured with the notochord or with SHH-N (Fig. 2A). In contrast, we had observed that both Sfrp1 and Sfrp3 expression were highly upregulated in the presence of BMP4 recombinant protein (C. S. Lee and C.-M. Fan, unpublished; and see Discussion). In parallel, we also assessed the expression of Ptc1 and Pax1, known downstream genes of SHH-N, in the RNA samples of the same explants for comparison in these induction assays (Fig. 2A).

Results of the expression pattern survey and the in vitro assay suggest that of the four known Sfrps, only SFRP2 is likely to be the WNT antagonistic factor utilized by SHH-N within the ventral somite. The moderate induction of Sfrp2 by SHH-N led us to further examine the sensitivity of its response to SHH-N alone and in combination with other factors. When psm explants were cultured with increasing concentrations of purified SHH-N, increasing levels of Sfrp2 expression were observed (Fig. 3A; N=3). Similar to Pax1 and Ptc1, Sfrp2 induction could be observed when the concentration of SHH-N applied was 25 ng/ml or more. Sfrp2 upregulation plateaued around 200 ng/ml of SHH-N and no more than eightfold induction was ever observed (after normalization to β-actin; Fig. 3A) even with excess SHH-N. Compared with the linear response of Ptc1 and Pax1 to the concentration of the added SHH-N (Fig. 3A), the response of Sfrp2 was not robust. This observation is consistent with the moderate increase of Sfrp2 expression in the sclerotome versus psm observed by in situ hybridization.

Fig. 2.

Sfrp2 expression in the presomitic mesoderm can be upregulated by the notochord and SHH-N. (A) E9.5 psm explants were dissected and cultured in vitro for 24 hours either alone, in the presence of notochord of the same axial level (+n), or with recombinant SHH-N (+SHH-N, 500 ng/ml). Explant samples were harvested and prepared for RT-PCR using oligonucleotide primers for Sfrp1-4, Pax1, Ptc1 and β-actin (N=4). All PCR reactions were performed with 23 cycles of amplification in the presence of [γ-32P]dCTP. For details, see Materials and Methods. (B) To quantitate the Sfrp2 expression level, the incorporated radioactive counts of the Sfrp2 PCR products (cpm, counts per minute; the left vertical axis) using RT reactions starting with the same amount of RNA harvested from uninduced (black squares) and induced (500 ng/ml SHH-N; black diamonds) were plotted against the PCR cycle number used (the horizontal axis). The conditions of the PCR were chosen in order that the products were exponentially amplified between 19 and 25 cycles. Data are presented as log2(cpm); the right vertical axis is in red. The log2 values of the SHH-N treated and untreated sample counts taken from the left axis are presented as red diamonds and red squares, respectively. 23 cycles was chosen as a mid-point for quantification for this study. (C) The Sfrp2 expression level in the untreated explant cultures (−SHH-N, top panel) or explants treated with 500 ng/ml of SHH-N (+SHH-N, bottom panel) was assessed by radioactive section in situ hybridization using a [35S]UTP-labeled anti-sense Sfrp2 probe (N=4). Bar, 50 μm.

Fig. 2.

Sfrp2 expression in the presomitic mesoderm can be upregulated by the notochord and SHH-N. (A) E9.5 psm explants were dissected and cultured in vitro for 24 hours either alone, in the presence of notochord of the same axial level (+n), or with recombinant SHH-N (+SHH-N, 500 ng/ml). Explant samples were harvested and prepared for RT-PCR using oligonucleotide primers for Sfrp1-4, Pax1, Ptc1 and β-actin (N=4). All PCR reactions were performed with 23 cycles of amplification in the presence of [γ-32P]dCTP. For details, see Materials and Methods. (B) To quantitate the Sfrp2 expression level, the incorporated radioactive counts of the Sfrp2 PCR products (cpm, counts per minute; the left vertical axis) using RT reactions starting with the same amount of RNA harvested from uninduced (black squares) and induced (500 ng/ml SHH-N; black diamonds) were plotted against the PCR cycle number used (the horizontal axis). The conditions of the PCR were chosen in order that the products were exponentially amplified between 19 and 25 cycles. Data are presented as log2(cpm); the right vertical axis is in red. The log2 values of the SHH-N treated and untreated sample counts taken from the left axis are presented as red diamonds and red squares, respectively. 23 cycles was chosen as a mid-point for quantification for this study. (C) The Sfrp2 expression level in the untreated explant cultures (−SHH-N, top panel) or explants treated with 500 ng/ml of SHH-N (+SHH-N, bottom panel) was assessed by radioactive section in situ hybridization using a [35S]UTP-labeled anti-sense Sfrp2 probe (N=4). Bar, 50 μm.

Fig. 3.

Sfrp2 expression is upregulated by SHH-N. (A) Titration of the responsiveness of Sfrp2 transcription to recombinant SHH-N. Psm explants were cultured in the absence or presence of various concentrations of recombinant SHH-N (12.5-400 ng/ml, indicated at the top). Sfrp2, Pax1, Ptc1 and β-actin expression were assessed in these samples by RT-PCR. Note that Sfrp2, Pax1 and Ptc1 are upregulated in response to 25 ng/ml of SHH-N and above, except that basal expression of Ptc1 and Sfrp2 was maintained in culture without additional SHH-N. (B) Sfrp2 upregulation by SHH-N can be antagonized by the SHH-N inhibitors, Forskolin, IBMX and 5E1 antibody. Psm explants were treated with 100 ng/ml of SHH-N so that the induction of Sfrp2 could be easily assayed. Forskolin (100 μM), IBMX (100 μM), or the functional blocking antibody of SHH-N, 5E1 (17.5 μg/ml), were included in these cultures. The induction of Sfrp2, Pax1 and Ptc1 by SHH-N was inhibited in the presence of these reagents.

Fig. 3.

Sfrp2 expression is upregulated by SHH-N. (A) Titration of the responsiveness of Sfrp2 transcription to recombinant SHH-N. Psm explants were cultured in the absence or presence of various concentrations of recombinant SHH-N (12.5-400 ng/ml, indicated at the top). Sfrp2, Pax1, Ptc1 and β-actin expression were assessed in these samples by RT-PCR. Note that Sfrp2, Pax1 and Ptc1 are upregulated in response to 25 ng/ml of SHH-N and above, except that basal expression of Ptc1 and Sfrp2 was maintained in culture without additional SHH-N. (B) Sfrp2 upregulation by SHH-N can be antagonized by the SHH-N inhibitors, Forskolin, IBMX and 5E1 antibody. Psm explants were treated with 100 ng/ml of SHH-N so that the induction of Sfrp2 could be easily assayed. Forskolin (100 μM), IBMX (100 μM), or the functional blocking antibody of SHH-N, 5E1 (17.5 μg/ml), were included in these cultures. The induction of Sfrp2, Pax1 and Ptc1 by SHH-N was inhibited in the presence of these reagents.

Unlike Pax1 and similar to Ptc1, Sfrp2 maintained a basal level of expression in cultured psm. It is possible that the minimal amount of SHH-N associated with the dissected psm was sufficient to maintain the ‘basal’ Sfrp2 expression in vitro. To address this, we applied antagonists to SHH-N to the explant cultures (Fan et al., 1995; McMahon et al., 1998). When Forskolin, IBMX or anti-SHH-N function-blocking antibody 5E1 were included in the medium to inhibit SHH-N, we observed a three-to fourfold downregulation but not complete absence of Sfrp2 expression after normalization to β-actin (compare lanes 1, 3, 5 and 7 of Fig. 3B; N=4), suggesting that Sfrp2 expression in the cultured psm was partially but not entirely dependent on SHH-N. Importantly, the SHH-N-upregulated Sfrp2 expression could be inhibited by Forskolin, IBMX and 5E1, indicating the directness of SHH-N action (compare lanes 2, 4, 6 and 8 of Fig. 3B; N=4). The expression of Pax1 and Ptc1 was also examined in the same explants.

While the SHH-N-induced Pax1 expression was completely abolished by the inhibitors at the applied concentrations, some Ptc1 expression was still detectable, indicating that both Ptc1 and Sfrp2 genes have basal transcription that is independent of SHH-N. The reduction of Ptc1 and Sfrp2 expression in the psm in the presence of SHH-N inhibitors strongly suggests that psm is already under the influence of SHH-N and that Ptc1 and Sfrp2 transcription may in fact respond to concentrations of SHH-N lower than 25 ng/ml. It is perhaps significant that if the five-to eightfold induction of Sfrp2 by SHH-N recombinant protein and the three-to fourfold reduction of Sfrp2 by SHH-N inhibitors are multiplied, the maximal response of Sfrp2 to SHH-N may reach approximately 15-to 32-fold.

Noggin and BMP4 play minor roles in regulating Sfrp2 expression in the presomitic mesoderm explants

It is also possible that the remaining Noggin in the dissected tissue is sufficient to maintain Sfrp2 expression since the notochord expresses Noggin, which has also been implicated in sclerotome specification (McMahon et al., 1998). Surprisingly, when recombinant Noggin was included, we observed no upregulation of Sfrp2 expression; rather, there was a consistent slight downregulation (compare lanes 1 and 3 of Fig. 4; N=8). BMP4 has been shown to antagonize the function of SHH-N and Noggin in sclerotome specification (McMahon et al., 1998). We therefore tested the effect of BMP4 in our explant culture. When BMP4 was applied, we did not observe any reduction but a slight upregulation of Sfrp2 expression (compare lanes 1 and 5 of Fig. 4; N=8). Higher concentrations of BMP4 (up to 100 ng/ml) in the culture medium rarely induced Sfrp2 expression by twofold (not shown). Whether this slight modulation by Noggin and BMP4 is significant remains to be determined. When SHH-N and BMP4 or SHH-N and Noggin were applied together, upregulation of Sfrp2 was still observed to the same extent as SHH-N applied alone (compare lanes 2, 4 and 7 of Fig. 4; N=8). Thus, within the confines of our culture conditions, Sfrp2 expression in the somitic mesoderm is primarily upregulated by SHH-N. We also monitored the expression of Ptc1 and Pax1 from the same explant RNA samples (N=3). Similiar to the previous report (McMahon et al., 1998) on Pax1 regulation, Noggin had a minimal effect in activating Ptc1 expression, and synergized with SHH-N to activate Ptc1 expression further (Fig. 4; compare lanes 1 and 3, and 2 and 4). Consistent with this, BMP4 reduced Ptc1 expression slightly (lane 5, Fig. 4). Furthermore, Ptc1 activation by SHH-N was greatly reduced by BMP4 (lane 7, Fig. 4). In the same RNA samples, SHH-N- and Noggin-induced Pax1 expression was completely eliminated in the presence of 10 ng/ml of BMP4. Thus, although Sfrp2, Ptc1, and Pax1 are all SHH-N inducible sclerotome markers, their regulated expression in the somitic mesoderm displays differential sensitivity to BMP4 and Noggin.

Fig. 4.

Sfrp2 induction by SHH-N is not inhibited by BMP4. Psm explants were cultured in the presence or absence of SHH-N at 100 ng/ml in combination with either BMP4 (10 ng/ml) or Noggin (100 ng/ml) for 24 hours. Combinations of these recombinant proteins applied in culture are indicated at the top of the figure. RNA samples from the same explant cultures were subjected to RT-PCR to assess the expression levels of Sfrp2, Pax1 and Ptc1. β-actin was assayed as a control to normalize the expression and RNA recovery.

Fig. 4.

Sfrp2 induction by SHH-N is not inhibited by BMP4. Psm explants were cultured in the presence or absence of SHH-N at 100 ng/ml in combination with either BMP4 (10 ng/ml) or Noggin (100 ng/ml) for 24 hours. Combinations of these recombinant proteins applied in culture are indicated at the top of the figure. RNA samples from the same explant cultures were subjected to RT-PCR to assess the expression levels of Sfrp2, Pax1 and Ptc1. β-actin was assayed as a control to normalize the expression and RNA recovery.

SFRP2 can counteract the dermomyotome-inducing activity of WNT1 and WNT4

WNT1-expressing cells can induce dermomyotome markers in vitro and can be counteracted by SHH-N expressing cells (Fan et al., 1997). One possibility is that SHH-N upregulates SFRP2 expression in the sclerotome (see above) to prevent WNTs from functioning ventrally. To test whether SFRP2 can inhibit WNT1’s dermomyotome-inducing activity, we modified the experimental design from the previously described paradigm (Fan et al., 1997). Instead of placing cell aggregates expressing the antagonizing signals on the opposite sides of the psm, we placed them diagonally. This design allowed us to assess even a weak antagonistic effect at the immediate junction between two opposing signals (Fig. 5B).

Fig. 5.

SFRP2 inhibits WNT1’s dermomyotome-inducing activity. (A) Secreted SFRP2 protein was produced by transfected COS cells. Supernatant (supe) and detergent-extracted cell lysate (RIPA) from transfected COS cells (con., transfected with control plasmid without a cDNA insert; Sfrp2myc, transfected with myc-tagged Sfrp2 expression plasmid) were assessed for SFRP2 production by SDS-PAGE/western blotting. Equivalent fractions of supe and detergent extracts were loaded in each lane to compare the ratio of cell-associated versus secreted SFRP2 protein. 9E10 antibody, which recognizes the myc epitope, was used to detect the myc-tagged SFRP2. The circle indicates the small amount of SFRP2 protein detected in the extracts. (B) The dermomyotome-inducing activity of WNT1-expressing RHW1 cells is compromised by SFRP2-expressing cells (bracketed area). The experimental set-up is shown diagrammatically at the top. Two pieces of psm were aligned side by side as a square-shaped responding tissue. RHW1 or parental RatB1a cells were cut into a long strip of similar length to the explants and placed to the left of the explants. Their dermomyotome-inducing activity was challenged by COS cells that did or did not express SFRP2myc placed at the bottom of the explants, diagonal to the RHW1 or RatB1a cells. Cultures were incubated for 24 hours prior to processing for DIG whole-mount in situ hybridization using the dermomyotomal marker Pax3 as a probe.

Fig. 5.

SFRP2 inhibits WNT1’s dermomyotome-inducing activity. (A) Secreted SFRP2 protein was produced by transfected COS cells. Supernatant (supe) and detergent-extracted cell lysate (RIPA) from transfected COS cells (con., transfected with control plasmid without a cDNA insert; Sfrp2myc, transfected with myc-tagged Sfrp2 expression plasmid) were assessed for SFRP2 production by SDS-PAGE/western blotting. Equivalent fractions of supe and detergent extracts were loaded in each lane to compare the ratio of cell-associated versus secreted SFRP2 protein. 9E10 antibody, which recognizes the myc epitope, was used to detect the myc-tagged SFRP2. The circle indicates the small amount of SFRP2 protein detected in the extracts. (B) The dermomyotome-inducing activity of WNT1-expressing RHW1 cells is compromised by SFRP2-expressing cells (bracketed area). The experimental set-up is shown diagrammatically at the top. Two pieces of psm were aligned side by side as a square-shaped responding tissue. RHW1 or parental RatB1a cells were cut into a long strip of similar length to the explants and placed to the left of the explants. Their dermomyotome-inducing activity was challenged by COS cells that did or did not express SFRP2myc placed at the bottom of the explants, diagonal to the RHW1 or RatB1a cells. Cultures were incubated for 24 hours prior to processing for DIG whole-mount in situ hybridization using the dermomyotomal marker Pax3 as a probe.

When a myc-tagged Sfrp2 cDNA (Rattner et al., 1997) was expressed in COS cells, secreted SFRP2myc protein was detected in the supernatant by the anti-myc antibody 9E10 (Fig. 5A). In contrast to Frzb-1/SFRP3, we observed little SFRP2myc protein in the detergent-extracted fraction. COS cells transfected with Sfrp2myc expression or control plasmid, RatB1a cells, and RatB1a cells expressing WNT1 (RHW1) were prepared as cell-aggregates for assembling the explant culture. RatB1a cells did not induce the psm to display the dermomyotomal program as assessed by expression of Pax3 using whole-mount in situ hybridization (Fig. 5B; Fan et al., 1997). When RHW1 cells were placed diagonally to the control COS cells, Pax3 expression was detected in cells lining the length of RHW1 cells spanning an induction range of approximately 100 μm (N=8). In contrast, when COS cells expressing SFRP2myc were used (Fig. 5B), WNT1’s dermomyotome-inducing activity remained in cells away from the SFRP2 source, but was severely compromised at the junction near the SFRP2-expressing cells (ranging from 50-150 μm, bracketed region; N=6). When a recently characterized dermomyotomal-marker gene c38 (C. S. Lee and C.-M. Fan, unpublished) was assayed under the same experimental conditions, similar results were obtained (not shown).

We have previously shown that WNT1, WNT3A and WNT4 all display dermomyotome-inducing activity (Fan et al., 1997). Here, we tested whether SFRP2 can inhibit all or only selected members of the WNT family. Previously reported Wnt3a and Wnt4 expression vehicles in COS cells did not appear to produce sufficient ‘active’ recombinant proteins to induce high enough levels of marker gene expression to allow analysis by our current whole-mount in situ protocol. We instead turned to the 3T3 cells stably expressing WNT7A, WNT4 and WNT3A, all of which have been shown to be active in mediating kidney tubule induction (Kispert et al., 1998). Identical to the aforementioned experimental set-up, WNT7A-, WNT4- and WNT3A-expressing cells were placed diagonally to control or SFRP2myc-producing COS cells, and Pax3 expression in the psm explants was assayed. Consistent with published results (Fan et al., 1997), both WNT4- and WNT3A-expressing cells could induce the dermomyotomal program in vitro, while the control WNT7A-expressing (N=8, Fig. 6) and parental 3T3 cells (data not shown) could not. In fact, neither WNT7A-expressing nor parental 3T3 cells could induce expression of other dermomyotomal markers such as Sim1, Pax7 and c38 in the psm by RT-PCR assay (not shown). These results indicate that not all WNT family members can induce the dermomyotome program in vitro.

Fig. 6.

SFRP2 inhibits the dermomyotome-inducing activity of WNT4 (bracketed area) but not that of WNT3A. The same experimental set-up described in Fig. 5B was used to test the selectivity of SFRP2 to various WNTs. WNT7A (3T3-WNT7A), WNT4 (3T3-WNT4) and WNT3A (3T3-WNT3A)-expressing 3T3 cell aggregates were placed to the left of the explants. Their dermomyotome-inducing activity was challenged by COS cells that did or did not express SFRP2myc placed at the bottom of the explants, diagonal to the WNTs-expressing cells. 3T3-WNT3A 1/9 cell clumps used as the WNT3A source contained a mixture of one part WNT3A-expressing cells and eight parts parental 3T3 cells. Cultures were incubated for 24 hours prior to processing for DIG whole-mount in situ hybridization using the dermomyotomal marker Pax3 as a probe. Abbreviations are as in Fig. 5.

Fig. 6.

SFRP2 inhibits the dermomyotome-inducing activity of WNT4 (bracketed area) but not that of WNT3A. The same experimental set-up described in Fig. 5B was used to test the selectivity of SFRP2 to various WNTs. WNT7A (3T3-WNT7A), WNT4 (3T3-WNT4) and WNT3A (3T3-WNT3A)-expressing 3T3 cell aggregates were placed to the left of the explants. Their dermomyotome-inducing activity was challenged by COS cells that did or did not express SFRP2myc placed at the bottom of the explants, diagonal to the WNTs-expressing cells. 3T3-WNT3A 1/9 cell clumps used as the WNT3A source contained a mixture of one part WNT3A-expressing cells and eight parts parental 3T3 cells. Cultures were incubated for 24 hours prior to processing for DIG whole-mount in situ hybridization using the dermomyotomal marker Pax3 as a probe. Abbreviations are as in Fig. 5.

WNT4-expressing 3T3 cells exerted on average a 50 μm range of dermomyotome inducing activity. Importantly, its inducing activity was compromised next to the SFRP2-expressing COS cells (ranging from 30-80 μm; N=8; Fig. 6, bracketed area). Surprisingly, SFRP2 was not able to inhibit the dermomyotome-inducing activity of WNT3A-expressing cells (N=6, Fig. 6). Based on the distance of induction (over 200 μm), WNT3A-expressing cells displayed stronger activity than WNT1- and WNT4-expressing cells. It was therefore possible that these cells produced more WNT3A protein than the COS cell-derived SFRP2 could counteract in this assay system. To investigate this possibility, we tested cell aggregates with different ratios of WNT3A-expressing and parental 3T3 cells to reduce the level of WNT3A source. At 1/9 dilution, the mixed cell clumps induced the dermomyotome/Pax3 expression at a range between 50 and 100 μm, albeit very weakly. Still, no inhibition of WNT3A activity by SFRP2 could be observed (N=6, Fig. 6), indicating that SFRP2 selectively inhibits WNT1 and WNT4 but not WNT3A.

To date, the competitive two-gradient model between SHH-N and WNTs to subdivide the somitic field into sclerotome and dermomyotome is the most straightforward one (as proposed by Fan et al., 1997; Marcelle et al., 1997). However, data presented here suggest that in addition to inducing the transcription factor Pax1 to mediate sclerotome differentiation (Fan and Tessier-Lavigne, 1994) and upregulating its receptor Ptc1 to boost its own signal (Goodrich et al., 1996; this work), SHH-N also coordinately upregulates SFRP2 to exclude WNT1/4 activity in the ventral somites. Since SFRP2 does not inhibit all dorsally derived WNT proteins (e.g. WNT3A), it is unlikely that SFRP2 functions to block all WNT activities. It is more likely that SFRP2 in the sclerotome functions selectively to modulate the types and the concentrations of the WNT proteins present in the somitic mesoderm as a secondary step of SHH-N-mediated sclerotome specification and/or maintenance.

Among the four known SFRPs, the SFRP3 orthologue Frzb-1 is the first to be shown to bind and inhibit the function of WNT1 and WNT8 (Finch et al., 1997; Leyns et al., 1997; Wang et al., 1997a), and its CRD domain is sufficient for binding (Lin et al., 1997). There is also evidence that Frzb-1/SFRP3 functions to inhibit only some of the WNT proteins but not others (including WNT3A) (Wang et al., 1997b). Moreover, Frzb-1/SFRP3 can associate with WNT5A but does not inhibit its function, indicating that physical association by no means correlates with antagonistic action (Lin et al., 1997). Similarly, our function-blocking data provide evidence that SFRP2’s inhibitory effect in the somitic mesoderm is selective to WNT1 and WNT4, but not WNT3A. Although we do not know the precise binding partnership between SFRP2 and the WNT proteins, our data strongly suggest that SFRP2 associates with WNT1 and WNT4. In fact, physical association between SFRP2 and WNT4 has been demonstrated (Lescher et al., 1998), supporting the functional data presented here. To date, physical association between SFRP2 and WNT1 has not been demonstrated in vitro. It is possible that psm and somites provide an auxilliary factor that facilitates the binding between SFRP2 and WNT1 and is not present in the cultured cell lines. Although it would be of interest to document the binding relationships between SFRP2 and all WNT proteins, the precedent of non-functional association between SFRP3 and WNT5A indicates that such data may not necessarily define the inhibitory targets of SFRP2.

It is intriguing that SFRP2 inhibits WNT1 but not WNT3A in our in vitro assay, while in vivo the two WNTs appear to have redundant patterning functions in the CNS (Ikeya et al., 1997) and in the somitic mesoderm (Ikeya and Takada, 1998). It is possible that a yet to be identified SFRP(s) or a homolog of WIF-1 (Hsieh et al., 1999) is also upregulated by SHH-N in the sclerotome and functions specifically to inhibit WNT3A and possibly other WNTs. Together with SFRP2, they may inhibit all WNT function in the ventral somite. However, it should be noted that transcripts of the newly identified mouse Sfrp5 gene (Chang et al., 1999) cannot be detected in the psm of E9.5 embryos even with 35 cycles of RT-PCR (C.-M. Fan, unpublished). It is equally possible that WNT1 and WNT3A do have subtle functional differences in vivo and the selective inhibitory effect of SFRP2 can help refine the ratios of various WNTs in the somitic environment to specify sub-cell types within the somites (e.g. intercostal versus intervertebral muscles, ribs and vertebrae, etc). It is also possible that SFRPs at times function as carrier proteins for WNTs for long-range diffusion or as protector proteins to prolong the half-life of WNTs until they encounter their receptors. However, in our assay, we did not observe any potentiation of the range of induction of WNT1, WNT4 or WNT3A in the presence of SFRP2. Although Wnt1 and Wnt4 single mutants have no reported somitic phenotype and the information regarding the double mutant is not available, it remains possible that Sfrp2 mutant mice may display a sclerotome-related phenotype.

Our interpretation of SFRP2 function in the somitic mesoderm is based on its in vivo RNA expression pattern, its in vitro regulated expression and its in vitro overexpression activity. Although the SFRP2myc expressed by COS cells is soluble and abundant, the biochemical properties and the expression level of SFRP2 in vivo are unknown. Nor do we know of the diffusibility and availability of SHH-N and each WNT member in the somitic field. In fact, compared to its expression in the neural tube and mesonepheric mesenchyme, Sfrp2 expression in the somites is not at very high levels. Thus, we propose that SFRP2 only functions locally in the sclerotome where it is expressed to exclude WNT1/4 function as a secondary safety net/back-up mechanism employed by SHH-N. The moderate levels of Sfrp2 expression in the sclerotome may be necessary to prevent disruption of the dermomyotome, which is located immediately adjacent to the sclerotome.

It is to be noted that Sfrp1 and Sfrp3 are also expressed in the developing somite in dynamic fashions. For example, Sfrp1 is detected in the psm and the lateral dermomyotome (data not shown), and Sfrp3 in the psm and neural crest, albeit transiently (Hoang et al., 1998). We have evidence that both of these genes are upregulated by BMP4 and this induction can be blocked by Noggin in our in vitro explant assay (data not shown). Although Sfrp1 is unexpectedly also induced by SHH-N, this induction appears to be regulated independently from the BMP4 pathway, as only an additive effect was observed when the two factors were applied together (data not shown). Intriguingly, overexpression of SFRP3 in the mouse causes myogenic defects without affecting dermomyotomal marker Pax3 expression (Borello et al., 1999). Whether BMPs utilize SFRP3 and possibly SFRP1 to antagonize WNT function in the presomitic cells emerging from the primitive streak, to prevent premature myogenesis in the psm, has yet to be tested. It is possible that the newly identified WIF-1 (Hsieh et al., 1999) also plays a role in somite patterning in addition to its role in somitogenesis. Interestingly, at the early steps of somite differentiation, SFRP2 appears to be employed by SHH-N to modulate WNT function in the sclerotome, while SFRP3 has been shown to be expressed in the chondrium and can induce cartilage when ectopically applied (Finch et al., 1997). It is tempting to speculate that the chondrogenic lineage utilizes SFRPs (e.g. SFRP2 early and SFRP3 late) to reduce or inhibit WNTs (at least a selected few) for proper differentiation. Whether SFRP3 is utilized by members of the extended TGF-β family (e.g. BMPs) or IHH to modulate WNT activity during chondrogenesis remains to be tested.

Since SHH-N may increase SFRP2 levels to reduce WNT1/4 function, one may speculate that WNTs downregulate Sfrp2 expression in the dermomyotome to assure their own activity dorsally. However, this negative transcriptional regulation has not been observed in vitro (data not shown), consistent with the observed low levels of Sfrp2 expression in the dermomyotome in vivo. It is equally possible that WNTs activate their own receptor(s) to strengthen their signals in the dermomyotome and at the same time also activate a yet to be identified interacting inhibitor of SHH-N to prevent its function in the dermomyotome. Such auto-augmentation and mutual exclusion mechanisms employed by two competitive signals are appealing strategies for stably dividing an unspecified field.

We thank Dr J. Nathans for generously providing the Sfrp plasmids and Dr R. Nusse for the WNT1 expressing cell line. We also thank Drs D. Koshland and A. Spradling for critically reading the manscript. C.-M. Fan is an Alfred P. Sloan Scholar, a Beckman Young Investigator, a John Merck Scholar and a Damon Runyon Scholar.

Bhanot
,
P.
,
Brink
,
M.
,
Harryman Samos
,
C.
,
Hsieh
,
J.-C.
,
Wang
,
Y.
,
Macke
,
J. P.
,
Andrew
,
D.
,
Nathans
,
J.
and
Nusse
,
R.
(
1996
).
A new member of the frizzled family from Drosophila functions as a Wingless receptor
.
Nature
382
,
225
230
.
Borello
,
U.
,
Coletta
,
M.
,
Tajbakhsh
,
S.
,
Leyns
,
L.
,
De Robertis
,
E. M.
,
Buckingham
,
M.
and
Cusso
,
G.
(
1999
).
Transplancental delivery of the Wnt antagonist Frzb1 inhibits development of caudal paraxial mesoderm and skeletal myogenesis in mouse emberyos
.
Development
126
,
4247
4255
.
Capdevila
,
J.
,
Tabin
,
C.
and
Johnson
,
R. L.
(
1998
).
Control of dorsoventral somite patterning by Wnt-1 and beta-catenin
.
Dev. Biol.
193
,
182
194
.
Chang
,
J. T.
,
Esumi
,
N.
,
Moore
,
K.
,
Li
,
Y.
,
Zhang
,
S.
,
Chew
,
C.
,
Goodman
,
B.
,
Rattner
,
A.
,
Moody
,
S.
,
Stetten
,
G.
,
Campochiaro
,
P. A.
and
Zack
,
D. J.
, (
1999
).
Cloning and characterization of a secreted frizzled-related protein that is expressed by the retinal pigment epithelium
.
Hum. Mol. Genet.
8
,
575
583
.
Chiang
,
C.
,
Litingtung
,
Y.
,
Lee
,
E.
,
Young
,
K. E.
,
Corden
,
J. L.
,
Westphal
,
H.
and
Beachy
,
P.
(
1996
).
Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function
.
Nature
383
,
407
413
.
Christ
,
B.
and
Ordahl
,
C. P.
(
1995
).
Early stages of chick somite development
.
Anat. Embryol.
191
,
381
396
.
Chuang
P. T.
and
McMahon
,
A. P.
(
1999
).
Vertebrate Hedgehog signalling modulated by induction of a Hedgehog-binding protein
.
Nature
397
,
617
621
.
Fan
,
C.-M.
,
Lee
,
C. S.
and
Tessier-Lavigne
,
M.
(
1997
).
A role for WNT proteins in induction of dermomyotome
.
Dev. Biol.
191
,
160
165
.
Fan
,
C.-M.
and
Porter
,
J. A.
,
Chiang
,
C.
,
Chang
,
D. T.
,
Beachy
,
P. A.
and
Tessier-Lavigne
,
M.
(
1995
).
Long-range sclerotome induction by Sonic Hedgehog: Direct role of the amino-terminal cleavage product and modulation by the cyclic AMP signaling pathyway
.
Cell
81
,
457
465
.
Fan
,
C.-M.
and
Tessier-Lavigne
,
M.
(
1994
).
Patterning of mammalian somites by surface ectoderm and notochord: Evidence for sclerotome induction by a Hedgehog homolog
.
Cell
79
,
1175
1186
.
Finch
,
P. W.
,
He
,
X.
,
Kelley
,
M. J.
,
Üren
,
A.
,
Schaudies
,
R. P.
,
Popescu
,
N. C.
,
Rudikoff
,
S.
,
Aaronson
,
S. A.
,
Varmus
,
H. E.
and
Rubin
,
J. S.
(
1997
).
Purification and molecular cloning of a secreted, Frizzled-related antagonist of Wnt action
.
Proc. Natl. Acad. Sci. USA
94
,
6770
6775
.
Frohman
M. A.
,
Boyle
,
M.
and
Martin
,
G. R.
(
1990
).
Isolation of the mouse Hox2.9 gene: analysis of embryonic expression suggests that positional information along the anterior-posterior axis is specified by mesoderm
.
Development
110
,
589
607
.
Goodrich
,
L. V.
,
Johnson
,
R. L.
,
Milenkovic
,
L.
,
McMahon
,
J. A.
and
Scott
,
M. P.
(
1996
).
Conservation of the hedgehog/patched signaling pathway from flies to mice: induction of a mouse patched gene by Hedgehog
.
Genes Dev.
10
,
301
312
.
Harlow
,
E.
and
Lane
,
D.
(
1988
).
Antibodies: A Laboratory Manual.
Cold Spring Harbor, New York
:
Cold Spring Harbor Laboratory Press
.
He
,
X.
,
Saint-Jeannet
,
J. P.
,
Wang
,
Y.
,
Nathans
,
J.
,
Dawid
,
I.
and
Varmus
,
H.
(
1997
).
A member of the frizzled protein family mediating axis induction by Wnt5a
.
Science
275
,
1652
1654
.
Hoang
,
B. H.
,
Thomas
,
J. T.
,
Abdul-Karim
,
F. W.
,
Correia
,
K. M.
,
Conlon
,
R. A.
,
Luyten
F. P.
and
Ballock
,
R. T.
(
1998
).
Expression pattern of two frizzled-related genes, Frzb-1 and Sfrp-1, during mouse embryogenesis suggests a role for modulating action of Wnt family members
.
Dev. Dyn.
212
,
364
372
.
Hsieh
,
J. C.
,
Kodjabachian
,
L
,
Rebbert
,
M. L.
,
Rattner
,
A.
,
Smallwood
,
P. M.
,
Samos
,
C. H.
,
Nusse
,
R.
,
Dawid
,
I. B.
and
Nathans
,
J.
(
1999
).
A new secreted protein that binds to Wnt proteins and inhibits their activities
.
Nature
398
,
431
436
.
Ikeya
,
M.
,
Lee
,
S. M.
,
Johnson
,
J. E.
,
McMahon
,
A. P.
and
Takada
,
S.
(
1997
).
Wnt signaling required for expansion of neural crest and CNS progenitors
.
Nature
389
,
966
970
.
Ikeya
,
M.
and
Takada
,
S.
(
1998
).
Wnt signaling from the dorsal neural tube is required for the formation of the medial dermomyotome
.
Development
125
,
4969
4976
.
Johnson
,
R. L.
,
Laufer
,
E.
,
Riddle
,
R. D.
and
Tabin
,
C. J.
(
1994
).
Ectopic expression of Sonic Hedgehog alters dorsal ventral patterning of the somites
.
Cell
79
,
1165
1174
.
Keynes
,
R. J.
and
Stern
,
C. D.
(
1988
).
Mechanisms of vertebrate segmentation
.
Development
103
,
413
429
.
Kispert
,
A.
,
Vaninio
,
S.
and
McMahon
,
A. P.
(
1998
).
Wnt-4 is a mesenchymal signal for epithelial transformation of metanepheric mesenchyme in the developing kidney
.
Development
125
,
4225
4234
.
Leimeister
,
C.
,
Bach
,
A.
and
Gessler
,
M.
(
1998
).
Developmental expression patterns of mouse sFRP genes encoding members of the secreted frizzled related protein family
.
Mech. Dev
.
75
,
29
42
.
Lescher
B.
,
Haenig
,
B.
and
Kispert
,
A.
(
1998
).
sFRP-2 is a target of the Wnt-4 signaling pathway in the developing metanepheric kidney. Dev. Dyn.
213
,
440
451
.
Leyns
,
L.
,
Bouwmeester
,
T.
,
Kim
,
S.-H.
,
Piccolo
,
S.
and
De Robertis
,
E. M.
(
1997
).
Frzb-1 is a secreted antagonist of Wnt signaling expressed in the Spemann organizer
.
Cell
88
,
747
756
.
Lin
,
K.
,
Wang
,
S.
,
Julius
,
M. A.
,
Kitajewski
,
J.
,
Moos
Jr.,
M.
and
Luyten
,
F. P.
(
1997
).
The cysteine-rich frizzled domain of Frzb-1 is required and sufficient for modulation of Wnt signaling
.
Proc. Natl. Acad. Sci. USA
94
,
11196
11200
.
Marcelle
,
C.
,
Stark
,
M. R.
and
Bronner-Fraser
,
M.
(
1997
).
Coordinate actions of BMPs, Wnts, Shh and Noggin mediate patterning of the dorsal somite
.
Development
124
,
3955
3963
.
Mayr
,
T.
,
Deutsch
,
U.
,
Kühl
,
M.
,
Drexler
,
H. C. A.
,
Lottspeich
,
F.
,
Deutzmann
,
R.
,
Wedlich
,
D.
and
Risau
,
W.
(
1997
).
Fritz: a secreted frizzled-related protein that inhibits Wnt activity
.
Mech. Dev.
63
,
109
125
.
McMahon
,
J. A.
,
Takada
,
S.
,
Zimmerman
,
L. B.
,
Fan
,
C.-M.
,
Harland
,
R. M.
and
McMahon
,
A. P.
(
1998
).
Noggin-mediated antagonism of BMP signaling is required for growth and patterning of the neural tube and somite
.
Genes Dev
.
12
,
1438
1452
.
Rattner
,
A.
,
Hsieh
,
J.-C.
,
Smallwood
,
P. M.
,
Gilbert
,
D. J.
,
Copeland
,
N. G.
,
Jenkins
,
N. A.
and
Nathans
,
J.
(
1997
).
A family of secreted proteins contains homology to the cysteine-rich ligand-binding domain of frizzled receptors
.
Proc. Natl. Acad. Sci. USA
94
,
2859
2863
.
Slusarski
,
D. C.
,
Corces
,
V. G.
and
Moon
,
R. T.
(
1997
).
Interaction of Wnt and a Frizzled homologue triggers G-protein-linked phosphatidylinositol signaling
.
Nature
390
,
410
413
.
Stone
,
D. M.
,
Hynes
,
M.
,
Armanini
,
M.
,
Swanson
,
T. A.
,
Gu
,
Q.
,
Johnson
,
R. L.
,
Scott
,
M. P.
,
Pennica
,
D.
,
Goddard
,
A.
,
Phillips
,
H.
,
Noll
,
M.
,
Hooper
,
J. E.
,
de Sauvage
,
F.
and
Rosenthal
,
A.
(
1996
).
The tumour-suppressor gene patched encodes a candidate receptor for Sonic hedgehog
.
Nature
384
,
129
134
.
Wang
,
J.
,
Macke
,
J. P.
,
Abella
,
B. S.
,
Andreasson
,
K.
,
Worley
,
P.
,
Gilbert
,
D. J.
,
Copeland
,
N. G.
,
Jenkins
,
N. A.
and
Nathans
,
J.
(
1996
).
A large family of putative transmembrane receptors homologous to the product of the Drosophila tissue polarity gene frizzled
.
J. Biol. Chem.
271
,
4468
4476
.
Wang
,
S.
,
Krinks
,
M.
,
Lin
,
K.
,
Luyten
,
F. P.
and
Moos
Jr.,
M.
(
1997a
).
tFrzb, a secreted protein expressed in the Spemann organizer, binds and inhibits Wnt-8
.
Cell
88
,
757
766
.
Wang
S.
,
Krinks
,
M.
and
Moos
Jr.,
M.
(
1997b
).
Frzb-1, an antagonist of Wnt-1 and Wnt-8, does not block signaling by Wnts-3A, -5A or -11
.
Biochem. Biophys. Res. Commun.
236
,
502
504
.
Wardoz
,
A.
and
Nusse
,
R.
(
1998
).
Mechanisms of Wnt signaling in development
.
Annu. Rev. Cell Dev. Biol.
14
,
59
88
.