Wnts constitute a family of vertebrate genes encoding ligands essential to signaling in early development (Cadigan and Nusse, 1997). These secreted glycoproteins act via members of the frizzled gene family (Bhanot et al., 1996; Yang-Snyder et al., 1996; He et al., 1997) of G-protein-coupled receptors (GPCR) (Liu et al., 1999b, 2001; Ahumada et al., 2002). Signaling downstream of some Frizzled homologs in response to Wnt-1 or Wnt-8 leads to activation of the phosphoprotein Dishevelled, repressing the function of glycogen synthase kinase-3β (GSK-3) activity (Miller et al., 1999; Polakis, 2000). In the absence of Wnt, GSK-3 phosphorylates β-catenin, reducing its stability and abundance. Wnt signaling represses GSK-3 activity, and β-catenin accumulates in the nucleus, where it binds to members of the lymphoid-enhancer factor and T-cell factor (Lef/Tcf) classes of architectural high mobility group box transcription factors to activate genes involved in early development. Wnt-5A, in contrast, does not alter β-catenin levels in some cellular contexts, including mouse F9 cells; rather, it stimulates phosphodiesterase, decreases intracellular [cyclic GMP], increases intracellular [Ca²⁺], and activates Ca²⁺/calmodulin-dependent protein kinase II, calcineurin, and other downstream effectors (Kuhl et al., 2000a).

Frizzleds display seven hydrophobic, membrane-spanning segments, characteristic of members of the superfamily of GPCRs (Liu et al., 1999b, 2001; Malbon et al., 2001; Ahumada et al., 2002). Analysis of Wnt-sensitive pathways has been hindered by the lack of purified, biologically active Wnts and the need to make use of conditioned media from cells secreting Wnts. To overcome this obstacle, we constructed a chimeric receptor with the ligand-binding and transmembrane segments from the β2-adrenergic receptor (β₂AR) and the cytoplasmic domains from rat Frizzled-1 (Rfz1) and Frizzled-2 (Rfz2). Stimulation of mouse F9 clones expressing the Rfz1 chimera (β₂AR/Rfz1) with the β-agonist isoproterenol stimulates the canonical features of Wnts on β-catenin [e.g., stabilization of β-catenin and activation of β-catenin-sensitive transcription factors (such as Lef/Tcf) and formation of primitive endoderm (PE)] in the same manner stimulated in cells expressing the Rfz1 by use of conditioned media containing Wnt-8 (Liu et al., 2001).

Stimulation of the β₂AR/Rfz2 chimera, in contrast to the stimulation of the Rfz1 chimera, results in increased intracellular [Ca²⁺] and decreased intracellular concentrations of cyclic GMP, in the same manner stimulated in cells expressing the Rfz2 by use of conditioned media containing Wnt-5A (Ahumada et al., 2002). Stimulation of the Rfz1 chimera does not lead to changes in intracellular levels of Ca²⁺ or cyclic GMP, whereas stimulation of the Rfz2 does not lead to appreciable stabilization of β-catenin and activation of the Lef/Tcf-mediated transcription (Liu et al., 1999b; Kuhl et al., 2000b; Wang and Malbon, 2003). Thus, the ability of each chimera to stimulate downstream signaling mimics with high fidelity the effects of the Wnts mediated through authentic, wild-type Frizzled-1 or Frizzled-2. Having validated the use of the chimeric receptors and avoided the complications of using media conditioned with other hormones, growth factors, etc., we were well positioned to perform gene profiling to address fundamental questions in Wnt-Frizzled signaling. Mammals display 18 or more separate genes for Wnts as well as for Frizzleds, and there exists overlap in the ability of several Wnts to activate more than one Frizzled. Making use of these novel Rfz1 and Rfz2 chimeras, however, enabled us to pose the question: what genes are activated in response to stimulation of a single specific Frizzled? We focus on activation of two well known members of the Frizzled family, making use of β₂AR/Rfz1 and β₂AR/Rfz2 chimera to specifically activate downstream pathways of gene activation of Frizzled-1 and Frizzled-2, respectively. Employing DNA microarrays, real-time PCR, and protein expression analyses, we scanned potential target genes in response to the activation of Frizzled-1 versus Frizzled-2 in totipotent, mouse F9 teratocarcinoma cells in culture. Uncomplicated by the use of conditioned media containing Wnts and other growth factors as well as the ability of purified Wnts to activate multiple Frizzleds, this is the first study of gene profiling after activation of a single Frizzled. We report on gene clusters that are activated only by stimulation of Rfz1 chimera, genes activated only by stimulation of Rfz2, and some genes regulated downstream of activation of either Rfz1 or Rfz2.

Materials and Methods

F9 Cell Culture and Stably Transfected Clones. The mouse F9 teratocarcinoma cells were obtained from the American Type Culture Collection (Manassas, VA), propagated, and stably transfected using LipofectAMINE (Invitrogen, Carlsbad, CA). The cells were transfected with pCDNA3 expression vector (Invitrogen) alone (empty vector) or pCDNA3 harboring either β₂AR/Rfz1 (hereafter “Rfz1 chimera”) or β₂AR/Rfz2 (hereafter “Rfz2 chimera) (Liu et al., 2001). Some 10 to 20 independent clones that were resistant to G-418 were propagated in each transfection. The expression of the mRNA for each of the target proteins was measured indirectly via RT-PCR. The clones displaying the highest level of expression of mRNA for the chimera were employed for the studies. The levels of receptor expressed for the stably transfected clones was determined by equilibrium binding assays of the β₂AR binding domain of the chimera, using the radioligand iodocyanopindolol and intact cells (Liu et al., 2001). The mean values of receptor expression were as follows: Rfz1 chimera, 10.6 ± 2.1 fmol/10⁵ cells, and Rfz2 chimera, 20.6 ± 4.4 fmol/10⁵ cells (Liu et al., 1999a,b). F9 clones stably expressing a Frizzled chimera were induced to PE by the addition of 10 μM isoproterenol (Sigma, St. Louis, MO) and continued growth for 4 to 5 days. This concentration of isoproterenol was shown to yield maximal activation of Frizzled-1 and Frizzled-2 chimera in downstream read-outs (Liu et al., 1999a,b). In some cases, F9 cells were transiently transfected with the dominant-negative interfering mutant form of Tcf4 (DN-Tcf4) as detailed elsewhere (Kolligs et al., 1999; Liu et al., 2002). Activation of Rfz1, Rfz2, or their respective chimeras does not influence intracellular concentrations of cyclic AMP. Incubation media was supplemented, however, with the protein kinase A inhibitor KT-5720 (1 μM) to suppress any signal from the small complement of endogenous β₂-adrenergic receptors expressed by F9 cells.

Affymetrix Oligonucleotide Array. Total cellular RNA was isolated with RNA STAT-60 (Tel-Test, Inc., Friendswood, TX) from either β₂AR/RFz-1 or β₂AR/Rfz2 stably transfected F-9 cells at times 0, 3.5, 6, 16, and 45 h after 10 μM isoproterenol treatment as described previously (Liu et al., 2001). Subsequent RNA processing procedures and Affymetrix Murine Genome Array U74Av2 application followed according to the manufacturer's protocol (Affymetrix, Santa Clara, CA) by the University DNA Microarray Facility (Stony Brook, NY). The data sheet for Affymetrix Murine Genome Array for the U74Av2 chips can be found at http://www.affymetrix.com/products/arrays/specific/mgu74.affx. Data were analyzed using Affymetrix Genechip and GeneSpring software. High quality reproducible gene profile was facilitated by the analysis of time-courses from cells in which either Rfz1 or Rfz2 chimera were stimulated and then followed over time. The time points selected for analysis in the time-courses were chosen to make optimal the detection of the genes regulated by either of the Frizzled pathways. The entire gene profiling time course was performed twice using different clones on separate occasions to create a list of likely gene targets for validation by real time quantitative polymerase chain reaction (QPCR). The analysis of the data was performed at the University DNA Microarray Center (http://www.osa.sunysb.edu/udmf/) and with the help of the Bioinformatics Core Facility (http://www.osa.sunysb.edu/bioinformatics/index.html) at SUNY-Stony Brook. Software employed in the analyses included the GeneSuite software of Affymetrix and the GeneSpring software.

Real-Time Quantitative PCR. Real-time quantitative reverse transcription-PCR amplification was performed on the DNA Engine Opticon continuous fluorescence detection system (MJ Research Inc., Boston, MA) by monitoring the increase of fluorescence produced by the incorporation of SYBR green to the double-stranded DNA. The melting curve analysis was performed at the end of each PCR reaction to ensure that there was only one specific product. The PCR primers were designed with Primer3 (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). The first-strand cDNA was synthesized from 5 μg of total RNA using random hexamers with Superscript II reverse transcriptase (Invitrogen) and diluted to 0.3 ml of water. For a 20-μl PCR, 8 μl of cDNA template was mixed with forward and reverse primers (6.25 pmol of each primer for each reaction) and 2× SYBR Green PCR Master Mix (QIAGEN, Valencia, CA). The reaction was run at the setting program (40 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min). Each cDNA sample was performed in duplicate gene-specific PCR. The gene specific primers were detailed in Table 1. For quantification purpose, the relative percentage changes were normalized in each sample against the expression of a gene that is not regulated by development, cyclophilin A. Typically, the target genes identified by Affymetrix-based gene profiling of the time-courses were then subjected to validation by employing real-time PCR analysis of three separate, detailed time-courses of wild-type F9 cells (or cells harboring empty expression vector), as well as of those clones expressing either the Rfz1 or the Rfz2 chimera. The graphs shown represent the mean values of the data from the three individual real-time PCR analyses. Variance and statistical significance were analyzed using SRS PRISMA software (Lion Biosciences, Cambridge, MS). The variance observed among the three separate real-time PCR determinations routinely was less than 20% of the mean values. The S.E.M. bars are not included on the graphs because of the dense, data-rich nature of the time-courses displayed for the real-time PCR analyses. The real-time PCR proved invaluable in the validation of the gene chip data. As will be noted, some genes highlighted in the DNA microarray were not validated as targets for Rfz1 or Rfz2-mediated pathways and consequently were deleted from further studies at the protein and function levels.

Antisense Oligodeoxynucleotides and Transfection Studies. Phosphorothioate-modified, high-performance liquid chromatography-purified oligodeoxynucleotides (ODNs) were obtained from QIAGEN Operon (Alameda, CA). The sequences were detailed in Table 2. The F9 clones expressing the β₂AR/Rfz1 chimera were propagated on 12- or 24-well plates and allowed to attach overnight. The clones were treated with antisense or mis-sense ODNs 48 to 72 h in advance of challenge with 10 μM isoproterenol.

Western Blot Analysis. For immunoblot analysis, the following polyclonal antibodies were used: rabbit polyclonal anti-ephrin A3, goat polyclonal anti-Nip3, goat polyclonal anti-JAB, rabbit polyclonal anti-Msx-1, rabbit polyclonal anti-Lefty, goat polyclonal anti-STAM, rabbit polyclonal anti-Msx-2, goat polyclonal anti-REST/NRSF, and rabbit polyclonal anti-Erk 1 (Santa Cruz Biochemicals, Santa Cruz, CA). Rabbit polyclonal MyD118 was kindly provided by Dr. Dan Liebermann (Temple University, Philadelphia, PA). Cell extracts were denatured, separated by SDS-PAGE, and electroblotted onto nitrocellulose membranes. After blocking, the blots were incubated with the primary antibody for 2 h, washed, incubated with secondary, horseradish peroxidase-coupled antibody of choice for 1 h, washed again, and then visualized by enhanced chemiluminescence. Equal loading was demonstrated by reprobing the membrane with anti-β-tubulin monoclonal antibody (Sigma).

Results

Cellular RNA was isolated from F9 clones stably expressing either β₂AR/Rfz1 or β₂AR/Rfz2. Clones were stimulated with the β-agonist isoproterenol (10 μM) for 0 to 45 h and subjected to gene profiling (Table 3). Stimulation of clones expressing β₂AR/Rfz1 chimera with β-agonist provokes the formation of primitive endoderm. This PE formation is maximally induced with 10 μM β-agonist and specifically blocked by simultaneous treatment with the β-adrenergic antagonist propranolol (10 μM) (Liu et al., 2001). The gene chips included more than 6000 functionally characterized genes of the Mouse UniGene set (build 74) and more than 6000 expressed sequence tag clusters on the murine chip. The approach adopted was to perform DNA microarray from multiple clones in which either the Rfz1 or the Rfz2 pathway was activated and to study the gene profiles from detailed time-courses, rather than from a single time point. The initial times selected for sampling of gene activation were 0, 3.5, 6, 16, and 45 h after stimulation of clones expressing one or the other chimera. Displayed in Table 3 are the names of the regulated genes, their accession numbers, the maximal values of change (-fold increase), and the times at which peak change occurred for each gene. Fewer than 30 genes were activated at >1.5-fold over unstimulated clones, in response to activation of either Rfz1 or Rfz2 chimeras. Based upon the DNA microarray data, several of the target genes were found to be activated only by β₂AR/Rfz1 chimera (e.g., Dvl-2) or by β₂AR/Rfz2 chimera (e.g., MmZw10), whereas other genes (e.g., Lefty1, MyD118, and STAM) were activated by stimulation of either Rfz1 or Rfz2 chimera. We validated putative target genes using QPCR employing cellular RNA isolated from clones stimulated without or with β-agonist for 1 to 45 h (Table 4). Stimulation of either Frizzled chimera activated Msx-1, Msx-2, CBP/P300-associated factor, ephrin A3, and Nip3 genes. Follistatin, Lefty1, and STAM were early targets of gene activation by Rfz1, whereas REST/NRSF, Ubc4/5E2, nucleophosmin, and Groucho were early targets for gene activation by Rfz2.

The results of QPCR confirmed most but not all (e.g., Dishevelled-2 and sFRP-1) of the target genes identified by DNA microarray and extended our description of the temporal up-regulation of these RNAs (Fig. 1A). We grouped those genes whose expression was identified first by DNA microarray and later confirmed independently by QPCR analysis of multiple, separate time-courses into clusters of genes activated by Rfz1 chimera only (Fig. 1A), by Rfz2 chimera only (Fig. 1D), and by both Rfz1 and Rfz2 chimeras (Fig. 1G). To facilitate data presentation, the traces are grouped temporally and displayed in one of three different colors. RNA for the activin-binding protein follistatin, the TGFβ-related Lefty1 gene product implicated in early embryonic patterning in vertebrates, and signal-transducing adaptor molecule of Janus kinase STAM RNAs increased 1- to 3-fold within 1 to 2 h of Rfz1 activation and declined thereafter within 4 h (Fig. 1A, red traces). Lefty1 and STAM RNA levels increased later; Lefty1 RNA reached peak levels (∼9-fold over basal) by 30 h after stimulation of Rfz1 chimera. RNAs for Janus kinase-binding protein JAB (SOCS1), and IgA receptor Fcer Ig, in contrast, displayed a later onset at 2 to 5 h after Rfz1 stimulation, declining at the 5- to 10-h period, maintaining nearly 2-fold elevation to 30 h (Fig. 1A, green traces). The mitogen-activated protein kinase Erk 1 and growth suppressor/pro-apoptotic MyD118 RNAs, in contrast, peaked at 20 h after Rfz1 chimera stimulation (Fig. 1A, blue traces). Treating the clones expressing the Rfz chimera with the β-adrenergic antagonist propranolol (10 μM) blocked activation observed in response to isoproterenol of the following genes studied: CBP/P300-associated factor, ephrin A3, Lefty1, msx2, MyD118, and Nip3. Thus, we demonstrated clusters of genes that were being activated in response to activation of the Rfz1 pathway in these mouse F9 cells.

Several controls were employed throughout these studies. Studies with wild-type F9 cells were included to detect any gene expression not attributable to the activation of either the Rfz1 chimera (Fig. 1B) or the Rfz2 chimera (Fig. 1E). Activation of these target genes by wild-type F9 cells in response to isoproterenol was not observed. The inactivation of several of these genes noted in the wild-type cells in response to isoproterenol (Fig. 1, B and E) was sensitive to treatment with the antagonist propranolol (data not shown). The results from multiple experiments reveal no significant activation of the genes stimulated in clones expressing the Rfz1 chimera (Fig. 1A) in the similarly treated wild-type F9 cells (Fig. 1B). In addition, we performed QPCR of the Rfz1-sensitive genes in the clones expressing the Rfz2 chimera (Fig. 1C). Confirming the results from the DNA microarray (Table 3), the QPCR data revealed no significant activation of the Rfz1-sensitive genes after activation of the Rfz2 chimera (Fig. 1C). Thus, good agreement exists between the data from the time-courses analyzed by DNA microarray and independently from the studies employing QPCR.

Activation of Rfz2 chimera caused short-term stimulation of RNA levels of the zinc-finger gene-specific REST/NRSF; the Groucho corepressor that mediates negative transcriptional regulation in association with DNA-binding proteins in development; the ubiquitin-conjugating enzyme (E2) homologous to the yeast UBC4 Ubc4/5E2; and the abundant nucleolar phosphoprotein nucleophosmin genes (Fig. 1D, red traces) with RNA levels declining below basal within 20 to 30 h (Fig. 1D). The activation of the Wnt-Ca²⁺ pathway by the Rfz2 chimera in response to isoproterenol was sensitive to antagonism with 10 μM propranolol (Liu et al., 1999b, 2001; Ahumada et al., 2002). Likewise, treating these clones with propranolol blocked the activation of gene expression stimulated by isoproterenol (data not shown). Analysis of these Rfz2-sensitive genes in wild-type F9 cells revealed no significant activation of these target genes by isoproterenol (Fig. 1E). Similarly, analyses by QPCR of the genes activated by Rfz2 were performed in clones expressing the Rfz1 (Fig. 1F). The activation of Rfz1, which led to enhanced expression of one set of genes subjected to analysis by DNA microarray and QPCR (Table 3 and Fig. 1A), did not lead to any significant activation of the genes stimulated in response to activation of Rfz2 chimera (Fig. 1F).

Interestingly, RNA levels of some target genes for activation in response to either Rfz1 or Rfz2 stimulation (e.g., homeobox-containing genes in development Msx-1, Msx-2, CBP/P300-associated factor, the ephrin A3 ligand for the Eph tyrosine kinase, as well as E1B 19K/Bcl-2-binding protein Nip-3) displayed similarities in their time-courses (Fig. 1G). Although Rfz1 and Rfz2 display highly homologous (∼85% identity) cytoplasmic sequences (Wang et al., 1996), they activate both unique and common sets of genes. Analysis in wild-type F9 cells of this subset of genes regulated in response to stimulation of either Rfz1 or Rfz2 revealed no significant gene activation (Fig. 1H).

Although QPCR studies enabled validation of a major subset of the Rfz-sensitive genes identified by DNA microarray and gene profiling, further validation of regulated expression of the target proteins was crucial to detailed analysis of Rfz-stimulated genes. At the protein level, immunoblotting was employed to validate the QPCR data. Immunoblots of whole-cell extracts prepared from clones expressing the Rfz1 chimera activated with isoproterenol confirmed Rfz1-stimulated expression of key genes (Fig. 2). Activation of Rfz1 chimera led to increases in the several gene products: JAB, the coactivator of activator protein-1 transcription factor central to angiogenesis (Endo et al., 1997); and MyD118, a gene product involved in negative growth control, JNK suppression, and apoptosis (Abdollahi et al., 1991) (Fig. 2A). STAM, the signal-transducing adaptor in Janus tyrosine kinase-signal transducer and activator of transcription signaling and Lefty1, a TGFβ-related gene of early embryonic patterning were likewise up-regulated in response to activation of the Rfz1 chimera (not shown). The gene for the mitogen-activated protein kinase Erk 1 displayed activation and increased mRNA expression in response to stimulation of the Rfz1 chimera (Fig. 1A); however, its protein expression remained essentially unchanged for the first 20 h after stimulation (Fig. 2A).

A number of target genes, including REST/NRSF, Groucho, nucleophosmin, Ubc4/5E2, and AA968123, were activated by Rfz2, but not Rfz1. Several of these same genes were reported to be activated in the human teratocarcinoma cell line NCCIT treated with conditioned media prepared from mouse L-cells transfected with the expression vector for Wnt-3A (Willert et al., 2002). At the protein level, Rfz2 activation was found to provoke increased expression of a number of these targets, including the zinc-finger gene-specific repressor element REST/NRSF and the transcriptional corepressor of Wnt-Frizzled-1 signaling, Groucho 1 (Cavallo et al., 1998) (Fig. 2B). Several target genes demonstrated activation in response to either Rfz1 or Rfz2, especially members of the Msx homeobox gene products, Msx-1 and Msx-2, involved in inductive tissue interactions necessary for vertebrate organogenesis (Fig. 2C). Although activation of both Frizzleds activated Msx-1 expression, activation of the Rfz1 chimera stimulated an earlier, more robust response than did activation of the Rfz2 pathway. Members of this group of target genes (e.g., Msx-1, Msx-2, and CBP/P300-associated factor) also were activated by treatment of NCCIT cells with conditioned media containing Wnt-3A (Willert et al., 2002), thus providing good agreement from these DNA microarray analysis and recent studies performed with conditioned media containing authentic Wnt-3A. The genes for Nip3, a proapoptotic mitochondrial protein (Bruick, 2000) and for the ephrin A3 ligand for the Eph receptor tyrosine kinase were shown to be new members targeted by activation of both Rfz1 and Rfz2 chimera (Fig. 1G, Table 4).

We next analyzed the role of several target gene products in the ability of Rfz1 to stimulate early features of development in these F9 embryonic stem cells. The roles of Frizzled-1 activated target genes were investigated in the formation of PE. F9 clones were treated with either antisense or missense (as a control) ODNs targeting the gene product of interest. The ability of the Rfz1 chimera to promote formation of primitive endoderm was determined in the clones made deficient in the target gene (Table 5, Fig. 3). Formation of PE was established by positive staining for the cytokeratin endoA, a hallmark protein for PE, with the TROMA-1 anti-body (Liu et al., 2001). Treatment of the clones with ODN antisense, but not mis-sense, to JAB, Nip3, or MyD118 suppressed the expression of each (not shown) and blocked the ability of the activated Rfz1 to stimulate PE formation (Table 5, Fig. 3). These observations demonstrated an obligatory role of JAB, Nip3, and MyD118 as key genes in the formation of PE in response to stimulation of the Frizzled-1 pathway. Interestingly, for two target genes, Msx-1 and STAM, the treatment with antisense ODNs increased rather than decreased PE formation, suggesting functions of both Msx-1 and STAM as de-differentiation factors in Frizzled-1 signaling.

Activation of the Lef/Tcf-dependent transcriptional response has been shown to be an essential early response for Rfz1 signaling in development (Cadigan and Nusse, 1997; Barker et al., 2000; Liu et al., 2001). We evaluated to what degree, if any, the β-catenin-sensitive Lef/Tcf transcriptional response to Rfz1 regulates the Rfz1-sensitive genes MyD118, STAM, Msx-1, and JAB. F9 clones stably expressing Rfz1 chimera were transiently transfected with an expression vector harboring the dominant-negative form (DN-Tcf4) of Tcf4 (Kolligs et al., 1999). The expression of DN-Tcf4 effectively blocks genes regulated by the Lef/Tcf-sensitive transcriptional pathway and has been shown previously to block Frizzled-1 stimulated PE formation in F9 embryonic stem cells (Liu et al., 2001). Using QPCR, we analyzed the level of mRNA for the targeted genes in clones transiently transfected with either 1 or 2 μg of DN-Tcf4 plasmid DNA (Fig. 4, A and B). Graded expression of DN-Tcf4 was found to produce a graded block of the Rfz1-stimulated activation of MyD118, Nip3, and STAM genes (Fig. 4A). The results justified a closer inspection of the Frizzled-1 sensitive genes for the presence of the DNA motif (A/T)(A/T)CAA(A/T)GG, conserved for Tcfs in vertebrates, Drosophila melanogaster, and the worm Caenorhabditis elegans (Roose and Clevers, 1999). This canonical motif was observed in the follistatin and MyD118 genes, whereas the motif with one 5′ end substitution was found in early genes for STAM and Lefty1, as well as in JAB and ephrin A3. Interestingly, the genes for CBP/P300-associated factor, Msx-1, and Msx-2 display the same DNA motif but were up-regulated by activation of either Rfz1 or Rfz2 chimera. Up-regulation of JAB and Msx-1 by Frizzled-1, in contrast, was shown to be insensitive to DN-Tcf4 expression (Fig. 4B), revealing novel Lef/Tcf/β-catenin-insensitive elements of Frizzled-1 signaling.

Discussion

Gene profiling of the Wnt-Frizzled signaling pathways is essential if we are to more fully understand how these pathways control cell polarity, embryonic induction, specificity of cell fate, and diseases, such as cancer, attributed to lesions in one or more of the Wnt-Frizzled pathways (Wang and Malbon, 2003). More than 20 Wnt ligands are expressed during mammalian development. The overarching goal of these studies was to dissect out Frizzled-specific pathways, using a strategy that obviates the ability of Wnts to activate multiple Frizzleds, a gene family that itself has more than 18 members in mammals. We made use of the chimeric GPCRs composed of the transmembrane and exofacial domains of the β₂AR to which Frizzled cytoplasmic domains were substituted, creating functional β₂AR/Rfz1 and β₂AR/Rfz2 chimeras. These unique studies provide the first insights into transcriptional responses to Frizzled-specific signaling GPCRs that mediate key aspects of early vertebrate development (Malbon et al., 2001). First and perhaps most importantly, we show both common as well as unique patterns of transcriptional responses after activation individually of Rfz1 and of Rfz2, highly homologous GPCRs (Bhanot et al., 1996). The transcription factors REST/NRSF, Groucho, and Ubc4/5E2 were shown previously to be up-regulated by treating NCCIT human embryonic carcinoma cells with conditioned media containing Wnt-3A (Willert et al., 2002) and by activation of Rfz2 chimera, but not by Rfz1 (current study). The use of conditioned media containing Wnt-3A to stimulate cells for gene profiling is fraught with limitations (Willert et al., 2002), but it is interesting to note the overlaps between the gene profiling reported by others and the data derived from the activation of a single Frizzled pathway. Most of the genes noted above have been linked in aspects of early development, although not specifically Frizzled-1 or Frizzled-2 signaling. The strategy that was employed (i.e., creating β₂AR/Frizzled chimeras) enabled precise definition of the target genes regulated by activation of a single Frizzled, avoiding the pitfalls of the use of conditioned media with variable and indeterminate amounts of active Wnts and other growth factors.

Activation of these Frizzled-specific chimeras also revealed early (e.g., Lefty1, JAB, Fce Ig, STAM, and follistatin) and intermediate-to-late (e.g., MyD118, p44 MAPK) genes that are up-regulated by activation of the Rfz1-pathway specifically and not by activation of the Rfz2-pathway. These target genes are regulated by neither activation of Rfz2 (Table 4) nor treatment of the NCCIT cells with Wnt-3A (Willert et al., 2002). Lefty1 regulates TGFβ signaling (Ulloa and Tabibzadeh, 2001; Tabibzadeh, 2002), whereas STAM and JAB are key factors in Janus kinase signaling (Duhe and Farrar, 1998). The up-regulation of STAM by activation of the Frizzled-1-specific chimera was blocked by expression of DN-Tcf4, implicating the Wnt/β-catenin pathway in the up-regulation of STAM. MyD118 (GADD45b), involved in negative growth control and apoptosis (Vairapandi et al., 2002), is up-regulated late in the response to Frizzled-1 chimera activation and is also sensitive to expression of DN-Tcf4. Nip3, a Bcl-2-binding protein involved in apoptosis (Bruick, 2000), up-regulated by activation of either Rfz1 or Rfz2, is also sensitive to DN-Tcf4. The gene profiling of Frizzled-1 revealed in the current study enabled the identification of a cluster of genes whose structures include promoter elements sensitive to elevation of cellular levels of β-catenin. The ability of expressed DN-Tcf4 to block Frizzled-1 activation of these genes and to block the formation of primitive endoderm provides compelling evidence for the essential role of activation of this cluster of genes in development.

Finally, although the genes for the Janus tyrosine kinase-signal transducer and activator of transcription regulator JAB and the homeobox-containing Msx-1 genes in development displayed a Tcf-like DNA binding motif and were activated by Frizzled-1-specific chimera, expression of DN-Tcf4 had no effect on their activation, revealing β-catenin-independent signaling of Frizzled-1 at the level of gene expression. Activation of the Frizzled-2 chimera, which was not observed to significantly activate Lef/Tcf-sensitive reporter genes in these mouse F9 cells, led to activation of Msx-1 gene, but with a much later onset than for activation of Frizzled-1.

The value of gene profiling is that it provides a high throughput strategy with which to analyze the regulation of a large number of genes. In the current work, we attacked the problem of gene profiling of the Wnt-Frizzled pathways, which have been largely resistant to study in the absence of active, purified Wnts. Even as the problem of purified Wnts is solved, the question of how to activate a single, specific Frizzled and not a collection of Frizzleds that may share the ability to be activated by one or more Wnts remains daunting. The GPCR chimera strategy employed for these studies provides an invaluable approach toward dissecting the Frizzled-specific pathways, pharmacologically teased one from the other. We have used Frizzled-specific chimera, DNA microarray, real-time PCR and protein analysis in tandem with the Frizzled-1-specific and Frizzled-2-specific chimeric receptors to gene profile Frizzled signaling. A new appreciation for Frizzled-1-specific versus Frizzled-2-specific regulation of gene expression has emerged using this strategy. These studies demonstrate a proof-of-concept that will enable dissection of the individual roles that Frizzleds play in gene expression, development, and human disease.