Characterization and functional analysis of the murine Frat2 gene.

The Frat1 proto-oncogene was first identified as a gene contributing to tumor progression in T-cell lymphomas induced by retroviral insertional mutagenesis with the Moloney murine leukemia virus. The biological function of Frat remained elusive until its Xenopus homologue GBP was isolated as a glycogen synthase kinase 3 (GSK3)-binding protein and was shown to be an essential component of the maternal Wnt-signaling pathway. To date two Frat homologues have been described in the mouse, Frat1 and Frat3. The proteins encoded by these two genes are 84% identical. Here we describe the cloning and characterization of a third murine Frat homologue, Frat2, which is the mouse ortholog of human FRAT2. Frat1 and Frat2 are juxtaposed on chromosome 19 in a chromosomal organization conserved between man and mouse. We show that Frat1 and Frat2 are phosphorylated, which is the first evidence that these proteins are subject to posttranslational modification. Like Frat1, Frat2 is able to bind to GSK3beta. However, a side-by-side comparison of the murine Frat proteins for their capacity to induce signaling through beta-catenin/T-cell factor reveals that Frat2 is a less potent activator of the canonical Wnt pathway. Frat2 protein accumulates to higher levels upon transfection into 293T cells than either Frat1 or Frat3. Thus, whereas Frat1 may be a core component of canonical Wnt-signaling, Frat2 might very well be part of a divergent intracellular GSK3beta pathway.

The Frat1 proto-oncogene was originally isolated in a retroviral insertional mutagenesis screen aimed at the identification of genes that were involved in later stages of T-cell lymphomagenesis (1). After infection of newborn E-Pim1 or H2K-Myc transgenic mice with the Moloney murine leukemia virus, the resulting primary tumors were grafted to syngeneic hosts. When the proviral integration patterns of primary and transplanted tumors were compared up to 30% of the transplanted tumors showed outgrowth of a population of cells with a similar common integration site. Cloning of the gene affected by integration of the provirus into this genomic locus resulted in the identification of Frat1. Although Frat1 was shown to convey a strong selective advantage to lymphoma cells in vivo (1), its biological function remained unknown until its Xenopus homologue GBP was identified as a GSK3 1 -binding protein (2).
GBP was shown to be part of the maternal Wnt pathway in Xenopus, since depletion of the endogenous pool of GBP in the oocyte prevented the formation of a normal body axis in developing embryos. Like GBP, Frat is able to induce secondary axis formation upon ectopic expression in developing Xenopus embryos by stabilizing ␤-catenin levels (2,3). Frat/GBP competes with axin for the same binding site on GSK3␤ (4 -7). It is generally presumed that Frat functions to titrate GSK3␤ away from the scaffolding complex containing APC and axin, thus preventing the phosphorylation and subsequent degradation of ␤-catenin (8). As a result, Frat/GBP is a potent activator of the canonical Wnt pathway.
To learn more about the function of Frat in mammalian development, we have generated Frat1 knockout mice in which most of the Frat1-coding sequence has been replaced by a lacZ marker gene (3). Despite the fact that Frat1 is expressed in a broad range of neural and epithelial tissues during embryonic development as well as in adult animals, Frat1-deficient animals showed no gross abnormalities. At that time we described the existence of Frat3. Because it shares 84% amino acid identity with Frat1, it was hypothesized to exhibit compensatory activity in its absence. Frat3, however, appears to be present as an imprinted gene only in mice and rats due to a relatively recent transposition event (9,10). The human genome instead harbors FRAT2, which is less conserved to FRAT1 (2,11,12).
Here we report the cloning and characterization of the murine Frat2 gene. It bears close resemblance to its human counterpart and is less homologous to Frat1 than Frat3. We have performed an extensive side-by-side analysis of all three murine Frat genes with regard to their ability to induce canonical Wnt signaling in mammalian cells and show that Frat2 is less able to do so in comparison to Frat1 despite the fact that Frat2 protein accumulates to much higher levels. In addition, we show that Frat1 and Frat2 are phosphorylated, which is the first evidence for posttranslational modification of this family of proteins.

MATERIALS AND METHODS
Cloning of Frat2-A Frat2 cDNA clone was identified in the NCBI EST data base (IMAGE clone 318660) and used to design primers. Using Frat2-ForA (5Ј-CGGTAGATCCCAGGTCCTC-3Ј) and Frat2-RevA (5Ј-AGAGACCGGGAACCTTGC-3Ј) Frat2 was then PCR-amplified from murine kidney cDNA. The Frat2 sequence has been deposited in the NCBI data base under GenBank TM accession number AY518895.
Generation of Expression Constructs-The complete Frat1-, Frat2-, and Frat3-coding sequences were cloned in-frame with an N-terminal Myc tag into pGlomyc3.1, which contains 5Ј and 3Ј globin UTRs and the pCDNA1.1 polylinker in the backbone of pCDNA3.1. All cloning procedures were carried out according to standard techniques. Hybrid fusions between Frat1 and Frat2 were cloned by swopping a BssHII/-XbaI fragment, an EcoNI/XbaI fragment, or an Eco47III/XbaI fragment. To introduce an Eco47III site by silent mutation into Frat1, a HindII-I/EcoRI fragment was replaced with an HindIII/EcoRI fragment containing this mutation, which had been generated using the forward 5Ј-GCAAAGCTTCCCGCACACCCGTTCCTCGGGCCTCTGAGCGCTC-CAG-3Ј and reverse 5Ј-GGTGTTCTTGAGGCTGG-3Ј primers. A Frat1 C-terminal deletion mutant was cloned by deleting coding sequences downstream of the NruI site. The integrity of all constructs was verified by restriction enzyme digestion and sequence analysis on an ABI sequencer.
Cell Culture and Transfection-COS7 or 293T cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen) under 5% CO 2 at 37°C in humidifying conditions. On the day before transfection cells were plated in 12-well tissue culture plates. Cells were transfected with a total amount of 500 ng of DNA per well using FuGENE (Roche Applied Science).
Protein Assays-Cells were harvested 48 h after transfection by lysis in radioimmune precipitation assay buffer supplemented with protease inhibitors (Roche Applied Science). Protein concentration was determined using a colorimetric assay (Bio-Rad), and equal amounts of protein were run on a 12% SDS-PAGE gel and analyzed on Western blot using ECL (Pierce). Antibodies were used recognizing the Myc tag (9E10, 1:5000, Invitrogen), GSK3␤ (1:2500, Transduction Laboratories), or ␤-galactosidase (1:2500, Chapel). Secondary antibodies were goat-anti-mouse-horseradish peroxidase (HRP) (1:5000, BIOSOURCE) or swine-anti-rabbit-HRP (1:2000, Dako). For immunoprecipitation, COS7 lysates were incubated at 4°C with a polyclonal antibody directed against GSK3␤ (Santa Cruz). Immunocomplexes were pulled down by incubation with protein G-Sepharose, after which samples were washed in radioimmune precipitation assay buffer to remove unbound protein, resuspended in radioimmune precipitation assay buffer, boiled, and analyzed on gel. For phosphatase treatment, 293T lysates were incubated with the indicated amounts of protein phosphatase (New England Biolabs) at 30°C for 90 min, boiled, and analyzed on gel. For cycloheximide (CHX) experiments, 293T cells were treated for the indicated amounts of time with 100 g/ml cycloheximide (ready made, Sigma) before lysis in radioimmune precipitation assay buffer. Protein levels were evaluated by densitometry following Western blotting.
Luciferase Assay-293T cells were transfected as described above with TOPFLASH, pGlomyc-Frat1, Frat2 or Frat3, human TCF4, and ␤-galactosidase (Clontech). Cells were harvested 48 h after transfection in reporter lysis buffer (Promega) and analyzed with luciferase assay reagent (Promega) according to the manufacturer's instructions in a TopCounter (Packard Instrument Co.). To control for transfection efficiency ␤-galactosidase activity was determined using ortho-nitrophenyl-␤-D-galactopyranoside (Sigma). The same lysates were analyzed on Western blot after the addition of sample buffer and electrophoresis on a 12% SDS-PAGE gel.
Immunofluorescence-Cells were plated on glass chamber slides and fixed in 4% paraformaldehyde at 48 h after transfection. Cells were permeabilized with 0.2% Triton X-100 and incubated with a primary antibody recognizing the Myc tag (1:200, Invitrogen) and a fluorescein isothiocyanate-conjugated secondary antibody (1:500, Molecular Probes).

Cloning and Characterization of the Frat2
Gene-In search of a putative mouse homologue of the human FRAT2 gene, we identified an EST clone in the NCBI data base with a high degree of homology to FRAT2. Using this clone to design primers, we were able to PCR-amplify Frat2 from murine kidney. Comparison of the PCR products from cDNA or genomic clones revealed no difference in size, indicating that the Frat2 gene does not contains introns, similar to murine Frat1 and Frat3 and human FRAT1 and FRAT2. The Frat2 gene (GenBank TM accession number AY518895) contains an open reading frame of 696 nucleotides encoding a polypeptide of 231 amino acids (Fig. 1A). The transcript encoded by Frat2 contains a long 3Ј-UTR with four destabilizing motifs ATTT(A) upstream of the polyadenylation signal (AATAAA). Sequence alignment of the three murine Frat proteins reveals that Frat2 is the most distant family member of the three with 68% amino acid identity to Frat1. By comparison, Frat3 shares 84% identity with Frat1 at the protein level (Fig. 1B). The human and mouse Frat2 homologues are 76% identical at the amino acid level (Fig. 1C), which is comparable with the degree of conservation between mouse Frat1 and human FRAT1. Despite its reduced similarity to the other Frat homologues, Frat2 does contain the conserved domains that are also present in Xenopus and zebra fish GBP. The C-terminal domain contains the conserved IKEA box, which forms the binding site for GSK3␤ in the Frat1 protein and Xenopus GBP (2).
A detailed Ensembl data base search revealed two Fratrelated sequences that mapped to the same sequence contig. Analysis of the genomic organization at this locus revealed that Frat1 and Frat2 lie only 18 kilobases apart on mouse chromosome 19 in opposite transcriptional orientation (Fig. 1D). This genomic organization has recently been confirmed for human FRAT1 and FRAT2, which lie 16 kilobases apart in the syntenic region of human chromosome 10 (13).
To monitor the expression of Frat1, Frat2, and Frat3, we set up a gene-specific RT-PCR (Fig. 1E). Primers for each of the three homologues were designed in the 3Ј-UTR of the genes. No RT control reactions were included for each sample to exclude DNA contamination (results not shown). All three genes are expressed in embryonic stage E12.5-E16.5. Frat1 and Frat2 also show an overlapping expression pattern in tissues from adult FVB mice. In contrast, Frat3 is expressed at much lower levels as well as in a reduced number of tissues, as has been reported previously (10).
Frat2 Binds to GSK3␤-It has previously been shown that GSK3␤ can be co-immunoprecipitated with Frat1 using a polyclonal antibody raised against mouse Frat1 (15). Because the IKEA box, which is required for binding to GSK3␤, is conserved between all Frat family members across species, we expected Frat2 to bind GSK3␤ as well. To test this, we cloned all three murine Frat genes into a mammalian expression vector inframe with an upstream Myc tag. After transient transfection of these constructs into COS7 cells, all three proteins can be readily detected by Western blot analysis. Myc-tagged Frat2 has an apparent molecular mass of around 35 kDa on an SDS-PAGE gel compared with 40 kDa for Myc-tagged Frat1 and Frat3 ( Fig. 2A). To test for binding of Frat2 to GSK3␤, we sought to co-immunoprecipitate Frat2 with GSK3␤. As shown in Fig. 2A, Frat2 can indeed be co-precipitated with GSK3␤ from the same COS7 lysates.
Frat2 Is Phosphorylated-Upon closer analysis Frat2 and Frat3 were found to migrate as a doublet on SDS-PAGE gels. Doublets often represent fractions of the same protein with a differential phosphorylation status. When protein lysates from 293T cells transfected with Frat2 were treated with protein phosphatase, the upper band disappeared, indicating that this band represents phosphorylated Frat2 (Fig. 2B). Because the lower band appeared to be affected by the phosphatase treat-ment as well, the originally observed doublet may have consisted of hypo-and hyperphosphorylated Frat2. Although Frat1 was not seen as a doublet on SDS-PAGE gels, we did observe a change in mobility after treatment with protein phosphatase (Fig. 2B). Thus, both Frat1 and Frat2 are phosphoproteins. Because Frat1 and Frat2 contain three and five sites, respectively, that match the reported consensus motif for GSK3 (X(S/T)XXXS, in which the C-terminal serine must be prephosphorylated (16 -20)), we tested whether this was the kinase responsible for the observed phosphorylation of Frat2. After transfection of 293T cells with Myc-Frat2, the cultures were treated with LiCl (21), a known inhibitor of GSK3 (Fig.  2C). The addition of LiCl resulted in phosphorylation of GSK3 on Ser-9, which is indicative of inhibition of its kinase activity (22), but we did not observe a change in the migration of Myc-Frat2 in cultures treated with LiCl compared with controls that were treated with NaCl. This indicates that, although GSK3 might still phosphorylate Frat, it does not result in the hyperphosphorylated form observed on denaturing gels.
Frat2 Activates the Canonical Wnt Pathway-Given the high degree of homology between the three murine Frat proteins, we compared the ability of all three to induce Wnt signaling through ␤-catenin/TCF in a TOPFLASH reporter assay, which uses luciferase activity as a readout for activation of consensus TCF binding sites cloned upstream of the reporter gene (23). Frat1 has previously been shown to function as an activator of canonical Wnt signaling using this assay (15). As shown in Fig. 3A, we found that Frat2 and Frat3 are also able to activate the TOPFLASH reporter in 293T cells, albeit to a lesser extent than Frat1 (p Ͻ 0.005 for Frat2). However, when we compared the Frat protein levels present in these 293T lysates, we found that Myc-Frat1 and Myc-Frat3 were present at much lower levels compared with Myc-Frat2 (Fig. 3A) even though ␤-galactosidase, which was co-transfected to control for transfection efficiency, was expressed at similar levels (not shown). By transfecting increasing amounts of Frat we studied the concentration-dependent response of the TOPFLASH reporter. Comparable amounts of Frat1 and Frat2 protein induced different levels of reporter activity (Fig. 3B). Thus, Frat2 is a less potent activator of Wnt signaling through ␤-catenin/TCF compared with Frat1 and Frat3 even if present at much higher protein levels. Because Frat3 is only present in mice and rats and is observed as a doublet on denaturing gels. After treatment with protein phosphatase, migration of both Frat1 and Frat2 is affected. C, phosphorylation by GSK3␤ is not responsible for the observed mobility shift. After treatment with LiCl, GSK3␤ becomes phosphorylated on Ser-9, which renders it inactive. However, Frat2 was still observed as a doublet on SDS/ PAGE gels. since its behavior in the TOPFLASH assay was intermediate to that of Frat1 and Frat2, we decided to focus on the latter two homologues.
Frat2 Is More Stable than Frat1-The observed difference in protein levels could be accounted for by a difference in protein half-life. To test this, we treated 293T cells expressing Myc-Frat1 or Myc-Frat2 with CHX to block protein synthesis and monitored the decay patterns by Western blot analysis. We consistently observed that substantial amounts of Frat2 remained present after prolonged CHX treatment. In contrast, Frat1 could hardly be detected at the same time point (Fig. 4A). When the change in protein levels was followed over time, both Frat1 and Frat2 levels gradually decreased in the first 2 h after CHX treatment (Fig. 4B). After 6 h of CHX treatment the decrease in Frat1 levels was more pronounced than the decrease in Frat2 levels. The observed pattern of protein degradation was independent of the starting levels of protein. Also when the initial levels of Frat1 were higher than those of Frat2 (Fig. 4B, bottom) the levels of Frat2 diminished much less than those of Frat1. There was no obvious difference in the stability of hyper-and hypophosphorylated Frat2 (Fig. 4, A and B). To determine the half-life of Frat1 and Frat2 more accurately, we quantified the amounts of protein by Western blot analysis and densitometry. Fig. 4C depicts the average of 3 or 4 experiments for each time point. After a 6-or 7-h treatment with CHX, Frat1 levels had decreased by more than 90%. In contrast, we observed that the levels of Frat2 protein had only dropped by 50%. Therefore, we conclude that Frat2 is considerably more stable than Frat1 due to a longer protein half-life.
Given its interaction with GSK3␤ and the increasing evidence that different pools of this kinase exist within the cell (24) we speculated that Frat2 might be more stable as a result of its presence in an intracellular pool, where it could be shielded from destruction. After transfection of 293T cells we studied the intracellular localization of Myc-Frat1 and Myc-Frat2 by immunofluorescence (Fig. 4D). Both Frat1 and Frat2 were predominantly located in the cytoplasm when cells were left untreated. After CHX treatment for 6 h before fixation, the overall fluorescence intensity decreased. In fact, the results confirmed the earlier Western blot analysis, with a more substantial decline in Frat1 signal compared with Frat2. However, FIG. 3. Frat2 is less efficient in inducing canonical Wnt-signaling than Frat1. A, Frat2 is able to activate a TOP-FLASH reporter in 293T cells. All luciferase reporter assays were performed in triplicate and performed at least three times. A representative experiment is shown. Frat2 induces reporter activity less well than Frat1 (*, p Ͻ 0.005 as determined by Student's t test). When the same lysates were analyzed on Western blot (WB) for expression of the proteins Frat2 was found to accumulate to higher levels than Frat1 or Frat3 even though it was less active in inducing Wnt signaling. B, the transfection of increasing concentrations of Frat (10,25,50,75, and 100 ng of DNA for each homologue) shows the concentration-dependent response of the TOPFLASH reporter. Again, the absolute protein levels of Frat2 are higher than those of Frat1 and Frat3. Similar levels of Frat protein are obtained upon the transfection of 100 ng of Frat1 and 25 ng of Frat2.
we did not observe a pool of Frat2 with a specific subcellular localization that was more resistant to destruction.
The C Terminus of Frat1 Conveys Protein Instability-Given the apparent discrepancy between Frat2 protein stability on the one hand and its poor ability to induce canonical Wnt signaling on the other hand, we searched for specific regions in Frat1 and Frat2 that might determine either of these characteristics. The amino acid sequence of Frat2 is most divergent from Frat1 in the central part of the protein, where Frat2 lacks a stretch of amino acids present in Frat1 and in the C-terminal part downstream of the IKEA box (Fig. 5A). We generated fusion proteins (shown schematically in Fig. 5A) by cloning Frat1 and Frat2 sequences in-frame at a BssHII restriction site (A), at the EcoNI site in the IKEA box (B), or at an Eco47III site, which was first introduced into Frat1 by site-directed silent mutagenesis (C). All constructs were tested in the TOP-FLASH assay to ascertain that they still represented functional proteins. Although all fusion proteins retained activity, none of them mimicked the behavior of either Frat1 or Frat2. Instead, most constructs showed intermediate or low activation of the TOPFLASH reporter (Fig. 5B). Thus, we were unable to define a distinct domain solely responsible for optimal activation of signaling through ␤-catenin. However, we found that Frat1 sequences downstream of the IKEA box substantially reduced the stability of the proteins (Fig. 5C). Constructs containing Frat1 C-terminal sequences (2A1 and 2B1) accumulate to levels comparable with Frat1. In contrast, constructs composed of a Frat1 N terminus and a Frat2 C terminus (2A1B2 and 1B2) show a much higher stability. This observation could either be explained by a destabilizing motif in Frat1 or a stabilizing motif in Frat2. We therefore generated a Frat1 construct that lacked the Frat1 C terminus (Frat1⌬C, Fig. 5A).
This mutant showed a much higher stability than full-length Frat1 (Fig. 5D), indicating the presence of a destabilizing motif in Frat1 downstream of the NruI site (D, Fig. 5A). Frat1⌬C could also activate the TOPFLASH reporter (results not shown), but the levels of reporter gene induction were in the same range as those for most fusion proteins and did not reach the high values observed for Frat1. Thus, whereas a distinct C-terminal domain is responsible for the observed differences in protein stability, the optimal activation of canonical Wntsignaling likely requires the interaction of multiple domains.

DISCUSSION
Here we report the cloning and characterization of the murine Frat2 gene, which represents the third member of this highly conserved family of intronless genes. Whereas Frat3 shares 84% amino acid homology with Frat1, Frat2 is only 68% identical. Interestingly, the human and mouse Frat2 homologues are 76% identical at the amino acid level, suggesting that they share a unique function distinct from Frat1 and Frat3, conserved during evolution. Like Frat1 and Frat3, Frat2 does not contain any known protein domains or functional motifs except for the IKEA sequence, which represents the binding site to GSK3␤.
We found that Frat1 and Frat2 are subject to phosphorylation. This is the first report on posttranslational modification of this family of proteins. Given the fact that Frat1 and Frat2 both bind to GSK3, we considered GSK3 to be an attractive candidate to phosphorylate Frat. In addition, Frat1 contains three and Frat2 contains five potential GSK3 phosphorylation sites. However, LiCl treatment did not result in an observable mobility shift of Frat proteins on SDS-PAGE gels, indicating that GSK3 does not cause the change in phosphorylation re-FIG. 4. Frat2 is more stable than Frat1. A, transiently transfected 293T cells were treated with CHX for 7 h, resulting in a substantial decrease in Frat1 but not Frat2 protein levels. To start out with similar levels, 4 -5 times more Frat1 than Frat2 DNA was transfected. WB, Western blot. B, Western blot analysis of protein stability reveals that the difference in protein levels at later time points is irrespective of the initial net levels. Two representative experiments are shown. C, quantitation of protein levels after CHX treatment of transiently transfected 293T cells by Western blot densitometry does not show a significant difference in protein half-life at early time points. After 6 -7 h of CHX treatment, Frat1 levels dropped to less than 10%, whereas Frat2 levels remain at 50%. The graph shows the average of 3 or 4 experiments for each time point. D, immunofluorescence of transfected 293T cells reveals a predominantly cytoplasmic localization for Frat1 and Frat2. An overall decrease in fluorescence signal is observed after 6 h of CHX treatment, albeit more pronounced for Frat1 than for Frat2. sponsible for the mobility shift of these proteins. Nevertheless, we cannot rule out that GSK3 phosphorylates Frat. Although the functional importance of the observed phosphorylation is currently unknown, it might prove to be an important mechanism for activating or inactivating these proteins. Several kinases, including casein kinase I and II, have been implicated in the phosphorylation and regulation of Wnt-pathway components (8,(25)(26)(27)(28)(29)(30)(31)(32)34). A better understanding of the biological function of Frat2 will help us to identify kinases affecting its phosphorylation status.
We also show that Frat1, Frat2, and Frat3 have partially overlapping activities. In agreement with the earlier reported ability of different members of the Frat/GBP family to bind to GSK3␤, we found that all three murine Frat proteins are able to bind to GSK3␤ and induce signaling through ␤-catenin/TCF, as determined in a TOPFLASH luciferase reporter assay. However, Frat1 is by far the more potent activator of canonical Wnt-signaling despite the fact that Frat2 accumulates to much higher levels. Because Frat1 and Frat2 are expressed from the same expression vector and detected with an antibody that recognizes an N-terminal Myc tag, transcriptional and translational events can be largely ruled out as a cause of the observed differences in protein levels between the two homologues. Therefore, the observed difference between Frat1 and Frat2 protein levels appears to represent an intrinsic difference between the two homologues. Because neither Frat1 nor Frat2 contain known functional motifs that provide any further clues about its function or regulation, we sought to map the region that might be responsible for the difference in Frat protein accumulation. We found that the C terminus downstream of the GSK3 binding site controls protein levels, although the very C-terminal tail downstream of the Eco47III site was unable to confer (de)stabilization. In addition to controlling protein levels, this same region also appears to be partially responsible for the activation of Wnt signal transduction, since Frat1 proteins containing a Frat2 C terminus never reached the levels of reporter activation achieved by wild type Frat1 despite the fact that protein levels were higher. Although this may be FIG. 5. Frat1 contains a destabilizing motif in the C terminus. A, amino acid alignment and schematic representation of the Frat1 and Frat2 proteins. Frat2 lacks a stretch of amino acids in the central part of the protein (between sites A and B) and has a shorter C-terminal tail. In addition, the sequence downstream of the IKEA box containing the GSK3␤ binding site is very divergent between the two. A, B, and C indicate the BssHII, EcoNI, and Eco47III sites, respectively, which were used to generate the Frat1 and Frat2 hybrid proteins ("swops") by cloning Frat1 N-terminal sequences upstream of Frat2 C-terminal sequences and vice versa. D indicates the NruI site that was used to generate the Frat1 C-terminal deletion mutant. B, all swops were tested in the TOPFLASH luciferase reporter assay for their ability to induce Wnt-signaling, which at the same time indicates that the hybrid proteins are functional. No distinct domain can be defined that seems solely responsible for optimal activation of the reporter. WB, Western blot. C, Western blot analysis shows that the higher levels of Frat2 protein are due to sequences in the C terminus downstream of the IKEA sequence (compare for example 2A1 to 2A1B2 and 2B1 to the reverse 1B2 swop). D, Western blot analysis reveals a destabilizing motif in the Frat1 C terminus. A Frat1 C-terminal deletion mutant accumulates to higher protein levels than full-length Frat1.
due to suboptimal folding of the fusion proteins, it is more likely that optimal binding to GSK3␤ requires the interaction of multiple domains. The amino acid sequence in Frat2 that allows accumulation of Frat protein does not contain a distinct sequence motif. The corresponding region in Frat1 contains a leucine-rich stretch. In addition Frat1 contains a lysine in this region that is not conserved in Frat2. This residue could be involved in ubiquitination and make the protein a better target for degradation, thereby leaving Frat2 unaffected. Indeed, a Frat1 C-terminal deletion mutant lacking this stretch accumulates to higher levels than full-length Frat1. Taking the stability of the fusion proteins into account it is most likely that the segment between the IKEA box and the Eco47III site is at least partially responsible for the instability of Frat1.
Endogenous Frat1 mRNA and protein is hardly if at all detectable. This might be caused by the destabilizing motifs in the 3Ј-UTR of the mRNA, thus providing mRNA and protein with regulatory motifs that reduce their half-life. The Frat2 gene contains similar destabilizing motifs in the 3Ј-UTR and can, therefore, be envisioned to be regulated in a similar fashion. However, since Frat is considered to act as a very potent inhibitor of GSK3, there appears to be an apparent discrepancy between the observation of high Frat2 protein levels and a reduced capacity to activate the TOPFLASH reporter. The difference in protein levels can be best explained by a difference in protein half-life. Indeed, we found Frat2 to be more stable than Frat1 upon prolonged CHX treatment. Unfortunately, the variation associated with the densitometry measurements at early time points prevented us from determining whether this difference in degradation is also observed upon shorter periods of CHX treatment or whether Frat2 is divided over both a stable and unstable pool.
Although we did not find a difference in the intracellular localization between Frat1 and Frat2, the apparent stability of Frat2 could reflect its association with other proteins that shield Frat2 from destruction. Furthermore, the fact that the higher levels of Frat2 activate the TOPFLASH reporter less well than Frat1 lends further support to the notion that the two proteins likely have partially overlapping but otherwise distinct functions in the cell. The fact that we do not find any proviral integrations affecting the Frat2 gene in Moloney-induced tumors might indicate that overexpression of this gene does not yield a substantial growth advantage to tumor cells as does activation of Frat1.
In fact, Frat2 might not act in the canonical Wnt-signaling pathway at all but in one of the numerous other GSK3-dependent cellular activities (14,33,35). In addition to binding to GSK3, Frat1 has been reported to associate with the Wntpathway component Dishevelled (15). It will be interesting to see whether Frat2 shares this feature.
It should be stressed that most studies establishing a role for Frat in Wnt signaling have relied on overexpression of the protein in vitro. Therefore, it has not been formally shown that Frat is part of the canonical Wnt pathway in that it signals through ␤-catenin in mammals during embryonic development or in normal physiology. The fact that no Frat orthologues have been found in Caenorhabditis elegans and Drosophila melanogaster, whereas the gene is highly conserved in Xenopus, zebra fish, mice, and men suggests that Frat might either provide an addi-tional layer of control in the Wnt pathway only to be found in vertebrates or that it is part of a completely different signaling pathway. Additional in vivo studies are needed to shed more light on the physiological role of the different Frat proteins.