The mid-hindbrain boundary (MHB) harbors an important organizing center for the adjacent brain regions. Here, we present evidence that the receptor protein tyrosine phosphatase λ (RPTPλ) is part of the complex molecular network that maintains and shapes the MHB region. RPTPλ is expressed in a tight band of cells in the caudal midbrain, anterior to the transverse ring of Wnt1 expression. Forced expression of RPTPλ across the mid-hindbrain region repressed expression of Wnt1, whereas RNA interference-mediated knock-down of RPTPλ resulted in expansion and distortion of the Wnt1 domain. When ectopically expressed in the mesencephalon, RPTPλ specifically inhibited the induction of Wnt1 expression after subsequent stimulation with Fgf8. Reduced Wnt1 expression after RPTPλ transfection correlated with a decrease in Ras- mitogen-activated protein kinase activity at the MHB. We further show that in the embryonic midbrain, RPTPλ can bind to β-catenin, a central component of the canonical Wnt signaling pathway. Overexpression of RPTPλ suppressed the activity of a β-catenin responsive promoter in the midbrain and reduced progenitor cell proliferation. Cotransfection of Wnt1 or of a stabilized form of β-catenin together with RPTPλ partially rescued the RPTPλ-mediated proliferation defect. Together, these data suggest that RPTPλ may play a dual role in the control of midbrain development: as a negative modulator of Fgf8-induced Wnt1 expression at the MHB, which may help to confine the Wnt1 domain to it characteristic tight ring at the MHB; and as an inhibitor of canonical Wnt signaling through interaction with and presumably sequestration of β-catenin.
The mid-hindbrain (MHB) (isthmic) organizer, located near the constriction between the developing mesencephalic and metencephalic vesicles, acts as a secondary organizer for the control of growth and pattern formation of the adjacent midbrain and hindbrain (Nakamura et al., 1986; Martinez and Alvarado-Mallart, 1990; Martinez et al., 1991). Its activity is regulated by an interdependent network of nuclear factors and signaling proteins, which includes the transcription factors En1/En2, Pax2/Pax5, Lmx1b, Hes1/Hes3, iro1/iro7, and members of the Fgf and Wnt families of secreted proteins (Bally-Cuif et al., 1992; Rowitch and McMahon, 1995; Araki and Nakamura, 1999; Shamim et al., 1999; Adams et al., 2000; Hirata et al., 2001; Itoh et al., 2002; Matsunaga et al., 2002; O'Hara et al., 2005).
Fibroblast growth factor 8 (Fgf8), expressed in the rostral hindbrain adjacent to the MHB, is of central importance for midbrain and hindbrain development. Beads soaked in recombinant Fgf8 and implanted in prosomeres p1, p2 of the diencephalon, in the midbrain or hindbrain can mimic the inductive capacity of isthmic transplants (Crossley et al., 1996; Martinez et al., 1999; Shamim et al., 1999; Irving and Mason, 2000). Conversely, zebrafish ace mutants, which lack functional Fgf8, and mice hypomorphic for Fgf8 fail to maintain gene expression at the MHB and display massive defects in midbrain and cerebellar development (Meyers et al., 1998; Reifers et al., 1998). An important target of Fgf8 at the MHB is the secreted glycoprotein Wnt1. After isthmic organizer formation, Wnt1 expression is restricted to a narrow band of cells in the caudal mesencephalon, anterior to the Fgf8 expression domain (Bally-Cuif et al., 1995; Hollyday et al., 1995). Wnt signaling is critically involved in mid-hindbrain development, because targeted deletion or tissue-specific overexpression of Wnt1 at the MHB lead to massive perturbation of mid-hindbrain growth (Thomas and Capecchi, 1990; McMahon et al., 1992; Panhuysen et al., 2004).
Reversible protein phosphorylation on tyrosine residues is a key mechanism underlying intracellular signal transduction. Protein tyrosine phosphatases (PTPs) antagonize the activity of protein tyrosine kinases (PTKs) and thereby limit the duration and intensity of the PTK signal (Stoker and Dutta, 1998; den Hertog, 1999; Paul and Lombroso, 2003; Tonks, 2006). Receptor protein tyrosine phosphatase λ (RPTPλ; also called RPTPψ), a member of the type IIb family of receptor tyrosine phosphatases, is expressed in a spatially and temporally dynamic manner in the embryo and has been shown recently to be necessary for somitogenesis and convergent extension during gastrulation (Aerne et al., 2003; Aerne and Ish-Horowicz, 2004). We reported previously the developmental expression of RPTPλ in the CNS and have shown that its expression is particularly dynamic anterior to the isthmic constriction (Badde et al., 2005). Here, we have taken a gain-of-function and loss-of-function approach to show that RPTPλ can modulate mesencephalic development through a dual mechanism: it inhibits the induction of Wnt1 expression by Fgf8, which may help to restrict the Wnt1 domain to its characteristic tight ring at the MHB, and binds to β-catenin and appears to thereby negatively modulate canonical Wnt signaling.
Materials and Methods
Expression constructs and in ovo electroporation.
cDNAs encoding full-length mouse RPTPλ [National Center for Biotechnology Information (NCBI) accession number U55057] or RPTPλ-ΔIC (corresponding to nucleotides 352–2701 of NCBI accession number U55057) were fused to a triple HA-tag and cloned into the expression vector pMES, which includes an internal ribosomal entry site (IRES)-green fluorescent protein (GFP) cassette (Swartz et al., 2001) or into the vector pMIWIII, which lacks IRES-GFP (Suemori et al., 1990). The coding sequence of chick RPTPλ was cloned using degenerate primers based on published sequences of the mouse, rat, and human orthologs and found to be mostly identical to NCBI accession number AY147868 (sequences are available on request). The coding sequences of chick Fgf8b, mouse Wnt1, or mouse En1 were cloned into pMIWIII. For dnRasN17-pMES, a dominant-negative form of the Ras protein, which contains a serine to asparagine mutation at residue 17, was inserted into pMES (Clontech). Axin-pMES carried the full-length coding region of chick Axin, a generous gift from F. Costantini (Columbia University Medical Center, New York, NY) (Zeng et al., 1997). For caβ-catenin-pMIWIII, a stabilized form of β-catenin was cloned into pMIWIII (Funayama et al., 1995). For the rescue experiment with RPTPλ and Wnt1, the coding sequence of GFP in RPTPλ-pMES was replaced by a single ClaI site (RPTPλ-pMEC). The coding region of Wnt1 was then cloned into this ClaI site resulting in RPTPλ-Wnt1-pMEC. Unless otherwise noted, 1–2 μg/μl of each construct were injected into the neural tube of Hamburger–Hamilton stage 9–10 (HH9–HH10) chick embryos (White Leghorn) (Hamburger and Hamilton, 1992). Electroporation was performed as described, except that five pulses of 10 V for 50 ms were given to facilitate DNA uptake (Schulte et al., 1999). To visualize DNA uptake, 0.5–0.8 μg/μl of plasmids expressing enhanced GFP (GFP-pMIWIII), Discosoma red fluorescent protein (dsRed-pCMV), or a nuclear version of red fluorescent protein (nRFP-pCAGGS) were coelectroporated. To control for possible unspecific effects of the procedure, pMES (which carries IRES-GFP) or GFP-pMIWIII was introduced. For consecutive electroporation into the same region of the neural tube (see Fig. 3A), GFP with or without the full-length form of RPTPλ or the catalytically inactive allele RPTPλ-ΔIC were introduced into the mesencephalic vesicle at HH9–10. Six hours after initial transfection, a time window known to be sufficient for robust transgene expression (Momose et al., 1999), an expression plasmid carrying the coding sequence of Fgf8b together with a plasmid carrying dsRed were electroporated into the same area that had received the first electroporation 6 h earlier. Only embryos that exhibited extensive coexpression of both fluorescent markers, indicating that both transgenes were targeted to the same region of the neural tube, were chosen for additional analysis. For TOP:dgfp reporter analysis, 1 μg/μl of the TOP:dgfp reporter construct, which drives expression of destabilized green fluorescent protein off the TOPFLASH promoter/enhancer (Korinek et al., 1997; Dorsky et al., 2002), were electroporated into the mesencephalic vesicle alone or in combination with 1.5 μg/μl RPTPλ-pMIWIII or 1.5 μg/μl RPTPλ-ΔIC-pMIWIII. GFP expression was monitored and documented 24 h later.
Heparin-coated acrylic beads (Sigma-Aldrich) were soaked in 0.015–0.15 mg/ml mouse recombinant Fgf8b protein (R&D Systems) overnight at 4°C, washed in sterile PBS, and split with forceps. Appropriately sized bead fragments were implanted into the lateral wall of the mesencephalic vesicle of HH11 chick embryos. The embryos were collected 24 h after bead implantation and processed for in situ hybridization.
RNA in situ hybridization.
To ensure specificity of the probe, a partial sequence of chick RPTPλ was used for all in situ hybridization experiments (Badde et al., 2005). RNA probes for cGrg4, cSprouty2, and cSef were cloned from chick HH15–20 whole head total RNA by reverse transcriptase-PCR with gene-specific primers (primer sequences are available on request). The cDNAs used to generate in situ probes for cEn1, cFgf8, cLmx1b, cOtx2, cPax2, cPax5, or cWnt1 were gifts from C. Tabin (Harvard Medical School, Boston, MA), C. Logan (University of Calgary, Calgary, Alberta, Canada), or H. Rohrer (Max Planck Institute for Brain Research, Frankfurt, Germany). In situ hybridization on whole embryos was performed as described previously (Schulte et al., 1999). For open book preparation, the mesencephalic vesicle was opened along the dorsal midline and flat-mounted. Two-color in situ hybridization was performed as described with the exception that the first transcripts were detected using 5-bromo-4-chlor-indolyl-phosphate (BCIP) and nitroblue–tetrazolium–chloride as chromophores, which results in a dark blue/purple precipitate, whereas the second transcripts were detected with BCIP alone, which results in a bright blue/turquoise precipitate (Schulte and Cepko, 2000). Specimens were photographed with a Zeiss StemiSV11 microscope and a Canon PowerShot G5 camera or with a Leica MZ12 microscope and a Leica MPS60 camera. Brightness and contrast were adjusted in Adobe Photoshop (Adobe Systems). In Figure 1, C, D, D′, I′, and I″ were false colored in Adobe Photoshop for better visual contrast.
Localization of ectopically expressed RPTPλ-HA or RPTPλ-ΔIC-HA was visualized on cryosections using a polyclonal anti-HA antibody (1:1000; Roche Diagnostics). GFP expression was confirmed using a polyclonal antibody (1:1000; Invitrogen). The mitotic index was determined with a polyclonal anti-phosphorylated Histone H3 (pH3) antibody (1:1000; Upstate Biotechnology). All antibodies were diluted in PBS containing 5% Chemiblocker (Millipore Bioscience Research Reagents) and 0.5% Triton X-100 before use. Microscope fields with a magnification of 40× were documented either with an Axioplan2 (Zeiss), a FluoView 1000 confocal microscope (Olympus), or a LSM5 Pascal confocal microscope (Zeiss). Immunohistochemical detection of diphosphorylated Erk1/2 with the monoclonal anti-dpErk antibody (clone M-8159 Sigma-Aldrich) was performed as described by Corson et al. (2003).
For RNA interference (RNAi)-mediated knock-down of RPTPλ or Wnt1 expression, the psiSTRIKE-U6 Hairpin Cloning System was used (Promega). The short hairpin RNA target sequences directed against nonoverlapping regions of the cRPTPλ cytoplasmic domain were GTCAACATGACCAAAGCAA (RPTPλ_si-a) and GCTTCAAGCAGGAGTATGA (RPTPλ_si-b). The efficiency and specificity of the RNAi targeting constructs were assessed by luciferase reporter assay. Human embryonic kidney 293T (HEK293T) cells were cotransfected with the RNAi targeting constructs RPTPλ_si-a, RPTPλ_si-b, targeting constructs containing randomized sequences of RPTPλ_si-a or RPTPλ_si-b, respectively (GAACTACGTAAACCAGACA or GCATAGACGAATCGGTATG), or an unrelated targeting construct (GAAGGTACACGAATTATGT) together with a psiCHECK-2 reporter construct, which contained the intracellular portion of cRPTPλ fused to Renilla luciferase (1.5 μg of each construct/six-well plate; Promega). After 48 h, luciferase activity was measured using the Dual-Luciferase Reporter 1000 Assay System (Promega) in a GloMax96 plate Luminometer (Promega). For in vivo analysis, RPTPλ_si-a or RPTPλ_si-b was electroporated into the neural tube at a concentration of 1.5 μg/μl together with GFP-pMIWIII (0.8–0.5 μg/μl) at the developmental ages indicated. Short hairpin RNA target sequences against Wnt1 were GGTCATCTACGGCAACAAA and GCGCCTCGAGGGTCATCTA. Simultaneous transfection of both RNA targeting constructs (1.5 μg/six-well dish) repressed the activity of a psiCHECK-2 reporter construct, which contained the coding region of chick Wnt1 fused to Renilla luciferase, to 34% compared with a targeting construct carrying the randomized sequence CGAATCAGTCAGTCCAGAA (data not shown).
RPTPλ-HA-pMES or RPTPλ-ΔIC-HA-pMES was electroporated into the midbrain at HH10. After 24 h, the midbrain was dissected in ice-cold PBS and lysed in 150 mm NaCl, 10 mm Na-Phosphate buffer, pH 7.2, 1% NP-40, 1% Desoxycholate, 0.1% SDS, 50 mm Na-F, 0.2 mm Na-orthovanadate, and protease inhibitors (Complete tablets; Roche Diagnostics). Precipitation of the transgenes was achieved using a monoclonal, agarose-coupled anti-HA antibody (Sigma-Aldrich). Immunoprecipitation and Western blot analysis were performed as described previously (Mühleisen et al., 2006). Primary antibodies used for Western blot analysis were rat anti-HA high-affinity antibody (1:1000; Roche Diagnostics) or polyclonal anti-β-catenin (1:1000; Upstate Biotechnology). For stripping off the first set of antibodies, the blots were incubated in 0.1 m glycine, pH 2.5, for 30 min at 37°C.
Analysis of cell proliferation and programmed cell death.
RPTPλ-HA-pMES or pMES were electroporated into the midbrain of HH9 chick embryos. After 24 h, bromodeoxyuridine (BrdU) was injected into the midbrain vesicle; labeling duration was 2 h. Detection of BrdU-positive cells was performed on 15-μm-thick cryosections using the BrdU Labeling and Detection Kit I (Roche Diagnostics) and colabeling with a GFP-specific antibody to visualize transfected cells. For analysis of programmed cell death, RPTPλ-pMIWIII together with nRFP-pCAGGS, or nRFP-pCAGGS alone, was electroporated into the midbrain vesicle. Detection of apoptotic cells was performed on cryosections using the Fluorescein In Situ Cell Death Detection Kit (Roche Diagnostics).
RPTPλ is expressed in a domain tightly associated with the ring of Wnt1 expression at the MHB and is regulated by Fgf8
We reported previously that between the 19 and 40 somite stages (HH13–HH19), RPTPλ is expressed in a tight, transverse ring anterior to the isthmic constriction and to the transverse domain of Wnt1 expression (Fig. 1A,B) (Badde et al., 2005). To correlate the precise location of this ring to that of Wnt1 at the organizer, RPTPλ- and Wnt1-specific transcripts were visualized together in the same specimen by two-color in situ hybridization. Expression of the two molecules was found in tightly associated domains at HH14 (Fig. 1C) and HH19 (Fig. 1D,D′). Although both expression domains appear to slightly overlap, in all specimens analyzed, RPTPλ-specific transcripts were found rostral to the ring of Wnt1 expression that abuts the MHB. Therefore, during this developmental period, the Wnt1 domain at the MHB is bordered by a ring of RPTPλ-expressing cells at its anterior side. Based on these expression data, the RPTPλ expression domain at the MHB may lie within the future preisthmic domain, the caudal-most aspect of the mesencephalic alar plate, which is characterized by coexpression of Otx2, Pax2, and possibly Wnt1 (Hidalgo-Sánchez et al., 2005).
As a first step to understanding RPTPλ function at the MHB, we investigated whether its expression was regulated by the secreted protein Fgf8. Fgf8 is expressed in rhombomere 1 adjacent to the isthmic organizer and can induce organizer characteristic gene expression and cell fate changes when ectopically applied in the caudal diencephalon or mid-hindbrain region (Crossley et al., 1996; Martinez et al., 1999). Transfection of 1 μg/μl to 0.1 μg/μl of an expression plasmid carrying the splicing isoform Fgf8b into the mid-hindbrain region of chick embryos has been shown previously to induce ectopic cerebellar development, whereas transfection of 0.01 μg/μl promoted tectal growth (Sato et al., 2001). We introduced an Fgf8b expression plasmid at different concentrations into the mid-hindbrain region together with an expression plasmid carrying GFP for visualization (Fig. 1F–G′). Expression of Wnt1 and RPTPλ at the MHB were unaltered after transfection with the GFP expressing plasmid alone compared with wild-type (WT) control embryos, demonstrating that in ovo electroporation per se does not affect gene expression at the MHB (Fig. 1E) (data not shown) (n = 36 of 37 for RPTPλ; n = 12 of 12 for Wnt1). Consistent with our observation that RPTPλ is not expressed in the hindbrain region during normal development, we found that widespread electroporation across the mid-hindbrain territory of 1 μg/μl of the Fgf8b-pMIWIII expression plasmid, a concentration known to stimulate cerebellar development, effectively repressed RPTPλ expression at the MHB (n = 9 of 10) (Fig. 1F,F′). Electroporation of 0.01 μg/μl of the Fgf8b expression plasmid, a concentration that enhanced midbrain growth in a previous study (Sato et al., 2001), did not alter RPTPλ expression (n = 7 of 7) (data not shown). However, RPTPλ-specific transcripts were upregulated in the midbrain vesicle at an intermediate concentration of 0.05 μg/μl Fgf8b-pMIWIII (n = 3 of 5) (Fig. 1G,G′). Fgf8b misexpression at either concentration was unable to stimulate RPTPλ expression in the hindbrain, supporting the idea that metencephalic tissue is not permissive for RPTPλ expression (n = 18 of 18) (Fig. 1F–G′) (data not shown). Together, these results suggest that RPTPλ expression at the isthmic organizer requires a distinct Fgf8 signaling level, which is below the level required for metencephalic development but above the level that stimulates tectal growth. Fgf8 signal transduction at the MHB appears to be mediated primarily through the Ras-mitogen-activated protein (MAP) kinase pathway (Corson et al., 2003; Sato and Nakamura, 2004). To confirm the dependency of RPTPλ expression on Fgf8 signaling at the MHB, we misexpressed a dominant-negative form of Ras (dnRasN17), which was shown previously to effectively block Fgf8 signal transduction at the isthmic organizer (Sato and Nakamura, 2004). RPTPλ expression was consistently lost in cells forced to express dnRasN17, demonstrating that RPTPλ expression at the MHB indeed requires Fgf8/Ras-MAP kinase pathway activity (Fig. 1H,H′) (n = 8 of 9).
Next, we targeted the Fgf8b expression plasmid (at a concentration of 1 μg/μl) to a narrow region surrounding the isthmic organizer and analyzed the expression of RPTPλ and Wnt1 together in the same specimen 24 h later (Fig. 1I–I″). The domain of RPTPλ expression was shifted rostrally around the region targeted for Fgf8 misexpression (recognizable by the fluorescence of the coelectroporated GFP; n = 9 of 10) (Fig. 1I, arrow), consistent with the idea that RPTPλ expression is inhibited by strong Fgf8b signals but induced at a distance from the Fgf8b source, where Fgf8b protein levels are expected to be lower. When Wnt1- and RPTPλ-specific transcripts were visualized together, both expression domains were arranged in concentric circles, with the RPTPλ domain (arrow) always located anterior to the Wnt1 domain (Fig. 1I′,I″, arrowhead) (n = 6 of 6). Fgf8-induced upregulation of Wnt1 and RPTPλ thus mimics their normal spatial relationship at the MHB, where the Wnt1 domain separates the Fgf8 and RPTPλ expression domains (Badde et al., 2005) (Fig. 2G). This raises the possibility that the threshold of Fgf8 signaling differs from the induction of Wnt1 and RPTPλ expression at the MHB, and that Wnt1 expression requires higher levels of Fgf8 signaling than expression of RPTPλ. To investigate this in more detail, we narrowed down the concentrations of Fgf8 necessary for inducing Wnt1 and RPTPλ expression, respectively, by implanting fragments of beads soaked in different concentrations of Fgf8b. As reported previously, Wnt1 expression was induced around a bead fragment soaked in 0.15 mg/ml Fgf8b (Fig. 2A,A′, arrowheads) (n = 12 of 12) (Crossley et al., 1996). Induction of RPTPλ expression was also consistently observed, yet its expression domain was located at a distance to the Fgf8b soaked bead and must therefore lie distal of the ring of Wnt1 expression, which generally surrounded the Fgf8 source (Fig. 2B,B′, arrows) (n = 15 of 15). When the Fgf8b concentration was lowered to 0.05 mg/ml, some ectopic Wnt1 expression could still be observed surrounding the Fgf8 soaked bead, and the ring of RPTPλ-specific transcripts was located closer to the bead (Fig. 2C,D) (n = 5 of 8 for Wnt1; n = 7 of 7 for RPTPλ). Fgf8b at 0.015 mg/ml was not sufficient to induce Wnt1 expression (Fig. 2E) (n = 1 of 3) but could still elicit robust ectopic RPTPλ expression (Fig. 2F,F′) (n = 4 of 4). Although additional work is clearly needed to fully understand Fgf receptor signaling dynamics at the MHB, these results strongly suggest that transcriptional activation of Wnt1 and RPTPλ requires different levels of Fgf8 signaling.
Overexpression of RPTPλ suppresses Wnt1 expression at the MHB
To examine the function of RPTPλ at the MHB, we misexpressed the murine homolog of RPTPλ in the area surrounding the isthmic constriction (Fig. 3A). By fusing the coding region of RPTPλ to a triple HA-tag, the ectopically expressed RPTPλ protein could be distinguished from its endogenous form by the use of an HA-specific antibody. As expected for a transmembrane protein, RPTPλ-HA staining was concentrated to cellular plasma membranes, suggesting that the recombinant protein was processed normally (Fig. 3B). When GFP alone was ectopically expressed in the MHB region, Wnt1 expression was unaltered (Fig. 3C) (n = 15 of 16). In contrast, after ectopic expression of RPTPλ-HA, the normal Wnt1 expression pattern was markedly disturbed (Fig. 3D,E) (n = 17 of 20). In embryos in which RPTPλ-HA was misexpressed in a broad region surrounding the isthmic constriction, Wnt1 expression was reduced or lost from the electroporated side of the embryo, suggesting that strong RPTPλ expression inhibits Wnt1 expression at the MHB (Fig. 3D, arrowhead) (n = 12 of 20). In some embryos, particularly in those where RPTPλ-HA expression was directly targeted to the organizer (and just to the region where Wnt1 would normally be expressed), a rostral displacement or split of the Wnt1 expression domain around the RPTPλ-HA expressing cells was observed (Fig. 3E, arrowheads) (n = 5 of 20). Identical results were obtained when an untagged form of the protein was misexpressed, indicating that the presence of the triple-HA tag was not critical for RPTPλ function (data not shown). To test for the possible contribution of the cytoplasmic part of RPTPλ, the domain most likely to be essential for its phosphatase activity and the interaction with components of intracellular signal transduction pathways, we constructed truncated forms of RPTPλ in which a triple HA-tag replaced the juxtamembrane and phosphatase domains (RPTPλ-ΔIC-HA). When overexpressed in the mesencephalic vesicle, RPTPλ-ΔIC-HA staining was concentrated to cellular plasma membranes, indicating correct targeting of the truncated protein (Fig. 3F). In contrast to the full-length form of RPTPλ, electroporation of RPTPλ-ΔIC-HA did not perturb Wnt1 expression (Fig. 3G) (n = 21 of 22), indicating that the cytoplasmic domain was required for RPTPλ function at the MHB.
To further characterize the potential role of RPTPλ in modulating gene expression at the MHB, we analyzed the expression of other MHB marker genes after electroporation of RPTPλ-HA. The expression domain of Fgf8 was normal 24 h after misexpression of RPTPλ (n = 6 of 6) (supplemental Fig. 1A,B, available at www.jneurosci.org as supplemental material). The altered Wnt1 expression that we observed 24 h after RPTPλ transfection can therefore not be a secondary effect of an altered Fgf8 expression profile at the MHB. In addition, forced expression of RPTPλ did not affect expression of Lmx1b (n = 16 of 17), En1 (n = 8 of 8), Pax2 (n = 9 of 11), Pax5 (n = 9 of 10), Otx2 (n = 4 of 4), Sef1 (n = 7 of 8), sprouty2 (n = 7 of 7), or Grg4 (n = 5 of 5) within 24 h after transfection (supplemental Fig. 1C–K, available at www.jneurosci.org as supplemental material) (data not shown).
RPTPλ interferes with Fgf8-mediated Wnt1 induction
Next, we took advantage of the fact that Fgf8 can mimic the organizer's activity when ectopically expressed in adjacent brain regions and performed two consecutive rounds of in ovo electroporation within the mesencephalic alar plate (Fig. 4A,A′). Electroporation of the Fgf8b expression plasmid into an embryo without previous manipulation reproducibly led to ectopic Wnt1 expression (Fig. 4B,B′) (n = 5 of 5). Mock-transfection with GFP or dsRed did not alter Wnt1 expression (Fig. 4G). Overexpression of Fgf8b in embryos already experiencing 6 h of GFP or RPTPλ-ΔIC expression also resulted in robust upregulation of Wnt1 expression (Fig. 4C,D) (n = 5 of 5 for GFP; n = 7 of 8 for RPTPλ-ΔIC). In contrast, when full-length RPTPλ was misexpressed in the mesencephalic vesicle (Fig. 4E), only 22% of the transfected embryos exhibited robust upregulation of Wnt1 after Fgf8b electroporation (Fig. 4F,F′) (n = 4 of 18). Because overexpression of RPTPλ at the MHB had specifically perturbed the expression of Wnt1 but not that of MHB marker genes like En1, Pax2, or Pax5 (supplemental Fig. 1, available at www.jneurosci.org as supplemental material), we analyzed expression of these molecules after sequential RPTPλ/Fgf8b electroporation. In agreement with our previous results, expression of all three molecules was strongly induced by Fgf8, regardless of previous overexpression of RPTPλ (supplemental Fig. 1L–O, available at www.jneurosci.org as supplemental material) (n = 6 of 7 for En1; n = 5 of 5 for Pax2; n = 5 of 5 for Pax5). Together, these results suggest that strong RPTPλ expression specifically interferes with the ability of cells in the embryonic mesencephalon to initiate Wnt1 expression in response to Fgf8 signals. Considering that, during normal development (or in response to local Fgf8 application), RPTPλ expression always extended over the Wnt1 domain at its rostral side, our data suggest that RPTPλ may function to rostrally restrict the domain of Wnt1 expression at the MHB.
If RPTPλ expression at the MHB indeed serves to restrict Wnt1 expression, one would expect that reducing RPTPλ activity at the MHB should allow Wnt1 expression to expand anteriorly. To test this idea, two RNAi constructs targeted to nonoverlapping sequences in the intracellular domain of RPTPλ were designed. The activity and specificity of the constructs were confirmed by luciferase reporter assays compared with the effects of an RNAi construct targeted to an unrelated protein or with that of RNAi constructs containing randomized targeting sequences (Fig. 5A). Repression by unrelated or randomized sequences was negligible (Fig. 5A). Electroporation of either RPTPλ_si-a (n = 8 of 9) (Fig. 5D) or RPTPλ_si-b (n = 6 of 9) (data not shown) but not of the randomized control targeting construct random_a (n = 5 of 5) (Fig. 5C) into the MHB region at HH10 markedly reduced the endogenous expression of RPTPλ. Knock-down of RPTPλ also led to an anterior expansion of Wnt1-positive cells into the domain where RPTPλ is normally expressed (Fig. 5G–G″,I,J) (n = 9 of 14). This effect was more pronounced near the dorsal midline than in the ventral midbrain, presumably because the domain of RPTPλ expression is broader and any effect of knock-down of its expression may therefore be more profound in the dorsal midbrain (Fig. 5E,K). Wnt1 expression at the MHB, however, was not only expanded rostrally, but expression was also scattered within the rostral half of its normal domain in some embryos (Fig. 5F–J). We do not know the molecular basis for this effect at present. A possible explanation may come from the fact that RPTPλ and Wnt1 expression at the MHB appear to slightly overlap during normal development (Figs. 1C-D′, 5E). It is therefore possible that RPTPλ stabilizes the Wnt1 expression border in the small region where both proteins are normally coexpressed. In addition, because the extracellular part of RPTPλ contains multiple cell adhesion-like motives, reduced RPTPλ expression at the MHB may alter cell adhesive properties critical for the maintenance of the coherent cellular organization within its normal expression domain. It is worth pointing out that disorganization of the Wnt1 expression domain was also observed after targeted deletion of Fgf receptor 1 from the MHB territory (Trokovic et al., 2003).
At least three signaling pathways, the Ras-MAP kinase, phosphatidyl inositol 3 kinase, and phospholipase Cγ pathway, can be activated after Fgf receptor stimulation, and of the three, the Ras-MAP kinase pathway has been demonstrated to be important for mid-hindbrain development (Corson et al., 2003; Sato and Nakamura, 2004; Dailey et al., 2005). Because signal transduction in each of these pathways requires protein phosphorylation on tyrosine residues, it is possible that RPTPλ, because of its tyrosine phosphatase activity, interrupts signal transduction through dephosphorylation of intermediate signaling components. To test whether RPTPλ impinges on Ras-MAP kinase signaling, we visualized phosphorylation of the pathway intermediate Erk1/2 with the phosphorylation-specific antibody dpErk1/2 (Corson et al., 2003). Six hours after transfection of 2 μg/μl RPTPλ-pMES but not after transfection of pMES alone, dpErk1/2 labeling intensity was slightly reduced in the right electroporated side of the electroporated embryos (Fig. 5N–P) (n = 6 of 9). Repression of Wnt1 expression by RPTPλ could also be detected 6 and 8 h after RPTPλ transfection (Fig. 5Q–T) (n = 10 of 18). In contrast to Wnt1 but in agreement with our previous results (supplemental Fig. 1, available at www.jneurosci.org as supplemental material), we did not observe any major disturbance of En1 expression 6 or 8 h after RPTPλ transfection (Fig. 5U,V) (data not shown) (n = 11 of 11). These findings indicate that the influence of RPTPλ on Wnt1 expression at the MHB may, at least in part, result from modulation of Ras-MAP kinase signaling intensity by RPTPλ.
RPTPλ interacts with β-catenin in the embryonic midbrain and inhibits the activity of a β-catenin responsive promoter
RPTPλ expression in the developing midbrain is highly dynamic during the first 4 d of chick development (Figs. 1, 6A–C) (Badde et al., 2005). Before HH12 (16 somites), RPTPλ is expressed throughout the mesencephalic vesicle (Fig. 6A) (Badde et al., 2005). Between HH12 and HH19, RPTPλ expression is lost from the midbrain alar plate but is maintained in a tight transverse domain rostral to the Wnt1 domain (Figs. 1A,C, 6B) (Badde et al., 2005). Thereafter, RPTPλ expression is induced once again in the mesencephalic vesicle and covers the entire mesencephalic alar plate from late E4 onwards, without crossing into the isthmic organizer region (Figs. 1D, 6C, arrowhead) (Badde et al., 2005). Because of this prominent “on-off” expression in the midbrain alar plate, we speculated that RPTPλ may contribute to midbrain development by mechanisms in addition to restricting Wnt1 expression at the MHB organizer.
The juxtamembrane domain of RPTPλ exhibits a high degree of homology to the cytoplasmic domain of E-cadherin (Wang et al., 1996). Moreover, the human homolog of RPTPλ, PCP2, colocalizes with β-catenin at sites of cell-to-cell contact in human adenocarcinoma cells and coprecipitates with β-catenin in PC-12 cells overexpressing both proteins (Yan et al., 2002). The interaction of both proteins occurs independently of the phosphorylation state of β-catenin, but β-catenin can be dephosphorylated on tyrosine residues by PCP2 (Yan et al., 2002). We therefore investigated whether β-catenin and RPTPλ may physically interact in the developing mesencephalon. We overexpressed either RPTPλ-HA or RPTPλ-ΔIC-HA in the mesencephalic vesicle for 24 h and subsequently immunoprecipitated RPTPλ bound proteins using an antibody directed against the HA-epitope (Fig. 6D,E). Overexpressed RPTPλ-HA was detected as a discrete band of ∼185 kDa recognized by the HA-specific antibody (Fig. 6D, top panel, lysate). After stripping and reprobing of the blot, endogenous β-catenin expression could also be readily detected (Fig. 6D, bottom panel, lysate). Precipitation with an unspecific antibody or with agarose-coupled protein-G did not deplete RPTPλ-HA or β-catenin from the extracts (Fig. 6, preclear). After immunoprecipitation with the HA-specific antibody, the 185 kDa band was cleared from the supernatant (Fig. 6D, top panel, sup) and enriched in the immunoprecipitate (Fig. 6D, top panel, IP) (n = 5 of 5). β-Catenin was also present in the immunoprecipitates (Fig. 6D, bottom panel, IP) (n = 5 of 5). Neither β-catenin nor RPTPλ could be precipitated with an antibody not directed against HA or β-catenin (data not shown). When the truncated allele RPTPλ-ΔIC-HA, which lacks the cytoplasmic part of the protein and thus the putative interaction domain with β-catenin, was electroporated into the mesencephalic vesicle, a specific band of 105 kDa was detected (Fig. 6E, top panel, lysate). After immunoprecipitation with the HA-specific antibody, RPTPλ-ΔIC-HA could be successfully immunoprecipitated but failed to enrich β-catenin (Fig. 6E, IP) (n = 5 of 5). These data demonstrate that RPTPλ can coprecipitate β-catenin from mesencephalic extracts via its intracellular domain.
We next investigated whether the observed interaction of β-catenin with RPTPλ was biologically relevant for canonical Wnt signaling. β-Catenin plays an essential role in the structural organization of cells by linking the cytoplasmic domain of type I cadherins, such as E-cadherin, to the actin cytoskeleton and, in addition, functions as a key component of the canonical Wnt signaling pathway. Loss of E-cadherin-mediated cell adhesion increased canonical Wnt signaling, presumably because, in the absence of E-cadherin, more β-catenin is available for Wnt signal transduction (Orsulic et al., 1999; Stockinger et al., 2001). We therefore wondered whether overexpression of RPTPλ in the mesencephalon may sequester a portion of β-catenin at the plasma membrane, making it unavailable for Wnt signaling. To test this idea, we examined the activity of the β-catenin responsive reporter TOP:dgfp in the midbrain, alone or in the presence of ectopically expressed RPTPλ. TOP:dgfp contains four consensus Lef-binding sites and a minimal c-Fos promoter-driving expression of a destabilized form of GFP (dGFP) (Dorsky et al., 2002). Transgenic zebrafish that stably express dGFP under control of this promoter exhibit intense GFP expression throughout the entire midbrain alar plate, demonstrating that the canonical Wnt/β-catenin pathway is active in this region of the neural tube (Dorsky et al., 2002). When the TOP:dgfp reporter was introduced into the MHB region of HH11 chick embryos by in ovo electroporation, strong GFP expression in the midbrain alar plate could be observed 24 h later (at HH15–16), demonstrating that the midbrain alar plate is a domain of active Wnt/β-catenin signaling in the chick at this developmental age (Fig. 6F–H,N) (n = 8 of 8). Therefore, at HH15–HH16, neither RPTPλ nor Wnt1 are expressed in the mesencephalic alar plate, but the Wnt/β-catenin pathway is active. We therefore reasoned that misexpression of RPTPλ in the midbrain alar plate at HH15–H16 should allow us to analyze a possible influence of RPTPλ on canonical Wnt signal transduction that is independent of its ability to influence Wnt1 expression at the MHB. When RPTPλ was coelectroporated with TOP:dgfp, GFP fluorescence intensity was reduced severely (Fig. 6I–K,N) (n = 10 of 11). Transfection of RPTPλ-ΔIC, which lacks the intracellular domain critical for coprecipitation of β-catenin, did not suppress GFP fluorescence (Fig. 6L–N) (n = 6 of 7). These results demonstrate that RPTPλ can suppress the activation of this β-catenin responsive reporter in vivo.
The normal growth and development of the mid-hindbrain region depends on activation of the canonical Wnt signaling pathway (McMahon and Bradley, 1990; Thomas and Capecchi, 1990; Brault et al., 2001; Panhuysen et al., 2004). To investigate whether ectopic RPTPλ expression or RNAi-mediated knock-down of expression altered midbrain growth, RPTPλ expression was targeted to the right side of the mesencephalic vesicle by in ovo electroporation into HH9 embryos, and the consequence of this manipulation on midbrain growth was assessed 24 h later (Fig. 7). We chose HH9 embryos for these experiments, because this developmental time is before the downregulation of endogenous RPTPλ expression in the mesencephalon, which occurs after HH12 (Fig. 6A–C) (Badde et al., 2005). Introduction of RPTPλ-pMES into the HH9 mesencephalic vesicle therefore maintains RPTPλ expression in the midbrain beyond HH12, whereas introduction of an RNAi targeting construct leads to precocious loss of RPTPλ expression. The homeodomain Meis2 was used as a marker for the tectal anlage, because its transcripts are excluded from the roof plate, allowing us to unambiguously determine the location of the dorsal midline and estimate the size of the left and right halves of the mesencephalic vesicle. After forced expression of GFP or of the truncated allele RPTPλ-ΔIC-HA together with GFP, the right and left halves of the tectal anlage appeared to have normal size (Fig. 7A,B) (n = 5 of 5 for GFP; n = 6 of 6 for RPTPλ-ΔIC-HA). However, in embryos that had been electroporated with an expression plasmid containing RPTPλ, the size of the right electroporated tectal anlage was reduced severely when compared with the left, nonelectroporated side of the same embryo (Fig. 7C,C′) (n = 15 of 16). Conversely, the right tectal anlage was slightly enlarged in embryos transfected with the specific RNAi targeting construct RPTPλ_si-a (Fig. 7D) (n = 6 of 7).
To determine whether this growth reduction was the result of decreased cell proliferation and/or enhanced cell death, BrdU pulse-labeling experiments were performed. We delivered RPTPλ-HA using the electroporation vector pMES, which contains an IRES-GFP cassette allowing us to visualize individual RPTPλ-HA transfected cells by their GFP fluorescence (Swartz et al., 2001). Two hours after BrdU injection, 51% (±1.85 SEM) of the cells forced to express GFP, but only 27.2% (±1.41 SEM; p < 0.002, Student's t test) of RPTPλ/GFP expressing cells had incorporated the BrdU label (Fig. 7E–F″). Similar to mock transfected or nontransfected embryos, apoptotic cell nuclei were rarely observed 24 h after misexpression of RPTPλ, indicating that RPTPλ electroporation does not compromise cell survival within this time frame (Fig. 7G–I) (n = 3). Knock-down of RPTPλ expression at HH9 slightly increased the percentage of mitotic cells, visualized by detecting pH3 from 11.7% (±1.29 SEM) after introduction of GFP alone to 14.2% (±1.45 SEM) after transfection of the RNAi targeting construct RPTPλ_si-a together with GFP (Fig. 7J,K) (supplemental Fig. 2E, available at www.jneurosci.org as supplemental material) (n = 4). In conclusion, these results suggest that RPTPλ functions as negative modulator of mesencephalic growth primarily through modulation of progenitor cell proliferation.
Because RPTPλ could precipitate (and possibly sequester) the Wnt-signaling intermediate β-catenin from mesencephalic extracts (Fig. 6D,E), we wondered whether cotransfection of β-catenin or Wnt1 together with RPTPλ would rescue the proliferation defect observed after introduction of RPTPλ alone. To test this idea, we introduced either pMES (expressing only GFP), RPTPλ-pMES (expressing RPTPλ together with GFP), RPTPλ-Wnt1-pMEC together with GFP-pMIW (which results in coexpression of RPTPλ, Wnt1, and GFP), or RPTPλ-pMES together with caβ-catenin-pMIW (which results in coexpression of RPTPλ, GFP, and a stabilized form of β-catenin) into the mesencephalic vesicle at HH9 and determined the mitotic index of midbrain progenitor cells 24 h later (Fig. 7L,M) (supplemental Fig. 2A–D, available at www.jneurosci.org as supplemental material) (n = 9 for pMES, RPTPλ-pMES; n = 3 for RPTPλ-Wnt1-pMEC, RPTPλ-pMES plus β-catenin-pMIW). Transfection of RPTPλ together with GFP significantly reduced the percentage of mitotic cells compared with GFP transfected controls. Cotransfection of Wnt1 or stabilized β-catenin together with RPTPλ and GFP partially restored the mitotic index (Fig. 7M, gray bars). Therefore, the negative influence of RPTPλ on midbrain progenitor cell proliferation could, at least in part, be alleviated by cotransfection of Wnt1 or of the Wnt1-signaling intermediate β-catenin. To determine whether RPTPλ modulates midbrain progenitor cell proliferation in a cell-autonomous or noncell-autonomous manner, we compared the percentage of pH3-positive cell nuclei in the transfected (GFP expressing) cell population under each of these experimental conditions with the percentage of pH3+ cells in nontransfected (GFP-negative) neighboring cells (Fig. 7M, black bars). The mitotic index of cells that were located in close proximity to cells transfected with RPTPλ plus GFP did not differ from that of the neighbors of GFP transfected cells, suggesting that RPTPλ acts cell autonomously on midbrain progenitor cell proliferation. We also did not observe a significant difference in the percentage of pH3+ cell nuclei in neighbors of RPTPλ/GFP/β-catenin transfected cells compared with neighbors of GFP-expressing cells. In contrast, the percentage of pH3+ nuclei in cells located near RPTPλ/Wnt1/GFP transfected cells was clearly increased. This observation is consistent with the notion that the secreted molecule Wnt1 may influence cell proliferation in a cell-nonautonomous manner.
Our finding that overexpression of RPTPλ interfered with Fgf8-mediated induction of Wnt1 expression at the MHB as well as with the activation of the Wnt/β-catenin-dependent TOP:dgfp reporter in the mesencephalon raises the question of whether the reduced TOP:dgfp reporter activity in RPTPλ transfected embryos may be a secondary consequence of their diminished Wnt1 expression or, as suggested above, may reflect a more direct inhibition of Wnt/β-catenin signaling, presumably through interaction of RPTPλ with β-catenin. To distinguish between these possibilities, we first decreased Wnt1 expression at the MHB through RNAi-mediated knock-down and monitored TOP:dgfp activity in the mesencephalic vesicle 24 h later (supplemental Fig. 3A–H′, available at www.jneurosci.org as supplemental material) (n = 15). Although electroporation of Wnt1-specific RNAi targeting constructs reduced Wnt1 expression in vivo, GFP expression of the TOP:dgfp reporter appeared normal. Second, we blocked Wnt signaling in the midbrain through forced expression of the known signaling antagonist Axin (Zeng et al., 1997). Wnt1 expression at the MHB was normal 24 h after transfection (supplemental Fig. 3I,J, available at www.jneurosci.org as supplemental material) (n = 10). Based on these observations, we suggest that RPTPλ may modulate Wnt1 expression and signal transduction in the midbrain through two separate mechanisms.
The midbrain alar plate undergoes massive expansion between HH12 and HH19 when RPTPλ expression is absent from this tissue. Cells around the MHB in embryonic day 10.5 (E10.5) to E11.5 mouse embryos are less mitotically active than cells in the midbrain and rhombomere 1 (Trokovic et al., 2005). A similar domain of reduced proliferation also exists at the chick MHB at HH15 and HH20 (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). Although the ring of RPTPλ expression seen during this developmental period does not cover this domain entirely, RPTPλ expression nevertheless appears to localize within this domain.
Here, we present evidence based on gain-of-function and knock-down experiments that RPTPλ can counteract transcriptional activation of Wnt1 in the embryonic midbrain. In addition, we show that RPTPλ can physically interact with β-catenin in the midbrain, inhibit the activity of a β-catenin responsive promoter, and suppress mesencephalic cell proliferation. RPTPλ therefore emerges as a potential modulator of Wnt1 expression and signaling at the MHB organizer.
RPTPλ is part of the complex molecular interactions that maintain and shape the MHB region
A complex spatiotemporal pattern of gene expression has been described at the developing mid-hindbrain boundary. The MHB is bordered by a narrow, transverse ring of Wnt1 expression at its anterior side and by a ring of Fgf8 expression at its posterior side. Several proteins, including Pax2/5 and En1/2, are expressed in nested, symmetric gradients across the MHB region. Fgf8 signaling induces activation of the Ras-MAP kinase pathway, which is required for isthmic organizer activity (Sato and Nakamura, 2004). Fgf feedback inhibitors, such as Spry2, Sef, and Mkp3, interfere with Ras-MAP kinase signaling downstream of receptor tyrosine kinases at different steps within the signal transduction pathway and are thought to limit the lateral expansion of the Fgf signal at the MHB (Gross et al., 2001; Nutt et al., 2001; Fürthauer et al., 2002; Lin et al., 2002; Tsang et al., 2002; Kawakami et al., 2003; Echevarria et al., 2005). Maximal Ras-MAP kinase activation occurs within the domain of Fgf8 expression but tapers off into the mesencephalic and metencephalic vesicles (Corson et al., 2003; Sato and Nakamura, 2004). The shapes of the Pax2/5 and En1/2 expression domains reflect the gradient of Ras-MAP kinase activity at the MHB. In contrast, Wnt1 expression is confined to a tight ring of cells in the mesencephalon and is absent from the metencephalon. Metencephalic tissue is not permissive for Wnt1 expression, explaining the asymmetric nature of the Wnt1 expression domain (Kikuta et al., 2003). It is yet unclear, however, as to why Wnt1 expression is restricted to such a discrete band of cells and does not, like that of Pax2/5 or En1/2, taper off into the mesencephalic vesicle. As we show here, the receptor tyrosine phosphatase λ is expressed in a ring of cells rostral to the Wnt1 domain and inhibits Wnt1 expression when overexpressed. RPTPλ thus has the potential to restrict Wnt1 expression to its characteristic tight ring at the MHB.
RPTPλ expression itself is induced by Fgf8b in a concentration-dependent manner. When Fgf8-induced expression of Wnt1 and RPTPλ was monitored successively in the same specimen, ectopic expression of RPTPλ could always be observed anterior to the ring of Wnt1 expression. Induction of Wnt1 and RPTPλ expression by Fgf8 thus mimics their normal spatial relationship at the MHB, where the Wnt1 expression also separates the domain of RPTPλ expression from that of Fgf8. In addition, after implantation of Fgf8b releasing beads, ectopic RPTPλ expression could be observed at a Fgf8b concentration that was not sufficient to induce Wnt1 expression. We conclude from these results that different dosages of the Fgf8 signal from the MHB are required for transcriptional induction of RPTPλ and Wnt1, respectively. Examples for dosage dependency of Fgf function during embryogenesis have been described previously and include telencephalic and cerebellar development, patterning of the foregut, and skeletal development (Sato et al., 2001; Storm et al., 2003, 2006; Hajihosseini et al., 2004).
RPTPλ specifically inhibited Wnt1 expression but not that of other MHB marker genes when overexpressed in the MHB region. In addition, when cells in the midbrain were forced to express RPTPλ, they failed to upregulate Wnt1 expression after subsequent stimulation by Fgf8, yet they readily expressed En1, Pax2, or Pax5 under identical experimental conditions. RPTPλ therefore appears to specifically interfere with Fgf8-induced transcriptional activation of Wnt1 at the MHB. How can this high degree of functional specificity be achieved mechanistically? Expression of many MHB marker genes, including Otx2, Gbx2, Pax2/5, En1/2, and Wnt1, is negatively affected by overexpression of a dominant-negative form of Ras (Sato and Nakamura, 2004). Yet, we observed recently after transfection of different concentrations of a constitutive active form of Ras (caRas) that transcriptional activation of Wnt1 at the MHB could only be induced at a narrow concentration range of caRas, whereas transcriptional activation of En1 was observed over a concentration range spanning more than an order of magnitude (Vennemann et al., 2008). Induction of Wnt1 at the organizer thus appears to be more sensitive to changes in the strength of Ras-MAP kinase pathway activation than other MHB marker genes. This raises the possibility that tight, local regulation of Ras-MAP kinase activity may be a prerequisite for Wnt1 expression at the MHB and that RPTPλ may contribute to this regulation through modulation of Ras-MAP kinase signaling. Notably, in mouse embryos in which Fgf8 expression was specifically lost in the MHB territory from the five somite stage onwards, the transverse ring of Wnt1 expression in the caudal mesencephalon was also absent, whereas the expression domains of Fgf8, Otx2, Pax2, or En1 appeared normal at early stages of embryogenesis (Chi et al., 2003). Genetic loss of Fgf8 at the MHB, however, caused massive cell death, an effect we did not observe within 24 h after RPTPλ transfection. We cannot rule out that RPTPλ inhibits Wnt1 expression by interfering with still unknown factor(s) that are specifically required for transcriptional Wnt1 activation in Fgf8 responding cells. Regardless of the exact molecular nature of this inhibition, the experiments described here suggest a model in which Fgf8 induces Wnt1 and RPTPλ expression in the midbrain at different signaling thresholds and consequently at different distances to the MHB, so that the Wnt1 domain is always bordered by RPTPλ-expressing cells at its rostral side. One function of RPTPλ at the MHB may thus be to act as a Ras-MAP kinase feedback inhibitor to restrict the anterior expansion of the transverse Wnt1 domain (Fig. 8).
RPTPλ binds to β-catenin and modulates canonical Wnt1 signaling at the MHB
In addition to restricting Wnt1 expression at the MHB, RPTPλ also seems to play a role in modulating Wnt1 signal transduction. In the canonical or Wnt/β-catenin pathway, Wnt binding to frizzled receptors and LRP coreceptors inhibits degradation of the cytoplasmic pool of β-catenin, allowing it to translocate to the nucleus, where it acts as coactivator for LEF/TCF transcription factors to promote target gene expression (Logan and Nusse, 2004). In addition to its role in Wnt signaling, β-catenin is a structural link between cadherins/α-catenin and the actin cytoskeleton. Cadherin bound β-catenin can be freed to participate in Wnt signaling and vice versa, a switch that appears to be regulated by phosphorylation of β-catenin on tyrosine residues (Brembeck et al., 2004; Nelson and Nusse, 2004; Schambony et al., 2004; Lilien and Balsamo, 2005). It was shown previously that β-catenin interacts with the juxtamembrane domain of the human homolog of RPTPλ and can be dephosphorylated by this protein in vitro (Yan et al., 2002). As shown here, a similar interaction of RPTPλ and β-catenin also occurs in the embryonic chick midbrain. In addition, when full-length RPTPλ was overexpressed, we found that two functional readouts of canonical Wnt signaling in the midbrain, activation of a β-catenin responsive promoter and neuroepithelial cell proliferation, were reduced. This proliferation defect could be partially reversed by cotransfection of Wnt1 or of a stabilized form of β-catenin. It is intriguing to speculate that RPTPλ may antagonize canonical Wnt signaling through binding and potentially dephosphorylation of β-catenin and ultimately serve as a negative regulator of cell proliferation in the developing midbrain (Fig. 8). Decreasing Wnt1 expression at the MHB through transfection of Wnt1-specific RNAi targeting constructs did not notably reduce β-catenin responsive reporter activity in the midbrain, nor did misexpression of Axin, a known inhibitor of the Wnt signaling pathway, in the mesencephalon disturb the transverse ring of Wnt1 expression at the organizer. These observations suggest that RPTPλ acts on Wnt1 expression and signal transduction through two independent mechanisms.
How can this dual effect be explained? Protein tyrosine phosphatases antagonize the activity of protein tyrosine kinases and influence a broad spectrum of physiological processes. For instance, RPTPλ (also termed RPTPψ) is required for Delta/Notch-dependent oscillatory gene expression in the presomitic mesoderm as well as for convergent extension during gastrulation through an as yet unknown mechanism (Aerne and Ish-Horowicz, 2004). It is therefore possible that RPTPλ impinges on multiple signaling pathways through dephosphorylation of different signaling components. In addition, the ring of RPTPλ expression at the MHB may coincide with that of another receptor tyrosine phosphatase, RPTPζ/β, a keratan sulfate modified protein, which has been suggested to modify signal transduction at the MHB through influencing the diffusion of morphogenetic signals of the Wnt and Fgf families (Canoll et al., 1993; Hamanaka et al., 1997). Although the precise role of RPTPλ in any of these processes will only be fully understood once its substrates are identified, our results suggest that RPTPλ is part of the complex molecular network that governs development of the midbrain region.
This work was supported by the International Max-Planck Research School Program in Structure and Function of Biological Membranes (Frankfurt, Germany) (A.B.). We thank C. Tabin (Harvard Medical School, Boston, MA), C. Logan (University of Calgary, Calgary, Alberta, Canada), C. Krull (University of Michigan, Ann Arbor, MI), H. Rohrer (Max-Planck Institute for Brain Research, Frankfurt, Germany), F. Costantini (Columbia University Medical Center, New York, NY), and R. Dorsky (University of Utah, Salt Lake City, UT) for reagents; C. Ziegler for excellent technical assistance; G. O'Sullivan for critically reading this manuscript; and A. Vennemann, E. Dohle, and A. Poplawski for experimental help.
- Correspondence should be addressed to Dorothea Schulte, Department of Neuroanatomy, Max Planck Institute for Brain Research, Deutschordenstrasse 46, 60528 Frankfurt, Germany.