Volume 25, Issue 3, Pages 441-454 (February 2007) Balancing BMP Signaling through Integrated Inputs into the Smad1 Linker  Gopal Sapkota, Claudio Alarcón, Francesca M. Spagnoli, Ali H. Brivanlou, Joan Massagué  Molecular Cell  Volume 25, Issue 3, Pages 441-454 (February 2007) DOI: 10.1016/j.molcel.2007.01.006 Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 1 FGF Inhibition of BMP Signaling Requires Smurf1 (A) Alignment of the linker regions of R-Smads and identification of conserved sites of interest. (B) Activity of a BMP/Smad1 reporter construct in WT and SMAD1L MEFs in the presence of BMP2 and/or basic FGF. Renilla luciferase was used for normalization. (C) Immunoblot analysis of WT and Smad1L MEFs treated with BMP and/or FGF using antibodies against the indicated proteins and phosphoproteins. P-ERK, phosphorylated, activated ERK. (D) qRT-PCR analysis of FoxO4 or Smurf1 mRNA levels in cells transfected with either an siRNA against FoxO4 or an siRNA against Smurf1. (E and F) The BMP reporter activity in MEFs or C2C12 cells transfected with the indicated siRNAs. (G) C2C12 cells were transfected with siFoxO4 or siSmurf1 and, 48 hr later, treated with the indicated additions. Expression of BMP target genes ID1 and SnoN was analyzed by qRT-PCR. (H and I) C2C12 cells transfected either with siFoxO4 or siSmurf1 were incubated with the indicated additions for 48 hr. Alkaline phosphatase activity was measured enzymatically (H) and immunocytochemically (I). Data are presented as mean ± SD (n = 3) (B and D–H). Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 2 Smad1 Phosphorylation at the Linker Region Downstream of Various Pathways (A) HaCaT cells deprived of serum for 16 hr were treated with the indicated stimuli and lysed. The Smad1 immunoprecipitates or whole lysates were subjected to western immunoblotting analyses with antibodies against Smad1, phospho-Ser206 Smad1, tail-phosphorylated Smad1 (Smad1-TP), or activated forms of the MAPKs ERK, JNK, and p38. (B) HaCaT cells were incubated with BMP and/or EGF in the presence or absence of the MEK inhibitor U0126, added 1 hr prior to growth factor additions. (C) HaCaT cells were treated with BMP or exposed to UV radiation in the presence or absence of p38 MAPK inhibitor (SB203580) and/or JNK inhibitor (SP600125). (D) HaCaT cells were incubated with BMP or EGF for the indicated times, and lysates were immunoblotted with the indicated antibodies. (E) HaCaT cells were treated with BMP (45 min), EGF (10 min), EGF followed by BMP, or BMP followed by EGF and separated into nuclear and cytosolic fractions, which were analyzed for the presence of linker- and tail-phosphorylated Smad1. Histone H1B (nuclear) and α-tubulin (cytosolic) served as markers of cell fractionation. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 3 Smurf1 Selectively Binds to Linker-Phosphorylated Smad1 (A) Scheme of the in vitro binding assay to determine interactions between GST-Smad1 or ERK-phosphorylated GST-Smad1 (GST-Smad1-PL) and test proteins. (B) GST-Smad1 or GST-Smad1-PL immobilized to GSH beads was incubated with HEK293 lysates expressing the indicated HA- or Flag epitope-tagged proteins for 2 hr in the cold. Bound proteins were eluted from the beads and detected by western immunoblotting with the indicated antibodies. The incorporation of 32P was detected by autoradiography, and the amount of GST-Smad1 was detected by Coomassie staining. (C) As above, except that immobilized GST-Smad1 WT or GST-Smad1 mutated to alanine (EPSM) or aspartic acid (EPAM) at the four MAPK sites was tested for its ability to bind Flag-Smurf1. (D) As in (B), except that WT GST-Smad1 or GST-Smad1[AAAY] (AY) mutant, phosphorylated or not phosphorylated by ERK, was subjected to interaction assay with Flag-Smurf1. (E) HEK293 cells transfected with HA-Smurf1 were incubated with BMP or EGF. To determine interaction between Smurf1 and Smad1 in vivo, HA-Smurf1 immunoprecipitates were subjected to western immunoblotting with the indicated antibodies. Immunoblots of lysate inputs with Smad1-TP, P-ERK1/2, Smad1, and HA antibodies were run to verify BMP and EGF stimulations as well as loading controls. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 4 Linker Phosphorylation of Smad1 Enables Smurf1-Mediated Ubiquitination (A) HEK293 cells were cotransfected with vectors encoding HA-tagged ubiquitin and Flag-tagged Smad1 (WT or EPSM mutant), and the cells were lysed 36 hr later. The lysates or Flag immunoprecipitates were subjected to western immunoblotting with anti-HA or anti-Flag antibodies. (B) Transfected 293 cells expressing Flag-Smad1, the Flag-Smad1[EPSM], or Flag-Smad1[AAVA] were stimulated with EGF or BMP for 1 hr. Western immunoblotting with P-PXS∗P demonstrated constitutive linker phosphorylation of the overexpressed Smad1, independent of BMP or EGF stimulation or ERK activity. (C) As in (A), except that HA-Smurf1 (WT) or the catalytically inactive mutant HA-Smurf1[DD] was cotransfected with Flag-Smad1 (WT or EPSM) and HA-ubiquitin. (D) The indicated Flag-Smad1 mutants were coexpressed with HA-Smurf1 and Myc-ubiquitin. Flag immunoprecipitates or lysates were subjected to western immunoblotting with the indicated antibodies. AAAY, Smad1 mutated at the PY motif (PPAY to AAAY mutation). AAVA, Smad1 mutated at the C-terminal receptor phosphorylation sites (SSVS mutated to AAVA). (E) As in (A), except that HA-ubiquitin was coexpressed with the indicated mutants of Flag-Smad1 to assess the ability of each mutant to be ubiquitinated. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 5 ERK-Mediated Phosphorylation Primes the Smad1 Linker for Phosphorylation by GSK-3 (A) Immobilized GST-Smad1, ERK phosphorylated or not ERK phosphorylated, was incubated with GSK-3β and γ-32P-ATP. The proteins were run on SDS-PAGE, Coomassie stained, and subjected to autoradiography. (B) As in (A), except that the indicated GST-Smad1 mutants were used as substrates for GSK-3. (C) As in Figure 3A, except that ERK/GSK-3-phosphorylated GST-Smad1 was used in the interaction assay with HA-Nup214 and HA-Smurf1. An antibody raised against the Smad1-phospho-Thr202 was used as a control for phosphorylation of Thr202 residue by GSK-3. (D) Extracts from 293 cells transfected with Flag-Smad1, Flag-Smad1[EPSM], or Flag-Smad1[GPSM] were immunoblotted with Smad1-phospho-Thr202 and Smad1-phospho-Ser206 antibodies (GPSM, GSK-3 phosphorylation sites in Smad1 mutated as follows: S210A, T202V, and S191A). (E) As in Figure 4A, except that Flag-Smad1[WT], Flag-Smad1[EPSM], or Flag-Smad1[GPSM] mutants were cotransfected with HA-ubiquitin and the relative ubiquitination was analyzed by immunoblotting Flag immunoprecipitates with anti-HA antibody. Flag immunoblotting was used as loading control. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 6 Smurf1 Limits Phospho-Smad1 Accumulation In Vivo in Xenopus Embryogenesis (A) Neural phenotype in embryos injected at eight-cell stage into both dorsal-animal blastomeres with 30 ng of Smurf1-MO. Smurf1-MO-injected embryos at late neurula stage show defects in neural tube closure (open arrows) and reduced pigmentation in the anterior neural plate (asterisks). Late phenotypes including reduction in telencephalic structures (arrowhead), micropthalmy (open arrowheads), and shortening of the trunk are visible in injected embryos at tadpole stage. (B) Two-cell-stage embryos were injected into the animal pole of both blastomeres with 30 ng of Smurf1-MO or were left uninjected. Lysates from whole embryos at different developmental stages were analyzed by immunoblotting with antibodies against the indicated targets, including anti-α-tubulin as loading control. (C) Explants from prospective neural ectoderm of uninjected and Smurf1-MO-injected embryos were dissected at stage 10, cultured, lysed at stage 14, and subjected to immunoblotting assays. The results from two separate experiments are presented. (D) The intensity of the tail-phosphorylated and linker-phosphorylated signals in (C) was quantified and normalized relative to the total Smad1 signal. Error bars represent deviation from the mean of two independent experiments. (E) Two-cell-stage embryos were left uninjected, or 100 pg XSmurf1 mRNA was injected into each blastomere with indicated amounts of Smad1[WT] or Smad1[EPSM] mRNAs. Animal cap explants were analyzed at neurula stage 18 by RT-PCR for expression of XAG and Otx2, while ornithine decarboxylase (ODC) served as control. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions

Figure 7 Smurf1 Binding Inhibits Nucleoporin Interaction and Nuclear Translocation of Smad1 (A) GST-Smad1 proteins on GSH beads were incubated with HA-Smurf1 or Flag-Smurf1, and then incubated in lysates expressing Flag-Smurf1 or HA-Nup214, respectively. Bound proteins were eluted and subjected to immunoblotting with anti-HA or anti-Flag antibodies. The amount of GST-Smad1 was analyzed by Coomassie staining. (B) HEK293 cells expressing Flag-Smad1, alone or together with HA-Smurf1[DD], were incubated with BMP. Nuclear and cytosolic fractions were separated and analyzed for the presence of Smad1-TP and Flag-Smad1. Histone H1B and α-tubulin served as cell fractionation controls. (C) HeLa cells expressing HA-Fyn-Smurf1, which is targeted to cytoplasmic locations including focal adhesions, were treated with EGF, BMP, or EGF followed by BMP. Cells were immunostained with anti-HA or anti-Smad1 antibodies. DAPI staining was used to visualize nuclei. Image merging demonstrates colocalization of HA-tagged protein and endogenous Smad1 in EGF-treated cells. (D) Cells expressing Fyn-Smurf1[DD] treated with BMP or EGF and BMP from multiple fields from four slides were analyzed for Smad1 distribution and are plotted as a percentage of cells expressing Smad1 in both cytosolic and nuclear fractions or only in the nuclear fraction for each condition. (E) Model depicting the molecular interactions and possible fates of linker-phosphorylated Smad1. Linker phosphorylation triggered by MAPK activators such as FGF and EGF primes Smad1 for recognition and polyubiquitination by Smurf1, and degradation. Bound Smurf1 additionally inhibits the interaction of Smad1 with nucleoporin Nup214, rendering Smad1 resistant to nuclear accumulation in response to BMP. In the absence of MAPK activation, BMP induces nuclear accumulation and activation of Smad1. However, this is followed by BMP-induced linker phosphorylation and Smurf1 binding in the nucleus, which targets Smad1 for degradation, providing negative feedback in the BMP pathway. Nuclear Smad1 C-terminal phosphatases (not depicted) also contribute to terminate Smad1 activation. Molecular Cell 2007 25, 441-454DOI: (10.1016/j.molcel.2007.01.006) Copyright © 2007 Elsevier Inc. Terms and Conditions