Figure 1.
TGFβ and BMP activated Smads bind to the same aa459–472 segment of (mouse) Sip1.
A) Sip1 SBD sequence conservation in five different vertebrates, and schematic representation of the peptide aptamers used in this study. The numbering shown for the aptamers at the right applies to the mouse protein sequence; the histograms reflect strong versus weaker conservation amongst the animal species. The aptamer A1 represents the insert in the Trx protein scaffold of the entire 51 aa-long SBD, as defined previously [12]; the aptamers A2–A8 are C- and/or N- terminal truncations of A1. B–D) Co-IP experiments in extracts from transfected HEK293T cells show that both TGFβ (B) and BMP (C,D) activated (Flag-tagged) Smads bind only to the aptamers A1, A3 and A4, indicating that the shared region shared between A3 and A4 is responsible for the interaction. E) Flag-tagged Smad3 and Smad1 proteins were specifically bound by a 14 aa-long sequence represented by the A6 aptamer (sequence in red in panel A), and comprehending the tandem repeat (QxVx)2 (for details, see main text). The activation of the TGFβ family pathway was assessed using p-p38MAPK or pAKT levels; Tubulin detection was used as loading control. Panel B was acquired using a Digital Chemiluminescence System (Bio-Rad; see Materials and Methods).
Figure 2.
3D modelling of the Sip1 SBD sequence and structural features of the 459–472 sequence.
Two orientations of the same model are depicted. Panels A and A′ represent the “cartoon model” highlighting the structural elements present in the SBD of WT Sip1. Panels B and B′ are surface representation with volume occupied by the domain. C and C′ show the combination of panels A/A′ and B/B′. The SBD structure was obtained after one-to-one threading process using the dvpra1a.pdb structure against the 417–503 sequence of Sip1. The overall structure of the SBD shows the presence of several α-helixes bundled and coiled-coil regions in a closed conformation. The (QxVx)2 sequence (as part of aa459–472; red/yellow α-helix panels A and A′) is on the solvent-exposed surface and two polar chains protrude from this α-helix, suggesting a potential role as mediator of the interaction with the hydrophobic corridor in the Smad MH2 domain. In blue: the previously defined SBD aa437–487; red: aa459–472 (with the (QxVx)2 sequence); yellow: Q461, V463, Q465 and V467 residues (see main text for further details). X,Y and Z axes are depicted showing the orientation of the two sets of panels.
Figure 3.
Effect of mutagenesis of specific residues within the Sip1 SBD.
Again, two orientations of the same model are depicted (see Figure 2): panels A and A′ represent the “cartoon model” highlighting the structural elements present in the Sip1 SBD; panels B and B′ are surface representation with volume occupied by the domain; C and C′ show the combination of panels A/A′ and B/B′. X,Y and Z axes are depicted showing the orientation of the two sets of panels. No alteration of the overall structural model is seen (compared to the structure presented in Figure 2), but the mutation significantly impairs the candidate contact surface with the Smad MH2 domain.
Figure 4.
Electrostatic properties of the predicted models for the SBD of Sip1 WT (panel A) and mutant (panel B).
Potential isocontours are displayed across a range of −5 to +5 kT/e, depicting negative potential areas (in red). Cartoon representations of the models are also shown, with key residues indicated in stick representation (panels A′ and B′).
Figure 5.
Overlay of Sip1 WT (cyan) and (AxAx)2 mutant (ochre) aa459–472 (panel A), highlighting the absence of polar chains protruding from the solvent-exposed surface, and the global SBD domains (panel B).
In A blue: WT SBD residues; lime: SBD residues that are mutated to A. Arrowheads indicate highly disordered regions of the two superimposed structures.
Figure 6.
Full-length Sip1(AxAx)2 mutant no longer binds to activated Smads, while its DNA-binding property is preserved.
A) Site-specific mutagenesis of the QxVx repeats (see Results) in the SBD of full-length (Myc-tagged) Sip1 causes loss of Smad interaction in ligand-activated cells (shown are Smad1 and Smad3). B) Sip1(AxAx)2 protein is predominantly nuclear, like WT Sip1. Lamin C was used as marker for the nuclear fraction and Tubulin serves as control to verify the presence of cytoplasmic contamination in the lysates. C) DNA-IP using a segment encompassing the Sip1-binding, E-box containing segment of the Xbra2 promoter demonstrated that Sip1(AxAx)2 is still able to bind to its cognate DNA target sequence [13].
Figure 7.
BMP/TGFβ-induced activation of SBE4-based luciferase based-promoter is maintained in cells overexpressing the (AxAx)2 mutant.
HEK293T cells were transfected with WT or SBD mutant Sip1 encoding plasmids in presence (+) or absence of a c.a.Alk3 (panel A) or c.a.Alk4 (panel B) receptor. Increasing concentrations of Sip1 WT plasmid result in a down regulation of BMP or TGFβ induced gene response, effect that is attenuated by co-transfection of the c.a.Alk3/Alk4 encoding plasmids. On the other hand, the Smad binding mutant Sip1(AxAx)2 is no longer able to repress such gene response. Blots show the expression levels of the Sip1 (WT and mutant) encoding plasmids transfected. Images were acquired using a Digital Chemiluminescence System (Bio-Rad).
Figure 8.
Sip1(AxAx)2 represses Cdh1 promoter based reporter activity in a c.a.Alk4/activated Smad-independent and dominant-negative, DNA-binding fashion.
A) Luciferase reporter assay showing the inhibition of the transcriptional activity of the Cdh1 promoter mediated by WT Sip1 (as already described in several studies). Upon transfection of the cells with a plasmid encoding the Sip1(AxAx)2 mutant, no repressive effect was observed on the co-transfected Cdh1 promoter in the absence of c.a.ALK4. In addition, the WT Sip1 dose-dependent repression in the presence of c.a.ALK4 is lost upon co-transfection of the Sip1(AxAx)2 mutant. B) Competition assay in cells co-transfected with WT Sip1 (human; Flag-tagged) and mutant Sip1(AxAx)2 encoding plasmids at different ratios (total amount of transfected DNA in all cases is 50 ng). Restoration of the levels of Cdh1 promoter activity is seen in the presence of the Sip1(AxAx)2 SBD mutant, which acts as a DNA-binding dominant-negative Sip1 (in terms of Smad interaction), suggesting that the transcriptional repression mediated by Sip1 is strictly Smad-dependent and requires Smads bound as co-factors. Error bars represent standard deviations.
Figure 9.
Smad-Sip1 interaction is active during TGFβ-induced EMT in NMe cells.
Reduction of endogenous Sip1/Zfhx1b levels using specific siRNAs (A–C) and rescue using WT Sip1 (D). Different concentrations of Sip1(AxAx)2 (E–G) or ZnF (F) mutant encoding plasmids do not show a marked inhibition of the EMT process in TGFβ1-stimulated cells. Only a significant reduction in the formation of actin stress fibers (graph in panel I) could be observed with increasing concentration of the Sip1(AxAx)2 mutant. Blue DAPI; green: F-actin; red: anti-Sip1 (panels A–C), anti-myc (panels D–G), anti-Flag (panel H). Panels A′ to H′ show the stress fibers images quantified (see Materials and Methods for details) in panel H. *p<0.05; **p<0.01, Student t-test.
Figure 10.
GABAergic interneuron migration to the cortex is mediated by Sip1 in a Smad-dependent way.
The focal electroporation experiment is schematically represented in panel E (for more details, see Materials & Methods, and see van den Berghe et al., 2013). Sip1 WT or Sip1 SBD domain mutant construct (AxAx)2, and ZnF, were co-electroporated with a conditional dsRed-encoding plasmid (CALNL) to mark targeted cells in Sip1;RCE|Nkx2-1 brain slices (E13.5). After 3 days in vitro (DIV), only 3.10% ±0.73% (n = 9 slices) of the Sip1 KO interneurons in the control condition (CALNL only) is able to reach the cortex compared to 23.77% ±2.06% (n = 28 slices) when a Sip1 WT construct is electroporated. The Sip1 domain mutants are not able to rescue the interneuron migration, Sip1(AxAx)2 mutant: 7.06±0.77% (n = 29 slices) and Sip1 ZnF mutant: 4.55% ±0.75% (n = 18 slices). Quantification is shown in panel F. Error bars represent the SEM of 2 independent experiments, *p<0.0001, Chi-square test.