Homodimerization of RBPMS2 through a new RRM-interaction motif is necessary to control smooth muscle plasticity

Abstract

In vertebrates, smooth muscle cells (SMCs) can reversibly switch between contractile and proliferative phenotypes. This involves various molecular mechanisms to reactivate developmental signaling pathways and induce cell dedifferentiation. The protein RBPMS2 regulates early development and plasticity of digestive SMCs by inhibiting the bone morphogenetic protein pathway through its interaction with NOGGIN mRNA. RBPMS2 contains only one RNA recognition motif (RRM) while this motif is often repeated in tandem or associated with other functional domains in RRM-containing proteins. Herein, we show using an extensive combination of structure/function analyses that RBPMS2 homodimerizes through a particular sequence motif (D-x-K-x-R-E-L-Y-L-L-F: residues 39-51) located in its RRM domain. We also show that this specific motif is conserved among its homologs and paralogs in vertebrates and in its insect and worm orthologs (CPO and MEC-8, respectively) suggesting a conserved molecular mechanism of action. Inhibition of the dimerization process through targeting a conserved leucine inside of this motif abolishes the capacity of RBPMS2 to interact with the translational elongation eEF2 protein, to upregulate NOGGIN mRNA in vivo and to drive SMC dedifferentiation. Our study demonstrates that RBPMS2 possesses an RRM domain harboring both RNA-binding and protein-binding properties and that the newly identified RRM-homodimerization motif is crucial for the function of RBPMS2 at the cell and tissue levels.

Figure 1.

Dimeric structure of RBPMS2 in solution. (A, B) RBPMS2, CPO and MEC-8 proteins harbor evolutionary conserved RNA recognition motif (RRM) domain. (A) Phylogenetic tree using human RBPMS2 protein as query (in red). (B) Phylogenetic tree using the RBPMS2 RRM domain (amino acids from 27 to 117) as query (in red). After each leaf end follow the protein names, the species names (H.sa.: Human; M.mu.: Mus musculus; G.ga.: Gallus gallus; D.me.: Drosophila melanogaster; C.el.: Caenhorabditis elegans), the E-value and the percentage of identity regarding the query. The accession number of each protein is presented on the right column of the figure. Scale bar represents evolution distance of leaf branches (A.U.). The figure was prepared using iTOL (http://itol.embl.de). (C) Sequence-structure alignment of the RRM domain in RBPMS2 homologs. The N-terminus (residues 27–117) of human RBPMS2 was aligned with the corresponding segments in the sequences of RBPMS2, CPO or MEC-8 from different metazoan species (human: Homo sapiens; danre: Dano rerio; galga: Gallus gallus; xenle: Xenopus leavis; drome: Drosophila melanogaster; apime: Apis mellifica; culpi: Culex pipens; Bruma: Brugia malayi; caeel: Caenorhabditis elegans). Arrowheads indicate the predicted RNA binding site and asterisks the newly identified dimerization interface. Residue numbering corresponds to the human RBPMS2 sequence. The figure was made using ESPRIPT (http/espript.ibcp.fr). Overall structure of wild-type human RBMPS2-Nter homodimers (D) and detailed view of the dimerization interface (E) seen in a ribbon diagram. The secondary structure and loops involved in the dimerization are shown as red-to-blue (subunit A) and orange (subunit B) ribbons. Red indicates highly conserved amino acid where blue labels less conserved residues. The side chains that stabilize the dimeric interface are shown as sticks using the CPK color convention. Residues are numbered according to the human RBMPS2 sequence. The figures were prepared using Pymol (http://pymol.sourceforge.net). (F) Experimental small-angle X-ray scattering curve (logarithm of intensity in arbitrary units as a function of the momentum transfer range s in Å⁻¹) for RBPMS2-Nter measured at 1.9 mg/ml (green crosses), with its fitting theoretical curve (red continuous line) back-calculated from the RBPMS2-Nter NMR structure (Supplementary Figure S2). Blue dots represent the relative error bound. The χ² value of the fit is 1.115.

Figure 2.

RBPMS2 dimers are formed in vitro and in vivo independently of RBPMS2 interaction with RNA. (A) Schematic representation of the different RBPMS2 clones isolated by Y2H screening with human RBPMS2 as bait. The RRM domain of RBPMS2 is between amino acids 32 and 105. (B) Immunoprecipitation with rabbit anti-Myc antibodies (lanes 3 and 6) or without (lanes 2 and 5) of protein lysates from DF-1 cells that express human HA-RBPMS2 or HA-TC10 and Myc-RBPMS2 or not. Lanes 1 and 4: 10% of total protein extracts from cells that express only HA-RBPMS2 or HA-TC10. Co-immunoprecipitation of HA-RBPMS2 was monitored by immunoblotting with mouse anti-HA antibodies (upper panel). The efficiency of immunoprecipitation was monitored by immunoblotting with rabbit anti-Myc antibodies (lower panel). (C) Co-immunoprecipitation of HA-RBPMS2 and Myc-RBPMS2 dimers in the absence of RNA. Protein lysates from DF-1 cells that express HA-RBPMS2 alone or with Myc-RBPMS2 (lanes 2–5) were incubated with 50-μg/ml RNase A at room temperature for 30 min (lanes 4 and 5) or left untreated (lanes 1–3) and then immunoprecipitated with rabbit anti-Myc antibodies (lanes 3 and 5) or without (lanes 2 and 4). Lane 1: 10% of total cell extracts from cells that express only HA-RBPMS2. Co-immunoprecipitation was monitored by immunoblotting with mouse anti-HA antibodies (upper panel). The immunoprecipitation efficiency was monitored by immunoblotting with rabbit anti-Myc antibodies (lower panel). (D) Analysis of the interaction of Myc-RBPMS2 with HA-RBPMS2 by proximity ligation assays (PLAs) in DF-1 cells that express Myc-RBPMS2 with HA-RBPMS2 or HA-TC10, or Myc-NICD and HA-RBPMS2, or Myc-RBPMS2 alone. HA-tagged proteins were detected with anti-mouse HA antibodies (in green) and Myc-tagged proteins with anti-rabbit Myc antibodies (in red). Interactions between proteins were detected with Duolink PLA labeled in magenta. Images were collected by confocal microscopy. Bars, 10 μm.

Figure 3.

The mutation of Leucine 49 into Glutamic acid (L49E) in the RRM of human RBPMS2 inhibits homodimerization, but does not alter binding to NOGGIN mRNA. (A) Analysis of the interaction of Myc-RBPMS2 with HA-RBPMS2 by Duolink PLA in DF-1 cells that co-express Myc-RBPMS2 or Myc-RBPMS2-L49E and HA-RBPMS2, or Myc-RBPMS2 and HA-TC10. Ha-tagged proteins were detected with anti-mouse HA antibodies (in green) and Myc-tagged proteins with anti-rabbit Myc antibodies (in red). Protein interactions were detected with Duolink PLA labeled in magenta. Images were collected by confocal microscopy. Bars, 10 μm. (B) Immunoprecipitation of RBPMS2 homodimers. Protein lysates from DF-1 cells that express HA-RBPMS2 and Myc-RBPMS2 (lanes 2 and 3) or Myc-RBPMS2-L49E (lanes 4 and 5) were immunoprecipitated with rabbit anti-Myc antibodies (lanes 3 and 5) or without (lanes 2 and 4). Lane 1: 10% of total cellular extracts from cells that express HA-RBPMS2 alone. Co-immunoprecipitation was monitored by immunoblotting with mouse anti-HA antibodies (upper panel). Immunoprecipitation efficiency was monitored by immunoblotting with rabbit anti-Myc antibodies (lower panel). (C) Subcellular localization of human RBPMS2 and RBPMS2-L49E. HEK293 cells that express Myc-RBPMS2 or Myc-RBPMS2-L49E were detected with anti-EiF3n (eukaryotic translation initiation factor 3n is present in stress granule) and rabbit anti-Myc antibodies. Myc-RBPMS2 and Myc-RBPMS2-L49E show similar cytoplasmic localization. (D) Experimental small-angle X-ray scattering curve (logarithm of intensity in arbitrary units as a function of the momentum transfer range s in Å⁻¹) for RBPMS2-Nter-L49E measured at 1.1 mg/ml (green crosses), with its fitting theoretical curve (red continuous line) back-calculated from the RBPMS2-Nter-L49E NMR structure (Supplementary Figure S2). Blue dots represent the relative error bound. The χ² value of the fit is 1.096. (E) EMSA binding assays using a fixed high concentration of 161-nt NOGGIN RNA (100 nM) were performed with increasing concentrations of RBPMS2-Nter and RBPMS2-Nter-L49E ranging from 0.1 to 5 μM on a same gel and detected with SYBR^® Green EMSA nucleic acid gel stain. Note that RBPMS2-Nter forms defined RNA/protein complex as soon as 1 μM and two complexes at 5 μM (complexes I and II), whereas RBPMS2-Nter-L49E forms very diffuse bands (bracket and arrow). Free 161-nt RNA is indicated with a red arrow.

Figure 4.

RBPMS2 homodimerization is required for RBPMS2 function. (A) Stomachs from E9 chicken embryos after retroviral misexpression of RCAS-GFP alone (negative control) or of RCAS-RBPMS2 and RCAS-GFP. The presence of retroviruses was confirmed by direct observation of GFP expression. Sustained expression of chicken RBPMS2 leads to specific stomach malformations: hypertrophied proventriculus (arrows) and denser and malformed gizzard (arrowheads). (B) Stomachs from E9 chicken embryos after retroviral misexpression of RCAS-GFP (control) or of the RCAS-RBPMS2-L40E mutant that cannot homodimerize with RCAS-GFP. White arrows and arrowheads indicate, respectively, the gizzard and the proventriculus. (C) Whole-mount in situ hybridization analysis of NOGGIN expression in stomachs from E9 chicken embryos after retroviral misexpression of RCAS-GFP (control), RCAS-RBPMS2 and RCAS-GFP, or RCAS-RBPMS2-L40E and RCAS-GFP. Infection was confirmed by direct observation of GFP expression. (D) Glutaraldehyde crosslink assay of protein extract from DF1 cells infected either with RCAS-GFP (positive control, left panel), RCAS-Myc-RBPMS2 or RCAS-Myc-RBPMS2-L40E (right panels), followed by SDS-PAGE separation and revealed by rabbit anti-GFP or anti-Myc antibodies. Anti-GFP antibody revealed GFP monomers and dimers (near 55 kDa, black arrow), as expected. Anti-Myc antibody revealed Myc-RBPMS2 monomers at 27 kDa on both Myc-RBPMS2 and Myc-RBPMS2-L40E expressing cell protein extract, but only Myc-RBPMS2 expressing cell protein extract presents Myc-RBPMS2 dimerization at 55 kDa (red arrow). Note the below 50-kDa bands (asterisk) on Myc-RBPMS2 expressing cell protein extract correspond to the induction of endogenous avian MYC protein.

Figure 5.

RBPMS2 homodimerization is necessary to dedifferentiate digestive SMCs. (A) Immunofluorescence analysis of primary SMC cultures infected with RCAS-empty, RCAS-Myc-RBPMS2 and RCAS-Myc-RBPMS2-L40E retroviruses for 3 days. Nuclei were visualized with Hoechst. Anti-Calponin antibodies were used as a marker of SMC differentiation and anti-Myc antibodies to identify cells infected by retroviruses that express chicken RBPMS2 or RBPMS2-L40E. (B) Quantification of mitotic cells using anti-phosphorylated Histone 3-Ser10 (PH3) antibodies in primary SMC cultures infected with retroviruses that express RBPMS2 or RBPMS2-L40E retroviruses or controls (RCAS-empty) for 7 days. Values are the mean ± standard error of the mean of two independent experiments (RCAS-RBPMS2 or RCAS-RBPMS2 L40E versus RCAS-empty). **for P < 0.01; n.s. for no statistically significant. (C) Schematic representation of the eEF2 clones isolated by Y2H screening with human RBPMS2 as bait. Human eEF2 harbors one elongation factor domain (EF2) and three U5 small nuclear ribonucleoprotein domains (snRNP). (D) Immunoprecipitation with mouse anti-HA antibodies of protein lysates from HEK293 cells that express human HA-RBPMS2, HA-TC10 or not. Co-immunoprecipitation of endogenous eEF2 was monitored by immunoblotting with rabbit anti-eEF2 antibodies (upper panel). The efficiency of immunoprecipitation of HA-RBPMS2 and HA-TC10 was monitored by immunoblotting with mouse anti-HA antibodies (middle panel). Lower panel: 10% of total protein extracts from cells of each condition monitored by immunoblotting with rabbit anti-eEF2 antibodies. (E) Analysis of the interaction of Myc-RBPMS2 with HA-eEF2 by Duolink PLA in HEK293 cells that co-express Myc-RBPMS2 or Myc-RBPMS2-L49E and HA-eEF2. HA-eEF2 was detected with anti-mouse HA antibodies (in red) and Myc-tagged proteins with anti-rabbit Myc antibodies (in green). Protein interactions were detected with Duolink PLA labeled in magenta. Bars, 50 μm.

Figure 6.

Model of the action of RBPMS2 homodimerization in the SMC dedifferentiation process. Homodimeric RBPMS2 complex interacts with eEF2, an essential member of the eukaryote translational machinery that allows the accumulation of NOGGIN mRNA that will be further translated to inhibit BMP pathway activity (9) and to trigger the dedifferentiation of the SMCs (A). Disruption of RBPMS2 homodimerization through Leu to Glu substitution in position 49 leads to monomeric RBPMS2 proteins unable to interact with eEF2 and inefficient to induce the dedifferentiation process (B).