Molecular basis of ALK1-mediated signalling by BMP9/BMP10 and their prodomain-bound forms - PubMed (original) (raw)

Molecular basis of ALK1-mediated signalling by BMP9/BMP10 and their prodomain-bound forms

Richard M Salmon et al. Nat Commun. 2020.

Abstract

Activin receptor-like kinase 1 (ALK1)-mediated endothelial cell signalling in response to bone morphogenetic protein 9 (BMP9) and BMP10 is of significant importance in cardiovascular disease and cancer. However, detailed molecular mechanisms of ALK1-mediated signalling remain unclear. Here, we report crystal structures of the BMP10:ALK1 complex at 2.3 Å and the prodomain-bound BMP9:ALK1 complex at 3.3 Å. Structural analyses reveal a tripartite recognition mechanism that defines BMP9 and BMP10 specificity for ALK1, and predict that crossveinless 2 is not an inhibitor of BMP9, which is confirmed by experimental evidence. Introduction of BMP10-specific residues into BMP9 yields BMP10-like ligands with diminished signalling activity in C2C12 cells, validating the tripartite mechanism. The loss of osteogenic signalling in C2C12 does not translate into non-osteogenic activity in vivo and BMP10 also induces bone-formation. Collectively, these data provide insight into ALK1-mediated BMP9 and BMP10 signalling, facilitating therapeutic targeting of this important pathway.

PubMed Disclaimer

Conflict of interest statement

N.W.M. and W.L. are co-founders of Morphogen-IX. D.J.G. and J.R. are co-founders of RxCelerate Ltd. J.S.B. and J.R. are full-time employees of RxCelerate Ltd. The remaining authors declare no competing interests.

Figures

Fig. 1

Fig. 1. Pro-BMP9 and pro-BMP10 are equivalent ALK1-ligands.

a Dose-dependent signalling assays in PAECs. Serum-starved PAECs were treated with different ligands at 2.48 pM (white bars), 8.27 pM (light grey bars) and 27.3 pM (dark grey bars) (using monomer molecular weight, equivalent to 0.03, 0.1 and 0.33 ng ml−1 BMP9 GF-domain concentration) for 1 h. Changes in the ID1 gene expression were monitored using RT-qPCR. Data were presented as fold change relative to untreated cells, and means ± SEM of three independent experiments are shown. Source data are provided as a Source Data file. b–d Volcano plots comparing changes in global gene expression in PAECs after pro-BMP9 or pro-BMP10 treatment. Serum-starved PAECs were treated with 25 pM of pro-BMP9 or pro-BMP10 (purity can be found on SDS-PAGE with silver staining in Supplementary Fig. 8a, lanes 1 and 4) for 1.5 h before RNA was extracted for microarray analysis. Four different primary PAEC lines were used. Red dots above the dashed line represent the changes in target genes with adjusted p values of less than 0.05. Several representative target genes are highlighted in c and d. Full list of genes can be found in Supplementary Data 1 and 2. e Affinity measurements of BMP9 and BMP10 for ALK1 using Biacore. A CM5 Biacore chip was immobilised with ALK1 dimer (ALK1-Fc) or monomer (in-house purified ALK1 ECD, purity can be seen in Supplementary Fig. 8a, lane 7). The sensorgrams of BMP9, pro-BMP9, BMP10 and pro-BMP10 binding raw data (in black lines) were overlaid with a global fit to a 1:1 model with mass transport limitations (red lines). f A summary of kinetic parameters for ligand-receptor interactions derived from the Biacore measurements in e.

Fig. 2

Fig. 2. Crystal structure of human BMP10:ALK1 complex at 2.3 Å.

a Crystal structure of BMP10 (cyan) in complex with ALK1 (magenta), overlaid with the structure of BMP9:ALK1:ActRIIb complex (PDB:4FAO, in grey and semi-transparent). Mol A and Mol B are the two BMP monomers whose interfaces with ALK1 (Mol C) were analysed in b. b Comparison of the buried interface upon complex formation between BMP10:ALK1 and BMP9:ALK1 (from 4FAO). Total buried surface area and the contributing residues were calculated using PDBePISA server.

Fig. 3

Fig. 3. Specificity determinants in the BMP9 and BMP10 subfamily.

a Sequence alignment of representative ALK-binding BMPs. GF-domain sequences of ALK1-binding BMP9 and BMP10, ALK6-binding BMP14, ALK3-binding BMP2 and BMP4, as well as ALK2-binding BMP6 and BMP7 are aligned. Lines over and below the sequences highlight the residues at the type I and type II receptor-binding surface based on BMP10:ALK1 and BMP9:ALK1:ActRIIb structures, respectively. Asterisk (*) marks the residues that are conserved among at least 6 out of 7 aligned BMPs. Residues preferentially conserved between BMP9 and BMP10 are highlighted, in cyan for those at the type I site (conserved region 1), in blue for those at the type II site (conserved region 2) and in yellow for those outside receptor binding surface (conserved region 3). BMP10 D338 and P366 are also highlighted in cyan because they make conserved interactions with ALK1 in the crystal structure (Fig. 4). b Residues from conserved regions 1–3 plotted on BMP10 structure and labelled with full length proBMP10 residue numbers. Fifteen residues from conserved regions 1–3 are shown in spheres, coloured accordingly. The first Gly from conserved region 3 is not modelled in the crystal structure, and hence not plotted. c An overlay of BMP10 (grey) onto the structures of BMP9 (gold, from 4FAO), BMP2 (green, from 2GOO) and BMP7 (cyan, from 1M4U) is shown from the side view (left) and the top view (right). The red arrows indicate the unique insertion in BMP9 and BMP10.

Fig. 4

Fig. 4. Conserved region 1 and ALK1-specificity determinants.

a ALK1-binding residues are mapped onto the BMP10 surface (grey), with those conserved across all BMPs in Fig. 3a coloured in red, those from the BMP9 and BMP10 conserved region 1 in cyan, and other variable residues in yellow. b ALK1 (magenta cartoon) binding to BMP10 (surface), with residues interacting with BMP10 shown in sticks. c–f Detailed interactions between BMP10 and ALK1. g Sequence alignment of four BMP-binding type I receptors, ALK1, ALK2, ALK3 and ALK6, with the four specificity-determining residues in ALK1 highlighted in yellow. Loop 4 and loop 5 are the two loops surrounding the 310 helix (Supplementary Fig. 4). h Overlaid structures of BMP type I receptors. The structure of ALK1 in the BMP10:ALK1 complex (magenta) is overlaid onto ALK1 in BMP9:ALK1 complex (PBD:4FAO, orange), ALK3 (PDB:2GOO, light green) and ALK6 (PDB:3EVS, light grey). ALK1 residues highlighted in g are shown in sticks.

Fig. 5

Fig. 5. Conserved region 2 and type II site analysis.

a ActRIIb-binding residues (based on PDB:4FAO, ActRIIb in semi-transparent cartoon) are mapped onto BMP9 surface (grey), with those conserved across all BMPs in Fig. 3a coloured in red, those from conserved region 2 in blue, and other variable residues in yellow. b Type II binding surface of BMP9 (left) and BMP10 (right), showing as electrostatic surface (generated in PyMOL, red representing negatively charged and blue positively charged surface). ActRIIb is shown in orange, semi-transparent cartoon. c Residues from BMP9 conserved region 2 make three backbone β-sheet and one sidechain H-bond interactions with ENG (PDB:5HZW, ENG in green, BMP9 in cyan). BMP10 is overlaid onto BMP9 and shown in grey, with four conserved region 2 residues shown in blue spheres. Sidechains of other residues are omitted for clarity. d Sequence alignment of human BMP10 prodomain (hBMP10_pro) with mouse BMP9 prodomain (mBMP9_pro) and human BMP9 prodomain (hBMP9_pro). Residues at the BMP9-binding surface are highlighted in yellow and those that make direct interactions with BMP9 GF-domain are marked with *. Residues that make main chain interactions are also marked with ^. Only the prodomain regions that interact with BMP9 GF-domain are shown, and full-length alignment of hBMP9_pro and hBMP10_pro can be found in Supplementary Fig. 5. e Residues in conserved region 2 of BMP9 make four backbone H-bond β-sheet interactions with prodomain (PDB:4YCG; prodomain in orange, BMP9 in cyan. BMP10 is overlaid on BMP9 and shown in grey. Four conserved BMP10 residues are in blue spheres).

Fig. 6

Fig. 6. ALK1 can form a complex with pro-BMP9.

a Overall structure of the human pro-BMP9:ALK1 complex at 3.3 Å (BMP9 in green, ALK1 in magenta, prodomain in orange) overlaid onto the pro-BMP9 structure (4YCG, grey, semi-transparent). b Backbones of the BMP9:ALK1 portion from the pro-BMP9:ALK1 structure (BMP9 in green, ALK1 in magenta) overlaid onto the same region in the BMP9:ALK1:ActRIIb structure (4FAO, grey). c Overlay of the two prodomains from the pro-BMP9:ALK1 structure (shown in cartoon and coloured in orange and grey respectively) and that from 4YCG (in ribbon, cyan). d In the pro-BMP9:ALK1 structure, the conserved region 2 in BMP9 makes the same four backbone H-bond interactions with the prodomain as shown in Fig. 5e. Four residues in the conserved region 2 are shown in blue spheres and labelled with BMP10 numbering. e–h Analysis of complex formation by analytical gel filtration. Purified pro-BMP9, pro-BMP9 mixed with ALK1, pro-BMP9 mixed with sENG and sENG were run separately on an S200 10/300 gel filtration column which was pre-equilibrated with 20 mM Tris.HCl, 150 mM NaCl, pH 7.4. e Gel filtration traces. The arrows indicate the elution volumes of the standards. Numbers 1-6 indicate the 6 peaks which were analysed by SDS-PAGE. f Middle fraction from each peak was run on an SDS-PAGE. Identities of the proteins on the SDS-PAGE are indicated using coloured circles. g Consecutive fractions from each gel filtration experiment were run on a non-reducing SDS-PAGE and immunoblotted using an anti-BMP9 prodomain antibody. Each analytical sample run was repeated at least one more time with fraction checked on SDS-PAGE to ensure reproducibility. h Cartoon diagrams, using the same colouring scheme as the circles in f, to illustrate that mixing pro-BMP9 with ALK1 leads to the formation of pro-BMP9:ALK1 complex, whereas mixing pro-BMP9 with sENG leads to the displacement of the prodomain which can be readily detected as a different peak in the gel filtration.

Fig. 7

Fig. 7. CV2 does not inhibit BMP9 signalling.

a Structural analysis. BMP10 (grey, with conserved region 2 residues in blue spheres) was overlaid onto the BMP2:CV2 structure (PDB:3BK3, CV2 in magenta and BMP2 in green). Four mainchain H-bonds that stabilise the BMP2:CV2 β-sheet interaction are shown. BMP9 has the same conformation as BMP10 in this region. b, c CV2 does not inhibit BMP9 signalling in PAECs. Serum-starved PAECs were treated with BMP9 or pro-BMP9 (at 1 ng ml−1 GF-domain concentration) without or with CV2 at 10-fold, 20-fold, 50-fold or 250-fold molar excess for 15 min to assess Smad1/5 phosphorylation using immunoblots (b) or for 1 h to assess ID1 gene expression using qPCR (c). One representative of three independent experiments is shown in b. Band intensity was quantified using Image J (version 1.51s). d CV2 inhibits BMP4 signalling in PASMCs. Serum-starved PASMCs were treated with BMP4 (25 ng ml−1) without or with CV2 at indicated molar excess for 15 min. Immunoblots and quantification were carried out as above. N = 3 independent experiments and one representative blot is shown. e CV2 inhibits BMP2 but not BMP9 signalling in C2C12 cells. Serum-starved C2C12 cells were treated with BMP2 (130 ng ml−1) or BMP9 (25 ng ml−1) without or with CV2 at the indicated molar excess for 68 h. ALP activity in the cell lysate were analysed (see Methods section). N = 7 independent experiments. For all panels, means ± SEM are shown. d, e One-way ANOVA for each BMP treatment group, Dunnett’s post hoc analysis against BMP alone-treated controls. Source data are provided as a Source Data file.

Fig. 8

Fig. 8. Modifying BMP9 signalling specificity by mutagenesis.

A panel of BMP9 mutants were generated, and tested in vitro and in vivo as described in the Methods. a Mutant proteins were subject to endothelial cell signalling assays (at 0.3 ng ml−1 GF-domain concentration) using induction of ID1 gene expression in hPAECs as a readout, and osteogenic signalling assays (at 10 ng ml−1 GF-domain concentration) using ALP induction in C2C12 cells as a readout. Data were normalised to WT BMP9 and shown as fold change relative to WT upon mutation. Each treatment condition was repeated in 3–7 independent experiments alongside untreated and WT controls. The exact N number for each condition is given under the column. Means ± SEM are shown. b Recombinant WT pro-BMP9, pro-BMP9 D366E, pro-BMP10, as well as BMP2 GF-domain were subject to in vivo heterotopic bone-forming assays in the presence and absence of cardiotoxin. Each data point represents the HO result from an independent injection in one leg. N number for each treatment condition is given under each column. Data are presented as % ossification volume relative to the average of BMP2-treated controls. Means ± SEM are shown. c Representative CT images (left) and histological staining (right) of in vivo formed heterotopic bones after stimulation of indicated BMP molecules in the presence and absence of cardiotoxin. B: osteoid matrix; M: muscle cells. Scale bar = 500 µm.

References

    1. Seki T, Yun J, Oh SP. Arterial endothelium-specific activin receptor-like kinase 1 expression suggests its role in arterialization and vascular remodeling. Circ. Res. 2003;93:682–689. doi: 10.1161/01.RES.0000095246.40391.3B. - DOI - PubMed
    1. David L, Mallet C, Mazerbourg S, Feige JJ, Bailly S. Identification of BMP9 and BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in endothelial cells. Blood. 2007;109:1953–1961. doi: 10.1182/blood-2006-07-034124. - DOI - PubMed
    1. Walsh DW, Godson C, Brazil DP, Martin F. Extracellular BMP-antagonist regulation in development and disease: tied up in knots. Trends Cell Biol. 2010;20:244–256. doi: 10.1016/j.tcb.2010.01.008. - DOI - PubMed
    1. Bidart M, et al. BMP9 is produced by hepatocytes and circulates mainly in an active mature form complexed to its prodomain. Cell. Mol. Life Sci. 2012;69:313–324. doi: 10.1007/s00018-011-0751-1. - DOI - PMC - PubMed
    1. Jiang H, et al. The prodomain-bound form of bone morphogenetic protein 10 is biologically active on endothelial cells. J. Biol. Chem. 2016;291:2954–2966. doi: 10.1074/jbc.M115.683292. - DOI - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources