Point mutations predicted to interfere with phosphorylation of the BMP4 prodomain selectively interfere with BMP4 homodimer but not BMP4/7 heterodimer activity.

(A-B) Schematic illustrating sequential cleavage of BMP4 homodimers (A) and BMP4/7 heterodimers (B). (C) Sequence alignment of a portion of the prodomain of human (h), mouse (m), zebrafish (z), Xenopus (x) and chick (c) BMP4 to illustrate the conserved S-X-G FAM20C recognition motif. (D) RNA encoding wild type or point mutant forms of BMP4 were injected alone or together with BMP7 near the dorsal marginal zone (DMZ) of 4-cell embryos. DMZ and ventral marginal zone (VMZ) explants were isolated at stage 10 and pSmad1 levels were analyzed by immunoblot as illustrated. Blots were reprobed with actin as a loading control. (E-F) Quantitation of relative pSmad1 levels normalized to actin in at least 3 independent experiments (mean +/- SD, data analyzed using an unpaired t-test). (G) RNA encoding wild type or point mutant forms of BMP4 were injected alone or together with BMP7 near the animal pole of two-cell embryos. Ectodermal explants were isolated at stage 10 and tbxt levels were analyzed by semi-quantitative RT-PCR as illustrated. (H-I) Quantitation of relative tbxt levels normalized to odc in at least 3 independent experiments (mean +/- SD, data analyzed using an unpaired t-test).

Bmp4S91C homozygotes die during mid-embryogenesis and show reduced BMP activity in multiple tissues

(A-F) Photograph of E10.5 or E11.5 wild type (A, D) or mutant (B, C, E, F) littermates. Scale bars in Panel A and D apply across each row. (G-J) E10.5 wild type (G, J) or Bmp4S91C/S91C mutant littermates (H, I, K, L) carrying a BRE-LacZ transgene were stained for ß- galactosidase activity to detect endogenous BMP pathway activation. Embryos from a single litter were stained for an identical time under identical conditions. A minimum of three embryos of each genotype were examined and results shown were reproduced in all. G-I show right side and J-L the left side of the same embryos. VPM; ventral posterior mesoderm, BA; branchial arches, SM; somitic mesoderm. Scale bar corresponds to 1mm in all panels. (M-O) Expression of Nkx2.5 was analyzed by whole mount in situ hybridization in E10.5 wild type (M) and mutant (N, O) littermates. Photographs of hearts dissected free of embryos are shown. Scale bar in M applies across the row. RA; right atrium, LA; left atrium, RV; right ventricle, LV; left ventricle, OFT; outflow tract.

Progeny from Bmp4S91C/+ or Bmp4E93G/+ intercrosses, or from Bmp4-/+ and Bmp4E93G/+ intercrosses at P28.

Progeny from Bmp4S91C/+ intercrosses, or from Bmp4-/+ and Bmp4E93G/+ Intercrosses at embryonic ages.

Bmp4-/E93Gmutants die during embryogenesis with defects in ventral body wall closure, small or absent eyes and heart defects.

(A-G) Photographs of E13.5 (A-C) or E14.5 (D-G) wild type (A, D) or mutant (B, C, E-G) littermates. The position of the liver inside (A-E, F) or outside (C, G) of the abdomen is indicated. (H, I) Hematoxylin and eosin stained coronal sections of hearts dissected from E14.5 wild type (H) or Bmp4-/E93G littermates (I). Asterisk denotes VSD and arrowheads indicate thin, non-compacted ventricular wall (I). Scale bar applies across each row.

E93G and S91C mutations lead to accumulation of BMP4 precursor protein and reduced levels of cleaved ligand and pSmad1 in vivo.

(A-H) Levels of pSmad1 and BMP4 were analyzed in protein lysates isolated from E10.5 Bmp4S91C (A-D) or Bmp4E93G (E-H) homozygotes, heterozygotes or wild type littermates. (I-P) Levels of pSmad1, BMP4 precursor protein and cleaved BMP4 ligand were analyzed in cell lysates and conditioned media of MEFs isolated from E13.5 Bmp4S91C heterozygotes or wild type littermates (I-L) or from Bmp4E93G homozygotes, heterozygotes or wild type littermates (M-P). Representative blots (A, E, I, K) and quantitation of protein levels normalized to actin (mean +/- SD, data analyzed using an unpaired t-test) (B-D, F-H, J-L, N-O) are shown.

BMP4E93G and BMP4S91C precursor proteins are O-glycosylated and exit the ER.

(A, B) MEFs were isolated from E13.5 Bmp4S91C heterozygotes, Bmp4E93G homozygotes, or wild type littermates.Protein lysates were left untreated or were treated with O-glycosidase and α- neuraminidase to remove O-linked glycosylation (A) or with EndoH or PNGase to remove N-linked glycosylation (B). Immunoblots were probed with antibodies directed against BMP4 to detect BMP4 precursor protein. Bands corresponding to Endo H-sensitive (asterisks) and Endo H- resistant PNGase-sensitive BMP4 (arrowheads) are indicated in B.

Hypothetical models for how prodomain phosphorylation regulates proteolytic maturation of BMP4.

(A) Model 1: Phosphorylation of BMP4 by Fam20C within the TGN directs subcellular trafficking of BMP4 out of the golgi to the cell surface via a route that is distinct from the pathway taken by furin. Furin- and BMP4-containing vesicles fuse in a region adjacent to the cell surface where furin cleaves BMP4 to release the active ligand. (B) Model 2: Fam20C, BMP4 and furin traffic together to a membrane proximal subcellular compartment where Fam20C becomes catalytically activated (lightening bolt) and phosphorylates the prodomain of BMP4. Phosphorylation enables furin to access and cleave the consensus motif on BMP4. (C) Structure of BMP4 precursor monomer predicted by alphafold. N-terminus in blue, C-terminus in red. Boxed region includes Ser91 and the furin consensus motif. (D-E) Close up of boxed region in C illustrating the close proximity between Ser91 (pink letters) within the prodomain in its unphosphorylated (D) or phosphorylated (E) form and the -R287-R-R-A-K-R292 furin consensus motif at the C-terminus of the prodomain. Red arrows indicate the site of furin cleavage. (F-G) Spacefill representation of the of the structures of BMP4 precursor protein homodimers (F) or BMP4 (light gray) and BMP7 (dark gray) precursor protein heterodimers (G) predicted by alphafold with the location of furin recognition motifs denoted in purple.