Fig 2: Prodomain-bound BMP10 is a highly stable complex. Analytical gel filtration analysis of semi-purified pBMP10 in TBS (A), TBS with NaCl at 1 m final concentration (B) or TBS with 1 m GuHCl (C). pBMP10 was pre-incubated in each buffer for 30 min before being loaded onto a Superdex S200 10/300 size-exclusion column pre-equilibrated in the same buffer. A control protein carbonic anhydrase (24 kDa) that does not interact with either BMP10 GFD or its prodomain was added to the pBMP10 before loading. Blue dextran (2000 kDa, black arrow) was run separately in each buffer to indicate the void volume. Note the protein peaks shifted slightly between the runs, potentially due to proteins interacting with the column matrix differently in different buffer systems. Points 1–11 on the traces correspond to consecutive fractions, which were run in lanes 1–11 of immunoblotting analyses probed with either anti-BMP10 prodomain or anti-BMP10 antibodies. Peaks X and Y were TCA precipitated, ran on a non-reducing SDS-PAGE and Coomassie Blue stained to reveal the identity of the peaks. Peak Z is carbonic anhydrase.

Fig 3: Administration of BMP10, but not BMP9, prevents AVM formation in Bmp9/10-dKO. a Schematic representation of BMP protein administration and Bmp10 gene deletion into control and Bmp9/10-dKO mice. Arrowheads indicate injection of 50 μg tamoxifen at P1–P4 and BMP9 or BMP10 proteins daily from P1. b–d CD31 and SMA staining on retinas isolated from P8 control (PBS-treated CreER-negative Bmp10-iKO, b, BMP9-treated Bmp9/10-dKO (c), and BMP10-treated Bmp9/10-dKO (d) mice. Arrows mark arteriovenous shunts. a artery, v vein. Scale bars, 500 μm. e and f Quantification of retinal vascular outgrowth (e, n = 5 to 12 retinas from 5 to 6 mice per group) and AVM number (f, n = 6 to 12 retinas from 5 to 6 mice per group). Data are mean ± SEM. Statics Kruskal–Wallis test followed by Dunn’s post hoc test. g and h Hemoglobin levels (g, n = 5 to 6 mice per group) and HW/BW ratio (h, n = 3 to 5 mice per group) at the end day of the study. Data are mean ± SEM. One-way ANOVA followed by Tukey’s post hoc test

Fig 4: Bmp10 deficiency leads to the development of arteriovenous malformation in neonatal retinas. a Schematic representation of Bmp10 deletion. Arrowheads indicate intragastric tamoxifen injections at P2 and P3 and the analysis at P7. b CD31 and SMA staining of control, Bmp9-KO, Bmp10-iKO, and Bmp9/10-dKO P7 retinas. a artery, v vein. Scale bars, 200 μm. c–e Quantification of the vessel radial length (c, n = 8 to 14 retinas per group), artery diameter (d, n = 26 to 31 arteries from 6 to 8 retinas per group), and vein diameter (e, n = 25 to 34 veins from 6 to 8 retinas per group). Data are mean ± SEM. One-way ANOVA followed by Tukey’s post hoc test. f Representative images of the latex dye-perfused retinas of control, Bmp9-KO, Bmp10-iKO, and Bmp9/10-dKO mice at P7. Arrows mark arteriovenous shunts. a artery, v vein. Scale bars, 500 μm. g–j Scatter plots measuring Hb (g), RBC (h), HCT (i), and MCV (j) to assess hemorrhage. Data represent individual mice and mean ± SEM. One-way ANOVA followed by Tukey’s post hoc test (n = 6 to 13 mice per group)

Fig 5: BMP10 prodomain does not inhibit BMP10 activity in endothelial cell lines. A, titration of BMP10 GFD activities in HPAECs (left) and HAECs (right). Increasing concentrations of BMP10 GFD as indicated were used to treat the serum-starved cells. After 1 h treatment, cells were harvested and the phosphorylation of Smad1/5 and the induction of ID1 and ID3 proteins were analyzed using immunoblotting. B, BMP10 GFD was pre-incubated with increasing amounts of BMP10 prodomain in same molar ratio as in Fig. 1B before applying to serum-starved endothelial cells in (B) HPAECs, (C) HAECs, and (D) HMEC-1. Remaining activity of BMP10 was measured by phosphorylation of Smad1/5 with total Smad1 as a loading control. One representative blot from four repeats is shown for all experiments. Prodomain inhibitions were quantified and analyzed as in Fig. 1B and shown below.