Competition between type I activin and BMP receptors for binding to ACVR2A regulates signaling to distinct Smad pathways

Background Activins and bone morphogenetic proteins (BMPs) play critical, sometimes opposing roles, in multiple physiological and pathological processes and diseases. They signal to distinct Smad branches; activins signal mainly to Smad2/3, while BMPs activate mainly Smad1/5/8. This gives rise to the possibility that competition between the different type I receptors through which activin and BMP signal for common type II receptors can provide a mechanism for fine-tuning the cellular response to activin/BMP stimuli. Among the transforming growth factor-β superfamily type II receptors, ACVR2A/B are highly promiscuous, due to their ability to interact with different type I receptors (e.g., ALK4 vs. ALK2/3/6) and with their respective ligands [activin A (ActA) vs. BMP9/2]. However, studies on complex formation between these full-length receptors situated at the plasma membrane, and especially on the potential competition between the different activin and BMP type I receptors for a common activin type II receptor, were lacking. Results We employed a combination of IgG-mediated patching-immobilization of several type I receptors in the absence or presence of ligands with fluorescence recovery after photobleaching (FRAP) measurements on the lateral diffusion of an activin type II receptor, ACVR2A, to demonstrate the principle of competition between type I receptors for ACVR2. Our results show that ACVR2A can form stable heteromeric complexes with ALK4 (an activin type I receptor), as well as with several BMP type I receptors (ALK2/3/6). Of note, ALK4 and the BMP type I receptors competed for binding ACVR2A. To assess the implications of this competition for signaling output, we first validated that in our cell model system (U2OS cells), ACVR2/ALK4 transduce ActA signaling to Smad2/3, while BMP9 signaling to Smad1/5/8 employ ACVR2/ALK2 or ACVR2/ALK3. By combining ligand stimulation with overexpression of a competing type I receptor, we showed that differential complex formation of distinct type I receptors with a common type II receptor balances the signaling to the two Smad branches. Conclusions Different type I receptors that signal to distinct Smad pathways (Smad2/3 vs. Smad1/5/8) compete for binding to common activin type II receptors. This provides a novel mechanism to balance signaling between Smad2/3 and Smad1/5/8. Supplementary Information The online version contains supplementary material available at 10.1186/s12915-022-01252-z.


Fig. S3 Smad activation as a function of time by ActA, BMP9 or BMP2 in U2OS cells.
Experiments were conducted as in Fig. S2, except that a single saturating concentration of each ligand was used (4 nM for ActA and BMP9, 10 nM for BMP2), and stimulation was for varying times (from 0 to 90 min). a, c, e Representative blots of stimulation by ActA (a), BMP9 (c), or BMP2 (e). b, d, f Quantification of ActA signaling to Smad2/3 (b), BMP9 signaling to Smad1/5/8 (d), and BMP2 signaling to Smad1/5/8 (f). The bands were visualized by ECL and quantified by densitometry (see Methods). Data are mean ± SEM (3 independent experiments in each case) of the ratio of pSmad2/3 (or pSmad1/5/8) over -actin. The values obtained for 30 min stimulation with each ligand (which were the highest in all cases) were taken as 1.  Fig. 2 by myc-ACVR2A alone or together with an HA-tagged type I receptor (or empty vector). After 24 h, the cell surface levels of the indicated receptors were measured by the point confocal method, as described under Methods and in Fig. 1a, using the FRAP setup under identical non-bleaching conditions. Results are mean  SEM of 30 independent measurements (each on a different cell) under each condition. a Level of myc-ACVR2A alone or in the presence of HA-tagged type I receptors. b Level of HA-type I receptors expressed alone or together with myc-ACVR2A. No significant differences were found between the expression levels of singly-expressed myc-ACVR2A or the same receptor coexpressed with HA-tagged ALK2/3/4/6. Similarly, coexpression of myc-ACVR2A with any of the HA-tagged receptors did not alter the level of the HA-tagged receptor (one-way ANOVA and Bonferroni post-hoc test; P > 0.1).

Fig. S5
Patch/FRAP studies do not detect interactions between myc-ACVR2A/HA-TRII or myc-ALK4/HA-ALK2. COS7 cells were cotransfected with myc-ACVR2A and HA-TRII or empty vector (a-d), or with myc-ALK4 and HA-ALK2 or empty vector (e-h). After 24 h, live cells were subjected to the IgG-mediated patching protocol (panels a, b, e, f) as in Fig. 2, resulting in the HA-tagged receptor patched and labeled by Alexa 488-GR IgG (IgG HA), whereas the myctagged receptor is labeled by monovalent Alexa 546-GαM Fab'. In control experiments, the IgG HA labeling was replaced by Fab' HA labeling. FRAP studies were conducted as in Fig. 2. a, b Average Rf (a) and D values (b) show no effect of crosslinking HA-TRII on Rf or D of myc-ACVR2A. Bars depict the average values (mean ± SEM); the number of measurements (each conducted on a different cell) is shown on each bar. No significant differences were detected between the Rf or D values (P > 0.29; one-way ANOVA and Bonferroni post-hoc test). c, d The cell surface levels of myc-ACVR2A and HA-TRII alone or together are similar (Student's twotailed t-test; P > 0.3), as measured by the point confocal method (see Methods and Fig. 1a). Results are mean  SEM of 30 independent measurements (each on a different cell) under each condition. e, f Average Rf (e) and D values (f) showing no effect of crosslinking HA-ALK2 on Rf or D of myc-ALK4. Bars, mean ± SEM values, with the number of measurements shown on each bar. No significant differences were found between the Rf or D values under all conditions (P > 0.6; oneway ANOVA and Bonferroni post-hoc test). g, h The cell surface levels of myc-ALK4 and HA-ALK2 alone or together are similar (Student's two-tailed t-test; P > 0.8); measurements were as in panels c, d.  Fig. 2d, second bar from the left, as indicated to the right of the panel). No significant differences were found between the D values (panel b; one way ANOVA with Bonferroni posthoc test, P > 0.9). Coexpression of HA-TRII with ALK4 and myc-ACVR2A did not alter the reduction in Rf of the latter as measured upon coexpression of myc-ACVR2A with ALK4 alone (full line arrow; ***, P < 10 -4 , one way ANOVA and Bonferroni post-hoc test), and this reduction was unaffected by IgG crosslinking of HA-TRII (n.s. = non significant; P > 0.9). This indicates that HA-TRII is not competing with ALK4 for binding myc-ACVR2A. c, d Point confocal measurements of the cell surface levels of coexpressed myc-ACVR2A and HA-TRII with or without coexpression of untagged ALK4. Bars represents mean  SEM of 30 independent measurements. No significant differences were observed between the levels of any receptor pairs compared (Student's two-tailed t-test; P > 0.4).

Fig. S8
ActA does not induce significant signaling to Smad1/5/8 in U2OS cells. Cells were starved (2 h, 1% serum) and stimulated (30 min, 37 °C) with 4 nM ActA or left in starvation medium (control). They were then lysed, subjected to SDS-PAGE and immunoblotted for pSmad1/5/8, tSmad1 and -actin. a A blot of a representative experiment. b Quantification of ActA signaling to Smad1/5/8. The bands were quantified by ECL and densitometry. Data are mean ± SEM of the pSmad1/5/8 over -actin ratio of 4 independent experiments. The value obtained for ActA-stimulated cells was taken as 1. No significant pSmad1/5/8 formation was detected following stimulation with ActA (P > 0.5; Student's two-tailed t-test).

Fig. S9 The ALK2/3 inhibitor LDN212854 inhibits BMP9-mediated pSmad1/5/8 formation in U2OS cells.
Cells were starved (2 h, 1% serum) and incubated for the second hour of starvation with LDN212854 (1.3 or 2.5 nM) or with 0.5% DMSO (vehicle). LDN212854 inhibits ALK2 and ALK1 (the latter of which is hardly expressed in U2OS cells; Additional file 2: Table S1) with IC50 of 1.3 and 2.4 nM, respectively, and ALK3 to a lower degree [67]. The incubation with the inhibitor was followed by stimulation (or not; control) for 30 min at 37 °C with BMP9 (4 nM). The cells were then lysed and immunoblotted for pSmad1/5/8, tSmad1 and -actin as in Fig. 6. a A representative blot. b Quantification of BMP9 signaling to pSmad1/5/8. The bands were visualized by ECL and quantified by densitometry. Data are mean ± SEM of the pSmad1/5/8 over -actin ratio of 3 independent experiments for each concentration. The value obtained for untreated cells stimulated with BMP9 was taken as 1. pSmad1/5/8 formation in response to BMP9 was strongly inhibited by LDN212854 at both 2.5 and 1.3 nM concentrations, suggesting that most of the signaling to pSmad1/5/8 in the U2OS cells is mediated via ALK2 with a possible contribution of ALK3. Asterisks indicate significant differences between the pairs marked by the brackets, using one-way ANOVA and Bonferroni post-hoc test (**, P < 0.02).

Fig. S10 Signaling activity of HA-ALK2 and HA-ALK4.
To test whether HA-ALK2 and HA-ALK4 induce signaling, U2OS cells were transfected with HA-ALK2, HA-ALK4 or empty vector (control). After 24 h, cells were starved (2 h, 1% serum) and stimulated (30 min, 37 °C) with the indicated ligands (4 nM ActA or BMP9) or left in starvation medium. They were then lysed, subjected to SDS-PAGE and immunoblotted for pSmad1/5/8, tSmad1 (for BMP9 signaling), or for pSmad2/3 and tSmad2/3 (for ActA signaling); -actin served as loading control. a, c Representative blots. b, d Quantification of BMP9 signaling to Smad1/5/8 by HA-ALK2 (b) and of ActA signaling to Smad2/3 by HA-ALK4 (d). The bands were quantified by ECL and densitometry. Data are mean ± SEM of the pSmad over β-actin ratio of 3 independent experiments in each case. The value obtained for ligand-stimulated control cells was taken as 1. In both cases, a significant increase was observed in the signaling following transfection with the respective HAtagged receptor (one-way ANOVA and Bonferroni post-hoc test; *, P < 0.05).

Fig. S12 Untagged ALK4 and type I BMP receptors compete for signaling to Smads via
ACVR2. Experiments were done as in Fig. 8, except that the cells were transfected with the respective untagged receptors instead of the HA-tagged receptors. After activation with ActA (4 nM, 30 min) or BMP9 (4 nM, 30 min) the cells were lysed and immunoblotted for pSmad2/3, tSmad2/3 (for ActA signaling) and pSmad1/5/8, tSmad1 (for BMP9 signaling) along with -actin. a, c Representative blots showing the effect of untagged ALK2 and ALK3 overexpression on ActA signaling to Smad2/3 (a) or of ALK4 overexpression on BMP9 signaling to pSmad1/5/8 (c). b, d Quantification of the effects of untagged ALK2 or ALK3 on ActA-mediated pSmad2/3 formation (b) and of untagged ALK4 on BMP9-mediated pSmad1/5/8 formation (d). Data are mean ± SEM of the relevant pSmad over β-actin ratio of 3 independent experiments in each case. The value obtained for control cells stimulated by the respective ligand was taken as 1. Asterisks indicate significant differences between the pairs marked by the brackets, using one-way ANOVA and Bonferroni post-hoc test (*, P < 0.03; **, P < 8 x 10 -4 ; ***, P < 10 -4 ).

Fig. S13
Signaling competition by ALK4 and type I BMP receptors occurs also at lower ligand concentrations. Experiments were done as in Fig. 8, except that the ligand concentrations used for stimulation (30 min) were lowered from 4 nM to 1 nM. a, c Representative blots showing the effects of overexpressing HA-ALK2 or -ALK3 on ActA signaling to Smad2/3 (a), or of HA-ALK4 overexpression on BMP9 signaling to pSmad1/5/8 (b). The expression of the various HAtagged receptors was probed by blotting for the HA tag using the HA-7 antibody (a, b). b, d Quantification of ALK2 or ALK3 effects on ActA-mediated pSmad2/3 formation (b) and of ALK4 on BMP9-mediated pSmad1/5/8 formation (d). Data are mean ± SEM of the relevant pSmad over β-actin ratio of 3 independent experiments in each case. The value obtained for control cells stimulated by the respective ligand was taken as 1. Asterisks indicate significant differences between the pairs marked by the brackets, using one-way ANOVA and Bonferroni post-hoc test (*, P < 0.03; **, P < 0.01; ***, P < 5 x 10 -4 ). Data are mean ± SEM of the relevant mRNA levels of 3 independent experiments in each case. The value obtained for the cells transfected with the respective HA-tagged receptor was taken as 1. In both cases, no significant differences were observed between the control and the transfected samples (Student's two-tailed t-test. n.s. = non significant; P > 0.3).