Article Pathogenic ACVR1R206H activation by Activin A-induced receptor clustering and autophosphorylation Anassuya Ramachandran1,*,†,‡ , Merima Mehić1,† , Laabiah Wasim2, Dessislava Malinova2 , Ilaria Gori1 , Beata K Blaszczyk3, Diana M Carvalho4, Eileen M Shore5 , Chris Jones4 , Marko Hyvönen3 , Pavel Tolar2,# & Caroline S Hill1,** Abstract receptor (Taylor et al, 2014b). FOP is a monogenic, autosomal domi- nant disease caused by germline mutations in ACVR1 and character- Fibrodysplasia ossificans progressiva (FOP) and diffuse intrinsic ized by progressive, episodic heterotopic ossification of soft tissue pontine glioma (DIPG) are debilitating diseases that share causal (Shore et al, 2006; Furuya et al, 2008; Kaplan et al, 2009). DIPG, in 206 mutations in ACVR1,aTGF-β family type I receptor. ACVR1R H is a contrast, is an invariably fatal childhood glioma arising in the ventral frequent mutation in both diseases. Pathogenic signaling via the pons of the hindbrain, with a median survival of 9–12 months. SMAD1/5 pathway is mediated by Activin A, but how the mutation Approximately 20% of DIPG tumors harbor somatic ACVR1 muta- triggers aberrant signaling is not known. We show that ACVR1 is tions (Buczkowicz et al, 2014; Fontebasso et al, 2014; Taylor et al, essential for Activin A-mediated SMAD1/5 phosphorylation and is acti- 2014a; Wu et al, 2014; Mackay et al, 2017). vated by two distinct mechanisms. Wild-type ACVR1 is activated by TGF-β family ligands require two different receptors for signal- 206 the Activin type I receptors, ACVR1B/C. In contrast, ACVR1R H activa- ing, type I and type II receptors, which are thought to function as a tion does not require upstream kinases, but is predominantly acti- heterotetramer, comprising two type I and two type II receptors. vated via Activin A-dependent receptor clustering, which induces its Ligand binding brings the receptors together, whereupon the consti- auto-activation. We use optogenetics and live-imaging approaches to tutively active type II receptor phosphorylates and activates the type demonstrate Activin A-induced receptor clustering and show it I receptor (Wrana et al, 1994). The activated type I receptors then requires the type II receptors ACVR2A/B. Our data provide molecular phosphorylate the intracellular effectors of the pathway, the mechanistic insight into the pathogenesis of FOP and DIPG by linking receptor-activated SMADs (R-SMADs). Phosphorylated R-SMADs the causal activating genetic mutation to disrupted signaling. form complexes with SMAD4 and SMAD heterotrimers accumulate in the nucleus to regulate gene expression (Miller & Hill, 2016). Keywords Activin A; ACVR1R206H; DIPG; FOP; receptor clustering There are five type II receptors and seven type I receptors, and dif- Subject Categories Cancer; Signal Transduction ferent ligands use different combinations of receptors to signal. The DOI 10.15252/embj.2020106317 | Received 22 July 2020 | Revised 25 March identity of the SMADs phosphorylated by a particular ligand–recep- 2021 | Accepted 26 March 2021 tor complex is dictated by the type I receptor engaged (Massague, The EMBO Journal (2021)e106317 2012). Thus, ACVR1B and ACVR1C, which bind Activins and Nodal, and TGFBR1, which binds TGF-βs, induce phosphorylation of SMAD2 and SMAD3. The other type I receptors, which bind BMPs Introduction and GDFs, induce phosphorylation of SMAD1, SMAD5, and SMAD9 (Massague, 2012). ACVR1 is in this latter group and has a prefer- Fibrodysplasia ossificans progressiva (FOP) and diffuse intrinsic ence for BMPs (Miller & Hill, 2016). pontine glioma (DIPG) are devastating human diseases, both associ- An exception to these rules emerged with the discovery that ated with activating mutations in ACVR1, a type I TGF-β family TGF-β signaling through ACVRL1 and ACVR1 led to SMAD1/5 1 Developmental Signalling Laboratory, The Francis Crick Institute, London, UK 2 Immune Receptor Activation Laboratory, The Francis Crick Institute, London, UK 3 Department of Biochemistry, University of Cambridge, Cambridge, UK 4 Division of Molecular Pathology, The Institute of Cancer Research, Sutton, UK 5 Departments of Orthopaedic Surgery and Genetics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA *Corresponding author. Tel: +64 9 923 1431; E-mail: [email protected] **Corresponding author. Tel: +44 203 796 1251; E-mail: [email protected] †These authors contributed equally to this work ‡Present address: Department of Molecular Medicine and Pathology, University of Auckland, Auckland, New Zealand #Present address: Division of Infection and Immunity, Institute of Immunity and Transplantation, University College, London, UK ª 2021 The Authors. Published under the terms of the CC BY 4.0 license The EMBO Journal e106317 | 2021 1 of 22 The EMBO Journal Anassuya Ramachandran et al phosphorylation (Goumans et al, 2002; Lebrin et al, 2004; Daly initiation of heterotopic ossification in mice. Notably, this pathologi- et al, 2008; Ramachandran et al, 2018). TGF-β-induced SMAD1/5 cal signaling could be blocked using the soluble type II receptor phosphorylation is required for cell proliferation in endothelial cells extracellular domains, acting as ligand traps or Activin A neutraliz- (Goumans et al, 2002; Lebrin et al, 2004) and for anchorage- ing antibodies (Hatsell et al, 2015; Lees-Shepard et al, 2018). independent growth, cell migration, and epithelial-to-mesenchymal Here, we unravel the mechanism whereby Activin A signals transition (EMT) in other cell types (Miettinen et al, 1994; Daly through ACVR1R206H. We use extensive genome editing in HEK293T et al, 2008; Liu et al, 2009; Ramachandran et al, 2018). The underly- cells, as well as glioma cells from DIPG patients to both model the ing mechanism of receptor activation in this case represents a depar- disease mutation, ACVR1R206H, and resolve the role of additional ture from the classical mechanism and involves a type I receptor receptors in Activin A-induced ACVR1R206H activity. We provide activating another type I receptor. We recently demonstrated that strong evidence that type I receptor activation of other type I recep- TGFBR1, activated as a result of TGF-β stimulation, can phosphory- tors, as we previously established for TGF-β, is a general mechanism late and activate ACVR1, which subsequently phosphorylates and not restricted to TGF-β receptors. We demonstrate that wild- SMAD1/5 (Ramachandran et al, 2018). It is not yet known whether type ACVR1 (hereafter ACVR1WT) is essential for Activin A-induced this mechanism of activating ACVR1 is unique to TGF-β or whether SMAD1/5 phosphorylation, and that similar to TGFBR1, ACVR1B/C other receptor complexes that normally signal through SMAD2/3 activate ACVR1WT downstream of Activin A to mediate this can also fulfill this function. phosphorylation. In contrast, ACVR1R206H is predominantly auto- The mutations in ACVR1 that lead to FOP occur in the cytoplas- activated by clustering, independent of the kinase activity of type II mic portion of the receptor and are concentrated around both the receptors or ACVR1B/C. Finally, we use live-imaging approaches to GS domain, a glycine-serine rich juxtamembrane region, and the demonstrate Activin A-induced receptor clustering and show that kinase domain (Han et al, 2018). Although approximately 13 muta- this requires the type II receptors, ACVR2A/B. These results eluci- tions have been identified, almost 97% of cases of FOP have a single date the signaling mechanism of ACVR1R206H, which is crucial for amino acid change, Arg206His (hereafter ACVR1R206H) (Shore et al, understanding the pathogenesis of FOP and DIPG. 2006; Han et al, 2018). This same mutation accounts for about a fifth of DIPG mutations (Han et al, 2018). ACVR1 mutations in FOP and DIPG, and in particular Results ACVR1R206H, lead to enhanced SMAD1/5-induced signaling (Fukuda 206 et al, 2009; Shen et al, 2009; van Dinther et al, 2010; Mucha et al, ACVR1R H exhibits increased responsiveness to Activin A, 2018). However, the underlying mechanism has proven controver- relative to wild-type ACVR1 sial. One possible explanation has centered around the immuno- philin FKBP1A (also known as FKBP12), which binds the GS To dissect the mechanism whereby the mutant ACVR1R206H receptor domain of the type I receptors, maintaining them in an inactive signals, we introduced this point mutation into the endogenous conformation (Wang et al, 1994; Okadome et al, 1996). Disease ACVR1 locus in a clone of HEK293T cells (hereafter referred to as mutations in ACVR1 were thought to disrupt binding of FKBP1A, parental) using CRISPR/Cas9 genome editing. A heterozygous clone thereby removing pathway inhibition. However, the vast majority of (HET) and several homozygous clones (HOM1 and HOM2) were ACVR1 mutants retain the ability to bind FKBP1A and can be inhib- obtained (Fig EV1A and B). We first characterized the responsive- ited by it, suggesting that this mechanism does not explain the ness of these clones to different TGF-β family ligands (Activin A, enhanced activity of ACVR1R206H (Groppe et al, 2011; Chaikuad Activin B, BMP4, BMP7, BMP4/7 heterodimer, and BMP2), using et al, 2012; Machiya et al, 2018). SMAD1/5 phosphorylation (pSMAD1/5) as a readout. Treatment of The roles of type II receptors and activating ligands in the activity parental cells for 1 h with all these ligands resulted in induction of of ACVR1R206H are also not fully understood. In Drosophila, the pSMAD1/5, with Activin A being the least efficient (Figs 1A and