- Research article
- Open Access
Structural basis for potency differences between GDF8 and GDF11
- Ryan G. Walker1,
- Magdalena Czepnik1,
- Erich J. Goebel1,
- Jason C. McCoy1,
- Ana Vujic2,
- Miook Cho2, 3,
- Juhyun Oh2, 3,
- Senem Aykul4,
- Kelly L. Walton5, 8,
- Gauthier Schang6,
- Daniel J. Bernard6,
- Andrew P. Hinck7,
- Craig A. Harrison5, 8,
- Erik Martinez-Hackert4,
- Amy J. Wagers2, 3,
- Richard T. Lee2 and
- Thomas B. Thompson1, 9Email author
© Thompson et al. 2017
Received: 21 November 2016
Accepted: 18 January 2017
Published: 3 March 2017
Growth/differentiation factor 8 (GDF8) and GDF11 are two highly similar members of the transforming growth factor β (TGFβ) family. While GDF8 has been recognized as a negative regulator of muscle growth and differentiation, there are conflicting studies on the function of GDF11 and whether GDF11 has beneficial effects on age-related dysfunction. To address whether GDF8 and GDF11 are functionally identical, we compared their signaling and structural properties.
Here we show that, despite their high similarity, GDF11 is a more potent activator of SMAD2/3 and signals more effectively through the type I activin-like receptor kinase receptors ALK4/5/7 than GDF8. Resolution of the GDF11:FS288 complex, apo-GDF8, and apo-GDF11 crystal structures reveals unique properties of both ligands, specifically in the type I receptor binding site. Lastly, substitution of GDF11 residues into GDF8 confers enhanced activity to GDF8.
These studies identify distinctive structural features of GDF11 that enhance its potency, relative to GDF8; however, the biological consequences of these differences remain to be determined.
The transforming growth factor β (TGFβ) superfamily of secreted proteins comprises more than 30 structurally related, yet functionally distinct proteins that play critical roles in embryological tissue development and patterning, wound healing, and adult tissue maintenance (reviewed in [1–5]). The TGFβ superfamily can be divided into three subclasses: TGFβs, bone morphogenetic proteins (BMPs), and activins/inhibins. Growth/differentiation factors 8 and 11 (GDF8/myostatin and GDF11/BMP11, respectively) are two closely related members of the activin/inhibin subclass that share ~90% sequence identity in their mature, C-terminal signaling domains and ~52% identity in their N-terminal prodomains. GDF8 and GDF11 bind to similar receptors [6–8] and extracellular antagonists [6, 9, 10], leading to the assumption that mature GDF8 and GDF11 ligands are functionally indistinguishable. While it is clear that GDF8 and GDF11 share many commonalities, including overlapping or redundant roles in certain biological processes [11–17], there is accumulating evidence that the two ligands may be functionally distinct [13, 18–23].
Gdf8 is expressed postnatally by skeletal and cardiac muscle and therein negatively regulates skeletal muscle mass by suppressing both the number and size of individual muscle fibers [6, 18, 19, 24]. In contrast, GDF11 appears to act more broadly, regulating anterior/posterior patterning and development of multiple organs/tissues [11, 13]. Many tissues express Gdf11 postnatally, including the spleen, pancreas, kidney, and skeletal muscle [11, 25–28]. However, determination of GDF11’s exact role in the adult has remained elusive due to the embryonic lethality of Gdf11 -/- mice [11, 13]. In stark contrast, Gdf8 -/- mice survive into adulthood and have a profound hypermuscular phenotype, which can be recapitulated in wild-type mice using natural occurring antagonists of GDF8, such as follistatin (FS), follistatin-like 3 (FSTL3), and growth/differentiation factor-associated serum protein 1 (GASP1) [6, 29–33]. Interestingly, Gdf8 -/- /Gdf11 -/- mice have exaggerated homeotic axial transformations compared to Gdf11 -/- mice, suggesting that GDF8 and GDF11 have redundant functions in skeletal patterning . However, muscle-specific knockout of Gdf11 does not result in significant increases in muscle mass and circulating GDF11 does not overcome the hypermuscular phenotype found in Gdf8 -/- mice, suggesting that GDF8 and GDF11 do not serve redundant roles in regulating skeletal muscle mass . Thus, while it is clear that loss of one ligand compared to the other yields drastically different phenotypes, it has been argued that these differences relate primarily to differential localization of ligand expression and do not reflect differences in ligand signaling.
There have been conflicting reports comparing the biological functions of GDF8 and GDF11, with some question as to whether the two ligands are functionally equivalent [16, 17, 20–23, 43, 44]. Specifically, GDF11 was identified as a circulating factor that declines with age, and exogenous delivery of the mature ligand reversed many pathologies associated with aging [22, 23], including reducing age-related cardiac hypertrophy , improving skeletal muscle performance and repair , improving olfaction , and increasing neurogenesis . Conversely, other reports have challenged these claims, arguing that GDF8 and GDF11 are functionally and biologically equivalent [16, 17, 43, 44]. However, it is difficult to compare the results, given that many of these studies did not utilize the same experimental design, sources of recombinant proteins, or strategies for detection of GDF8 and GDF11 proteins (recently reviewed in [5, 45]). Therefore, the extent of functional overlap between mature GDF8 and GDF11 without the prodomain remains to be determined. Furthermore, a rigorous side-by-side biochemical and biological comparison of mature GDF8 and GDF11 has yet to be performed to determine if distinct features exist between these growth factors.
In this study, using a variety of in vitro and in vivo systems, we show that GDF11 is significantly more potent than GDF8 and that the enhanced activity is due to differences in type I receptor utilization. In addition, we present three new X-ray crystal structures of apo-GDF8, apo-GDF11, and GDF11 in complex with follistatin 288 (FS288). These structures, including an additional recently solved apo-GDF11 crystal structure , reveal that differences between GDF11 and GDF8 cluster in the type I receptor binding epitope. Intriguingly, our structural analysis has revealed unique and alternate conformations of both GDF8 and GDF11, suggesting that both ligands are inherently flexible. GDF8/GDF11 chimeras, in which particular GDF11 residues are substituted into the GDF8 sequence, show that increased potency can be conferred to GDF8. Taken together, our results demonstrate that, despite the high sequence identity between mature GDF8 and GDF11, the ligands indeed possess different signaling properties.
GDF11 is a more potent ligand than GDF8
Recent reports have attributed unique in vivo properties to GDF11 relative to GDF8 [20–23]. Therefore, we first aimed to determine if signaling differences exist between these two growth factors using HEK293 (Fig. 1b) and HepG2 (Fig. 1c) cells stably or transiently transfected, respectively, with the SMAD3-responsive (CAGA)12 luciferase reporter [47–51]. Side-by-side titration of mammalian-derived (see “Methods”) ligands revealed that GDF11 was more potent than GDF8 in both cell lines (Fig. 1b and c; Additional file 1: Table S1). Remarkably, we determined that the half-maximal effective concentration 50 (EC50) values for GDF11 in HEK293 and HepG2 cells were 0.08 nM and 3.4 nM, respectively, compared to 0.48 nM and 5.4 nM for GDF8, respectively (Fig. 1c; Additional file 1: Table S1). Interestingly, the maximal SMAD3 response achieved by GDF11 was ~ fourfold higher compared to GDF8 in HepG2 cells (Fig. 1c; Additional file 1: Table S1), suggesting that there may be differences in ligand potency and/or in the repertoire of receptors and their relative utilization. Furthermore, we found that GDF11 (EC50 0.03 nM) more potently stimulated the release of follicle-stimulating hormone (FSH) than GDF8 (EC50 0.08 nM) in murine LβT2 pituitary gonadotrope cells (Fig. 1d; Additional file 1: Table S1). Finally, GDF11 activated SMAD3 with ~ a sevenfold lower EC50 compared to GDF8 (6.5 nM versus 45.7 nM, respectively) in A204 cells transfected with the (CAGA)12 luciferase reporter (Ravindra Kumar, personal communication). These results are consistent with previous reports, which likewise indicated that GDF11 is a more potent activator of SMAD3-dependent signaling than GDF8 [16, 44]. Note that ligands from different sources were utilized and were directly compared using the SMAD3-responsive (CAGA)12 luciferase reporter assay (Additional file 2: Figure S1). For clarity, we have indicated the ligand sources utilized for each study in the figure legends.
Given that GDF11 stimulated greater SMAD3 activation in a concentration-dependent fashion compared to GDF8, we wanted to determine if this difference was maintained when we performed a pulse-chase exposure to each ligand (Fig. 1e and f; Additional file 3: Figure S2). To examine this, we performed two separate experiments. In the first, we treated HEK293 cells with GDF8 or GDF11 at three different concentrations (25, 125, and 620 pM) for the indicated times (5–180 min), after which the culture medium was removed and replaced with ligand-free culture medium for the duration of the experiment. At 24 h (1440 min), the cells were lysed and assayed for SMAD3-dependent luciferase activity (Fig. 1e; Additional file 3: Figure S2). GDF11 elicited a stronger response compared to GDF8 in both a concentration- and a time-dependent manner. In fact, at lower concentrations and exposure times, treatment with GDF11, but not GDF8, resulted in productive signaling (25 pM, Fig. 1e). A second experiment was performed where the cells were treated for the indicated time (60–1440 min) and immediately assayed for activity (Fig. 1f; Additional file 3: Figure S2). Interestingly, at the lower concentration (25 pM; Fig. 1f), we observed a significant difference in SMAD3 activation compared to GDF8 at most time points. These differences were less pronounced or non-existent at higher concentrations (Fig. 1f; Additional file 3: Figure S2). Together, these results indicate that the cellular response to mature GDF8 and GDF11 can be significantly different depending on the concentration and duration of exposure.
Modulation of GDF8 and GDF11 activity by known extracellular antagonists
Analysis of known antagonists to GDF8 and GDF11 by luciferase reporter assay
IC50 (nM) 95% CI a
Log[IC50(M)] ± SEM b
Hill slope 95% CI
0.21 to 0.23
−9.7 ± 0.01
−5.17 to −3.43
0.27 to 0.34
−9.5 ± 0.03
−6.73 to −2.81
1.08 to 1.30
−8.9 ± 0.02
−2.19 to −1.68
3.72 to 4.35
−8.4 ± 0.02
−6.79 to −4.02
0.25 to 0.24
−9.6 ± 0.01
−4.90 to −4.03
0.29 to 0.39
−9.1 ± 0.02
−5.49 to −3.20
3.31 to 5.07
−8.4 ± 0.05
−2.15 to −1.14
14.98 to 22.46
−7.7 ± 0.04
−2.18 to −1.16
Structure of GDF11 bound to FS288
X-ray data collection and refinement statistics (molecular replacement)
GDF11:FS288 (native) a
apo-GDF8 (native) a
apo-GDF11 (native) a
P 21 21 21
P 21 21 21
P 32 2 1
Unit cell dimension
a, b, c (Å)
56.0, 59.1, 288.5
29.6, 77.7, 119.5
65.2, 65.2, 101.9
α, β, γ (°)
90, 90, 90
90, 90, 90
90, 90, 120
Mn (I/σI) b
R work (%), R free d (%)
No. of atoms (molecules)
B-factors (average, Å2)
Bond lengths (Å)
Bond angles (°)
GDF8 and GDF11 structural differences: follistatin-bound versus apo states
We resolved the structure of apo-GDF8 to 2.25 Å (Fig. 5a and Table 2) and the apo-GDF11 structure to 1.90 Å (Fig. 5b, bottom; Table 2) using X-ray crystallography, thus allowing a direct comparison to the recently published apo-GDF11 structure (Fig. 5b, middle) . Unexpectedly, resolution of the apo-GDF8 structure revealed a unique conformation such that the fingertips of each monomer are positioned proximally in a “closed” conformation compared to the classical distal positioning or “open” conformation observed for many ligands, including structures of FS-bound GDF8 (Fig. 5a, top) [38, 47, 54]. While the structure of apo-GDF8 adopted an unexpected “closed” conformation, a similar conformation has been observed for activin A when in complex with ActRIIB . In both cases, the wrist region is disordered and not resolved in the crystal structure (Fig. 5a). Our new apo-GDF11 structure, solved under similar crystallization conditions as apo-GDF8, adopted an exaggerated “open” conformation such that the fingers extend past the horizontal plane (Fig. 5b). This conformation shares a similar exaggerated “open” conformation to the crystal structure of GDF8 bound to a neutralizing antibody (Fig. 5a, bottom) . However, the wrist region in our apo-GDF11 structure is also disordered and not resolved in the crystal structure, similar to apo-GDF8 (Fig. 5b, bottom). The previously published apo-GDF11 crystal structure adopts the classic “open” conformation with an ordered wrist helix that includes the additional interchain hydrogen bond between Q62A and the pre-helix loop observed in the FS288-bound GDF11 crystal structure. Taken together, crystallization of both GDF8 and GDF11 in their apo states reveals intrinsic structural flexibility in these ligands.
GDF11 signals more effectively through type I receptors than GDF8
Analysis of ActRIIB-ECD on GDF8 and GDF11 activity by luciferase reporter assay
IC50 (nM) 95% CI a
Log[IC50(M)] ± SEM b
Hill slope 95% CI
12.24 to 13.66
−7.9 ± 0.02
−1.99 to −1.47
12.61 to 15.46
−7.9 ± 0.02
−1.79 to −1.35
2.60 to 3.38
−8.5 ± 0.03
−2.82 to −1.65
9.40 to 11.125
−8.0 ± 0.02
−10.46 to −5.63
Next, we determined if GDF8 and GDF11 signal through similar type I receptors. Numerous studies have shown that GDF8 and GDF11 utilize the type I receptors ALK4 and ALK5 [6–8]. In addition, GDF11 can signal through ALK7 , although it is unclear if GDF8 also utilizes this receptor. To our knowledge, a study directly comparing signaling by both ligands via the three receptors has not been reported. Therefore, we assessed the relative (CAGA)12 luciferase reporter response to GDF8 or GDF11 following transient expression of individual type I receptors in RIB L17 cells (Fig. 6c and d). This cell line lacks ALK5 expression and expresses low levels of ALK4 [7, 54, 63]. The level of ALK7 expression in these cells has not been reported. As expected, activin A robustly signaled in cells transfected with ALK4 [63, 64], activin B signaled through both ALK4 and ALK7 [65, 66], and TGFβ3 specifically signaled through ALK5 (Fig. 6c) [67, 68]. Consistent with previous reports, GDF8 and GDF11 readily signaled through ALK4 and ALK5 [6–8, 54]. Additionally, we confirmed that GDF11 signaled through ALK7 (Fig. 6c and d) . Strikingly, GDF11 induced greater reporter responses than GDF8 via ALK4, ALK5, or ALK7 (Fig. 6c and d). We should note that activin A, activin B, GDF11, and to a lesser extent GDF8, stimulated some reporter activity in the absence of exogenous receptors, suggesting low levels of endogenous ALK4 and ALK7 expression by these cells (Fig. 6c). Consistent with this idea, activin B, GDF8, and GDF11 induction of reporter activity was completely blocked by addition of the ALK4/5/7 small molecule inhibitor SB431542 (Fig. 6d) [69–72]. Importantly, signaling could be rescued by transfection of an SB431542-resistant form (S270T) of human ALK7 (Fig. 6d), confirming that all three ligands can signal via ALK7 with relative potencies of activin B > GDF11 > GDF8. Collectively, our results indicate that GDF11 can generate a greater SMAD3-dependent signal compared to GDF8 through all three type I receptors.
Given that type I receptor expression is not ubiquitous in all cells, we next wanted to determine if differences in ALK4/5/7 combinations would further distinguish or normalize response differences between GDF8 and GDF11. Different receptor combinations modestly enhanced signaling relative to individual receptors (Fig. 6c). Interestingly, when ALK4 and ALK5 were co-transfected, there was no statistically significant difference in response to GDF8 and GDF11 (at the concentration tested), in contrast to when either receptor was expressed alone (Fig. 6c). Co-transfection of ALK4 and ALK7 resulted in a more robust response to GDF11 over GDF8. Taken together, these results further suggest that ligand sensitivity can be driven, in part, by the receptor(s) expressed on a particular cell type, thereby providing a possible avenue by which cells may discriminate between highly similar ligands, such as GDF8 and GDF11.
To clarify these results, we attempted to detect direct binding of either ligand to ALK5-ECD, Fc-ALK5-ECD, and Fc-ActRIIB using SPR (Fig. 7b; Additional file 5: Figure S4). In this experiment, the receptor was immobilized on the SPR biosensor chip with standard amine coupling or captured using Protein A. As expected, both ligands readily bound the Fc-ActRIIB-ECD (Additional file 5: Figure S4D) [6–8]. Qualitatively, GDF11 interacted with both ALK5-ECD and Fc-ALK5-ECD in a dose-dependent manner and exhibited a fast association and rapid dissociation (Fig. 7b; Additional file 5: Figure S4A–C, E, F). In contrast, GDF8 did not bind to ALK5-ECD or Fc-ALK5-ECD in a dose-dependent fashion (Fig. 7b; Additional file 5: Figure S4E, F). Given that ALK5 binding to TGFβ is significantly enhanced through a cooperative interaction with TβRII [73, 74], we speculated that ActRIIB might improve GDF8 or GDF11 binding to ALK5-ECD. Interestingly, co-injection of ActRIIB-ECD with GDF11 did not appear to alter the association or dissociation with ALK5-ECD compared to GDF11 alone (Fig. 7b right two panels; Additional file 5: Figure S4B, C). However, we did observe a detectable improvement in binding of GDF8 to ALK5-ECD in the presence of ActRIIB-ECD (Fig. 7b; Additional file 5: Figure S4B, C). These results suggest that GDF11 binding to ALK5 is likely more favorable compared to GDF8. However, it is clear that additional experiments will be necessary to fully resolve the interactions between these ligands and their cognate type I receptors.
Substitution of GDF11-like residues into GDF8 enhances ligand activity
To rule out that activity differences were attributed to differences in the relative amounts of protein expression, we produced and purified wt (wt GDF8 prodomain:mature GDF8) and selected mutant latent complexes (wt GDF8 prodomain:mature GDF8/GDF11 chimera). To determine their specific activity relative to wt GDF8 (Fig. 8b and c), we used two methods known to activate the latent complex: acid activation (treatment with HCl to pH 2.5; ) and activation with a member of the TLD family, such as tolloid-like 2 (TLL2) . Overall, our results from treatment with purified protein were consistent with the results from transient transfection (Fig. 8b and c). As expected, little to no luciferase activity was detected in cells treated with the latent complexes unless they were acid activated (Fig. 8b) or applied to cells transfected with TLL2 (Fig. 8c). In addition, we performed the experiments in the presence of GASP1, a specific antagonist of GDF8 and GDF11 [10, 42, 76], to confirm that the signal was specific to wt GDF8 or GDF8/GDF11 chimeras (Fig. 8b and c). Both activation methods revealed similar findings with the exception of the acid-activated empty vector sample (EV; conditioned medium from empty vector transfected and mock purified). The acid-activated EV sample showed a significant response that could not be attenuated with GASP1 (Fig. 8b), indicating possible contamination by additional ligands or pH-dependent artifacts. Despite this possibility, most of the signal with GDF8 or GDF8/GDF11 chimeras was attenuated by GASP1, suggesting that the majority of the signal was specific to the ligands added. Moreover, the higher background signal was not observed in the TLL2 experiment (Fig. 8c). Together, these data support the conclusions that mature GDF11 is more active than GDF8 and that GDF8 activity can be enhanced through substitution of specific residues from GDF11.
GDF11 is a more potent inducer than GDF8 of phosphorylated SMAD2/3 in mouse primary skeletal myoblasts in vitro and the mouse myocardium in vivo
Previous studies indicate that GDF8 and GDF11 have both divergent and overlapping roles in developmental processes and tissue maintenance in adult animals (reviewed in [1–5]). While the role of GDF11 during development is well defined, its role in the adult remains poorly understood. However, recent evidence suggests that GDF11 is an important circulating factor that declines with age and exogenous treatment with GDF11 can reverse age-related effects including reducing cardiac hypertrophy [22, 23], improving skeletal muscle regeneration , and promoting neurogenesis . However, other studies have contradicted certain aspects of these claims, fueling significant controversy on the functional similarities and differences between GDF11 and GDF8 [16, 17, 43, 44]. Controversy notwithstanding, in this study, we investigated the signaling activity and structural features of the mature GDF11 and GDF8 ligands.
At the outset of our studies, we made the unexpected observation that GDF11 is a more potent signaling ligand than GDF8. This observation was consistently and reproducibly observed across multiple laboratories, using different ligand preparations, in multiple cell lines and cultured primary cells, and recapitulated in vivo in the mouse myocardium. The current dogma in the literature suggests that there are no differences in the signaling potencies of GDF8 and GDF11 and that the ligands have identical signaling properties [16, 17, 43, 44]. Our results are consistent with this idea, but only at the highest ligand concentrations examined. Interestingly, prior studies also appeared to show that GDF11 is more potent or active than GDF8 [16, 44], yet many conclusions from these studies maintain the notion that GDF8 and GDF11 signal and function identically. The conclusion that GDF8 and GDF11 are indistinguishable may derive from the fact that subsequent comparisons were performed at concentrations that were above the EC100 for both ligands, thus resulting in identical transcriptional outcomes. In fact, we show that at high concentrations of GDF8 and GDF11, SMAD3-dependent readout is indistinguishable under certain conditions (see early time points in Fig. 1e and f). However, our data also reveal that GDF11 can elicit a significant response under conditions where GDF8 appears to have little to no activity. These results show that the responses generated by GDF8 and GDF11 are highly dependent on the concentrations of ligand and the available receptors, emphasizing that GDF8 and GDF11 are not equivalent. To be clear, these data neither support nor refute claims made by other groups regarding the functional outcomes resulting from GDF8 or GDF11 signaling. Rather, they demonstrate that the biochemical responses elicited by GDF8 and GDF11 at equivalent concentrations are significantly different under the conditions we tested. The physiological relevance, if any, which may result from potency differences, requires additional investigation.
Given the nearly identical sequence of their mature ligands, we were initially puzzled by the potency differences between GDF8 and GDF11, especially since most of the changes are relatively conserved. Therefore, we turned to X-ray crystallography to determine the structures of GDF11 in complex with FS288 in addition to GDF8 and GDF11 in their apo forms. While this manuscript was in preparation, an additional apo-GDF11 structure was resolved , thus providing the opportunity to compare multiple structures of each ligand in their FS288-bound and apo states. As expected, the overall structures of the ligands when bound to FS288 are highly similar with minor changes in the FSND helix of FS288, a region previously shown to be able to accommodate different ligands [47, 48, 52–55]. Comparison between GDF8 and GDF11 revealed that differences cluster within and around the type I receptor binding surface. In particular, differences are observed in residues within or bordering the wrist helix, an area important for type I receptor binding (; reviewed in ).
To explore if these differences are the basis for increased activity, we used a chimeric approach that introduced GDF11-specific residues into GDF8. Apart from the pre-helix loop, the two most prominent differences between the structures of GDF11 and GDF8 are located at each end of the wrist helix and include GDF11 residues Q62 and G100. In both cases, substitution of the individual GDF8 residues H62 and A100 with the GDF11 residues Q62 and G100 resulted in increased activity over wild-type GDF8. Moreover, combination of these two mutations (H62Q/A100G) resulted in an additional gain in activity over the single mutations alone, suggesting an additive effect. In GDF11, these residues appear to stabilize the dimer interface where an additional hydrogen bond is formed between Q62 and the backbone of the pre-helix loop. Thus, increased dimer stability or stability of the interaction between the wrist and pre-helix loop might be one explanation for the increased ligand activity of GDF11, though we cannot rule out the possibility that these residues directly interact with the receptors. However, given that activity differences are observed over multiple receptors, it is more likely that potency differences are a result of inherent ligand differences (e.g., differences in ligand flexibility/conformation). Furthermore, introducing the GDF11-specific residues in the pre-helix loop (F49Y/V50M/L52M), a region known to be important for receptor specificity (; reviewed in ), also results in increased activity. Additionally, introduction of GDF11 fingertip residues, D89 and Q91, into GDF8 also significantly increased ligand activity. However, our structural data do not provide a molecular explanation for how these variances confer differences in activity. Nevertheless, they do suggest that the potency differences between GDF8 and GDF11 are a result of a combination of the residue differences within and around the type I binding interface.
Resolution of the apo-GDF8 and apo-GDF11 crystal structures revealed that both ligands exhibit significant conformational flexibility. Interestingly and despite the high sequence identity, nearly identical crystallization conditions resulted in two distinct conformations. Although these conformations may be a result of the crystallization process, it supports the idea that GDF8 and GDF11 are flexible enough to be trapped in alternate states. However, an additional apo-GDF11 structure revealed the ligand in the classic “open” conformation and contained similar molecular contacts within the wrist region that we observed in the FS288-bound state . Furthermore, under similar crystallization conditions, additional TGFβ superfamily growth factors have been solved in their apo states and adopt the classic “open” conformation (reviewed in ). For example, structures of several BMP ligands in the apo state all exhibit the “open” configuration [77–81]. The extreme “closed” conformation of GDF8 is not entirely unprecedented, as one structure of activin A in complex with ActRIIB showed a similar “closed” configuration . This was initially somewhat controversial; however, multiple activin A structures have since been determined and support the notion that the activin A dimer is flexible [82, 83]. It should be noted that this is not the first example of GDF8 captured in an alternate conformation. Recently, the co-crystal structure of GDF8 bound to a neutralizing antibody was solved where the GDF8 ligand adopts an exaggerated “open” conformation , resembling the apo-GDF11 structure presented here. Furthermore, similar to activin A and GDF8, ligands of the TGFβ subclass also exhibit significant flexibility [60, 84–86]. In fact, the biological activity of different TGFβ isoforms, which also share high overall sequence identity, was shown to correlate with the rigidity of the dimer, specifically the wrist helix . However, we cannot discount that the crystallization process has trapped these ligands in alternate conformations. Additional solution-based biochemical approaches are needed to better understand ligand dynamics. Nevertheless, differences in the wrist region of GDF11 could contribute to the increased ligand potency over GDF8 by either stabilizing the dimer interface or presentation of the wrist/pre-helix loop to facilitate differential direct interaction with the type I receptors, or a combination thereof.
Our data strongly indicate that GDF11 is a more potent signaling ligand compared to GDF8. While this observation has not been explicitly realized in the literature, there is evidence suggesting that our results are consistent in that GDF11 is a more potent ligand than GDF8 [16, 44]. However, a recent study utilizing a cell line with modified and potentially functionally inactivated type II and type I receptors showed that GDF8 and GDF11 are nearly indistinguishable in terms of their potential for type II and type I heterodimerization . In this same study and using another assay, there was little difference in the potentiation of downstream SMAD2/3 responsive elements between GDF8 and GDF11 . Interestingly, we observed that differences in SMAD2/3 responsiveness between GDF8 and GDF11 were less pronounced in the HepG2 cell line, the same cell line utilized in . We observed more robust differences between GDF8 and GDF11 in other cell lines and in vivo. Therefore, a potential explanation for these differences may be due to the receptor profile of a particular cell type or tissue in addition to other confounding factors such as co-receptors. Given that GDF8 may not utilize the type I receptors as effectively as GDF11, it is tempting to speculate that both potency and receptor utilization differentiate the biological actions of GDF11 and GDF8.
In conclusion, we present evidence supporting the notion that GDF8 and GDF11, despite their high sequence identity, are not functionally equivalent. Our data show that GDF11 is a more potent activator of SMAD2/3 in vitro and in vivo. GDF8 has been considered a somewhat unique ligand of the TGFβ family due to dual utilization of the type I receptors ALK4 and ALK5. Our data suggest that GDF11 possesses these same attributes but, due to differences in amino acid composition, utilizes these receptors more effectively to initiate signal transduction. Apart from additional regulatory alterations between GDF8 and GDF11, these ligands may have evolved differences in relative potency through selective pressure in order to provide an evolutionary advantage.
Purified proteins utilized in this study
Proteins used in this study were commercially purchased or produced and purified as previously described with minor alteration [47, 48, 50, 52–55, 88, 89]. Unless otherwise noted, experiments were performed with proteins produced and purified by the authors. Purchased proteins were as follows: mature GDF8 (R&D Systems; Cat. no. 788-G8-CF and PeproTech; Cat. no. 120-00), mature GDF11 (R&D Systems; Cat. no. 1958-GD-010-CF and PeproTech (Cat. no. 120-11); Fc-ALK5 (R&D Systems; Cat. no. 3025-BR/CF), Fc-ActRIIB (R&D Systems; Cat. no. 339-RB/CF). Mature ligands purchased from R&D Systems were reconstituted in 4 mM HCl, 0.01% bovine serum albumin (BSA). Mature ligands purchased from PeproTech utilized in the in vitro studies were reconstituted in 4 mM HCl, 0.01% BSA. Produced and purified proteins are as follows: mature GDF8 [6, 18, 47, 50, 54, 89], mature GDF11 , latent GDF8/GDF11 chimeras (see below), mature activin A [52, 55, 62], mature activin B (see below), mature TGFβ3 , TβRII , FS288 [52, 54], FSTL3 [47, 55], GASP1 and GASP2 [50, 90], ActRIIB-ECD , and ALK5-ECD .
Chinese hamster ovary (CHO)-DG44 cells were co-transfected with linearized plasmids containing full-length human activin B (pcDNA3) and dihydrofolate reductase (pMT2). The transfected cells underwent multiple rounds of clonal selection performed with increasing concentrations of up to 1.24 μM methotrexate (MTX) to generate a stable cell line expressing activin B. High expressing clones were isolated and adapted to suspension culture in SFM4CHO-Utility (HyClone). Activin B was produced by adjusting the cells at 1 × 106 cells/mL to 5 mM sodium butyrate and culturing for 8 days, after which the conditioned medium was concentrated and purified as previously described .
GDF11-like changes were introduced into a pRK5 plasmid containing wild-type full-length human GDF8 using site-directed mutagenesis. GDF8/GDF11 chimeric protein was transiently produced using HEK293T co-transfected with human furin in pcDNA4 using polyethylenimine MAX (Polysciences, Inc.). Transfection proceeded for 4 h in culture medium followed by exchange into serum-free media. Conditioned media were collected 72–96 h later and concentrated 10 times. The concentrated media were applied to a Superdex S200 column (Amersham Biosciences). Fractions containing the latent complex were pooled and concentrated. Initial protein concentration and subsequent normalization were done by quantification (ImageJ) of SDS-PAGE colloidal Coomassie-stained gels.
GDF11:FS288 complex crystal structure determination
Purified GDF11 (Acceleron Pharma) was mixed with FS288 at a 1:2.5 molar ratio and purified over a Superdex 200 column (Amersham Biosciences) similar to previous purification of the GDF8:FS288 complex . The protein was then concentrated to 6.6 mg/mL and mixed 1:1 in a hanging drop experiment with a solution containing 125 mM phosphate citrate pH 4.2, 18% (w/v) EtOH, and 1% (w/v) PEG 1000. Diffraction experiments were performed at the Argonne National Laboratory Advanced Photon Source 23ID beamline and processed as previously described . Phasing was performed by molecular replacement using Phaser and the CCP4 suite using one dimer of GDF8 and one monomer of FS288 as the search model (PDB ID: 3HH2; ). Refinement was carried out with Refmac and Phenix. Coordinates have been deposited in the PDB (PDB ID: 5JHW).
apo-GDF8 crystal structure determination
Crystals of apo-GDF8 were obtained while attempting to generate crystals of the GDF8:GASP1 complex. Purified mature GDF8 [50, 54] was mixed at a 1:1.5 molar ratio with purified GASP1 [50, 90], and the complex was purified as previously described . The complex was concentrated to 10 mg/mL and mixed 1:1 in a hanging drop experiment with a well solution containing 100 mM MES pH 6.0 and 40% (w/v) 2-methyl-2,4-pentanediol (MPD). Crystals were readily obtained in the presence of noticeable protein precipitant. The resultant crystals contained only GDF8. Diffraction experiments were performed at the Argonne National Laboratory Advanced Photon Source 21ID beamline. Phasing was performed by molecular replacement using MolRep and the CCP4 suite using one monomer of GDF8 as the search model (PDB ID: 3HH2; ). Refinement was carried out with Refmac and Phenix. Coordinates have been deposited in the PDB (PDB ID: 5JI1).
apo-GDF11 crystal structure determination
Crystals of apo-GDF11 were obtained while attempting to generate crystals of the GDF11:ActRIIB complex. Purified mature GDF11, provided by Acceleron Pharma, was mixed at a 1:2.1 ratio with purified ActRIIB-ECD, and the complex was sized using a Superdex S75 column (Amersham Biosciences). The complex was concentrated to 7.5 mg/mL and mixed 1:1 in a hanging drop experiment with a well solution containing 5 mM CaCl2, 0.1 M sodium acetate pH 3.75, and 33% MPD. Similar to apo-GDF8, crystals were readily obtained in the presence of noticeable precipitant. The resultant crystals contained only GDF11. Diffraction experiments were performed at the Argonne National Laboratory Advanced Light Source 23-ID-D beamline. Phasing was performed by molecular replacement using Phaser and the CCP4 suite using one monomer of GDF11 as the search model (PDB ID: 5E4G). Refinement was carried out with Refmac and Phenix. Coordinates have been deposited in the PDB (PDB ID: 5UHM).
Luciferase reporter assays
HEK293 activity/potency and inhibition assays
The assays using the HEK293-(CAGA)12 luciferase reporter cells (initially derived from RRID: CVCL_0045) were performed as previously described [47–50]. Cells were seeded in a 96-well plate and grown for 24 h. For the activity comparison assays, the growth medium was then removed and replaced with serum-free medium + 0.1% BSA containing a twofold serial dilution series of mature GDF8 or GDF11 for 18 h. Inhibition assays were performed in a similar fashion, except that the ligand was held at a final concentration of 0.62 nM and then mixed with twofold serial dilutions of antagonist (FS288, FSTL3, GASP1, GASP2, ActRIIB-ECD, Fc-ActRIIB; R&D Systems). The cells were lysed and luminescence was recorded immediately using a Synergy H1 Hybrid plate reader (BioTek). The activity data were imported into GraphPad Prism and fit using a non-linear regression with a variable slope to calculate the EC50 or IC50.
For these transfection assays, the HEK293-(CAGA)12 luciferase reporter cells were plated in a 96-well plate at ~2 × 104 cells/well, grown for 24 h, and transiently co-transfected with plasmids containing GDF8/GDF11 chimeric constructs (25–100 ng; pRK5), human TLL2 (50 ng; pcDNA3), and human furin (50 ng; pcDNA4) using Mirus LT-1 transfection reagent. Empty pRK5 vector was added for a total of 200 ng DNA transfected/well. Transfection proceeded overnight followed by removal of growth medium in exchange for serum-free medium. Cells were cultured for an additional 24 h and assessed for luciferase activity as described above. The cells were plated and cultured similarly for the experiments where exogenous complexes were added to the cells. Following acid activation with 1 M HCl, the complexes were neutralized with 1 M NaOH and diluted in serum-free medium + 0.1% BSA. The culture medium was removed and the media containing the complexes was added to the cells. After 18–24 h, the cells were lysed and assessed for luciferase activity. For the TLL2 activation experiments, the cells were transfected with 50 ng TLL2 24 h prior to treatment with complexes diluted in serum-free media + 0.01% BSA. Cells were lysed 18–24 h later and assessed for luciferase activity.
HepG2 and LβT2 activity assays
The luciferase and hormone assays in HepG2 (ATCC; Cat. no. HB-8065; RRID: CVCL_0027) and LβT2 (kindly provided by Dr. Pamela Mellon, UCSD; RRID: CVCL_0398) cells were performed as previously described [51, 92]. Ligands were purchased from commercial sources for these experiments.
RIB L17 receptor utilization assays
The assays using RIB L17 (kindly provided by Dr. Joan Massagué; RRID: CVCL_0596) cells were performed as previously described with some minor alterations . To increase the experiment scale, the assay was performed in a 96-well plate. The cells were plated at ~2 × 104 cells/well and grown for 24 h. The cells were then co-transfected with a total of 100 ng DNA containing the (CAGA)12 luciferase reporter construct (gift from Dr. Anita Roberts ) and receptor containing plasmids (pRK5 rat ALK4, pRK5 rat ALK5, pcDNA3 human ALK7) alone or in combination using Mirus LT-1 transfection reagent. The ALK7 S270T variant was produced via mutagenesis with the following primer set: forward, GACTCAACTTTGGCTGGTAACTGAATATCATGAACAGGG; reverse, CCCTGTTCATGATATTCAGTTACCAGCCAAAGTTGAGTC. Empty pRK5 vector was added to normalize the total DNA concentration. Transfection proceeded overnight in culture medium followed by media exchange to serum-free media + 0.1% BSA containing 0.62 nM mature activin A, activin B, GDF8, GDF11, or TGFβ3. After 8 h, the cells were lysed and assessed for activity.
Surface plasmon resonance (SPR) studies
SPR analysis was performed similarly to previous studies [47, 50]. Briefly, experiments and protein dilutions were carried out in HBS-EP+ buffer (10 mM HEPES, pH 7.4, 500 mM NaCl, 3.4 mM EDTA, 0.05% P-20 surfactant, 0.5 mg/mL BSA) at 37 °C on a Biacore T200 optical biosensor system. All experiments were performed using a CM5 biosensor chip. Proteins were either immobilized by standard amine chemistry according to the manufacturer’s protocol or captured by using immobilized Protein A. For the specific experimental design, see the Results section. Data were analyzed using BIAevaluation software version 1.0.
Heparin affinity analysis
Determination of heparin affinity was performed as previously described . Briefly, 100 μg of FS288 alone or in complex with GDF8 or GDF11 was applied to a 1 mL HiTrap column (Amersham Biosciences) and eluted with a linear 2 M NaCl gradient over 120 column volumes.
Primary skeletal myoblasts
Primary skeletal myoblasts were isolated from limb muscles of mice aged 8–12 weeks. After dissection, limb muscles were washed in ice-cold phosphate-buffered saline (PBS). Next, tissues were removed from PBS, minced with surgical scissors, and placed in digestion media (0.2% collagenase type II, 0.05% dispase in DMEM) for 15 min while shaking at 37 °C. After digestion, muscle slurries were triturated until smooth and then digested for additional 8 min. 5 mL filtered donor bovine serum was added to stop the enzymatic reaction and the digestion, and the mixture was triturated again until smooth followed by addition of PBS and centrifuged at 1600 rpm for 5 min. Pellets were resuspended in PBS and filtered through a 70-μm cell strainer and centrifuged at 1600 rpm for 5 min. Pellets were resuspended in growth media (20% horse serum, 1% Glutamax, 1% penicillin/streptomycin, 5 ng/mL bFGF (Sigma-Aldrich) in F10 media) and added to coated culture dishes (0.2% rat-tail collagen, 5 μg/mL natural mouse laminin). Media were replaced after 48 h of culture. Cells were grown for another 48 h before the culture was enriched for myoblasts. Detached myoblasts were collected and centrifuged at 1200 rpm for 5 min and then re-plated on freshly coated culture dishes. Purity of the myoblast culture was assessed microscopically, and a second round of pre-plating was performed to maximize the purity before treating primary myoblasts with commercially purchased GDF8 or GDF11 (PeproTech; Cat. no. 120-00 and Cat. no. 120-11, respectively). Western analysis on cell lysates was performed using phosphorylated SMAD2/3 (Cell signaling; Cat. no. 8828S; Lot 6), total SMAD2/3 (Cell signaling; Cat. no. 3102S; Lot 9), and GAPDH (Santa Cruz; Cat. no. sc-25778; Lot I3015) primary antibodies. Antibody detection was performed with horseradish peroxidase-conjugated antibodies (Cell signaling; Cat. no. 7074S; Lot 25) and enhanced chemiluminescence (Amersham™ GE Healthcare; Cat. no. 45-002-401).
In vivo injection of ligands into mice
All animal studies were performed as approved by the Harvard Committee on Animals. Adult (1-year-old) C57Bl/6 male mice were obtained from Charles River, and intravenously injected (by tail vein injection) with 0.5 mg/kg GDF11 (PeproTech) or 0.5 mg/kg, 1 mg/kg, 2 mg/kg and 4 mg/kg GDF8 (R&D Systems) or saline as control. Ligands were reconstituted in water at a concentration of 1 mg/mL and diluted in saline prior to injection. Heart tissue was collected 1 h post injection. Whole heart protein lysates were obtained by homogenizing the heart in RIPA buffer freshly supplemented with 1 mM PMSF and protein phosphatase inhibitor 2 and 3 (Sigma-Aldrich). 40 g total protein was loaded in NuPAGE 4-12% Bis-Tris gels (LifeTechnologies). Following transfer, membranes were blocked with non-fat dry milk for 1 h at room temperature and successively incubated with primary pSMAD2 antibody (Millipore, Cat. no. AB3849; Lot 2649232) and total SMAD2/3 antibody (Cell Signaling Technology, Cat. no. 8685P; Lot 4) overnight at 4 °C. Proteins were detected with horseradish peroxidase-conjugated antibodies (BioRad Laboratories; Cat. no. 172-1019; Lot L006328 A) and enhanced chemiluminescence (Amersham™ GE Healthcare, Cat. no. RPN2236).
We would like to thank Ravindra Kumar for helpful discussions and generously providing GDF11 protein for our crystallographic studies and biochemical assays. We also thank members of the Thompson laboratory for helpful discussions regarding the manuscript.
This study was supported, in part, by the National Institutes of Health, the National Health and Medical Research Council Australia Grant, Michigan State University, the Paul F. Glenn Center for the Biology of Aging, the Muscular Dystrophy Association, the University of Cincinnati Graduate Dean Fellowship, and the American Heart Association (R01AG047131, R01AG040019, and R03AG049657 to RTL; Paul F. Glenn Center Grant and R56AG048917, R56AG052979 and R01AG048917 to AJW; R41AR06880401 to EMH; GM58670 and CA172886 to APH; CIHR MOP-133394 to DJB; 1078907 to CAH; R01GM114640 and Muscular Dystrophy Association Grant 240087 to TBT; Graduate Dean Fellowship and 12PRE11790027 to RGW).
Availability of data and materials
The atomic coordinates for the crystal structures (GDF11:FS288 = 5JHW, apo-GDF8 = 5JI1, apo-GDF11 = 5UHM) presented in this work are available in the RCSB Protein Data Bank (http://www.rcsb.org/pdb/home/home.do). The remaining datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.
RGW and TBT conceived the idea and designed the experiments. RGW prepared all figures and wrote the initial draft of the manuscript, performed the X-ray crystallography and structural refinement for the FS288:GDF11 complex and apo-GDF8 structures, performed the SPR, purified all components, and designed and optimized the initial HEK293 (CAGA)12 and RIB L17 cell-based assays. TBT edited the manuscript and helped with data interpretation and X-ray data collection and structural refinement. MC (University of Cincinnati) helped perform the HEK293 (CAGA)12 and RIB L17 cell-based assays. JCM helped with the RIB L17 cell-based assays. EJG performed the X-ray crystallography and structural refinement for the apo-GDF11 structure. SA and EMH designed and performed the HepG2 experiment and analysis. KLW and CAH performed and designed the LβT2 cell-based assays. GS and DJB provided reagents and input for the RIB L17 assay. APH provided ALK5-ECD and TGFβ3 proteins and input for subsequent binding studies. MC (Harvard University), JO, and AJW designed and performed the primary skeletal myoblast experiments. AV and RTL designed and performed the in vivo experiments. All authors revised and edited the manuscript. All authors read and approved the final manuscript.
TBT is a consultant for Acceleron Pharma. Harvard University and Brigham and Women’s Hospital have filed for intellectual property on GDF11 listing AJW and RTL as inventors. The other authors report no competing interests.
Consent for publication
Ethics approval and consent to participate
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Lee S-J. Extracellular regulation of myostatin: a molecular rheostat for muscle mass. IEMAMC. 2010;10:183–94.View ArticleGoogle Scholar
- Antsiferova M, Werner S. The bright and the dark sides of activin in wound healing and cancer. J Cell Sci. 2012;125:3929–37.View ArticlePubMedGoogle Scholar
- Pardali E, ten Dijke P. TGFβ signaling and cardiovascular diseases. Int J Biol Sci. 2012;8:195–213.View ArticlePubMedPubMed CentralGoogle Scholar
- Brazil DP, Church RH, Surae S, Godson C, Martin F. BMP signalling: agony and antagony in the family. Trends Cell Biol. 2015;25:249–64.View ArticlePubMedGoogle Scholar
- Walker RG, Poggioli T, Katsimpardi L, Buchanan SM, Oh J, Wattrus S, et al. Biochemistry and biology of GDF11 and myostatin: similarities, differences, and questions for future investigation. Circ Res. 2016;118:1125–42.View ArticlePubMedGoogle Scholar
- Lee S-J, McPherron AC. Regulation of myostatin activity and muscle growth. Proc Natl Acad Sci. 2001;98:9306–11.View ArticlePubMedPubMed CentralGoogle Scholar
- Rebbapragada A, Benchabane H, Wrana JL, Celeste AJ, Attisano L. Myostatin signals through a transforming growth factor-like signaling pathway to block adipogenesis. Mol Cell Biol. 2003;23:7230–42.View ArticlePubMedPubMed CentralGoogle Scholar
- Andersson O, Reissmann E, Ibáñez CF. Growth differentiation factor 11 signals through the transforming growth factor-beta receptor ALK5 to regionalize the anterior-posterior axis. EMBO Rep. 2006;7:831–7.PubMedPubMed CentralGoogle Scholar
- Hill JJ, Davies MV, Pearson AA, Wang JH, Hewick RM, Wolfman NM, et al. The myostatin propeptide and the follistatin-related gene are inhibitory binding proteins of myostatin in normal serum. J Biol Chem. 2002;277:40735–41.View ArticlePubMedGoogle Scholar
- Hill JJ, Qiu Y, Hewick RM, Wolfman NM. Regulation of myostatin in vivo by growth and differentiation factor-associated serum protein-1: a novel protein with protease inhibitor and follistatin domains. Mol Endocrinol. 2003;17:1144–54.View ArticlePubMedGoogle Scholar
- McPherron AC, Lawler AM, Lee S-J. Regulation of anterior/posterior patterning of the axial skeleton by growth/differentiation factor 11. Nat Genet. 1999;22:260–4.View ArticlePubMedGoogle Scholar
- Gamer LW, Cox KA, Small C, Rosen V. Gdf11 is a negative regulator of chondrogenesis and myogenesis in the developing chick limb. Dev Biol. 2001;229:407–20.View ArticlePubMedGoogle Scholar
- McPherron AC, Huynh TV, Lee S-J. Redundancy of myostatin and growth/differentiation factor 11 function. BMC Dev Biol. 2009;9:24.View ArticlePubMedPubMed CentralGoogle Scholar
- Trendelenburg A-U, Meyer A, Rohner D, Boyle J, Hatakeyama S, Glass DJ. Myostatin reduces Akt/TORC1/p70S6K signaling, inhibiting myoblast differentiation and myotube size. Am J Physiol Cell Physiol. 2009;296:C1258–70.View ArticlePubMedGoogle Scholar
- Jeanplong F, Falconer SJ, Thomas M, Matthews KG, Oldham JM, Watson T, et al. Growth and differentiation factor-11 is developmentally regulated in skeletal muscle and inhibits myoblast differentiation. OJMIP. 2012;2012:127–38.View ArticleGoogle Scholar
- Egerman MA, Cadena SM, Gilbert JA, Meyer A, Nelson HN, Swalley SE, et al. GDF11 increases with age and inhibits skeletal muscle regeneration. Cell Metab. 2015;22:164–74.View ArticlePubMedPubMed CentralGoogle Scholar
- Smith SC, Zhang X, Zhang X, Gross P, Starosta T, Mohsin S, et al. GDF11 does not rescue aging-related pathological hypertrophy. Circ Res. 2015;117:CIRCRESAHA.115.307527–932.Google Scholar
- McPherron AC, Lee S-J. Regulation of skeletal muscle mass in mice by a new TGF-beta superfamily member. Nature. 1997;387:83–90.View ArticlePubMedGoogle Scholar
- McPherron AC, Lee S-J. Double muscling in cattle due to mutations in the myostatin gene. Proc Natl Acad Sci. 1997;94:12457–61.View ArticlePubMedPubMed CentralGoogle Scholar
- Katsimpardi L, Litterman NK, Schein PA, Miller CM, Loffredo FS, Wojtkiewicz GR, et al. Vascular and neurogenic rejuvenation of the aging mouse brain by young systemic factors. Science. 2014;344:630–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Sinha M, Jang YC, Oh J, Khong D, Wu EY, Manohar R, et al. Restoring systemic GDF11 levels reverses age-related dysfunction in mouse skeletal muscle. Science. 2014;344:649–52.View ArticlePubMedPubMed CentralGoogle Scholar
- Loffredo FS, Steinhauser ML, Jay SM, Gannon J, Pancoast JR, Yalamanchi P, et al. Growth differentiation factor 11 is a circulating factor that reverses age-related cardiac hypertrophy. Cell. 2013;153:828–39.View ArticlePubMedPubMed CentralGoogle Scholar
- Poggioli T, Vujic A, Yang P, Macias-Trevino C, Uygur AN, Loffredo FS, et al. Circulating growth differentiation factor 11/8 levels decline with age. Circ Res. 2015;118:29–37.View ArticlePubMedPubMed CentralGoogle Scholar
- Zimmers TA, Davies MV, Koniaris LG, Haynes P, Esquela AF, Tomkinson KN, et al. Induction of cachexia in mice by systemically administered myostatin. Science. 2002;296:1486–8.View ArticlePubMedGoogle Scholar
- Wu H-H, Ivkovic S, Murray RC, Jaramillo S, Lyons KM, Johnson JE, et al. Autoregulation of neurogenesis by GDF11. Neuron. 2003;37:197–207.View ArticlePubMedGoogle Scholar
- Harmon EB, Apelqvist ÅA, Smart NG, Gu X, Osborne DH, Kim SK. GDF11 modulates NGN3+ islet progenitor cell number and promotes β-cell differentiation in pancreas development. Development. 2004;131:6163–74.View ArticlePubMedGoogle Scholar
- Kim J, Wu H-H, Lander AD, Lyons KM, Matzuk MM, Calof AL. GDF11 controls the timing of progenitor cell competence in developing retina. Science. 2005;308:1927–30.View ArticlePubMedGoogle Scholar
- Liu J-P. The function of growth/differentiation factor 11 (Gdf11) in rostrocaudal patterning of the developing spinal cord. Development. 2006;133:2865–74.View ArticlePubMedGoogle Scholar
- Bogdanovich S, Krag TOB, Barton ER, Morris LD, Whittemore L-A, Ahima RS, et al. Functional improvement of dystrophic muscle by myostatin blockade. Nature. 2002;420:418–21.View ArticlePubMedGoogle Scholar
- Whittemore L-A, Song K, Li X, Aghajanian J, Davies M, Girgenrath S, et al. Inhibition of myostatin in adult mice increases skeletal muscle mass and strength. Biochem Biophys Res Commun. 2003;300:965–71.View ArticlePubMedGoogle Scholar
- Rodino-Klapac LR, Chicoine LG, Kaspar BK, Mendell JR. Gene therapy for Duchenne muscular dystrophy: expectations and challenges. Arch Neurol. 2007;64:1236–41.View ArticlePubMedGoogle Scholar
- Kota J, Handy CR, Haidet AM, Montgomery CL, Eagle A, Rodino-Klapac LR, et al. Follistatin gene delivery enhances muscle growth and strength in nonhuman primates. Sci Transl Med. 2009;1:6ra15–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Rodino-Klapac LR, Haidet AM, Kota J, Handy C, Kaspar BK, Mendell JR. Inhibition of myostatin with emphasis on follistatin as a therapy for muscle disease. Muscle Nerve. 2009;39:283–96.View ArticlePubMedPubMed CentralGoogle Scholar
- Thies RS, Chen T, Davies MV, Tomkinson KN, Pearson AA, Shakey QA, et al. GDF-8 propeptide binds to GDF-8 and antagonizes biological activity by inhibiting GDF-8 receptor binding. Growth Factors. 2001;18:251–9.View ArticlePubMedGoogle Scholar
- Scott IC, Blitz IL, Pappano WN, Imamura Y, Clark TG, Steiglitz BM, et al. Mammalian BMP-1/tolloid-related metalloproteinases, including novel family member mammalian tolloid-like 2, have differential enzymatic activities and distributions of expression relevant to patterning and skeletogenesis. Dev Biol. 1999;213:283–300.View ArticlePubMedGoogle Scholar
- Ge G, Hopkins DR, Ho W-B, Greenspan DS. GDF11 forms a bone morphogenetic protein 1-activated latent complex that can modulate nerve growth factor-induced differentiation of PC12 cells. Mol Cell Biol. 2005;25:5846–58.View ArticlePubMedPubMed CentralGoogle Scholar
- Anderson SB, Goldberg AL, Whitman M. Identification of a novel pool of extracellular pro-myostatin in skeletal muscle. J Biol Chem. 2008;283:7027–35.View ArticlePubMedGoogle Scholar
- Yadin D, Knaus P, Mueller TD. Structural insights into BMP receptors: specificity, activation and inhibition. Cytokine Growth Factor Rev. 2016;27:13–34.View ArticlePubMedGoogle Scholar
- Gamer LW, Wolfman NM, Celeste AJ, Hattersley G, Hewick R, Rosen V. A novel BMP expressed in developing mouse limb, spinal cord, and tail bud is a potent mesoderm inducer in Xenopus embryos. Dev Biol. 1999;208:222–32.View ArticlePubMedGoogle Scholar
- Trexler M, Banyai L, Patthy L. A human protein containing multiple types of protease-inhibitory modules. Proc Natl Acad Sci. 2001;98:3705–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Trexler M, Banyai L, Patthy L. Distinct expression pattern of two related human proteins containing multiple types of protease-inhibitory modules. Biol Chem. 2002;383:223–8.View ArticlePubMedGoogle Scholar
- Kondás K, Szláma G, Trexler M, Patthy L. Both WFIKKN1 and WFIKKN2 have high affinity for growth and differentiation factors 8 and 11. J Biol Chem. 2008;283:23677–84.View ArticlePubMedPubMed CentralGoogle Scholar
- Rodgers BD, Eldridge JA. Reduced circulating GDF11 is unlikely responsible for age-dependent changes in mouse heart, muscle, and brain. Endocrinology. 2015;156:3885–8.View ArticlePubMedGoogle Scholar
- Hinken AC, Powers JM, Luo G, Holt JA, Billin AN, Russell AJ. Lack of evidence for GDF11 as a rejuvenator of aged skeletal muscle satellite cells. Aging Cell. 2016;15:582–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Harper SC, Brack A, MacDonnell S, Franti M, Olwin BB, Bailey BA, et al. Is growth differentiation factor 11 a realistic therapeutic for aging-dependent muscle defects? Circ Res. 2016;118:1143–50.View ArticlePubMedGoogle Scholar
- Padyana AK, Vaidialingam B, Hayes DB, Gupta P, Franti M, Farrow NA. Crystal structure of human GDF11. Acta Crystallogr F. 2016;72:160–4.View ArticleGoogle Scholar
- Cash JN, Angerman EB, Kattamuri C, Nolan K, Zhao H, Sidis Y, et al. Structure of myostatin · follistatin-like 3: N-terminal domains of follistatin-type molecules exhibit alternate modes of binding. J Biol Chem. 2012;287:1043–53.View ArticlePubMedGoogle Scholar
- Cash JN, Angerman EB, Keutmann HT, Thompson TB. Characterization of follistatin-type domains and their contribution to myostatin and activin A antagonism. Mol Endocrinol. 2012;26:1167–78.View ArticlePubMedPubMed CentralGoogle Scholar
- Cash JN, Angerman EB, Kirby RJ, Merck L, Seibel WL, Wortman MD, et al. Development of a small-molecule screening method for inhibitors of cellular response to myostatin and activin A. J Biomol Screen. 2013;18:837–44.View ArticlePubMedGoogle Scholar
- Walker RG, Angerman EB, Kattamuri C, Lee Y-S, Lee S-J, Thompson TB. Alternative binding modes identified for growth and differentiation factor-associated serum protein (GASP) family antagonism of myostatin. J Biol Chem. 2015;290:7506–16.View ArticlePubMedPubMed CentralGoogle Scholar
- Aykul S, Martinez-Hackert E. Transforming Growth Factor-β family ligands can function as antagonists by competing for type II receptor binding. J Biol Chem. 2016;291:10792–804.View ArticlePubMedGoogle Scholar
- Thompson TB, Lerch TF, Cook RW, Woodruff TK, Jardetzky TS. The structure of the follistatin:activin complex reveals antagonism of both type I and type II receptor binding. Dev Cell. 2005;9:535–43.View ArticlePubMedGoogle Scholar
- Lerch TF, Shimasaki S, Woodruff TK, Jardetzky TS. Structural and biophysical coupling of heparin and activin binding to follistatin isoform functions. J Biol Chem. 2007;282:15930–9.View ArticlePubMedGoogle Scholar
- Cash JN, Rejon CA, McPherron AC, Bernard DJ, Thompson TB. The structure of myostatin:follistatin 288: insights into receptor utilization and heparin binding. EMBO J. 2009;28:2662–76.View ArticlePubMedPubMed CentralGoogle Scholar
- Stamler R, Keutmann HT, Sidis Y, Kattamuri C, Schneyer A, Thompson TB. The structure of FSTL3.activin A complex. Differential binding of N-terminal domains influences follistatin-type antagonist specificity. J Biol Chem. 2008;283:32831–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Thompson LD, Pantoliano MW, Springer BA. Energetic characterization of the basic fibroblast growth factor-heparin interaction: identification of the heparin binding domain. Biochemistry. 1994;33:3831–40.View ArticlePubMedGoogle Scholar
- Hashimoto O, Nakamura T, Shoji H, Shimasaki S, Hayashi Y, Sugino H. A novel role of follistatin, an activin-binding protein, in the inhibition of activin action in rat pituitary cells. Endocytotic degradation of activin and its acceleration by follistatin associated with cell-surface heparan sulfate. J Biol Chem. 1997;272:13835–42.View ArticlePubMedGoogle Scholar
- Sidis Y, Schneyer AL, Keutmann HT. Heparin and activin-binding determinants in follistatin and FSTL3. Endocrinology. 2005;146:130–6.View ArticlePubMedGoogle Scholar
- Sidis Y, Mukherjee A, Keutmann H, Delbaere A, Sadatsuki M, Schneyer A. Biological activity of follistatin isoforms and follistatin-like-3 is dependent on differential cell surface binding and specificity for activin, myostatin, and bone morphogenetic proteins. Endocrinology. 2006;147:3586–97.View ArticlePubMedGoogle Scholar
- Huang T, Schor SL, Hinck AP. Biological activity differences between TGF-β1 and TGF-β3 correlate with differences in the rigidity and arrangement of their component monomers. Biochemistry. 2014;53:5737–49.View ArticlePubMedPubMed CentralGoogle Scholar
- Apgar JR, Mader M, Agostinelli R, Benard S, Bialek P, Johnson M, et al. Beyond CDR-grafting: structure-guided humanization of framework and CDR regions of an anti-myostatin antibody. MAbs. 2016;8:1302–18.View ArticlePubMedGoogle Scholar
- Thompson TB, Woodruff TK, Jardetzky TS. Structures of an ActRIIB:activin A complex reveal a novel binding mode for TGF-β ligand:receptor interactions. EMBO J. 2003;22:1555–66.View ArticlePubMedPubMed CentralGoogle Scholar
- Attisano L, Cárcamo J, Ventura F, Weis FM, Massagué J, Wrana JL. Identification of human activin and TGF beta type I receptors that form heteromeric kinase complexes with type II receptors. Cell. 1993;75:671–80.View ArticlePubMedGoogle Scholar
- Cárcamo J, Weis FM, Ventura F, Wieser R, Wrana JL, Attisano L, et al. Type I receptors specify growth-inhibitory and transcriptional responses to transforming growth factor beta and activin. Mol Cell Biol. 1994;14:3810–21.View ArticlePubMedPubMed CentralGoogle Scholar
- Tsuchida K, Nakatani M, Yamakawa N, Hashimoto O, Hasegawa Y, Sugino H. Activin isoforms signal through type I receptor serine/threonine kinase ALK7. Mol Cell Endocrinol. 2004;220:59–65.View ArticlePubMedGoogle Scholar
- Bernard DJ, Lee KB, Santos MM. Activin B can signal through both ALK4 and ALK7 in gonadotrope cells. Reprod Biol Endocrinol. 2006;4:52.View ArticlePubMedPubMed CentralGoogle Scholar
- Franzén P, ten Dijke P, Ichijo H, Yamashita H, Schulz P. Cloning of a TGFβ type I receptor that forms a heteromeric complex with the TGFβ type II receptor. Cell. 1993;75:681–92.View ArticlePubMedGoogle Scholar
- Bassing CH, Yingling JM, Howe DJ, Wang T, He WW, Gustafson ML, et al. A transforming growth factor beta type I receptor that signals to activate gene expression. Sci Am Assoc Adv Sci. 1994;263:87–9.Google Scholar
- Callahan JF, Burgess JL, Fornwald JA, Gaster LM, Harling JD, Harrington FP, et al. Identification of novel inhibitors of the transforming growth factor beta1 (TGF-beta1) type 1 receptor (ALK5). J Med Chem. 2002;45:999–1001.View ArticlePubMedGoogle Scholar
- Inman GJ, Nicolás FJ, Callahan JF, Harling JD, Gaster LM, Reith AD, et al. SB-431542 is a potent and specific inhibitor of transforming growth factor-beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol Pharmacol. 2002;62:65–74.View ArticlePubMedGoogle Scholar
- Laping NJ, Grygielko E, Mathur A, Butter S, Bomberger J, Tweed C, et al. Inhibition of transforming growth factor (TGF)-beta1-induced extracellular matrix with a novel inhibitor of the TGF-beta type I receptor kinase activity: SB-431542. Mol Pharmacol. 2002;62:58–64.View ArticlePubMedGoogle Scholar
- Ogunjimi AA, Zeqiraj E, Ceccarelli DF, Sicheri F, Wrana JL, David L. Structural basis for specificity of TGFβ family receptor small molecule inhibitors. Cell Signal. 2012;24:476–83.View ArticlePubMedGoogle Scholar
- Groppe J, Hinck CS, Samavarchi-Tehrani P, Zubieta C, Schuermann JP, Taylor AB, et al. Cooperative assembly of TGF-β superfamily signaling complexes is mediated by two disparate mechanisms and distinct modes of receptor Binding. Mol Cell. 2008;29:157–68.View ArticlePubMedGoogle Scholar
- Radaev S, Zou Z, Huang T, Lafer EM, Hinck AP, Sun PD. Ternary complex of transforming growth factor-beta1 reveals isoform-specific ligand recognition and receptor recruitment in the superfamily. J Biol Chem. 2010;285:14806–14.View ArticlePubMedPubMed CentralGoogle Scholar
- Wolfman NM, McPherron AC, Pappano WN, Davies MV, Song K, Tomkinson KN, et al. Activation of latent myostatin by the BMP-1/tolloid family of metalloproteinases. Proc Natl Acad Sci. 2003;100:15842–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Kondás K, Szláma G, Nagy A, Trexler M, Patthy L. Biological functions of the WAP domain-containing multidomain proteins WFIKKN1 and WFIKKN2. Biochem Soc Trans. 2011;39:1416–20.View ArticlePubMedGoogle Scholar
- Scheufler C, Sebald W, Hülsmeyer M. Crystal structure of human bone morphogenetic protein-2 at 2.7 Å resolution. J Mol Biol. 1999;287:103–15.View ArticlePubMedGoogle Scholar
- Keller S, Nickel J, Zhang J-L, Sebald W, Mueller TD. Molecular recognition of BMP-2 and BMP receptor IA. Nat Struct Mol Biol. 2004;11:481–8.View ArticlePubMedGoogle Scholar
- Nickel J, Kotzsch A, Sebald W, Mueller TD. A single residue of GDF-5 defines binding specificity to BMP receptor IB. J Mol Biol. 2005;349:933–47.View ArticlePubMedGoogle Scholar
- Saremba S, Nickel J, Seher A, Kotzsch A, Sebald W, Mueller TD. Type I receptor binding of bone morphogenetic protein 6 is dependent on N-glycosylation of the ligand. FEBS J. 2008;275:172–83.View ArticlePubMedGoogle Scholar
- Healey EG, Bishop B, Elegheert J, Bell CH, Padilla-Parra S, Siebold C. Repulsive guidance molecule is a structural bridge between neogenin and bone morphogenetic protein. Nat Struct Mol Biol. 2015;22:458–65.View ArticlePubMedPubMed CentralGoogle Scholar
- Greenwald J, Vega ME, Allendorph GP, Fischer WH, Vale W, Choe S. A flexible activin explains the membrane-dependent cooperative assembly of TGF-β family receptors. Mol Cell. 2004;15:485–9.View ArticlePubMedGoogle Scholar
- Harrington AE, Morris-Triggs SA, Ruotolo BT, Robinson CV, Ohnuma S-I, Hyvönen M. Structural basis for the inhibition of activin signalling by follistatin. EMBO J. 2006;25:1035–45.View ArticlePubMedPubMed CentralGoogle Scholar
- Pellaud J, Schote U, Arvinte T, Seelig J. Conformation and self-association of human recombinant transforming growth factor-beta3 in aqueous solutions. J Biol Chem. 1999;274:7699–704.View ArticlePubMedGoogle Scholar
- Bocharov EV, Korzhnev DM, Blommers MJJ, Arvinte T, Orekhov VY, Billeter M, et al. Dynamics-modulated biological activity of transforming growth factor beta3. J Biol Chem. 2002;277:46273–9.View ArticlePubMedGoogle Scholar
- Shi M, Zhu J, Wang R, Chen X, Mi L, Walz T, et al. Latent TGF-β structure and activation. Nature. 2011;474:343–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Khalil AM, Dotimas H, Kahn J, Lamerdin JE, Hayes DB, Gupta P, et al. Differential binding activity of TGF-β family proteins to select TGF-β receptors. J Pharmacol Exp Ther. 2016;358:423–30.View ArticlePubMedGoogle Scholar
- Jiang M-S, Liang L-F, Wang S, Ratovitski T, Holmstrom J, Barker C, et al. Characterization and identification of the inhibitory domain of GDF-8 propeptide. Biochem Biophys Res Commun. 2004;315:525–31.View ArticlePubMedGoogle Scholar
- Cawley SM, Sako D, Davies MV, Devine M, Grinberg AV, Quisel JD, et al. Transforming growth factor-β superfamily ligand trap ACE-536 corrects anemia by promoting late-stage erythropoiesis. Nat Med. 2014;20:408–14.View ArticlePubMedGoogle Scholar
- Lee Y-S, Lee S-J. Regulation of GDF-11 and myostatin activity by GASP-1 and GASP-2. Proc Natl Acad Sci. 2013;110:E3713–22.View ArticlePubMedPubMed CentralGoogle Scholar
- Schmelzer CH, Burton LE, Tamony CM, Schwall RH, Mason AJ, Liegeois N. Purification and characterization of recombinant human activin B. Biochim Biophys Acta. 1990;1039:135–41.View ArticlePubMedGoogle Scholar
- Makanji Y, Harrison CA, Stanton PG, Krishna R, Robertson DM. Inhibin A and B in vitro bioactivities are modified by their degree of glycosylation and their affinities to betaglycan. Endocrinology. 2007;148:2309–16.View ArticlePubMedGoogle Scholar
- Dennler S, Itoh S, Vivien D, ten Dijke P, Huet S, Gauthier JM. Direct binding of Smad3 and Smad4 to critical TGFβ-inducible elements in the promoter of human plasminogen activator inhibitor-type 1 gene. EMBO J. 1998;17:3091–100.View ArticlePubMedPubMed CentralGoogle Scholar
- Gore AJ, Philips DP, Miller WL, Bernard DJ. Differential regulation of follicle stimulating hormone by activin A and TGFB1 in murine gonadotropes. Reprod Biol Endocrinol. 2005;3:73.View ArticlePubMedPubMed CentralGoogle Scholar
- Lee KB, Khivansara V, Santos MM, Lamba P, Yuen T, Sealfon SC, et al. Bone morphogenetic protein 2 and activin A synergistically stimulate follicle-stimulating hormone beta subunit transcription. J Mol Endocrinol. 2007;38:315–30.View ArticlePubMedGoogle Scholar
- Allendorph GP, Vale WW, Choe S. Structure of the ternary signaling complex of a TGF-beta superfamily member. Proc Natl Acad Sci. 2006;103:7643–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA. Electrostatics of nanosystems: application to microtubules and the ribosome. Proc Natl Acad Sci. 2001;98:10037–41.View ArticlePubMedPubMed CentralGoogle Scholar