Crystal structure of a multi-domain human smoothened receptor in complex with a super stabilizing ligand

The Smoothened receptor (SMO) belongs to the Class Frizzled of the G protein-coupled receptor (GPCR) superfamily, constituting a key component of the Hedgehog signalling pathway. Here we report the crystal structure of the multi-domain human SMO, bound and stabilized by a designed tool ligand TC114, using an X-ray free-electron laser source at 2.9 Å. The structure reveals a precise arrangement of three distinct domains: a seven-transmembrane helices domain (TMD), a hinge domain (HD) and an intact extracellular cysteine-rich domain (CRD). This architecture enables allosteric interactions between the domains that are important for ligand recognition and receptor activation. By combining the structural data, molecular dynamics simulation, and hydrogen-deuterium-exchange analysis, we demonstrate that transmembrane helix VI, extracellular loop 3 and the HD play a central role in transmitting the signal employing a unique GPCR activation mechanism, distinct from other multi-domain GPCRs.

Despite the fact that closely related structures have been recently published, I believe this paper has the potential to both confirm and extend the previous structural studies of Smo, but will require significant additional work. My main concern is that the paper has almost no functional tests of the structural model using structure-guided mutagenesis to assess effects on Hedgehog signaling. Mutational analysis to test structural predictions was an integral part of the recent comparable papers describing Smo structures (pmids 27545348 and 27437577) and is essential to maximizing the impact of this study. In addition, there are several other issues that need to be addressed or clarified: Major Comments: 1. The effect of the E194M mutation (present in their crystallization construct) on the signaling activity of Smo should be assessed.

2.
A major source of novelty in the paper is the design of a super -stabilizing SMO ligand, and the discussion that this structure could lead to new drugs to combat Hh cancers. It would be important to test whether TC114 can also block signaling by drug resistant mutants of SMO, like D473H, which is resistant to vismodegib.
2. In their discussion of the comparison between the cholesterol-bound, vismo-bound and TC114-bound structures, they describe a role for ECL3, including the glyca n modification and I496, in occluding the sterol-binding site in the CRD. This leads the authors to suggest that the "outward tilt of helix VI and ECL3, and the displacement of the HD are likely to trigger a conformational change that activates the receptor." Since the mechanisms of SMO activation in response to sterols or Shh are still unclear, testing these models would be very valuable. For instance, the authors could make mutations that remove the glycan, or mutate I496, or even simply truncate the portion of ECL3 that is supposed to occlude the sterolbinding groove and ask what the effect is on signaling by SMO at baseline or in response to Shh, oxysterols, or cholesterol. If ECL3-occlusion is necessary for stabilizing the inactive state one might expect these mutations to be activating.
3. The normal-mode analysis and the HDX studies seem to be contradictory. The former suggests that the ECL3 and HD have significant flexibility but there is not much deuterium exchange seen in these regions in the HDX experiment. Since cholesterol has recently been shown to be a CRD agonist, it would be important compare the effect of 20S-OHC to cholesterol. HDX represents a unique approach, one the authors' could use more extensively to characterize conformational changes in SMO in response to ligands. One major issue is that the SMO protein used for HDX analysis includes a heterologous Rubredoxin protein inserted into ICL3, which will likely change the Smo conformational ensemble probed by HDX, especially on the cytoplasmic face (indeed exactly where the authors find most of their HDX effects). HDX experiments should be repeated on a SMO protein having native, unperturbed ICLs (or the fusion protein used should be evaluated to ensure that it displays intact Hedgehog signaling activity using a cell-based assay).
4. Related to the point #3 above the normal-mode analysis is likely to be unreliable because it does not account for the fact that SMO is in a membrane environment. Though computationally more intensive, the MD analysis should be redone simulating a membrane environment.
5. Two recent papers (PMIDs: 27705744 and 27545348) have suggested that cholesterol is a direct activating ligand for Smo, yet there is little discussion of this agonist effect. How might this impact the conformational changes proposed by the authors?
6. The authors suggest that the HD transmits the signal between the CRD and the 7TMD. The basis for this argument is not clear, nor is it tested with any mutagenesis experiments. The paper by Byrne et. al. shows mutagenesis data that argues against this hypothesis: mutations in the CRD itself (at the interface between the CRD and the rest of the molecule) leads to constitutive activity. Also, mutations of the disulfide bond in the HD domain (that should disrupt the HD conformation) lead to constitutive activity. The authors mistakenly state that this disulfide mutant cannot be activated; Byrne et. al. show that this mutant can be activated by the native ligand Shh.
Minor comments 1. The authors need to be much more careful in their efforts at citing the literature. Several papers that are highly relevant to this study and used in this study are not cited. Most important are papers showing that oxysterol/cholesterol bind to the CRD and describi ng the structures of the SMO CRD: PMIDs: 23954590, 27437577, 24351982, and 27705744. As one example amongst many, the structure of the zebrafish SMO CRD (from PMID 27437577) was used by the authors to solve the synchrotron structure, but this paper is never cited. These studies should be cited at the appropriate places throughout the manuscript. extracellular loop 3 (ECL3) and the HD play a central role in transmitting the signal from CRD to TMD employing a unique GPCR activation mechanism". This is a slight exaggeration. I agree that they show the CRD has a differing position when bound to different ligands, that the hinge domain is flexible and that different ligands impact HDX of SMO regions, particularly in the cytoplasmic side of the TMD. However, as they point out, the direct observation of TMD conformation upon ligand binding remains elusive. I think their data support the idea, but do not directly demonstrate it and so this should be toned down.
The authors provide sufficient detail in methods and descriptions for other researchers to be able to reproduce this work. Moreover, the data are in line with that in other publications.
Minor comments: • Page 3 Gli should be GLI • It would be interesting to know why the authors chose to co-express SMO-FLA with vismodegib and not the TC114 compound (is it less stable?) • Page 9 -Nature communications is a broad audience journal and so the authors should provide more explanation of their normal mode analysis for non-experts to interpret their conclusions -What is normal mode analysis? what would be expected? What has been tested? What assumptions are made? • Similarly, the HDX analysis might benefit from further clarity such as explaining that HDX occurs in solvent accessible parts of the protein, so changes in conformation upon ligand binding that reveal or mask protein regions can be measured. • In support of the notion that TMD conformational changes are critical for SMO activation are oncogenic mutations within transmembrane helices, particularly on the cytoplasmic side e.g. W535Ldo these studies reveal anything about how these mutations might activate SMO? • Page 11 "….while the CRD is *a* relatively.." • Page 13: Citation(s) should be provided in the discussion about drug resistance/ mutations in the TMD region • What are the technical reasons for why Byrne et al identified bound cholesterol and the authors did not? • Figure 3A: On the right side of the CRD there is a pink glycan (N188?)it is not clear which chain it is associated with (presumably pink), therefore the asparagine side chain should be shown.

Reviewer #3 (Remarks to the Author):
This manuscript describes the X-ray crystal structure of the Smoothened receptor, and while the structure of this multi domain receptor has recently been recently published by the Seibold laboratory, this current study provides new mechanistic details that complements that published work. Overall this is a well written and interesting story using a combination of X-ray crystallography, biophysical analysis, and molecular dynamics to characterise this important GPCR receptor. They also describe a novel SMO ligand TC114 that allowed for stabilisation of the receptor structure. The authors clearly describe the differences between their structure and the Siebold lab structures in the results of the manusc ript, however, it the Siebold structures are something that should be mentioned in the introduction of the manuscript as well.
There are a number of corrections that may strengthen and clarify points within the manuscript. With these changes I am supportive of eventual publication in Nature Communications.
Major points 1. The authors discuss the changed orientation of the CRD domain in the their two structures ( Fig. 1b), as well as comparing their orientation to the published structures from the Siebol d lab (Fig. 3A). From these structures it is very difficult to clearly observe the tilt in the CRD. This is particularly challenging in Fig. 3A, if the authors could more clearly label how each CRD in the different structures is tilted compared to their XF EL structure it might be more clear.
2. How do the HDX-MS decreases seen on the cytoplasmic face compare to previous studies using HDX-MS to examine ligand bound GPCR complexes?
3. The HDX-MS data is shown fitted to a curve in Fig. S7. These fits appear to be very imprecise, and there is no discussion of how these fits were generated. I would recommend completely removing these lines.
4. The authors discuss that the HD and ECL3 would likely act as a hinge between the CRD and TMD. HDX in this study is only used to compare the apo and TC114 bound states. The authors do not discuss the H/D exchange rates for this region compared to the stable CRD and TMD domains. It would be expected that these hinge regions would be much more dynamic and the secondary structure elements here would undergo much more rapid deuteration. This needs to be discussed in the text, as the authors describe in the abstract that By combining the structural data, computer modelling, and hydrogen -deuterium-exchange analysis, we demonstrate that transmembrane helix VI, extracellular loop 3 (ECL3) and the HD play a central role in transmitting the signal from CRD to TMD employing a unique GPCR activation mechanism, I am not sure how the TC114 H/D exchange data demonstrates the role of the ECL3 and HD transmitting this signal from CRD to TMD. This could be addressed either by a new figure in the supplement showing that indeed the hinge region is more flexible than would be expected.
Minor points -In Fig 1A the primary sequence domain organisation is shown at a non -linear scale, and is confusing that the 30 amino acid hinge domain is the same length as the >300 amino acid TMD. 1

Reviewer #1:
My main concern is that the paper has almost no functional tests of the structural model using structure-guided mutagenesis to assess effects on Hedgehog signaling. Response: We have subsequently conducted a panel of cell-based functional tests of the structural model with structure-guided mutants to assess effects on Hedgehog signaling. The SMO cell assays are very challenging due to the usually low signals with or without ligand stimulation, especially for mutant tests. We spent significant efforts to set up this assay system and have the key mutants or treatment investigated. Details of these results have been added to the manuscript on page 6, 7, 10, 12. We have also included additional authors to the paper to reflect the contributions of this new work.
Major comments: 1. The effect of the E194M mutation (present in their crystallization construct) on the signaling activity of Smo should be assessed. Response: We tested the E194M mutant in a cell-based luciferase reporter assay, and compared the mutant with WT Smo. The results indicate that the E194M mutation elicits enhanced receptor activity to the WT. This result is mentioned on page 7 in the revised manuscript, and reported in Supplementary Fig. 2b.

2.
A major source of novelty in the paper is the design of a super-stabilizing SMO ligand, and the discussion that this structure could lead to new drugs to combat Hh cancers. It would be important to test whether TC114 can also block signaling by drug resistant mutants of SMO, like D473H, which is resistant to vismodegib. Response: In the original manuscript, we tested the antagonist activity of TC114 and showed that its IC50 is comparable to its prototype LY2940680. In the revised manuscript, we tested TC114 antagonist activity on the SMO drug resistant mutant D473H and W535L. We showed that TC114 could block agonist SAG-induced signaling to these two mutants with an IC50 ~4-fold higher than WT, by a luciferase reporter assay. These results are mentioned on page 6 in the revised manuscript and reported in Supplementary Fig. 2a. 3. In their discussion of the comparison between the cholesterol-bound, vismobound and TC114-bound structures, they describe a role for ECL3, including the glycan modification and I496, in occluding the sterol-binding site in the CRD. This leads the authors to suggest that the "outward tilt of helix VI and ECL3, and the displacement of the HD are likely to trigger a conformational change that activates the receptor." Since the mechanisms of SMO activation in response to sterols or Shh are still unclear, testing these models would be very valuable. For instance, the authors could make mutations that remove the glycan, or mutate I496, or even simply truncate the portion of ECL3 that is supposed to occlude the sterol-binding groove and ask what the effect is on signaling by 2 SMO at baseline or in response to Shh, oxysterols, or cholesterol. If ECL3-occlusion is necessary for stabilizing the inactive state one might expect these mutations to be activating.

Response:
In order to test our structure-based activation model, we made a panel of point mutations on the HD, ECL3 as well as glycan-modification residue. We characterized their effects on signaling by SMO at baseline conditions as well as in response to Shh and oxysterols. This data support our model that ECL3-occlusion is necessary for stabilizing the inactive state, as point mutations N493Q and I496R both showed increased basal activity of SMO signaling. Point mutations on the HD, V198R and K204A, can block the 20(S)-OHC, but not Shh, induced signaling. These results are described on page 10 and 12 in the revised manuscript and reported in Supplementary Fig. 2. 4. The normal-mode analysis and the HDX studies seem to be contradictory. The former suggests that the ECL3 and HD have significant flexibility but there is not much deuterium exchange seen in these regions in the HDX experiment. Since cholesterol has recently been shown to be a CRD agonist, it would be important compare the effect of 20S-OHC to cholesterol. HDX represents a unique approach, one the authors' could use more extensively to characterize conformational changes in SMO in response to ligands. One major issue is that the SMO protein used for HDX analysis includes a heterologous Rubredoxin protein inserted into ICL3, which will likely change the Smo conformational ensemble probed by HDX, especially on the cytoplasmic face (indeed exactly where the authors find most of their HDX effects). HDX experiments should be repeated on a SMO protein having native, unperturbed ICLs (or the fusion protein used should be evaluated to ensure that it displays intact Hedgehog signaling activity using a cell-based assay). Response: After an internal discussion, we agree that normal-mode analysis is quite preliminary and not appropriately suited for analyzing our model. We therefore deleted the normal-mode analysis part from the manuscript. As for the HDX analysis, we agree that such experiments with the non-fusion SMO protein would be more valuable and, hence, we tried very hard to conduct the experiment with the SMO protein that has native, unperturbed intracellular loops. However, the non-fusion SMO protein shows extremely poor expression (<0.05mg/L on yield) and reduced protein stability, compared to the Rubredoxin-fusion construct that we used for original HDX characterization. The poor stability of this protein makes it challenging to characterize with HDX, as the sequence coverage by MS is low. While we are continuing to optimize the HDX conditions, we think additional studies will be required to draw a comprehensive conclusion, which will be the focus pf follow-up work. In parallel, we tested the effect of 20(S)-OHC treatment on the HDX results, and found no difference on the H/D exchange rate, likely due to some non-specific interaction between 20(S)-OHC and the detergent micelles.
Based on these experimental trials, we propose to maintain the original HDX results and conclusions in the manuscript. TC114 showed very clear protection in the indicated cytoplasmic region and we have confidence reporting this from the current data/results. 5. Related to the point #3 (should be #4) above the normal-mode analysis is likely to be unreliable because it does not account for the fact that SMO is in a membrane environment. Though computationally more intensive, the MD analysis should be redone simulating a membrane environment. Response: Since the normal-mode analysis does not represent SMO in a membrane environment, we have replaced it with a long 1-μs MD simulation analysis with the protein embedded in a membrane environment. This analysis shows the overall stable conformation of the SMO CRD in the presence of cholesterol, similar to the 100-ns MD result reported in the paper by Byrne et al. Interestingly, the MD result showed that the cholesterol-stabilized CRD leans toward the membrane plane, in agreement with our structural comparison shown in Fig. 3a. This result is also consistent with our HDX analysis, where we do not see much H/D exchange at the CRD and TMD regions. We conclude that further investigation with a more intensive and thorough MD analysis is required, which will become the central focus of a follow-up paper. The result of suggested MD analysis is described on page 10 in the revised manuscript and reported in Supplementary Fig. 6. 6. Two recent papers (PMIDs: 27705744 and 27545348) have suggested that cholesterol is a direct activating ligand for Smo, yet there is little discussion of this agonist effect. How might this impact the conformational changes proposed by the authors? Response: According to the two recent papers (PMIDs: 27705744 and 27545348) and previous papers discussing SMO activation mechanisms, cholesterol and its oxymetabolite, as well as cyclopamine adopt similar binding poses via a common 3-β hydroxyl mediated hydrogen bond network. Their agonistic functions for SMO were also indicated. In our cell based luciferase reporter assay, we stimulated the cells with 20(S)-OHC for its better solubility to circumvent the use of cyclodextrin (Supplementary Fig.  2c). Similarly, 20(S)-OHC seemed to be better for HDX experiments that required a very high concentration of ligand. In discussing our activation model, we mentioned both oxysterol (20(S)-OHC) and cholesterol in the revised manuscript.
7. The authors suggest that the HD transmits the signal between the CRD and the 7TMD. The basis for this argument is not clear, nor is it tested with any mutagenesis experiments. The paper by Byrne et. al. shows mutagenesis data that argues against this hypothesis: mutations in the CRD itself (at the interface between the CRD and the rest of the molecule) leads to constitutive activity. Also, mutations of the disulfide bond in the HD domain (that should disrupt the HD conformation) lead to constitutive activity. The authors mistakenly state that this disulfide mutant cannot be activated; Byrne et. al. show that this mutant can be activated by the native ligand Shh. Response: We thank the reviewer for pointing out this mistake. We have removed this statement on page 12 in the revised manuscript. We have included and discuss our own HD mutation results in the revised manuscript and reported this data in Supplementary  Fig. 2.   4 Minor comments 1. The authors need to be much more careful in their efforts at citing the literature. Several papers that are highly relevant to this study and used in this study are not cited. Most important are papers showing that oxysterol/cholesterol bind to the CRD and describing the structures of the SMO CRD: PMIDs: 23954590, 27437577, 24351982, and 27705744. As one example amongst many, the structure of the zebrafish SMO CRD (from PMID 27437577) was used by the authors to solve the synchrotron structure, but this paper is never cited. These studies should be cited at the appropriate places throughout the manuscript. Response: We appreciate the reviewer's kind reminder and have included the key citations that the reviewer suggested at the appropriate places in the revised manuscript. We have also carefully checked all the citations throughout the paper to make sure the references are appropriate and reflect the state of the field.
2. The negative-stain EM is of poor quality and does not add anything to the paper. I would recommend that it be removed. Response: The EM data has been removed from the revised manuscript.
3. The authors suggest that the XEFL structure is of higher quality however there is no clear evidence for this-the statistical table shows that they are essentially identical. Indeed both structures were solved in the same space group and appear very similar. I don't think the authors can conclude much about conformational changes based on a qualitative comparison of these structures and this discussion should be made much more conservative.

Response:
We have revised the comparison between the two structures to be more conservative and concise. We have removed the comparison between the XFEL and Synchrotron structures, and have focused the discussion on only the XFEL structure in the revised manuscript.
4. The comparison with the frizzled protein is speculative and does not add much to the story. The hinge domain is one of the least conserved regions of the Frizzled family. Response: We believe these results are worth reporting. We agree that the hinge domain is not well conserved among the Frizzled family. In our revised manuscript, we deleted two ambiguous sentences and made a more cautious comparison on the CRD part, with a focus on the two ligand binding pockets, between SMO and the Frizzled proteins which are the most conserved in the Frizzled GPCR family.

Reviewer #2:
General comments/suggestions: 1. Overall the manuscript is well written, however, the order of figures is confusing. It would be better to introduce the development and rationale behind the TC114 compound before presenting the SMO structure, which is bound to the molecule. Response: We appreciate the reviewer's comment and have changed the order of Figures 1  & 2 as well as related text in the manuscript to present the design of TC114 first followed by presentation of the overall SMO structure.
2. The SMO structural data are unfortunately limited in their novelty due to the recent publication of a similar structure (Byrne et al. 2016. Nature). However, this is undoubtedly an independent study and the authors have acknowledged and compared their findings to those in the literature.

Response:
We thank the reviewer for this comment. Our study is independent and we acknowledged and compared our structure with the reported findings in the literature. We believe our study helps to strengthen the scientific data about this important receptor and further helps to clarify the role of the receptor in signal transduction.
3. The authors provide a comparison of the structure of their CRD region to refute claims by Huang et al (Cell, 2016) that the conformational changes in the CRD upon ligand binding are sufficient to activate SMO. It appears the presence of the hinge domain and ECL3 stabilize the CRD, limiting conformational changes upon ligand binding. A similar point is made by Luchetti et al. (ELife, 2016), which was published a few days after submission of this manuscript to Nature Communications. The authors might consider discussing this recent publication in support of their observation. This is an important point when considering a model for how a putative ligand might activate/inactivate SMO. Response: We have reviewed the new eLife paper by Luchetti et al, and agree with their point on the limited CRD conformations upon ligand binding. Since cholesterol and 20(S)-OHC bind to the same site on SMO and induce the same functional activity, the discussion on 20(S)-OHC is also applicable for cholesterol. We have added an appropriate citation to their study on page 9. 4. In the abstract the authors state: "we demonstrate that transmembrane helix VI, extracellular loop 3 (ECL3) and the HD play a central role in transmitting the signal from CRD to TMD employing a unique GPCR activation mechanism". This is a slight exaggeration. I agree that they show the CRD has a differing position when bound to different ligands, that the hinge domain is flexible and that different ligands impact HDX of SMO regions, particularly in the cytoplasmic side of the TMD. However, as they point out, the direct observation of TMD conformation upon ligand binding remains elusive. I think their data support the idea, but do not directly demonstrate it and so this should be 6 toned down. Response: We have toned down the statement on TMD conformations when discussing the SMO activation mechanism on Page 10 & 12 as well as in the Abstract as suggested by the reviewer.
Minor comments: • Page 3 Gli should be GLI Response: Corrected.
• It would be interesting to know why the authors chose to co-express SMO-FLA with vismodegib and not the TC114 compound (is it less stable?) Response: Co-expression with vismodegib increased the protein expression yield; we also tested other ligands including TC114, but none of them worked better than vismodegib. We believe one key factor is the nitro group in TC114 which is very reactive.
• Page 9 -Nature communications is a broad audience journal and so the authors should provide more explanation of their normal mode analysis for non-experts to interpret their conclusions -What is normal mode analysis? what would be expected? What has been tested? What assumptions are made? Response: We have removed the normal-mode analysis according to Reviewer #1's comment and included MD data in its place.
• Similarly, the HDX analysis might benefit from further clarity such as explaining that HDX occurs in solvent accessible parts of the protein, so changes in conformation upon ligand binding that reveal or mask protein regions can be measured. Response: As suggested, we have added a short clarification on the HDX experiment on Page 10 in the revised manuscript.
• In support of the notion that TMD conformational changes are critical for SMO activation are oncogenic mutations within transmembrane helices, particularly on the cytoplasmic side e.g. W535Ldo these studies reveal anything about how these mutations might activate SMO? Response: We do not see substantial conformational changes in W535 or its neighboring residues among all solved SMO structures, including agonist-or antagonist-bound TMD structures, as well as the two-domain structures, so we cannot currently correlate an effect between oncogenic mutations and SMO activation based on structural observations. 7 • Page 11 "….while the CRD is *a* relatively." Response: We have removed this sentence in response to reviewer #1, point #4 above.
• Page 13: Citation(s) should be provided in the discussion about drug resistance/ mutations in the TMD region. Response: The citation (PMID: 26614022) was added on page 14.
• What are the technical reasons for why Byrne et al identified bound cholesterol and the authors did not? Response: In our crystal structure, the side chain of I496 from ECL3 is positioned to extend into the cholesterol binding site and this precludes cholesterol or oxysterol binding to the CRD. We have stated this on Page 8 in the revised manuscript.
• Figure 3A: On the right side of the CRD there is a pink glycan (N188?)it is not clear which chain it is associated with (presumably pink), therefore the asparagine side chain should be shown. Response: The side chain of N493 which contains the glycan modification is now shown in the revised Fig. 3A, 3B.

Reviewer #3:
Major points 1. The authors discuss the changed orientation of the CRD domain in the their two structures (Fig. 1b), as well as comparing their orientation to the published structures from the Siebold lab (Fig. 3A). From these structures it is very difficult to clearly observe the tilt in the CRD. This is particularly challenging in Fig. 3A, if the authors could more clearly label how each CRD in the different structures is tilted compared to their XFEL structure it might be more clear. Response: Fig. 3b has been revised in the manuscript to clearly show the CRD tilt (direction and angle) on the cholesterol-bound structure and Vismodegib-bond structure with respect to our XFEL structure. The synchrotron structure has been removed from this figure for clarity.
2. How do the HDX-MS decreases seen on the cytoplasmic face compare to previous studies using HDX-MS to examine ligand bound GPCR complexes? Response: The HDX-MS decreases observed on the cytoplasmic side in our SMO structure are comparable to previous studies on the ligand-bound or apo GCGR structure (PMID: 26227798, Fig. 2).
3. The HDX-MS data is shown fitted to a curve in Fig. S7. These fits appear to be very imprecise, and there is no discussion of how these fits were generated. I would recommend completely removing these lines. Response: The HDX-MS data curve fits were generated in prism (we added the sentence "data was fitted to a simple nonlinear regression (least squares) best fit model (X is log and Y is linear) using GraphPad Prism" in the revised manuscript on Page 20-the Methods session). If we had more time points between 0 and 10s (the earliest we could measure), then the curves would look much smoother. So we think these fits in our Fig.  S5 are precise. We have removed these lines to make it appear more reasonable according to the reviewer's suggestion.
4. The authors discuss that the HD and ECL3 would likely act as a hinge between the CRD and TMD. HDX in this study is only used to compare the apo and TC114 bound states. The authors do not discuss the H/D exchange rates for this region compared to the stable CRD and TMD domains. It would be expected that these hinge regions would be much more dynamic and the secondary structure elements here would undergo much more rapid deuteration. This needs to be discussed in the text, as the authors describe in the abstract that "By combining the structural data, computer modelling, and hydrogendeuterium-exchange analysis, we demonstrate that transmembrane helix VI, extracellular loop 3 (ECL3) and the HD play a central role in transmitting the signal from CRD to TMD employing a unique GPCR activation mechanism", I am not sure how the TC114 H/D exchange data demonstrates the role of the ECL3 and HD transmitting this signal 9 from CRD to TMD. This could be addressed either by a new figure in the supplement showing that indeed the hinge region is more flexible than would be expected. Response: We conducted a panel of functional assays with SMO mutants that contain structure-guided point mutations on the HD and ECL3 domains, respectively. The result are now included in Supplementary Fig. 2 and described on page 10, 12 in the revised manuscript supporting the observation that both HD and ECL3 are critical for signaling. For the past 2 months, we have conducted additional HDX-MS experiment using the SMO protein without an ICL fusion protein to probe conformational flexibility in a more native state. However, as noted in the reply to reviewer 1, the non-fusion SMO protein is very unstable and hence not amenable for HDX analysis due to low MS sequence coverage. We therefore cannot draw conclusion from the HDX analysis regarding the flexibility of HD domain; further investigation along this line is required and we are continuing this work in a series of follow-up studies. As such, we toned down the wording in the abstract to make the statement more precise and consistent.
Minor points -In Fig 1A the primary sequence domain organisation is shown at a non-linear scale, and is confusing that the 30 amino acid hinge domain is the same length as the >300 amino acid TMD. Response: A new Fig. 2A (originally Fig. 1A and here moved to Fig. 2A as suggested by Reviewer#2) is made to reflect the length of each domain.