Cooperative allostery and structural dynamics of streptavidin at cryogenic- and ambient-temperature

Multimeric protein assemblies are abundant in nature. Streptavidin is an attractive protein that provides a paradigm system to investigate the intra- and intermolecular interactions of multimeric protein complexes. Also, it offers a versatile tool for biotechnological applications. Here, we present two apo-streptavidin structures, the first one is an ambient temperature Serial Femtosecond X-ray crystal (Apo-SFX) structure at 1.7 Å resolution and the second one is a cryogenic crystal structure (Apo-Cryo) at 1.1 Å resolution. These structures are mostly in agreement with previous structural data. Combined with computational analysis, these structures provide invaluable information about structural dynamics of apo streptavidin. Collectively, these data further reveal a novel cooperative allostery of streptavidin which binds to substrate via water molecules that provide a polar interaction network and mimics the substrate biotin which displays one of the strongest affinities found in nature.

18. Fig. 6 C, mark the se-biotin. 19. Fig. 7. What is seen here? It is hard to identify the purpose of this figure.
20. Fig. 8. What is an ellipsoid structure? Do the authors mean that thermal ellipsoids were assigned to each (main chain carbon?) atom of the SA structure. What are the red boxes? It is not sufficient to refer to them in the text, they must be explained in the figure caption. Fig. 8C, highlight the biotin in the active site with a different color. 21. Fig. 9. Details such as "atoms N1, C2, C9 and O12 were selected as nodes" do not belong to a figure caption. The cross-correlation heat maps of the SA and SA-biotin seems to be identical. Point out differences, and explain in short in the figure caption. Fig. 9C and Fig. 9D have the same caption "Intrachain correlation differences 5JD2 over Apo_SFX structure (results)". Please check. MSQ fluctuations from the GNM are ok to show, but should be overlaid over observed B-factors (does this agree?).
Reviewer #2 (Remarks to the Author): The manuscript entitled "Cooperative Allostery and Structural Dynamics of Streptavidin at Cryogenicand Ambient-temperature" by Ayan et al., presents 2 structures of Apo Streptavidin, one at room temperature (SFX structure) and the second at cryo temperature (synchrotron). The authors performed a Gaussian Network Model analysis on the Apo structure SFX and on a previously solved structure of the Streptavidin bound to selenobiotin. The GNM analysis provides a dynamic insight of the Streptavidin cooperative allostery, but on the other hand, the GNM study did not seem to need a new apo SFX structure of Streptavidin as plenty of structures of Streptavidin are already available.
Minor comments: 1) Introduction: please add unit for the Kd: "Kd = 10^-13 to 10^-14 M" 2) Introduction: "The conformational dynamics of streptavidin and its interactions shows" I think it should be show and not shows 3) Introduction: "The conformational dynamics of streptavidin and its interactions shows … type of molecule [15]." I don't really understand this sentence, could you please re-formulate it? 4) Materials and methods: Data collection and analysis for cryo-synchrotron studies at SSRL: Is the crystal used for the cryo data from the same pool than the crystals for SFX? Same size? 5) Materials and methods: Data processing for SFX..: Could you please give more details how you define the resolution cut-off for the SFX data with CrystFEL? With CCTBX.XFEL, people use both CC1/2 and redundancy in the high-resolution shell. The C1/2 must decrease in a monotonic fashion and the redundancy for the high-resolution shell must be at least 10-fold. 6 Figure 11 and 12: are not mentioned at all in the manuscript 17) Supp table 1: please report Rmeas and not Rmerge. 18) Supp table 1: Statistic seems very poor for Cryo-Streptavidin for completeness. I would not publish a structure with 48% completeness in the high-resolution shell neither with 80% completeness overall. I would have a least 90% for the high-resolution shell and >95% overall. The CC1/2 is also quite low. You should cut off the resolution at a lower resolution to have a more acceptable completeness. 19) Supp table 1: Could you add redundancy and completeness for the high-resolution shell for SFX data? 20) Supp table 1: Note 1: One crystal was used for each dataset. This is true for the synchrotron data but untrue for the SFX data. 21) Supp table 2 to 6: I would have like to have the RMSD for the loop 3,4 added to the tables (maybe in parenthesis). 22) To make interaction matrices more readable, I would add where the L3,4 is in the sequence number.
Major comments: 1) From the manuscript, I understand there is no significant difference between the cryo-SSRL structure and the room temperature (RT) MFX structure. How can the authors justify the use of XFEL for solving the RT structure? Radiation damage should not be a main concern as there is no metal/ion in the protein. Using an XFEL for just a RT structure is an overkill as you can do it routinely in synchrotron.
2) What are the new features the SFX structure is bringing compared to the synchrotron structure or what have been done before?
3) The general feeling on this paper is that the GNM study could have been done with previously solved structure. I don't think the GNM has to use a RT structure. In molecular dynamic simulation, most of the model are issued from cryo-crystallography. For example, 1SWB and 5JD2 are also tetramer and could have been use for this study.
In conclusion, the novelty of this paper does not reside into the structures as they are very similar to previous ones and don't seem to bring anything new, but in the GNM analysis. The GNM analysis shows the importance of key residues in the mechanism of cooperative allostery. To confirm hypothesis from GNM, the authors suggest at the end of the discussion a time resolved study in the future. This kind of study will definitely justify a serial crystallography approach.
Reviewer #3 (Remarks to the Author): General Remarks: Streptavidin is a very important biotechnological tool, being able to capture biotin and biotin derivatives with affinities higher than other commonly used capturing molecules, i.e. antibodies, aptamers, etc. Albeit engineered derivatives were reported to work as monomers or dimers, the wildtype molecule, that coordinates four streptavidin monomers, holds the highest affinity. However, full occupancy of all four binding sites is still debated. The quaternary structure was solved several times, at varying resolutions, with different ligands, and by different approaches, x-ray crystallography, and later by cryoEM. Streptavidin can be considered a model system to study how multimeric proteins exert their function and how each monomer influences the others is still a focus of study, intrinsically challenging. In streptavidin, Trp120 appears to be important in the intradimer allostery. Ligand binding uses loop 3/4 appears to function as a "lid", closing over the binding pocket when biotinbound. However, in the apo-state L3/4 appears very flexible and many times unsolved. How the "lid" opening and closing propagates to the binding pocket and how this propagates to the neighboring monomer through trp120 remains under study.
The present study tackles the above questions using the latest systems to resolve crystal structures at ambient temperatures. This new approach is providing novel conformational states at atomic resolution of several proteins and also of highly complex systems as the 30S ribosome. The authors take a further step and solve the same apo structure using classical x-ray diffraction at cryogenic conditions. In addition, the authors compare their new structures to previously reported structures of streptavidin complexed with seleno-biotin/biotin or other apo states. Ayan et al. describe defined electron densities for L3/4 for the Apo structures, at both, ambient and cryogenic conditions. Interestingly, not all L3/4 loops appeared in the open state. 3 out of 4 lids appeared open while one was similar to biotin-bound closed lid. A comparison with the seleno-biotin bound structure, also obtained at ambient temperature, indicates a certain asymmetry among monomers. In the bound state, 3 out 4 lids were closed while one stayed open. Another interesting finding relies on the intramonomer conformational coupling between the lid and the binding site. For instance, the density of positive charges in the binding pocket decreases if the lid is closed, similarly to when biotin is bound. This indicates that redistribution of charges in the binding pocket is not mediated by biotin. Also, coordinated water molecules are affected, yet more dependent on biotin binding rather than the lid state. In general, the coordinated number of water molecules appears reduced when biotin bound. The Gaussian Network Model analysis highlights several intra-and interchain correlated conformational changes.
Altogether, the work of the DeMirci group provides new results and analysis defining how the tetrameric streptavidin orchestrates to capture biotin. The work is technically well executed by an expert team. Nevertheless, the manuscript requires a number of adjustments, here you can find some suggestions: Major concerns: Manuscript: 1) The main goal and objectives of the study are not clearly introduced to the reader. Also, the reason behind the necessity to solve the structure at ambient or physiological temperatures is not clear, what is expected? How is this approach providing new insights into other complexes/proteins? Why is it important to solve the Apo state? Introducing the reader to the above can tremendously increase the clarity of the manuscript.
2) The results section appears very slim. Seven main figures and seven supplementary figures are packed into two pages of text. This section could be extended with better descriptions of the main findings. Now they appear hidden within other less important results or comments. For example: in the 2nd paragraph, "We observed additional electron density that belongs to the residues in the L3,4 region which were not modeled in previous studies". This deserves a more detailed description of the result. Here go some examples where the authors could expand and explain better: 3) Page 5: "… we observed the minor conformational changes ..." The authors do not define or mention how many As imply minor of major conformational changes. 4) Page 7: "Slowest 10 modes reveals the globul residue motions", I am not sure if the authors meant "global" or "globular". 5) Page 8: "These results confirm biotin binding may not have a stabilization effect however contribute to the allostery of streptavidin" This is not clear,the authors may need to rephrase this sentence. Allostery is a property of certain proteins, then, it is not clear how the biotin binding might contribute to the allostery of streptavidin. Ligands can cause allosteric responses, they do not contribute to allostery. 6) Page 9: "However, our atomic resolution structural and GNM data suggest that there is a predisposed cooperative allosterism before binding of the first biotin molecule." This sentence is not clear, do authors mean that "predisposed cooperative allosterism" is a property of the streptavidin? Or do they imply the protein has a certain predisposition to bind, particularly, biotin with cooperative allosterism? 7) In general, the results segment could be described in more detail, likely considering rearranging some of the main figures. A good rule of thumb is that each figure should relate to a results segment, supporting a well described finding. Scientific: 11) I remained puzzled about the similarities between the ambient and cryo structures. What defines the conformational state of streptavidin? How much the crystallization process is influencing? What is the contribution of the increased temperature? 12) How do the new apo structures compare to other Apo-states that had been resolved previously? The comparison with PDB: 6J6K, obtained by cryoEM, shows different conformations of L3/4, all closed albeit obtained the Apo state (Fig. 4), how do you interpret this? 13) Page 9: "Together with observed conformational changes in the biotin binding pocket (Fig 5), apo streptavidin (PDB ID: 6J6K) obtained by CryoEM was superposed with our SFX structure (Supp Fig 3). The comparison of the structures suggests that the two techniques can capture alternative binding conformations and expand the conformational space sampling of the active site loop." Taking into account that the RMSD between the binding site residues of apo and holo structures is less than 0.4A, how can the authors assure these differences are only usual fluctuations of the residues, or error during the experimental elucidation of the structures, or they are actually real differences between the apo and holo states of streptavidin? 14) During the GNM analysis, the authors mention they used the 10 slowest modes to be correlated with the global motions of the protein, and the 10 fastest modes for localized motions. So, how many modes were in total calculated by the GNM analysis?
Minor points: 15) Supplementary Figure 1 refers to structures obtained in "2020", intended to the structures of this work. Please update to 2021 and indicate that they belong to "this work". Also, "2014" in the captions should be bold to maintain consistency. 16) The units of the KD is missing in 4th line of the second paragraph of the introduction. Should read: KD = 10-14 to 10-15 M 17) Reference 10 suggests that modifications in biotin may cause a "more "disordered" L 3/4 loop in avidin, rather than in streptavidin. Although both are highly conserved, there are differences, specially in L 3/4. Please, indicate what is the important role of L3/4 mentioned in the last sentence of the second paragraph of the introduction. 18) Nomenclature of the different structures analyzed: please keep consistency, specially related to the SFX (ambient) structures. Sometimes it is called "ambient structure", "SFX structure" "APO-SFX", and so on. 19) Bound not bounded, correct along the text.

Point by Point Responses to Reviewers' comments
Please find below our point-by-point responses to our Reviewers' comments written in blue text. All changes on the manuscript are also highlighted with yellow color on the main text and supplementary files.

RESPONSE:
We sincerely apologize for all the typos and other issues. We extensively reviewed our manuscript and fixed all the existing points raised by our Reviewer 1. Tables 2 to 6, there is not so much information in the individual tables. RESPONSE: As suggested by our Reviewer, we merged these Tables.

2.
Since all structures are essentially identical, explore, in short, why this is interesting. The Gaussian network decomposition seems to be added to boost significance. This reviewer has doubts that the Gaussian network analysis produces trustworthy results. At least the MSQ fluctuations from GNM can be superposed on the B-factors to see whether that agrees. RESPONSE: The structures are similar however they are not identical because the binding loop of selenobiotin (3/4 loop) conformations is distinct in the Apo_SFX vs 5JD2 (holo-SFX) structure. These differences in loop motions can also be detected with GNM analysis by analyzing the protein's global and local motions. Due to differences in the conformations, the network models are also distinct from each other and the Normal Mode Analysis of these models supports these data. Particularly, the binding of selenobiotin affects the residue fluctuations and their correlations which we discussed in detail. We actually selected the cutoff distance for calculating GNM based on the correlation with the b-factors in 5JD2. We rewrote the method section to clarify the steps in the calculation process and provided the missing details. We compared the theoretical temperature factors calculated with GNM to the experimental b-factors also by checking the Pearson correlation between them: the overall correlation was 0.785 in the 5JD2 structure and 0.646 in Apo-SFX. Higher cutoff in Apo-SFX would probably increase correlation at the binding site residues, however, we wanted to maintain compatibility in the analysis with the same parameter selection. As 3. It appears that this work is the result of an attempt to mix biotin with the SA crystals to follow the binding kinetics, but somehow this did not work (would have been great). RESPONSE: Thank you for reviewer 1's valuable comment and we apologize for not being able to explain our story line clearly. Our next obvious step as indicated by our Reviewer is to better understand the details of the binding kinetics and cooperative allosterism by performing time-resolved structural analysis by using the streptavidinbiotin fast-mixing kineto-crystallography technique. We would like to explain the main purpose of the current study more explicitly as follows: Streptavidin is a paradigm protein complex system with the highest affinity for its ligand among other proteins in its class. This reputation in affinity is due to the coordination of all monomers of streptavidin compared to the function of its subunits alone. Therefore, the binding kinetics of the streptavidin to the substrate is closely related to the occupancy of all four binding sites. Although a wide variety of structural studies have been carried out, the mystery of this high affinity still remains unsolved. In this study, we not only elucidated the high-resolution apo-structure of the protein at cryogenic and near-physiological temperatures but also emphasized several intra-and interchain correlated motions with GNM analysis. The dynamics of the structure obtained by the GNM analysis and the extensive polar interaction network in the binding region obtained by the Apo-SFX structure confirm for the first time presence of a novel cooperative allosterism of streptavidin. The choice of the Apo-structure in the study is, naturally, to observe the non-ligand-bound L3,4 conformation of this model system and compare it with our previous holo-SFX structure. In the X-ray crystallographic aspect of our study, the intermonomer function of Trp120 on structural cooperation and the effect of this on the loop 3/4 conformation, which acts like a "lid", was emphasized. Interestingly, not all L3/4 loops in the apo-structure were open. Only 3 lids were open and one was in a closed conformation such as ligandbound. We observed that this is related to the number of water molecules and hydrogen bonds in the binding site. When the lid is closed, we observe a decreased density of positive charges in the binding pocket, just as in the presence of the ligand. This demonstrates that redistribution of charges, the interaction of the coordinated water molecules and extensive polar interactions network in the binding pocket is not regulated by biotin but regulated with cooperative allosterism property of streptavidin. To confirm this hypothesis, we performed extensive GNM analysis, using both our apostructures and our previous holo-structure. The asymmetry between monomers in the ligand-bound structure and several intra-& intermonomer-related correlated motions with allostery provided more robust structural dynamics that define how streptavidin regulates to interaction with biotin via GNM analysis. All these results and our literature comparisons demonstrate the novel cooperative allosterism of streptavidin to bind to its ligand through water molecules that mimic the substrate and provide the polar network.  Fig 3)." RESPONSE: We would like to thank reviewer 1 for highlighting this obscure phrase. This sentence has been changed to "Similar to the selenobiotin-bound structure (Fig.  3), the cryo-EM structure of streptavidin in complex with biotin (PDB ID: 6j6j) were superposed with our Apo-SFX structure ( Supplementary Fig. 6)." 10. Please explain in more detail: "predisposed cooperative allosterism before binding of the first biotin molecule" RESPONSE: We apologize for this vague sentence. We fixed it as "However, our atomic-resolution structural and GNM data suggest that there is a predisposed property of streptavidin cooperative allosterism before binding of the first biotin molecule." 11. Define "vertical": "The next vertical step to better understand" RESPONSE: This part has been changed to "The next step to better understand…" 13. Please correct expression! As it is written, your crystals diffract with the packing material: "Together with this packing materials and techniques, we were able to get crystals diffracting 1.7 Å resolution." RESPONSE: ''Transport packing by using large quantities of cottons prevented physical damage of the crystals and successfully transportation to XFEL, was followed by diffraction to 1.7 Å resolution.'' 14. Please explain why molecular replacement with 'phaser' was necessary. What was the reason that model 5JD2 could not be used as a direct initial model that only needed to be adjusted by a rigid body refinement to fit to the XFEL and SSRL data? RESPONSE: The sentence re-formulated as " The structural dynamics of streptavidin and its interaction with small molecules demonstrate that we have a greater understanding of its structure-function relationship, which makes it easy to "plug and play" type of molecule.

4) Materials and methods: Data collection and analysis for cryo-synchrotron studies
at SSRL: Is the crystal used for the cryo data from the same pool than the crystals for SFX? Same size?
RESPONSE: The answer of question 4 from reviewer 2 is no. The crystals that are used for cryo and SFX structure come from the same batch, however, obtained with different crystallization techniques. While the large cryo-synchrotron crystal obtained with microbatch under oil, the micro crystals for SFX obtained by batch method. Additionally, unit cell parameters for both cryo and apo structure were indicated in Supplementary Table 1.

5) Materials and methods:
Data processing for SFX..: Could you please give more details how you define the resolution cut-off for the SFX data with CrystFEL? With CCTBX.XFEL, people use both CC1/2 and redundancy in the high-resolution shell. The C1/2 must decrease in a monotonic fashion and the redundancy for the high-resolution shell must be at least 10-fold. RESPONSE: Resolution cut-off for all crystallography data is generally based on CC* and not CC1/2. CC* should be > 0.5 (it is 0.87). The 10-fold redundancy criteria may be something the cctbx users abide by, but it was not considered in this paper due to data processing with CrystFEL for our SFX data. The redundancy in the high-resolution shell for SFX data is 63. So, if anything our SFX data is of higher resolution, the resolution cutoff is based on the Wilson plot (which is independent from CrystFEL processing) rather than moving the detector forward (it was not possible because of the chamber set-up). The Wilson statistics confirmed a good fit to the data down to a resolution of 1.7 Å. 6) Figure 1: I think you meant: to prevent crystal settling and not to prevent precipitation of proteins RESPONSE: We would like to thank reviewer 2 for his/her comment. Figure 1 is replaced as Supplementary Fig. 3. The legend of Supplementary Fig. 3 Table 1 with the high-resolution shell values for all reported criteria and apologize for this gross oversight in the original submission. (The error in the reported overall completeness value was due to an inaccuracy in setting the low-resolution limit when calculating the statistics, it was formerly mistakenly set to 50 Å instead of 48.49 Å. This has now also been corrected). We were not diffraction-limited for this experiment. It was, in fact, not possible to move the detector closer to the interaction region due to experimental constraints. The Wilson statistics confirmed a good fit to the data down to a resolution of 1.7 Å.  Major comments: 1) From the manuscript, I understand there is no significant difference between the cryo-SSRL structure and the room temperature (RT) MFX structure. How can the authors justify the use of XFEL for solving the RT structure? Radiation damage should not be a main concern as there is no metal/ion in the protein. Using an XFEL for just a RT structure is an overkill as you can do it routinely in synchrotron. 2) What are the new features the SFX structure is bringing compared to the synchrotron structure or what have been done before?

20) Supp
3) The general feeling on this paper is that the GNM study could have been done with previously solved structure. I don't think the GNM has to use a RT structure. In molecular dynamic simulation, most of the model are issued from cryocrystallography. For example, 1SWB and 5JD2 are also tetramer and could have been use for this study. RESPONSE: For major comments 1, 2 and 3: We would like to thank our Referee for bringing up those issues. We agree with our Reviewer according to the data obtained by SFX. The main purpose of this study was not to get only the RT structure of streptavidin without radiation damage, which may provide the most physiologically relevant data. The femtosecond data collection speed and high precision of the MFX instrument, low mosaicity of the small crystals collectively could improve the binding site residues' electron density. This may help to provide better conformation determination, which is crucial for ultimate GNM analysis. First, we compared the suggested ambient temperature synchrotron apo-structure of streptavidin (1SWB) with our Apo_SFX structure. 1SWB has 1.85 Å resolution, R-free 0.253, overall Robserved 0.174 while Apo_SFX structure has 1.7 Å resolution, R-free 0.2242 and Rwork 0.1904. The 1SWB structure has missing residues in chain B at 45-48th positions, in chain C and D at 46-48th positions as mentioned in Supplementary Fig. 15. Moreover, 1SWB was observed with lack of electron density at residues Gln24, Lue25, Val47, Glu51, Arg53 in different chains, thus this structure was not the best model for applying GNM analysis. Moreover, for a fair comparison, we compared the electron density of binding site residues between our Apo_SFX structure and with the latest synchrotron structure of apo streptavidin (PDB ID: 3RY1), but the electron density of the loops, unfortunately, was not enhanced significantly. Overall, 3RY1 has 1.03 Å resolution, R-free 0.135 and R-work 0.117, while Apo_SFX (PDB ID: 7EK8) has lower resolution. On the other hand, in our Apo_SFX data, some of the binding site residues' electron density enhanced, and continuous electron density without alternate side chains conformations was observed. Chain A at Apo_SFX structure between Ile30-Thr40, Ala46-Arg53, and Glu44 have better electron density and precise side-chain conformations compared to the 3RY1 structure as indicated in the following figure. In chain B, between Asn23-Gly26, Phe29-Leu39, Ser45-Gly48, Glu51-Val55, and Thr42 there is more precise electron density and lack of alternate side-chain conformations for Apo_SFX structure. In chain C Asn23-Leu25, Ile30, The32, Ala35, Glu44, and Glu51 have better and continuous electron density with less alternate conformations at Apo_SFX structure, however between Ala46-Ala50 and Arg53, the electron density is better at 3RY1, but similar conformation with Apo_SFX. Similarly, in Chain D Apo_SFX structure has a better density and precise conformations for Asn23-Leu25, Ile30, The32, Ala35-Ala38 (but not Asp36), and Thr42, however, for Glu44-Ser52 3RY1 structure has better electron density but similar conformation with Apo_SFX. The following figure was also added as Supplementary Fig. 16. The findings about 1SWB and 3RY1 comparison was also added in the result section. On the other hand, we would like to compare the closest data in GNM analysis. For this purpose the new Apo_SFX structure and our previous 5JD2 structure have the same X-ray source and RT data collection properties, that's why we minimized the temperature and device relevant differences and artifacts with those structures. This part was also mentioned at the end of the introduction section as "The new Apo_SFX structure, which has better resolution and electron density from previous ambient temperature apo-structures, and 5JD2 have the same X-ray source and ambient temperature data collection properties, which minimize the temperature and device relevant differences and artifacts with those structures for a proper comparison in GNM analysis." Moreover, as mentioned in the materials and methods section "For a better comparison for GNM analysis, 5JD2 and Apo_SFX crystals were obtained from the same batch with minimized artifacts such as crystallization conditions, mother liquor and protein sample." The figure was indicated as Supplementary Fig. 15 The figure was indicated as Supplementary Fig. 16.
In conclusion, the novelty of this paper does not reside into the structures as they are very similar to previous ones and don't seem to bring anything new, but in the GNM analysis. The GNM analysis shows the importance of key residues in the mechanism of cooperative allostery. To confirm hypothesis from GNM, the authors suggest at the end of the discussion a time-resolved study in the future. This kind of study will definitely justify a serial crystallography approach. ************************************************************************************** Reviewer #3 (Remarks to the Author): General Remarks: Streptavidin is a very important biotechnological tool, being able to capture biotin and biotin derivatives with affinities higher than other commonly used capturing molecules, i.e. antibodies, aptamers, etc. Albeit engineered derivatives were reported to work as monomers or dimers, the wild-type molecule, that coordinates four streptavidin monomers, holds the highest affinity. However, full occupancy of all four binding sites is still debated. The quaternary structure was solved several times, at varying resolutions, with different ligands, and by different approaches, xray crystallography, and later by cryoEM. Streptavidin can be considered a model system to study how multimeric proteins exert their function and how each monomer influences the others is still a focus of study, intrinsically challenging. In streptavidin, Trp120 appears to be important in the intradimer allostery. Ligand binding uses loop 3/4 appears to function as a "lid", closing over the binding pocket when biotin-bound. However, in the apo-state L3/4 appears very flexible and many times unsolved. How the "lid" opening and closing propagates to the binding pocket and how this propagates to the neighboring monomer through trp120 remains under study. The present study tackles the above questions using the latest systems to resolve crystal structures at ambient temperatures. This new approach is providing novel conformational states at atomic resolution of several proteins and also of highly complex systems as the 30S ribosome. The authors take a further step and solve the same apo structure using classical x-ray diffraction at cryogenic conditions. In addition, the authors compare their new structures to previously reported structures of streptavidin complexed with seleno-biotin/biotin or other apo states. Ayan et al. describe defined electron densities for L3/4 for the Apo structures, at both, ambient and cryogenic conditions. Interestingly, not all L3/4 loops appeared in the open state. 3 out of 4 lids appeared open while one was similar to biotin-bound closed lid. A comparison with the seleno-biotin bound structure, also obtained at ambient temperature, indicates a certain asymmetry among monomers. In the bound state, 3 out 4 lids were closed while one stayed open. Another interesting finding relies on the intramonomer conformational coupling between the lid and the binding site. For instance, the density of positive charges in the binding pocket decreases if the lid is closed, similarly to when biotin is bound. This indicates that redistribution of charges in the binding pocket is not mediated by biotin. Also, coordinated water molecules are affected, yet more dependent on biotin binding rather than the lid state. In general, the coordinated number of water molecules appears reduced when biotin bound. The Gaussian Network Model analysis highlights several intra-and interchain correlated conformational changes.Altogether, the work of the DeMirci group provides new results and analysis defining how the tetrameric streptavidin orchestrates to capture biotin. The work is technically well executed by an expert team. Nevertheless, the manuscript requires a number of adjustments, here you can find some suggestions: Major concerns: 1) The main goal and objectives of the study are not clearly introduced to the reader. Also, the reason behind the necessity to solve the structure at ambient or physiological temperatures is not clear, what is expected? How is this approach providing new insights into other complexes/proteins? Why is it important to solve the Apo state? Introducing the reader to the above can tremendously increase the clarity of the manuscript. RESPONSE: Cryogenic temperatures can introduce bias for structure determination as cryogenic temperatures may disrupt the overall protein backbone fold. The crystallographic data without radiation damage at ambient temperature provides better structure determination from small micro crystals by using ultrabright X-ray sources. Moreover, there are unsolved residues which are especially observed in the loop region and emphasized in the main text (discussion part in the last paragraph) as ''Ligand binding uses loop 3/4 appears to function as a "lid", closing over the binding pocket when biotin-bound. In the apo-state L3/4 appears very flexible and many times unsolved''. Moreover, the conformational changes in the binding pocket with and without ligand play a key role in the engineering of new biological tools using the streptavidin-biotin system. Ultrafast data collection speed and high precision of the MFX instrument could enhance this binding site residues' electron density, which provide better conformation determination, which is crucial for ultimate GNM analysis. Thus, while we were aware that there was an ambient temperature Apo-state structure of streptavidin (1SWB), we performed SFX for apo_streptavidin. 1SWB structure has missing amino acid residues at the crucial 3/4 loop at binding site (described in new supplementary fig. 15), however our Apo_SFX structure provides those residues and other missing ones with much improved electron density and resolution. As it is mentioned in our manuscript (in introduction section), "The new Apo_SFX structure, which has better resolution and electron density from previous ambient temperature structures, and holo 5JD2 structure have the same X-ray source and ambient temperature data collection properties, which minimized the temperature and device relevant differences and artifacts with those structures for a proper comparison in GNM analysis". By determining the ambient temperature apo-state streptavidin structure, we would like to have the closest template to compare ligand-bound state and apo-state comparison. For this purpose, "5JD2 and Apo_SFX crystals were obtained from the same batch with minimized artifacts such as crystallization conditions, mother liquor and protein sample." as newly mentioned in the materials and methods section.. This data was compared with the ligand-bound structure in GNM analysis to provide a better understanding of the effect of the ligand on the structure dynamics. Lastly, the indicated text (discussion part in the last paragraph) is added to the main text to emphasize the main goal of this study ''To determine and validate the accuracy of previous structures of streptavidin, SFX offers structural data without temperature and radiation side effects, leading to a solid template for future studies. The next step to better understand the details of this binding and cooperative allosterism is performing time-resolved structural analysis by using ultrabright and ultrafast XFELs.'' at the end of the discussion section. Also, "We re-evaluated the tetrameric structure of streptavidin by Gaussian Network Model We would like to thank the referee for this issue. We explained the minor and major differences in more detail with Supplementary Fig. 15 and 16 and in the result section as "First we compared the ambient temperature synchrotron apostructure of streptavidin (1SWB) with our Apo_SFX structure. 1SWB has 1.85 Å resolution, R-free 0.253 without R-free while Apo_SFX structure has 1.7 Å resolution, R-free 0.2242 and R-work 0.1904. The 1SWB structure has missing residues in chain B at 45-48th positions, in chain C and D at 46-48th positions as mentioned in Supplementary Fig. 15. Moreover, 1SWB was observed with lack of electron density at residues Gln24, Lue25, Val47, Glu51, Arg53 in different chains, thus this structure was not suitable for applying GNM analysis. Moreover, for a fair comparison, we compared electron density of binding site residues between our Apo_SFX structure and with the more recent synchrotron structure of apo streptavidin (PDB ID: 3RY1), but electron density of the loops, unfortunately, was not enhanced significantly. Overall, 3RY1 has 1.03 Å resolution, R-free 0.135 and R-work 0.117, while Apo_SFX (PDB ID: 7EK8) has 1.7 Å resolution, R-free 0.2242 and R-work 0.1904. On the other hand, in our Apo_SFX data, some of the binding site residues' electron density enhanced, and continuous electron density without alternate side chains conformations was observed. In particular, chain A at Apo_SFX structure between Ile30-Thr40, Ala46-Arg53, and Glu44 have better electron density and precise side-chain conformations compared to the 3RY1 structure as indicated in the following figure. In chain B, between Asn23-Gly26, Phe29-Leu39, Ser45-Gly48, Glu51-Val55, and Thr42 there is more precise electron density and lack of alternate side-chain conformations for Apo_SFX structure. In chain C Asn23-Leu25, Ile30, The32, Ala35, Glu44, and Glu51 have better and continuous electron density with less alternate conformations at Apo_SFX structure, however between Ala46-Ala50 and Arg53, the electron density is better at 3RY1, but similar conformation with Apo_SFX. Similarly, in Chain D Apo_SFX structure has a better density and precise conformations for Asn23-Leu25, Ile30, The32, Ala35-Ala38 (but not Asp36) and Thr42, however, for Glu44-Ser52 3RY1 structure has better electron density but similar conformation with Apo_SFX." 3) Page 5: "… we observed the minor conformational changes ..." The authors do not define or mention how many As imply minor of major conformational changes. RESPONSE: The manuscript has been changed with "we observed the minor conformational changes which are less than 1 Å at the residues…" for clearing this issue. 4) Page 7: "Slowest 10 modes reveals the globul residue motions", I am not sure if the authors meant "global" or "globular". RESPONSE: Thanks to the referee for pointing out this mistake. We apologize for the typo. We fixed it as global.

5) Page 8: "These results confirm biotin binding may not have a stabilization effect however contribute to the allostery of streptavidin" This is not clear,the authors may need to rephrase this sentence. Allostery is a property of certain proteins, then, it is not clear how the biotin binding might contribute to the allostery of streptavidin.
Ligands can cause allosteric responses, they do not contribute to allostery. RESPONSE: Thanks to the referee for pointing out this mistake. We fixed it as "These results confirm biotin binding may not have a stabilization effect however can cause the allosteric response of streptavidin" 6) Page 9: "However, our atomic resolution structural and GNM data suggest that there is a predisposed cooperative allosterism before binding of the first biotin molecule." This sentence is not clear, do authors mean that "predisposed cooperative allosterism" is a property of the streptavidin? Or do they imply the protein has a certain predisposition to bind, particularly, biotin with cooperative allosterism? RESPONSE: Thanks to the referee for pointing out this mistake. We fixed it as "However, our atomic resolution structural and GNM analysis suggest that there is a predisposed property of streptavidin cooperative allosterism before binding of the first biotin molecule." 7) In general, the results segment could be described in more detail, likely considering rearranging some of the main figures. A good rule of thumb is that each figure should relate to a results segment, supporting a well described finding. RESPONSE: We believe that our article has made good progress thanks to the pragmatic comments and suggestions of our reviewers. We tried to address all comments and suggestions, the organization and layout of the figures have been changed and also the results segment detailed. 8) Figure 1 does not add to the manuscript since the novel approach is not the main focus of the paper. Perhaps, Figure 1 Fig. 6, Fig. 8, Fig. 9). Perhaps it could be moved to supplementary figures.  23, 27, 45, 49, 88) are observed between the ambient and cryo structure (Fig.  2). These two structures come from the same crystallization condition (Pact Premier TM 100 mM MMT buffer pH 6.0 and 25 % w/v PEG 1500) and have the same space groups (P 1 21 1). The similar cell dimensions indicated below and Supplementary Table 1 However, we perform SFX to offer a more accurate model to minimize the temperature and device relevant differences and artifacts with those structures for a proper comparison in GNM analysis with the same X-ray source and ambient temperature data collection properties in this paper. Moreover, "5JD2 and Apo_SFX crystals were obtained from the same batch with minimized artifacts such as crystallization conditions, mother liquor and protein sample." to minimize the sample preparation bias as newly mentioned in the materials and methods section.  (Fig 5), apo streptavidin (PDB ID: 6J6K) obtained by CryoEM was superposed with our SFX structure (Supplementary Fig 3). The comparison of the structures suggests that the two techniques can capture alternative binding conformations and expand the conformational space sampling of the active site loop." Taking into account that the RMSD between the binding site residues of apo and holo structures is less than 0.4A, how can the authors assure these differences are only usual fluctuations of the residues, or error during the experimental elucidation of the structures, or they are actually real differences between the apo and holo states of streptavidin? RESPONSE: We would like to thank the reviewer for his/her comment. The RMSD was calculated for the overall structure, now we added the RMSD for the loop to provide a more detailed and clear expression of differences in the Supplementary Table 2 in the parenthesis. The major differences between 6J6K and our structure is generated because of the loop region (residues 42-52). We suggested that ligand binding using loop 3/4 appears to function as a "lid", closing over the binding pocket when biotinbound. The RMSD for the L3,4 is overall 3.572 (for chain A: 2.879, chain B: 4.484  chain C: 3.276 , chain D: 3.649) , leading to a major difference. 14) During the GNM analysis, the authors mention they used the 10 slowest modes to be correlated with the global motions of the protein, and the 10 fastest modes for localized motions. So, how many modes were in total calculated by the GNM analysis? RESPONSE: In GNM analysis for the N number of atoms selected, N-1 number of nonzero modes can be calculated. Thus, in this study, the selected atom number in 5JD2 structure with selenobiotin was 489; and 488 non-zero modes were calculated with GNM. Similarly, the selected atom number in the Apo-SFX structure is 476 (all carbon alpha atoms); and 475 non-zero modes were calculated with GNM. Also, the selected 10 slowest modes corresponded to the 10 modes with the smallest eigenvalue (high variance) and the 10 fastest modes correspond to the 10 modes with the highest eigenvalue (low variance). We show the high variance (bigger than 1) in the slowest 10 modes below. Furthermore, we had decided to only focus on the slow modes, so we removed any mention of the fast modes. We also rewrote the method section to provide the correct details and increase the reproducibility of the analysis. This paper could be a nice paper. However, there are grammatical mistakes, logical inconsistencies and odd expressions that make a review difficult if not impossible. This reviewer is saddened by the lack of attention to detail. This reviewer already pointed out numerous odd expressions in the previous version. The present version is not better. In order to make this publishable this reviewer urges the corresponding author, who spent most of his career in the US, to correct the text throughout and pay attention to expression and logical flaws. A number of examples are listed below. Please make sure that the paper is prepared to professional standard. odd sentence, please correct. How can structures have the same X-ray source? (maybe: structures are determined at the same X-ray source, or structures are determined from data collected at the same X-ray source).

15) Supplementary
108 The new Apo_SFX structure, which has better resolution 109 and electron density from previous ambient temperature apo-structures, and 5JD2 have the same 110 X-ray source odd sentence, please correct:  The changes the authors made to the manuscript improved both reading and clarity. I appreciated the clear answers in the rebuttal letter for all the different concerns raised by the reviewers. The introduction on GNM analysis is much better. The overall explanation the authors gave about the necessity for Apo-SFX structure is satisfactory. The second version of the manuscript is much better.
I still have few minor remarks: 1) I disagree with part of the answer in the rebuttal letter from my minor remark nº5. I was just curious about the criteria applied with CrystFEL to determine the resolution cut-off. I understand it is based on CC* which is quite specific for CrystFEL and XDS. I disagree with the sentence "Resolution cut-off for all crystallography data is generally based on CC* and not CC1/2." Different programs use different parameters, e.g. Dials uses CC1/2. Also, CC* is based on CC1/2. 2) Line 77: I don't understand the analogy to "plug and play" type of molecule 3) Line 92-94: I agree with the authors about the fact that "cryogenic temperature can introduce bias". Could the author backup this statement with reference(s)? 4) Lines 114, 280, 314, 361, 362, sup Fig 2 and 6…: Could you please either call in the manuscript the loop L3,4 or L3/4 loop? 5) Line 141: I don't understand "R-free 0.253 without R-free…" 6) Line 225: what is the cut-off distance? I suppose it is the cut-off for the correlation between residues. Why do you have 2 different cut-off distances? What is the unit? Å? 7) Line 307-309: Just being curious… You should have enough structural data to build an hybrid tetramer with each dimer having only one biotin bound. GNM studies on this tetramer may answer your hypothesis. Maybe for another paper... 8) Line 458: Replace ( by 28% 9) Line 465: I would add here the answer you gave me for my remark nº18 about the low completeness (~80%) in the high-resolution shell for the cryo structure. 10) Line 469: as you put the instrument used at LCLS, I would add the instrument BL-12-2 as well for SSRL. 11) Figure 4E: I don't really understand which one is 5JD2 and which one is Apo-SFX. In blue, you have close conformation so it should be 5JD2 and in red open conformation then Apo-SFX. But chain A from Apo-SFX is in closed conformation, hence my confusion for the Chain A in figure 4E. 12) Supplementary Fig. 1: What is the difference between green arrow and red dot? And what is the red arrow? 13) Supplementary Fig. 5: Yellow text is hard to read 14) Supplementary Fig. 11: Black square are not described in the legend 15) Supplementary The manuscript by Ayan et al. has increased in clarity from the previous version. Most concerns have been addressed. The novel Apo-SFX structure provides enhanced details of the tetrameric streptavidin, several segments and side chains are resolved continuously. Particularly, the 3,4 loop (L3,4) of each monomer is clearly resolved. The authors use GNM to explore the correlated movements of residues on each monomer as part of the Apo-SFX and compared to the biotin bound structure, also solved at ambient temperatures (PDB: 5JD2). However, the correlation between residues in L3,4 and others in the binding site, dimer interface, or dimer/dimer interaction appear to be unnoticed by GNM or they are absent. This, in turn, would indicate that the "Lid" movement does not propagate to other parts of the complex or it is not influenced by the rest of the complex. Nevertheless, the Apo-SFX structure shows that L3,4 is closed in one monomer out of three. The opposite occurs in the Holo-SFX structure, where one monomer shows an open lid albeit being bound to biotin. The authors and others interpret these findings as L3,4 being intrinsically flexible. However, this can be also explained by coordinated allosterism. In this case, the GNM should have evidenced such correlations.
Altogether, the work by the DeMirci group provides new insights into the functioning of the multimeric streptavidin.
Here, you can find several issues that were annotated during this revision.
Minor issues: Revise the order of supplementary figures. They do not appear orderly in the text.
Lines 65-67: "L3,4…. Is the crucial part of streptavidin that interacts with biotin and regulates the binding". References required. However, whether L3,4 loop regulates biotin binding or not may be overstated. Please, revise also the following sentence, appears redundant.
Lines 92-94: " However, the available apo-state structures are still limited as cryogenic temperatures can introduce bias for structure determination by disrupting the overall protein backbone fold." Review this sentence. First, there are apo-state structures available. Second, what is the evidence available for structural biases introduced by cryogenic temperatures?? Need some references here.  Table 1)." 11. l. 448. The authors admit in the rebuttal letter that they used phenix for thermal ellipsoid generation. This is not reflected here. ✓✓✓✓ Thank you Reviewer #1 for comment. Structure refinement was done via PHENIX by using TLS parameters, leading to the visualization of ellipsoid structure via PyMOL. The explanation is added to the method section which is titled as 'Temperature factor analysis and generation of ellipsoids'. The added sentence is indicated below with related reference.
-The generation of ellipsoid models via PyMOL [68] was enabled based on structure refinement with TLS parameters through PHENIX [69].

l. 495. what is a Kirchhoff matrix? ✓✓✓✓
It is basically the contact map that we define at the start of a Gaussian Network Model (GNM) analysis. It is an NxN symmetric matrix in which N is the atom number. The Hamiltonian potential calculated for GNM requires the Kirchoff matrix to be defined. In this study we used the python module ProDy for doing so; we simply defined the Kirchhoff matrices with the selected atom coordinates, selected a cut-off distance (7.3 Å) to assume pairwise interactions, and used the default spring constant of 1.0 for both our structures. We decided to add the information about the spring constant to the methods section as follows: "Default spring constant of 1.0 was used for both structures." We would like to thank reviewer 2 for his/her comment. The mentioned sentence was added to line 465. The sentence is indicated below.
-The resolution cutoff set to 1.1 Å without negatively impacting Rfree and Rwork (Supplementary Table 1) 10) Line 469: as you put the instrument used at LCLS, I would add the instrument BL-12-2 as well for

SSRL. ✓✓✓✓
The sentence is corrected with additional information as follows:.
-  Figure 4E: I don't really understand which one is 5JD2 and which one is Apo-SFX. In blue, you have close conformation so it should be 5JD2 and in red open conformation then Apo-SFX. But chain A from Apo-SFX is in closed conformation, hence my confusion for the Chain A in figure 4E. ✓✓✓✓ We believe our Reviewer 2 refers to Figure 8E and we first apologize for the confusion caused with the labels. To clarify, the labels in legends were changed from open to Apo-SFX (red) and from closed to 5JD2 (blue) for consistency which can be seen in the following figure. Although the conformations should be similar in chain A for both structures, water molecules in Apo-SFX structure are the main reason for similarity but they were not included in the GNM analysis. Selenobiotin fills the place of the waters in the Holo-SFX (5JD2) and we included atoms of this ligand in the analysis to see precisely the effect of the ligand. We think that is the reason for the fluctuation differences at binding sites between them. The following sentences were also added to the main text to clarify this issue:

11)
-Results: "Additionally, although the conformations should be similar in Apo-SFX chain A for both structures, fluctuation differences were observed between them at binding sites." -Discussion:"One reason for the striking difference in fluctuations is that selenobiotin was included in the GNM of Holo-SFX (5JD2) and, therefore, the ligand's effect on decreasing the flexibility of the region was more clear in the protein's global motions." Figure: The legends were changed from open to Apo-SFX (red) and from closed to 5JD2 (blue) for consistency.
12) Supplementary Fig. 1: What is the difference between green arrow and red dot? And what is the red arrow? ✓✓✓✓ The sentence is corrected with additional information as follows: "Red dots represent residues in contact with ligand (biotin), while inverted triangles that are colored in green and red represent functional residues of repeats." 13) Supplementary Fig. 5: Yellow text is hard to read ✓✓✓✓ Yellow text of Supplementary Fig. 7 (updated version) is replaced with a darker tone to increase the contrast.

15) Supplementary
The manuscript by Ayan et al. has increased in clarity from the previous version. Most concerns have been addressed. The novel Apo-SFX structure provides enhanced details of the tetrameric streptavidin, several segments and side chains are resolved continuously. Particularly, the 3,4 loop (L3,4) of each monomer is clearly resolved. The authors use GNM to explore the correlated movements of residues on each monomer as part of the Apo-SFX and compared to the biotin bound structure, also solved at ambient temperatures (PDB: 5JD2). However, the correlation between residues in L3,4 and others in the binding site, dimer interface, or dimer/dimer interaction appear to be unnoticed by GNM or they are absent. This, in turn, would indicate that the "Lid" movement does not propagate to other parts of the complex or it is not influenced by the rest of the complex. Nevertheless, the Apo-SFX structure shows that L3,4 is closed in one monomer out of three. The opposite occurs in the Holo-SFX structure, where one monomer shows an open lid albeit being bound to biotin. The authors and others interpret these findings as L3,4 being intrinsically flexible. However, this can be also explained by coordinated allosterism. In this case, the GNM should have evidenced such correlations.
Altogether, the work by the DeMirci group provides new insights into the functioning of the multimeric streptavidin.
Here, you can find several issues that were annotated during this revision.
Minor issues: