Solvent concentration at 50% protein unfolding may reform enzyme stability ranking and process window identification

As water miscible organic co-solvents are often required for enzyme reactions to improve e.g., the solubility of the substrate in the aqueous medium, an enzyme is required which displays high stability in the presence of this co-solvent. Consequently, it is of utmost importance to identify the most suitable enzyme or the appropriate reaction conditions. Until now, the melting temperature is used in general as a measure for stability of enzymes. The experiments here show, that the melting temperature does not correlate to the activity observed in the presence of the solvent. As an alternative parameter, the concentration of the co-solvent at the point of 50% protein unfolding at a specific temperature T in short \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${c}_{{U}_{50}}^{T}$$\end{document}cU50T is introduced. Analyzing a set of ene reductases, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${c}_{{U}_{50}}^{T}$$\end{document}cU50T is shown to indicate the concentration of the co-solvent where also the activity of the enzyme drops fastest. Comparing possible rankings of enzymes according to melting temperature and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${c}_{{U}_{50}}^{T}$$\end{document}cU50T reveals a clearly diverging outcome also depending on the specific solvent used. Additionally, plots of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${c}_{{U}_{50}}$$\end{document}cU50 versus temperature enable a fast identification of possible reaction windows to deduce tolerated solvent concentrations and temperature.

mixtures (Khmelnitsky, Y. L.; Mozhaev, V. V.; Belova, A. B.; Sergeeva, M. V.; Martinek, K. Eur.J. Biochem.1991, 198, 31).The authors should have considered this work and compared it with their own.5. Most importantly: the authors cherrypicked their data on four EREDs (ChrOYE1, OYE1, PpXenB, and TsOYE) of the 13 in the main text to make claims that are less firm or are strongly diminished when considering even all 13 ERED enzymes (not to say anything of ore EREDs or other enzymes) with data in the SI.As an example, SI Figure 6 shows F350/F330 data on the y-axis plotted against solvent content, here ethanol.Only 7 of the 13 EREDs show any usable transition!The applicability of the C,T,U50 approach to other systems, even other EREDs, seems tenuous at best and will not be approached with confidence.
For all these, and more, points, this manuscript should not be accepted.

Dear Authors
The manuscript Sorgenfrei et al. entitled "Solvent concentration at point of 50% protein unfolding   may reform enzyme solvent stability ranking and identification of process window" describes a novel approach to identify conditions for applying enzymes in the presence of water miscible co-solvents.Often the melting temperature is or was used to get a hint towards the process stability of enzymes also in the presence of solvents.However, as the authors state this is not always leading to a success, and one needs another more accurate routine to predict this process window.This important issue was now tackled by the authors and as role model they employed ene reductases.For this class of enzymes, a huge set of data are readily available from literature including thermo-as well as solvent stability, making it an ideal candidate for such an approach to establish a new routine to predict process applicability of enzymes in the presence of miscible co-solvents.Furthermore, the authors laboratories have a long-lasting experience with this enzyme family.All this expertise was used to generate a holistic new set of data to establish a new key parameter to evaluate enzyme stability towards solvents connected to enzyme performance.
This new concept will be of use for the community, especially in biocatalysis, but it may be of interest to a broader audience as well.Protein stability and folding is a very diverse field, and such supportive methods will help to uncover novel features among those molecules and their function.Thus, I can state this is a novel and original contribution to the field of high impact.
A hypothesis was formulated, and proper methods have been employed to answer the raised questions.The methods are well documented and sound reproducible.Needed standards to describe enzymes have been applied and especially for flavin dependent ene reductases.Those can be transferred to other enzymes as well.
In general, the manuscript is well prepared, scientific sound and all needed information are provided to follow the study.A new concept is presented and supported by a large set of data and well discussed.Proper conclusions were drawn from a sophisticated data analysis.
The reference list is appropriate and most relevant articles are cited; no excess self-citation determined.General formatting is of good quality and presentation of tables and figures very supportive.Nevertheless, some minor points need to be improved -nothing critical from my point of view.
In the following I will provide my comments and hope they are supportive and constructive: -This part of the sentence from line 24 is not sound: "showing depending also on the solvent itself" and might be rephrased.
-Line 78/79 while I agree that FMN is a true cofactor; NAD(P)H is rather a co-substrate for EREDs as it is consumed during catalysis and cannot be recycled by the enzyme in the same reaction course.
-The general observation towards Tm that DMSO is less effective while the alcohols are more effective as the chain length increases is quite interesting; was to some extent observed for oxidases as well (of course not that generalized as shown herein); it might have to do with the change in pKa and acidity of the HO-group.Can you comment on this? -Line 144 "spectroscopically" rather "spectrophotometrically" -Line 211 Add a space "(a) The …" -Line 263 as well as Line 275 and Line 285: correct "Equation ((3)." to "Equation (3)." -Line 398, correct "sequence" -Line 425, correct "temperature" -Ene reductase quantification; did you determine the FMN loading?As not all ene reductases are fully loaded, and what was used to set protein concentration (active sites; FMN-loaded or total protein?);specify.

Supplementary Material:
-The given activities are mostly presented as relative values; please indicate the 100% as U mg-1 or observed rate as s-1.
-SFIG3.: insert a space "pH 7.4using" to correct "pH 7.4 using" -SFIG4.: "cofactor" delete "-" General point: protein folding and unfolding is studied from many perspectives.Often unfolding curves are determined by titrating guanidine HCl or urea.I just wonder, if one could expect similar effects here as you observed by comparing unfolding and activity?From a protein point of view, you change in both cases the environment by a chemical agent leading to partial and later to total unfolding.I just wonder if your concept can be extended to other agents as well.

Reviewer #3 (Remarks to the Author):
This paper investigates parameters that influence enzyme activity in organic water miscible solvents.There is significant amount of literature suggesting that thermal stability can be taken as a surrogate for tolerance of organic solvent, but the authors show that the correlation is not always strong.They suggest that as an alternative the stability of the enzyme with respect to increasing amount of organis solvent should be used.They also show that solvent stability differes between solvents.Overall, these results are of interest in biotechnology, where organis solvents are often needs to solubilise hydrophobic substrates and allow for high substrate concentrations.
The study is very thorough with many variations of parameters considered and the conclusions are well justified.
My concern is with respect to the enzyme class selected -only EREDs were investigated, but the conclusions were drawn much more broadly.I would like to see data for other systemswith different protein folds and different co-factors.I assume all EREDs have the same fold?
Is it possible that the data are due to NADPH binding pockets being influenced by solvent?
The authors should comment on that.
It would be interesting to compare closely related variants with differern solvent tolerance and /or temperature stability.

Point by point answer
We thank all reviewers for the valuable comments to help us to improve our manuscript.

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): The authors sought to find a criterion that indicates a transi�on, most easily thought to be one from na�ve, folded protein to inac�ve, unfolded one, in aqueous-miscible organic solvent mixtures.They argue that the mel�ng temperature Tm, commonly used for protein transi�ons in aqueous medium, is not a good indicator for transi�ons in aqueous-organic mixtures.While this ra�onale is correct and the objec�ve is very worthwhile, the authors did not deliver on that promise.
1. C,T,U50 is not a simple criterion to be u�lized widely in the community.While the authors correctly place the transi�on in aqueous-organic mixtures at 50% unfolded (and 50% folded) protein (analogous to Tm), the inclusion of temperature as a parameter is unnecessarily cumbersome.While Tm also depends on the composi�on of the medium, such as with kosmotropes and chaotropes, it is measured and reported simply as Tm, and o�en measured at comparably standard condi�ons, such as in PBS buffer.Thus, C,U50 would completely suffice, provided it were to be measured at a standard temperature, say 25 or 30oC.Measurement at a standard temperature would actually result in larger amounts of organic solvent to be added to reach the spot, where [N] = [U].Thus, the inclusion of temperature as a parameter in fact obscures the impact of different organic solvents.
Response: We thank the reviewer for the feedback.Maybe it was not clear from the manuscript how the   50  was determined.The descrip�on was now revised making clear that temperature is intrinsically involved.Nevertheless, the data also show, that the temperature has a significant impact at which solvent concentra�on each enzyme is unfolding.Therefore, the temperature is of importance.The text in the manuscript was adapted as follows to clarify the mode of measurement: To obtain   50  values for enzyme/co-solvent combina�ons, samples containing the same enzyme at varied concentra�ons of solvent were followed via nano differen�al scanning fluorimetry (nanoDSF) experiments, thus measuring the unfolding at increasing temperature.NanoDSF is used to track the fluorescent signal arising from the internal tryptophan/tyrosine residues and has previously been used to iden�fy mel�ng points of enzyme libraries. 58From mul�ple nanoDSF experiments the   50  can be extracted by plo�ng the unfolding at a specific temperature versus solvent concentra�on (Fig. 4a and Fig. 5a).Consequently,… Indeed, the analysis of this value at a (to be) defined standard temperature yields already informa�on on the solvent tolerance, however, as other parameters (like temperature, but also any other salt/ingredients) influence the condi�ons, the measurement condi�ons will always depend on a specific case/condi�ons.Comparison can of course only be done using the same condi�ons.We agree that in case no nanoDSF machine is available and more tedious methods for the collec�on of the unfolding data like DSF has to be performed, just analyzing specific condi�on can be advisable.Anyway, it may be highly beneficial to have a broader set of stability data also at varied temperature to find op�mal opera�onal windows as exemplified in the paper.Thus, looking at more temperature and solvent concentra�ons together we provide a strategy suitable for the op�miza�on of reac�on condi�ons close to prac�cal requirements.
Finally, to show the applicability also for other solvents than the ones just mentioned (MeOH/EtOH),   50  values were also measured for selected EREDs in DMSO, DMF and npropanol (Table 3, Supplementary Table 4, Supplementary Fig. 12).THF was also tested, but it was noticed, that it was not tolerated by the enzymes investigated.From the data obtained (e.g.entries 1-4) it can nicely be seen that the   50  values change with the temperature, thus, the temperature has a clear impact.In general, the higher the temperature, the lower the   50  value.Additional experiments also indicated the applicability to other enzymes like transaminases (entries 17-24).The two transaminases investigated, one (S)-selective one originating from Arthrobacter citreus (ArS) and the other (R)-selective one from an Arthrobacter sp.(ArR) possess different structural folds.Also for these enzymes   50  values were successfully determined.a The error of the   50  was estimated to be about 2%.
In the Experimental sec�on: Transaminases ArS (internal plasmid number pEG29) and ArR (pEG234) were prepared as previously described. 66,67As an example for ArR, cells were disrupted using HEPES buffer (100 mM, pH 8 ) containing imidazol (20 mM) and PLP (1 mM).The same buffer was used for equilibra�ng the HisTag column, and also for the washing step a�er the crude extract was applied to the column.The ArR was eluted using the following buffer: 100 mM HEPS pH 8 containing 200 mM imidazol.A�er the protein containing frac�ons were pooled the puffer was changed (desal�ng PD10 column) to 10 mM HEPES pH 8. Aliquots of enzyme were stored at −20 °C un�l measurements.

Pressnitz, D. et al. Asymmetric amina�on of tetralone and chromanone deriva�ves employing
An alternative parameter for the selection of an organic solvent for a biotransformation is the denaturation capacity of a solvent as guidance across different enzymes 33 .
Due to these differences,   50  is worth to be considered in the ranking of enzymes.Compared to the denatura�on capacity parameter 33 , our approach also allows for a ranking of different enzymes in dependence of the solvent and not a ranking of co-solvents only.Especially when looking for the best enzyme of a library or analyzing variants in a protein engineering campaign,…
Reviewer #1 5. Most importantly: the authors cherrypicked their data on four EREDs (ChrOYE1, OYE1, PpXenB, and TsOYE) of the 13 in the main text to make claims that are less firm or are strongly diminished when considering even all 13 ERED enzymes (not to say anything of ore EREDs or other enzymes) with data in the SI.As an example, SI Figure 6 shows F350/F330 data on the y-axis ploted against solvent content, here ethanol.Only 7 of the 13 EREDs show any usable transi�on!The applicability of the C,T,U50 approach to other systems, even other EREDs, seems tenuous at best and will not be approached with confidence.
Response: Not surprisingly, we strongly disagree with this statement.We displayed only four representa�ve examples in the main paper to improve the readability and not for cherry-picking.All data is in the SI to which we refer to several �mes and we also discuss data of enzymes in detail that are only shown in the SI e.g.Tm of NerA in lines 103 + 118 (original version).In line 207-211 (original version) we also state that some enzymes gave unfolding data that was not possible to analyse in the described way.Line 252 states where to find more data and unfolding and ac�vity fits and which SI Figures need to be compared to see the rela�on of the unfolding and the loss of the ac�vity around the same co-solvent concentra�on.On top of that, for all values given, they were in all cases calculated from all recorded data, that could be included.In the main text it is always explained in detail, which data series were combined and which criteria were used to make this decisions.We actually men�oned limita�ons and reasons why seven could be interpreted (For some enzymes this method did not give an analysable unfolding curve (e.g.LacER, see Supplementary Fig. 6 and Supplementary Fig. 7), which was atributed for example to tryptophan residues with litle change in their local environment during the unfolding process.).This is a limit of the analy�cal method, not of the C,T,U50.

Reviewer #1
For all these, and more, points, this manuscript should not be accepted.Response: We thank the reviewer for the comments especially to include also the DC, which we have now done; we clarified that the measurement for   50  goes in hand with measuring the mel�ng temperature in a very simple set up using nanoDSF, and also clarified that we did not perform cherry picking but showed representa�ve data in the paper make it easier to read and underlined that all data is in the SI.

Dear Authors
The manuscript Sorgenfrei et al. en�tled "Solvent concentra�on at point of 50% protein unfolding   may reform enzyme solvent stability ranking and iden�fica�on of process window" describes a novel approach to iden�fy condi�ons for applying enzymes in the presence of water miscible cosolvents.O�en the mel�ng temperature is or was used to get a hint towards the process stability of enzymes also in the presence of solvents.However, as the authors state this is not always leading to a success, and one needs another more accurate rou�ne to predict this process window.This important issue was now tackled by the authors and as role model they employed ene reductases.For this class of enzymes, a huge set of data are readily available from literature including thermo-as well as solvent stability, making it an ideal candidate for such an approach to establish a new rou�ne to predict process applicability of enzymes in the presence of miscible co-solvents.Furthermore, the authors laboratories have a long-las�ng experience with this enzyme family.All this exper�se was used to generate a holis�c new set of data to establish a new key parameter to evaluate enzyme stability towards solvents connected to enzyme performance.
This new concept will be of use for the community, especially in biocatalysis, but it may be of interest to a broader audience as well.Protein stability and folding is a very diverse field, and such suppor�ve methods will help to uncover novel features among those molecules and their func�on.Thus, I can state this is a novel and original contribu�on to the field of high impact.
A hypothesis was formulated, and proper methods have been employed to answer the raised ques�ons.The methods are well documented and sound reproducible.Needed standards to describe enzymes have been applied and especially for flavin dependent ene reductases.Those can be transferred to other enzymes as well.
In general, the manuscript is well prepared, scien�fic sound and all needed informa�on are provided to follow the study.A new concept is presented and supported by a large set of data and well discussed.Proper conclusions were drawn from a sophis�cated data analysis.The reference list is appropriate and most relevant ar�cles are cited; no excess self-cita�on determined.General forma�ng is of good quality and presenta�on of tables and figures very suppor�ve.Nevertheless, some minor points need to be improved -nothing cri�cal from my point of view.
Response: We thank the reviewer for acknowledging the impact of the study and also for the suppor�ve statements.
Reviewer #2: In the following I will provide my comments and hope they are suppor�ve and construc�ve: -This part of the sentence from line 24 is not sound: "showing depending also on the solvent itself" and might be rephrased.

Response:
The sentence was rephrased: Comparing possible rankings of enzymes according to mel�ng temperature and   50  revealed a clearly diverging outcome also depending on the specific solvent used.
Reviewer #2: -Line 78/79 while I agree that FMN is a true cofactor; NAD(P)H is rather a co-substrate for EREDs as it is consumed during catalysis and cannot be recycled by the enzyme in the same reac�on course.
Response: We have rephrased the sentence and refer to NAD(P)H as cosubstrate now: 'For this study, ene-reductases (EREDs) using the cofactors FMN and NAD(P)H as cosubstrate were selected as model catalysts possessing high relevance for biocatalysis [34][35][36][37][38][39][40] .' Reviewer #2: -The general observa�on towards Tm that DMSO is less effec�ve while the alcohols are more effec�ve as the chain length increases is quite interes�ng; was to some extent observed for oxidases as well (of course not that generalized as shown herein); it might have to do with the change in pKa and acidity of the HO-group.Can you comment on this?
Response: We like very much the idea to link the trend of Tm to the pKa, although when we compared the order of effect with the pKa values, there is not a perfect fit: Methanol (pKa 15.2) < ethanol (pKa 16) < 2-propanol (pKa 17.1) < n-propanol (pKa 16.85).Probably the pKa is one parameter to consider maybe together also with volume… There is more research needed.As we do not spot a clear rela�on, we did not put any specula�ons to the manuscript.

Reviewer #2
Reviewer #3 (Remarks to the Author): This paper inves�gates parameters that influence enzyme ac�vity in organic water miscible solvents.
There is significant amount of literature sugges�ng that thermal stability can be taken as a surrogate for tolerance of organic solvent, but the authors show that the correla�on is not always strong.They suggest that as an alterna�ve the stability of the enzyme with respect to increasing amount of organis solvent should be used.They also show that solvent stability differes between solvents.Overall, these results are of interest in biotechnology, where organis solvents are o�en needs to solubilise hydrophobic substrates and allow for high substrate concentra�ons.
The study is very thorough with many varia�ons of parameters considered and the conclusions are well jus�fied.
My concern is with respect to the enzyme class selected -only EREDs were inves�gated, but the conclusions were drawn much more broadly.I would like to see data for other systems -with different protein folds and different co-factors.I assume all EREDs have the same fold?Is it possible that the data are due to NADPH binding pockets being influenced by solvent?The authors should comment on that.
Response: We thank the Reviewer for this comment and have done in the mean�me a significant amount of new measurements (now Table 3 and Supplementary Table 4) and added this data to the manuscript.As described for reviewer 1 at point 3, we included on the one had more/other solvents and also extended the examples to other type of enzymes, namely two transaminases, each with a different fold.As expected (at least for us), the concept was also applicable for the other type of enzymes.The stability depends on the structural elements and also whether the cofactor is bound or not bound but we did not inves�gate this point here.We have now included the text for the addi�onal enzymes (two transaminases), each with a different fold.
Finally, to show the applicability also for other solvents than the ones just mentioned (MeOH/EtOH),   50  values were also measured for selected EREDs in DMSO, DMF and npropanol (Table 3, Supplementary Table 4, Supplementary Fig. 12).THF was also tested, but it was noticed, that it was not tolerated by the enzymes investigated.From the data (entries 1-4) obtained it can nicely be seen that the   50  values change with the temperature, thus, the temperature has a clear impact.In general, the higher the temperature, the lower the   50  value.Additional experiments also indicated the applicability to other enzymes like transaminases (entries 17-24).The two transaminases investigated, one (S)-selective one originating from Arthrobacter citreus (ArS) and the other (R)-selective one from an Arthrobacter sp.(ArR) possess different structural folds.Also for these enzymes   50  values were successfully determined.
As the principles of this strategy are applicable for all enzymes having tryptophane/tyrosine in their sequence, this approach is not limited to EREDs and can be used for other types of enzymes, as we have shown here also for transaminases, and could have a major impact on the descrip�on of enzyme stability, selec�on of suitable enzymes from (commercial) libraries and in enzyme engineering campaigns, as well as on iden�fying opera�onal windows for reac�ons.

Reviewer #3
It would be interes�ng to compare closely related variants with differern solvent tolerance and /or temperature stability.
Response: Yes, we agree that it would be very interes�ng to analyse the solvent tolerance and temperature stability with respect to rela�on of the variants and we expect that this is a future applica�on for the   50  .Consequently, we think that for a detailed analysis another study with variants is needed.The candidates of this study were picked widely to cover most of the sequence space of the known EREDs and was expanded by two transaminases to show the effect on another class of enzymes as well.The inten�on was to implement a method and indeed we agree that there are many more op�ons for applica�ons.

Addi�onal changes not related to comments of reviewers
Due to maternity (Frieda Sorgenfrei) and paternity (Jeremy J. Sloan) leaves, we had a delay in answering and also "new" people did the nanoDSF measurements at BASF and were responsible for that.Consequently, the list of authors was extended by Niklas A. Mehner 2 , Thomas L. Ellinghaus 2 .
Furthermore, we spoted several typos, which were now corrected and are marked in the main paper with a yellow background.
Finally we added:

Table 3
Examples for     values