Heterologous Expression of Thermotolerant α-Glucosidase in Bacillus subtilis 168 and Improving Its Thermal Stability by Constructing Cyclized Proteins
and Zhiming Rao *Round 1
Reviewer 1 Report (Previous Reviewer 3)
The authors have made important improvements in their work, but there are still problems with two issues:
a) I agree that, as shown in Figure 4, now the protein is purified (not completely, but nicely). Unfortunately, having one band does not tell me that the isopeptide bond is formed. It just tells me that there is a protein of that size. I am still lacking an experiment that shows unequivocally that the isopeptide bond is formed, and with what efficiency. The new figures of the structures including the tags and the catchers show that the link is possible, without distorting the enzyme (I thank the authors for the figures and the clarification), but does not prove, experimentally, that the link is there. The main issue is that you could have an increase in thermal resistance through non-covalent interactions between the tag and catcher modules with the enzyme.
- regarding the simulations, the methods section is incomplete and has mistakes. The protein is described with the oplsaa forcefield, and the water molecules (not the protein) with the SPC model. There is no information on how long-range interactions were handled. On line 196 it says 0.15% NaCl, where I guess it should say 0.15M NaCl. The width of the water layer is not stated. Most importantly, I still do not find a clear definition of the proteins that are simulated. A supplementary table with the amino acid sequences of the proteins, marking clearly the tag, catcher, enzyme, and location of the isopeptide bond is needed. I still do not know what was submitted to Alpha Fold 2 for modeling, though I asked for this before. Also, I am still missing information on how the isopeptide bond was modeled in the forcefield. This is not a standard bond in regular forcefields for proteins, and requires parametrization. I raised this point from the first review.
- as a suggestion, in the figure caption for Figure 1, relate the colors to the ways of synthesis of IMOs in the text.
In summary, the section on the simulations just requires a thorough definition of the modeled and simulated systems, and the conditions for the runs. The experimental section still requires an assay that shows that the isopeptide bond is formed.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report (New Reviewer)
The manuscript ‘Heterologous Expression of Thermotolerant α-Glucosidase in Bacillus Subtilis 168 and improving Its Thermal Stability by Constructing Cyclized Proteins’ is an interesting and well-written manuscript. It describes the expression of an α-Glucosidase variant with improved thermal stability by cyclization in recombinant host cells. The SpyTag/SpyCatcher strategy has been used to mediate spontaneous isopeptide bond formation leading to cyclization.
The discussion part of the manuscript could have been a bit more elaborate
Why the wild-type enzyme activity is high at pH 8 and 9 when compared to improved variants? Substantiate
Why the relative activity of the variants is lower than wild-type enzymes at 40 to 60 ° C in fig. 5a. Whether the activity is compromised when the stability was improved? Substantiate the reasons
Discuss the future directions of the research at the end of the manuscript.
Minor comments:
Mention t1/2 half-life in line 74 when it first time appears
Describe T50 in line 77 when it appears the first time
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report (Previous Reviewer 3)
My concerns were addressed reasonably.
Minor issues follow:
- in the abstract: "After forming the cyclized 19 α-glucosidase by different isopeptide bonds (SpyTag/SpyCatcher, SnoopTag/SnoopCatcher, 20 SdyTag/SdyCatchr, RIAD/RIDD)." That is not a sentence, because it lacks a conjugated verb.
- in the abstract: "The optimal temperature if all cyclized AGL were" ... not were, was.
- in the supplementary material, Table S3, correct "Cather" to "Catcher".
In general, do a thorough grammar check, making sure that all sentences have conjugated verbs, and that there is subject-verb agreement.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Authors present results on the heterologous expression of two thermotolerant alpha glucosidases and their improvement via cyclization. In broad terms, the work is novel and interesting. However, there are some important issues, and several minor issues, that need to be addressed before the work can be accepted for publication.
Major comments
1) In the introduction section, there are several paragraphs where the information is not clear or technically imprecise. For example:
Lines 30-31: the glycosidic bonds are not linked to glucose groups, in fact they are the links between residues.
Lines 48-53. A transglycosylation reaction is being described, however, is termed hydrolysis.
Lines 56-57: This sentence ("the active center of ...different sugar substrates") is imprecise and does not really convey a clear meaning. Please rephrase.
2) Authors should address clearly and justify with literature sources the need for a thermotolerant alpha-glucosidase (e.g, lines 66-68 are not justified). In fact, I could not find the connection of ref. 14 (line 65) with the benefits of operating the process at high temperatures.
3) More information is required concerning the isopeptide bond leading to cyclized proteins, for example, to explain the rationale of SpyTag/SpyCatcher selection (also for the other C and N terminal additions).
Methods.
Methods are rather incomplete, which dificult the comprehension of the work. Moreover, the number of replicas per experiment (biological or analitycal) is not stated and this information is critical to assess the results.
4) No information regarding the purification of the crude extracts is presented. At the beginning of this section, a Ni-NTA column (not NI-NTA, since Ni stands for nickel) is mentioned, but there is no explanation about where and how was used.
5) Line 117. State how plasmids were verified.
6) More information regarding the cell disruption method is required (time, sonication power, temperature, use of chemicals, etc)
7) How was the alpha-glucosidase obtained in the determination of activity (2.2.3) obtained? Please clearly state was the enzyme was obtained and purified. This has important consequences for the calculation of enzyme activity. Therefore, after clarifiyng weather the activity was measured using a crude or purified extract, state cleary what is the "amount of the enzyme" in the activity definition. To ilustrate my concerns in this point, consider that Table 1 reports specific enzyme activities in U/mg, but is not clear if these values are per mg of total protein or mg of purified alpha-glucosidase.
8) Line 144: The mobile phase is made of 75% Acn. What is the remaining 25%?
9) Section 2.2.6: Further details are required, includinng for how long was the activity assayed to determine the optimal temperature, the thermal stability was assayed with ot without substrate (as it is well known that substrate binding can increase thermas stability), what was the pH for the activity determination after the pH stability assays.
10) Section 2.2.1: please state what is the conection between the cloned genes and agl or gsj. These acronyms appear at the beggining of the Results section, and the reader is clueless about its meaning.
Results
11) Figure 1. From Figure 1, it is clear that most of the enzyme remains with the broken cells precipitate, thus pointing to a rather ineficient cell disruption. Please discuse this point, both from a methodological point of view and a technological one (the implications of this for process scale-up). Same is tru for Figure 2. Please include in Figure 2 the SDS-PAGE results for the broken precipitate of the cells harboring the cyclized proteins plasmids.
12) Table 1: TAble 1 requires further explanations regarding the basis for the activity (U per mg of what, total protein?)
13) Lines 244-246. This analysis is not correct. The fact that cyclized AGLs showed higher activities at temperatures above 60C (compared to the wt-AGL) do not imply that they are more thermotolerant since I guess that the assay was performed in a short time to determine the optimal temperature.
14) Figure 4. How error bars were calculated? Improve image resolution.
15) Figure 5: I cannot understand the "jump" or discontinuity in the relative activity between pH 8 and 9. No information is presented in the text either.
Minor comments.
The quality and clarity of writting needs to be improved significantly. In this round of revision, I will not provide details due to the large amount of typos. Only a few examples:
Define: T_50, Tres, AGL, GSJ...
Please check correct use of Figure referencing in text (Figure 4.b in line 272 should be Figure 5b)
IMOs are plural, not singular (see for example line 32)
Author Response
Please see the attachment.
Author Response File: 
Author Response.doc
Reviewer 2 Report
Dear authors, the manuscript "Heterologous Expression of Thermotolerant α-Glucosidase in Bacillus Subtilis 168 and Improving Its Thermal Stability by Constructing Cyclized Proteins" is quite interesting and worth investigation. Please see some comments below:
1 - It is well-designed and discussed.
2 - Have you considered design of experiments in order to reach the optimum catalytic activity?
3 – Please add information about purification.
4 – Have you considerer complex model (Kinetic parameters)?
5 - Molecular dynamics simulation of cyclized α-glucosidase is quite great. Congratulations
Regards
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The authors present strategies to thermally stabilize thermotolerant alpha-glucosidases by constructing cyclized version using isopeptide bonds. This is interesting as having temperature-resistant versions of this enzyme would improve the synthesis of isomaltooligosaccharides, of relevance in food industry and feed for animals. They use two enzymes from two thermophiles, and end up optimizing only one of them (AGL), because the other one (GSJ) carried out a side reaction of hydrolysis in excess. This work is timely, but has the following issues in presentation and also in its execution:
- the acronyms for both alpha-glucosidases should be listed in materials and methods. Otherwise, the reader is left to look over the whole paper searching for the meaning of AGL and GSJ; these acronyms make no obvious sense.
- panose is written in two different ways (one and two "n").
- the explanation of the IMO synthesis process involving alpha-glucosidase is convoluted (which is the donor and which is the acceptor, for instance?). It requires rewriting; actually, the described function is not hydrolysis, but transglycosylation. Maybe a supplementary figure showing the three main synthesis routes would be useful.
- to justify using the N- and C-termini of AGL as the sites for cyclization, a figure (in the main text or supplementary) of the 3D structure of this enzyme should be included, showing the distance between the termini in the wild type version, and the convenience of each of the linkers in covering that distance without distorting the enzyme.
- it is not clear from the introduction why pH should also be a parameter of interest, or why would the optimum change upon cyclization of the enzyme.
- in the molecular dynamics section, how was the "cycle" indicated in AlphaFold2? This is a side chain-side chain link, with an extra 12 amino acids between them. More details are needed. While the engine for the simulations was Gromacs, the details of the force field and solvent used, and how the isopeptide was modeled, are lacking. A single run, of 40 ns, is insufficient sampling, as can be seen from current literature using MD simulations to determine stability. Also, from the results in figure 7, it is clear that the runs have not equilibrated properly.
- how are the authors sure that cyclization took place? This is not specified in the methods or the results section, other than a gel of the proteins.
- figure 4 shows error bars, but in the methods section there is no mention of replicas. Please indicate how many times the measurement was carried out and how. This applies to all activity measurements.
- if the optimal pH is 6 (figure 5), why was activity measured at pH 7 (figure 4)?
- the claim that cyclized AGL is more pH stable than the wild-type is not supported from figure 5b; the error bars cancel the differences. To prove the opposite, statistical analysis is needed.
- in the methods section (2.2.6) the authors state that kinetic parameters were measured at pH 7. In the results section, it says pH 6. Please correct accordingly.
- I am curious as to why the authors measure the kinetics of the hydrolysis reaction, instead of the one of interest (transglycosylase). It is conceivable that restricting the motion of the enzyme could alter binding to larger substrates in the transglycosydation reaction (compared to hydrolysis).
- if the linker for cyclization is far from the active site, and does not impede the functional motion of the enzyme, there is no reason to anticipate changes in catalytic efficiency. Once the error bars are taken into account, Kms and Kcats are identical, and the differences in Vmax are small. Figure 6 could include the location of the active site, and help in the discussion.
- the structures shown in close-up in figure 6 indicate irregular bond distances in all ASPs. This figure also drives home the need to explain more clearly how the cyclized versions were built, as the places for the linkages are not the same for all constructions: K11-D818, N28-K840, and D8-K819. They are indeed near the N- and C-termini, but each one is different.
- regarding figure 7a, RMSDs of 0.45nm and higher are indications of a problem in the simulation, in general. This is particularly clear for the SpyTag-AGL-SpyCatcher. This is a large enzyme, with subdomains that could be moving relatively to each other. The problem with RMSD is that it describes all these motions with just one number, and offers very little information (other than potential disasters).
-regarding figure 7b, rather than saying "most regions", map the relevant regions to the protein structure and discuss the relevance of tightening these regions for the stability of the protein.
- regarding figure 7c, the fluctuations in Rg should be analyzed by taking snapshots from the simulation with extreme values of Rg, to see what these values correspond to. Again, this is a large enzyme, and the sampled motions could be relevant to understand how the enzyme works.
- in the conclusions, it is very strange that the experimentally most promising version (Spy) is the least stable in the simulations. There is no correspondence between what is seen in the experiments and in the simulations.
- in the supplementary material, a diagram showing the construction of each version of cyclized AGL would be very useful (not just the primers).
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Most of the issues raised in the previous round of review were addressed. However, upon further reading, are several unresolved issues, raised mainly by Reviewer 3.
From my standpoint, the fact that evidence for protein cyclization beyond the modest increase in molecular weight (almost indistinguishable in the gels) is missing is a serious lack of evidence. And as raised by reviewer 3, the protein dynamic simulations do not fully interpret the results.
Moreover, the paper still presents faults in the report of the methodology used. For example, there is no hint on how the his-tag was included in the protein for Ni column purification.
Reviewer 3 Report
The authors addressed some of the comments, but not all of them. Therefore, serious problems remain.
Figure 1 is a nice addition, and helps a lot; I suggest using an arrow for fructose in the green pathway, to show that it leaves the reaction (as shown for the other pathways).
Regarding how the linkers were built, there is a steric problem. 3XGGGS allows for a ~12Å span assuming an overstretched chain, and as shown in the new supplementary figures, the link should span ~60Å. Furthermore, in the models generated by AlphaFold, it is clear that there is a severe conformational change imposed on the "blue" and "red" domains in order to achieve the linkage. It is not true that the enzyme is not altered in these models. If the authors superimpose the original "linear" enzyme with the cyclized versions, using the catalytic domain as reference, the differences should be obvious.
This ties in to the unresolved issue of how cyclization was determined. I checked the papers suggested in the author's response, and in those works, a thorough purification of the cyclized enzyme is carried out (by ITC, for instance), allowing for the quantification of the efficiency of the cyclization reaction. Here, what we see are crude gels, with no purification. Given the tiny change in molecular weight between the wild-type and the variants, even with the crude gels it is hard to know that the variant is the one expressed. Regardless of which version it is, the band is at the same height. Again, these SDS gels are not the way to validate that cyclization has taken place. In addition, an SDS gel with the purified enzymes should be shown.
There is a new figure of RMSD vs time. Were new simulations conducted? If so, why is the RGYR the same? The simulations are still too short, and the analysis is insufficient. What are the magenta boxes in the RMSF plot? Given that the starting structures are suspect, the simulations are not useful to explain the experiments in the manuscript.
In the conclusions, it still says that the Spy version is the best, and that the simulations support that. Figure 5b indicates that Snoop is the best.
