Improvement of L-Tryptophan Production in Escherichia coli Using Biosensor-Based, High-Throughput Screening and Metabolic Engineering

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript describes the engineering of an E. coli strain to maximize production of L-tryptophan. The authors describe the construction of a vector system for expression of the L-tryptophan biosensor, two rounds of ARTP mutagenesis of the L-tryptophan production strain followed by cell sorting and plate-based screening to identify the most productive mutant strains, sequencing analysis to identify key mutated genes and validation of the importance of these genes in L-tryptophan production, and further metabolic engineering (knockout/promotor engineering/gene copy number amplification). This methodology was shown to be effective in significantly enhancing L-tryptophan production in E. coli. The introduction section provides valuable context for the study by describing the economic value of L-tryptophan and previous work that has been done in engineering L-tryptophan production strains, however the introduction fails to effectively justify the need for this study in light of previous advances that have been made in this area (see comment below). It seems to this reviewer that the main advance of this study is the demonstration of the effectiveness of the employed engineering strategy, but this is not clear from the introduction as written. The materials and methods section is sufficiently detailed to allow for this work to be repeated in other laboratories. Although the optimized strain produced here is not competitive with existing strains for L-tryptophan production, this study serves as a strong proof of concept for the value and effectiveness of the metabolic engineering strategy employed, and also served to identify a new gene target for metabolic engineering in the context of L-tryptophan production. This manuscript would be of interest to the bioengineering community (as well as others), and should be published once the minor issues described have been addressed.

Introduction, Lines 54-56: The statement “…the process remains time consuming due to its reliance on manual assays…” is not clearly worded. What process is being referred to here? This statement also follows a description of other work that has resulted in other strains that can produce very high titers of L-tryptophan, and it is unclear how the quoted statement justifies this study in this light. This needs to be clarified to that the statement is clear and so that the study described here can be justified in light of previous work that has been done in the area.

Figure 2: The Y-axis is labeled as “Mortality rate”, while in the text the terminology used is “Lethality rate”.

Author Response

Response to Reviewer 1 Comments

1. Summary
Thank you for your comments concerning our manuscript entitled “Improvement of L-Tryptophan Production in Escherichia coli using Biosensor-based High-throughput Screening and Metabolic Engineering” (ID: fermentation-3591266). Those comments are all valuable and very helpful for revising and improving our paper, as well as having important guiding significance for our researche. We have carefully revised the manuscript and have provided a point-by-point response below. We very much hope that you will find the revised content acceptable for publication.

2. Point-by-point response to Comments and Suggestions for Authors
Comments and Suggestions for Authors: This manuscript describes the engineering of an E. coli strain to maximize production of L-tryptophan. The authors describe the construction of a vector system for expression of the L-tryptophan biosensor, two rounds of ARTP mutagenesis of the L-tryptophan production strain followed by cell sorting and plate-based screening to identify the most productive mutant strains, sequencing analysis to identify key mutated genes and validation of the importance of these genes in L-tryptophan production, and further metabolic engineering (knockout/promotor engineering/gene copy number amplification). This methodology was shown to be effective in significantly enhancing L-tryptophan production in E. coli. The introduction section provides valuable context for the study by describing the economic value of L-tryptophan and previous work that has been done in engineering L-tryptophan production strains, however the introduction fails to effectively justify the need for this study in light of previous advances that have been made in this area (see comment below). It seems to this reviewer that the main advance of this study is the demonstration of the effectiveness of the employed engineering strategy, but this is not clear from the introduction as written. The materials and methods section is sufficiently detailed to allow for this work to be repeated in other laboratories. Although the optimized strain produced here is not competitive with existing strains for L-tryptophan production, this study serves as a strong proof of concept for the value and effectiveness of the metabolic engineering strategy employed, and also served to identify a new gene target for metabolic engineering in the context of L-tryptophan production. This manuscript would be of interest to the bioengineering community (as well as others), and should be published once the minor issues described have been addressed.
Response: Thanks very much for your positive evaluation and devotedly suggestions. We have revised the manuscript according to your devotedly suggestions as below.

Comments 1: [Introduction, Lines 54-56: The statement “…the process remains time consuming due to its reliance on manual assays…” is not clearly worded. What process is being referred to here? This statement also follows a description of other work that has resulted in other strains that can produce very high titers of L-tryptophan, and it is unclear how the quoted statement justifies this study in this light. This needs to be clarified to that the statement is clear and so that the study described here can be justified in light of previous work that has been done in the area.]
Response 1: We thank you for raising this important point, which we have revised in lines 54-63 accordingly. In the above article we cited two examples of increasing L-Trp production in E. coli by metabolic engineering. Although higher titers of L-Trp-producing strains have been reported, it has been reported by other researchers that relying solely on metabolic engineering to enhance L-Trp production in E. coli is very time-consuming and labor-intensive. At this point, biosensor-based high-throughput screening strategies have garnered increasing attention due to their rapidity and efficiency. In the following, we present in detail the current status of research on biosensor-based high-throughput screening of L-Trp-producing strains. Thank you for your helpful suggestion.

Comments 2: [Figure 2: The Y-axis is labeled as “Mortality rate”, while in the text the terminology used is “Lethality rate”.]
Response 2: We feel sorry for our carelessness in our manuscript, the error is revised. Thanks for your correction (Figure 2).

Again, we thank the reviewer for the positive and constructive comments regarding our paper.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this manuscript, a tryptophan-overproducing E. coli strain underwent random mutagenesis using ARTP. The mutants were tested for tryptophan production using a biosensor based on a riboswitch and YFP that reports intracellular tryptophan levels, with the highest tryptophan-producing cells isolated based on their fluorescence using FACS. The authors performed two rounds of mutagenesis to increase tryptophan production and sequenced the mutant strain, finding that the ptsN gene plays a vital role in tryptophan production. Subsequently, they subjected the mutant strain to rational design modifications to achieve even higher production levels.
Overall, the text is well-structured. The authors' intentions remain clear throughout, and the publication achieves its objective. The authors made sound decisions that allowed them to improve tryptophan production; however, the main drawback of the manuscript is that some selection criteria are not sufficiently clear and require further explanation. The authors can improve their manuscript by considering the following points:
After completing the mutagenesis rounds, the authors modified the tnaB and yddG genes, which are part of the tryptophan transport system. Previous studies have indicated that changing these genes can lead to more tryptophan being stored outside the cell and less blockage of its production inside the cell. Because of this, the biosensors could lead to choosing strains that keep high levels of tryptophan inside the cells, which might not be the best approach. It would be fascinating to know if the authors used biosensors in strains Zh02, Zh03, or any strain obtained after the mutagenesis rounds to understand how the sensor behaves in these tryptophan-secreting strains. Additionally, they should add potential solutions or strategies to address this issue.
The authors identified ptsN as a gene affecting tryptophan production; they reached this conclusion by analysing DNA sequence changes in the mutants. However, they listed other mutations that could also favour tryptophan production, such as changes in the tyrR (10.3390/ijms241411866), ptsI (10.1016/j.mec.2021.e00167), and pstH (10.3390/ijms241411866, 10.1007/s10295-018-2020-x) genes. The way the authors set up their experiments shows that increasing the amount of ptsN helps make more tryptophan; however, they chose to insert ptsN into the ackA gene in strain Zh01. Turning off ackA changes how acetate is made, which leads to several changes in how the cell works, helping to produce more foreign proteins, increasing the number of cells in the culture, and providing more building blocks for tryptophan production. Are there additional experiments supporting the role of ptsN in tryptophan production? Did the authors clone ptsN in a neutral genome region, overexpress it through plasmids, or have a strain with the same genetic background as Zh01 but without ptsN to compare the effect caused by ackA deletion?
What criteria did the authors use to select the genes to modify in strains Zh04—Zh08? The authors indicated that it was a literature search and an in-depth analysis of metabolic pathways (lines 360-364); however, how did they choose the order of genes to mutate? Additionally, they indicated that there was a combination of promoter usage and copy numbers. Yet, for each modification, only one strain is shown, which is not explained if other strains had poor results.
What criteria did the authors use to perform only two rounds of mutagenesis?
In line 134, the word "resuscitated" is used; it could be replaced with "reactivated", "recovered", or "reanimated" to be more appropriate.
The authors should explain why they used the DH5α strain to test the pACYC184-727 plasmid, as described in line 216. I presume it was because they needed a control, and the other strains were overproducers.
Specify in Figure 1A what the columns represent; they appear to be visible light and fluorescence.
In Graph 1B, the error bars overlap. The authors could change the scale of one of the axes or use different colours to differentiate it. Although the size of the bars allows for distinguishing some overlaps, this does not work for strain trp02.
The authors seem to use an arrow to symbolise each reaction in Figure 5A. However, why do they only use one arrow for the serA, serB, and serC genes?
To avoid ambiguities, the authors should briefly mention the origin of the P4 and P3 promoters used in this work.
In the introduction, the authors cite the p15-ribo727 riboswitch, while in the methodology, they only call it fragment 727. It is advisable to maintain uniformity or, if there is a difference, to highlight it.
In Figure 4, "* for P < 0.05" is not used.

Author Response

Response to Reviewer 2 Comments

1. Summary
Thank you for your comments concerning our manuscript entitled “Improvement of L-Tryptophan Production in Escherichia coli using Biosensor-based High-throughput Screening and Metabolic Engineering” (ID: fermentation-3591266). Those comments are all valuable and very helpful for revising and improving our paper, as well as having important guiding significance for our researche. We have carefully revised the manuscript and have provided a point-by-point response below. We very much hope that you will find the revised content acceptable for publication.

2. Point-by-point response to Comments and Suggestions for Authors
Comments and Suggestions for Authors: In this manuscript, a tryptophan-overproducing E. coli strain underwent random mutagenesis using ARTP. The mutants were tested for tryptophan production using a biosensor based on a riboswitch and YFP that reports intracellular tryptophan levels, with the highest tryptophan-producing cells isolated based on their fluorescence using FACS. The authors performed two rounds of mutagenesis to increase tryptophan production and sequenced the mutant strain, finding that the ptsN gene plays a vital role in tryptophan production. Subsequently, they subjected the mutant strain to rational design modifications to achieve even higher production levels. Overall, the text is well-structured. The authors' intentions remain clear throughout, and the publication achieves its objective. The authors made sound decisions that allowed them to improve tryptophan production; however, the main drawback of the manuscript is that some selection criteria are not sufficiently clear and require further explanation. The authors can improve their manuscript by considering the following points.
Response: Thanks very much for your positive evaluation and devotedly suggestions. We have revised the manuscript according to your devotedly suggestions as below.

Comments 1: [After completing the mutagenesis rounds, the authors modified the tnaB and yddG genes, which are part of the tryptophan transport system. Previous studies have indicated that changing these genes can lead to more tryptophan being stored outside the cell and less blockage of its production inside the cell. Because of this, the biosensors could lead to choosing strains that keep high levels of tryptophan inside the cells, which might not be the best approach. It would be fascinating to know if the authors used biosensors in strains Zh02, Zh03, or any strain obtained after the mutagenesis rounds to understand how the sensor behaves in these tryptophan-secreting strains. Additionally, they should add potential solutions or strategies to address this issue.]
Response 1: Thank you very much for your interest in our constructed L-Trp biosensor pACYC184-727. We very much agree with your suggestion of “It would be fascinating to know if the authors used biosensors in strains Zh02, Zh03, or any strain obtained after the mutagenesis rounds to understand how the sensor behaves in these tryptophan-secreting strains”, which points to the direction of our subsequent research. 
On the one hand, genetic modifications targeting the tryptophan transporter system were accomplished on the basis of mutant strain GT3938 to enhance the L-Trp efflux capacity. The screening of the mutant strain using the biosensor was completed at this point. The modifications of genes tnaB and yddG did not interfere with the biosensor's ability to screen for high-titer tryptophan-producing strains. At the same time, as shown in Figure 1B, no significant effect on biosensor performance was observed in strains knocking out tnaB and yddG overexpression. 
On the other hand, the L-Trp biosensor has only been adopted as an auxiliary tool in the high-throughput screening stage. For the behavior of biosensors in response to extracellular secretion of L-Trp, Liu et al. have constructed a high-throughput screening platform by combining L-Trp biosensors and droplet microfluidics [1]. They successfully selected a mutant strain K3mu1 that produced 2.19 g/L L-Trp, which was 155.1% higher than KW003, from an ARTP-based genome-wide random mutation library. Meanwhile, the intracellular L-Trp titer of K3mu1 was 0.027 g/L, which was 1.37 times higher than that of KW003 (0.019 g/L), indicating that the concentration of extracellular L-Trp for K3mu1 and KW003 showed a positive relationship to their intracellular L-Trp levels. Thus, it was demonstrated that this L-Trp biosensor was equally responsive to L-Trp secreted into the extracellular compartment. 
We have added the outlook for future research in the manuscript based on your suggestions (Line 273-283). Meanwhile, we will further optimize the sensitivity and detection threshold of the L-Trp biosensor to achieve higher screening efficiency in our subsequent research work. Once again, we would like to express our gratitude for your constructive ideas and look forward to more communication with you.

1.    Liu, Y.; Yuan, H.; Ding, D.; Dong, H.; Wang, Q.; Zhang, D. Establishment of a Biosensor-based High-Throughput Screening Platform for Tryptophan Overproduction. ACS Synthetic Biology 2021, 10, 1373-1383, doi:10.1021/acssynbio.0c00647.

Comments 2: [The authors identified ptsN as a gene affecting tryptophan production; they reached this conclusion by analysing DNA sequence changes in the mutants. However, they listed other mutations that could also favour tryptophan production, such as changes in the tyrR (10.3390/ijms241411866), ptsI (10.1016/j.mec.2021.e00167) and pstH (10.3390/ijms241411866, 10.1007/s10295-018-2020-x) genes. The way the authors set up their experiments shows that increasing the amount of ptsN helps make more tryptophan; however, they chose to insert ptsN into the ackA gene in strain Zh01. Turning off ackA changes how acetate is made, which leads to several changes in how the cell works, helping to produce more foreign proteins, increasing the number of cells in the culture, and providing more building blocks for tryptophan production. Are there additional experiments supporting the role of ptsN in tryptophan production? Did the authors clone ptsN in a neutral genome region, overexpress it through plasmids, or have a strain with the same genetic background as Zh01 but without ptsN to compare the effect caused by ackA deletion?]
Response 2: Thank you for pointing this out. For the identification of the key gene ptsN, we obtained it by comparing the whole genome sequencing results of the mutant strain GT3938 and the starting strain G01. Any identified mutations were manually curated by using Breseq and IGV to eliminate false positives as well as ambiguous regions. The specific mutation types and genes involved are shown in table S3, table S4 and table S5 of Supplementary. Through a literature search, the ptsN gene and the genes you listed as tyrR, ptsI and ptsH were focused on for detection because they all appear in the L-Trp synthesis pathway. Among them, the ptsN gene has never been reported, so we have given it more attention. To further understand the effect of the ptsN gene on L-Trp synthesis, we successfully increased the L-Trp titer of the mutant strain GT3938 by 21.36% through the strategy of increasing the copy number of the ptsN gene. This was able to show the importance of overexpression of the ptsN gene for increasing L-Trp production, and the conclusion has never been reported. Based on your comments, we have added to the manuscript to further clarify why ptsN is being focused on (Line 304-309). In the future, we will follow up with more in-depth mechanistic analysis of the mutant strains obtained from the screening through transcriptomics, which will be more interesting research.
As you pointed out, we apologize for our carelessness in overlooking the effect of knocking out the ackA gene on L-Trp synthesis. To further validate the effect of ptsN gene on the synthesis of L-Trp, we re-overexpressed the ptsN gene in plasmid form as you suggested. As a result, strain GT3938-1 was constructed. Shake flask fermentation for 48 h showed that the L-Trp titer of the strain GT3938-1 was 1.18 g/L, with an OD600 of 15.53. The L-Trp titer of GT3938-1 was lower compared to that of strain zh01 but has still increased by 14.56% compared to that of strain GT3938. The results indicated that overexpression of the ptsN gene remained effective for the elevation of L-Trp titer, further eliminating the possible effects of ackA knockout. We have added relevant results to the manuscript (Line 310-311, Line317-320 and Figure 4). 
Again, we apologize for our carelessness and extend our highest appreciation to you for your devoted comments.

Comments 3: [What criteria did the authors use to select the genes to modify in strains Zh04—Zh08? The authors indicated that it was a literature search and an in-depth analysis of metabolic pathways (lines 360-364); however, how did they choose the order of genes to mutate? Additionally, they indicated that there was a combination of promoter usage and- copy numbers. Yet, for each modification, only one strain is shown, which is not explained if other strains had poor results.]
Response 3: We appreciate you pointing this out. For the order of genetic modification of strains zh04-zh08, we classified them according to the increase in precursor PRPP supply and serine supply. Firstly, to increase the supply of the precursor PRPP, we followed the PRPP synthesis pathway (Fig.5A) and successively modified prs (encoding PRPP synthetase), zwf (encoding glucose-6-phosphate dehydrogenase), gnd (encoding 6-phosphogluconate dehydrogenase) and rpiA (encoding ribulose phosphate isomerase A) genes. Strains zh04-zh07 were obtained. Thereafter, for increasing the supply of the precursor L-Ser, we co-overexpressed serB (encoding phosphoserine phosphatase) and serC (encoding phosphoserine aminotransferase) to obtain strain zh08.
Additionally, during the modification process, strains zh04-08 was modified only once at a time using promoter P4 and was not optimised for fine expression, thus only one strain was obtained at each step. For gene copy number, some of the modifications were overexpressed at the gene's original positions without increasing the copy number, while others were overexpressed at other positions and only one copy was added, and thus only one strain was obtained.
Thank you very much for your helpful suggestions. We have revised in the manuscript (Lines 384-390). 

Comments 4: [What criteria did the authors use to perform only two rounds of mutagenesis?]
Response 4: The screening criterion we adopted during the ARTP mutagenesis was the change in the L-Trp titer per OD600. Typically, screening of mutants using ARTP mutagenesis requires 2-3 rounds of stacked mutagenesis to obtain mutants that undergo significant positive changes. Firstly, the efficiency of ARTP mutagenesis decreases with the increase in the round of mutations. In the two rounds of ARTP stacking mutagenesis we performed, the L-Trp titer per OD600 of the mutant strain GT3938 was enhanced by 76.96% and 43.45%, respectively, which has been significantly changed. Secondly, strain GT3938 is still very much space for metabolic modification of the L-Trp metabolic pathway, and the modification of specific genes using metabolic engineering on the basis of ARTP mutagenesis can enhance the L-Trp production of mutant strain GT3938 even further. Consequently, we decided not to perform the third round of superimposed mutagenesis and directly used the GT3938 mutant strain obtained in the second round for subsequent metabolic modification.

Comments 5: [In line 134, the word "resuscitated" is used; it could be replaced with "reactivated", "recovered", or "reanimated" to be more appropriate.]
Response 5: Thank you for your helpful suggestion. We have checked and revised it in Line 132.

Comments 6: [The authors should explain why they used the DH5α strain to test the pACYC184-727 plasmid, as described in line 216. I presume it was because they needed a control, and the other strains were overproducers.]
Response 6: We appreciate you pointing this out. the DH5α strain carrying pACYC184-727 was employed to test the function of the L-Trp biosensor treated with L-Trp (Figure 1A). At the same time, strain trp01-03 was used to test the corresponding effect of the L-Trp biosensor in the endogenous production of L-Trp (Figure 1B). We have checked and revised all, such as lines 212-214.

Comments 7: [Specify in Figure 1A what the columns represent; they appear to be visible light and fluorescence.]
Response 7: Thank you for pointing this out. The left column in Figure 1A represents visible light and the right column represents fluorescence. We have made a revision to Figure 1A.

Comments 8: [In Graph 1B, the error bars overlap. The authors could change the scale of one of the axes or use different colours to differentiate it. Although the size of the bars allows for distinguishing some overlaps, this does not work for strain trp02.]
Response 8: Thank you for your helpful suggestion. We have made a revision to Figure 1B.

Comments 9: [The authors seem to use an arrow to symbolise each reaction in Figure 5A. However, why do they only use one arrow for the serA, serB, and serC genes?]
Response 9: Thank you for pointing this out. We have checked and made a correction in Figure 5A.

Comments 10: [To avoid ambiguities, the authors should briefly mention the origin of the P4 and P3 promoters used in this work.]
Response 10: Thanks to your suggestion and we agree with your point of view. We have cited references in the text to indicate the source of the P4 promoter (Line 98). And we have added the P3 and P4 promoter gene sequences in the supplementary table S2 and indicated their relative strengths in the exegesis.

Comments 11: [In the introduction, the authors cite the p15-ribo727 riboswitch, while in the methodology, they only call it fragment 727. It is advisable to maintain uniformity or, if there is a difference, to highlight it.]
Response 11: Thank you for pointing this out. However, the p15-ribo727 riboswitch mentioned in the introduction (Line 73) is the plasmid constructed in the report by Liu et al. It is fragment 727 that exercises the riboswitch function, so we refer to it as 727 here and reconstructed the plasmid named pACYC184-727 for this study.

Comments 12: [In Figure 4, "* for P < 0.05" is not used.]
Response 12: Thank you for pointing this out, we have removed the words ‘* for P < 0.05 and’ in Lines 338.

Again, we thank the reviewer for the positive and constructive comments regarding our paper.

Author Response File: Author Response.pdf