On the flexibility of the cellular amination network in E coli

Article activity feed

1. 

   Version published to 10.7554/elife.77492 on eLife

   Jul 25, 2022
2. 

   eLife

   Jul 7, 2022

   Author Response

   Reviewer #1 (Public Review):

   In this manuscript, the authors aim to evaluate the flexibility of the amination network in E. coli. To achieve this, they knock out key enzymes GDH and GOGAT (which supply the majority of the cell's fixed nitrogen by aminating 2-oxoglutarate to glutamate), creating a glutamate auxotroph strain (glut-aux). They first consider whether exogenous amino acids can either replace glutamate, create glutamate through transamination, or be converted directly into glutamate. They found that many amino acids rescued growth of glut-aux, either through conversion to glutamate (proline, via putA), or transamination to glutamate (many, via aspC), and validate this finding with isotopic nitrogen labeling and demonstrating concentration-growth rate dependence. Then, for some amino acids that didn't …

   Author Response

   Reviewer #1 (Public Review):

   In this manuscript, the authors aim to evaluate the flexibility of the amination network in E. coli. To achieve this, they knock out key enzymes GDH and GOGAT (which supply the majority of the cell's fixed nitrogen by aminating 2-oxoglutarate to glutamate), creating a glutamate auxotroph strain (glut-aux). They first consider whether exogenous amino acids can either replace glutamate, create glutamate through transamination, or be converted directly into glutamate. They found that many amino acids rescued growth of glut-aux, either through conversion to glutamate (proline, via putA), or transamination to glutamate (many, via aspC), and validate this finding with isotopic nitrogen labeling and demonstrating concentration-growth rate dependence. Then, for some amino acids that didn't initially rescue growth in the glut-aux strain, the authors engineer growth rescue through laboratory evolution, gene deletion, and exogenous transaminase overexpression. Finally, they propose that E. coli may accommodate non-canonical (non-glutamate) ammonium assimilation. Informed by the glut-aux rescue experiments, they engineer two strains that assimilate ammonium through alternative amino acids: aspartate (via native aspA overexpression) and leucine (via exogenous leucine dehydrogenase overexpression).

   Expanding the repertoire and characterization of auxotrophic microbial strains is an important goal for synthetic biology and metabolic engineering. By creating an E. coli glutamate auxotroph strain, demonstrating and expanding growth rescue on other amino acids, and engineering alternative ammonium entry points, the authors support their claims of flexibility and promiscuity in the cellular amination network. These claims are corroborated by comprehensive growth data and isotope labeling. While certain aspects of their investigation are not novel and the manuscript could benefit from more contextualizing, their findings will be of broad interest to researchers investigating nitrogen assimilation in microbes, and those seeking to engineer E. coli for bioproduction and novel metabolic circuits.

   Strengths:

   *The collection of growth rate data is comprehensive, and in combination with nitrogen isotope labelling, paints a clear picture of amine donation (and ammonium assimilation, in figure 7). The growth rate dependence experiments represent an impressive amount of work, and are particularly informative in the strain engineering experiments in figure 6.

   *The putA and aspC knockouts are elegant demonstrations of the specificity and promiscuity of E. coli's amination network, respectively. The contextualization with previous in vitro data was very informative, and reporting the minimal effect of the ybdL knockout demonstrated the importance of the glut-aux strain in assessing the promiscuity of various transaminases in their cellular context.

   *Engineering the growth of glut-aux on four amino acids that didn't originally rescue growth is impressive, particularly getting exogenous transaminases to work as intended. As mentioned in the manuscript, this shows the potential for this particular auxotroph strain to serve as a growth-based selection platform for alternative amine sources.

   *Engineering alternative ammonium assimilation through aspartate with a simple native AspA overexpression is a very strong demonstration of the flexibility of E. coli's amination network. This result may be useful for metabolic engineers looking to optimize E. coli for growth on formate and other low-energy substrates for the production of biofuel and high-value products.

   Weaknesses:

   *The framing of hypotheses for alternative routes of amine donor assimilation are clarifying to a reader unaware of the range of amine supplementation options available to E. coli: (i) replacement of glutamate, (ii) amine donation to 2-ketoglutarate to create glutamate, (iii) indirect amine donation to 2-ketoglutarate to create glutamate, and (iv) conversion to glutamate. However, outside of the figure 1 caption and lines 97-100 in the results section, the hypotheses are not mentioned again according to this classification. Directly after introducing figure 2, the authors discuss the possible ways that various amino acids rescue the growth of glutaux and hypothesize that amine transfer is responsible. It would be immediately helpful to organize this discussion by the i-iv classification system introduced in the preceding paragraph. Similarly, figures 5-7 can be classified in this way. For example, the action of PutA in figure 5, where you say that proline is "metabolized... to glutamate", is unclear, and presumably refers to being "metabolically converted to glutamate," hypothesis (iv).

   We thank the reviewer for this comment and agree that repeating the classification when specific mechanisms are described is indeed very helpful for understanding the context. Correspondingly, we now refer to these classifications whenever addressing rescue mechanisms in the text passages e.g. in lines 132, 133, 152, 189 – 191, 197, 213, 240, 394, 417, 468, 471.

   *Construction of a glutamate auxotroph strain, by deletion of gdhA and gltBD, is well-established (Dougherty et al, 1993), and has been standardized in the Keio collection (Baba et al, 2006). While it is critical that the authors used the same lambda-red recombinase strain for all deletions after making the glut-aux strain, it should be made clear to readers what has been done before by adding some context here.

   We thank the reviewer for drawing our attention to this relevant publication (https://doi.org/10.1128/jb.175.1.111-116.1993), which we now cite in lines 90-91. However, the publication describes a D-glutamate auxotrophic E. coli strain with nonsense mutations in the genes dga (glutamate racemase murI) and gltS (involved in Glu-transport). Thus, this is a very different strain to the one that is the center of our study, which is auxotrophic for L-glutamate.

   *Most of the experiments were conducted in M9 media with glycerol as the carbon and energy source. Glycerol is utilized by oxidative phosphorylation in E. coli, like glucose, but is not a preferred carbon source. However, glycerol leads to higher growth rates when amines are supplied by amino acids rather than ammonia, due to imbalances in 2-ketoglutarate (Anat Bren et al, Sci Rep 2016). Knowing that cellular pyruvate and 2-ketoglutarate concentrations are different depending on carbon and nitrogen source, and both are relevant for thermodynamic favorability in cellular amination networks, the authors should justify why glycerol is the carbon source used for most experiments, as they justify fumarate in figure 7.

   We agree with the reviewer that metabolite concentrations differ depending on which carbon source is fed. Ideally, glucose would be used as the most favored carbon source. However, the presence of glucose leads to catabolite repression and hence might not provide the best testing conditions when analyzing the growth effects of additionally provided amines. Hence, in our opinion glycerol, which can be fed together with other supplements, was the better choice to test growth rescuing effects of amino acids. We now explain this to the reader from line 114 - 116.

   *In figure 3, it's unclear why AFLMPST and R are the only proteinogenic amino acids that are analyzed for 15N labeling. One might assume it's a technical issue for the mass spectroscopy data, but the relatively small selection of both amino-donor amino acids and 15N fraction amino acids makes initial interpretation of the figure confusing. Emphasizing that the 15N measurements are representative of all proteinogenic amino acids, and the amino-donors are representative of all amino acids that rescued growth for glut-aux would help.

   We agree with the reviewer that the amino acids selected need to be explained. From line 157, we now explain that we show amino acids covering the aspartate, glutamate, and serine families as well as representatives of branched-chain, aromatic, and arginine which contains an amine derived from the δ amino group of glutamine that originated from free ammonium.

   Additionally, figure 3B could benefit from greater distinction between amine groups and ammonium ions. –

   We changed the figure according to the reviewer’s suggestion. Now in both Figure 1 and Figure 3B free ammonium is highlighted in boxes thus making it is easier to distinguish it to amino groups

   Also, there is presumably a typo in the "external amine donor" cartoon, with 14NH4+ in the grey circle rather than 14NH3+.

   The error was changed accordingly.

   *Adaptive laboratory evolution is not a fair description of how the authors found that a dadX mutation led to growth rescue of glut-aux+alaA on alanine (line 221). Although two weeks of growth may allow for evolution of E. coli in some cases, a single growth curve over two weeks is similar in duration and concept to some of the other (non-evolution) growth curve experiments (Figure 6C). Rather than being evolved from a series of mutations, the appearance of dadX mutants is much more likely the result of highly stringent selection on mutations acquired during outgrowth before the selection was applied. Given the inoculum size of ~106 cells from overnight culture, and E. coli's spontaneous mutation rate of ~10-3 mutations per genome per generation, there is a reasonable probability of isolating one or more dadX mutant cells in the inoculum, which then expanded over two weeks (rather than evolved), given the growth rate of those mutants evident in figure 6A. Labeling this experiment as a spontaneous mutant selection of glut-aux+alaA engineered strain would make the aims and outcome of the experiment more transparent. Alternatively, one could report the growth data from the experiment, if available, or conduct a selective plating of prepared glut-aux+alaA inocula on M9+alanine plates to show the existence of a small mutant population.

   After checking again we now show data from a platereader experiment in supplementary Figure 6-figure supplement 1A showing the emergence of spontaneous mutants after a shorter cultivation time (110 and 130 h). We changed the text according to the reviewer’s suggestion and describe the observation as spontaneous mutants, lines 259 - 260.

   *Beta-alanine and ornithine are important non-proteinogenic amino acids, but there are hundreds of others. It is unclear to the reader why they were selected for assessing amine donation to glut-aux, or why beta-alanine was selected for adding an exogenous transamination route. Stating the relevance of these amino acids to E. coli's amination network or metabolic engineering, or stating that they were serendipitous findings of rescue and no rescue of glut-aux by non-proteinogenic amino acids, would make the choices for strain engineering seem less arbitrary. Similarly, strains engineered to utilize glycine or serine as amine donors (fig 6), or aspartate or leucine as centers of ammonium fixation (fig 7), seem to be chosen arbitrarily out of many amino acids that did or did not initially rescue growth of glut-aux. Simply stating that these were the best (or worst) amine donors based on growth rescue in figure 2 would explain why the strain engineering was not systematic over all 20 proteinogenic amino acids for ammonium fixation or amine donation.

   We thank the reviewer for the constructive criticism. Our investigations first started with testing proteinogenic amino acids. After many of these amino acids rescued the growth of the glut-aux strain we investigated if also non-proteinogenic amino acids are able to complement the growth of the strain. Beta-alanine and ornithine were chosen because they are derived from aspartate (beta-alanine) and glutamate (ornithine), both belonging to families best-complementing growth of the glut-aux strain. We now comment on why we selected these amino acids in the text from line 110 - 114.

   In the paragraph describing engineering the use of beta-alanine, we now refer to the biotechnological interest of beta-alanine-derived products (lines 286 - 289).

   From line 321, we describe why engineering the use of glycine and serine as amine donors is relevant.

   In the ammonium assimilation section, we now comment from line 381 that aspartate and leucine represent amine donors supporting fast and very slow growth.

   Reviewer #2 (Public Review):

   The authors are trying to show the existence of a reversible amination network that allows nitrogen transfer via transaminases for synthesis of several amino acids. Nitrogen assimilation and distribution is known to start with ammonia assimilation via glutamate and glutamine synthesis, and subsequent transfer of the nitrogen via transaminases and amidotransferases. To demonstrate an amination network, i.e., reversible nitrogen transfer, the authors start with a glutamate auxotroph and provide a variety of compounds to determine which support growth. Growth implies the transfer of amino groups to glutamate. The authors show some pathways required for transfer of the amino acid nitrogen via genetic analysis. The concept of an amination network is clever since current thinking would suggest that nitrogen flows in one direction and is not reversible. The basic method is genetic which is sometimes supplemented with isotope dilution experiments. In addition to analysis of a possible amination network, experiments are presented that test whether alternate routes of ammonia assimilation (the source for amino groups) are possible. While the authors show that a reversible amination can exist (nitrogen flow from glutamate is known, while the authors show that nitrogen flow to glutamate is possible), they do not provide any evidence that the nitrogen flux to glutamate does exist in nature for wild-type strains. A genetic analysis with complex strains (multiple mutations) and very specific growth conditions cannot provide evidence for nitrogen flux to glutamate from other amino acids. Positive evidence requires a biochemical analysis: how much N15 from an amino acid is transferred to glutamate or other amino acids. Without such results, it cannot be established that amino groups can be transferred to glutamate at an appreciable level or that the amination network is reversible, which is an important conclusion of this work. Without such results, the proposed amination network is a theoretical possibility that is detectable only in genetically complex strains and specific medium. The impact of the work is more limited without a biochemical analysis.

   We agree with the reviewer’s comment that we use a synthetic system to study the amination network. The results obtained with this system do not necessarily describe the amination network operating in a wild-type strain, especially because of differing metabolite concentrations. Also, we are not claiming that these reverse reactions, e.g. for glutamate synthesis are happening in a wild-type E. coli and we are referring to the glut-aux strain in all relevant parts of the manuscript.

   Nevertheless, we believe that our approach of using the glut-aux strain as a readout reveals the potential of the network to interconvert amino acids into one another, although parts of the network may only become relevant under specific conditions in which glutamate availability is limiting.

   We made some amendments to the manuscript clarifying that we are using a synthetic strain for our analysis which may not reflect the intracellular amino acid concentrations in the wild-type (lines 145 – 148, lines 462 – 466).

   We agree that a further deep biochemical characterization of the amination network, as well as dynamic 15Nlabeling experiments, will reveal very valuable information about the connections within the amination network. However, these analyses are beyond the scope of this manuscript and cannot be easily done.

   Strengths and weaknesses.

   The results suggest that the amino group can be readily transferred between keto acids via a network of transaminases in strains. Several of the proposed reactions are novel and have not been previously described, such as tryptophan as an amino donor. The idea and experimental design are clever.

   We thank the reviewer for the support.

   This work does not discuss several relevant topics. Please notice that the comment is that several issues are not discussed or considered. The topics overlap.

   The activities of transaminases are important for this study but are not discussed. A useful summary of transaminase levels is provided in the following reference: Mol Microbiol . 2014 94:843-56. doi: 10.1111/mmi.12801. PMID: 25243376. The 3 most abundant transaminases are SerC, AspC, and IlvE. The results from that paper are consistent with many results of this work. In the cited work, all defects in transaminases that result in phenotypes were complemented with all transaminase genes. The cited paper is directly relevant to this work.

   We thank the reviewer for the comment. Now, when referring to overlapping transaminase activities we cite the work of Lal et al. who identified high enzymatic redundancies within E. coli transaminases (lines 229 – 233).

   There is no discussion of metabolite levels. This is important since the most abundant metabolite in wild type E. coli is undoubtedly glutamate (see work by Rabinowitz), which by mass action will provide the direction for transaminase reactions. For the mutant strains used, glutamate might not be the most abundant metabolite, and reversible transaminases conceivably will flow in an unphysiological direction. It is likely that metabolite levels are substantially perturbed in the mutants analyzed, and that the proposed amination network (nitrogen flow to glutamate) requires these metabolic perturbations. Several of these perturbations are likely to be effects on glutamine synthetase (GS). The GltBD mutant should prevent induction of the Ntr response. However, given the unusual conditions assayed, this is not a certainty. Several amino acids inhibit GS activity, including serine, glycine, alanine, histidine, and tryptophan. The levels of metabolites needed to drive reactions toward glutamate synthesis may never occur.

   We agree with the reviewer that our synthetic strain’s metabolome is most like different from the WT. From line 146 we clarify the difference in the glut-aux strain compared to the WT. Additionally, we mention the difference in the discussion from line 462.

   GS activity relies on glutamate availability and hence glutamate needs to be produced via the network first (regardless of which amine donor is provided). Thus, inhibition of GS would cause glutamine auxotrophy and hence a no-growth phenotype. However, since we see immediate growth on histidine and tryptophan, GS must not be inhibited completely. Thus we conclude that GS regulation is not stopping the strains from growing but might have an influence on growth velocity (from line 536).

   Additionally, growth with serine, glycine, or alanine was possible after engineering or adaptation. However, for the strain evolved towards growth with L-alanine, no mutations associable with GS regulation were observed.

   However, now we emphasized regulatory/inhibitory effects in lines 446 and 536 - 546, and give more explanations on amino acid toxicity (lines 447 - 449).

   There is no discussion of the regulation of the enzymes involved. For example, aspA is controlled by several factors (EcoCyc). Is the control of the relevant enzymes consistent with the proposed amino transfer, or does the cell require a novel form of regulation and/or a suppressor mutation? Do the transaminase levels change during these experiments?

   We thank the reviewer for the constructive criticism. Since both AspA and LeuDH were overexpressed from plasmids with synthetic promoters, we don’t expect native transcriptional regulation to be relevant (mentioned in line 539). Since in both cases we obtained immediate growth in selective conditions, we dont expect suppressor mutations to be required in the tested conditions. However, we recognize that a lack of regulatory control mechanisms like present for GDH / GS will result in less flexible metabolic adaptation to changing conditions (lines 541 - 543), thus limiting the utilization space for these ammonium assimilation mechanisms. Although we didn’t measure transaminase levels, in line with the summarizing model suggested by reviewer #3, we now suggest transaminase upregulation as a possible adaptation mechanism of the glutaux in the discussion (lines 486 - 493).

   The authors are imposing strong selective pressure (growth or no growth) and the possibility that suppressor mutations can rapidly accumulate is not discussed or assessed.

   The reviewer is right, mutations can lead to falsified results under selective pressure. That is why we chose our preculture conditions accordingly. This is now clarified at the beginning of the results section from line 118 - 121.

   The results of this paper make the implicit assumption that the transport of any amino acid that is added to the medium will not limit growth. Transport is undoubtedly often limiting. To avoid this problem, di- and tripeptides could have been used. Both are rapidly transported, and the amination network may prove to be larger than the results suggest. The use of dipeptides could increase the amino acids that can transfer amino groups, since their internal concentration would be higher. Experiments are not requested, but the authors should consider whether their proposed network is potentially larger. (It is understood that on one hand the reviewer is questioning whether a reversible network exists, and on the other hand that it may be larger. These are not incompatible since under conditions in which peptides are the amino source, the network may exist.)

   We thank the reviewer for this comment. In the discussion section we now refer to this point from line 445. Please note that we already mention in the in the results section that glutamate uptake might be limiting (from line 195 - 199).

   Furthermore, we discussed the topic of threonine, which can serve as a nitrogen source only if the expression of threonine dehydrogenase is induced by the presence of leucine (from line 449). For valine, we expect valinebased inhibition of acetohydroxy acid synthase needed for isoleucine biosynthesis to be toxic for all strains (line 451 - 452). For beta-alanine, we showed that transaminase availability is limiting growth, and that reengineering growth was independent from transport mechanisms (from line 509).

   The proposed alternate ammonia assimilation pathway has some interesting conceptual issues that should be addressed. Ammonia assimilation is necessarily at the interface of carbon/energy and nitrogen metabolism. The incredibly complex control of ammonia assimilation via glutamine and glutamate has layers upon layers of regulation that ensure that energy is not drained when all nitrogen-containing compounds are present at sufficient levels. Any alternate ammonia assimilation pathway in nature must take this into consideration. It is predicted that the constructed strains in this study will poorly handle many environmental stresses and changing nutrient content. These considerations are largely theoretical but limit the ability of alternate pathways to exist in nature, except perhaps under certain conditions. It is not suggested that these issues should be addressed experimentally but it would be important to acknowledge them.

   We agree with the reviewer that this is a very valuable point to discuss. We added a section to the discussion from line 536.

   Read the original source
3. 

   eLife

   Jul 7, 2022

   Evaluation Summary:

   In this paper, the authors demonstrate that a reversible amination network that allows nitrogen transfer via transaminases for synthesis of several amino acids can be constructed in laboratory strains through clever and carefully designed experiments. As a result, this work should be of interest to microbiologists, biochemists, synthetic biologists, and biotechnologists.

   (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

   Read the original source
4. 

   eLife

   Jul 7, 2022

   Reviewer #1 (Public Review):

   In this manuscript, the authors aim to evaluate the flexibility of the amination network in E. coli. To achieve this, they knock out key enzymes GDH and GOGAT (which supply the majority of the cell's fixed nitrogen by aminating 2-oxoglutarate to glutamate), creating a glutamate auxotroph strain (glut-aux). They first consider whether exogenous amino acids can either replace glutamate, create glutamate through transamination, or be converted directly into glutamate. They found that many amino acids rescued growth of glut-aux, either through conversion to glutamate (proline, via putA), or transamination to glutamate (many, via aspC), and validate this finding with isotopic nitrogen labeling and demonstrating concentration-growth rate dependence. Then, for some amino acids that didn't initially rescue growth in …

   Reviewer #1 (Public Review):

   In this manuscript, the authors aim to evaluate the flexibility of the amination network in E. coli. To achieve this, they knock out key enzymes GDH and GOGAT (which supply the majority of the cell's fixed nitrogen by aminating 2-oxoglutarate to glutamate), creating a glutamate auxotroph strain (glut-aux). They first consider whether exogenous amino acids can either replace glutamate, create glutamate through transamination, or be converted directly into glutamate. They found that many amino acids rescued growth of glut-aux, either through conversion to glutamate (proline, via putA), or transamination to glutamate (many, via aspC), and validate this finding with isotopic nitrogen labeling and demonstrating concentration-growth rate dependence. Then, for some amino acids that didn't initially rescue growth in the glut-aux strain, the authors engineer growth rescue through laboratory evolution, gene deletion, and exogenous transaminase overexpression. Finally, they propose that E. coli may accommodate non-canonical (non-glutamate) ammonium assimilation. Informed by the glut-aux rescue experiments, they engineer two strains that assimilate ammonium through alternative amino acids: aspartate (via native aspA overexpression) and leucine (via exogenous leucine dehydrogenase overexpression).

   Expanding the repertoire and characterization of auxotrophic microbial strains is an important goal for synthetic biology and metabolic engineering. By creating an E. coli glutamate auxotroph strain, demonstrating and expanding growth rescue on other amino acids, and engineering alternative ammonium entry points, the authors support their claims of flexibility and promiscuity in the cellular amination network. These claims are corroborated by comprehensive growth data and isotope labeling. While certain aspects of their investigation are not novel and the manuscript could benefit from more contextualizing, their findings will be of broad interest to researchers investigating nitrogen assimilation in microbes, and those seeking to engineer E. coli for bioproduction and novel metabolic circuits.

   Strengths:

   *The collection of growth rate data is comprehensive, and in combination with nitrogen isotope labelling, paints a clear picture of amine donation (and ammonium assimilation, in figure 7). The growth rate dependence experiments represent an impressive amount of work, and are particularly informative in the strain engineering experiments in figure 6.

   *The putA and aspC knockouts are elegant demonstrations of the specificity and promiscuity of E. coli's amination network, respectively. The contextualization with previous in vitro data was very informative, and reporting the minimal effect of the ybdL knockout demonstrated the importance of the glut-aux strain in assessing the promiscuity of various transaminases in their cellular context.

   *Engineering the growth of glut-aux on four amino acids that didn't originally rescue growth is impressive, particularly getting exogenous transaminases to work as intended. As mentioned in the manuscript, this shows the potential for this particular auxotroph strain to serve as a growth-based selection platform for alternative amine sources.

   *Engineering alternative ammonium assimilation through aspartate with a simple native AspA overexpression is a very strong demonstration of the flexibility of E. coli's amination network. This result may be useful for metabolic engineers looking to optimize E. coli for growth on formate and other low-energy substrates for the production of biofuel and high-value products.

   Weaknesses:

   *The framing of hypotheses for alternative routes of amine donor assimilation are clarifying to a reader unaware of the range of amine supplementation options available to E. coli: (i) replacement of glutamate, (ii) amine donation to 2-ketoglutarate to create glutamate, (iii) indirect amine donation to 2-ketoglutarate to create glutamate, and (iv) conversion to glutamate. However, outside of the figure 1 caption and lines 97-100 in the results section, the hypotheses are not mentioned again according to this classification. Directly after introducing figure 2, the authors discuss the possible ways that various amino acids rescue the growth of glut-aux and hypothesize that amine transfer is responsible. It would be immediately helpful to organize this discussion by the i-iv classification system introduced in the preceding paragraph. Similarly, figures 5-7 can be classified in this way. For example, the action of PutA in figure 5, where you say that proline is "metabolized... to glutamate", is unclear, and presumably refers to being "metabolically converted to glutamate," hypothesis (iv).

   *Construction of a glutamate auxotroph strain, by deletion of gdhA and gltBD, is well-established (Dougherty et al, 1993), and has been standardized in the Keio collection (Baba et al, 2006). While it is critical that the authors used the same lambda-red recombinase strain for all deletions after making the glut-aux strain, it should be made clear to readers what has been done before by adding some context here.

   *Most of the experiments were conducted in M9 media with glycerol as the carbon and energy source. Glycerol is utilized by oxidative phosphorylation in E. coli, like glucose, but is not a preferred carbon source. However, glycerol leads to higher growth rates when amines are supplied by amino acids rather than ammonia, due to imbalances in 2-ketoglutarate (Anat Bren et al, Sci Rep 2016). Knowing that cellular pyruvate and 2-ketoglutarate concentrations are different depending on carbon and nitrogen source, and both are relevant for thermodynamic favorability in cellular amination networks, the authors should justify why glycerol is the carbon source used for most experiments, as they justify fumarate in figure 7.

   *In figure 3, it's unclear why AFLMPST and R are the only proteinogenic amino acids that are analyzed for 15N labeling. One might assume it's a technical issue for the mass spectroscopy data, but the relatively small selection of both amino-donor amino acids and 15N fraction amino acids makes initial interpretation of the figure confusing. Emphasizing that the 15N measurements are representative of all proteinogenic amino acids, and the amino-donors are representative of all amino acids that rescued growth for glut-aux would help. Additionally, figure 3B could benefit from greater distinction between amine groups and ammonium ions. Also, there is presumably a typo in the "external amine donor" cartoon, with 14NH4+ in the grey circle rather than 14NH3+.

   *Adaptive laboratory evolution is not a fair description of how the authors found that a dadX mutation led to growth rescue of glut-aux+alaA on alanine (line 221). Although two weeks of growth may allow for evolution of E. coli in some cases, a single growth curve over two weeks is similar in duration and concept to some of the other (non-evolution) growth curve experiments (Figure 6C). Rather than being evolved from a series of mutations, the appearance of dadX mutants is much more likely the result of highly stringent selection on mutations acquired during outgrowth before the selection was applied. Given the inoculum size of ~106 cells from overnight culture, and E. coli's spontaneous mutation rate of ~10-3 mutations per genome per generation, there is a reasonable probability of isolating one or more dadX mutant cells in the inoculum, which then expanded over two weeks (rather than evolved), given the growth rate of those mutants evident in figure 6A. Labeling this experiment as a spontaneous mutant selection of glut-aux+alaA engineered strain would make the aims and outcome of the experiment more transparent. Alternatively, one could report the growth data from the experiment, if available, or conduct a selective plating of prepared glut-aux+alaA inocula on M9+alanine plates to show the existence of a small mutant population.

   *Beta-alanine and ornithine are important non-proteinogenic amino acids, but there are hundreds of others. It is unclear to the reader why they were selected for assessing amine donation to glut-aux, or why beta-alanine was selected for adding an exogenous transamination route. Stating the relevance of these amino acids to E. coli's amination network or metabolic engineering, or stating that they were serendipitous findings of rescue and no rescue of glut-aux by non-proteinogenic amino acids, would make the choices for strain engineering seem less arbitrary. Similarly, strains engineered to utilize glycine or serine as amine donors (fig 6), or aspartate or leucine as centers of ammonium fixation (fig 7), seem to be chosen arbitrarily out of many amino acids that did or did not initially rescue growth of glut-aux. Simply stating that these were the best (or worst) amine donors based on growth rescue in figure 2 would explain why the strain engineering was not systematic over all 20 proteinogenic amino acids for ammonium fixation or amine donation.

   Read the original source
5. 

   eLife

   Jul 7, 2022

   Reviewer #2 (Public Review):

   The authors are trying to show the existence of a reversible amination network that allows nitrogen transfer via transaminases for synthesis of several amino acids. Nitrogen assimilation and distribution is known to start with ammonia assimilation via glutamate and glutamine synthesis, and subsequent transfer of the nitrogen via transaminases and amidotransferases. To demonstrate an amination network, i.e., reversible nitrogen transfer, the authors start with a glutamate auxotroph and provide a variety of compounds to determine which support growth. Growth implies the transfer of amino groups to glutamate. The authors show some pathways required for transfer of the amino acid nitrogen via genetic analysis. The concept of an amination network is clever since current thinking would suggest that nitrogen flows …

   Reviewer #2 (Public Review):

   The authors are trying to show the existence of a reversible amination network that allows nitrogen transfer via transaminases for synthesis of several amino acids. Nitrogen assimilation and distribution is known to start with ammonia assimilation via glutamate and glutamine synthesis, and subsequent transfer of the nitrogen via transaminases and amidotransferases. To demonstrate an amination network, i.e., reversible nitrogen transfer, the authors start with a glutamate auxotroph and provide a variety of compounds to determine which support growth. Growth implies the transfer of amino groups to glutamate. The authors show some pathways required for transfer of the amino acid nitrogen via genetic analysis. The concept of an amination network is clever since current thinking would suggest that nitrogen flows in one direction and is not reversible. The basic method is genetic which is sometimes supplemented with isotope dilution experiments. In addition to analysis of a possible amination network, experiments are presented that test whether alternate routes of ammonia assimilation (the source for amino groups) are possible. While the authors show that a reversible amination can exist (nitrogen flow from glutamate is known, while the authors show that nitrogen flow to glutamate is possible), they do not provide any evidence that the nitrogen flux to glutamate does exist in nature for wild-type strains. A genetic analysis with complex strains (multiple mutations) and very specific growth conditions cannot provide evidence for nitrogen flux to glutamate from other amino acids. Positive evidence requires a biochemical analysis: how much N15 from an amino acid is transferred to glutamate or other amino acids. Without such results, it cannot be established that amino groups can be transferred to glutamate at an appreciable level or that the amination network is reversible, which is an important conclusion of this work. Without such results, the proposed amination network is a theoretical possibility that is detectable only in genetically complex strains and specific medium. The impact of the work is more limited without a biochemical analysis.

   Strengths and weaknesses.

   The results suggest that the amino group can be readily transferred between keto acids via a network of transaminases in strains. Several of the proposed reactions are novel and have not been previously described, such as tryptophan as an amino donor. The idea and experimental design are clever.

   This work does not discuss several relevant topics. Please notice that the comment is that several issues are not discussed or considered. The topics overlap.

   The activities of transaminases are important for this study but are not discussed. A useful summary of transaminase levels is provided in the following reference: Mol Microbiol . 2014 94:843-56. doi: 10.1111/mmi.12801. PMID: 25243376. The 3 most abundant transaminases are SerC, AspC, and IlvE. The results from that paper are consistent with many results of this work. In the cited work, all defects in transaminases that result in phenotypes were complemented with all transaminase genes. The cited paper is directly relevant to this work.

   There is no discussion of metabolite levels. This is important since the most abundant metabolite in wild type E. coli is undoubtedly glutamate (see work by Rabinowitz), which by mass action will provide the direction for transaminase reactions. For the mutant strains used, glutamate might not be the most abundant metabolite, and reversible transaminases conceivably will flow in an unphysiological direction. It is likely that metabolite levels are substantially perturbed in the mutants analyzed, and that the proposed amination network (nitrogen flow to glutamate) requires these metabolic perturbations. Several of these perturbations are likely to be effects on glutamine synthetase (GS). The GltBD mutant should prevent induction of the Ntr response. However, given the unusual conditions assayed, this is not a certainty. Several amino acids inhibit GS activity, including serine, glycine, alanine, histidine, and tryptophan. The levels of metabolites needed to drive reactions toward glutamate synthesis may never occur.

   There is no discussion of the regulation of the enzymes involved. For example, aspA is controlled by several factors (EcoCyc). Is the control of the relevant enzymes consistent with the proposed amino transfer, or does the cell require a novel form of regulation and/or a suppressor mutation? Do the transaminase levels change during these experiments?

   The authors are imposing strong selective pressure (growth or no growth) and the possibility that suppressor mutations can rapidly accumulate is not discussed or assessed.

   The results of this paper make the implicit assumption that the transport of any amino acid that is added to the medium will not limit growth. Transport is undoubtedly often limiting. To avoid this problem, di- and tri-peptides could have been used. Both are rapidly transported, and the amination network may prove to be larger than the results suggest. The use of dipeptides could increase the amino acids that can transfer amino groups, since their internal concentration would be higher. Experiments are not requested, but the authors should consider whether their proposed network is potentially larger. (It is understood that on one hand the reviewer is questioning whether a reversible network exists, and on the other hand that it may be larger. These are not incompatible since under conditions in which peptides are the amino source, the network may exist.)

   The proposed alternate ammonia assimilation pathway has some interesting conceptual issues that should be addressed. Ammonia assimilation is necessarily at the interface of carbon/energy and nitrogen metabolism. The incredibly complex control of ammonia assimilation via glutamine and glutamate has layers upon layers of regulation that ensure that energy is not drained when all nitrogen-containing compounds are present at sufficient levels. Any alternate ammonia assimilation pathway in nature must take this into consideration. It is predicted that the constructed strains in this study will poorly handle many environmental stresses and changing nutrient content. These considerations are largely theoretical but limit the ability of alternate pathways to exist in nature, except perhaps under certain conditions. It is not suggested that these issues should be addressed experimentally but it would be important to acknowledge them.

   Read the original source
6. 

   eLife

   Jul 7, 2022

   Reviewer #3 (Public Review):

   This article describes a series of experiments aimed at mapping in a comprehensive way the possible avenues by which E. coli is able to assimilate nitrogen into its metabolism. In particular, the authors address the question of whether (and how efficiently) different amino acids can be used by E. coli as a nitrogen source. The paper portrays an interesting and complex global picture of how the nitrogen assimilation capabilities are shaped by the connectivity of metabolism and the promiscuity of some enzymes. The results are based on a combination of genetic modifications and laboratory evolution experiments, all carefully planned to dive deep into each individual case. The evidence presented, in the form of growth curves and isotope labelling patterns, seems very solid and clear. In fact, I find it truly …

   Reviewer #3 (Public Review):

   This article describes a series of experiments aimed at mapping in a comprehensive way the possible avenues by which E. coli is able to assimilate nitrogen into its metabolism. In particular, the authors address the question of whether (and how efficiently) different amino acids can be used by E. coli as a nitrogen source. The paper portrays an interesting and complex global picture of how the nitrogen assimilation capabilities are shaped by the connectivity of metabolism and the promiscuity of some enzymes. The results are based on a combination of genetic modifications and laboratory evolution experiments, all carefully planned to dive deep into each individual case. The evidence presented, in the form of growth curves and isotope labelling patterns, seems very solid and clear. In fact, I find it truly beautiful that one can still learn so much about metabolism by measuring growth under cleverly chosen combinations of environmental conditions. I think that this manuscript is important, and will have a strong impact, both in terms of fundamental understanding of metabolism, and as a starting point for future metabolic engineering applications. The article is well written, and inspiring, as it makes the readers realize how the amination network is at the same time concentrated around a few key pathways, but also substantially diverse in its interactions with other cellular processes.

   Read the original source
7. 

   Version published to 10.1101/2022.01.25.477661v1 on bioRxiv

   Jan 27, 2022