Genetically controlled membrane synthesis in liposomes

Lipid membranes, nucleic acids, proteins, and metabolism are essential for modern cellular life. Synthetic systems emulating the fundamental properties of living cells must therefore be built upon these functional elements. In this work, phospholipid-producing enzymes encoded in a synthetic minigenome are cell-free expressed within liposome compartments. The de novo synthesized metabolic pathway converts precursors into a variety of lipids, including the constituents of the parental liposome. Balanced production of phosphatidylethanolamine and phosphatidylglycerol is realized, owing to transcriptional regulation of the activity of specific genes combined with a metabolic feedback mechanism. Fluorescence-based methods are developed to image the synthesis and membrane incorporation of phosphatidylserine at the single liposome level. Our results provide experimental evidence for DNA-programmed membrane synthesis in a minimal cell model. Strategies are discussed to alleviate current limitations toward effective liposome growth and self-reproduction.

In this manuscript the Danelon laboratory does a good job of developing a cell-free system for the synthesis of lipids. Such work is of clear importance to the synthesis of synthetic cells and may have additional biotechnological applications. The experiments are generally well designed and the data are convincing.
This group did publish previously in PLOS ONE similar work that demonstrated the synthesis of lipid. It is not immediately clear what the difference is with this submitted manuscript. The authors should do a better job in indicating what the new advance is.
I think this work is important and will be used by several laboratories as they try to build artificial cells. I do, however, have suggestions that the authors may wish to consider during revisions.
1. I do not believe that the authors ever demonstrated that the membrane proteins that were cell-free expressed actually partitioned to the membrane. Functional activity is certainly important, but it would be nice to demonstrate the localization of these proteins since that may aid future work. 2. Liposome growth was not demonstrated but could have easily been observed with FRET labeled lipids. This would strengthen the manuscript considerably since the goal appears to be growth. 3. The figures frequently lack statistical analysis and an indication of the number of replicates. Sometimes the units of the axes are missing. 4. The first two sentences of the abstract are probably a little too strong regarding what is needed for life. 5. The introduction begins by stating that life is an individual construct, but that is again, likely too strong. Evolution gives rise to populations. I know what the authors mean, but more care should be taken in how these ideas are expressed. 6. Last sentence of the introduction: "Our results provide experimental evidence for DNA-encoded homeostatic growth of a liposome-based artificial cell." But the authors do not demonstrate growth in this manuscript. 7. It is not clear why data was split between figure 2 and Supplementary Figure 11. The two should be combined together. 8. The degradation of material external to the liposomes was either with proteinase K or DNase I. Is there a reason why both weren't used together? 9. Line 5, page 10: It is unclear what is meant by "hidden in ensemble measurements." 10. Supplementary Figure 4: Why are the lipid concentrations in arbitrary units? 11. Supplementary Figure 4: The legend states "We hypothesized that this was caused by transcription termination read-through by the T7 RNAP, transcribing downstream PG synthesis genes even if those are under control of an orthogonal promoter." One way to avoid this problem is to clone the different modules in different directions. 12. Supplementary Figure 8: without error bars, the data are hard to interpret 13. Figure 3b: The difference between the dashed lines is confusing and not enough to appreciate the logic of the experiment.
Reviewer #2 (Remarks to the Author): In this study, Blanken and colleagues constructed a single plasmid encoding seven genes involved in lipid synthesis and showed protein synthesis using a reconstituted transcription/translation system (PURE system) from the information. The results showed that the 7 protein involved in lipid synthesis was successfully synthesized in the liposomes, and indicated the synthesis of the objective lipids. The direction in which this research aims is certainly important in the future of synthetic biology. However, questions remain regarding the achievement, novelty of the thesis, and conceptual advances.
The authors conclude in the title "membrane synthesis", but no evidence has been presented beyond lipid synthesis as the author mentioned that they failed to observe an increase of liposomes sizes by their system. This is in contrast to reference 19 that shows this point. Regarding novelty, previous studies have already developed technologies to combine seven or more genes into one plasmid. Expression and insertion to the membrane of the 7 proteins involved in lipid synthesis have been shown in their previous study. Although the only novel point of the manuscript is the expression of the 7 proteins in liposomes, there are many reports that showed normal protein synthesis (including membrane proteins as the author already reported) via the PURE system within liposomes. Due to the levels of achievement they set and lack of significant novelty and conceptual advances, I do not support the current manuscript for publishing in Nature Communication.
Other major comments L2, p3. Figure 5 shows synthesis of PS within GUVs. However, PS is also not part of the original membrane composition of E. coli system. L9, p3. No evidence of the growth of liposome is shown in this manuscript.
L27, p5. I think normalization using a different peptide fragment is a non-standard method in MSbased quantification. Do the authors have a reference or data that support the feasibility of the normalization? Figure 3&4. I wonder why the protease or DNase outside liposomes affect protein expression within liposomes. Does the author have experimental data to explain this point?

References
The authors should cite pioneer studies that use cell extract system within GUVs

Methods
The calibration method to compare fluorescent intensities among samples (gain or exposure time, laser strength, and standard sample) is missing. Therefore, I cannot evaluate the reliability of the quantification data from confocal microscopy.
Purity of LactC2-eGFP It seems too low purity to conclude something. I strongly suggest to add some data using more purified LactC2-eGFP through further purification steps.
Minor comments L2, p2. Please refer some reference. Reference 51. The title is incorrect. Furthermore, the study did not show "liposome division".
Reviewer #3 (Remarks to the Author): The manuscript by Blanken et al. demonstrated the possibility of membrane lipid synthesis within the encapsulated volume of lipid bilayer vesicles using cell-free expressed enzymes with transcriptional and metabolic regulation and subsequent incorporation of synthesized lipids into the vesicle membranes. Four major types of lipids were detected in bulk and in confinement. Both quantitative and qualitative assays were developed to assess the yield and localization of the different lipids. As mentioned in the manuscript, the authors claim the major significance of the study as the ability of the presented system to synthesize a repertoire of membrane lipids which are the same as the ones used to make liposomes in the first place from a so called "mini-genome". By changing the concentrations of the added polymerases and altering the membrane lipid compositions, the final yield and type of lipids made could be controlled. Quantitative techniques like mass spectrometry and lipidomics were used to detect and quantify yield of reaction products at different stages of the enzymatic pathways. Two types of optical detection assays were also developed to monitor localization of synthesized lipids on liposomal membranes when expressed in confinement. This is indeed a major advance in the field and could be an impactful approach for possible growth and division of synthetic cell-like liposomes (although growth has not been demonstrated here). While expression of enzymes with cell-free expression systems have been attempted before with successful production of lipids like PA and PS, the work is novel in its use of a mini-genome-like plasmid with seven genes and a combination of acyl-CoA with different chain lengths to generate hybrid lipid mixtures in a one pot reaction. Overall, the manuscript is well written and organized. The work is quite relevant in the field of in vitro lipid synthesis and use of cell-free expression as a tool for reconstitution of enzymatic pathways and membrane proteins. There are however certain limitations of the presented approach that need to be elucidated to provide the full scope of the work to the readers. My major concerns are as follows.
1. The authors have stated that the primary reason for building a polycistronic plasmid with all the genes for the Kennedy pathway was to prevent heterogeneity in expression due to variation in plasmid copies when encapsulating multiple plasmids in liposomes. However, using the same promotor to express all the enzymes for the PE pathway doesn't allow control over the extent of expression of individual proteins. Since acyl-CoA and CTP are consistently used by several proteins of this pathway, there is a competition for these resources and having equal amounts of each protein will lead to nonoptimal synthesis and low yield. Instead, using promotors with different strengths specific to the endogenous E. coli RNA polymerase can be beneficial to improve the rate and yield of the final products. Have the authors considered this option? 2. The plot in Fig. 2b shows similar yields of DOPE in the presence and absence of SP6 RNA polymerase for all values of DOPG concentration in the LUV membranes. The authors have mentioned that the PG synthesis pathway is hindered and results in low yield at high initial DOPG concentrations. However, at low DOPG concentrations, there seems to be no difference between the yield of DOPE with and without SP6 addition despite a significant amount of DOPG synthesis. Further discussion on this observation should be included. The authors should also comment that at 50% DOPG that the amount of DOPE synthesized actually dropped (or remained constant given the experimental error). The author showed only one representative experiment here and showed two repeats in Supp. Fig. 11, where only one of the two repeats had the same DOPG% and the other was very different (though a similar trend). It would be prudent to have the same experiments repeated three times to include for a main result in a manuscript.
3. There seems to be no difference between the number of vesicles with NBD signal on their membranes in the case with proteinase outside and the one with no proteinase (Supp Fig. 13). It has been previously shown that the rate of cell-free expression changes in confined volumes and varies with size of liposomes. But protein synthesis in bulk expression is much higher in yield and duration and so one might expect to see a much larger contrast in the NBD fluorescence with no proteinase in the outer CFE solution post rehydration of the lipid-coated beads. This is qualitatively shown in the bottom plot of Fig. 4b where the condition with no protease shows higher peaks for all the lipids species compared to the one with protease outside the liposomes. Was the outer CFE diluted and washed before incubation? I was also a bit surprised that the fluorescence intensity outside the liposomes in the 'no confinement' condition is not very high in Fig. 3a and 3b (the baseline between '1' and '2' in Figure 3b is not very large). I find the term 'no confinement' here may be confusing to the readers as I would think confinement here refers to encapsulation. I suggest the author change this description. Figure 3b should display a y-axis range. I am not familiar with where the active sites are for these various lipid modifying enzymes. PS is found mostly in the inner leaflet of natural cells and the asymmetry is possible by specific flippases that exposes PS to the outer leaflet. Given the detection of PS is by purified LactC2-GFP added to the outside and from the schematic cartoons shown, the authors seem to suggest that PS is made on the outer leaflet. I was surprised that there was no discussion on the asymmetry (or symmetry) of the leaflets, as this may introduce caveat in the interpretation of the results.
4. The authors noted that the amount of de novo synthesized lipids incorporated in the membrane was not sufficient to observe liposome growth. The authors should show this quantified data. The example from Fig. 5f gives the impression that the liposome grew. This data is 'hidden' in Fig. 5g where the authors displayed liposome radius in color (which to me is not a good way to present this). Since it seems like the authors have looked at many individual liposomes over time, it should be relatively straightforward to measure the size at time 0 and compared with time at 10 hours for the same liposome. This should be displayed as paired data as a separate figure panel.
Minor comments: 1. Abstract: The first statement should be reworded. Lipid, nucleic acid, and proteins are classes of macromolecule. Metabolism is a process and not an 'ingredient'.
4. When calculating NBD enrichment, quantification here be biased by the fact that not all vesicles express enzymes as shown in earlier figures. What are the criteria for selecting GVs? There is some info in the methods on this, but it might helpful to include this in the main text for the readers. There appears to be a large fraction of liposomes that don't express lipids and the authors should discuss this observation beyond this being inherent to gene expression in cell-sized compartments. Is this also the case when there is no proteinase or DNAse added outside? 5. Figure 3e. I found the color choices made the graph a bit difficult to read. It took me some time to decipher the different conditions. The magnified inset of d and e should have the y-axis scale labeled. 6. Page 6 line 30: '…….it relies also on the association-dissociation of PssA to the membrane'. Is there experimental evidence for this (from past literature or current work)? 7. Figure 3 caption: why is there heavy-isotope fraction of DOPE in the initial membrane? 8. Page 9 line 31: 'Concluding, it has been possible to…..'. This is a poor sentence, please rephrase. 9. Page 10 line 28: '…, whilst yielding a sufficiently…. ', no 'to'. 10. Have the authors investigated how much acyl-CoA is consumed over time? The authors commented on adding more than 100 µM of acyl-CoA is not possible due to low solubility. Given the groups expertise in mass spec, it might be possible to look at the consumption of acyl-CoA. I am still surprised with the addition of synthesized lipids that there is no apparent increase in liposome size.
11. There are a large number of supplemental figures. While they generally help make the case, I question whether 1) they all need to be shown, and 2) whether they support the claims made. While it is ok sometimes to show data from n = 1 experiments, I don't think this provides the level of rigorousness that would be needed, even if they are shown in supplemental figures. For instance, Supp. Fig. 4 is unimpressive, and one should really include the comparison of pGEMM7 on the same graph as a proper control. Supp. Fig. 8 needs to have other controls and repeated. Supp. Fig. 15 showed examples of linescan for 1 and 12% PS that aren't very different. There is also an issue with experimental design for Supp. Fig. 15e that looks to have missing data point and the legend is quite strange. Supp. Fig. 16c has no values and units on y-axis and shows a n = 1 experiment. Supp. Fig.  17: The authors claimed that compartmentalized PS synthesis results in a higher coefficient of variation in LacC2 signaling than when PS is directly included in vesicle membrane. The significance of this finding is unclear and not well placed in the context of this work. There are missing error bars for the eGFP sample. Why do 0% PS liposomes have LactC2 binding? Supp. Fig. 18d: The quantification shown here does not corroborate with the image shown in 18a. The fraction of LactC2-mCherry bound liposomes seems to be far less than 40%. Supp. Fig. 20 also has n = 1, even though the result is quite striking.
12. Supp. Review of NCOMMS-20-01773-T The authors present a manuscript which describes the cell-free expression of phospholipid-producing enzymes in liposomes. They can show that these enzymes are functional and convert lipid precursors into a variety of lipids, including those which constitute the liposome. The authors could confirm that this minimal cell systems can regulate a balanced production of different lipids via metabolic feedback. I was asked to comment on the technical aspects of the mass spectrometry experiments of this study. The successful expression of the enzymes in the liposomes was confirmed using a targeted proteomics approach, by tryptic digest of the peptides followed by identification via MS/MS. EF-Tu is a constant component of the PURE system and was used as a global internal standard. The proteins are partially identified by not many peptides (for example only one in the case of CdsA (P0ABG1)). But as these are membrane proteins, a lower coverage than for soluble proteins is to be expected, and the identification justified. The lipids synthesized in the liposomes were analyzed by LC-MS and could be distinguished from the originally present lipids as the precursors were labeled with 13C-G3P, resulting in a 3Da mass shift. Conversion of integrated peaks for the lipid fragments was into lipid concentrations was based on linear calibration curves obtained from dilution series of standards. One concern: The authors state "the average integrated counts of two injections was determined." Does that mean two injections of the same sample? To get reliable data, I would expect data from three independent experiments, as was done for the proteomics analysis. Apart from the point mentioned above, the proteomics and lipidomics experiments are performed in a technically sound manner, including the appropriate controls and support the conclusions of the authors, which present a study, which I consider overall a quite exciting step towards use of minimal cells for investigation of separated aspects in cell function.

Manuscript:
"Genetically controlled membrane synthesis in liposomes" by Blanken et al.  We concluded that membrane localization greatly enhances enzymatic activity for these proteins. In the present study, we showed in Fig. 2 that the lipid composition of the vesicles (% of PG content) modulates lipid production, presumably acting as an allosteric regulator of PssA and PlsB. These findings can best be explained if the involved enzymes associate with the membrane. In addition, the transmembrane nature of some enzymes, like CdsA, implies that the active protein partitions into the lipid bilayer.
In future works, it would be insightful to determine the localization of the different proteins in the membrane. Specifically, it would be relevant to determine whether all proteins are evenly distributed in the vesicle membrane or if they form functional clusters. We think, however, that such investigations go beyond the scope of the present study.

Liposome growth was not demonstrated but could have easily been observed with FRET labeled
lipids. This would strengthen the manuscript considerably since the goal appears to be growth.
Reply Both phospholipid synthesis and membrane incorporation of the output phospholipids have been unambiguously shown in this manuscript. Together, these two processes represent membrane growth. Due to the limited amount of acyl-CoA precursor that could be supplied, the total amount of synthesized lipids only resulted in a growth of ~5% of the membrane area. This value is modest relative to the 100% membrane expansion that is necessary for a full proliferation cycle, which is our long-term goal. Nonetheless, liposome growth was demonstrated.
We agree that FRET-based methods would be a powerful tool to quantify liposome growth, especially in a regime where the magnitude of membrane expansion is small. This technique has been widely used in liposome research and we are familiar with it. The outcome of such an assay will be an accurate assessment of the increase of the membrane area, which we know will range between 0% and ~10%. The knowledge of the 'exact' % increase has little added value, in our opinion, given that our next goal is to achieve substantial growth that is directly observable by standard microscopy methods, i.e. without FRET. Therefore, the quantitative results from FRET assays will not really help us improve the vesicle growth efficiency. For these reasons, we decided to not conduct FRET experiments in the context of this study.

The figures frequently lack statistical analysis and an indication of the number of replicates.
Reply For all figures in the main text, number of replicates is noted in the caption and, when appropriate, the individual data points are displayed in the graph. In the revised manuscript, Fig. 2b has been replotted to include data previously shown in Supp. Fig. 11, hence displaying the three repeats in the main text. In Reply We agree that the first sentence might be too strong if one considers primitive 'living' systems, i.e. intermediates in the evolutionary pathway from liveless molecules to contemporary organisms. However, lipid membranes, nucleic acids, proteins and metabolism are universal ingredients of modern-day life. To not exclude the possibility that protocells were of a different nature and to avoid a disputation about the essence of life, we have rewritten the first sentences to restrict our definition to the current biology: "Lipid membranes, nucleic acids, proteins, and metabolism are essential for modern cellular life. Synthetic systems emulating the fundamental properties of living cells must therefore be built upon these functional elements."

The introduction begins by stating that life is an individual construct, but that is again, likely too
strong. Evolution gives rise to populations. I know what the authors mean, but more care should be taken in how these ideas are expressed.
Reply It was not our intention to start with a provocative statement about the nature of life. It would certainly be more accurate to say that the simplest manifestation of life is in the form of individual cellular entities. To avoid ambiguity, and realising that it does not support a key message, we decided to remove this sentence.
6. Last sentence of the introduction: "Our results provide experimental evidence for DNA-encoded homeostatic growth of a liposome-based artificial cell." But the authors do not demonstrate growth in this manuscript.
Reply By demonstrating both the synthesis and membrane incorporation of bilayer-forming phospholipids, we have demonstrated vesicle growth. The fact that the increase in liposome size is below the diffraction limit of light does not mean that no growth has taken place. See also our reply to point 2. Figure 11. The two should be combined together.

It is not clear why data was split between figure 2 and Supplementary
Reply We have replotted Fig. 2 accordingly and we removed Supp. Fig. 11.

The degradation of material external to the liposomes was either with proteinase K or DNase I. Is there a reason why both weren't used together?
Reply The methods are mutually exclusive since proteinase K will degrade DNase I. Using either one of them was sufficient to shut down gene expression outside liposomes.

Line 5, page 10: It is unclear what is meant by "hidden in ensemble measurements."
Reply We modified the sentence as: "While LC-MS methods provide sensitive detection of multiple lipid species in a liposome population, information about lipid composition at the single vesicle level is lost due to vesicle solubilisation."

Supplementary Figure 4: Why are the lipid concentrations in arbitrary units?
Reply The y axis represents integrated peak intensity of 13 C-labelled lipids. It correlates with lipid concentration but it is not a direct measure of concentration. No absolute quantification was performed for this experiment, since we were interested here in relative amounts of lipids. In the revised figure, the data have been normalized to the signal of 12 C-DOPE to account for differences in MS sensitivity, and the y-axis label has been modified accordingly. Note that we have included data with pGEMM7 at the request of Reviewer #3. Reply This was also our reasoning and we accordingly modified pGEMM6 by cloning the modules in different directions, giving rise to the plasmid pGEMM7 that has been extensively used in this research. We have added the following sentence to the figure caption to clarify this point: "This finding prompted us to design the new construct pGEMM7, where the two sets of PE and PG pathway genes were cloned in opposite directions to ensure full orthogonality."

Supplementary Figure 8: without error bars, the data are hard to interpret
Reply The experiment has been performed only once. Therefore, no error bars are provided. When replacing pgpA with pgpC, the presence of synthesized DPPG could unambiguously be determined, demonstrating the activity of PgpC. We decided to remove this figure as it should be repeated for higher scientific accuracy and because it does not support a main claim.
13. Figure 3b: The difference between the dashed lines is confusing and not enough to appreciate the logic of the experiment.

Reply
We have replotted the graph with a logarithmic y-axis to more clearly display the decrease of external fluorescence upon addition of proteinase K or DNase.

Reviewer #2:
In this study, Blanken and colleagues constructed a single plasmid encoding seven genes involved in lipid synthesis and showed protein synthesis using a reconstituted transcription/translation system (PURE system) from the information. The results showed that the 7 protein involved in lipid synthesis was successfully synthesized in the liposomes, and indicated the synthesis of the objective lipids.
The direction in which this research aims is certainly important in the future of synthetic biology.
Reply We thank the reviewer for acknowledging the significance of the research area.
However, questions remain regarding the achievement, novelty of the thesis, and conceptual advances. The authors conclude in the title "membrane synthesis", but no evidence has been presented beyond lipid synthesis as the author mentioned that they failed to observe an increase of liposomes sizes by their system.

Reply
The fact that we achieved membrane synthesis is simply not questionable. We are surprised that the Reviewer missed this important aspect of our work. The results of both fluorescence-based assays (Figs. 4 and 5, and related supporting figures) unambiguously demonstrate that -at least a fraction of -the internally synthesized phospholipids are effectively incorporated to the membrane, yielding vesicle growth.
We've been very explicit throughout the article about the current limitations of our system toward effective growth from an internal phospholipid factory. As stated on page 14, lines 3-5, "the amount of de novo synthesized lipids incorporated in the membrane was not sufficient for directly observing liposome growth under an optical microscope." Further, on page 16, lines 31-32, we wrote "no visible membrane or volume expansion could unambiguously be measured by optical microscopy".
The new panel Fig. 5h is also meant to illustrate that no 'visible' increase of the vesicle radius was observed within the spatial resolution of the microscope. However, vesicle growth, be it modest, is obviously demonstrated, as commented above.
This is in contrast to reference 19 that shows this point. plasmid. What is truly novel in our work is the design and the use (not the construction per se) of a multi-cistron DNA that supports functional expression and regulation of a metabolic pathway in a cell-free system and, to a larger extent, within liposomes.

Expression and insertion to the membrane of the 7 proteins involved in lipid synthesis have been
shown in their previous study. Although the only novel point of the manuscript is the expression of the 7 proteins in liposomes, there are many reports that showed normal protein synthesis (including membrane proteins as the author already reported) via the PURE system within liposomes.
Reply PURE system expression of (membrane) proteins inside liposomes has already been achieved, including by our group. However, in the vast majority of the studies, expression of a single or two genes is reported. Several technical challenges must be overcome to express an entire catalytic pathway of several enzymes within liposomes. This is precisely what we have accomplished for the first time in this study, including also the implementation of regulatory mechanisms.
We object to the statement that "expression of the 7 proteins in liposomes […] is the only novel point of the manuscript". It is an unfair assessment of our work, in sharp contrast with the opinion of the other reviewers. Below, we summarize the key novel results.
We have previously shown the functional reconstitution of the seven lipid synthesis enzymes in the PURE system (Scott et al., 2016). Importantly, this was realized from individual DNA constructs, expressed at the outside of small unilamellar vesicles and only ensemble LC-MS data were provided (no single-vesicle resolution). Here, aided by the construction of the pGEMM7 mini-genome, we are able to recapitulate the full enzymatic pathway inside giant liposomes. By design, we equipped liposomes with genetic control over cell-free expression of the branched enzymatic pathway (Fig. 1).
All these aspects represent major conceptual advances towards an autonomously growing minimal cell. In addition, we show for the first time (i) in vitro reconstitution of the membrane-mediated control of PE vs. PG synthesis (Fig. 2), (ii) the one-pot synthesis of six lipid species (Fig. 3), (iii) the use of NBD-labelled precursors for cell-free synthesis of artificial NBD-labelled phospholipids (Fig. 4), (iv) the use of the lactC2-eGFP probe (Fig. 5). The latter two assays provide the first evidence of lipid synthesis and membrane incorporation on the single vesicle level by fluorescence microscopy and reveal liposome-to-liposome variability. Finally, the lactC2 probe allows us to measure the incorporation kinetics of newly synthesized PS (Fig. 5).
These novel points are therefore not only technical, but conceptual in nature and we are surprised to notice that they have been overlooked by the Reviewer.
Due to the levels of achievement they set and lack of significant novelty and conceptual advances, I do not support the current manuscript for publishing in Nature Communication.
Reply We hope that our point-by-point responses provide the necessary clarifications and a better positioning of our study.
Other major comments L2, p3. Figure 5 shows synthesis of PS within GUVs. However, PS is also not part of the original membrane composition of E. coli system.
Reply PS is an intermediate phospholipid species in the E. coli membrane. Accumulating PS by silencing expression of Psd was a necessary modification of the system to make use of the LactC2 probe. We have also demonstrated lipid synthesis and membrane incorporation of PE and PG by employing NBD-labelled lipid precursors (Fig. 4), which did not require a modification of the enzymatic pathway. Moreover, the LC-MS data reported in Figs. 2 and 3 show successful production of the two output lipids PE and PG when the pathway is fully activated.

L9, p3. No evidence of the growth of liposome is shown in this manuscript.
Reply This is not correct. Both phospholipid synthesis and membrane incorporation of the resultant phospholipids have been unambiguously shown in this manuscript. These two processes together represent membrane growth. Please, refer to page 7 of this letter for a more elaborate argumentation.
L27, p5. I think normalization using a different peptide fragment is a non-standard method in MSbased quantification. Do the authors have a reference or data that support the feasibility of the normalization?
Reply Using the same peptide fragment is only required for absolute quantification, which is not the purpose of this experiment. Reviewer #4, who was specifically asked to review our mass spectrometry methods, states that "the proteomics and lipidomics experiments are performed in a technically sound manner, including the appropriate controls and support the conclusions of the authors". Reviewer #4 also explicitly acknowledges the use of EF-Tu as an internal global standard. Reply In both Figs. 3 and 4, we observe a decrease in total lipid synthesis activity when gene expression is exclusively confined to the interior of liposomes by addition of protease or DNAse. We do not attribute this decrease to any deleterious effect of protease or DNase on internal gene expression, as suggested by the Reviewer. Inhibiting protein synthesis outside liposomes strongly reduces the total volume in which lipid production takes place. If one takes into account this drastic reduction in volume, the decrease in enzymatic activity measured at the population level is actually surprisingly low. On page 11, lines 30-33, we added: "the moderate increase (~50%) of the fraction of NBD-enriched liposomes when omitting the proteinase K (Fig. 4d) might be explained by an enhancement of enzymatic activity in liposome-confined reactions, as suggested above for lipid production at the population level (Fig. 3d,e)."

References
The authors should cite pioneer studies that use cell extract system within GUVs On page 25, lines 31-32: "Within data sets, identical imaging settings were used." On page 9, line 8: "[…] the same size and were acquired with identical imaging settings." On page 15, line 28: "Identical imaging settings were used for all acquired data." Purity of LactC2-eGFP It seems too low purity to conclude something. I strongly suggest to add some data using more purified LactC2-eGFP through further purification steps.
Reply Firstly, we quantified the level of nonspecific binding at different concentrations of purified LactC2-eGFP and we used reference liposomes with different amounts of PS for calibration (Supp. Fig. 15). From these experiments, we found the optimal concentration of LactC2-eGFP to specifically report PS exposed to the vesicle membrane. Note that the concentration of LactC2-eGFP was also determined by absorbance measurement of eGFP. Moreover, our results were confirmed using different batches of purified LactC2-eGFP and a different protein fusion (LactC2-mCherry, Supp. Figs. purity of LactC2-eGFP as shown in the lane 'BE' (buffer exchange) of Supp. Fig. 12a, is sufficient, although a relative low amount of protein was loaded onto this gel.
Regarding the presence of two to three prominent bands of the LactC2-mCherry fusion protein observed in Supp. Fig. 12b Additionally, we analyzed these LactC2-mCherry samples using anion exchange chromatography   Minor comments L2, p2. Please refer some reference.
Reply Reviewer #1 commented that the statement was too strong and we decided to remove this sentence.

This is indeed a major advance in the field and could be an impactful approach for possible growth and division of synthetic cell-like liposomes (although growth has not been demonstrated here).
Reply We thank the reviewer for her/his accurate summary of our work and for recognizing that it represents a major advance in the field. We highly appreciate the thorough review and extensive comments that helped us clarify several points in the revised manuscript.

Overall, the manuscript is well written and organized. The work is quite relevant in the field of in vitro lipid synthesis and use of cell-free expression as a tool for reconstitution of enzymatic pathways and membrane proteins.
Reply Thank you for pointing out the relevant and novel aspects of our studies.
There are however certain limitations of the presented approach that need to be elucidated to provide the full scope of the work to the readers. My major concerns are as follows.   Fig. 2b  Reply PE synthesis is downregulated by the high PE amount in the membrane, irrespective of the presence of a competing pathway. We stated in the main text "Interestingly, PE synthesis was reduced at low PG content, independent of the expression of the PG-synthesizing pathway branch (Fig. 2c, Supp. Fig. 11). This result indicates that the regulatory mechanism is not solely driven by competition between the two pathway branches but it relies also on the association-dissociation of PssA to the membrane". We think that the level of discussion is sufficient given the claims we make and the reported data sets. We'd rather avoid speculating about putative mechanisms that lack in vivo confirmation anyways.

The authors should also comment that at 50% DOPG that the amount of DOPE synthesized actually dropped (or remained constant given the experimental error). The author showed only one
representative experiment here and showed two repeats in Supp. Fig. 11, where only one of the two

repeats had the same DOPG% and the other was very different (though a similar trend). It would be prudent to have the same experiments repeated three times to include for a main result in a manuscript.
Reply We have replotted the main text figure to include all three experiments (new Fig. 2b-d). The fact that in one experiment the input PG% values are slightly different is not important as we pooled all data without calculating mean ± sdv values. Note that PG% values are measured parameters and are therefore also subject to experimental variability. We've been careful to base our statements on obvious trends of the data: (i) the concentration of synthesized PE increases with the proportion of input PG with and without SP6 RNAP (new Fig. 2b), (ii) the concentration of synthesized PG decreases as the % of input PG increases when the PG branch is activated (new Fig. 2c), and (iii) the total amount of synthesized lipids (PE+PG) increases as the % of input PG increases (new Fig. 2c).
Our data do not provide evidences that the amount of synthesized PE drops or remains constant at 50% input PG. The aggregated data clarify this point.

There seems to be no difference between the number of vesicles with NBD signal on their
membranes in the case with proteinase outside and the one with no proteinase (Supp Fig. 13) Fig. 4b where the condition with no protease shows higher peaks for all the lipids species compared to the one with protease outside the liposomes.

. It has been previously shown that the rate of cell-free expression changes in confined volumes and varies with size of liposomes. But protein synthesis in bulk expression is much higher in yield and duration and so one might expect to see a much larger contrast in the NBD fluorescence with no proteinase in the outer CFE solution post rehydration of the lipid-coated beads. This is qualitatively shown in the bottom plot of
Reply We would like to start by clarifying some of the reviewer's statements. Protein synthesis with PUREfrex2.0 in bulk expression is not much higher in duration and concentration compared to in-liposome reactions; see our study cited as ref 38, where a similar protocol for gene expressing liposomes was applied.
In Supp. Fig. 13 (now Supp. Fig. 11), the number of liposomes in the field of view is not large enough to conclude that NBD-positive liposomes is the same in both conditions, especially if one considers background signal from the NBD-substrate and liposome-to-liposome variability. It can be noticed that the fraction of NBD-enriched liposomes is higher with no proteinase compared to proteinase added outside (Fig. 4d), but not largely higher as one could expect given the much larger volume outside liposomes. Similarly, Fig. 3d,e shows that the concentration of synthesized lipids is higher with no proteinase, but not largely higher, as measured by LC-MS for six different lipid species. The latter observation has been discussed on page 8 lines 13-15. We attribute this effect to an enhancement of enzymatic activity inside the confined environment of the GUV, in line with the reviewer's remark that "the rate of cell-free expression changes in confined volumes". Two citations supporting this hypothesis have been added (i.e., refs. 4 and 38). The same hypothetical scenario applies for Fig. 4d, but we forgot to specify this. In the revised manuscript, on page 11, lines 30-33, we clarified this point by adding: "In addition, the moderate increase (~50%) of the fraction of NBDenriched liposomes when omitting the proteinase K (Fig. 4d) might be explained by an enhancement of enzymatic activity in liposome-confined reactions, as suggested above for lipid production at the population level (Fig. 3d,e)."

Was the outer CFE diluted and washed before incubation?
Reply On page 11, lines 22-23, we wrote: "After pGEMM7 expression, the liposomes were diluted to reduce the membrane signal coming from NBD-palmitoyl-CoA and NBD-LPA." As detailed in the Methods section, the sample of surface-immobilized liposomes was washed three times with buffer E to remove non-reacted NBD-palmitoyl-CoA. For HPLC (Fig. 4b), the (NBD-labelled) lipid fraction was extracted using methanol, without any washing steps. The corresponding protocol was also mentioned in the Methods section. For clarity, we have modified the caption of Fig. 4d on page 13, lines 1-2, as: "… in b (bottom). The samples were washed three times to remove non-reacted NBDpalmitoyl-CoA." Fig. 3a and 3b (the baseline between '1' and '2' in Figure 3b is not very large).

I was also a bit surprised that the fluorescence intensity outside the liposomes in the 'no confinement' condition is not very high in
Reply In this particular experiment, liposomes were diluted (2 μL in 7.5 μL total) to reduce their surface density and aid visualization. This was not clearly stated in the paper. Therefore, we added this information in the caption of Fig. 3a. We have previously observed that YFP expression inside individual GUVs can be significantly higher than in the surrounding bulk (see ref. 38) The fluorescence contrast is more pronounced here as the sample was diluted. Such an observation was also reported in the pioneer work of Nomura et al. (new ref. 4). These results further support a scenario where liposome-compartmentalized gene expression is enhanced compared to bulk reactions.

I find the term 'no confinement' here may be confusing to the readers as I would think confinement
here refers to encapsulation. I suggest the author change this description.
Reply We agree. We changed the term 'no confinement' in Fig. 3a and Supp. Fig. 10  Reply In Fig. 1a, the cartoon illustrates lipid production occurring from the outside of the liposomes because the PURE system was mixed with preformed SUVs. In Figs. 3c, 4a and 5a, the cartoons illustrate enzyme and lipid production from the inside of GUVs, with one of the precursors, the acyl-CoA, being supplied from the outside of the liposomes. In the latter configuration, we presume that a fraction of PS partitions in the outer leaflet where it is exposed to the LactC2 probe.
However, this does not necessarily imply that PS is synthesized on the outer leaflet and we did not intend to suggest so. Given that the PS-producing enzyme, PssA, is a membrane-associated protein that exists in both the membrane-bound and free states (Fig. 2a), it is more likely that synthesized PS is released from the active site into the inner leaflet. Subsequently, PS may flip to the outer leaflet, a process that is not energetically favourable and requires assistance of specialized enzymes in vivo.
However, the artificial bilayer of our liposomes is not as rigid and is more prone to transient defects compared to cell membranes (refs. 38 and 42). Therefore, membrane dynamic processes such as lipid flip-flop and translocation of small molecules may be less impaired in liposomes than in cells, facilitating partitioning of PS in the outer leaflet, and possibly of other synthesized lipids too. These differences in the membrane properties may also explain why the bilayer-spanning enzymes of the Kennedy pathway spontaneously insert in the membrane without an active machinery. Note also that we cannot rule out the possibility that LactC2-GFP permeates across the membrane and binds PS exposed to the lumen. We've been prudent to not comment on this aspect given the level of speculation. However, in the light of the reviewer's comment, we acknowledge that it requires some discussion which we included in the revised manuscript (see next comment).

I was surprised that there was no discussion on the asymmetry (or symmetry) of the leaflets, as this may introduce caveat in the interpretation of the results.
Reply Following on the previous comment, the question of the partitioning of the synthesized lipids in the inner and outer leaflets of the liposomes is highly relevant and deserves more attention in the Discussion. We've been prudent in the interpretation of our results. For instance, we assigned LactC2 membrane recruitment to a 'PS enrichment', without specifying its location. Fig. 5a illustrates the scenario where the externally supplied LactC2 binds to PS present in the outer leaflet. In our opinion, this is the most probable scenario. Nonetheless, we decided to provide a more complete discussion on page 16, lines 4-12.
"What is the distribution of internally produced lipids in the bilayer? Phosphatidylserine is likely synthesized on the inner leaflet of the liposome membrane [41]. Nevertheless, synthesized PS is detected on the outer leaflet, where it is exposed to the LactC2-eGFP probe. Flipping of phospholipids is not energetically favourable and requires assistance of specialized enzymes in vivo.
However, the artificial bilayer of our liposomes is not as rigid as the bacterial cell membrane and is more prone to transient defects [38,42]. Therefore, membrane dynamic processes, such as

The authors noted that the amount of de novo synthesized lipids incorporated in the membrane
was not sufficient to observe liposome growth. The authors should show this quantified data. The example from Fig. 5f gives the impression that the liposome grew. This data is 'hidden' in Fig. 5g where the authors displayed liposome radius in color (which to me is not a good way to present this).
Since it seems like the authors have looked at many individual liposomes over time, it should be relatively straightforward to measure the size at time 0 and compared with time at 10 hours for the same liposome. This should be displayed as paired data as a separate figure panel.
Reply We agree that our statement that "no visible membrane or volume expansion could unambiguously be measured by optical microscopy" should be supported by quantified data. As suggested, we have included a new panel in figure 5 (Fig. 5h), which displays the change in apparent radius between the start and end points of the kinetics reported in Fig. 5g. No significant increase in apparent radius could be observed. Moreover, we have extended the analysis shown in Supp. Fig. 19 to investigate the relationship between change in apparent radius and reaction rate, plateau time, and liposome size (Supp. Fig. 19e,f). No clear correlations were observed.
We added in the caption of Reply We rephrased as: "To establish a link between the lipid compartment and its internal content, liposome growth could be made conditional to encapsulated nucleic acids [12,16] or catalysts [17]."

Page 3 line 26: Enzymes downstream 'of' the pathway are….
Reply We changed as "Enzymes downstream in the pathway".
4. When calculating NBD enrichment, quantification here be biased by the fact that not all vesicles express enzymes as shown in earlier figures. What are the criteria for selecting GVs? There is some info in the methods on this, but it might helpful to include this in the main text for the readers.
Reply The Methods section has been moved from the supplemental information to the main text.
Moreover, we have added the following sentence on page 12, lines 21-22: "Liposomes were selected for unilamellarity based on the Texas Red channel."

There appears to be a large fraction of liposomes that don't express lipids and the authors should discuss this observation beyond this being inherent to gene expression in cell-sized compartments. Is this also the case when there is no proteinase or DNAse added outside?
Reply Non-expressing liposomes are also observed in the absence of proteinase or DNase (Fig. 3a) in the case of YFP expression. Supp. Fig. 11a (right panel) shows liposomes with no (at least no more than in negative control samples) NBD signal when gene expression was enabled inside and outside the vesicles. We complemented the discussion on page 16, lines 26-29, to account for other mechanisms that may contribute to the observed heterogeneity in lipid expression: "In the present experiments, other sources of heterogeneity in lipid enrichment may also contribute, such as a variability in the adsorption of acyl-CoA among liposomes upon resuspension of the precursor film.
Investigating the mechanisms leading to phenotypic differences will be important to further optimise the chain of reactions from genes to output lipids." 5. Figure 3e. I found the color choices made the graph a bit difficult to read. It took me some time to decipher the different conditions.
Reply We took a color map that is designed to provide as much contrast as possible, also for color blind people (https://cran.r-project.org/web/packages/viridis/vignettes/intro-to-viridis.html). We tried several options to plot the data from six different lipids in eight different conditions with three repeats, and we think this is the best we can do.

The magnified inset of d and e should have the y-axis scale labeled.
Reply The insets of Fig. 3d,e have been modified accordingly.
6. Page 6 line 30: '…….it relies also on the association-dissociation of PssA to the membrane'. Is there experimental evidence for this (from past literature or current work)?
Reply The transient association of PssA with the membrane is conditional to the presence of anionic lipids, as described in refs. 30-32. We have cited again these papers on page 6, line 32.

Figure 3 caption: why is there heavy-isotope fraction of DOPE in the initial membrane?
Reply In nature, various isotopes coexist for almost all elements. For example, 1.1% of carbon in nature is 13 C, having a mass 1 Da higher than the most prevalent 12 C. Because of this, molecules that are slightly heavier than calculated from the standard elemental weights are a common occurrence.
This becomes obvious when trying to distinguish between regular DOPE and 13 C-labelled DOPE, which is only 3 Da heavier. Apparently, ~1% of naturally occurring DOPE is 3 Da heavier than the atomic weight of 743.5 Da, and is therefore detected when scanning for the isotopically labelled 13 C-DOPE. For DOPG, two molecules of 13 C-labelled G3P are incorporated, causing a mass difference of 6 Da between the light DOPG and our reaction product. Apparently, DOPG with a 6-Da mass shift does not occur in nature in detectable amounts.
For clarity, we slightly rephrased the corresponding text on page 10, lines 6-8, as: "A small amount of DOPE was measured in samples where no acyl-CoA was supplied. This represents the naturally occurring heavy-isotope fraction of the DOPE contained in the initial liposome membrane." Reply We changed it.

Have the authors investigated how much acyl-CoA is consumed over time? The authors
commented on adding more than 100 µM of acyl-CoA is not possible due to low solubility. Given the groups expertise in mass spec, it might be possible to look at the consumption of acyl-CoA.
Reply We found that transitions of 514.6 → 78.9 and 1030.3 → 407.9 gave clear peaks for oleoyl-CoA. We provide below a chromatogram displaying the total ion count (black, major peak) and the two transitions (lower peaks), measured on a standard of pure oleoyl-CoA. The retention time is 1.4 min, indicating limited adsorption on the column. We have performed a repeat of the experiment shown in Fig. 2b, where we investigated the amount of oleoyl-CoA after synthesis. Compared to a sample without lipid synthesis, we found that between 1% (0% PG, +SP6 RNAP) and 22% (50% PG, +SP6 RNAP) of input oleoyl-CoA remained unreacted. This result indicates that a significant fraction of oleoyl-CoA was consumed.

I am still surprised with the addition of synthesized lipids that there is no apparent increase in liposome size.
Reply When preparing GUVs as shown in Fig. 5, we add 1 mg of lipid-coated beads per 1 µL of PURE system solution. Every gram of beads contains 3.33 mg of lipids, or equivalently, 4.24 µmol of lipids. So, our liposome solution has an initial lipid concentration of 3.33 µg/µL or 4.24 nmol/µL, corresponding to a lipid concentration of 4.24 mM. Prior to the microscopy experiments shown in In the ideal case, where all oleoyl-CoA (100 µM) is consumed, 50 µM of lipids are synthesized. This represents an increase in lipid concentration of (1.13 mM + 0.05 mM)/1.13 mM = 1.045, so 4.5 %.
Assuming that i) lipid synthesis is homogenously distributed, ii) all supplied lipids assemble into liposomes, and iii) all lipid species in the membrane have the same molecular surface area, the area of liposomes is also increased by 4.5%. Therefore, the diameter, which scales with the square root of the area, increases by 2.2%. For an average liposome diameter of 4 µm, this means a diameter increase of 88 nm, which is well below the detection limit of our light microscope.
Let's assume now that only 50% of the liposomes synthesize lipids and that all supplied oleoyl-CoA feeds the active liposomes. Then, a surface area increase of 9% is expected, so a diameter increase of 4.4%. For our 4 µm diameter liposomes, this results in a growth of 176 nm, still less than one pixel of 250 nm.
For the sake of the argument, one could calculate the increase in lipid concentration needed to grow For instance, Supp. Fig. 4 is unimpressive, and one should really include the comparison of pGEMM7 on the same graph as a proper control.
Reply Although the data may not look 'impressive', they prompted us to rethink and change our plasmid design. We therefore think it is useful to show it. We agree that including pGEMM7 data will aid interpretation of the result, and have therefore included the 0 AU and 4 AU SP6 RNAP data from Fig. 1c in the revised Supp. Fig. 4. We modified the caption as: "b, Synthesis of phospholipids from pGEMM6 (data with pGEMM7 are appended for comparison), […] Bars are average values from two independent repeats (three with pGEMM7), each represented by a different symbol." Supp. Fig. 8 needs to have other controls and repeated.
Reply We removed this figure.
Supp. Fig. 15 showed examples of linescan for 1 and 12% PS that aren't very different.
Reply We agree with this observation. As can be seen at the population level, the mean GFP rim intensity shown in the box plot aren't that different for 1% and 12% PS. This observation, combined with similar results for LactC2-mCherry (Supp. Fig. 18c), indicates that the relationship between the concentration of PS and LactC2-probe fluorescence intensity is nonlinear. In addition, it can be seen in Supp. Fig. 15c that the histograms of LactC2-eGFP fluorescence at 1% and 12% PS significantly overlap. We've been prudent to report the fraction of PS-enriched liposomes instead of absolute fluorescence of LactC2-eGFP/mCherry as a measure of PS synthesis.
To bring this observation to the attention of the reader, we added in the caption of Supp. Fig. 15c,d: "The distributions of LactC2-eGFP fluorescence at 1% and 12% PS significantly overlap.
[…] The relationship between LactC2-eGFP fluorescence intensity and PS concentration is nonlinear." There is also an issue with experimental design for Supp. Fig. 15e that looks to have missing data point and the legend is quite strange.
Reply No data points are missing from the graph, but it is true that the legend layout was confusing. We have clarified it.
Supp. Fig. 16c has no values and units on y-axis and shows a n = 1 experiment.
We have added y-axis labels.
Some of the experiments presented in Supp. Fig. 16c have been repeated multiple times, e.g. with linear pGEMM (+EcoR1). Herein, we only report data from a single experiment, where all conditions have been tested at once to minimize technical variability. The obtained results are very clear: PS only accumulates when the psd gene is inactivated, and no major difference is measured between the three strategies to silence gene expression outside liposomes. Further quantitation of the data was not necessary.
Supp. Fig. 17: The authors claimed that compartmentalized PS synthesis results in a higher coefficient of variation in LacC2 signaling than when PS is directly included in vesicle membrane. The significance of this finding is unclear and not well placed in the context of this work.
Reply This finding is significant because it reveals the heterogeneous nature of lipid production, which is an important property of the vesicle population.
This statistical analysis is relevant due to the inherent variability of LactC2-probe fluorescence intensity even in liposomes with predetermined amounts of purified PS (Supp. Fig. 15c,d). To be more explicit on the relevance of the coefficient of variation, we modified the text on page 13, lines 22-26 : "A wide distribution of eGFP intensity values in PS-synthesizing liposomes was measured (Fig.   5d). The coefficient of variation is ~2-fold higher than in control samples with a predetermined fraction of PS ( Supplementary Fig. 17). This result further supports the highly heterogeneous nature of liposome-encapsulated lipid synthesis." In addition, we have added the following sentence in the caption of Supp. Fig. 17: "The coefficient of variation provides a good measure of the excess liposome-to-liposome heterogeneity of membrane-incorporated PS caused by cell-free protein and lipid synthesis." We have clarified the significance of this finding by adding the following text in the Discussion on page 16, lines 26-29: "In the present experiments, other sources of heterogeneity in lipid enrichment may also contribute, such as a variability in the adsorption of acyl-CoA among liposomes upon resuspension of the precursor film. Investigating the mechanisms leading to phenotypic differences will be important to further optimise the chain of reactions from genes to output lipids." There are missing error bars for the eGFP sample.
Reply We wrote in the caption: "No error bar indicates that the experiment was performed once." We modified as: "Calibration experiments with LactC2-eGFP have been performed once". Two to three independent experiments were already conducted with LactC2-mCherry. We considered that one experiment with LactC2-eGFP was sufficient here because the two probes behave similarly, as shown in Supp. Figs. 15 and 18.

Why do 0% PS liposomes have LactC2 binding?
Reply Unspecific binding of membrane probes is a commonly observed phenomenon and is more prevalent at higher LactC2 concentrations. The working concentration of LactC2 probe should be chosen to minimize unspecific binding and provide a high signal-to-noise ratio. Thus, it must be high enough to ensure good signal, but not too high to specifically report on PS. In the caption of Supp. Supp. Fig. 18d: The quantification shown here does not corroborate with the image shown in 18a.
The fraction of LactC2-mCherry bound liposomes seems to be far less than 40%.
Reply We have replaced the panels shown in Supp. Fig. 18a,b to display a more representative field of view.
Supp. Fig. 20 also has n = 1, even though the result is quite striking.
Reply We decided to remove this figure, see previous point on page 27 of this letter.

Supp. Fig. 5: There is discrepancy between the caption and what is indicated in the main text (Page 5 line 16). Is this using E. coli cell-free expression or by PURE?
Reply The experiments displayed in the original Supp. Fig. 5  Reply Only one or two experiments were conducted, which is not enough to draw solid conclusions. We decided to remove this figure, see previous point on page 27 of this letter.
We appreciate the reviewer's efforts to provide extensive comments and constructive feedback.
But as these are membrane proteins, a lower coverage than for soluble proteins is to be expected, and the identification justified.
The lipids synthesized in the liposomes were analyzed by LC-MS and could be distinguished from the originally present lipids as the precursors were labeled with 13C-G3P, resulting in a 3Da mass shift.

Conversion of integrated peaks for the lipid fragments was into lipid concentrations was based on
linear calibration curves obtained from dilution series of standards.
Reply We thank the reviewer for the detailed summary and careful evaluation of our mass spectrometry methods.
One concern: The authors state "the average integrated counts of two injections was determined." Does that mean two injections of the same sample? To get reliable data, I would expect data from three independent experiments, as was done for the proteomics analysis.
Reply We refer here to two injections of the same sample. For each condition, three biological repeats (independent experiments) were performed, as mentioned in the captions of Figs. 1,2 and 3.
Elaborating, for every biological replicate, we have performed two injections, the average of which was used in further analysis. We opted for this approach since we noticed that carry-over from previous injections could cause biased lipid concentration measurements. By twice injecting the samples in random order, we have attempted to mitigate this effect. Furthermore, we sometimes observed a minor decrease in MS sensitivity during the course of our measurements. Randomly injecting twice, and measuring a standard curve before and after the samples for quantification, was applied to average out bias introduced by this.
Apart from the point mentioned above, the proteomics and lipidomics experiments are performed in a technically sound manner, including the appropriate controls and support the conclusions of the authors, which present a study, which I consider overall a quite exciting step towards use of minimal cells for investigation of separated aspects in cell function.
Reply Thank you for the positive comment.
Manuscript: "Genetically controlled membrane synthesis in liposomes" by Blanken et al. I think significance of the paper depends on the conclusion of whether membrane synthesis is actually achieved by proteins expressed within liposomes. Hence, this point should be supported by unquestionable data. Reply As stated by Reviewer #3, 'phospholipid synthesis and membrane incorporation are unambiguously shown'.
Major point: Membrane synthesis Although the authors claimed that the method the authors use (assessments by binding of fluorescent proteins) is enough to show insertion of a significant fraction of lipids in the membranes, their optical microscope observation that does not have a spatial resolution (~1 um) cannot provide evidence on a lipid bilayer structure with a thickness of about 5 nm. Therefore, at this stage, it has not been shown whether the fluorescence is derived from a membrane (lipid bilayer embedded in liposomes) structure or not, and cannot deny a possibility of other lipid structures such as a micelle on membranes. LC-MS can demonstrate the existence of synthesized phospholipids, but not show the existence of bilayer as the authors also described in the response.
Reply First, the spatial resolution of our microscope is ~250 nm, not 1 µm. Second, PS lipids are not stable in micellar structures. Because of their cylinder shape, they will preferentially form bilayers. In the absence of more solid arguments supporting an alternative mechanism, our observations are best explained by membrane synthesis through enzymes expressed within liposomes.
As an additional point of argumentation, the LactC2-eGFP probe has been tested on liposomes where PS was included as part of the liposome membrane ( Supplementary Fig. 15). This resulted in similar colocalization characteristics as when PS was synthesized, which supports our claim of membrane incorporation. Furthermore, the NBD experiments in the present study and the results reported for synthesized DPPA in Fig. 4 and Supplementary Fig. 3 of our previous research (ref. 23) support membrane incorporation.
Reviewer 1 also suggested a method like FRET, but the authors declined the suggestion due to technical problems. An additional experiment to support the conclusion of membrane synthesis is necessary. The authors should consider other demonstration experiments to show membrane synthesis such as time-lapse observation using super-resolution microscopes or clear visualization of lipid structures by electron microscopes.
Reply This is incorrect, we have not declined to perform this experiment because of 'technical problems'. In our reply to Reviewer #1, we wrote: "The knowledge of the 'exact' % increase has little added value." Transmission electron microscopy is not a suitable technique for >400-nm liposomes and the crowding protein environment (in particular ribosomes and some translation factors) will complicate direct visualization of lipid structures. Super-resolution fluorescence imaging will suffer from thermal membrane fluctuations. We are not saying that these two techniques are not relevant for future experiments, but it will be extremely challenging to reliably measure such a minute change of membrane surface area or liposome diameter. Anyways, as mentioned above, we do not think that additional experiments are necessary to strengthen our conclusion about membrane synthesis.
In any case, at the current manuscript, the synthesis of membranes are not "obviously demonstrated" as the author insisted in the responses.
Other points: i) Regarding the response to Reviewer1 The authors argue that the expression inside liposomes is a remarkable point compared to previous studies (I agree about this point), but however, the authors simultaneously cite the results of expression from the outside in previous studies as an evidence to show the correct insertion of the membrane proteins synthesized. There is no guarantee that the same thing is happening in the cases of expression from outside and inside. This point also suggests that the author's conclusions are not fully supported.
Reply Membrane insertion or association of the involved enzymes is a prerequisite for their functions. Phospholipid synthesis is demonstrated; hence, functional incorporation of the membrane proteins is indirectly demonstrated too. We do not claim that all expressed proteins are active and correctly inserted. However, a fraction definitely is, both when gene expression occurs inside or outside vesicles.
A possible difference in the efficiency of membrane incorporation between inside and outside expression, could result from a difference in membrane curvature between the LUVs (for which direct evidence of membrane protein insertion was provided, see ref. The revision by Blanken et al is well carried out and the modifications in the text made the content more clear (except for one place that I would suggest revising). Although there are questions that remain, I am in support of publication of this work in Nature Communications.
Reply We thank the reviewer for his/her positive opinion and for providing thoughtful suggestions.
With respect to the rebuttal letter, I do want to comment on the authors' response regarding demonstration of growth. All three technical content reviewers had commented on the insufficient demonstration of membrane growth. I think this was an important point in my first review, and I am ambivalent about the authors' response to this point. I agree that phospholipid synthesis and membrane incorporation are unambiguously shown. I think these are the hard data and should be interpreted as such. Evidence for growth, as presented, was not strong enough. Growth is a physical effect and carries a different meaning. Taking a living cell as an example (human would be a good one as well). Lipid synthesis and incorporation would be necessary for growth, but not sufficient. There are processes in cells that remove membrane and there is also a requirement to add volume for growth. I think most people would agree that a cell without physical growth is not considered 'growing', even if lipid synthesis and incorporation are taking place. I am not necessarily saying this argument applies to the current in vitro system, but growth needs to be substantiated by data (e.g. like the FRET expt reviewer 1 suggested).

Reply
We thank the reviewer for supporting our main claim that internal phospholipid synthesis and membrane incorporation are unambiguously shown. We understand the reviewer's viewpoint about the notion of 'physical growth' and decided to tone down the related claim in the revised version. Specifically, we modified the manuscript text below: "Strategies are discussed to alleviate current limitations toward more effective liposome growth and self-reproduction." (page 1, line 25) "Our results provide experimental evidence for DNA-encoded homeostatic growth membrane synthesis in a liposome-based artificial cell." (page 3, line 11) Synthesis of phospholipids from an internal machinery and their incorporation in the lipid bilayer inevitably results in liposome are essential steps toward physical growth. (page 17, line 4) I appreciate the back of envelope calculation by the authors to estimate consumption of oleoyl-CoA and used this to make the argument that vesicle growth would be sub-200 nm. One caveat of this analysis is that it depends on the initial lipid concentration. Even when I first read this manuscript, I was struck by the density of the liposomes appeared in the images. The author's ~5% lipid synthesis can be drastically increased by a lower initial lipid concentration (not oleoyl-CoA) -in essence by optimizing the starting conditions of the experiment.
Reply As the reviewer indicates, the relative growth is not only dependent on the amount of lipid synthesized, but also on the initial amount of phospholipid present as part of the liposome membrane. Therefore, lowering the liposome concentration might boost the magnitude of growth. This, however, has two main drawbacks.
• Practically, the current high vesicle density allows us to obtain measurements of many liposomes within a single experiment with a relatively low-throughput method as confocal microscopy. This has been paramount in our analysis of, for example, PS-enrichment (Fig. 4d,  Fig. 4e) and of the coefficient of variation ( Supplementary Fig. 17). Moreover, it is the best way to address the high liposome-to-liposome variability. Our previous observation that a small subset of liposomes exhibits enhanced gene expression phenotypes (ref. 38) encouraged us to perform high-content liposome measurements, to not miss potentially relevant lipid synthesis phenotypes. • Increasing the ratio of oleoyl-CoA:phospholipids by diluting liposomes before supplying oleoyl-CoA will be limited by the stability of liposomes because oleoyl-CoA acts as a detergent. A good compromise would be "to provide a continuous supply of low-concentration acyl-CoA", as we suggested on page 17, line 21.
I am satisfied with the authors' added quantified data to compare liposome sizes. In my second read of the manuscript, I did not feel there was an emphasis on demonstrating membrane growth. While this remains a weakness of the work, I am fine with it.
Reply Thank you.
My only other comment is related to the discussion on enzyme localization/membrane asymmetry. The additional text is fine with the exception of stating the possibility of LactC2-GFP permeating across the membrane. This is a pretty small protein and by stating this, what prevents other small proteins in the system from going in and out of the membrane?
Reply Translocation of the LactC2-GFP across the liposome membrane is a plausible scenario that we cannot totally exclude. We have shown permeation of the externally supplied low-molecular weight Cy5 molecule (Ref. 38) and of GFP (Ref. 42). In the latter study, we extensively discussed possible mechanisms in the supplemental information, section "3. Possible mechanisms for membrane permeability" and supplementary figures 5 & 6. In particular, we hypothesized that different reactional components in vesicles are engaged in functional multiprotein/nucleic acid complexes that are incapable of passing through the bilayer defects, even if their molecular weight is lower than that of GFP. This assumption is supported by experiments in continuous cell-free translation devices based on semipermeable ultrafiltration membrane with pore sizes (typical molecular mass cutoff is 30 kDa) theoretically large enough for the diffusion of smaller translation factors [*]. Although the vesicle experiments in Ref. 42 were conducted with a different lipid composition and a slightly different protocol for liposome formation, we believe that such considerations might be valid in the present study too.
[ To me, suggesting this possibility is more problematic in data interpretation. While I do not think there is data on which leaflet PS is produced in and how it may end up in the outer leaflet, the result presented here is that PS is detected on the outer leaflet. I believe there is literature out there that show enzyme-free lipid flipper under certain conditions and I cannot recall the details to know if it is applicable here. I would suggest that the author remove the statement on the possibility of LactC2-GFP permeates across the membrane as I don't think it is a sensible possibility in this context.

Reply
We have been prudent in our data interpretation. Even if LactC2-GFP were to cross the liposome membrane and bind to PS on the inner leaflet, none of our conclusions would be affected.
The main claim we make is that PS is produced from internally synthesized enzymes and that it incorporates in the liposome membrane, a process that can be imaged at the single-vesicle level with the LactC2-GFP probe. For the reasons mentioned above, and to give the reader all the elements to understand the experimental outcomes, we would like to keep the sentence on the possibility that LactC2-GFP may cross the membrane.
Minor comments: -pg 6, line 9: '…prior and posterior to data…' change to '….prior to and post data….' Reply We have modified the main text accordingly.
- Figure 4d: I am still struck by how low the percentage of enriched liposomes are….less than 20% in the case of no protease compared to 5% in the case of protease…..something to think about.

Reply
We agree that the difference in percentage of enriched liposomes as determined with the NBD assay is not spectacular. This is a result of the rather high background signal due to incorporation of unreacted NBD-p-CoA in the membrane. Even after washing, membranes in the negative control exhibit significant NBD fluorescence (~500 a.u., Supplementary Fig. 11). This necessitates the use of a rather stringent threshold to exclude false positives. This has the detrimental effect of likely underestimating the 'real' number of PS-enriched liposomes. Therefore, we complemented the NBD-based assay by the LactC2-based assay, which was found to be more specific and had lower background signal, at the concentration used. This allowed us to place a more accurate threshold, which revealed that in fact, >50% of liposomes are PS-enriched (Fig. 5e). This number is more in line with expectations based on previous work where roughly half of liposomes prepared in this manner where found to exhibit fluorescence as a result of the expression of yellow