Tools for engineering coordinated system behaviour in synthetic microbial consortia

Advancing synthetic biology to the multicellular level requires the development of multiple cell-to-cell communication channels that propagate information with minimal signal interference. The development of quorum-sensing devices, the cornerstone technology for building microbial communities with coordinated system behaviour, has largely focused on cognate acyl-homoserine lactone (AHL)/transcription factor pairs, while the use of non-cognate pairs as a design feature has received limited attention. Here, we demonstrate a large library of AHL-receiver devices, with all cognate and non-cognate chemical signal interactions quantified, and we develop a software tool that automatically selects orthogonal communication channels. We use this approach to identify up to four orthogonal channels in silico, and experimentally demonstrate the simultaneous use of three channels in co-culture. The development of multiple non-interfering cell-to-cell communication channels is an enabling step that facilitates the design of synthetic consortia for applications including distributed bio-computation, increased bioprocess efficiency, cell specialisation and spatial organisation.

The result of the analysis which uses both experiments and some mathematical modelling is encoded in a software tool that the authors propose for automated identification of orthogonal chemical communication channels and is made available as supplementary information to the manuscript.
I found the paper clear and well written. From a methodological viewpoint the authors tackle a problem of relevance particularly, as they recognize in the paper, for the current effort in Synthetic Biology aimed at constructing multicellular consortia. A striking example is that of multicellular control systems which the authors refer to on line 203 although without any further detail. (I would suggest adding a reference to recent work on this interesting topic.) The experimental work is well described and the mathematical models reasonable for the cognate and non-cognate pairs under investigation.
I would have liked to see more details on the software tool. For example, when the user specifies the conditions that need to be met what guarantee is there that a pair can be found that matches? Also typically how long it takes for the software to "converge" onto a solution? It might be that given the limited number of combinations to be examined the software works well on average but a more quantitative assessment of its performance and the underlying algorithm would enrich the paper if added to the supplementary info.
As a final consideration, I believe this is a methodological paper whose findings can be useful in several domains such as those identified in the manuscript. It is not going "to influence thinking in the field" but will certainly be a useful addition to the design and analysis tools available in the literature. I am therefore left wondering whether it might better fit a more specialist journal in the area of computational biology.
engineering coordinated system behaviour in synthetic microbial consortia", Nicolas Kylilis,

Zoltan A. Tuza, Guy-Bart Stan and Karen Polizzi
We thank the reviewers for taking the time to provide detailed feedback on our paper. We hope the responses detailed below will clarify any questions and will satisfy the concerns that were raised.  Response: We agree with the reviewers' comment that the Biobrick genetic part identification code used in the figure is not something that the average reader of the manuscript will be familiar with. To address this, we have now edited the figure so that the genetic construct includes generic descriptions of its constituent parts that should be familiar to most researchers in biological sciences. For example, the B0015 part has been relabelled as a double terminator, the B0034 part has been renamed RBS etc. Additionally, we have added a segment in the figure caption that provides the available identification codes (as they would appear in the Registry of Standard Biological Parts) for each genetic element. Figure 2A, the y axis needs to be defined. Figure 2B, I am assuming that these 6 plots correspond to the 6 promoters with the different colour codes, but these should be explained in the figure legend.

Comment #2 -
Response: In Figure 2A, the y-axis of the histograms has been now defined as the "number of events" to denote the type of data derived from flow cytometry experiments. In agreement with the suggestion regarding Figure 2B Figure 3A, but don't discuss it on the paper. I think it should be discussed in the paper. I think the illustration of the Anderson library should be removed from figure 3A. This is not relevant to show it as the setup is very simple. It should be discussed in the paper instead.

Response:
As suggested by the reviewer, the illustration for the Anderson promoter library has now been removed from Figure 3A. Additionally, after further consideration regarding the message we wanted to deliver to the reader in this figure, we removed the bar chart for the Anderson promoter library characterisation. Instead we now supply these data in the supplementary information ( Figure S2 and Table S3). Also, as suggested by the reviewer, we provide a more extended discussion on the Anderson promoter and the reasons why we carried out the characterisation of the mentioned promoters in the supplementary information (Page 7, lines 30-46). Regarding convergence, as illustrated in the flowchart of the software structure ( Figure S5), the software tool examines all possible solutions, and if no solution that satisfies userdefined specifications is available the algorithm terminates and displays a message to the user that "no solution was found to achieve the stated specifications". The user will then be able to "relax" their specifications to attempt to find a feasible solution given these new user-defined specifications.
Comment #3: As a final consideration, I believe this is a methodological paper whose findings can be useful in several domains such as those identified in the manuscript. It is not going "to influence thinking in the field" but will certainly be a useful addition to the design and analysis tools available in the literature. I am therefore left wondering whether it might better fit a more specialist journal in the area of computational biology.

Response:
We believe Nature Communications is the ideal outlet for our work because it will enable it to have wide visibility among our target audiences of synthetic biologists, biotechnologists, and bioengineers. We think our work, encompassing a large amount of new experimental data and the associated software tool (that was designed for the nonprogrammatically inclined end-user), will be useful to many researchers-equally for biological sciences researchers and mathematical modellers-who want to use quorum sensing in the design of synthetic systems. We suspect that submitting this work to a more specialist journal will make it much less accessible and visible to the broad sets of researchers and end-users interested in the currently hot topic of microbial consortia engineering, e.g. synthetic biologist, biotechnologists, and bioengineers.