Lateral transduction is inherent to the life cycle of the archetypical Salmonella phage P22

Lysogenic induction ends the stable association between a bacteriophage and its host, and the transition to the lytic cycle begins with early prophage excision followed by DNA replication and packaging (ERP). This temporal program is considered universal for P22-like temperate phages, though there is no direct evidence to support the timing and sequence of these events. Here we report that the long-standing ERP program is an observation of the experimentally favored Salmonella phage P22 tsc229 heat-inducible mutant, and that wild-type P22 actually follows the replication-packaging-excision (RPE) program. We find that P22 tsc229 excises early after induction, but P22 delays excision to just before it is detrimental to phage production. This allows P22 to engage in lateral transduction. Thus, at minimal expense to itself, P22 has tuned the timing of excision to balance propagation with lateral transduction, powering the evolution of its host through gene transfer in the interest of self-preservation.

Specific comments on figures/analysis follow below: 1) Throughout the manuscript it is not clear if replicates were done, and what exactly is being shown as far as replicates and error. This is true for all RNA and DNA-seq data (ie figs 1, 2, 4, S2 ect). The data points for each replicate should also be shown on each bar graph in addition to the error 2) It would be helpful to orient the experiments if the authors included a time course showing PFU production for the different strains, this would help put the RNA/DNA-seq in context 3) It is clear from the RNA & DNA-seq data that P22 TS mutant is already induced without the temp shift, I have a hard time believing those cells are not sick and producing a significant amount of phage -this makes me concerned about suppressor mutations that could be confounding the results somewhat 4) Figure 1 -the authors are using these data to conclude that int/xis are not expressed until late in infection in WT P22. However, it is unclear from the RNA-seq data alone that this is the casethis needs to be corroborated by analysis at the protein level. I also suggest that they provide a zoomed in view of the RNA-seq data on the int/xis operon -but again, this is not sufficient to conclude anything given that post transcriptional regulation may play a role and they are drawing their conclusions based on this data alone. I would also like to understand a little more about what is encoded downstream of xis that would need to be on late in infection, I'm sure there is a lot known and it would be helpful for the general reader who doesn't work on P22 to have a bit more information on where the structural operons are, what is known about regulators (1 repressor controls all operons ect). 5) Figure S2 is used to support the notion that excision is delayed, however this was very confusing: the text (lines 162-163) indicate the integration rate was calculated as a ratio between attL & attB, but that was not clear from the legend. A simple schematic would help. Line 167 -the authors indicate changes in the integration rate, however, no replicates/error is shown, so I have no idea if this is a meaningful change (it doesn't look like it is). Rather than rely exclusively on deep sequencing approaches -which are skewed if the biological entity is circular and you are mapping to a linear integrated reference for example, I suggest the authors include some PCR based assays to monitor excision/integration. These data are simple to generate and should support analyses already included, providing the model is correct. 6) Given the emphasis on erroneous conclusions being drawn from lab strains/commonly used mutants (ie TS P22 undergoes ERP, no lateral transduction), it is necessary for the authors to provide some evidence that their particular strain of P22 doesn't have any mutations that could be the source of these findings (ie mutations in Int/Xis, repressor ect). Ensuring that natural isolates of P22 lysogens have conserved seq to the strain used here would be valuable. 7) The relevance of the Infection P22 panel in figure 2 is questionable and misleading. Here, there should only be episomal replication -it is unclear why the authors are plotting this as if the phage is 'part' of genome. This gives the impression the profile is something to compare the P22/ts muts to, but this is an entirely different situation.
8) The conclusions (and emphasis) surrounding the difference between LT by wt P22 vs the TS mutant need to be tempered to the point of only suggesting a quantitative difference in the discussion (if desired). The authors clearly demonstrate LT occurs for both, but given that both the phage & the induction condition differs in the experiments, one cannot come to the conclusions stated in this paper. For example, perhaps MMC is a more robust inducer vs the TS shift, perhaps the TS lysogen is sick owing to high levels of spontaneous induction ect. Too many variables. 9) Line 210 "taken together, these results show that thermal induction resulted in the classical ERP program…" this is not supported by the data, the authors show that in the absence of inducer there is already HIGH levels of expression and replication, how is one to know if that phage also 'starts' with replication then excises if the culture is already/always in that induced state? 10) I would suggest combining Figure 3A & 4b, (4a is not particularly informative, can be moved to the supplement, or made clearer that b is a zoom in on the relevant data). Figure 4b is extremely convincing and nicely complements the transduction assays in fig 3a. 11) Figure 5 -it would have been easier to follow if 4b was reported as the PFU/ml relative to an uninduced control. This would much more clearly allow the reader to follow the assay, since that would require re-doing the experiment, I instead suggest adding a schematic to help orient the reader. 12) A point worth considering for the discussion is that although the authors do not see a fitness cost to delayed excision (which accompanies higher LT), their assays only report 1 round of infection. Subtle fitness decreases (as seen in 5b between 60-90 min that is not yet significant) will be exacerbated by repeated passaging and thus even subtle effects would be deleterious to a phage competing with a phage that doesn't do LT in nature. It doesn't seem feasible to do such competition experiments, unless the authors have a way of comparing otherwise isogenic phages that only differ in capacity undergo LT, but their conclusions are quite strong given the nature of the experiments they are able to perform. Even their model in Fig S1 looks as if the burst for phages undergoing LT is lower! It is very counterintuitive that this would not come at a cost, perhaps the other thing to consider is the relative abundance of virions for packaging vs DNA to be packaged. 13) Line 314 -this statement regarding unpublished data is not sufficient, if there is no data being shown to support these conclusions this statement should not included. Along those lines however, it would be interesting for the authors to examine the position of attachment sites for phages that engage in LT in natural strains, perhaps those flanking genes could be directly linked to host fitness/supporting increased virion production ect.

Reviewer #2 (Remarks to the Author):
This manuscript by Fillol-Salom is a continuation of a previous work from this lab demonstrating that prophages can hold a different program of lytic development, involving in-situ replication (while integrated within the host chromosome), DNA packaging and lastly excision (RPE), as opposed to ERP (originally shown for phage 80-alpha of S. aureus). In that work they also demonstrated that RPE promotes lateral transduction. In this manuscript the authors demonstrate that this phenomenon occurs also in P22 of Salmonella, but in the native phage and not in the temperature sensitive mutant (which is widely used). Further they show that the RPE process allows lateral transduction without compromising the phage titter, therefore contributing to the evolution of the host. This is an important work demonstrating the differences between the two P22 phages (native vs. the temperature sensitive, tsc229), their induction pathways (SOS versus temperature) and their impact on lateral transduction. The experiments are nicely presented and convincing. That said, i must say that I don't feel comfortable with the statement in the abstract and throughout the manuscript that the ERP program of the heat inducible mutant is an artefact (or the use of natural versus unnatural). One can argue that what we see in the heat sensitive mutant is the "natural" behaviour of the phage, and what we see in the native P22 upon SOS is an "adaptive" behaviour that is a result of a co-adaptation of the bacteria and the phage. Namely, it could be that the bacteria control the late excision and not the phage, and thus in respect of the phage, this is an unnatural behaviour (though eventually it benefits from it). In any case, I don't think that there are any artefacts here, but two different biological scenarios. The heat-sensitive mutant is less adapted to its host, as due to its mutation it responds to a new mechanism of induction, which is triggered by temperature. Besides this comment, I like the paper very much.
Additional comments that might help to improve the manuscript: I think that the title of the paper is not informative, i recommend to change it. Line 129 and through the manuscript: avoid the use of "unnatural". Fig 2S and the paragraphs describing it are a bit problematic: there are no error bars in the graphs, so it is not clear how significant are the results, the y-axis does not describe integration rate, but in best excision rate, and most importantly the calculation is misleading as it is based on the attL, while it is shown that there is in situ replication which amplifies the attL as well. Thus, the calculation actually represents (excised phages)/(in-situ amplified attL-phage+originally integrated phages+ chromosomal reads). In any case, I don't think this data is essential for the manuscript. line 194 : "delta-rep" should be "delta-pri" line 213-5: ES18 looks to me more similar to P22ts, where both in situ replication and episomal replication proceed simultaneously. Fig S4 is mentioned in the text after S5 and S6. The discussion section is too short and over-simplified.

Anat Herskovits
Reviewer #3 (Remarks to the Author): In this work Fillol-Salom et al examine and compare the induction of wild type phage P22 and a temperature sensitive mutant. They find differences in the way that the two phages enter the lytic cycle, with the wild type phage exhibiting a delayed excision and the temperature sensitive mutant phage excising early after induction. These differences result in much higher rates of lateral transduction for the wild type P22.
This work provides evidence that Salmonella phage P22 functions by a similar mechanism to the Staphylococcal phages previously characterized by the Penades lab (Science, 2018). They also provide some evidence that induction by DNA damage follows a different pathway than that induced by a temperature shift with a P22 temperature sensitive (P22-ts) mutant phage. This is a significant point to have made in the literature as phage ts mutants have been widely used in the past for phage studies and they may provide I found the manuscript to be well written and easy to follow. One notable problem though, is that the introduction is written with a real focus on the processes that are followed by phages like P22. This would be okay if it was made clear that this is how phages like P22 work. However, the way that many parts of the paper outline the background and conclusions make it sound like all phages are like P22. There are many variations within the phage world, with some of these differences should be noted in the text. For example: -not all temperate phages integrate into the bacterial chromosome -some are maintained as episomes -not all phages replicate by the standard rolling circle model as implied in the introductionphages P2 and P4 package replicated circular DNA, phage Mu transposes through the genome as it length DNA molecules.
-prophages switch to the lytic cycle for a variety of reasons, not just when their hosts begin to deteriorate. There are a number of studies that show that quorum sensing also can lead to a switch to the lytic cycle The authors state in line 88 that prophages excise and circularize early as the first step, and that this sequence "is believed to be universal for all integrated prophages". This statement is false for several reasons, and should be removed. E. coli phage Mu does not excise and circularize, it replicates by transposition and the genome is packaged from the bacterial chromosome. In addition, as stated later in the same paragraph, the Penades lab themselves showed that some Staphylococcal phages delay excision until later in their life cycle.
Line 119 -Many would argue that much of our current understanding of fundamental phage that the authors want to stress the importance of their system, but they should be careful not to oversell and to ignore or downplay the enormous body of work outside of Salmonella phage P22.
The authors end the introduction with the statement that "these results propose a new series of events in the life cycle of temperate prophages…" It's not clear to me what the new series of events are -it seems that this work is confirming that wild type P22 is acting in a similar manner to the in situ replication of Staphylococcal prophages that the Penades group previously characterized (2018, Science). Similarly, the statement that "delayed excision leading to LT are naturally parts of the phage life cycle" also seems to imply that this was discovered in this work. It seems to me that this work confirms a similar mechanism for this particular phage, but is not a new discovery itself.
In Fig 1, the P22 tsc229 induction time 0 (red) shows a lot of transcription. Why is this, if this timepoint is supposed to be before induction? It seems to suggest that this mutant was already spontaneously inducing before the temperature shift. When you compare this background to that seen for the wild type P22, there is a striking difference, with very little transcription observed at time 0. How can these be compared when the baseline levels are so disparate? Is there a way to decrease the levels of spontaneous induction from the P22-ts? What happens if the induction is more carefully controlled, for example, by the use of a dominant negative repressor? The worry, of course, is that some of what is being observed is because many of the cells in the P22-ts culture are already well into the lytic cycle due to the poor repression of the prophage. It's not clear to me that the observed "early" induction from the P22-ts lysogen is not merely an artifact due to the high levels of spontaneous induction. These analyses are being done on bulk cultures, which will be a mix of silenced lysogens and inducing lysogens at any point in time.
There is a statement that there is a general leakiness of transcription and spontaneous excision observed in P22 tsc299. What percent of cells are spontaneously inducing? The data presented in Fig S2 appear to show that there is little change in the excision of the P22-ts mutant over the 90 minute timecourse. There is a much larger decrease noted for P22 (>0.25) as compared to P22-ts (~0.05). Why is this? How does this play into the replication cycle? How many phages are made under each of these induction conditions? Are the titers similar for the two modes of induction? The authors note that the percent of integrated P22 changed course and increased after 60 minutes, but they fail to comment on the similar increase observed for the P22-ts mutant.
Why were the infection assays to assess GT not performed for P22-ts?
The analyses of lateral transduction for P22 and P22-ts need additional context in order to appreciate the significance of the numbers presented. What are the phage titers that are released by these two methods of induction? Are they equal? The authors note that there is a high ratio of tail-defective particles resulting from the thermal induction.
In the discussion it is stated that in situ packaging initiates from some of the integrated prophages and excision occurs from some of the others. Which data specifically support these statements? Is this being proposed from previous work that should be referenced?
The statement at the end of the discussion, that the results presented here re-write important concepts of phage biology is a very strong one. The 2018 Science paper that first characterized this mechanism of transfer certainly described a new paradigm. It seems that the work presented here extends the original observation that Staphylococcal phages can mediate LT through late prophage excision to Salmonella phages. They also show that there is a difference observed with the P22-ts mutant, but because of the high levels of spontaneous induction from the prophage, the significance of the difference is difficult to pin down.
Minor points: Figure 3 -what are metameric spans? Line 293-294 -phage Mu and transposable phages have been characterized in this way Line 313 -not all pac prophages integrate into the bacterial chromosome. Wording should be changed here.
Line 314 -what are the unpublished results? Do you mean the results presented here? Reviewer #1 (Remarks to the Author): . Corrected.
As per my original review -the schematic provided in the rebuttal for Fig S3 was not included in the revised figure. Again the description of the calculation in the text / legend is confusing; in the main text it appears the calculation would be: attL/(attL+attB); from the schematic perhaps this is correct, though the legend says sum of attL plus reads spanning an unlinked chromosome--is the unlinked chromosome-chromosome region attB or just an arbitrary location?
Following the reviewer s suggestion, we have now included the schematic figure (see new Figure S3a). We have also corrected the figure legend.
I am also disappointed that the authors are drawing these conclusions from N=1 sample in these data without any additional corroborating evidence (such as protein levels or PCR verification for the circularized junction as suggested in my first review) and that they justify this approach because reviewers either missed it or otherwise excused it in a previous publication. I do not agree that this overstep should persist.
Following the reviewer s suggestion, we have now validated all our results using qPCR (see new Fig. S6). We have obtained 3 replicates of each prophage (after induction), at different times points, and have analysed both prophage excision and circularisation.