Talk:PLOS/Architecture of the Escherichia coli nucleoid

Reviewer 1: Stella Stylianidou

The topic gives a good summary of what we know about how the bacterial nucleoid is organized, and it will be a great addition to wikipedia.

Major comments

The review goes through several findings about the main NAPs, in a few different sections. I think it would benefit the reader, if the knowledge was organized and brought together for all four main NAPs. Maybe with a table. They could compare in which phase of the E.coli life cycle each protein is thought to play role, how each one acts on the chromosome, whether they act locally or not etc.

The different models for the arrangement of the nucleoid should be added by the macrodomains section:
(i) the decodensed ter region stretched between the left and right poles (eg. Wang 06, Wiggins 10), (iii) ori and ter at the center, with the left and right arms folded (eg. Youngren 2014)
(ii) and the model of the OriC positioned at midcell and ter at the new pole before segregation starts (eg. Niki 2000)

The nucleoid dynamics section seem to be more about ‘NAPs and the bacterial life cycle’. It does not cover a lot about what is known regarding nucleoid dynamics: the mechanism of segregation, chromosomal loci motion etc.

Minor comments

line 52. Add a sentence explaining how brownian motion arises in DNA before explaining persistence length.

line 71. define plectonemic (it is defined much later in the paper)

line 109. ‘In the magnetic tweezers experiments’.. The magnetic tweezers should be mentioned in the previous sentences, when referring to the results of the experiments.

line 119 - (Fig. ) is missing

References to Figure 3 should be referring to Figure 2. Figure 3 does not seem to be discussed anywhere in the paper.

line 157 It should be added that there is a high intracellular concentration of IHF, which is why it is suggested by some researchers that IHF binds both specifically and non-specifically binding.

line 178. Regarding H-NS clustering, it should be pointed out that the lab that did the original experiment later pointed out that the clustering may have been exaggerated by the fusion (Wang S., 2014)

2.5 Fis has also been reported to form bridges and organize the branched plectonomes (Schneider R, 2001)

It would be helpful to replace the (See below) references with (See section X).

line 252. The sentence is a bit hard to read, maybe cut it in two.

line 283. Rephrase (remove motion?) : “Restrained supercoils can also act as barriers motion.”

Add in figure 4, a single twist, vs single writhe. Explain tw, wr in the figure and add a description. It would also help to add simpler definitions for twist and writhe, eg. number of times the the strand crossing the other strand and number of times the double helix crosses itself.

line 337. Needs reference: "Besides, Fis can modulate supercoiling by repressing gyrA and gyrB transcription through direct binding."

line 610. Please rephrase: ‘A 3D structure of DNA within nucleoid is linked to gene expression at perhaps different scales’.

line 635. It is also worth mentioning in this section that supercoiling is thought to contribute to transcriptional bursting

Response to Reviewer 1

Major comments

The review goes through several findings about the main NAPs, in a few different sections. I think it would benefit the reader, if the knowledge was organized and brought together for all four main NAPs. Maybe with a table. They could compare in which phase of the E. coli life cycle each protein is thought to play role, how each one acts on the chromosome, whether they act locally or not etc.

We agree. As suggested, we have added tables for abundance and DNA binding properties of NAPs. A part of Fig 2 has a table format.

The different models for the arrangement of the nucleoid should be added by the macrodomains section: (i) the decodensed ter region stretched between the left and right poles (eg. Wang 06, Wiggins 10), (iii) ori and ter at the center, with the left and right arms folded (eg. Youngren 2014) (ii) and the model of the OriC positioned at midcell and ter at the new pole before segregation starts (eg. Niki 2000) We would like to emphasize the point that nucleoid has spatial organization. The text has been greatly modified to keep that focus. Therefore, we believe it is not in the scope of the article to discuss different models of how macrodomains of the nucleoid arrange during replication phases.

The nucleoid dynamics section seem to be more about ‘NAPs and the bacterial life cycle’. It does not cover a lot about what is known regarding nucleoid dynamics: the mechanism of segregation, chromosomal loci motion etc.

We have changed the sub-heading to keep it specific to only growth-dependent dynamics.

Minor comments

line 52. Add a sentence explaining how brownian motion arises in DNA before explaining persistence length.

It is a good idea. Changed, as suggested

line 71. define plectonemic (it is defined much later in the paper)

Changed, as suggested

line 109. ‘In the magnetic tweezers experiments’.. The magnetic tweezers should be mentioned in the previous sentences, when referring to the results of the experiments.

Since, we have modified the text extensively, the suggested change is no longer relevant.

line 119 - (Fig. ) is missing

Fixed.

References to Figure 3 should be referring to Figure 2. Figure 3 does not seem to be discussed anywhere in the paper.

Fixed.

line 157 It should be added that there is a high intracellular concentration of IHF, which is why it is suggested by some researchers that IHF binds both specifically and non-specifically binding.

Changed, as suggested. We have a new table for intra-cellular concentration of NAPs.

line 178. Regarding H-NS clustering, it should be pointed out that the lab that did the original experiment later pointed out that the clustering may have been exaggerated by the fusion (Wang S., 2014).

Since the new 3C data does not support the model, we have removed the pertaining text; so, it is no longer relevant.

2.5 Fis has also been reported to form bridges and organize the branched plectonomes (Schneider R, 2001)

Added a sentence in the section 3.2.3 as suggested.

It would be helpful to replace the (See below) references with (See section X). line 252. The sentence is a bit hard to read, maybe cut it in two.

Text amplified as suggested.

line 283. Rephrase (remove motion?) : “Restrained supercoils can also act as barriers motion.”

Changed, as suggested.

Add in figure 4, a single twist, vs single writhe. Explain tw, wr in the figure and add a description. It would also help to add simpler definitions for twist and writhe, eg. number of times the the strand crossing the other strand and number of times the double helix crosses itself.

Changed, as suggested.

line 337. Needs reference: "Besides, Fis can modulate supercoiling by repressing gyrA and gyrB transcription through direct binding."

Reference added.

line 610. Please rephrase: ‘A 3D structure of DNA within nucleoid is linked to gene expression at perhaps different scales’.

Rephrased.

line 635. It is also worth mentioning in this section that supercoiling is thought to contribute to transcriptional bursting

Changed as suggested.

We thank Dr. Stylianidou for critical comments.

Reviewer 2: Linda J Kenney

Peer reviewer 2 comments and recommendations given as an annotated PDF of the submitted article:

File:The bacterial nucleoid Reviewer 2 comments and edits.pdf

Response to Reviewer 2

We have indeed incorporated all the suggested changes. We thank Dr. Kenney for careful reading of the manuscript and insightful comments.

Reviewer 3: Frederic Boccard

This review aims at presenting an overview of the different scales of nucleoid organization in E. coli and at describing the mechanisms at work to shape the DNA molecule into a nucleoid. Overall, even though the text refers to many of the most relevant publications in the field, I am not enthusiastic about the review that appears rather as a catalogue of papers without connecting or prioritizing the important data; all the data are presented at the same level of importance and it will be very difficult to the non-specialist readers to apprehend the key issues. There are no real connections between the data presented in the different sections or results obtained with different approaches.

Main comments

Title : Change “The Bacterial Nucleoid” by the “The E. coli nucleoid”

E. coli chromosome organization is paradoxical in many ways. Even though many principles may apply to different bacteria, several aspects are specific to E. coli.

1) Abstract: Existence of 400 plectonemic loops: l 25-27

The mostly negatively supercoiled DNA results in branched and plectonemic structures. Genetic studies revealed the presence of stochastic boundaries between these structures, allowing DNA interactions in cis between sites located more than 100 kb apart (Higgins et al., 1996). Later, it has been shown that RNA polymerase was able to block the movement of plectonemic supercoils in the transcribed track, thus generating supercoil diffusion barriers (Booker et al., 2010). Those constraints are probably responsible for segmenting the chromosome into “chromosomal interaction domains” identified through 3C/Hi-C (ranging in length from 30 to 400 kb); indeed, these domains often display highly expressed genes at their boundaries in Caulobacter crescentus, Bacillus subtilis, Vibrio cholerae and E. coli. Therefore, new results with HiC methods confirmed the main conclusion obtained with the resolution assay, i.e. highly expressed genes are impeding supercoiling slithering and prevent assembly of a synapse between res sites. I would be less affirmative concerning the size of 10-kb for the plectonemic loops in regard to results with HiC/3C maps; I would rather discuss how Higgins and Cozzarelli’s groups concluded about this 10-kb estimates and try to replace their results and interpretations in the new context provided by 3C results.

The results obtained using the different approaches should be discussed together and authors should propose a model

2) 3D structure that dictates gene expression

It is not clear to this referee whether we can say in a general sense that 3D structure dictates gene expression in prokaryotes. From the 3C results, it is clear that gene expression generates specific DNA structures. To my knowledge, evidences in prokaryotes for the opposite are specific cases and examples should be mentioned (for example, bending by Fis, IHF, lacI, or other factors allowing contacts between several factors activating transcription, etc). Noteworthy, this type of interactions is also involved in other processes involving DNA metabolism such as replication initiation, recombination, transposition, etc.

3) All sections §2 are not very exciting. As it is difficult to see the connections with the others scales (10 kb and > 100 kb), it lacks interest. Try to connect with results from other sections.

4) In § 2.1, most data are in vitro data. I am not sure that these results describe the in vivo effect of HU on the chromosome. In vivo results concerning nucleoid organization should be described as well as results obtained with 3C approaches.

5) §2.2 should be positioned after the various NAPs

6) §2.3 l 48. The sentence “IHF is more a specific DNA binding protein” should be rephrased. Most of IHF molecules (>99%) are bound non-specifically to bulk DNA in vivo (see Murtin et al., (1998) J. Mol. Biol. 284: 949). Affinity for specific and non-specific binding is known and should be cited (ibid). It is clear from these data that the main role of IHF (and probably of HU or Fis) in nucleoid global organization roots from its ability to bind non-specifically and in a dynamic way to bulk DNA.

The role of IHF binding at some high-affinity specific chromosomal targets could be discussed. The consensus binding site is given for Fis in § 2.5, not for IHF; why ?

7) § 2.4: H-NS. Results from 3C-seq obtained in wt and hns mutant should be discussed.

8) In several places, the Figure is not numbered (§2.1; § 2.5)

9) Figure 4 is not understandable by non-specialists without a detailed legend.

10) Figure 5: the authors should correlate this “loop model” with results obtained from i) Resolution assays and ii) HiC contact maps. See my comment on the existence of 10-kb plectonemic loops (for example, see Figure 1D or Figure S21 in Le et al. (2013) Science 342:731).

11) From the figure 6, it will not be obvious to non-specialists how the tracking of RNA polymerase generates supercoiled domains.

12) Figure 8: the model results from in vitro analyses mostly performed with MukB. As MukEF are required for MukBEF activity, one may question the importance of the results obtained for MukBEF activity. I would suggest the authors to rather discuss models about MukBEF activity in relation with its interaction with Topo IV/MatP and the mechanism used by other bacterial condensins (see for example results obtained by Sherratt’s group). The review should give more emphasis on the activity of MukBEF activity through the formation of dimers of MukB dimers. If the results with MukB are informative about how this protein interacts with the DNA, it does not give cues about condensin activity and its role in nucleoid organization. I would delete Fig. 8B-8C and replace by a model of condensin/MukBEF activity.

13) Page 14: Nucleoid dynamics

Many changes occurring in stationary phase should result from the reduced transcription, as evidenced by the changes in boundaries of CIDs. To my knowledge, the effect of other factors has not been documented using the 3C-method. Such experiments should shed light on the role of various factors in organizing the nucleoid in stationary phase.

14) Page 14: authors should connect results obtained in 3C experiments with the model of repression by H-NS.

15) Nucleoid-membrane connections

There are now several examples of proteins bound specifically to the chromosome that connect it to cellular structures in various bacteria. In E. coli, SlmA and MatP play a role in the control of chromosome segregation and cell division. This should be mention.

Minor points

Figure 1: Reference is Le Gall, not Gall Le

Page 2, l 47: remove “excellent”

Figure 3: Fis in red only for exponential phase. Why is H-NS green in exponential phase and red in stationary phase?

Response to Reviewer 3

This review aims at presenting an overview of the different scales of nucleoid organization in E. coli and at describing the mechanisms at work to shape the DNA molecule into a nucleoid. Overall, even though the text refers to many of the most relevant publications in the field, I am not enthusiastic about the review that appears rather as a catalogue of papers without connecting or prioritizing the important data; all the data are presented at the same level of importance and it will be very difficult to the non-specialist readers to apprehend the key issues. There are no real connections between the data presented in the different sections or results obtained with different approaches.

We appreciate the comment. We have massively revised the text to address the criticisms of the reviewer.

Main comments

Title : Change “The Bacterial Nucleoid” by the “The E. coli nucleoid”

E. coli chromosome organization is paradoxical in many ways. Even though many principles may apply to different bacteria, several aspects are specific to E. coli.

We have changed title to “Architecture of the Escherichia coli nucleoid”

1) Abstract: Existence of 400 plectonemic loops: l 25-27

The mostly negatively supercoiled DNA results in branched and plectonemic structures. Genetic studies revealed the presence of stochastic boundaries between these structures, allowing DNA interactions in cis between sites located more than 100 kb apart (Higgins et al., 1996). Later, it has been shown that RNA polymerase was able to block the movement of plectonemic supercoils in the transcribed track, thus generating supercoil diffusion barriers (Booker et al., 2010). Those constraints are probably responsible for segmenting the chromosome into “chromosomal interaction domains” identified through 3C/Hi-C (ranging in length from 30 to 400 kb); indeed, these domains often display highly expressed genes at their boundaries in Caulobacter crescentus, Bacillus subtilis, Vibrio cholerae and E. coli. Therefore, new results with HiC methods confirmed the main conclusion obtained with the resolution assay, i.e. highly expressed genes are impeding supercoiling slithering and prevent assembly of a synapse between res sites. I would be less affirmative concerning the size of 10-kb for the plectonemic loops in regard to results with HiC/3C maps; I would rather discuss how Higgins and Cozzarelli’s groups concluded about this 10-kb estimates and try to replace their results and interpretations in the new context provided by 3C results. The results obtained using the different approaches should be discussed together and authors should propose a model

We have extensively revised the text to discuss “chromosomal interaction domains” and have additionally discussed a role of RNAP in creating CID boundaries as well supercoiling diffusion barriers.

2) 3D structure that dictates gene expression

It is not clear to this referee whether we can say in a general sense that 3D structure dictates gene expression in prokaryotes. From the 3C results, it is clear that gene expression generates specific DNA structures. To my knowledge, evidences in prokaryotes for the opposite are specific cases and examples should be mentioned (for example, bending by Fis, IHF, lacI, or other factors allowing contacts between several factors activating transcription, etc). Noteworthy, this type of interactions is also involved in other processes involving DNA metabolism such as replication initiation, recombination, transposition, etc.

We have modified text to clarify these issues.

3) All sections §2 are not very exciting. As it is difficult to see the connections with the others scales (10 kb and > 100 kb), it lacks interest. Try to connect with results from other sections.

We have modified text to address the criticism. We hope the reviewer now agrees with the changes.

4) In § 2.1, most data are in vitro data. I am not sure that these results describe the in vivo effect of HU on the chromosome. In vivo results concerning nucleoid organization should be described as well as results obtained with 3C approaches.

We argue that in vitro data provide possible mechanism of actions for NAPs. Therefore, the in vitro data cannot be ignored as long as controlled experiments are done. Otherwise mechanisms would never be known. Besides, in vitro data provide foundation for future studies to address how these proteins function in vivo. With regard to “in vivo results concerning nucleoid organization”, we do discuss the 3C results and role of NAPs in the section 4.

5) §2.2 should be positioned after the various NAPs

Changed as suggested.

6) §2.3 l 48. The sentence “IHF is more a specific DNA binding protein” should be rephrased. Most of IHF molecules (>99%) are bound non-specifically to bulk DNA in vivo (see Murtin et al., (1998) J. Mol. Biol. 284: 949). Affinity for specific and non-specific binding is known and should be cited (ibid). It is clear from these data that the main role of IHF (and probably of HU or Fis) in nucleoid global organization roots from its ability to bind non-specifically and in a dynamic way to bulk DNA. The role of IHF binding at some high-affinity specific chromosomal targets could be discussed. The consensus binding site is given for Fis in § 2.5, not for IHF; why?

We have modified text to discuss both specific and non-specific binding of IHF. We believe most of the IHF molecules occupy specific sites in the genome in the growth phase because number of the known binding motif (Prieto et al 2012) and number of IHF molecules are very close. However, IHF is likely to bind non-specifically in the stationary phase as the abundance increases by five-fold. This is apparent in the ChIP-Seq signal of IHF in the growth and stationary phase (Fig. 3). We have included a consensus binding site for IHF as well. In fact, we have organized the known binding motifs and the specific and non-specific affinities of NAPs in a table.

7) § 2.4: H-NS. Results from 3C-seq obtained in wt and hns mutant should be discussed.

In the revised text, we decided to completely eliminate discussion on the role of H-NS in DNA-DNA connections because the recent data (Lioy VS et al 2018 Cell) do not support these interactions.

8) In several places, the Figure is not numbered (§2.1; § 2.5)

Fixed.

9) Figure 4 is not understandable by non-specialists without a detailed legend.

We have added a detailed description in the legend.

10) Figure 5: the authors should correlate this “loop model” with results obtained from i) Resolution assays and ii) HiC contact maps. See my comment on the existence of 10-kb plectonemic loops (for example, see Figure 1D or Figure S21 in Le et al. (2013) Science 342:731).

We have a figure as suggested and discussed the model in the text.

11) From the figure 6, it will not be obvious to non-specialists how the tracking of RNA polymerase generates supercoiled domains.

We have modified the figure as well the description to make it clear.

12) Figure 8: the model results from in vitro analyses mostly performed with MukB. As MukEF are required for MukBEF activity, one may question the importance of the results obtained for MukBEF activity. I would suggest the authors to rather discuss models about MukBEF activity in relation with its interaction with Topo IV/MatP and the mechanism used by other bacterial condensins (see for example results obtained by Sherratt’s group). The review should give more emphasis on the activity of MukBEF activity through the formation of dimers of MukB dimers. If the results with MukB are informative about how this protein interacts with the DNA, it does not give cues about condensin activity and its role in nucleoid organization. I would delete Fig. 8B-8C and replace by a model of condensin/MukBEF activity.

It is an interesting comment. We have removed these figures and included a figure for loop extrusion model by a MukBEF dimer of dimers.

13) Page 14: Nucleoid dynamics

Many changes occurring in stationary phase should result from the reduced transcription, as evidenced by the changes in boundaries of CIDs. To my knowledge, the effect of other factors has not been documented using the 3C-method. Such experiments should shed light on the role of various factors in organizing the nucleoid in stationary phase.

We agree. We also argue at the same time that NAPs such as IHF, HU, and Dps also participate in the reorganization.

14) Page 14: authors should connect results obtained in 3C experiments with the model of repression by H-NS.

We argue that 3C results do not relate directly to transcriptional repression by H-NS.

15) Nucleoid-membrane connections

There are now several examples of proteins bound specifically to the chromosome that connect it to cellular structures in various bacteria. In E. coli, SlmA and MatP play a role in the control of chromosome segregation and cell division. This should be mention.

Added, as suggested.

Minor points

Figure 1: Reference is Le Gall, not Gall Le

Corrected.

Page 2, l 47: remove “excellent”

Removed.

Figure 3: Fis in red only for exponential phase. Why is H-NS green in exponential phase and red in stationary phase?

Fixed.

We thank Dr. Boccard for providing insightful and critical comments on the manuscript.

Second review by Reviewer 3: Frederic Boccard

Review concerning the revised version of “Architecture of the Escherichia coli nucleoid” by Verma, Qian and Adhya.

The review has been considerably edited and is now suitable for publication as it gives a nice and complete overview of different processes contributing to the architecture of the E. coli nucleoid. The figures have also been improved and now more accessible to a wide audience.

Main comment

I have only one comment, it concerns the description of IHF (paragraph 2.2). As it stated, it might be understood that IHF molecules are bound to specific sites in exponential phase and that the additional IHF molecules synthesized in stationary phase are bound to non-specific sites. There is an equilibrium between IHF binding to specific sites (with a high affinity) and to less- or non-specific sites (with lower affinities). The increase of IHF would increase the level of occupancy of specific and non-specific sites but I am not sure it would dramatically change the occupancy ratio at specifc vs non-specific sites (see Murtin et al., (1998); as shown by Ditto et al. (1994), a small reduction in the IHF amount has strong effects on specific sites in support of this assumption).

Also, the number of ihf sites estimated by Grainger et al. is much more reduced compared to the one proposed by Prieto et al.

Minor points

§ 4.2 and 4.3.3: indicate Bliska and Cozzarelli (1987) as the first in vivo evidence for the presence of plectonemes in E. coli DNA.

Page 20: change § 4.3.1 by 5.3.1

Page 23: it may be indicated that the physical clustering of rrn operons is not revealed by 3C-seq (Lioy et al. 2018) suggesting the absence of close physical contacts.

Response to Second review by Reviewer 3: Frederic Boccard

We are happy that the first revision was satisfactory. We thank Dr. Boccard for taking time to give the second review.

I have only one comment, it concerns the description of IHF (paragraph 2.2). As it stated, it might be understood that IHF molecules are bound to specific sites in exponential phase and that the additional IHF molecules synthesized in stationary phase are bound to non-specific sites. There is an equilibrium between IHF binding to specific sites (with a high affinity) and to less- or non-specific sites (with lower affinities). The increase of IHF would increase the level of occupancy of specific and non-specific sites but I am not sure it would dramatically change the occupancy ratio at specifc vs non-specific sites (see Murtin et al., (1998); as shown by Ditto et al. (1994), a small reduction in the IHF amount has strong effects on specific sites in support of this assumption).

Also, the number of ihf sites estimated by Grainger et al. is much more reduced compared to the one proposed by Prieto et al.

We regret that we disagree with arguments of the main comment. The equilibrium hypothesis (mass action) is valid only if there are two forms of IHF - one binds to specific sites and the other to non-specific sites – and there is an equilibrium between the two forms. There is no evidence to assume that. If IHF is limiting compared to the number of specific sites in log phase, they would bind mostly to the specific sites (with 100-fold higher affinity compared to the affinity of non-specific sites). In stationary phase cells when IHF is made in excess, the latter would bind to non-specific sites after saturating the specific sites. As a result, the depletion of IHF at log phase would affect the specific sites but depletion at stationary phase would first affect the non-specific sites and then specific sites. The difference between the two studies, Grainger et al. and Prieto et al. could be due to resolution. Grainger et al is a low resolution micro-array study.

§ 4.2 and 4.3.3: indicate Bliska and Cozzarelli (1987) as the first in vivo evidence for the presence of plectonemes in E. coli DNA.

We have included the reference.

Page 20: change § 4.3.1 by 5.3.1

we have fixed the typo.

Page 23: it may be indicated that the physical clustering of rrn operons is not revealed by 3C-seq (Lioy et al. 2018) suggesting the absence of close physical contacts.

We have added a statement and highlighted this contradiction

Editorial recommendations and Wikification

There are a few edits that would take best advantage of the Wikipedia-integrated format:

• Addition of 'wikilinks' for important concepts and terms ([[w:page|words to display]] produces: words to display)
• Inline formatting of references (using the Quotation Mark icon when editing)
• Upload of the images via this page https://commons.wikimedia.org/wiki/Special:UploadWizard

Further info at the Author Guide
T Shafee (talk) 17:47, 30 October 2018 (PDT)