Cell boundary confinement sets the size and position of the E. coli chromosome
Fabai Wu, Pinaki Swain, Louis Kuijpers, Xuan Zheng, Kevin Felter, Margot Guurink, Debasish Chaudhuri, Bela Mulder, Cees Dekker
Overview
Wu et al. are demonstrating the importance of longitudinal confinement to determine the size and the position of the chromosome in E. coli.
Variations of cell sizes, from short cells to long non dividing cells were obtained through genetic manipulations. Fluorescence imaging reveals compaction or expansion of the nucleoid size in relation with cell length. SIM images show a dynamic structural organization of the chromosome whereas the nucleoid exhibits a very accurate position at mid-cell. The authors are using a coarse grained simulation that fits the observed experimental data and propose that molecular crowding are determinant for chromosome organization.
Feedback
The article was chosen among recent bioRxiv papers to be discussed in our ITQB preprint journal club. People appreciated the elegant systems used by the authors to reach very long and very short E. coli cells and generally agree with the interpretation of the experimental data. Molecular dynamics simulations were debated because of some uncertainty about how crowders were added.
Most importantly, this paper raised new questions that could be interesting for future work:
- What would happen to the chromosome size, position and organization in a cell that does not have cell wall (L-form) or has a different shape?
- To observe in vivo the effect of the crowders, it was suggested to play with the osmolarity of the medium. The increase in crowder concentration could be investigated by a FRET sensor like the ones developed by Boersma et al. — https://doi.org/10.1038/nmeth.3257 — Since we expect crowders to be mostly rRNA and tRNA, RplA fusion should help to visualize the potential localization of crowders in long cells.
Major concerns
Regarding the nucleoid visualization
Figure 1: It was not completely clear for us why nucleoid HU-mYPet is constant. The signal is diluted because of the nucleoid length, but should not we expect some change in the number of HU-mYPet molecules bound to the chromosome, because of (1) a change in HU-mYPet concentration in the perturbed growth conditions, or (2) changes in local DNA structure that influence specific or non-specific binding? Given how interesting it is that the total number of bound HU-mYPet stays constant, we would like to see a more complete description of how this was calculated. Particularly: how was background estimated and subtracted?
We find the assignment of a helix to the observations in Fig. S3B to be somewhat unfounded, and also that this is not an actual prediction of the model: in the model there is much more implicit DNA besides the backbone. Fig. 1 of the Chaudhuri 2012 PRL paper shows that even though the backbone arranges as a helix overall DNA distribution is expected to be more uniform.
Regarding imaging quantification
In general, we felt that more details about experimental and analytical protocols were needed for the reader. For example, some of the analysis could be affected by photobleaching, such as the translocation speed in shorter cells.
Regarding simulations
We had some difficulty interpreting the simulation methods; we would like to have more information on the simulation algorithm and initial conditions in addition to the model, ideally with code made available. We assumed the simulations were similar to Chaudhuri 2012 PRL, with crowding molecules and confinement volume increased at some defined time steps. It was also not clear how the cell walls are moved upon stretching. Is it relative to the chromosome position or to the cell center, or to some other reference?
We wondered whether chromosome positioning at the ½ or ¼ and ¾ positions is an inevitable consequence of (1) how crowding molecules were added and (2) a near impossibility of crowding molecules moving from one pole to the other (Simulation averages in Fig 3D and 6B, and the snapshot in S3A, show zero crowding density within the DNA). Regarding (1) the Widom insertion parameters seem likely to yield equal rates of insertion on either side of the chromosome; because of (2) the system then naturally evolves to a centered chromosome position. In the presence of two chromosomes insertion between the chromosomes is twice as likely, yielding an inter chromosome excluded volume twice a large as the polar volumes. Chromosome positioning then seems a consequence of model design choices. It would be interesting to discuss those in their physiological context: can the chromosome be considered an impassable plug for crowder molecules? Are such molecules expected to be equally expressed on either side of the chromosome? And could this hypothetical crowder compartmentalization by the chromosome have other physiological consequences?
Minor concerns
- Figure 1: 4585 for 12 minutes intervals, this makes 212 cells. Maybe this should be clearly stated.
- Figure 1: Journal club attendees asked for a graph of the different cell length distribution, since it seems that cells longer than 17 um are becoming rare.
- Figure 2: Why are there two ter sites at some time points (69’, 75’, 81’, 96’)?
- Figure S3: An overlay image showing cell contours determined from phase contrast or membrane dye would help to visualize the poles. Maybe 3D-SIM would help to clearly see the helix?
- Journal club members would have appreciated more details about the roles of hns, slmA, and fis; e.g. how slmA-FtsZ interaction and fis transcription regulation could impact the chromosome extent.
Technical notes
- Page 7, line 31, typo “local DNA”
- Figure S1: Legends for C and D are inverted