Fig 1.
Lateral diffusion in the cytoplasm of E. coli.
A diffusion map obtained with SMdM is overlayed with a schematic of the cytoplasm of the cell. The top panel highlights the effect of confinement on the measured diffusion, leading to lower diffusion coefficients near the boundaries of the cell. The bottom panel represents the effect of the perceived viscosity by diffusing proteins. Since diffusion scales with the complex mass, bigger particles will be affected more by the crowding of the cytoplasm than smaller molecules and move relatively more slowly, leading to the deviation from the Einstein-Stokes equation. The left panel represents our current hypothesis on the observed slowdown at the cell poles compared to the cell center, with accumulation of aggregated or misfolded proteins impairing the diffusion in these regions.
Fig 2.
Simulation-based Reconstructed Diffusion.
(A) Algorithm representation of SbRD. (B) Diffusion maps of a spherocylinder obtained by analyzing a Smoldyn simulation created with an input diffusion coefficient of 20 μm2/s. Maps are obtained via SMdM analysis (top) and via SbRD (center). The difference between the SbRD map and the SMdM map is depicted in the bottom panel. (C) Comparison of the dependence of the ratio of Doutput / Dinput in a simulated spherocylinder when analyzing the centermos 100 nm2 area (top) and the cell pole (bottom) with SMdM and SbRD. The gray dotted line represents the ideal case in which the ratio of Doutput over Dinput is one. The relevant range of diffusion coefficients for proteins in the cytoplasm is highlighted in green.
Fig 3.
Maximum likelihood-based detection method.
(A) Field of view and fitting of billiards around the identified cells, using a maximum likelihood estimation method. The cell indicated by a red arrow was discarded because it has too many displacements. (B) Details of the fitting process. In the top panel, the initial guess used for fitting, encompassing all the points clustered as a single cell, is represented in grey, while the final fitting is colored. For cells that just completed division, the initial guess encompasses both cells. Due to the abnormal length of the cell, the fitting routine is automatically performed with two billiards to detect both cells. Using the fitting information, it is possible to identify the newly formed cell pole (white dots) and the old one (red dots). The bottom panel shows the billiard used to describe the shape of cell 7. Since cells are represented as billiards, it is possible to obtain accurate estimates of their length and radius, which allow distinguishing the cell poles and cell center for every cell. For cells that just divided and two billiards overlapping, the intersection points are calculated (green dots) and used to draw a line (green line), which is then used to properly model the spherocylinder in the SbRD routine. (C) Comparison of Maximum likelihood method and Voronoi clustering for cell detection. Voronoi clustering cannot properly distinguish cells that are too close to each other. (D) Comparison of the apparent diffusion coefficients obtained with SMdM by analyzing the central region (top) and the poles (bottom) of cells identified with Voronoi clustering and with our maximum likelihood method, from images acquired in our previous work [14]. Curves are obtained via kernel density estimation.
Table 1.
Lateral diffusion coefficients of cell center and cell poles obtained via SMdM and SbRD for constructs fused to mEos3.2.
The columns show the name of the protein, their complex mass, their diffusion values obtained via SMdM (Dapp) for cell center and cell poles, and the confinement-corrected diffusion values obtained via SbRD (Dcc). The Uniprot ID is provided for every protein, except for mEos3.2, for which the Fpbase ID is given.
Fig 4.
Comparison of the diffusion values obtained via SMdM and via SbRD.
Comparison of the apparent diffusion coefficient obtained via SMdM and of the confinement-corrected diffusion coefficient for both the cell center (top) and the cell poles (bottom) for the dataset of proteins tagged with mEos3.2 [14]. Asterisks indicate statistical significance obtained via a Mann-Whitney U test for non-normally distributed data.
Fig 5.
Cell pole analysis of diffusion.
(A) Comparison of the ratios of the diffusion coefficients at the poles and center of the cell for data analyzed by SbRD and SMdM. The black dotted line shows the case when the diffusion at the poles and center is equal. For simulated data (orange) the ratio obtained with SbRD is equal to 1 for each protein. For microscopy data (blue) we find a positive correlation between the ratio obtained by SMdM and SbRD. (B) Ratios of diffusion at the poles and the cell center obtained by SMdM and SbRD. (C) Ratios of diffusion at the poles and cell center by SbRD for control cells, and cells treated with 250 μg/ml of erythromycin or 500 μg/ml rifampicin. (D) Diffusion coefficients at the cell center (blue) and poles (orange) for control, erythromycin- and rifampicin-treated cells. (E) Distribution of differences in diffusion coefficients between the newly formed cell pole and the old cell pole. The orange line represents the average of the distribution, the black dashed line is the zero. All curves are obtained via kernel density estimation. Statistical significance is indicated with asterisks.
Fig 6.
Accumulation of aggregated structures at the cell poles and correlation with aging.
A cell with an old pole, with slower diffusion, and a new cell pole, with faster diffusion is shown at the top. As the cell cycle progresses this difference is maintained. When the cell divides and the septation ring forms (red), the two daughter cells will inherit the two cell poles. One of the cells will have the oldest pole as its old pole, while the other will have the new pole of the mother as its old pole.