Bacterial flagella grow through an injection-diffusion mechanism

Results

Bicolor labeling

A total of 381 bicolor filaments were recorded with the strain EM800 and were plotted with the green part in function of the orange part (Fig. 3.1A). Every points represents a orange filament with a green part that had 1 hour to grow at 37°C. Orange segments without green part were not taken into account possibly indicating cell lysis. Since green-only segment show new filaments growing only during the second growth period, the time when they started growing could not be determined and they were not considered in the analysis. The different length were binned and the mean was computed for each bin. Fig 3.1B shows the graph of those means. The number of bins was determined using a Matlab function which combine 3 different methods of determining bins and taking the mean of those 3.
2 strains were developped to determine whether a chain of FliC could be formed in the central channel of the filament as proposed by Evans and al.[71]. When Arabinose is absent, the strains behave exactly like EM800. When the promoter is added, they express fusion proteins of a flagellar substrate with an unrelated C-terminal protein domain (beta-lactamase). This means that if a chain of fliC can occur, the presence of arabinose would break that chain in the central channel of the filament.
Fig. 3.2 shows the average plots of each strain with and without the presense of arabinose.
Scatter plots for each strain are available in Supplemental Material. The averages were obtained using the same method as the EM800 strain. One would expect a clear difference between the plots with and without the promoter if a chain of flagellar substrates was formed in the central channel of the filament. However we can’t see a clear trend as the 2 graphs show the same downward trend with a same growth rate. Since those 2 strains without arabinose behave like EM800, we would also expect a similar behavior. By comparing Fig. 3.1 and Fig.
3.2, we can see a similar pattern a decreasing growth rate versus initial length.
Similar experiments were done with a wild-type strain (EM2011). Fig. 3.3 shows the results obtained with 255 filaments measured. Experiments in the presence of arabinose were also conducted as a control and results are in the supplementary material.

Tricolor labeling

To ensure that the second part did not stop growing during the 1-hour period, we also performed 3-color labeling. The order of labeling was changed however since for an unknown reason, the labeling with AF546 in the cell gave enourmous backgroung noise compared to the fluorescent signal of the filaments. Hence, we performed the first 2 rounds of labeling in the tube and only the third one was done in the cell. A graph of the results of 122 filaments is shown in Fig. 3.4. Only filaments with 3 distinct segments are shown in the graph. A graph of filaments with only the 2 first segments and lacking a third one is available in the supplementary material.

Discussion and Conclusion

This study focused on the rate of growth of bacterial filaments as a function of its length. A previous study by Turner and al. [48] found that E.coli filaments’ rate of growth was independent of its length. Those results were used by Evans and al. [71] to suggest a chain mechanism model for the growth of bacterial filaments. By looking at the graphs of the length of filaments and especially the graphs of the means, we can clearly see a downward trend in the rate of growth which is in contrast to Turner and al. [48]. Since our initial experiments were done using a modified strain missing part of the flagellar motor (EM800), we did it again with a wildtype strain (EM2011) in which we saw a slower rate of growth, but a downward trend nonetheless.
To further our investigation of the chain model, 2 strains were modified to have a secreted protein with an unrelated C-terminal at the end of it. Those specific substrates would be secreted only in the presence of the promoter arabinose. Those unrelated terminal (FlgL-bla and FlgM-bla for EM1281 and EM1282 respectively), when secreted, prevent the formation of a chain of FliC. Those two substrate continue to be secreted by the bacteria and excreted through the filament’s central channel by the export apparatus during the growth of the filament [47]. Since FliC is secreted much more than those two substrates, we would expect a slowing down of the rate of growth be it dependent or independent of its length since the presence of either FlgL-bla or FlgM-bla would break a proposed chain. Without the presence of arabinose, the normal substrates FlgL and FlgM are secreted and the strains behave exactly like EM800.
We can see in Fig. 3.2A and B that the rate of growth decreases in a similar fashion as EM800 whether arabinose is present or not which should not be the case is a chain of protein was present. We would expect the blue ’.’ marker to be lower on the graph indicating a slower rate of growth. However, we see that even when the unrelated C-terminal substrates are secreted, the rate is the same in both strains with and without arabinose. We thus conclude that the proposed chain mechanism model might not explain exactly what is happening.
Finally, since our results are so different from [48], we wanted to be sure that we did not measure length of dead filaments. To do so, we repeated the same experiments, but we added a third period of growth with a third labeling. This ensures the filament grew for the whole hour during the second period of growth. We can see in Fig. 3.4 that there is still a downward trend in the rate of growth of the second part versus the first one. The red ’x’ marker show the third labeling versus the previous two added together. Of course, like in the bicolor case, we cannot say that all filament grew for the whole hour but, nonetheless, we can see a clear downward trend.