(l to r) Howard Berg, Pushkar Lele, and Basarab Hosu
Contact with a surface is believed to trigger bacterial swarmer-cell differentiation, and the flagellar motor is thought to be a potential candidate for the mechano-sensor. However, the molecular mechanisms of the sense of touch remain unknown and are considered the holy grail of bacterial swarming studies. Previous investigations of mechanosensing in bacteria have focused on indirect mechanical stimulation, achieved by changing the hardness (or stiffness) of the growth media. Such assays enable the measurements of long-term effects of stimuli on gene expression and swarming behavior, but have failed in determining the mechanisms of mechano-sensing.
At the length-scales of a bacterium, viscous forces dominate inertial forces (~ low Reynolds numbers). In such conditions, when a tiny biological motor moves or rotates an object such as the flagellum, the action of the motor is balanced by the viscous drag (friction) on the object, i.e., the viscous load. The larger the size of the object, the larger the viscous load. To determine how biological motors sense and respond to mechanical stimuli, one can directly stimulate the motor by changing the amount of load, and then measure the immediate effects on the motor structure and its dynamics.
To study mechanosensing in flagellar motors, we applied a direct mechanical stimulus by suddenly increasing the size of the object driven by the motor. To do this, first we focused on stuck cells with motors which rotated very short filaments (~ 50-100 nm) at ~300 Hz. By the virtue of its small size, a short flagellum offers little resistance to the motor rotation (~ zero load). To such a filament, we attached a 1 µm latex bead using optical tweezers. At the instant of attachment, the viscous load on the motor increased suddenly and by a large magnitude, since now the motor rotated a much larger object.
Following the load change, the motors initially slowed down to ~ 6-8 Hz, however, over the next few minutes they adapted by increasing speeds in a stepwise manner to ~ 60 Hz. Single-motor TIRF imaging of YFP-MotB (part of a stator force-generating unit) revealed that this mechano-response occurred by the addition of new force-generating units (stator-remodeling). From the initial speed (immediately following the load change), we inferred that motors running at near zero loads are driven by 1 or at most 2 force-generating units. Those at high loads are driven by 6-11 force-generating units. Thus, the number of stator elements or force-generators driving the motor depends on the viscous load. To our knowledge, this is the only molecular motor to respond to changes in viscous loads in such a fashion.
The bead-attachment experiments confirmed that the cell surface is not involved in mechanosensing, since it is the motor and not the exposed surface of a stuck cell that is subjected to the stimulus. Stator-remodeling was also observed in strains deleted for the fliC gene, indicating that the filaments themselves are not needed for mechanosensing. It is known from our previous studies that the makeup of the motor switch complex depends on the direction of motor rotation. To test whether the direction of rotation influenced stator-remodeling, we repeated these experiments in strains with motors locked in the CW or CCW state. Stator-remodeling was observed in these strains, as well as in strains deleted for fliL or for genes in the chemotaxis signaling pathway, ruling out their role in mechanosensing. These results strongly suggest that the stators themselves act as dynamic mechanosensors. They likely recruit additional stator units by perturbing stator-binding sites following a load-change.
Inhibition of flagellar rotation in Vibrio parahaemolyticus leads to replacement of one kind of flagellum (polar) by another (lateral). Mechano-sensing by the motor in Proteus mirabilis has been proposed to play an important role in swarmer cell differentiation. Our results suggest that the stator senses the change in mechanical load and recruits additional stator elements in response. How gene expression might be linked to stator conformation remains an interesting question.