Harvard University COVID-19 updates

Department News



I have been intrigued by bacterial flagellar motors since 1973, when Bob Anderson and I argued that bacterial flagellar filaments are rigid helices driven at their base, rather than flexible structures that propagate bending waves (1); although, the physics of thrust generation is essentially the same (2).  This realization, that bacterial flagella actually rotate, came more than ten years before it was realized that the membrane ATPase also is a rotary device, pumping protons out when hydrolyzing ATP and taking them in to power its synthesis.

The rotary motor is like most electrical motors, for example, a fan.  But it is much smaller, only a tenth of the wavelength of blue light in diameter.  It has a rotor connected to a drive shaft passing through bushings embedded in the outer layers of the cell wall (the peptidoglycan and lipopolysaccharide layers, respectively) and then connected to the filament (the propeller) by a flexible coupling (the hook).  There are force-generating elements bolted onto the peptidoglycan (the rigid framework of the cell wall), which are powered by the same proton electrochemical gradient that powers ATP synthesis; although, some marine organisms use sodium ions instead.  The disc upon which other rotor components are mounted is called the MS-ring (for Membrane and Supra membrane), initially thought to be two rings in tandem (3).  The MS-ring is embedded in the cytoplasmic membrane, which is fluid.  The innermost motor components comprise the C-ring (for Cytoplasmic), which is really three rings, named for the proteins FliG, FliM, and FliN.  The force-generating elements exert forces on FliG, while FliM and FliN, with FliG, comprise a switch complex that controls the direction of rotation.  A phosphorylated signal generated by membrane chemoreceptors (CheY-P) binds to FliM and then FliN, increasing the probability that motors spin clockwise (CW, viewing the drive shaft from the outside of the cell).  When flagella spin CW, cells tumble in place rather than swim smoothly, an event that enables them to hunt chemical nutrients.  But that is another story.

We know from linking markers to hooks or filaments, or linking hooks or filaments to glass slides, that these external components rotate relative to the rigid framework of the cell wall.  More than 15 years ago, I began to wonder whether the innermost components, e.g., FliG, FliM, and FliN, rotate at a similar speed.  One might determine this by embedding a fluorescent protein in the FliN ring, say, observing its fluorescence with polarized light (a probe beam), then bleaching about half of the fluorophores by turning up the intensity by a large factor for a period of time short compared the period of the ring’s rotation.  The fluorophores that are excited by the probe beam will be bleached by the flash, and the fluorophores whose transition moments are perpendicular to those of the bleached components will tend to survive.  Provided that the fluorescent protein is not free to rotate relative to the C-ring, the unbleached fluorophores will line up with the probe beam a quarter turn of the rotor after the flash and then at every half turn thereafter.  Thus, one would expect the fluorescence excited by the probe beam to drop suddenly (at the time of the flash), and then ring at twice the motor rotation frequency.  However, in a population of motors, given the dispersion in rotation speeds, this signal will rapidly decay, as different motors get out of step.

It took 15 years, but we finally succeeded.  The C-ring, as represented by FliN, rotates at the same speed as the hook and the filament (4).  The hero of the piece is Gabriel Hosu, a postdoctoral fellow, an electronics enthusiast with degrees in medicine and biophysics!  He realized that we might overcome signal-to-noise problems by looking at single motors on cells with hooks but no filaments: these rotate at about 300 Hz.  To get a useful signal, this requires looking at several hundred motors and logging their fluorescence over 15-microsecond spans about every 150 microseconds.  Veda Nathan, a research technician, solved the fluorescent labeling problem by linking the fluorescent protein YFP at both its N- and C-termini internally in FliN, obtaining motors that remain fully functional.

Were we crazy spending so much time on this experiment?  Not really, because our interest in fluorescence led to other things: the visualization of the motion of flagellar filaments following small-molecule fluorescence labeling (5), the use of fluorescent-protein fusions to determine the locations of components of the chemotaxis machinery (6), and the development of a fluorescence resonance energy transfer (FRET) technique to monitor, in vivo, the activity of the cell’s receptor kinase, the enzyme that converts the inactive signaling molecule, CheY, to its active form, CheY-P (7).  We also used fluorescence recovery after photobleaching (FRAP) and total internal reflection fluorescence (TIRF) microscopy to study motor adaptation, e.g. (8).  When motors spin predominantly CCW, they recruit FliM and FliN, thus increasing their sensitivity to CheY-P, adapting to the output of the chemotaxis signaling pathway.  When viscous loads increase, they add more force-generating elements, thus increasing their output power.  So we are dealing with a remarkably dynamic machine.

Read more in PNAS or download PDF
1.  Berg HC, Anderson RA (1973) Bacteria swim by rotating their flagellar filaments. Nature 245(5425):380–382. PDF

2.  Berg HC (1993) Random Walks in Biology (Princeton 1993) Fig 6.3, p 79.Princeton University Press

3.  DePamphilis ML, Adler J (1971) Fine structure and isolation of the hook-basal body complex of flagella from Escherichia coli and Bacillus subtilis. J Bacteriol 105(1):384–395. PDF
4.  Hosu BG, Nathan VSJ, Berg HC (2016) Internal and external components of the bacterial flagellar motor rotate as a unit. Proc Natl Acad Sci USA 113(17):4783-4787. PDF
5.  Turner L, Ryu WS, Berg HC (2000) Real-time imaging of fluorescent flagellar filaments. J Bacteriol 182(10):2793-2801. PDF
6.  Sourjik V, Berg HC (2000) Localization of components of the chemotaxis machinery of Escherichia coli using fluorescent protein fusions. Molec Microbiol 37(4):740-751. PDF
7.  Sourjik V, Berg HC (2002) Receptor sensitivity in bacterial chemotaxis. Proc Natl Acad Sci USA 99(1):123-127. PDF
8.  Yuan J, Branch RW, Hosu BG, Berg HC (2012) Adaptation at the output of the chemotaxis signalling pathway. Nature 484(7393):233-236.PDF

View Howard Berg’s Faculty Profile

Howard C. Berg (l) and Basarab G. Hosu

Howard C. Berg (l) and Basarab G. Hosu