Finding the Spark

Abstract

It began, as with many good things, at a happy hour. Adam Cohen, a young assistant professor asked whether rhodopsins could be used to optically sense voltage. In the heady days of 2009, channel rhodopsin had just been unveiled as a voltage actuator in neurons. Adam had the insight to question whether rhodopsins could be run in reverse; could optical changes in a protein relay the cellular voltage state using light? This was one of the earliest lessons I learned under his mentorship, and the first piece of advice in this retrospective—turning a scientific question or statement on its head can be the basis for many fantastic research projects.

Joel M. Kralj

Step 1: The Beginning, the Idea

It began, as with many good things, at a happy hour. Adam Cohen, a young assistant professor asked whether rhodopsins could be used to optically sense voltage. In the heady days of 2009, channel rhodopsin had just been unveiled as a voltage actuator in neurons. Adam had the insight to question whether rhodopsins could be run in reverse; could optical changes in a protein relay the cellular voltage state using light? This was one of the earliest lessons I learned under his mentorship, and the first piece of advice in this retrospective—turning a scientific question or statement on its head can be the basis for many fantastic research projects.

At the time of that happy hour, I was about to defend my PhD from Boston University (BU), and Adam's question led to the offer of a postdoc. It was the first step of my journey into the world of bacterial electrophysiology.

Step 2: Creating a Tool to Measure Bacterial Electrophysiology

The second step of the journey involved creating a tool to measure voltage in bacteria. Continuing the fortunate series of events, Adam and I decided that we would create optical electrophysiology tools from rhodopsins. My doctoral work in the laboratory of Ken Rothschild at BU was focused on understanding the molecular motions that underlie rhodopsin function, so I was excited to leverage my old skills in this new way.

Rhodopsins are a class of photoactive proteins distributed ubiquitously in nature (Fig. 1).¹ These membrane proteins all use retina (half of the beta-carotene that gives rise to the orange color of carrots) to absorb photons. They are so widespread and useful for life on earth, they are thought to have evolved multiple times in an instance of convergent evolution. Rhodopsins then use the energy from the photon to provide some useful function for the host such as moving ions across a gradient or relaying visual information. They are, by my knowledge, the world's smallest solar energy collectors.

FIG. 1.

Crystal structure of archaerhodopsin-3. Structure taken from Bada Juarez et al.¹ Seven alpha helices are shown in green that span the plasma membrane. The retinal chromophore is attached to a lysine residue through a Schiff base shown in orange. The electric field in the plasma membrane exerts a force on a proton near the chromophore that tunes the optical properties that, in turn, relays the membrane voltage.

The main characteristic of rhodopsins that inspired the optogenetics revolution was that they couple light and ionic gradients. In channel rhodopsin, the energy from an absorbed photon induces a change in the shape of the protein that allows ions to flow, changing the voltage in the cell. Thus, by application of light, one can actuate cellular voltage and, in the case of neurons and cardiomyocytes, induce action potential firing. Halorhodopsin, another member of the rhodopsin zoo, uses the absorbed photon energy to pump a negatively charged chloride ion into the cell, creating a more negative voltage. This protein will then induce a more negative voltage (hyperpolarization) upon the application of light, and inhibit action potential firing in neurons and cardiomyocytes. Given these published data, how would we convert the actuators into sensors?

The same characteristic of electrical light coupling that gave rise to optogenetics is what led us to our first proposal for a rhodopsin voltage sensor. Proteorhodopsin, my favorite rhodopsin from the Rothschild laboratory, uses light to pump protons out of the cell. However, we knew that the optical properties (color) of proteorhodopsin are tuned by the distance of a proton to the retinal chromophore. The scheme to create a sensor relied on the fact that cells induce extremely large electric fields across the cell membrane. A proton is a charged particle and will accordingly feel a force from an electric field. If the force is large enough, the proton will move and change the optical properties of the protein. Thus, if our sensor changes colors, we will know that the voltage in the cell changed (Fig. 2). As Adam remarked, our scheme worked beautifully in Powerpoint, so we decided to press forward.

FIG. 2.

“Blinking” Escherichia coli circa 2011. The rhodopsin fluorescence from individual cells is highly variable and spontaneously active in glucose containing medium. Each panel is a single cell imaged for 50 s, followed by ∼6 min in the dark. These traces show the varied types of voltage transients in bacteria as compared with the stereotyped action potential of neurons and cardiomyocytes. A higher fluorescence intensity corresponds to membrane depolarization (closer to 0 mV).

Step 3: The Egg and the Chicken

As trained physicists and aspiring biologists, both Adam and I decided that a reasonable place to start was cloning proteorhodopsin mutants in Escherichia coli and testing optical changes with pH response. A now trivial piece of data speaks of our journey down this path, and the nature of scientific research. During our quest to prove the voltage sensitivity of rhodopsins, we came across gramicidin as a means to porate and depolarize bacteria. The experiment was simple—image rhodopsin fluorescence from E. coli, add in gramicidin, and record the increase upon depolarization. Indeed, we saw transient flashes within the cells. However, when we sat down and stared at the videos very carefully, the “flashes” clearly came from aggregated particles floating in solution and were not cell intrinsic. This was a failure. Within a week after the gramicidin experiments, we were convinced that we could see fluorescence intensity changes upon changes in environmental pH. As an aspiring biologist, however, I had been running all my experiments in phosphate-buffered saline. Adam had another clever insight, perhaps we should add to the imaging medium? With that insight, we observed the fluorescence intensity “blinking” in E. coli (Fig. 2) and, after a few days, narrowed down a strong pH dependence of the blinks as well as their dissipation by a proton ionophore.² Upon that now obvious suggestion, we were knee deep into bacterial electrophysiology. The second piece of advice that has helped me dramatically and relayed by this story—watch as much of your video data with your eye before trusting image processing; your brain is an astounding neural net and you can observe subtle differences such as particles floating in solution well before a complicated segmentation and extraction algorithm.

The rhodopsin fluorescence transients in E. coli were a chicken/egg problem. We thought we had an optical voltage sensor, and we thought bacteria were showing something that resembled action potentials, but how do we prove one or the other? We could not use patch clamp on live bacteria due to the outer membrane, and proteorhodopsin did not express in mammalian cells. Our goals were clear, but we needed clever ideas and hard work. It was at this time that Daniel Hochbaum (physicist) and Adam Douglass (biologist) joined the project, and together, they helped supply the clever ideas, incredible tenacity, and tolerance for failure that would push us over the finish line. Ultimately, we were able to use paperclip electrodes in a flow cell to induce a transient voltage to measure fluorescence and voltage, proving we had a sensor. A series of control experiments with other dyes proved the transients were indeed voltage driven. The clarity of hindsight has taught me that without Daniel and Adam, I would never have succeeded. And so I offer yet another piece of advice—recognize that science is mostly failure, but that having help and emotional support from colleagues can keep a smile during the failures and make the successes ecstatic. Despite all of the experiments that never worked, the setups built and taken apart with no real data collected, the late nights and bright days in dark optics rooms, Daniel and Adam made this time fun for me.

Step 4: Finding the Biology in Bioelectricity

After dabbling in mammalian cell electrophysiology, I started my own laboratory at the University of Colorado—Boulder, with one of our foci on uncovering physiological roles of bacterial electrophysiology. There were several interesting questions to ponder, including, why would a bacterial cell ever want to depolarize? Voltage, a component of the proton motive force, is required to power the F1F0-ATPase, and any time spent depolarized is time spent not making adenosine triphosphate (ATP). Then we realized that our bacteria are the spoiled brats of the prokaryotic world in that they want for nothing. They were grown in Luria-broth and glucose at the exact optimal temperature for growth, so if they spend a little time not making ATP, they are totally fine. The changes in voltage could then be used in a time-sharing scheme to send other cellular signals. We hypothesized that the voltage was being used to send important signals, but our problem was figuring out the right circumstances where a bacterium would want to use electrophysiology as a signal.

Giancarlo Bruni joined my laboratory as a graduate student specifically to investigate bacterial electrophysiology, and his dedicated work in the laboratory pushed the importance of bacterial electrophysiology into our consciousness. With his previous training in zebrafish neuroscience, he put 2 and 2 together and started to think about voltage as the start of a signal. In neurons and cardiomyocytes of metazoans, voltage depolarization always precedes calcium influx, so he asked whether cytoplasmic calcium also changed with bacterial depolarization. Sure enough, with the help of very good fluorescent calcium sensors,³ we were able to show that E. coli are bona fide electrically excitable cells.⁴ All of the intense electrical processing of the nervous system was now actually invented in bacteria.

We still had the trouble of finding biological relevance to voltage and calcium transients, and so Giancarlo threw the kitchen sink of environmental conditions at E. coli to see perturbed electrophysiology. Through all the kitchen sink experiments, Giancarlo kept telling me that he was seeing more transients when the cells were sandwiched under agarose as compared with when they were adhered with poly-l-lysine (PLL). I shooed him away and told him that when I was a whippersnapper graduate student, I saw transients with PLL all the time. Finally, he took the advice of his instinct instead of mine, and he performed an experiment that proved to me his instinct was correct. Giancarlo took a movie of cells adhered to glass with PLL, and then very gently applied the agarose pad on top. The result was incredible—bacteria lit up with calcium transients by this change in the mechanical environment. With his movie in hand and me finally believing, Giancarlo finished out all the necessary measurements to show that E. coli recognize their mechanical environment (being squished), and they use voltage and calcium to relay that signal. The last piece of advice I give with a touch of hesitation, though I paraphrase from Linus Pauling—don't always trust the intuition of your mentors. Our work with mechanosensation only occurred because Giancarlo had the insight to perform the experiments rather than just take my word.

A second condition that dramatically altered voltage and calcium in bacteria was the addition of antibiotics. This made good sense to us given the tight coupling between ATP and ion pumps and transporters; any compound that affects a process consuming large amounts of ATP (DNA synthesis, protein synthesis, etc.) would be likely to affect the ionic homeostasis. The high brightness of the GCaMP calcium sensor enabled us to record calcium homeostasis over long periods. The results of adding aminoglycosides, a bactericidal protein translation inhibitor, shocked even us (Fig. 3). The small calcium fluctuations gave rise to massive swings from high to low calcium in the presence of the antibiotic. Watching the movies of treated cells was so mesmerizing, we had to delve into the mechanisms that led to this phenotype. Again Giancarlo delved into the experiments and literature, and we were able to show that voltage is critical not for the uptake, but only for the killing of bacteria by aminoglycosides.⁵

FIG. 3.

Calcium homeostasis is disrupted upon treatment with aminoglycosides. Each color is the GCaMP fluorescence trace over time from an individual cell from untreated (left) or treated (right) Escherichia coli. Kanamycin or control was added at time t = 0. GCaMP.

Steps 5 to Many: Future Experiments in Bacterial Electrophysiology

Voltage in bacteria has long been known and appreciated, and it has been shown to have roles in energy production, cell division, persistence, biofilm communication, among others. However, I believe the field of bacterial electrophysiology is only beginning to unfold. I believe that new tools, whether optical or electrical, along with a growing appreciation of ionic communication,⁶ will flesh out the role of bioelectricity in bacteria. Some answers that I am especially looking forward to involve the myriad ways that nature has harnessed ionic flux for signaling and cell maintenance, beyond which we are comfortable with in the nervous system. I am curious to know whether voltage is used as a mechanism for communication across kingdoms. In nitrogen-fixing plants, calcium signals from commensal bacteria induce a nuclear response in root cells.⁷ Does our microbiome talk to us through the language of electricity? In the same way that electroceuticals⁸ have the potential to alter medicine, will we be able to actuate bacterial electric fields in a way to enhance desirable properties? Finally, can we trace a clear evolutionary path from bacterial electrophysiology to that amazing system of the human brain? How many properties are unique to us, and how many developed with the single-cell organism? These questions, and all those I have not yet thought to ask, all started from a first step of seeing a blinking bacterium on a microscope.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was funded by the Searle Scholars Program and the National Institutes of Health New Innovator Award (1DP2GM123458) to J.M.K.