Summary: In this paper, they're interested in finding new optical actuators for controlling the membrane polarization of the cell (and particularly neurons). they begin by investigated a class of class of rhodopsins from cryptophytes, as opposed to previous channelrhodopsins which have been obtained for chlorophytes (green algae). Cryptophytes are also algae, but in a different kingdom (chromalveolata) than green algae (archaeplastida), thus they're probably quite significantly diverged.
They focus very quickly and with little justification (probably not their fault, Science doesn't provide much space) on Guillardia theta, a cryptophyte alga with a fully-sequenced genome. In this genome, there are 53 proteins sharing similiarity to microbial rhodopsins.
They then equally quickly focus on 3 of these 53 hits (why? possibly due to higher similarity to channelrhodopsins? I'm not clear on this), for which they express in HEK293 cells. It's not clear how much of the protein they actually express, but their construct includes at least the 7-transmembrane domains of three G. theta proteins. In -60mV-clamped cells, two out of three of these generate currents when illuminated. They call these two proteins GtACR1 and GtACR2.
Interestingly, GtACR1 and GtACR2 generate ~4x more photocurrent than ChR2, under the same illumination conditions in -60mV-clamped HEK293 cells, and the photocurrent is inward. This confused me for a little while, because I knew they were going to go on to claim that these were anion channels, and my usual expectation is that the predominant anion (Cl-) has an equilibrium potential around -80 to -90mV. But that depends both on the concentration of Cl- in the bath and in the cell, and under their experimental conditions, these two were close to equal (yielding an equilibrium potential much close to 0 mV). Thus when clamping at -60mV, one would expect an anion-selective channel to have a reversal potential closer to 0 and thus a net inward (depolarizing) current upon illumination, not an outward (hyperpolarizing) current.
Okay, now back to the key points about GtACR1 and GtACR2. In HEK293 cells, they generate near half-maximal current at illumination intensities between 0.01 and 0.1mW/mm^2 (more sensitive than Chronos and significantly more sensitive than Chrimson). The wavelength of this light doesnt seem to be stated though, and it matters because these two rhodopsins have distinct spectral peaks - GtACR2 is maximally sensitive to 460nm light and basically doesnt respond past 550nm, whereas GtACR1 responds maximally at about 510 and doesn't respond to light beyond 620nm.
To show whether these are anion or cation channels, they then just vary the bath ionic conditions. Changing the bath concentration of an ion changes its equilibrium potential, which will change the reversal potential of the channel if the channel conducts that ion. To show these are not cation channels, they show that replacing any cation with N-methyl-glucamine (a large impermeable cation) does not shift the reversal potential. In contrast, changing anion concentrations (and replacing the lost ions with aspartate) drastically changed the reversal potentials for all the anions tested, including larger anions like nitrate. However, neither channel changed its reversal potential in response to changes in sulfate, which could be due to either steric exclusion or that the channel does not conduct divalent anions.
Finally, they put the constructs into actual neurons, and compare to Arch (a proton pumping archaerhodopsin that pumps protons out of cells and thus hyperpolarizes them) and ChloC, a chloride conducting channelrhodopsin. GtACR2 produces significantly more current at the same light levels, and requires drastically lower light (2-3 orders of magnitude) to achieve the same outward current than Arch and ChloC. It's also much faster than ChloC, and because it has significantly less cation permeability than ChloC, it has a significantly more negative reversal potential, enabling hyperpolarization even for already-somewhat-hyperpolarized neurons.
Questions left over:
- what about those other 50 proteins? Or other cryptophytes outside of Guillardia theta?
- how much further could these be optimized by mutagenesis and rational engineering? should we be expecting the next iteration of these actuators soon?
- what domains of the original genes were left out of GtACR1/2?