Bright ligand-activatable fluorescent protein for high-quality multicolor live-cell super-resolution microscopy

We introduce UnaG as a green-to-dark photoswitching fluorescent protein capable of high-quality super-resolution imaging with photon numbers equivalent to the brightest photoswitchable red protein. UnaG only fluoresces upon binding of a fluorogenic metabolite, bilirubin, enabling UV-free reversible photoswitching with easily controllable kinetics and low background under Epi illumination. The on- and off-switching rates are controlled by the concentration of the ligand and the excitation light intensity, respectively, where the dissolved oxygen also promotes the off-switching. The photo-oxidation reaction mechanism of bilirubin in UnaG suggests that the lack of ligand-protein covalent bond allows the oxidized ligand to detach from the protein, emptying the binding cavity for rebinding to a fresh ligand molecule. We demonstrate super-resolution single-molecule localization imaging of various subcellular structures genetically encoded with UnaG, which enables facile labeling and simultaneous multicolor imaging of live cells. UnaG has the promise of becoming a default protein for high-performance super-resolution imaging.


Derivation of Analytical Solutions for On-switching Kinetics
The on-switching rate (kon) was extracted from the recovery of fluorescence intensity after complete bleaching (Fig. 2e). Since UnaG can have a number of distinct conformations including two fluorescent states, the fluorescence intensity did not increase in a simple exponential manner. Therefore, we built a model including multiple fluorescent and nonfluorescent states for the fluorescence switching of UnaG to find the on-switching rate constants for the fluorescence recovery (Supplementary Figure 2a).
In our model, two fluorescent conformations of UnaG (holoUnaG1 and holoUnaG2) are bleached by absorption of blue light via photo-oxidation of bilirubin (BR) bound on UnaG, and produce a nonfluorescent product (damUnaG) that is a complex of UnaG protein and photo-oxidized BR (OxBR). Dissociation of OxBR leaves apoUnaG alone with empty binding pocket for other fresh BRs. Finally, apoUnaG is transformed to the holoUnaG1 upon the binding of fresh BR to recover the fluorescence, and holoUnaG1 spontaneously transits to holoUnaG2.
Here, kda is the rate constant of the dissociation of OxBRs from damUnaG, and kah is the rate constant of the binding of the BR to apoUnaG. kh and kh ' are the rate constants of spontaneous transitions between holoUnaG1 and holoUnaG2.
A graphical scheme for all the possible reactions in Supplementary Equation 1-4 is presented in Supplementary Figure   2a. Then, the rates of formation/dissociation of each population can be described as follows. Note that the fluorescence recovery becomes a bi-exponential growth if the kda is much larger than 1 s -1 at the simulated time scale and resolution (and also at the experimental time scale and resolution), whereas a sigmoidal-like behavior is observed with slower dissociation rate. In formal case, kda can be treated as the infinite, and Supplementary Equation 13 is reduced as follow. on-switching rate constants (kon = kah ' ) by assuming that the dissociation rate of OxBR is much faster than 1 s -1 and is hard to precisely measure with our time resolution (1 s).  under strong blue light irradiation, interestingly, when both of GLOX and BR are included in the buffer solution. One of possible mechanism is that the excess BR absorbs the blue light instead of the cells, and the GLOX helps to reduce the cytotoxic reactive oxygen species (ROS) by depleting the dissolved oxygen and also by depleting the hydrogen peroxide. In this condition, the dead cells showed both symptoms of freezing and apoptosis. g Summary for the phototoxicity in various buffer compositions using the fraction of live cells. In our SML imaging conditions, most the cells can survive more than 3 hours under <15 min of strong 488-nm illumination. h ROS generation by the 300 W cm -2 of 488-nm laser irradiation in various buffer compositions measured by a fluorescent ROS indicator. A small area in a UnaG-expressed live Cos7 cell was illuminated by the blue light for 10 min, and the relative fluorescence intensity of the ROS indicator was measured from an unilluminated area of the same cell before and after illumination. While BR did not affect the ROS concentration, GLOX clearly decreased the concentration of ROS. Error bars: standard deviations (n = 7 individual measurements for a; 9 individual measurements for h).. Figure 19 SML imaging of clathrin-coated pits (CCP) with UnaG-CLC with endogenous concentration of bilirubin without supplementing bilirubin in the imaging buffer. a CCPs in a fixed Cos7 cell whose ring shapes were resolved. The image was reconstructed from 20,000 frames acquired at 10 ms exposure time for total 200 s. b CCPs in a live Cos7 cell. The image was reconstructed from 800 frames acquired at 50 ms exposure time. The long acquisition time of total 40 s enough for the pits to move and to change shape in live cells resulted in motion blur.

Supplementary
To visualize the movement and evolution of pits, we temporally color-coded the SML image as indicated in the color bar underneath the SML image. In particular, the pit in the lower left corner of the SML image showed color rainbow from blue on the left to red on the right, indicating directional motion from the left to the right. Scale bars: 500 nm for a and b; 100 nm for the inset in a.