Structure-guided design and functional characterization of an artificial red light-regulated guanylate/adenylate cyclase for optogenetic applications

Abstract

Genetically targeting biological systems to control cellular processes with light is the concept of optogenetics. Despite impressive developments in this field, underlying molecular mechanisms of signal transduction of the employed photoreceptor modules are frequently not sufficiently understood to rationally design new optogenetic tools. Here, we investigate the requirements for functional coupling of red light-sensing phytochromes with non-natural enzymatic effectors by creating a series of constructs featuring the Deinococcus radiodurans bacteriophytochrome linked to a Synechocystis guanylate/adenylate cyclase. Incorporating characteristic structural elements important for cyclase regulation in our designs, we identified several red light-regulated fusions with promising properties. We provide details of one light-activated construct with low dark-state activity and high dynamic range that outperforms previous optogenetic tools in vitro and expands our in vivo toolkit, as demonstrated by manipulation of Caenorhabditis elegans locomotor activity. The full-length crystal structure of this phytochrome-linked cyclase revealed molecular details of photoreceptor-effector coupling, highlighting the importance of the regulatory cyclase element. Analysis of conformational dynamics by hydrogen-deuterium exchange in different functional states enriched our understanding of phytochrome signaling and signal integration by effectors. We found that light-induced conformational changes in the phytochrome destabilize the coiled-coil sensor-effector linker, which releases the cyclase regulatory element from an inhibited conformation, increasing cyclase activity of this artificial system. Future designs of optogenetic functionalities may benefit from our work, indicating that rational considerations for the effector improve the rate of success of initial designs to obtain optogenetic tools with superior properties.

Keywords: adenylate cyclase (adenylyl cyclase); adenylyl cyclase; allosteric regulation; bacteriophytochrome; guanylate cyclase (guanylyl cyclase); hydrogen-deuterium exchange; light regulation; optogenetics; photoreceptor; photosensor.

Figure 1.

Domain architecture and construct design. A, domain architectures of the DrBphP bacteriophytochrome (PAS-GAF-PHY) with the natural histidine kinase effector (HisK) and the guanylate cyclase Cya2 (GC), which is attached to a light–independent CHASE2 extracellular sensory domain, separated by a transmembrane domain (TM). The cyclase core is N terminally preceded by a CTE. The C terminus is explicitly shown to point out the difference between the initial series of constructs featuring the phytochrome fused to the C terminally-truncated cyclase (PagCΔC/PaaCΔC) and the final series including the native C terminus (PagC/PaaC). PaaCΔC and PaaC are fused to a Cya2 E488K variant exhibiting AC activity. B, left: screening of PaaC constructs in cyclase-deficient E. coli on LB agar supplemented with X-Gal under red light illumination (L) and in the dark (D). Green coloration of the colonies indicates adenylate cyclase activity. Right: amino acid sequence spanning the fusion point between phytochrome (green) and cyclase (blue) of each of the tested constructs. Underlined residues highlight the generally hydrophobic a and d positions in the (abcdefg)-heptad coiled-coil repeats.

Figure 2.

UV-visible spectra of light–regulated constructs and activity plots of PagC constructs. A–C, UV-visible absorption spectra of PagC, PagC-1, and PagC+7, respectively. Full lines show Pr state spectra, dashed lines correspond to the Pfr spectra. The dot-dashed trace for PagC+7 shows a Pfr spectrum from 500 to 900 nm that was recorded in the presence of constant 660 nm illumination with a CCD-based spectrophotometer to minimize thermal reversion to Pr during data acquisition. D, dark-state recoveries of the three constructs indicated in % Pfr over the course of 6 h. E, enzymatic activities of PagC, PaaC, and cyclase-only variants. White bars indicate light-state activity and black bars show the activity in the dark-state. Hatched bars correspond to the cyclase-only variants. Brackets above the bars show the fold-difference between light/dark activities as well as an indication for the substrate specificity. Initial rates were measured at 1 mm GTP/ATP for different time points in a time frame where overall substrate conversion was below 10%. Error bars show the error of the estimate for linear fits of cNMP production weighted by the reciprocal of the square of the standard deviation from three experimental replicates.

Figure 3.

Structure of PaaCΔC in the Pr state. A, overview of the crystal structure of PaaCΔC in cartoon representation with individual domains colored according to the scheme in Fig. 1A. The biliverdin chromophores (orange) and their covalent attachment sites (Cys-24) are shown in stick representation. B, close-up view of the linker-CTE region with residues involved in coiled-coil formation and in packing of the short CTE helix highlighted in stick representation.

Figure 4.

Comparison of the PaaCΔC cyclase domain to other cyclase structures. A, superposition of various cyclase structures crystallized with their respective CTEs shows that the cyclase core dimer in PaaCΔC aligns well with all other structures (blue, PaaCΔC; yellow, PDB 4CLF (human soluble AC); orange, PDB 5M2A (bPAC from Beggiatoa sp.); green, PDB 5O5K (Cya from Mycobacterium intracellulare)). Structures were aligned by superposition of all chains “A” and 4CLF residues 1–263. B, close-up of the linker-CTE region from the cyclase structures shown in A. Unlike in other cyclases, the short CTE helix is “folded back” and residues in the loop preceding the short helix interact with the ^GCβ4–β5 tongue, which is displaced compared with other structures. 4CLF residues 236–262 are not displayed and clipping planes have been adjusted for clarity.

Figure 5.

HDX-MS analysis of PagCΔC reveals changes in conformational dynamics upon red light illumination. Left, PaaCΔC structure colored according to the differences in deuterium incorporation of ΔDrel upon red light illumination (Drel, light; Drel, dark) after 10 s of deuterium exchange. Red coloration indicates an increased, blue a reduced deuterium incorporation, respectively, as indicated by the scale bar at the top left. Right, relative deuterium uptake (D_rel) of representative peptides from the ^GCβ4–β5 tongue (a), the linker helix spanning the fusion point between PHY domains and CTE (b), and the PHY tongue region (c). The corresponding sequences are shown on top of the plots with absolute and construct-specific numbering. Data are shown as the mean of three independent measurements and error bars correspond to the standard deviation. The lower parts depict abundance distributions of deuterated species ranging from undeuterated to fully exchanged amide protons in the respective peptides.

Figure 6.

PaaC application in vivo is comparable with IlaC22 k27 but has slightly reduced “dark-state” activity. A, two independent lines each were created where IlaC22 k27 or PaaC were expressed in cholinergic motor neurons (red arrow indicates the ventral cord; white arrows indicate the autofluorescence of the intestines). B, behavioral analyses of worm activity were performed using the WorMotel system. Activity was quantified during illumination with nonactinic green light and activating red light where the light regime consisted of 5 min of green light followed by 5 min of red light in two rounds. WT worms did not respond significantly to the altering light conditions, whereas the transgenic lines expressing IlaC22 k27 or PaaC were significantly more active under red light. Additionally, the IlaC22 k27-expressing strains but not the PaaC-expressing strains were significantly more active under the initial green light exposure compared with WT animals (*, p < 0.05, WT n = 108, IlaC22 k27 n = 116, PaaC n = 216 animals, 18 trials). A time lapse movie of frames from 10-s intervals is provided in Movie S1. C, to measure dark-state activation we quantified locomotor activity under green light for an extended period of time (300 min) and found that both IlaC22 k27- and PaaC-expressing strains were more active than WT and not significantly different from one another (**, p < 0.001, WT, n = 18; IlaC22 k27, n = 18, PaaC n = 32 animals, 3 trials).