Current Biology Magazine Current Biology 25, R733–R752, August 31, 2015 ©2015 Elsevier Ltd All rights reserved R749 A photosensitive degron enables acute light-induced protein degradation in the nervous system Anke Hermann 1 , Jana Fiona Liewald 1 , and Alexander Gottschalk 1, * Acutely inducing degradation enables studying the function of essential proteins. Available techniques target proteins post-translationally [1], via ubiquitin or by fusing destabilizing domains (degrons), and in some cases degradation is controllable by small molecules [2–4]. Yet, they are comparably slow, possibly inducing compensatory changes, and do not allow localized protein depletion. The photosensitizer miniature singlet oxygen generator (miniSOG), fused to proteins of interest, provides fast light-induced protein destruction, e.g. affecting neurotransmission within minutes [5], but the reactive oxygen species (ROS) generated also affect proteins nearby, causing multifaceted phenotypes. A photosensitive degron (psd), recently developed and characterized in yeast [6], only targets the protein it is fused to, acting quickly as it is ubiquitin-independent, and the B-LID light-inducible degron was similarly shown to affect protein abundance in zebrafish [7]. We implemented the psd in Caenorhabditis elegans and compared it to miniSOG. The psd effectively caused protein degradation within one hour of low intensity blue light (30 μW/mm²). Targeting synaptotagmin (SNT-1::tagRFP::psd), required for efficient neurotransmission, reduced locomotion within 15 minutes of illumination and within one hour behavior and miniature postsynaptic currents (mPSCs) were affected almost to the same degree seen in snt-1 mutants. Thus, psd effectively photo-degrades specific proteins, quickly inducing loss-of-function effects without affecting bystander proteins. The psd, fused to the carboxyl terminus of the target protein, combines the light-switchable LOV2 domain with a carboxy- terminal degradation sequence from mouse ornithine decarboxylase (cODC; Figure S1A in Supplemental Information, published with this article online). The cODC contains a Cys–Ala motif for proteasomal recognition [8]. Upon illumination, the LOV domain undergoes conformational changes, unmasking the cODC. We fused red- fluorescent proteins, either with psd, or with an inactive version harboring a mutated cODC(C3A). Fusion proteins were expressed in body wall muscle, alongside GFP, for normalization and to control for photobleaching. Soluble tagRFP::psd or mCherry::psd fusions and GFP were evenly distributed (Figure S1B,E,G; top); however, blue illumination (470 nm, 30 μW/mm², 1 h) largely diminished tagRFP and mCherry fluorescence, leaving dot-like accumulations (Figure S1B,G; bottom, insets), while GFP remained unaffected and uniform, just as controls kept in dark (quantification: Figure S1C,D,F,H; illumination up to 6 h did not further reduce tagRFP). The mutated psd(C3A) did not induce degradation (Figure S1E–F); instead, it resulted in higher tagRFP signal, indicating that the psd may be partially active in the dark [6]. Our data show that the psd functions in C. elegans, enabling light-induced depletion of soluble proteins following one hour of illumination. Next, we explored whether the psd could deplete a membrane protein acting in an immediately observable neuronal process. Previously, light- induced inactivation of synaptotagmin, the Ca 2+ sensor for fast synchronous release of synaptic vesicles (SVs) [9], mediated by miniSOG and targeting the SV fusion machinery, affected behavior and synaptic transmission [5]. We expressed SNT-1::tagRFP::psd pan-neuronally, generating punctate fluorescence throughout the nervous system; this was significantly diminished (~75%; Figure S1I–J) by illumination (470 nm, 30 μW/mm², 1 h). We assessed effects of acute SNT-1 degradation by a pharmacological assay reporting on acetylcholine (ACh) release from motoneurons. The ACh- esterase inhibitor aldicarb induces progressive paralysis, which is delayed Correspondence in neurotransmission mutants [10]. We compared snt-1(md290) nulls, expressing SNT1::tagRFP::psd, and wild-type animals expressing SNT- 1::miniSOG::Citrine [5] (Figure S2A,B). Following illumination (100 μW/mm², 470 nm, 30’), animals expressing SNT- 1::miniSOG showed significant aldicarb resistance (Figure S2C), unlike dark controls. The snt-1(md290) mutants showed aldicarb resistance, while the md290; SNT-1::tagRFP::psd rescue strain was paralyzed like the wild type; illuminating this strain (30 μW/mm², 470 nm, 1 h) induced aldicarb resistance essentially as in snt-1(md290) (Figure S2D). Thus, psd-induced SNT-1 depletion affected neurotransmission just as the absence of endogenous SNT-1, and was comparable to miniSOG-induced inactivation of SNT-1 (and bystanders). Synaptotagmin function is required for normal locomotion, allowing acute effects of its loss to be assessed. Thus, we analyzed swimming locomotion of wild-type; SNT-1::miniSOG, or md290; SNT-1::tagRFP::psd animals, in a time-dependent manner during illumination (470 nm, 100 μW/mm² for miniSOG, and 30 μW/mm² for psd; Figure 1A,B; also higher light intensities were tested, but 30 μW/mm² sufficed; Figure S2E–G). In addition to mild exhaustion of animals due to prolonged swimming, all illuminated animals showed largely reduced locomotion, leading to complete paralysis for SNT-1::miniSOG within 30 minutes; for SNT-1::tagRFP::psd rescue animals, the thrashing rate slowed to the level of md290 within one hour (Figure 1A,B); illumination up to four hours evoked no stronger effects. Recovery was observed after 16 hours in dark (Figure 1C,D), indicating replacement of inactivated/ degraded proteins. While miniSOG also affects bystander proteins (here, the neurotransmission machinery [5]), psd only affects the tagged protein, as demonstrated in a control experiment: when wild-type animals, expressing SNT-1::tagRFP::psd and endogenous SNT-1, were illuminated (Figure S2H), locomotion was unaffected. This feature of psd, i.e. targeting only the fusion protein, is a distinctive advantage over miniSOG, enabling very precise interventions. An issue with psd is that LOV-domain switching is never 100%; thus, achieved protein reduction