Genetically-Encoded Neuronal Indicator and Effector (GENIE) Project
Transcription
Genetically-Encoded Neuronal Indicator and Effector (GENIE) Project
Genetically-Encoded Neuronal Indicator and Effector (GENIE) Project Steering Committee (2015-2016) Reed George, Senior Director, Scientific Services Vivek Jayaraman, Group Leader Douglas Kim, Program Scientist Wyatt Korff, Associate Director and Program Scientist, Team Projects Loren Looger, Group Leader Nelson Spruston, Scientific Program Director and Laboratory Head Karel Svoboda, Group Leader Research Staff Stephan Brenowitz, Senior Scientist (voltage sensor development) Hod Dana, Research Specialist (mouse imaging) Benjamin Fosque, Research Technician (protein biochemistry) Jeremy Hasseman, Research Specialist (molecular cloning) Bei-Jung Lin, Research Specialist (mouse imaging) Yi Sun, Research Specialist (Drosophila imaging) Getahun Tsegaye, Research Specialist (molecular cloning) Collaborating Scientists Adam Cohen, Associate Professor, Department of Chemistry and Chemical Biology, Harvard University (voltage sensor development) William Schafer, Group Leader, Medical Research Council, Laboratory of Molecular Biology, Cambridge University (C. elegans imaging) Eric Schreiter, Senior Scientist, Looger Lab (calcium integrator development) 1 1.0 Summary For more than 15 years, fluorescent proteins have been engineered to report neural activity. Neural circuits process information on spatial scales ranging from single synapses to large assemblies of neurons at time scales from milliseconds to months. Imaging based on fluorescent protein sensors spans all of these temporal and spatial scales. Although protein sensors are now routinely used to image neuronal activity in vivo, they still have major limitations. The GENIE project develops prototype protein sensors into the best-in-class tools for in vivo neurophysiology. Our focus is on high-impact tools that will benefit from protein engineering and large-scale screening technology. The project will continue to optimize green and red calcium indicators. Red indicators facilitate deeper imaging in vivo and multi-color imaging. We have also initiated a program to engineer genetically-encoded voltage sensors and switches that detect the coincidence of light triggers and calcium. 2.0 Background The goal of the GENIE Project is to engineer fluorescent sensors for imaging of neuronal activity in the intact nervous system. Understanding the function and plasticity of neural circuits requires measurements of neural activity over a large range of spatial and temporal scales. Calcium indicators have been especially useful for neurobiology. In most neurons, action potentials (APs) are tightly coupled to large (20-fold) and rapid (rise time <1 millisecond) increases in intracellular free calcium concentration. These calcium transients can be measured as fluorescence changes in neuronal somata (1, 2), dendrites (3, 4), and axons (5, 6) in vitro and in vivo as a proxy for the underlying electrical activity. Calcium indicators are also useful in non-spiking, graded potential neurons (7, 8), although optimal indicator parameters may differ. Additionally, excitatory synaptic transmission is typically associated with large-amplitude (20- to 100-fold) increases in calcium concentration in small synaptic compartments (e.g. dendritic spines). These synaptic calcium transients can be used to quantify synaptic transmission in vitro (9) and in vivo (10, 11). Calcium 2 imaging therefore provides a versatile tool to probe neural activity over timescales of milliseconds to months and spatial scales of micrometers to millimeters. Genetically-encoded calcium indicators (GECIs) occupy a niche complementary to single unit electrophysiology. Single unit methods have better sensitivity and time resolution. A major goal of the GENIE Project is to optimize GECIs to achieve signal-to-noise ratios (SNRs) that rival electrophysiological methods. We have developed the general-purpose GFP-based GCaMP6 sensors (11), which under favorable conditions detect single spikes in neuronal populations in vivo. GECIs with even higher sensitivity will allow larger populations of neurons to be imaged more rapidly and under more challenging conditions (e.g. during behavior). High sensitivity comes at the price of limited dynamic range; high-sensitivity indicators saturate at modest activity levels, making them unsuitable for monitoring activity patterns in neurons with high spike rates (~100 Hz). Existing GCaMP6 indicators are not sufficiently sensitive to measure single APs in some cell types, such as parvalbumin-positive interneurons in rodents. Beyond GCaMP6, we will develop a suite of green indicators tailored to specific applications, including imaging of large populations of neurons, single-spike detection across neuron types, measurement of spike rates over a wide dynamic range, and imaging spike times with high-speed indicators. We are also developing red fluorescent GECIs (12, 13). There are four major reasons why red indicators are of particular interest. First, longer wavelength excitation light penetrates deeper into tissue. In addition, red fluorescence is absorbed much less than green fluorescence, especially in mammalian tissue, giving red fluorophores a competitive advantage for in vivo imaging. Red GECIs (R-GECIs) therefore promise deeper and less invasive imaging compared to green GECIs. Second, many transgenic animals with green fluorescence have been created; RGECIs can be used in these animals. Third, R-GECIs are more easily combined with channelrhodopsin-2 (ChR2)-based manipulations of neural circuits. Fourth, combinations of red and green GECIs promise simultaneous imaging of multiple distinct neural structures (14). We will thus develop a suite of sensitive R-GECIs. Calcium imaging provides only an indirect report of neural activity. Electrical measurements are required to understand how neuronal input is converted to neuronal output, the heart of neuronal 3 computation (15). Calcium is not an accurate proxy for subthreshold electrical dynamics of neurons. Electrode-based measurements of membrane potential have been central to neuroscience research for many years, but microelectrodes are invasive and difficult to use. A long-standing goal has been to develop imaging methods for membrane potential (16). However, despite several attempts by multiple groups, the best available protein voltage sensors still suffer from small fluorescence changes in response to APs, low brightness, poor targeting to the plasma membrane, large capacitive load on the membrane, and/or slow response kinetics (17-20). Existing GEVI scaffolds require substantial improvement before they can be used to study neural coding in vivo. We have begun a program to develop genetically-encoded voltage indicators (GEVIs) for subcellular imaging. A major goal of neuroscience is to identify neural ensembles, groups of neurons that are coactive during specific phases of behavior (21). Tools to mark these ensembles in vivo, and to target them for later characterization or manipulation are lacking. Promoters from immediate early genes such as c-fos have been harnessed to trigger transgene expression in active cells, but these methods lack sensitivity, respond with slow kinetics, and are modulated by many physiological parameters other than neural activity (22). The GENIE project has developed switches that detect the coincidence of light and calcium to mark neurons that are active during experimenter-defined periods. We propose to develop similar switches to activate gene expression, either through transcriptional regulation of a target gene or through recombination. In mammalian systems, stable long-term expression of protein sensors remains a major challenge. High protein concentrations are required for in vivo imaging (23)(GECIs, 10-100 micromolar). Viral infection using AAV viral vectors is currently the gene delivery method of choice (24). Although AAV produces sufficient expression, expression levels continue to ramp up over months (6-fold from 1 to 6 months of expression) (11), eventually causing cytotoxicity and changes in neuronal behavior (24). GECIs can be expressed at constant levels over 10 months in transgenic mice, without signs of cytotoxicity (25). However, current schemes (e.g. CAG promoter-GCaMP3 targeted to the Rosa26 gene locus) still fail to provide sufficient expression levels for many applications. The lack of suitable methods for controlled long-term expression might be the single most important impediment to progress in using geneticallyencoded sensors in rodents. We will develop novel transgenic mice for Cre-dependent gene 4 expression, optimized for in vivo imaging. In addition, we will optimize sensors for protein stability, which may provide several benefits: boosting expression levels, reducing cytomorbidity and decreasing levels of inactive sensor, which contributes to fluorescence background but not signal. 3.0 Progress Report (September 2012-present) 3.1 Assay development Underlying the development of GCaMP6 and other advances in GECI development has been a neuronal culture assay for characterizing GECI variants rapidly in the relevant cell type (Fig. 1AC). This was necessary because neurons have fast and small-amplitude calcium accumulations that are difficult to model in non-neuronal systems. Primary neuronal cultures in 96-well plates are transduced with GECI variants and a nuclear fluorescent protein to control for expression level. APs are triggered using extracellular stimulation electrodes, while neurons are imaged using fluorescence microscopy. The output of the assay is the fluorescence dynamics of multiple individual neurons to a variety of trains of APs (26). This assay therefore allows characterization of large numbers of GECI variants under relevant conditions. Over the last three years we quadrupled the throughout of our assay (December 2010: 0.99 variants tested/day; February 2014: 4.5 variants tested/day). A major change is that gene transduction is now accomplished by DNA electroporation into primary neurons instead of viral infection. Transfections are performed using a liquid handling robot. Screening is now routinely performed in 96-well format instead of 24-well format. We mutagenize scaffolds using gene assembly cloning methods (27). Amino acid positions are selected for mutagenesis based on the crystal structures of sensors (28-30). Substitutions are made at single positions, tested in the culture assay, and the best point mutants are combined. Individual, sequenced clones are assayed in multiple wells. The sensitivity of the neuronal culture assay is limited by variability in neuronal physiology. A 100% improvement over R-GECO1.0 or GCaMP3 can be detected for a 1AP response with eight replicates (p<0.01). The screening cost is $36 per variant (without labor). Throughput is limited 5 Figure 1. GCaMP6 engineering. (A) Neuronal culture screening platform. (B) Schematic of stimulation geometry. (C) Expression vector with human synapsin-1 promoter (syn), GECI variant, internal ribosome entry site (IRES), nuclear localization signal fused with fluorescent protein (nls-fluorescent protein), and woodchuck hepatitis virus post-transcriptional regulatory element (WPRE). (D) Optimization of GCaMP for detecting neuronal activity. GCaMP SNR after 1AP was estimated for GCaMP, GCaMP1.3, GCaMP1.6, GCaMP2, GCaMP3, G-GECO1, GCaMP5G, and GCaMP6s. Note that SNR was estimated across different types of assays published by multiple groups and is thus a rough estimate. Data from (2, 11, 24, 47, 56-60). (E) GCaMP structure (29, 30) and mutations (red) found in different GCaMP6 variants. CaM-binding peptide M13 from myosin light chain kinase (M13, yellow), linker 1 (gray), circularly-permuted EGFP (cpEGFP, green), linker 2 (gray), CaM (blue), calcium ions (gray spheres). (F) Screening results for GCaMP6. Top, fluorescence change in response to 1 AP (vertical bars, ΔF/F0; green bar, OGB1-AM, left; black bars, single GCaMP mutations; red bars, combinatorial mutations; blue, GCaMP6 indicators) and significance values for different AP stimuli (color plot). Middle, half decay time after 10 APs. Bottom, resting fluorescence, F0 normalized to nuclear mCherry fluorescence. Red line, GCaMP3 level; green line, GCaMP5G level; blue line, OGB1-AM level. (G) Single AP Responses averaged across multiple neurons and wells for GCaMP3, 5G, 6f, 6m, 6s, and OGB1. (H) GCaMP6 performance in the mouse visual cortex in response to a single spike. Left, schematic of the experiment. Right, median fluorescence change in response to 1 AP for different calcium indicators. Shading corresponds to s.e.m., n = 9 (GCaMP5K, data from (47)), 11 (GCaMP6f), 10 (GCaMP6m), 9 (GCaMP6s) cells. GCaMP5K and GCaMP5G have similar properties. 6 Figure 2. Fly visual system assay. (A) Schematic of the fly visual assay. Female flies (3-5 days old) are glued onto a custom-built pyramid and 2-photon imaged through a window in the cuticle on the head. Visual stimulation is delivered through a modified digital-mirror-device that reflects a dispersed laser beam (532nm, DPSS) onto a pattern on a screen in front of the fly. A signal is also acquired from a silicon photodiode that detects changes in light intensity to synchronize visual stimulation with imaging. Laser light is isolated from GCaMP fluorescence through spectral filtering and physical masking. Inset: Schematic of the optic lobe, including lamina monopolar cell L2, where the axons terminating in the second layer in the medulla are imaged (dashed rectangle); and the lobula plate tangential cell HS, where dendritic trees are imaged (dashed rectangle). (B) GCaMP6f is faster and more sensitive than GCaMP6s and GCaMP5G in L2. L2 is a graded potential neuron that depolarizes in response to light offset. Whole-field flickering with different durations and duty cycles are used as visual stimulation. Gray bars denote light offset. (C) GCaMP6f responses to whole-field flickering visual stimulation flashed at various frequencies. (D) Representative traces of GCaMP6s and GCaMP6f in the HS neuron to moving gratings tilted at different angles, moving at different speeds and with different contrast (single condition shown). In these large non-spiking neurons, GCaMP6s responses are larger than GCaMP6f. GCaMP6f captures the stimulus onset and offset and is more similar to the envelope of the intracellular recording curves of this neuron. (E) Direction tuning of the imaged dendritic area of HS neuron measured with GCaMP6f (blue) and GCaMP6s (red). (F) Frequency tuning of the imaged dendritic area of HS neuron measured with GCaMP6s. (G) Contrast tuning of the imaged dendritic area of HS neuron measured with GCaMP6s. by the cloning rate (7.7 constructs cloned/day for RCaMP and R-GECO; an improvement of a factor of 6 over the GCaMP cloning rate, 1.3 constructs cloned/day). The neuronal culture assay has excess capacity. 7 Fly visual system The GENIE team has validated the performance of GECIs in adult Drosophila by comparing odor responses in projection neurons (PNs) of the fly antennal lobe. Although recording the activity of PNs in specific glomeruli in this early olfactory system assay has been useful for characterizing GECI performance in vivo (11), it has some limitations. First, odor delivery is notoriously difficult to control with precision and consistency. Second, although it is possible to perform cell-attached/loose-patch recordings from PNs, each glomerulus is innervated by multiple PNs, and a single GAL4 driver typically labels most or all of them, making it difficult to rigorously relate glomerular calcium activity to electrical activity in a single PN (31). Thus, we have begun to develop a visual system assay for testing GECIs. We have validated the assay by recording optic lobe responses in fly lamina and lobula plate neurons (Fig. 2). Going forward, we will increasingly focus on spiking neurons, and specifically interneurons similar to H1 and H2 in the optic lobe (32). We have chosen this class of neurons for their large size and tractability for electrophysiology, well-characterized visual responses, and the high dynamic range of firing rates that can be induced with appropriately controlled visual stimulation. We have identified candidate GAL4 lines that target such large neurons and are in the process of characterizing their visual responses using 2-photon imaging and electrophysiology. 3.2 Sensitive and fast GCaMPs We developed the general-purpose GCaMP6 family of indicators with improved SNR for spike detection in vivo (Fig. 1D-H) (11). We characterized close to 500 structure-guided mutations, with as many as 11 combined mutations in GCaMP3. Many of these mutations were at the interface between GFP and calmodulin (CaM), which was extensively remodeled. We identified GCaMP variants with 3-fold faster rise times and 2-fold decreased decay times. The kinetic improvements are based on a mutation in the M13/CaM interface. This CaM bump combines modularly with sensitivity-increasing mutations. The GCaMP6 sensors show single AP responses that equal or exceed those of Oregon Green Bapta-1 (OGB1), the most widely used calcium indicator. GCaMP6s detects all active neurons in the visual cortex in vivo. GCaMP6s, 6m, and 6f (for slow, medium, and fast) differ in their response kinetics, with 6f having the 8 Figure 3. GCaMPs with varying affinity and cooperativity. (A) Schematic and representative traces of the NMJ assay in Drosophila larvae. In a fillet preparation, motor neuron axons are cut and electrically stimulated with a suction electrode, and type Ib/s boutons are wide-field imaged with an EMCCD camera. Images are segmented and analyzed in response to stimulation. Scale bars, 5 µm. (B) Fluorescence changes (∆F/Fapo) from calcium titrations of purified GCaMP protein variants. (C) Frequency tuning curves for low affinity variants (EF-hand mutants), GCaMP6, and GCaMP5G. The averaged maximum ΔF/F0 at different stimulation frequencies are plotted. Different variants are color-coded according to their rank order of dissociation constant (KD). Note that the low affinity variants are right-shifted and have shallower slopes. (D) Decay kinetics for the low affinity variants, GCaMP6, and GCaMP5G. The low affinity variants are faster than previous GCaMPs including GCaMP6f. (E) Rise kinetics for the low affinity variants, GCaMP6, and GCaMP5G. The low affinity variants are faster than previous GCaMPs including GCaMP6f. fastest kinetics of any useful GECI. In 2013, the GCaMP6 indicators were amongst the most widely requested reagents at Addgene.org (481 GCaMP6s requests as of March 2014; 48th most requested plasmid ever) and the University of Pennsylvania Viral Vector Core. GCaMP6expressing flies were amongst the most requested fly strains at the Bloomington Drosophila Stock Center (148 GCaMP6m requests in 2013; most requested stock out of 37,125 stocks in 2013). Beyond GCaMP6, we have begun to engineer GCaMP6 variants with low resting fluorescence, to maximize SNR for large-scale in vivo imaging in rodents, higher resting fluorescence for applications in flies, and GCaMP6 variants with varying calcium affinities and binding 9 cooperativity, to enable activity imaging in neurons with graded responses and fast-spiking neurons (Fig. 3). 3.3. FRET-based GECIs We have completed a collaboration with Oliver Griesbeck (MPI Martinsried), resulting in the creation of the best-in-class troponin-based FRET GECIs (33). Ratiometric GECIs have advantages for imaging moving preparations, such as C. elegans. We focused on the TN-XXL ratiometric (CFP/YFP) indicator scaffold. TN-XXL uses a calcium-binding domain, troponin, which is orthogonal to CaM in terms of structure and perhaps toxicity profile (34). We screened 121 TN-XXL-based variants from the Griesbeck lab using neuronal culture screening. The resulting ‘Twitch’ sensors have 8-fold greater single AP responses compared with TN-XXL (33). 3.4 R-GECIs We have focused on optimizing two scaffolds, RCaMP and R-GECO, which are based on mRuby (35) and mApple (36), respectively. We first targeted our engineering to improve sensitivity (including SNR and brightness) and kinetics. Overall, the strategy employed is similar to GCaMP6 (11). Libraries of point mutations are being generated in a structure-guided manner (Fig. 4) (12). Targets for mutagenesis include the calcium-binding domains, the binding pocket in CaM for M13, and the interface between CaM and fluorescent protein. More than 700 RCaMP and 500 R-GECO variants have been characterized in the neuronal culture assay thus far. A novel R-GECO variant reliably detects single APs in the cell culture assay (Fig. 5G-L). Some variants have kinetics in the range of GCaMP6f (~25 ms half rise time). Structure-guided engineering of the RCaMP and R-GECO scaffolds has improved the 1AP SNR in cultured neurons by 7- and 5-fold, respectively, thus far (Fig. 4 and 5). Compared to GFP, red fluorescent proteins, and the derived red calcium sensors, suffer from photostability and brightness problems. Under 2-photon excitation, RFPs tend to bleach rapidly or show reversible dark states. The peak brightness per molecule (an experimental quantity proportional to quantum efficiency (QE) and inversely related to bleaching rate) (37) is substantially smaller for R-GECO1.0 and RCaMP1h compared to GCaMP (~33% and ~50%, respectively) (12). In addition, R-GECO1.0 is transiently induced to become red fluorescent after 10 Figure 4. R-GECI variants. (A) RCaMP structure (12). M13 peptide (orange), linker 1 (gray), circular permuted mRuby (red), linker 2 (yellow), and CaM (purple). (B) RCaMP amino acid positions that were mutated (red). (C) Screening results, 700 RCaMP1h variants. Top, fluorescence change in response to 1 AP (vertical bars, ∆F/F0; green bars at right, GCaMP5G, OGB1, GCaMP6s, respectively; black bars, RCaMP1h variants; and significance values for different AP stimuli (color plot). Middle, half decay time after 10 APs. Bottom, resting fluorescence, F0 normalized to nuclear GFP fluorescence. Red line, RCaMP1h level; green line, GCaMP5G level; blue line, GCaMP6s level. (D) R-GECO1.0 structure (12). M13 peptide (orange), linker 1 (gray), circular permuted mRuby (red), linker 2 (yellow), and CaM (purple). (E) R-GECO1.0 amino acid positions that were mutated (red). (F) Screening results, 500 R-GECO1.0 variants. Top, fluorescence change in response to 1 AP (vertical bars, ∆F/F0; green bars at right, GCaMP5G, OGB1, GCaMP6s, respectively; black bars, R-GECO1.0 variants; and significance values for different AP stimuli (color plot). Middle, half decay time after 10 APs. Bottom, resting fluorescence, F0 normalized to nuclear GFP fluorescence. Red line, R-GECO1.0 level; green line, GCaMP5G level; blue line, GCaMP6s level. 11 Figure 5. Performance of selected R-GECI variants in neuronal culture. (A) Responses averaged across multiple neurons and wells for RCaMP1h (black), OGB1 (green), RCaMP1h variant 471 (blue), and variant 692 (red). Fluorescence changes in response to 1 AP. Error bars correspond to s.e.m. (B) Responses to 10 APs. (C) Comparison of RCaMP1h, variants, and OGB1-AM as a function of AP number at 83 Hz (colors as in A). (D) SNR. (E) Half decay time. (F) Half rise time (after stimulus offset). (G) Responses averaged across multiple neurons and wells for R-GECO1.0 (black), OGB1 (green), R-GECO1.0 variant 783 (blue), variant 819 (red), and variant 1027 (magenta). Fluorescence changes in response to 1 AP. (H) Responses to 10 APs. (I) Comparison of R-GECO1.0, variants, and OGB1-AM as a function of AP number at 83 Hz (colors as in G). (J) SNR. (K) Half decay time. (L) Half rise time (after stimulus offset). exposure to blue light, which makes it incompatible with the action spectrum of ChR2. We hope to address these issues over the next year. Despite these challenges, one novel R-GECO variant performs well in mouse V1 neurons (Fig. 6). We expect to release next generation RCaMP and R-GECO variants by the end of 2014. 3.5 GEVIs Over the last fifteen years several different types of fluorescent protein sensors of voltage have been developed. One scaffold is based on archaerhodopsin-3 (Arch) (38). We modified our neuronal culture assay for screening fast voltage sensors. This required illumination with a highintensity laser (638 nm; ~1 kW/cm2) and imaging with a high-speed camera (EM-CCD, 500 fps, 128x128 pixels). A limited set of Arch variants was tested in collaboration with Adam Cohen 12 (Harvard). Wild-type Arch-GFP exhibited a 1AP ΔF/F0 response of 15%, whereas the D95E, Y, and N variants showed responses in the 3-7% range. More importantly, Arch quantum efficiency is tiny, making it very challenging to use in thick specimens. The quantum efficiency is also likely intrinsic to the retinal chromophore and unlikely to be substantially increased by protein engineering. We have therefore abandoned the Arch scaffold. Instead we will focus on fluorescent protein-based GEVIs (17) or perhaps in hybrid Arch-FP sensors (39). 3.6 Calcium-activated and light-dependent switch prototypes We have developed a genetically-encoded, light-gated integrator of calcium concentration for neurons (Calcium-modulated Photoconvertible Ratiometric Integrator, CaMPARI) (Fig. 7). CaMPARI allows the tagging of neurons that are active during a user-specified epoch of illumination (milliseconds to seconds). CaMPARI is a photoswitchable protein (based on EosFP (40)) where the photoswitching has been engineered to be calcium-dependent. CaMPARI survives tissue fixation and can thus be imaged post hoc in histological preparations of fly and other brains (Fig. 7H-K). This allows post hoc, brain-wide mapping of neurons that were active during defined epochs, defined by illumination in vivo. We have made incremental improvements to light-activated Cre recombinase (41) (Fig. 8). Light-activated Cre recombinase could be made dependent on neural activity (e.g. calcium) in the future to provide a powerful tool to switch genes on/off in active neurons. 3.7 Transgenic/Targeted Mice GCaMP6s and GCaMP6f transgenic mice were produced using Thy1-GCaMP6-WPRE (42) and Camk2a-GCaMP6 (43) expression cassettes. Most founders (>25 lines) showed negligible expression. Five Thy1 mouse lines were found suitable for in vivo imaging, with high expression levels in the hippocampus, neocortex, and other structures (Fig. 9). Neurons did not show any cytopathology even after long-term expression of GCaMP6. These mice have been made available at The Jackson Laboratory. Unfortunately, Thy1 mice show frustratingly inconsistent expression patterns across brain regions and neurons. Thus each line is suitable only for a small set of brain areas. A stable, retargetable, and tunable strategy for mouse transgenesis is needed for in vivo imaging. 13 Figure 6. Performance of novel R-GECO variant in mouse visual cortex. (A) Top, schematic of the experiment. Bottom, field of view showing neurons expressing R-GECO1.0 (left) and RGECO variant 850 (right). (B) Example traces from neurons expressing R-GECO1.0. Single sweeps (gray) and averages of sweeps (red) are overlaid. Directions of grating motion (8 directions) are shown above traces (arrows). (C) Example traces from neurons expressing R-GECO variant 850. Single sweeps (gray) and averages of sweeps (black) are overlaid. (D) The fraction of cells scored as responding to visual stimulation when loaded with different R-GECO1.0 and R-GECO variant 850. Error bars correspond to s.e.m. (E) The distribution of fluorescence changes across cells at the preferred orientation. 14 4.0 Research Plan 4.1 Specific Aims for the Next Project Period 4.1.1 GCaMPs (FY2015 commitment: 21%, FY2016 commitment: 16%) • Engineer and characterize 500 additional GCaMP variants. • Screen for high SNR GCaMP variants. • Screen for GCaMP variants with lower affinity and Hill coefficient. • Screen for fast GCaMP variants. • Develop and test low-F0 variants for imaging large volumes in vertebrates. • Develop and test higher-F0 variants for imaging in the fly brain. • Develop a protein stability assay and screen for stable GCaMP variants. • Test the 10 best variants in vivo. 4.1.2 R-GECIs (FY2015 commitment: 32%, FY2016 commitment: 32%) • Engineer and screen 1,000 RFP-based calcium indicator variants. • Screen for sensitive, fast and photostable variants. • Test the 10 best variants in vivo. 4.1.3 GEVIs (FY2015 commitment: 37%, FY2016 commitment: 37%) • Benchmark existing scaffolds. • Establish a high-throughput platform for evaluating protein voltage sensors. • Engineer and screen 100,000 voltage sensors • Test most promising 200 variants in cultured neurons • Test 10 best variants in mouse, Drosophila, and zebrafish in vivo. 4.1.4 Calcium-Activated and Light-Dependent Switch Prototypes (FY2015 commitment: 5%, FY2016 commitment: 5%) • Prototype two light and activity-dependent switches (light-sensitive Cre recombinase and transcription factors). 15 • Combine switches with reporter constructs to control neuronal ensembles (e.g. using ChR2). 4.1.5 Transgenic/Targeted Mice (FY2015 commitment: 5%, FY2016 commitment: 11%) • Engineer mice with high and stable neuronal expression of GCaMP6, RCaMP, and RGECO. 4.2 Experimental Plan 4.2.1 GCaMPs Sensitive GCaMPs The SNR of GECIs is largely determined by the fluorescence change per unit of activity (i.e. an AP, or a synaptic input) (44-46). This fluorescence change, ∆F/F, in turn depends on the brightness of the calcium-bound bright form, the fraction of indicator that is calcium-bound, and the brightness of the calcium-free indicator. Increasing ∆F/F can be achieved by tweaking three parameters: 1) Increase the affinity for calcium. This causes a larger ∆F/F per AP if spike rates are low. Increasing affinity was a major part of the GCaMP6 design strategy (11). One drawback of further pushing in this direction is that small stimuli will bring the indicator close to saturation, reducing the effective dynamic range. 2) Maximize fluorescence of the calcium-bound indicator. Screening for higher QE was also part of the GCaMP5 (47) and GCaMP6 (11) design strategies. However, the brightness of calciumbound GCaMP6 is already higher than GFP and it is therefore unlikely that this parameter will yield additional improvements. 3) Minimize resting fluorescence. Our screen has already revealed linker mutations with reduced resting fluorescence, without a large reduction of the peak fluorescence. In the past we have not focused on these mutations because our strategy was to produce indicators with appreciable resting fluorescence, which is critical for structural imaging. However, transgenic animals now allow imaging in a background in which neurons, nuclei, or other structures of interest are 16 Figure 7. Performance of CaMPARI in vitro and in fly neurons. (A) Schematic of a fluorescent protein that preferentially undergoes 400 nm light-induced green-to-red photoconversion in the presence of high calcium ion concentrations. (B) Schematic of the primary structure of CaMPARI illustrating the domain organization of nuclear export signal (NES), CaM, linker 1 (L1), mEos2 variant C-terminal portion (mEos2-C), circular permutation linker (cpL), mEos2 variant N-terminal portion (mEos2-N), and CaM-binding fragment from myosin light chain kinase (M13). (C) Timecourse of red fluorescence appearance during exposure of CaMPARI to photoconversion light in the presence of saturating calcium (closed circles) or in the presence of calcium (open circles). Lines represent single exponential fits to the photoconversion timecourse. (D) CaMPARI photoconversion rate (calculated as in (C)) as a function of different concentrations of free calcium ions. Black line represents a sigmoidal fit to the data. (E) Green and red (purple false color) composite epifluorescence micrograph (bottom) of primary neurons expressing CaMPARI after a 2 s photoconversion light pulse (top). (F) Same as (D) except that 2 s of 80 Hz field electrode stimulation of the neurons accompanied the 2 s photoconversion light pulse. (G) Quantitation of the red/green fluorescence ratio in neurons photoconverted for 2 s concomitant with 2 s of field electrode stimulation to induce AP firing at different frequencies. The black line is the best linear fit to the data. Values represent means across at least 10 independent measurements, normalized to the group that was not stimulated. (H) Schematics of olfactory assay with CaMPARI in Drosophila. Adult flies are tethered to a holder contained within an air-tight chamber. Odors are delivered through an olfactometer into the chamber and rapidly evacuated . UV photo-conversion light (405nm, ~4mW/mm2) is delivered through the objective onto the head. Odor delivery duration is 1 second, while UV light is delivered in four 500 millisecond windows with a 200 ms interstimulus interval. After the experiment, brains are dissected and imaged by confocal microscopy. (I) Control experiments where UV light illuminates for up to 300 s in total. (J)When fly is presented with phenylacetic acid, VL2a glomerulus is photoconverted into red. (K) When fly is presented with geosmin, DA2 glomerulus is photoconverted into red. 17 labeled with red fluorescence protein, obviating the need for substantial resting fluorescence. Reducing resting fluorescence even by a factor of two will have a substantial impact on SNR, increasing the yield in cellular imaging experiments by several-fold. We believe that this is the most fruitful path forward. We will therefore create GCaMPs with low resting fluorescence for large-scale, high SNR imaging. 200 linker variants of GCaMP6s and 6f will be screened. We aim to engineer variants maintaining the peak fluorescence of GCaMP6s and 6f, while lowering F0 by 5-fold. In flies, where it is common to use Gal4 drivers to target GECIs to sparse neural populations in neuropil (rather than cell bodies), substantial resting fluorescence remains desirable to easily locate and identify neural structures for imaging. Higher resting fluorescence is particularly important in behaving flies, in which high illumination intensity can artificially excite neural tissue and result in behavioral artifacts. In addition, the use of a co-labeling FP (e.g. RFP) complicates the use of multiple GECIs for simultaneous imaging of disparate neural populations. Thus, moderate-to-high resting fluorescence will continue to be a consideration when screening GCaMPs and R-GECIs for use in Drosophila. Stable GCaMPs GCaMP6 sensors have to be used at high concentrations (approximately 10-150 micromolar) for in vivo imaging. These protein concentrations are among the highest normally achieved in the brain and are difficult to mimic in transgenic mice. Instead GCaMPs are typically delivered using viral vectors such as AAV. AAV-mediated expression has undesired consequences. Protein expression increases monotonically with time, ultimately leading to cytotoxic GCaMP concentrations. The high expression levels required interfere with expression of endogenous proteins. Protein concentrations depend on the rate of production and the rate of degradation. To increase protein concentrations without increasing expression, we will engineer GCaMPs with increased protein stability. We anticipate that more stable GCaMPs will allow higher protein concentrations at modest expression rates. This will allow use of transgenic methods for expression and imaging with reduced cytotoxicity. We will test 200 more GCaMP6s and 6f variants to improve protein stability. This will include altering potential ubiquitination sites, improving thermodynamic stability, and optimizing codon 18 usage for mice and flies. Variants will be screened by transient transfection of HEK293T cells, application of a protein synthesis inhibitor (cycloheximide), and then epifluorescence imaging of or fluorescence-activated cell sorting (FACS) to measure GCaMP variant fluorescence and fluorescence of a co-expressed mCherry control. We will also directly measure extent of degradation through the proteasomal pathway, through use of inhibitors such as lactacystin (24). The goal will be to achieve a 5-fold increase in stability leading to a 5-fold increase in protein concentration, while maintaining GCaMP6 SNR levels. The best variants will be tested in cultured neurons and in vivo in mice and flies. Linear GCaMPs In fly larval neuromuscular junction (NMJ) assays, GCaMP6 responses were as much as 3-fold increased over those of GCaMP5G with single AP stimulation. But responses were not improved at higher frequencies (80-160 Hz, Fig. 3C). Although it may not be possible to reduce the Hill coefficient to the level of the best synthetic indicators (e.g., OGB1, with Hill coefficient of 1) and produce linear responses across low and high calcium concentrations, we hope to produce a family of indicators with different KD's that span the dynamic range of typical neuronal firing rate regimes and calcium concentrations. To enable improved calcium imaging over a wider range of stimuli in fly neurons, we will undertake engineering of (1) a series of lower affinity indicators that extend across the calcium range of various types of fly neurons, and (2) a single indicator with much wider dynamic range. This will be done by testing 50 more mutants with mutations at one or more of the 4 calciumbinding sites of CaM and the M13/CaM interface. Calcium-binding mutations have been previously demonstrated to lower affinity and cooperativity (48). The reduction of cooperativity of calcium binding could aid in widening the dynamic range of calcium sensing, which we have begun to explore (Fig. 3). The M13/CaM interface can affect calcium-binding and unbinding rates. The goals will be to create (1) a series of 5-10 indicators with calcium KD's from 500 nanomolar to 10 micromolar, while maintaining GCaMP6 SNR levels, and (2) an indicator with a GCaMP6s-like calcium KD but a Hill slope of 1 to 1.5, while maintaining GCaMP6s SNR levels. The kinetics of lower KD mutants could be improved compared to GCaMP6, which may allow instantaneous firing rates to be tracked in high-firing-rate neurons. 19 Figure 8. Light-induced Cre recombinase. (A) Schematic of split Cre halves interacting on DNA using light-sensitive dimerization domains from CRY2 and CIBN proteins. (B) Schematic of split Cre halves. Control (*) constructs with CRY2-Linker-Cre Nterminal domain ending at amino acid position 104 (CRY2 L CreN-104) and CIBN-Linker Cre C-terminal domain (CIBN L CreC-106) beginning with amino acid position 106 as in (Kennedy M J 2010). Constructs with various split sites are also shown. (C) Light-induced recombination of CAG-lox-stop-lox-GFP reporter construct. Various combinations of split Cre constructs were co-transfected with a reporter construct into HEK293T cells. Cells were exposed to pulsed blue light (2 s on, 180 s off) for 0, 1, 24 h. A red fluorescent protein transfection control construct was also used. Percent recombination was the ratio of the number of green to red cells. Milestones For FY2015, we will engineer low-F0 GCaMPs such that 5-fold more neurons can be imaged in vivo with GCaMP6s-like SNR for mouse and limited fly applications. For FY2016, stable GCaMPs will be produced that have 5 times greater stability of GCaMP6 when expressed in vivo and are less toxic. GCaMPs with lower calcium affinity and cooperativity will be made that span the calcium ranges of various fly high frequency and analog neurons for improved activity detection. 4.2.3 R-GECIs We propose to engineer improved RCaMPs and R-GECOs. After making considerable gains in amplitude and speed (Fig. 4 and 5), we will next address the following design parameters: Peak brightness is an easily measurable parameter that is highly relevant to imaging. It is proportional to the QE and related inversely to the bleaching cross-section (37). We will maintain ∆F/F0 and increase peak brightness. So far our protein engineering has focused on sites 20 away from the chromophore of RCaMP and R-GECO (Fig. 4A, D). We will first identify highest SNR and highest SNR/highest speed variants and then mutate near the chromophore. We believe that given the high level of sequence homology between red FPs it should be possible to graft chromophore-forming residues and surrounding side-chains from recently improved FPs (e.g. mCrimson, mCardinal) without having to start from scratch with the new scaffolds. It is likely that subsequent optimization will be required after chromophore grafting, but such designs may improve brightness or photostability in an efficient manner. Photophysical measurements will be performed in collaboration with the JFRC Applied Physics and Instrumentation Group (APIG) in crude protein preparations. The goal will be obtain improvements in photobleaching rates and photoswitching. Spectral separation from green fluorescent calcium indicators will also be tested. We will also explore long Stokes' shift R-GECIs that can be co-excited with GCaMP but yield easily separable emission. Lower F0 for improved imaging of neuronal populations. We seek to lower the background fluorescence from inactive neurons and processes in order to increase SNR. A co-expressed GFP could be used to identify sensor-expressing neurons in vivo. Re-engineering of the M13/FP and FP/CaM linkers will be done to lower F0 while maintaining the peak fluorescence, similar to the strategy for low-F0 GCaMPs. The goal will be to increase SNR so that thousands of neurons in vivo can be imaged simultaneously at high speed in mouse brains. Lower affinity for wider dynamic range in fly neurons. The gains in RCaMP and R-GECO response amplitudes have been obtained through driving the apparent calcium KD to lower levels than for the parent sensors. The response amplitudes are improved for low numbers of action of potentials. But for high numbers of APs, responses are not much improved. By mutating individual or several of the 4 calcium-binding domains in CaM, we will create an affinity series of RCaMP and R-GECO sensors. Our goal is also to widen the dynamic range of calcium detection by reducing the cooperativity of calcium binding. Milestones For FY2015, RCaMP and R-GECO variants will be produced with these properties: high fidelity detection of single AP spikes in vivo, brightness sufficient to image from thousands of neurons, at high speed, spectrally compatible with GCaMP and ChR2. For FY2016, RCaMP and R-GECO 21 variants will be produced with these properties: lower F0 to reduce background signal for improved population imaging, and lower affinity and wider dynamic range for activity monitoring of high frequency fly neurons. 4.2.5 GEVIs Benchmarking existing voltage sensors. Several fluorescent protein voltage sensors have been developed. These sensors differ in topology, voltage sensor moiety, contrast mechanism, etc. We will begin by benchmarking the best versions of the existing scaffolds, including the singlewavelength sensors ArcLight (19), ASAP1 (18), VSFP3.1 (49), and FRET-based sensors VSFP2.3 (49), VSFP-Butterfly (49) and Mermaid (50). We currently do not plan to pursue Archbased sensors because of their low fluorescence quantum yield (38) that is inherent to the retinal and not likely to be significantly improved by protein engineering. Benchmarking will be based on fluorescence microscopy and whole-cell recordings in HEK cells (Fig. 10) and in primary neuronal cultures. Screening assay for voltage sensors. In contrast to GECIs, the mechanism underlying fluorescence changes are not understood in GEVIs. Large-scale mutagenesis and engineering, coupled with a high-throughput assay for GEVI function, will be required to make substantial advances. We will thus develop a non-neuronal assay that will provide higher screening rates at reduced cost. The assay requires robust dividing cells, rather than primary cultured neurons. Also needed are cells with a stereotyped, timed voltage response that can be evoked synchronously across a population in a well. This is critical because for a high-throughput assay we would like to avoid single cell, microscopy-based assays and instead interrogate entire wells. Ideally, cells will be efficiently transfectable. One candidate system is a HEK cell line stably expressing NaV1.3 and Kir2.1 (‘HEK-NK’) (51). These cells exhibit spontaneous activity of AP-like spikes when grown to confluency, and can be induced to fire by field stimulation (Fig. 11). Alternatively we will generate a HEK cell line expressing ChR2 or one of its red-shifted variants (e.g. Chrimson, ReaChR). Such a line would permit simultaneous optical control and readout of 22 Figure 9. GCaMP6 transgenic mice. (A) Epifluorescence image of coronal brain section (bregma -4.3 mm) from a Camk2a-GCaMP6f transgenic mouse (founder line GP7.7). Unstained 50-µm section. (B) Thy1-GCaMP6f-WPRE (founder line GP5.17). (C) Thy1-GCaMP6f-WPRE (founder line GP5.5). Look up table with fluorescence signal scale (arbitrary units). (D) Confocal image of somatosensory cortex and hippocampus from a Camk2aGCaMP6f transgenic mouse (founder line GP7.7). (E) Thy1-GCaMP6f-WPRE (founder line GP5.17). (F) GCaMP fluorescence level per neuron measured from confocal optical sections of transgenic mice and AAV-transduced mice from motor (blue bars), barrel (black bars), or visual cortex (red bars). Quantification of cortical layer 2/3 (left graph). Quantification of cortical layer 5 (right graph). (G) Functional imaging of GCaMP6f in visual cortical neurons in Thy1-GCaMP6f-WPRE (founder line GP5.17) transgenic mice. Example traces from a neuron expressing GCaMP6s. Single sweeps (gray) and averages of sweeps (red) are overlaid. Directions of grating motion (8 directions) are shown above traces (arrows). Layer 2/3 neuron (left graph). Layer 5 neuron (right graph). 23 membrane potential. The choice between these systems will depend on the consistency of voltage control. A factor limiting the throughput of our current neuron-based imaging assay is that each well is interrogated in series by high-resolution imaging. The high-throughput GEVI assay will measure fluorescence in all 96 wells at once (52), increasing throughput by two orders of magnitude, at the expense of single cell resolution. This will allow screening 100,000 variants per year. A secondary component of screening will be measurement of surface expression of voltage sensors using an antibody assay and varying permeabilization conditions. Because of the time and expense involved in such an assay, this method will be applied to a subset of sensors identified in the activity-based screen with improved voltage sensing. After identifying residues in GEVIs that improve voltage signals, systematic changes in the signal sequence involved in membrane targeting will be examined to maximize membrane insertion. Screening libraries of sensor variants generated by large-scale site-directed mutagenesis. To date, testing of GEVI scaffolds has been done only with modest numbers of variants (dozens). Therefore it is not possible to predict which existing scaffold holds the greatest promise for improvements with large-scale engineering and screening. Based on existing data we propose to divide our screening resources roughly as follows: 40% ASAP1, 40% ArcLight, 20% other scaffolds, including new scaffolds. Thus our main effort will be to improve existing scaffolds, and a smaller but significant effort will be dedicated to demonstrating feasibility and function of fundamentally new GEVI scaffolds that hold promise for vast performance improvements. Selection of variants during screening will be based on several criteria to generate significant performance improvements in optical imaging of neuronal membrane potential with improved sensitivity as well as spatial and temporal resolution. Performance metrics that will be measured are sensitivity (amplitude and SNR), linearity, and speed. Secondary screens will also probe capacitative loading by the sensor gating charge. Developing GEVIs based on novel scaffolds. Despite more than a decade of work, GEVIs are far less mature than GECIs. The mechanisms coupling voltage change to fluorescence change are unknown. There is little structural information that guides the paths for optimization. It is likely 24 that the best GEVI scaffolds in the long run may not currently exist. Use of FP chromophores (as opposed to retinal in Arch, etc.) ensures that probes are bright. The voltage paddle of the ArcLight/VSFP family may not be an ideal voltage sensor for several reasons including capacitative loading. We will develop additional scaffolds, including direct placement of the GFP chromophore into the membrane, through targeted modification of the surface to increase lipophilicity and membrane intercalation. Potential designs include "stapling" FPs between transmembrane helices alongside specific intra-cellular and extra-cellular targeting domains; inserting FPs into the middle of pore-forming proteins, such as has been done inside designed cavitated chaperonins (53); or by reengineering FP surface residues to be more hydrophobic and/or be targeted for fatty acid attachment (the reverse of this approach has been used successfully to reengineer G-protein coupled receptors (obligate membrane proteins) to become soluble proteins for high-throughput drug screening and crystallization (14). Our goal is to develop a set of voltage sensors that reliably detect a single AP and have a range of decay kinetics (1, 10, 100 milliseconds) suitable for a variety of experimental situations. The goals of the project will be met when we have indicators that can measure AP trains in populations of neurons in the intact mouse and Drosophila brain, and that can be used to measure subthreshold electrical activity in dendritic arbors of individual neurons. Milestones For FY2015, we will produce a standardized benchmarking and screening assay, and the performance of current voltage indicators (Arch, ArcLight, ASAP, VSFP) will be compared. For FY2016, optimized voltage indicator variants will be made that can detect subthreshold activity of hundreds of neurons in vivo. Improved voltage sensor variants will be engineered for high fidelity tracking of neuronal spikes at 100 Hz in vivo. 4.2.7. Transgenic/targeted mice We will optimize the Rosa26-CAG-lox-STOP-lox-GCaMP6s-WPRE expression cassette to increase expression levels. Linker sequences between the CAG promoter and sensor start codon will be re-engineered to allow for more efficient transcription and/or translation. Thousands of constructs with altered linker sequences will be tested for expression level in transient 25 Figure 10. Performance of GEVIs in HEK cells. (A) Topology of ArcLight-Q239 protein (19) at membrane with intracellular FP. (B) Topology of ASAP1 protein (18) at membrane with extracellular FP. (C) Voltage clamp of HEK293T cell expressing ArcLight-Q239. Voltage steps (-100 to 40 mV from -60 mV holding potential; top) and transmembrane currents (bottom). (D) Voltage clamp of HEK293T cell expressing ASAP1, with same conditions as C. (E) ArcLight-Q239 fluorescence responses to voltage steps from -60 mV holding potential (fluorescence traces, top; ∆F/F0 traces after correction for bleaching, bottom). (F) ASAP1 fluorescence responses to voltage steps, under same conditions as E. (G) ArcLight-Q239 fluorescence changes versus voltage steps from -60 mV holding potential. (H) ASAP1 fluorescence changes versus voltage steps. (I) ∆F/F0 responses to voltage steps from -60 to 40 mV from HEK cells expressing either ArcLight-Q239 (blue, n=4) or ASAP1 (green, n=5). (J) Kinetics for the fast component of the rise (τ-on) and decay (τoff) time for GEVIs. (K) SNR calculated as the ratio of the peak fluorescence at 40 mV divided by the standard deviation of the resting fluorescence at -60 mV after background subtraction. (L) Bleaching measured as the percentage loss of fluorescence after illumination for 2 s as in E and F. 26 transfection assays in HEK293T tissue culture cells by epifluorescence imaging and FACS. A Cre recombinase-expressing construct and co-expressed RFP control (to normalize for transfection level) will be transfected transiently. The Rosa26-CAG variants with the highest fluorescence will be selected for use in gene targeting and mouse production for next generation GCaMPs, RCaMPs, and R-GECOs. In the absence of an improved strategy, we will continue using Thy1 transgenic expression cassettes and selecting for founder lines with suitable expression patterns and levels. We will also continue using improved expression systems (e.g. TIGRE insulator, Tet-regulatable, lox-gated system (54)) in collaboration with other laboratories. Milestones For FY2015, thousands of Rosa26-CAG-lox-STOP-lox-GCaMP6s-WPRE variants will be screened in tissue culture cells. For FY2016, Rosa26-targeted mice with optimized cassettes will be produced with high and stable GCaMP6s, GCaMP6f, RCaMP, and R-GECO expression levels comparable to AAV-mediated expression in cortical neurons. 4.2.9 Calcium-activated and light-dependent switch prototypes Our goal is to create proteins that transduce neural activity (i.e. calcium) and a user-defined stimulus (i.e. light) into activation of a downstream transgene, either through significantly upregulated transcription of the target locus, or through recombination of the locus to become compatible with transcriptional activation. Several promising scaffolds and approaches will be tested, including Cre for recombination, tetracycline repressor (TetR) and TAL effectors for transcriptional activation, and CRISPR/Cas9 for genome editing. For Cre recombinase, we will pursue alternative topologies to the published split, wild-type enzyme (41) that we have characterized to date (Fig. 8). Cre must bind to both DNA strands at both ends of a target locus, so that a split Cre requires the simultaneous assembly of 8 molecular fragments onto double-stranded DNA to potentially achieve a recombination event. All lightgated, split-Cre designs tested to date have suffered from very low levels of activation. For this reason, further experimentation with Cre will involve full Cre domains and potentially tandem dimers, drastically reducing the number of molecules, which should improve efficacy. Calciumdependence will be introduced into Cre recombinase by several strategies: computational 27 Figure 11. Spiking HEK cells. (A) Schematic of HEK-NK cells that stably express voltage-gated sodium channel NaV1.3 and the inwardly-rectifying potassium channel KIR2.1 under drug selection and are electrically-coupled by endogenous gap junctions (51). (B) Voltage steps (-60 to 0 mV) delivered from a -60mV holding potential activate a slowlyinactivating sodium current. (C) Current steps (-40 to 100 pA) were applied to an HEK-NK cell. Hyperpolarizing and subthreshold steps generated a passive response. Steps of 20 pA or greater evoked spikes with threshold of -30 mV. (D) Extracellular field stimulation evoked synchronous spikes in a population of HEK-NK cells transfected with ArcLight-Q239 (Top). One cell was recorded in current clamp mode (Top left, pipette attached). ROIs were selected from the epifluorescence image (Top right). (Middle) Field stimuli of 1 ms duration were delivered at 15 second intervals. Ten consecutive spikes are shown (red traces). Inset shows synchrony of the spike onset (black traces). (Bottom) ∆F/F0 responses for 5 cells selected shown at top, obtained during a single field stimulus trial. 28 calcium-binding-site design into the Cre itself, or by grafting M13/CaM as in the GCaMP, RCaMP, R-GECO, and CaMPARI scaffolds. Low background levels of enzyme activity will be retained by both design and directed evolution. We have preliminary data with improved activity several-fold over the starting molecules (Fig. 8). Recently, light-gated versions of TAL effectors have been introduced, taking advantage of the cryptochrome (CRY)/CIB interaction to bring together a TAL DNA-binding domain and a transcriptional activation motif (55). We will expand on this scaffold by introducing calciumdependence into the gene activation step, similarly through either computational design or by CaM/M13 grafting. A similar strategy to that used to generate the light-gated TALs may work for TetR and its derivatives TetOn and TetOFF, wherein the DNA-binding domains and activator/repressor modules are contained in the same protein. We will also explore the modification of endogenous mouse/fly/etc. transcription factors to introduce these characteristics. Finally, recent advances in understanding of the structure/function of the CRISPR/Cas9 microbial genome-editing enzyme system creates the possibility of engineering in neural activity-dependence to this tool as well. Such a reagent would be useful for targeted manipulation of neurons in culture, for instance (the dependence on guide RNA delivery limits in vivo utility). Milestones For FY2015, we will produce assays to test and screen calcium switch scaffolds based on light/calcium-sensitive Cre recombinase and transcription factors. For FY2016, we will demonstrate prototype light-sensitive Cre recombinase and transcription factors that can switch on ChR2 gene expression based on activity and light gating in vivo in mice and flies. 4.3 Publication and Reagent Distribution Plan Our goal is to produce well-characterized and quality-controlled reagents and make them immediately and widely available to the research community. For example, GCaMP5 was available at Addgene.org for almost one year before publication of the relevant manuscript (47). 29 4.3.1 Timeline of Reagent Distribution 0-9 Months After Discovery of a New Sensor Characterization in vitro and in vivo requires at least 9 months (including creation of viral vectors, transgenic flies, and worms) after the initial discovery. During this time period, reagents will be made available to key collaborators who are testing and calibrating the sensor using previously agreed upon procedures and standards; one obvious example would be collaborators who test in species/systems not represented on the project team. Additionally, laboratories at JFRC will receive access to the reagents for use at JFRC. Collaborators including JFRC laboratories will not distribute the reagents and will not publish results using the sensors prior to publication of the sensors by the project. Our goal is to complete testing within 9 months of initial characterization in cultured neurons. Transgenic mouse production will be initiated during this period, but production and testing will take considerably longer. Reagents will be distributed to other rodent labs if they plan to create transgenic mice that are not the same as those created at JFRC by the project, mainly to avoid duplication of effort, and are willing to widely share these mice on a pre-publication basis (i.e. via deposition at The Jackson Laboratory). 9 Months After Discovery • DNA reagents will be deposited in Addgene, Inc. • Viruses will be deposited in the University of Pennsylvania Viral Vector Core. • Flies will be submitted to the Bloomington Drosophila Stock Center and/or • JFRC's shared Drosophila resource after validation. Transgenic mice will be deposited at The Jackson Laboratory as soon as they have been characterized, no later than 9 months after germline transmission. 30 Reagents that have been tested, but are not being pursued further, will be advertised on the JFRC website and made available through Addgene, Inc., or the University of Pennsylvania Viral Vector Core. 4.7.2 Publication Policies The generation and characterization of new reagents will be published as a collaboration among the relevant GENIE members and other contributors at JFRC and elsewhere. The GENIE Project at JFRC will be listed as the first address, unless an alternative is mutually agreed upon. If the manuscript is accepted less than 9 months after discovery, all materials will be distributed without restrictions as described above. 31 5.0 References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Stosiek C, Garaschuk O, Holthoff K, & Konnerth A (2003) In vivo two-photon calcium imaging of neuronal networks. Proc Natl Acad Sci U S A 100(12):7319-7324. Reiff DF, et al. (2005) In vivo performance of genetically encoded indicators of neural activity in flies. J Neurosci 25(19):4766-4778. Seelig JD & Jayaraman V (2013) Feature detection and orientation tuning in the Drosophila central complex. Nature. Svoboda K, Denk W, Kleinfeld D, & Tank DW (1997) In vivo dendritic calcium dynamics in neocortical pyramidal neurons. Nature 385:161-165. Petreanu L, et al. (2012) Activity in motor-sensory projections reveals distributed coding in somatosensation. Nature 489(7415):299-303. Cox CL, Denk W, Tank DW, & Svoboda K (2000) Action potentials reliably invade axonal arbors of rat neocortical neurons. Proc Natl Acad Sci U S A 97(17):9724-9728. Seelig JD, et al. (2010) Two-photon calcium imaging from head-fixed Drosophila during optomotor walking behavior. Nat Methods 7(7):535-540. Haag J & Borst A (2000) Spatial distribution and characteristics of voltage-gated calcium signals within visual interneurons. Journal of neurophysiology 83(2):1039-1051. Yuste R & Denk W (1995) Dendritic Spines as Basic Functional Units of Neuronal Integration. Nature 375(6533):682-684. Jia H, Rochefort NL, Chen X, & Konnerth A (2010) Dendritic organization of sensory input to cortical neurons in vivo. Nature 464(7293):1307-1312. Chen TW, et al. (2013) Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499(7458):295-300. Akerboom J, et al. (2013) Genetically encoded calcium indicators for multi-color neural activity imaging and combination with optogenetics. Frontiers in molecular neuroscience 6:2. Zhao Y, et al. (2011) An expanded palette of genetically encoded Ca(2) indicators. Science 333(6051):1888-1891. Perez-Aguilar JM, et al. (2013) A computationally designed water-soluble variant of a Gprotein-coupled receptor: the human mu opioid receptor. PloS one 8(6):e66009. London M & Hausser M (2005) Dendritic computation. Annu Rev Neurosci 28:503-532. Grinvald A & Hildesheim R (2004) VSDI: a new era in functional imaging of cortical dynamics. Nat Rev Neurosci 5(11):874-885. Knopfel T (2012) Genetically encoded optical indicators for the analysis of neuronal circuits. Nature reviews. Neuroscience 13(10):687-700. St-Pierre F, et al. (2014) High-fidelity optical reporting of neuronal electrical activity with an ultrafast fluorescent voltage sensor. Nature neuroscience. Jin L, et al. (2012) Single action potentials and subthreshold electrical events imaged in neurons with a fluorescent protein voltage probe. Neuron 75(5):779-785. Gong Y, Li JZ, & Schnitzer MJ (2013) Enhanced Archaerhodopsin Fluorescent Protein Voltage Indicators. PloS one 8(6):e66959. Wallace DJ & Kerr JN (2010) Chasing the cell assembly. Current opinion in neurobiology. Cruz FC, et al. (2013) New technologies for examining the role of neuronal ensembles in drug addiction and fear. Nature reviews. Neuroscience 14(11):743-754. 32 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Huber D, et al. (2012) Multiple dynamic representations in the motor cortex during sensorimotor learning. Nature 484(7395):473-478. Tian L, et al. (2009) Imaging neural activity in worms, flies and mice with improved GCaMP calcium indicators. Nat Methods 6(12):875-881. Zariwala HA, et al. (2012) A Cre-Dependent GCaMP3 Reporter Mouse for Neuronal Imaging In Vivo. The Journal of neuroscience : the official journal of the Society for Neuroscience 32(9):3131-3141. Wardill TJ, et al. (2013) A neuron-based screening platform for optimizing geneticallyencoded calcium indicators. PloS one 8(10):e77728. Gibson DG, et al. (2009) Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat Methods 6(5):343-345. Ding J, Luo AF, Hu L, Wang D, & Shao F (2014) Structural basis of the ultrasensitive calcium indicator GCaMP6. Science China. Life sciences 57(3):269-274. Akerboom J, et al. (2009) Crystal structures of the GCaMP calcium sensor reveal the mechanism of fluorescence signal change and aid rational design. J Biol Chem 284(10):6455-6464. Wang Q, Shui B, Kotlikoff MI, & Sondermann H (2008) Structural basis for calcium sensing by GCaMP2. Structure 16(12):1817-1827. Jayaraman V & Laurent G (2007) Evaluating a genetically encoded optical sensor of neural activity using electrophysiology in intact adult fruit flies. Front Neural Circuits 1:3. Egelhaaf M & Warzecha AK (1999) Encoding of motion in real time by the fly visual system. Current opinion in neurobiology 9(4):454-460. Thestrup T, et al. (2014) Optimized ratiometric calcium sensors for functional in vivo imaging of neurons and T lymphocytes. Nat Methods 11(2):175-182. Direnberger T, et al. (2012) Biocompatibility of a genetically encoded calcium indicator in a transgenic mouse model. Nat Commun 3. Kredel S, et al. (2009) mRuby, a Bright Monomeric Red Fluorescent Protein for Labeling of Subcellular Structures. PloS one 4(2). Shaner NC, et al. (2008) Improving the photostability of bright monomeric orange and red fluorescent proteins. Nat Methods 5(6):545-551. Mutze J, et al. (2012) Excitation spectra and brightness optimization of two-photon excited probes. Biophysical journal 102(4):934-944. Kralj JM, Douglass AD, Hochbaum DR, Maclaurin D, & Cohen AE (2012) Optical recording of action potentials in mammalian neurons using a microbial rhodopsin. Nat Methods 9(1):90-95. Gong Y, Wagner MJ, Zhong Li J, & Schnitzer MJ (2014) Imaging neural spiking in brain tissue using FRET-opsin protein voltage sensors. Nat Commun 5:3674. McKinney SA, Murphy CS, Hazelwood KL, Davidson MW, & Looger LL (2009) A bright and photostable photoconvertible fluorescent protein. Nat Methods 6(2):131-133. Kennedy MJ, et al. (2010) Rapid blue-light-mediated induction of protein interactions in living cells. Nat Methods. Caroni P (1997) Overexpression of growth-associated proteins in the neurons of adult transgenic mice. J Neurosci Methods 71(1):3-9. Tsien JZ, et al. (1996) Subregion- and cell type-restricted gene knockout in mouse brain. Cell 87(7):1317-1326. 33 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. Yasuda R, et al. (2004) Imaging calcium concentration dynamics in small neuronal compartments. Science's STKE : signal transduction knowledge environment 2004(219):pl5. Sankaranarayanan S, De Angelis D, Rothman JE, & Ryan TA (2000) The use of pHluorins for optical measurements of presynaptic activity. Biophysical journal 79(4):2199-2208. Wilt BA, Fitzgerald JE, & Schnitzer MJ (2013) Photon shot noise limits on optical detection of neuronal spikes and estimation of spike timing. Biophysical journal 104(1):51-62. Akerboom J, et al. (2012) Optimization of a GCaMP Calcium Indicator for Neural Activity Imaging. The Journal of neuroscience : the official journal of the Society for Neuroscience 32(40):13819-13840. Sun XNR, et al. (2013) Fast GCaMPs for improved tracking of neuronal activity. Nat Commun 4. Akemann W, Lundby A, Mutoh H, & Knopfel T (2009) Effect of voltage sensitive fluorescent proteins on neuronal excitability. Biophysical journal 96(10):3959-3976. Tsutsui H, Karasawa S, Okamura Y, & Miyawaki A (2008) Improving membrane voltage measurements using FRET with new fluorescent proteins. Nat Methods 5(8):683-685. Park J, et al. (2013) Screening fluorescent voltage indicators with spontaneously spiking HEK cells. PloS one 8(12):e85221. Hempel CM, et al. (2011) A system for performing high throughput assays of synaptic function. PloS one 6(10):e25999. Paavola CD, et al. (2006) A versatile platform for nanotechnology based on circular permutation of a chaperonin protein. Nanotechnology 17(5):1171-1176. Zeng HK, et al. (2008) An inducible and reversible mouse genetic rescue system. Plos Genet 4(5). Liu JY, et al. (2012) Efficient and Specific Modifications of the Drosophila Genome by Means of an Easy TALEN Strategy. J Genet Genomics 39(5):209-215. Nakai J, Ohkura M, & Imoto K (2001) A high signal-to-noise Ca(2+) probe composed of a single green fluorescent protein. Nat Biotechnol 19(2):137-141. Pologruto TA, Yasuda R, & Svoboda K (2004) Monitoring neural activity and [Ca2+] with genetically encoded Ca2+ indicators. J Neurosci 24(43):9572-9579. Ohkura M, Matsuzaki M, Kasai H, Imoto K, & Nakai J (2005) Genetically encoded bright Ca2+ probe applicable for dynamic Ca2+ imaging of dendritic spines. Anal Chem 77(18):5861-5869. Tallini YN, et al. (2006) Imaging cellular signals in the heart in vivo: Cardiac expression of the high-signal Ca2+ indicator GCaMP2. Proc Natl Acad Sci U S A 103(12):47534758. Mao T, O'Connor DH, Scheuss V, Nakai J, & Svoboda K (2008) Characterization and subcellular targeting of GCaMP-type genetically-encoded calcium indicators. PloS one 3(3):e1796. 34