A light- and calcium-gated transcription factor for imaging and manipulating activated neurons

Abstract

Activity remodels neurons, altering their molecular, structural, and electrical characteristics. To enable the selective characterization and manipulation of these neurons, we present FLARE, an engineered transcription factor that drives expression of fluorescent proteins, opsins, and other genetically encoded tools only in the subset of neurons that experienced activity during a user-defined time window. FLARE senses the coincidence of elevated cytosolic calcium and externally applied blue light, which together produce translocation of a membrane-anchored transcription factor to the nucleus to drive expression of any transgene. In cultured rat neurons, FLARE gives a light-to-dark signal ratio of 120 and a high- to low-calcium signal ratio of 10 after 10 min of stimulation. Opsin expression permitted functional manipulation of FLARE-marked neurons. In adult mice, FLARE also gave light- and motor-activity-dependent transcription in the cortex. Due to its modular design, minute-scale temporal resolution, and minimal dark-state leak, FLARE should be useful for the study of activity-dependent processes in neurons and other cells that signal with calcium.

Figure 1

Engineering the calcium and light responses of FLARE. (A) FLARE scheme. FLARE components in the dark, low Ca⁺² state (left) and in the light-exposed, high Ca⁺² state (right). The evolved LOV domain (eLOV) undergoes a reversible conformational change upon blue light exposure that allows steric access to an adjoining peptide^,, in this case, a protease recognition sequence. On the left, the transcription factor (red) is tethered to the plasma membrane, sequestered from the cell nucleus. On the right, the coincidence of neuronal activity (which leads to rises in cytosolic calcium) and blue light causes eLOV to “uncage” the protease cleavage site, and brings the protease into proximity of its cleavage site, via the intermolecular calmodulin-calmodulin binding peptide interaction. Consequently, the transcription factor is irreversibly cleaved from the plasma membrane, translocates to the nucleus, and activates transcription of the reporter gene of interest (FP, fluorescent protein). (B) Summary of constructs tested to optimize the calcium response. None of these contain the light-sensitive LOV/eLOV domain, which is introduced later. For testing in HEK cells, we used Gal4 as the transcription factor and the transmembrane domain of CD4 to target it to the plasma membrane. We tested three different calmodulin (CaM) binding peptides (CaMbp), two different TEV protease (TEVp) cleavage sites (TEVcs), and two different forms of TEV protease (wild-type and C-terminally truncated). For the CaMbps, M1 is the M13 peptide with a A14F “bump” mutation that complements the “hole” mutations in our CaM sequence (F19L/V35G, which reduce CaM affinity for endogenous CaM effectors). M2 and M3 CaMbps are derived from CaMKII and reported to have reduced CaM affinities in the low calcium state^–. For TEVcs, the lower affinity sequence is derived from Tango, with a K_m of 240 μM and k_cat 0.84 min⁻¹, and the higher affinity sequence has a K_m of 50 μM and k_cat of 1.9 min⁻¹ . (C) Results from testing constructs in (B) in 12 combinations, under low and high calcium conditions in HEK cells. Gal4 transcription factor drove the expression of Citrine fluorescent protein, whose intensity was quantified by microscopy in 8–10 fields of view per condition (>250 cells per field of view). To elevate cytosolic calcium, HEK were treated with 5 mM CaCl₂ in the presence of 2 μM ionomycin for 5 minutes; cells were then returned to regular media and Citrine was imaged 12 hours later. Signal-to-noise (S/N) ratios at top are the ratios of mean Citrine intensities (black horizontal lines) under high versus low calcium conditions. This experiment has been replicated once. (D) Crystal structure of 16 kD Avena sativa LOV2 domain (PDB:2V1A), the basis of FLARE’s light gate. Structure in the dark state is shown. Upon blue light irradiation, the bound flavin of LOV2 rapidly (<1 sec) conjugates to Cys48, leading to a conformational change that unwinds the C-terminal Jα helix. This results in increased steric access to peptides fused to Jα. For FLARE engineering, the residues shown as dark blue sticks at the C-terminal end of Jα were targeted for replacement by the TEV cleavage site (“biting back”). This structure also highlights the mutations we discovered via directed evolution of the LOV domain (Figure 2). The 5 mutations in evolved LOV (eLOV) are rendered in cyan space-filling mode. (E) Summary of LOV-TEVcs fusions tested in FLARE (X in TEVcs = Y or L). LOV here is the published AsLOV2 mutant (G528A/N538E), not our eLOV. (F) Results from testing six LOV-TEVcs fusion constructs in HEK cells. Each construct was tested under 4 conditions and Citrine expression was quantified as in (C). Calcium was elevated by 5 minute CaCl₂ and ionomycin treatment as in (C). Light treatment was 5 minutes of 467 nm blue light at 60 mW/cm², 33% duty cycle (2 sec light every 6 sec). A star marks the fusion construct with the best performance in this assay (LOV(-2) fused to higher affinity TEVcs). Black horizontal bars are mean values. This experiment has been replicated once.

Figure 1

Engineering the calcium and light responses of FLARE. (A) FLARE scheme. FLARE components in the dark, low Ca⁺² state (left) and in the light-exposed, high Ca⁺² state (right). The evolved LOV domain (eLOV) undergoes a reversible conformational change upon blue light exposure that allows steric access to an adjoining peptide^,, in this case, a protease recognition sequence. On the left, the transcription factor (red) is tethered to the plasma membrane, sequestered from the cell nucleus. On the right, the coincidence of neuronal activity (which leads to rises in cytosolic calcium) and blue light causes eLOV to “uncage” the protease cleavage site, and brings the protease into proximity of its cleavage site, via the intermolecular calmodulin-calmodulin binding peptide interaction. Consequently, the transcription factor is irreversibly cleaved from the plasma membrane, translocates to the nucleus, and activates transcription of the reporter gene of interest (FP, fluorescent protein). (B) Summary of constructs tested to optimize the calcium response. None of these contain the light-sensitive LOV/eLOV domain, which is introduced later. For testing in HEK cells, we used Gal4 as the transcription factor and the transmembrane domain of CD4 to target it to the plasma membrane. We tested three different calmodulin (CaM) binding peptides (CaMbp), two different TEV protease (TEVp) cleavage sites (TEVcs), and two different forms of TEV protease (wild-type and C-terminally truncated). For the CaMbps, M1 is the M13 peptide with a A14F “bump” mutation that complements the “hole” mutations in our CaM sequence (F19L/V35G, which reduce CaM affinity for endogenous CaM effectors). M2 and M3 CaMbps are derived from CaMKII and reported to have reduced CaM affinities in the low calcium state^–. For TEVcs, the lower affinity sequence is derived from Tango, with a K_m of 240 μM and k_cat 0.84 min⁻¹, and the higher affinity sequence has a K_m of 50 μM and k_cat of 1.9 min⁻¹ . (C) Results from testing constructs in (B) in 12 combinations, under low and high calcium conditions in HEK cells. Gal4 transcription factor drove the expression of Citrine fluorescent protein, whose intensity was quantified by microscopy in 8–10 fields of view per condition (>250 cells per field of view). To elevate cytosolic calcium, HEK were treated with 5 mM CaCl₂ in the presence of 2 μM ionomycin for 5 minutes; cells were then returned to regular media and Citrine was imaged 12 hours later. Signal-to-noise (S/N) ratios at top are the ratios of mean Citrine intensities (black horizontal lines) under high versus low calcium conditions. This experiment has been replicated once. (D) Crystal structure of 16 kD Avena sativa LOV2 domain (PDB:2V1A), the basis of FLARE’s light gate. Structure in the dark state is shown. Upon blue light irradiation, the bound flavin of LOV2 rapidly (<1 sec) conjugates to Cys48, leading to a conformational change that unwinds the C-terminal Jα helix. This results in increased steric access to peptides fused to Jα. For FLARE engineering, the residues shown as dark blue sticks at the C-terminal end of Jα were targeted for replacement by the TEV cleavage site (“biting back”). This structure also highlights the mutations we discovered via directed evolution of the LOV domain (Figure 2). The 5 mutations in evolved LOV (eLOV) are rendered in cyan space-filling mode. (E) Summary of LOV-TEVcs fusions tested in FLARE (X in TEVcs = Y or L). LOV here is the published AsLOV2 mutant (G528A/N538E), not our eLOV. (F) Results from testing six LOV-TEVcs fusion constructs in HEK cells. Each construct was tested under 4 conditions and Citrine expression was quantified as in (C). Calcium was elevated by 5 minute CaCl₂ and ionomycin treatment as in (C). Light treatment was 5 minutes of 467 nm blue light at 60 mW/cm², 33% duty cycle (2 sec light every 6 sec). A star marks the fusion construct with the best performance in this assay (LOV(-2) fused to higher affinity TEVcs). Black horizontal bars are mean values. This experiment has been replicated once.

Figure 2

Directed evolution of LOV domain to improve light gating in FLARE. (A) Selection scheme. A >10⁷ library of AsLOV2 variants was displayed on the yeast surface as a fusion to Aga2p protein. The TEVp cleavage site ENLYFQ_▲Y (higher affinity) was fused to LOV’s C-terminal end, and HA and Flag are flanking epitope tags. The positive selection (green) enriches mutants with low Flag staining (i.e., high TEVcs cleavage) after protease treatment in the light. The negative selection (red) enriches mutants with high Flag staining (i.e., low TEVcs cleavage) after prolonged protease treatment in the dark. (B) Graph summarizing yeast library characteristics after each round of selection. Accompanying FACS plots in Figure SI-2. Grey bars indicate the fraction of yeast cells in quadrant Q2 (out of all cells in Q2 + Q4) after 3 hours of TEV protease incubation in the dark. Quadrants are defined in panel A. Yellow bars indicate the fraction of yeast cells in Q4 (out of all cells in Q2 + Q4) after 1 hour of TEV protease incubation in blue light. This experiment has been performed once. (C) FACS analysis of original AsLOV2 (top) and our evolved eLOV (bottom) on yeast. eLOV provides superior protection of TEVcs against TEVp cleavage in the dark state (left). eLOV has 5 mutations relative to AsLOV2 (G528A/N538E mutant), shown in Figure 1D. This experiment has been replicated 2 times. (D) Comparison of original AsLOV2 (G528A/N538E mutant, top) and our evolved eLOV (bottom) in HEK cells, in the context of FLARE. Constructs were TM(CD4)-CaMbp(M2)-(e)LOV-TEVcs(high affinity)-Gal4 and CaM-TEVp(truncated). Gal4 drives expression of the fluorescent protein Citrine. High calcium (5 minutes) and light conditions were the same as in Figure 1F. Anti-V5 staining detects expression of CaM-TEVp. S/N ratios on right are based on mean Citrine intensities across >500 cells from 10 fields of view per condition. Scale bars, 20 μm. This experiment has been replicated 2 times.

Figure 3

FLARE optimization and testing in neurons. (A) Summary of sequential improvements and changes to FLARE. F1 and F2 are earlier versions of the tool. Complete sequence of final FLARE tool in Figure SI-7. (B) Comparison of tool versions in neurons. tTA transcription factor drives expression of mCherry. To elevate cytosolic calcium, half of the culture medium was replaced with fresh neurobasal media (of identical composition), and mixed by gentle pipetting. GCaMP5f imaging showed that this treatment produced calcium transients for 10 minutes or more (Figure SI-8). Low calcium samples were not treated. Light stimulation was for 10 minutes using 467 nm light at 60 mW/cm², 33% duty cycle (2 sec light every 6 sec). Each datapoint corresponds to a single field-of-view with >40 neurons. Red bars denote mean values. This experiment has been performed once. (C) Confocal imaging of FLARE tool in rat cortical neuron cultures at DIV20. Constructs were introduced by AAV viral transduction at DIV13. Calcium and light conditions were identical to those in (B). 18 hours after treatment, neurons were fixed, stained with anti-V5 antibody (to visualize CaM-TEVp expression), and imaged. Quantification shows that the light-to-dark signal ratio is 121, and the high-to-low calcium signal ratio is 10. This experiment has been replicated 5 times. (D) Confocal imaging of FLARE tool after electrical stimulation. Neurons were transduced with AAVs at DIV10 and imaged at DIV17. Electrode stimulation parameters were 3-second trains consisting of 32 1-millisecond 50 mA pulses at 20 Hz for a total of 15 minutes. Light was co-applied for 15 minutes at 467 nm, 60 mW/cm², 10% duty cycle (0.5 sec every 5 sec). Neurons were fixed, stained, and imaged 18 hours later. Quantification shows that the light-to-dark signal ratio is 17, and the high-to-low calcium signal ratio is 23. This experiment has been replicated 3 times. (E) FLARE sensitivity/time course. DIV18 neurons expressing FLARE were untreated, or activated by electrical stimulation (same parameters as in (D)) or media change (90% of culture medium exchanged) for 4, 8, or 15 minutes with simultaneous application of blue light (467 nm, 60 mW/cm², 10% duty cycle (0.5 sec light every 5 sec)). S/N values reflect mean mCherry intensity ratios with versus without neuronal activity, across 10 fields of view per condition. Figure SI-9 shows a repeat of the media change time course with additional timepoints and accompanying fluorescence images. Errors, s.e.m. This experiment has been replicated 2 times with the media change protocol and once with the electrical stimulation protocol. (F) FLARE is highly specific for simultaneous (top row) rather than sequential (middle and bottom rows) light and calcium inputs. DIV10 cortical neurons expressing FLARE components were activated by electrical stimulation and blue light (same conditions as in (D)). In the case of sequential inputs, a 1 minute pause (to allow reversal of eLOV or calcium-sensing domains) separated the two inputs. Three separate fields of view are shown per condition. This experiment has been replicated once. (G) Mutagenesis experiments to probe FLARE mechanism. Conditions were the same as in (B). Control constructs contain mutations in calcium-binding, CaM-binding, and light sensitive regions, as described. Red bars denote mean values. This experiment has been replicated once. All scale bars, 100 μm.

Figure 3

FLARE optimization and testing in neurons. (A) Summary of sequential improvements and changes to FLARE. F1 and F2 are earlier versions of the tool. Complete sequence of final FLARE tool in Figure SI-7. (B) Comparison of tool versions in neurons. tTA transcription factor drives expression of mCherry. To elevate cytosolic calcium, half of the culture medium was replaced with fresh neurobasal media (of identical composition), and mixed by gentle pipetting. GCaMP5f imaging showed that this treatment produced calcium transients for 10 minutes or more (Figure SI-8). Low calcium samples were not treated. Light stimulation was for 10 minutes using 467 nm light at 60 mW/cm², 33% duty cycle (2 sec light every 6 sec). Each datapoint corresponds to a single field-of-view with >40 neurons. Red bars denote mean values. This experiment has been performed once. (C) Confocal imaging of FLARE tool in rat cortical neuron cultures at DIV20. Constructs were introduced by AAV viral transduction at DIV13. Calcium and light conditions were identical to those in (B). 18 hours after treatment, neurons were fixed, stained with anti-V5 antibody (to visualize CaM-TEVp expression), and imaged. Quantification shows that the light-to-dark signal ratio is 121, and the high-to-low calcium signal ratio is 10. This experiment has been replicated 5 times. (D) Confocal imaging of FLARE tool after electrical stimulation. Neurons were transduced with AAVs at DIV10 and imaged at DIV17. Electrode stimulation parameters were 3-second trains consisting of 32 1-millisecond 50 mA pulses at 20 Hz for a total of 15 minutes. Light was co-applied for 15 minutes at 467 nm, 60 mW/cm², 10% duty cycle (0.5 sec every 5 sec). Neurons were fixed, stained, and imaged 18 hours later. Quantification shows that the light-to-dark signal ratio is 17, and the high-to-low calcium signal ratio is 23. This experiment has been replicated 3 times. (E) FLARE sensitivity/time course. DIV18 neurons expressing FLARE were untreated, or activated by electrical stimulation (same parameters as in (D)) or media change (90% of culture medium exchanged) for 4, 8, or 15 minutes with simultaneous application of blue light (467 nm, 60 mW/cm², 10% duty cycle (0.5 sec light every 5 sec)). S/N values reflect mean mCherry intensity ratios with versus without neuronal activity, across 10 fields of view per condition. Figure SI-9 shows a repeat of the media change time course with additional timepoints and accompanying fluorescence images. Errors, s.e.m. This experiment has been replicated 2 times with the media change protocol and once with the electrical stimulation protocol. (F) FLARE is highly specific for simultaneous (top row) rather than sequential (middle and bottom rows) light and calcium inputs. DIV10 cortical neurons expressing FLARE components were activated by electrical stimulation and blue light (same conditions as in (D)). In the case of sequential inputs, a 1 minute pause (to allow reversal of eLOV or calcium-sensing domains) separated the two inputs. Three separate fields of view are shown per condition. This experiment has been replicated once. (G) Mutagenesis experiments to probe FLARE mechanism. Conditions were the same as in (B). Control constructs contain mutations in calcium-binding, CaM-binding, and light sensitive regions, as described. Red bars denote mean values. This experiment has been replicated once. All scale bars, 100 μm.

Figure 4

Functional reactivation of neurons marked by FLARE and in vivo testing. (A) Scheme for activity “replay” in neuron culture. The coincidence of blue light and high calcium activate FLARE, resulting in expression of opsin(ChrimsonR)-mCherry in subsets of neurons. To re-activate FLARE-marked neurons, red light is applied to stimulate opsin, resulting in cytosolic calcium rises, which can be detected with the GCaMP5f real-time fluorescence calcium indicator. (B) Imaging results from experiment performed as in (A). Cultured rat neurons were transduced with FLARE AAV viruses and GCaMP5f lentivirus at DIV13. At DIV19, neurons were treated with blue light (467 nm at 60 mW/cm², 10% duty cycle (0.5 sec light every 5 sec)) and electrode stimulation (15 minutes of 3-second long trains, each consisting of 32 1-millisecond 48 mA pulses at 20 Hz) for 15 minutes total. 18 hours later, at DIV20, GCaMP5f fluorescence timecourses were recorded (for the 6 indicated cells) while stimulating the ChrimsonR channelrhodopsin with pulses of red 568 nm light as indicated. The bottom image set shows a negative control in which electrical stimulation was withheld at DIV13, but blue light was applied. Scale bars, 50 μm. This experiment has been replicated once. (C) Scheme for testing FLARE in the mouse brain. Concentrated AAV viruses encoding FLARE components (in addition to BFP, an infection marker), were injected into the motor cortex of adult mice (both left and right hemispheres, as shown). After five days of expression, blue light was delivered to the right hemisphere via implanted optical fiber (single 30 minute session of 467 nm light at 0.5 mW, 50% duty cycle (2 sec light every 4 sec)), while mice were running on an exercise wheel or anesthetized. 24 hours later, mice were perfused for imaging analysis. (D) Two representative brain sections from experiment in (C), for anesthesized mouse (top) and wheel running mouse (bottom). Right hemisphere was illuminated for 30 minutes, whereas left hemisphere was in the dark. Activated FLARE drives expression of mCherry. BFP is an AAV infection marker. (E) Quantitation of brain imaging data. For each brain hemisphere with BFP signal above background, we quantified the total ChrimsonR-mCherry fluorescence intensity across 7 consecutive brain sections around the virus injection site. 21–63 brain sections were analyzed from 3–9 mice per condition. Light + running animals have significantly higher mCherry expression than light + anesthesia animals (Kolmogorov-Smirnov Test, p = 0.013). Alternative presentation of data shown in Figure SI-11. (F) Whole-cell patch-clamp electrophysiology was used to record from ChrimsonR-mCherry-expressing neurons in the mouse brain 24 hours after light + running stimulation. Neurobiotin was injected into the patched neuron. (G) Sample traces showing action potentials elicited in response to 5 ms pulses of 589 nm light (red ticks) delivered at 1 Hz (upper panel) or 10 Hz (lower panel). Scale bars = 20 mV, 500 ms. Experiments in (D)-(G) have each been performed once.

Figure 4

Functional reactivation of neurons marked by FLARE and in vivo testing. (A) Scheme for activity “replay” in neuron culture. The coincidence of blue light and high calcium activate FLARE, resulting in expression of opsin(ChrimsonR)-mCherry in subsets of neurons. To re-activate FLARE-marked neurons, red light is applied to stimulate opsin, resulting in cytosolic calcium rises, which can be detected with the GCaMP5f real-time fluorescence calcium indicator. (B) Imaging results from experiment performed as in (A). Cultured rat neurons were transduced with FLARE AAV viruses and GCaMP5f lentivirus at DIV13. At DIV19, neurons were treated with blue light (467 nm at 60 mW/cm², 10% duty cycle (0.5 sec light every 5 sec)) and electrode stimulation (15 minutes of 3-second long trains, each consisting of 32 1-millisecond 48 mA pulses at 20 Hz) for 15 minutes total. 18 hours later, at DIV20, GCaMP5f fluorescence timecourses were recorded (for the 6 indicated cells) while stimulating the ChrimsonR channelrhodopsin with pulses of red 568 nm light as indicated. The bottom image set shows a negative control in which electrical stimulation was withheld at DIV13, but blue light was applied. Scale bars, 50 μm. This experiment has been replicated once. (C) Scheme for testing FLARE in the mouse brain. Concentrated AAV viruses encoding FLARE components (in addition to BFP, an infection marker), were injected into the motor cortex of adult mice (both left and right hemispheres, as shown). After five days of expression, blue light was delivered to the right hemisphere via implanted optical fiber (single 30 minute session of 467 nm light at 0.5 mW, 50% duty cycle (2 sec light every 4 sec)), while mice were running on an exercise wheel or anesthetized. 24 hours later, mice were perfused for imaging analysis. (D) Two representative brain sections from experiment in (C), for anesthesized mouse (top) and wheel running mouse (bottom). Right hemisphere was illuminated for 30 minutes, whereas left hemisphere was in the dark. Activated FLARE drives expression of mCherry. BFP is an AAV infection marker. (E) Quantitation of brain imaging data. For each brain hemisphere with BFP signal above background, we quantified the total ChrimsonR-mCherry fluorescence intensity across 7 consecutive brain sections around the virus injection site. 21–63 brain sections were analyzed from 3–9 mice per condition. Light + running animals have significantly higher mCherry expression than light + anesthesia animals (Kolmogorov-Smirnov Test, p = 0.013). Alternative presentation of data shown in Figure SI-11. (F) Whole-cell patch-clamp electrophysiology was used to record from ChrimsonR-mCherry-expressing neurons in the mouse brain 24 hours after light + running stimulation. Neurobiotin was injected into the patched neuron. (G) Sample traces showing action potentials elicited in response to 5 ms pulses of 589 nm light (red ticks) delivered at 1 Hz (upper panel) or 10 Hz (lower panel). Scale bars = 20 mV, 500 ms. Experiments in (D)-(G) have each been performed once.