2.1.

Plasmids

Plasmids were cloned with conventional molecular biology techniques. Some constructs were cloned into both pcDNA and pAAV vectors for testing in HEK293 cells and primary neurons, respectively. Constructs are listed in Table 2.

Table 2

Plasmids used in this paper.

2.2.

Virus

AAVs carrying LMOs were produced by transfecting subconfluent HEK293FT cells per 10-cm culture dish with $24\mu \mathrm{g}$ of the helper plasmid pAd delta F6, $20\mu \mathrm{g}$ of the serotype plasmid AAV2/9, and $12\mu \mathrm{g}$ of the LMO plasmid using lipofectamine 2000. After 72 h, the supernatant was harvested from culture plates and filtered at $0.45\mu \mathrm{m}$. Virus was purified from cells and supernatant according to previously described methodology,²⁸ but without the partitioning step in the aqueous two-phase system. Virus was dialyzed against PBS (w/o Ca, Mg) overnight at 4°C, using FLOAT-A-LYZER G2, MWCO 50 kDa, followed by concentration in Amicon Ultra-0.5 379 mL centrifugal filters. Viral titers were determined by quantitative PCR for the woodchuck hepatitis post-transcriptional regulatory element. Preparations with titers around $1\times {10}^{13}\mathrm{vg}/\mathrm{mL}$ were used in the study.

2.3.

Cell Culture

2.4.

HEK Cell Electrophysiology

Whole cell voltage patch clamp was performed on HEK293 cells 24 to 72 h after transfection. A coverslip was transferred to a recording chamber (RC26-GLP, Warner Instruments) on the stage of an upright microscope (BX51WI, Olympus) and perfused with external buffer ($1.5\mathrm{mL}/\min$) containing (in mM): 150 NaCl, 3 KCl, 10 HEPES, 2 ${\mathrm{CaCl}}_2$, 2 ${\mathrm{MgCl}}_2$, and 20 D-glucose (pH 7.4, $\sim 300\text{-}315\mathrm{mOsm}/\mathrm{kg}$) at a temperature of $37\pm 1°\mathrm{C}$. Intracellular patch solution contained (in mM): 130 K-gluconate, 8 KCl, 15 HEPES, 5 ${\mathrm{Na}}_2$-phosphocreatine, 4 ${\mathrm{Na}}_2$-ATP, 0.3 ${\mathrm{Na}}_2$-GTP (pH 7.25, $\sim 295$ to $300\mathrm{mOsm}/\mathrm{kg}$). Borosilicate glass micropipettes were prepared using a vertical pipette puller (PC-100, Narishige) and had resistances from 2 to 6 MΩ. Epifluorescence microscopy was used to identify cells expressing the construct (constructs contained red or green fluorescent proteins). Opsins were excited through the objective (LUMPLFLN40XW, 433 NA 0.8, Olympus) using a metal halide light source (130 W, U-HGLGPS, Olympus) with filter cubes for blue (Ex/Em: 480/530 nm, U-MNIBA3, Olympus), green (Ex/Em: 540/600, U-MWIGA3, Olympus), and red (Ex/Em: 580 to 630/645 to 695, ET-Alexa633 filter set 605/50x, 670/50M w/BX2 cube) excitation. An electronic shutter (Lambda SC, Sutter Instruments) was used to control light delivery to cells and limit exposure to 1 s. Photocurrents were evoked with blue, green, or red light to match the peak bioluminescence emission of the luciferase tethered to the opsin. At maximum, light intensity from the light source irradiance was (in $\mathrm{mW}/{\mathrm{cm}}^2$) 16.8, 36.5, and 39.9, respectively, as measured with a light meter (ThorLabs).

Recordings were conducted at $-60\mathrm{mV}$ holding voltage using a Multiclamp 700b amplifier and Digidata 1440 digitizer with pClamp 10 recording software (Molecular Devices). After sealing the cells, photocurrent response was measured using a gap free protocol followed by addition of luciferin via perfusion. Luciferins included h-coelenterazine (hCTZ, Nanolight Technology, #301), native-coelenterazine (CTZ, Nanolight Technology, #303), and e-coelenterazine (eCTZ, Nanolight Technology, #355). Luciferin was diluted to a $500\mu \mathrm{M}$ working stock in ACSF ($500\mu \mathrm{L}$ total) for CTZ, hCTZ, and eCTZ. Final concentration in perfusion chamber was $\sim 100\mu \mathrm{M}$. Luciferin was chosen based on the most efficient substrate for the respective luciferase in each construct.

2.5.

Multi Electrode Array Recordings

MEA2100-Lite-System (Multichannel Systems, Germany) was used for all MEA recordings with Multichannel Experimenter software at a sample rate of 10,000 Hz. Consistently spiking neurons were used for recordings between DIVs 14 and 25; only cultures showing spontaneous electrophysiological activity were used. All‐trans retinal (R2500; Sigma‐ Aldrich, St. Louis, Missouri) was added to the culture medium to $1\mu \mathrm{M}$ final concentration before electrophysiological recordings. Prior to recording, all reagents were pre‐warmed to 37°C. MEAs were transferred from the ${\mathrm{CO}}_2$ incubator to the heated MEA2100 head stage maintained at 37°C, and the cultures were allowed to equilibrate for 5 to 10 min. The headstage was situated on a support stage with a PlexBright optical patch cable positioned from below to illuminate the center of the MEA dish. The patch cable was connected to a Plexon optogenetic LED fiber-coupled module (blue, 465 nm; green, 525 nm; lime, 550 nm) controlled by a PlexBright LD-1 single channel LED driver for light stimulation of cultures at different wavelengths. A micropipette was used to add luciferin or vehicle with the reagent drop gently touching the liquid surface, creating a time-locked artifact in the recordings. After recording, the media in the wells were replaced with fresh pre-equilibrated and pre-warmed NB-Plain media, and the cultures were used for another round of recording the next day.

2.6.

Data Analysis

3.1.

Coupling Efficiency of LMOs

To evaluate the performance of LMOs, whole-cell patch clamp recordings were obtained from cultured HEK293 cells expressing these constructs. For each recording, opsins were activated by an arc lamp to obtain the maximal photocurrent and by luciferin (CTZ, hCTZ, eCTZ) to obtain the bioluminescence-induced current [Fig. 2(a)]. We began by measuring, under voltage clamp conditions ($-60\mathrm{mV}$), the photocurrents evoked by exciting the opsin actuators directly via illumination by an arc lamp, choosing a filter cube matching the absorption peak of the opsin. As shown in the example in Fig. 2(b), for a cell expressing LMO7, illumination evoked an inward photocurrent due to activation of VChR1. This current reached a peak within a few ms and exhibited a sustained plateau that persisted for as long as the excitation light was maintained [1 s in the example shown in Fig. 2(b)]. To determine the effectiveness of the LMO for chemogenetic control of membrane potential, we measured the ability of the luciferin to induce currents in these cells. Infusion of luciferin into the recording chamber yielded bioluminescence emission and simultaneous activation of an inward current [Fig. 2(b)]. This inward current generally tracked the time course of the luciferin presence in the chamber. We then used these measurements to determine the efficiency of coupling between the luciferase light source and the opsin actuator. We employed a simplified version of the procedure introduced by Berglund et al.³ In brief, we compared the magnitude of currents induced by luciferin to those evoked by maximal exposure to light from an arc lamp by dividing the measured luciferin-induced current by the maximal photocurrent, yielding the coupling efficiency [Fig. 2(c)]. This value defines the intrinsic efficiency of coupling between luciferase and opsin and takes into account cell-to-cell variations in LMO expression. Figure 2(d) shows examples of non-expressing control HEK293 cells patched to determine the current response to light of different wavelengths and to different luciferins (CTZ, hCTZ, eCTZ). Examples are representative of five cells recorded for each experiment.

Fig. 2

Quantifying LMO efficacy. (a) Membrane potential changes elicited chemo (CTZ)- or opto (LED, arc lamp)-genetically can be measured in HEK293 cells expressing an LMO. (b) In whole cell voltage patch clamp, inward current is determined in cells, here expressing LMO7, exposed to luciferin (hCTZ) and to light from an arc lamp (photocurrent). (c) Equation used to quantify LMO efficacy using currents measured in patch clamp. (d) Absence of effects of various luciferins and wavelengths of physical light source on the current of non-expressing HEK293 cells (controls, $n=5$).

3.2.

Brightness of Light Emitter

In previous studies, we saw a continuous improvement of LMO function when replacing a luciferase with a brighter one. A 10-fold improvement in coupling efficiency was achieved when switching out wildtye GLuc (LMO2) with the brighter variant sbGLuc (LMO3), leading to the first excitatory LMO that allowed non-invasive manipulation of animal behavior.⁵ This trend continued, though not as dramatically, when we replaced sbGLuc in LMO3 with a luciferase-fluorescent protein Förster resonance energy transfer (FRET) probe (NCS2, a fusion of eKL9h and mNeonGreen), resulting in LMO7.²¹ Here, we took advantage of a new FRET probe, GeNL_SS, a bright light emitter obtained by extensive molecular evolution of a GeNL–SSLuc fusion protein. Tethering GeNL_SS to VChR1 resulted in LMO10. Direct comparison of VChR1-based LMOs 3 (sbGLuc) and 7 (NCS2) with 10 (GeNL_SS) showed an improvement in coupling efficiency (CE) at 0.57 ($n=5$) when compared to LMO7 (CE = 0.52; $n=5$) and LMO3 (CE = 0.31; $n=5$) [Figs. 3(a)–3(d)].

Fig. 3

Increasing brightness of the light emitter. (a) Schematics of LMO3, LMO7, and LMO10 with the same opsin (VChR1) and different light emitters (sbGLuc, NCS2, GeNL_SS). (b) Fluorescent images (left), photocurrent patch clamp traces (middle), and luciferin-induced patch clamp traces (right) for LMO3, LMO7, and LMO10. (c) Coupling efficiencies of LMO3, LMO7, and LMO10 ($n=5$).

3.3.

Sensitivity of Light Sensor

We previously also saw an improvement of LMO function when replacing an opsin with a more light-sensitive one. Again, a 10-fold improvement in coupling efficiency was observed from LMO1 (Gluc-ChR2) to LMO2 (Gluc-VChR1).³ Thus, we expected to see an increase in coupling efficiency when we fused NCS2, the luciferase used successfully in LMO7 (NCS2-VChR1), channelrhodopsin ChRger3²³ in LMO8. The ChRger series of excitatory opsins was optimized through a machine learning platform yielding high-photocurrent ChRs with high light sensitivity. While ChRger3 displayed a large response to lamp stimulation, it completely lacked a response to bioluminescence [$n=5$, Figs. 4(c) and 4(d)].

Fig. 4

Increasing sensitivity of the light sensor. (a) Schematics of LMO7 and LMO8 with the same luciferase-opsin fusion protein (NCS2) and different opsins (VChR1, ChRger3). (b) Fluorescent images (left), photocurrent patch clamp traces (middle), and luciferin-induced patch clamp traces (right) for LMO7 and LMO8. (c) Coupling efficiencies of LMO7 and LMO8 (n=5).

We next paired our currently brightest light emitter, GeNL_SS, with ChRmine, an opsin identified in a marine organism-based genomic screen for new classes of microbial opsins.²⁴ ChRmine was reported to have a hundred-fold improved operational light sensitivity compared with other fast, red-shifted variants. While combining GeNL_SS with VChR1 in LMO10 outperformed LMO7 (NCS2-VChR1), the combination with ChRmine in LMO11 (GeNL_SS-ChRmine) led to further improved coupling efficiency from 0.57 (LMO10) to 0.65 (LMO11) [$n=5$, Figs. 5(a)–5(d)].

Fig. 5

Increasing both brightness of the light emitter and sensitivity of the light sensor. (a) Schematics of LMO10 and LMO11 with brightest luciferase-opsin fusion protein (GeNL_SS) and opsins VChR1 and the super light sensitive ChRmine. (b) Fluorescent images (left), photocurrent patch clamp traces (middle), and luciferin-induced patch clamp traces (right) for LMO10 and LMO11. (c) Coupling efficiencies of LMO10 and LMO11 ($n=5$).

3.4.

Arrangement of Moieties

To compare different configurations of how the light emitter is attached to the opsin, we generated variants of LMO7 (NCS2-VChR1) with either NCS2 tethered to the C-terminus instead, LMO7.2 (VChR1-NCS2), or with an additional light source tethered to the C-terminus, LMO7.3 (NCS2-VChR1-NCS2) [Fig. 6(a)]. All LMO7 expressing cells showed both photocurrent and luciferin-induced current responses of variable magnitudes with a coupling efficiency of 0.56 ($n=5$) [Figs. 6(c) and 6(d)]. However, the C-terminal fusion of the light emitter in LMO7.2 permits chemogenetic activation with a diminished bioluminescent response reflected in the lower coupling efficiency of 0.30 ($n=5$) [Figs. 6(c) and 6(d)]. In contrast, of five LMO7.3 expressing cells, three completely lacked a response to lamp stimulation, and two had a minimal response [Figs. 6(c) and 6(d)]. None of the cells showed any luciferin-induced current. This diminished response is reflected in the nonfunctional (n.f.) coupling efficiency of LMO7.3 ($n=5$).

Fig. 6

Different arrangements of moieties in LMOs. (a) Schematics of NCS2 tethered to VChR1 at the N-terminal (LMO7), the C-terminal (LMO7.2), and the N- and C-terminal (LMO7.3). (b) Fluorescent images (left), photocurrent patch clamp traces (middle), and luciferin-induced patch clamp traces (right) for LMO7, LMO7.2, and LMO7.3. (c) Coupling efficiencies of LMOs 7, 7.2, and 7.3 ($n=5$).

3.5.

Wavelength Compatibility

We previously generated a series of LMOs that utilize the red-shifted Renilla luciferase variant RLuc8.6-535(W156F) fused to an evolved red fluorescent protein (mCherry22.0, derived from mCherry with the mutations F11Y, Y38F, D59H, S62P, M66Q, L83F, K92T, V96E, W143T, E144T, T147S, I161T, Q163W, D174E, N196D, I197R, S203Y, and G219S; to be described fully in a manuscript currently in preparation), tethered to the C-terminus of various opsins, analogous to iLMO2.⁴ mCherry22.0 has peak excitation at 561 nm and peak emission at 593 nm with a quantum yield of $\sim 0.70$, making it an excellent FRET acceptor for RLuc8.6-535. In LMO12, the peak emission of this red light emitter, 593 nm, closely matches the peak absorption of the opsin, ChRmine (585 nm²⁴). Even though LMO12 is constructed with a C-terminal fusion, we compared its performance to LMO11, which uses the light emitter GeNL_SS (peak emission around 520 nm) as an N-terminal fusion to ChRmine [Fig. 7(a)] since these two LMO constructs contain the same opsin. While the photocurrent generated by light exposure from the arc lamp remained the same for both LMOs, the coupling efficiency of LMO12 (CE = 0.16, $n=5$) was vastly diminished compared to LMO11 (CE = 0.65, $n=5$) [Figs. 7(b) and 7(c)], indicating that wavelength compatibility alone cannot outcompete other structural requirements of LMO performance.

Fig. 7

Matching emission and excitation spectra of moieties in LMOs. (a) Schematics of LMO11 (GeNL_SS-ChRmine) and LMO12 (ChRmine-mCh22.0-RLuc8.6). (b) Fluorescent images (left), photocurrent patch clamp traces (middle), and luciferin-induced patch clamp traces (right) for LMO11 and LMO12. (d) Coupling efficiencies of LMOs 11 and 12 ($n=5$).

3.6.

LMO Performance in Activating Neuronal Populations

We used multi-electrode extracellular recordings to examine the ability of LMOs to activate neuronal populations. For these experiments, we transduced rat embryonic cortical neurons in MEA wells with AAV-hSyn-LMO, using a virus titer that yielded transduction of most plated neurons by LMOs. Figure 8 shows the effects of LED, vehicle, and CTZ treatments in spontaneously firing cultures on neuronal firing rates in LMO7 and LMO11 expressing cortical cultures. Different sets of neurons were assessed for each of the pre- and post-comparisons. Implementation of a permutation test, followed by a Bonferroni correction for multiple comparisons, provided a robust assessment of the significance of the observed changes between pre- and post-treatment conditions in the various groups.

Fig. 8

Multi-electrode recordings from cortical neurons transduced with AAV-hSyn-LMO7 and -LMO11. Average spiking frequencies of MEA channels before any treatment (pre) and after application (post) of physical or biological light, or vehicle. Spiking frequency increased when neurons were exposed to an LED light and with application of $10\mu \mathrm{M}$ CTZ but did not change with application of vehicle.

In the LMO7 expressing neurons, the LED treatment resulted in a significant increase in firing rates from pre-treatment (mean: 6.92, 95% CI: [5.86, 7.97]) to post-treatment (mean: 12.63, 95% CI: [10.31, 14.95]), with a corrected $p$-value of 0.0000. As expected, the vehicle treatment did not exhibit a significant effect (pre-treatment mean: 5.09, 95% CI: [4.27, 5.90]; post-treatment mean: 4.39, 95% CI: [3.63, 5.15], $p>0.05$). CTZ treatment by addition of hCTZ to the cultures for a final concentration of $10\mu \mathrm{M}$ increased firing rates notably (pre-treatment mean: 8.52, 95% CI: [7.26, 9.78]; post-treatment mean: 14.54, 95% CI: [11.94, 17.13], corrected $p$-value = 0.0002).

Similarly, in the LMO11 group, the LED treatment indicated a profound elevation in firing rates (pre-treatment mean: 4.48, 95% CI: [4.14, 4.81]; post-treatment mean: 34.75, 95% CI: [29.11, 40.39], corrected $p$-value of 0.0000. The vehicle treatment showed a slight increase that did not reach statistical significance (pre-treatment mean: 6.75, 95% CI: [5.88, 7.61]; post-treatment mean: 7.10, 95% CI: [6.16, 8.04], corrected $p$-value p>0.05). In contrast, the CTZ treatment showed a significant enhancement of neural activity (pre-treatment mean: 12.91, 95% CI: [12.20, 13.62]; post-treatment mean: 16.65, 95% CI: [15.13, 18.17], corrected $p$-value = 0.0002).

These results indicate that the optogenetic elements in LMOs 7 and 11 can be activated by both physical and biological light sources and that bioluminescence can provide sufficient excitatory drive to increase spontaneous activity in a population of neurons.

3.7.

LMO Coupling Efficiency Versus Bioluminescence Light Emission

We previously found that radiance decreased for opsin-tethered light emitters, consistent with a loss of photon emission to direct energy transfer to the nearby opsin chromophore.²¹ We further observed a larger decrease for LMO7 (NCS2-VChR1) versus LMO3 (sbGLuc-VChR1). This is consistent with FRET from the light emitter to opsin being more efficient for NCS2 to VChR1 (LMO7) than for sbGLuc to VChR1 (LMO3).

We saw similar tendencies with the new set of LMOs tested here (Fig. 9). The two LMOs that basically lacked coupling efficiencies, LMO7.3 (NCS2-VChR1-NCS2) and LMO8 (NCS2-ChRger3), displayed the highest per cell bioluminescence, with LMO7.3 showing the highest light emission, probably due to the two copies of NCS2 per opsin. In contrast, LMOs with efficient coupling (LMO10, LMO11) had decreased radiance, presumably due to better FRET compatibility with the opsin chromophore (GeNL_SS > VChR1, GeNL_SS > ChRmine).

Fig. 9

Coupling efficiency versus bioluminescence produced by LMOs. Coupling efficiencies as determined throughout the above studies are plotted on the left, and bioluminescence measured from each LMO is plotted to the right. Bioluminescence produced by each LMO was measured in HEK293 cells in a plate reader with $100\mu \mathrm{M}$ concentration of luciferin ($n=8$ for each LMO). Readings were taken immediately after addition of substrate. Bioluminescence was normalized to expression of fluorescent reporter in each LMO using fluorescence ROI analysis in ImageJ.