Full-field exposure of larval zebrafish to narrow waveband LED light sources at defined power and energy for optogenetic applications

Abstract

Background:

Optogenetic approaches in transparent zebrafish models have provided numerous insights into vertebrate neurobiology. The purpose of this study was to develop methods to activate light-sensitive transgene products simultaneously throughout an entire larval zebrafish.

New method:

We developed a LED illumination stand and microcontroller unit to expose zebrafish larvae reproducibly to full field illumination at defined wavelength, power, and energy.

Results:

The LED stand generated a sufficiently flat illumination field to expose multiple larval zebrafish to high power light stimuli uniformly, while avoiding sample bath warming. The controller unit allowed precise automated delivery of predetermined amounts of light energy at calibrated power. We demonstrated the utility of the approach by driving photoconversion of Kaede (398 nm), photodimerization of GAVPO (450 nm), and photoactivation of dL5**/MG2I (661 nm) in neurons throughout the CNS of larval zebrafish. Observed outcomes were influenced by both total light energy and its rate of delivery, highlighting the importance of controlling these variables to obtain reproducible results.

Comparison with existing methods:

Our approach employs inexpensive LED chip arrays to deliver narrow-waveband light with a sufficiently flat illumination field to span multiple larval zebrafish simultaneously. Calibration of light power and energy are built into the workflow.

Conclusions:

The LED illuminator and controller can be constructed from widely available materials using the drawings, instructions, and software provided. This approach will be useful for multiple optogenetic applications in zebrafish and other models.

Introduction

Light-activated transgene products provide experimental opportunities to convert cellular labels (Hatta et al., 2006; Patterson and Lippincott-Schwartz, 2002), control transgene expression (Mruk et al., 2020), integrate calcium fluxes (Fosque et al., 2015), and induce cellular oxidative damage (He et al., 2016; Qian et al., 2019), each with temporal precision defined by the time at which samples are exposed to narrow-waveband light sources. This technology has found particular utility in the transparent zebrafish model (White et al., 2008), in which optogenetic and chemoptogenetic approaches in the intact animal have provided numerous insights into the biology and pathophysiology of the vertebrate nervous system (Alyenbaawi et al., 2021; Dukes et al., 2016; McLean and Fetcho, 2009; Teh et al., 2010; Xie et al., 2020).

Optogenetic reagents can be activated with high spatial resolution in transgenic zebrafish. For example, a laser scanning confocal microscope system allowed photoactivation of a fluorophore within individual neuronal mitochondria of the intact larval CNS, with organelle-level precision (Dukes et al., 2016). However, in many circumstances, it is desirable to activate an optogenetic transgene product throughout the zebrafish. Examples include: simultaneously labeling or inducing biochemical events in specific cellular populations that are widely distributed throughout the zebrafish, but targeted selectively using lineage-specific cis-regulatory elements (McLean and Fetcho, 2009; Xie et al., 2020); or maintaining colonies of transgenic lines whose transgene product is not visible by epifluorescence microscopy prior to photoactivation (Dukes et al., 2016). These applications necessitate whole-larval illumination. Current approaches for delivering narrow waveband light experimentally to entire zebrafish frequently employ a microscope with light sources intended for fluorescence imaging (McLean and Fetcho, 2009) or a dedicated optogenetics light source based on a laser or LED system. These solutions present some limitations. Delivery of light through a microscope demands substantial investment in optical equipment that remains unavailable for image acquisition while samples are being exposed, a particular challenge when long exposure times are necessary, or when a large number of experimental groups or conditions will be tested. Dedicated optogenetics light sources are costly and many are optimized specifically for activating channel opsins rather than the broader range of applications used in zebrafish. Furthermore, both methods present potential limitations in area of coverage, flatness of illumination field, and power. The limited lifetime of some light sources is also a potential concern, particularly for prolonged exposures using expensive equipment. Finally, activation of optogenetic reagents may be dependent on both the total amount of light energy and its rate of delivery. Methods to control these parameters are critical to optimize the experimental approach and enhance reproducibility.

We recently reported a novel transgenic zebrafish model in which production of singlet oxygen (¹O₂) within neuronal mitochondria of the intact living brain was driven by a chemoptogenetic photosensitizer (Xie et al., 2020). We used inexpensive, high-power, narrow-waveband LED chip arrays to direct light-induced ¹O₂ production. This presented several challenges, including: heat dissipation; light beam alignment; control of LED source to sample distance; light path optimization; ensuring a flat illumination field for exposing whole zebrafish larvae; and precise control of exposure energy and its rate of delivery for experimental replicability. To address these challenges, we developed an illuminator stand and controller. Here, we provide sufficiently detailed documentation to allow assembly of updated versions of the illuminator and controller from widely available materials, source code for programming the controller, and instructions for its operation. We verified that the illuminator stand provides a sufficiently flat illumination field, avoids sample bath heating, and that the microcontroller unit allows accurate automated control of light power, exposure time, and total exposure energy. We activated three different optogenetic reagents throughout the CNS of transgenic zebrafish in vivo, employing LED arrays with non-overlapping wavebands. These examples demonstrate the utility of the light stand and controller for driving optogenetic events simultaneously throughout transgenic zebrafish and illustrate situations in which controlling light power and energy is important for experimental outcome and replicability.

Materials and methods

LED light sources:

The study used 100W LED chip arrays (Chanzon, 1DGL-JC-100W-395, 1DGL-JC-100W-BL, 1DGL-JC-100W-660; Amazon; electronic components and suppliers are listed in supplemental tables S1 and S2). Holes were drilled and tapped in a CPU heat exchanger/fan assembly (Rosewill, RCX-Z90-AL), the back of the LED chip was coated in thermal paste (Arctic Silver, 5 AS5–3.5G), and the chip bolted through its mounting holes to the heatsink. Power supply leads were soldered to the LED terminals to connect to the power output of the controller, and a reflector/collimator assembly (Chanzon, 1LED-LENS-44MM-KIT) was mounted to the front of the LED chip.

Light stand:

The light stand was built from 6061 aluminum alloy (Wallace Metals, Pittsburgh, PA; drawings for the parts are provided in supplemental figure S1). The base and stage were fabricated on a Bridgeport 3-axis milling machine. For the base, ½-inch thick 6061 aluminum plate was squared to dimensions, and holes were drilled and counterbored to accept ¼ −20 socket head cap screws to attach the lower end of the legs. For the stage, ¼-inch thick 6061 aluminum plate was squared to dimensions, holes were drilled to accept ¼−20 socket head cap screws, then a second set of holes drilled and tapped to accept M4 × 0.7 × 50mm bolts to attach the LED/CPU cooler assembly. Finally, the center recess was milled to dimensions to accept the glass insert. The legs were fabricated on a Hardinge lathe from 1-inch diameter 6061-aluminum round stock. The ends were faced to square, turned to length, and then drilled and tapped to a ¼−20 thread, to ¾ inch depth. After fabrication, the pieces were anodized in Black Hard Coat, to comply with industry standards, and assembled (supplemental figure S2). The four legs were attached to the baseplate from the underside using ¼−20 × 1” bolts, such that the socket head caps sat completely within the counterbore, allowing the base to rest firmly on a flat surface. The stage and shade clips were then attached to the upper ends of the legs using ¼−20 × 1” bolts. Next, the LED/collimator/heatsink/fan assembly was suspended from the underside of the stage through the mounting holes on the CPU cooler, using M4 × 0.7 × 50mm bolts, 0.7mm washers and nuts, and the height adjusted so that the upper side of the collimator lens did not protrude into the stage aperture. Finally, the glass insert was placed into the recessed area around the aperture on the upper side of the stage. A light-tight, opaque black cardboard shade was made to cover the stage area completely in order to prevent light leakage and operator exposure during use (supplemental figure S3). The clips were designed to hold the shade firmly at its back, allowing it to be gently lifted from the front to access the stage.

Controller:

The controller was built inside an ABS plastic junction box (LeMotech, #Lm201803261124; parts are listed in supplemental table S2) which was cut and drilled to provide connector and component mounting holes prior to assembly (supplemental figure S4). 1/8” thickness ABS plastic sheet was cut to sit inside the bottom of the junction box, and electronic components were mounted to the sheet using nylon standoff spacers. Components were then connected according to the circuit diagram provided (supplemental figure S5). The CPU cooling fan and microcontroller, which both require 12VDC input, were powered from the same 21V or 30V power source used for the LED array through a voltage stepdown board (DROK, #LM2596). The LED power circuit was switched by a MOSFET module (Anmbest, #MD114) and the MOSFET gate was controlled by a digital PWM output from an ATmega328P microcontroller board (Arduino Uno R3; Arduino USA Store). The user interface comprised a 4-digit 7-segment display with I2C backpack (#1002; Adafruit.com), with user inputs provided by momentary-on push buttons (Gikfun, 12mm) and a 10kΩ rotary potentiometer (Uxcell, #a15040700ux0380). External connections to the power supply, LED and fan were made through sockets (Switchcraft, #L722AS; Mouser electronics) mounted on the junction box, and plugs (Switchcraft, #S761KS17; Mouser Electronics) soldered to the connecting leads. Firmware for the microcontroller was developed using the Arduino IDE (source code is supplied in supplemental data). An operating manual for the controller is provided as supplemental material.

Light and temperature measurements:

Spectra for each of the light sources were measured using a Stellarnet spectrometer with SpectraWiz software (BLK-CXR; Stellarnet, Tampa, FL). Custom MATLAB scripts were used to calculate peak ± half width at half height wavelengths. Power was measured using a FieldMaxII-TO power meter (SKU 1098579; Coherent, Santa Clara, CA) with a low power (100W – 1W) thermopile sensor (PS10Q, Coherent). The sensor was positioned with its aperture in the center of the stage facing the light source and zeroed in darkness prior to making measurements. The meter reading was allowed to stabilize for 20 seconds at each new illumination level prior to recording data. Images of illumination patterns were made using a monochrome USB 3.0 camera (BFS-U3-13Y3M; FLIR, Wilsonville, OR), with a 28mm lens (Pentax; B+H, New York) mounted on a copy stand directly above the stage. Images were captured using SpinView (v.1.23.0.27; FLIR). The lens was focused on an opaque plastic screen placed over the stage aperture, then set to f11. Gamma and gain were set to 1 and 0 respectively, and the exposure time adjusted until there were no saturated pixels with the light stand in its brightest configuration, following which exactly the same settings were used for every experiment. For each different experimental configuration, 20 identical images were captured and averaged in MATLAB to eliminate pixel noise. Normalized color scale images were made by scaling each pixel to the mean value of the brightest 0.5% of pixels in the image. Sample bath temperature measurements were made by immersing an epoxy-coated waterproof thermistor (#3950NTC; Adafruit.com) in water, in the same dishes used for zebrafish experiments. The thermistor was incorporated into a voltage divider circuit connected to an analog input port on an Arduino Uno board; the potential difference across the fixed resistor was recorded in 10-bit binary at 1s intervals, converted to V, then used to calculate the thermistor resistance. Temperature was then determined by linear interpolation of the thermistor lookup table using custom MATLAB scripts. Water temperature was equilibrated to room temperature prior to starting each experimental recording; recordings of water at 85°C or 4°C equilibrating to room temperature over time confirmed that the thermistor allowed accurate and reliable continuous monitoring of water bath temperature.

Zebrafish

All experiments were completed following full review and approval from the University of Pittsburgh Institutional Animal Care and Use Committee. Transgenic zebrafish lines were Tg(elavl3:gal4-vp16)^zc87 (Stevenson et al., 2012), Tg(UAS:kaede)^rk8 (Hatta et al., 2006), Tg(elavl3:GAVPO)^st1003 (Mruk et al., 2020), Tg(UAS:egfp) (Ilin et al., 2021), Tg(eno2:gal4ff)^pt425 (Xie et al., 2020), Tg(UAS:COX4-COX8-dL5**-mCer3)^pt427 (Xie et al., 2020). Zebrafish embryos were raised in E3 buffer (5mM NaCl, 0.17mM KCl, 0.33mM CaCl₂, 0.33mM MgSO₄; all chemicals were supplied by Sigma, St. Louis, MO) at 28.5°C. Transgenic zebrafish were identified by expression of fluorescent reporters using a stereozoom microscope (MVX10; Olympus, Center Valley, PA). Zebrafish expressing Kaede or GAVPO were raised in the dark prior to photoactivation. Zebrafish expressing dL5** were raised under cyclic illumination of 10 hours dark:14 hours light, with a green (515 nm) safelight that does not activate the dL5**/MG2I complex. MG2I was synthesized as described previously (He et al., 2016) and added to the larval E3 buffer at 72hpf at a final concentration of 500nM.

Zebrafish were exposed to light in one of four different experimental configurations (table 1 and supplemental figure S6): (i) in a 35mm dish with a 10mm glass coverslip bottom (MatTek, Ashland, MA), anesthetized in 0.015% tricaine and immobilized in 1.5% low melting point agarose with the dorsal surface of the head in contact with the coverslip (Kaede experiments; single zebrafish; maximum light fall off across zebrafish < 2%); (ii) in a 35mm dish with a 14mm glass coverslip bottom, free swimming in a 250 μL meniscus of E3 buffer located on the coverslip (GAVPO experiments; 6 zebrafish in the same dish; maximum light fall off across coverslip < 6%); (iii) in a 35mm dish, free swimming in 2 mL of E3 buffer (GAVPO experiments 16 hour exposure only; 6 zebrafish in the same dish; maximum light fall off across entire dish < 25%); (iv) in a 35mm dish with a 14 mm glass coverslip bottom, anesthetized in a 250 μL meniscus of 0.015% tricaine in E3 buffer located on the coverslip (dL5**/MG2I experiments; up to 18 zebrafish in the same dish; maximum light fall off across coverslip < 6%). The dish was centered on the glass stage of the light stand for exposure.

Table 1:

Summary of experimental design:

Figure	Fluorophore (light wavelength)	Age at exposure	Arena diameter	Zebrafish/arena	Anesthetic	Medium	Light stand configuration
3	Kaede (398 nm)	3 dpf	10 mm	1	0.015% tricaine	1.5 % LMP agarose mounting + E3	#1
4	GAVPO (450 nm)	4 dpf	14 mm or 35mm	6	None	E3 250 μL meniscus on coverslip or 2mL	#1 or #2
5	dL5**/MG2I (661 nm)	5 dpf	14 mm	18	0.015% tricaine	E3 250 μL meniscus on coverslip	#1

Intravital microscopy:

Zebrafish larvae were anesthetized in 0.015% tricaine and mounted in 1.5 % low melting point agarose in E3, either on the side or inverted with the dorsal surface of the head in contact with the 10mm coverslip of a 35mm glass bottom dish (MatTek, Ashland, MA) and the solidified agarose covered in E3 buffer, as reported previously (Van Laar et al., 2020). Z-planes were acquired from below through the glass coverslip, using a Nikon Ti2E inverted microscope with a resonance laser scanning confocal system (AXR; Nikon). Brain images were acquired to a depth of 200μm from the dorsal surface of the brain using a 20x long working distance water-immersion objective (CFI Apo Lambda-S, NA 0.95; Nikon). Whole larval images were acquired from the side of the zebrafish using a 4x objective (CFI Plan Fluor, NA 0.13; Nikon). Image analysis, generation of maximum projection Z-plane grayscale, color intensity scale and ratiometric images were carried out using NIS-Elements (Nikon Instruments, Melville, NY). Quantitative data were analyzed, and graphs plotted using Prism 9.4.1 (GraphPad, San Diego, CA).

Neurobehavioral analysis:

Locomotor function was analyzed as reported in our prior work (Cario et al., 2011; Zhou et al., 2014). Larvae were transferred to 96 well plates with black well surrounds and glass bottoms under green LED safelight illumination using a large-bore Pasteur pipette with a flame-polished aperture, then acclimatized to the recording chamber for 30 minutes at 28.5°C. The visual motor response was elicited using green (515nm, 120 lx) light (Burton et al., 2017) and recorded with a USB 3.0 camera (#FL3-U3-13Y3M-C; Point Grey Research, Richmond, BC, Canada) under infrared illumination (#BL812–880; Spectrum Illumination, Montague, MI). Video recordings were analyzed offline using our published open-source MATLAB applications LSRtrack and LSRanalyze (Cario et al., 2011). All data were derived from recordings with < 5% total tracking errors.

Results:

Light Stand

We developed a light stand that illuminates larval zebrafish uniformly for optogenetics applications, without heating the water bath, using powerful but inexpensive narrow-waveband LED chip arrays (figure 1; instructions and drawings to build the stand from readily available materials are provided in supplemental figures S1 – S3). The LED array was mounted on a CPU heatsink/cooling fan assembly and suspended below a glass stage that supported the sample (figure 1A, B). This configuration allowed the spacing between the light source and sample to be adjusted by raising or lowering the LED/cooler assembly using locking nuts, while maintaining a fixed distance once set. For a sample mounted in a glass-bottom dish, the light path passed through air, the glass stage, and the coverslip, avoiding variability resulting from light absorption or scattering in sample buffer or mounting medium. The absence of direct contact between the LED and stage prevented heat transfer by conduction, and the constantly circulating air below the stage, driven by the cooling fan, prevented heat transfer by convection (figure 1B). We previously showed that sample buffer in a 35mm dish was not heated by exposure to 661 nm light at 160 mW/cm² (irradiant flux through the plane of the stage; subsequently referred to as ‘power’ for brevity) for 60 minutes (Xie et al., 2020). We investigated whether different light wavelengths or brighter light sources would transfer heat to the sample dish, using a thermistor and voltage divider circuit to monitor sample bath temperature as previously described (Xie et al., 2020). The thermistor probe was wrapped in aluminum foil to mitigate energy transfer directly to the probe by radiation from the light source, so that the readings reflected the water temperature, as confirmed by cooling and heating curves for water with initial temperatures of 85°C and 4°C respectively in the absence of light (figure 1C). After 60 minutes exposure at a power of up to 250mW/cm² (900 J/cm² radiant exposure at the plane of the stage, referred to as ‘energy’ for brevity), LED sources with peak wavelengths of 450 nm (blue) or 661 nm (far red) caused < 1°C increase in water temperature relative to room temperature baseline (figure 1C; the final temperature readings for each LED at each power tested are shown as a function of total cumulative energy exposure in the inset panel; note the superimposed data points resulting from closely-aligned values for both light wavelengths). 60 minutes of exposure to either light source at 500 mW/cm² (1.8 kJ/cm² cumulative energy) caused ≈ 5°C warming of sample water, although the final temperature remained below the standard water temperature used for housing larval zebrafish (28.5°C). Optogenetic events can be induced experimentally well below these power and energy levels, in a range that does not cause appreciable water heating. The experiments shown below demonstrate robust optogenetically-induced changes at energy exposures less than 100 J/cm², and at powers below 200 mW/cm².

Figure 1:

LED light stand provides full-field illumination to larval zebrafish without heating the water. A: Photograph of the light stand. B: Schematic lateral projection of the light stand illustrating the locations of key components and the direction of airflow across the heatsink and heat exchanger. C: Bath temperature in a 35mm MatTek dish placed in the center of the stage. The graphs show recordings during 1 hour of exposure to a 405nm blue LED (left) or 661nm far red LED (right) at the powers indicated (expressed as mW/cm² at the stage). Curves for water with initial temperatures of 85°C or 4°C (gray) equilibrating to room temperature are shown as controls to confirm that the thermistor functioned as expected. The inset panel shows the relationship between the final bath temperature after 1 hour of light exposure, and the total light energy passing through the stage during the experiment (expressed as kJ/cm²); blue-outline squares show data from the 450nm light source and red-outline circles from the 661nm source. D: The spatial Illumination profile across the stage was measured by photographing a white plastic screen from above, while a 661nm LED was illuminated with the stand in three different configurations: #1. LED 25mm from stage with collimator lens and reflector (left column of images), #2. LED without lens or reflector, 25mm from stage (center column); #3. LED without lens or reflector 5mm from stage (right column). Top row: 8-bit grayscale images averaged from 20 pictures, using the same acquisition settings for each configuration. Bottom row: relative intensity color maps, in which each pixel was scaled to the brightest 0.5% of pixels in the averaged image for each configuration to compare spatial fall-off. The distance from the center of the stage is shown in mm. The white bars show the approximate size of a larval zebrafish at 5 dpf. E: Graph of relative light intensity as a function of distance from the center of the stage, calculated from the intensity colormap data in panel D. Colored lines are shown for each different configuration (configuration #1, black; configuration #2, blue; configuration #3, red). Verification of the photographic data from configuration #1 using a second method – positioning the center of a power meter sensor at different distances from the center of the stage – is shown as large green squares. For comparison, the coverslips of the standard sizes of glass-bottomed dishes used for zebrafish imaging, and a picture of a larval zebrafish at 5dpf, are superimposed below.

We next addressed the uniformity of illumination across the stage by replacing the glass stage insert with a translucent white plastic screen and photographing the illumination pattern from above. We compared three different optical configurations (figure 1D): #1 – LED array 25mm from stage, separated by a collimator/reflector assembly (leftmost column of images); #2 – LED 25mm from stage but reflector and collimator removed (middle column); #3 – LED 5mm from stage with no reflector or collimator (right column; this configuration was used in our previous work (Xie et al., 2020)). Averaged grayscale images taken with the same acquisition settings showed that all three configurations illuminated the center of the stage more than the edges (figure 1D). To compare the spatial fall-off in illumination between configurations, we re-calculated pixel values relative to the brightest area in each image (defined as the mean value of the 0.5% pixels with highest 8-bit grayscale values; color scale images, bottom row of figure 1D). Overall, the collimator/reflector assembly (configuration #1) provided the brightest illumination at the center of the stage (figure 1D, upper row of images) but showed significant fall-off at the edges of the stage. Removing the collimator and reflector (configuration #2) mitigated fall-off but greatly reduced power. Moving the LED without the collimator closer to the stage (configuration #3) improved power but led to steeper fall-off than the other configurations. We confirmed the photographic findings for configuration #1 using a power meter. The sensor was positioned precisely at one of four linear distances from the center of the stage (figure 1E, green squares show the mean of 4 measurements at each distance). There was strong agreement between measurements obtained using the two methods.

The illumination profile at the center of the stage in configuration #1 was sufficiently flat to irradiate larval zebrafish in their entirety; fall off in light intensity was < 2% from the center of the stage along the length of a larval zebrafish at 5dpf, which is likely to be experimentally insignificant (figure 1D, E). Across a 14mm coverslip, light fall-off was < 6% from the center of the stage, allowing simultaneous exposure of multiple larvae to comparable amounts of light energy (figure 1E; supplemental figure S6). For prolonged exposures, where the lower power necessary to deliver light energy does not require use of a collimator, configuration #2 provided a flatter illumination profile allowing free swimming zebrafish to be exposed in 35mm dishes with maximum light fall-off < 25% (see supplemental figure S6; the motility of the zebrafish during these prolonged exposures ensures each of the larvae receives similar exposure).

Controller

We next constructed a controller unit (figure 2A) to allow precise and reproducible exposure of the samples to light (figure 2B). The controller is based on an ATmega328P microcontroller board and, in addition to supplying electrical power to both the LED and the cooling fan, provides four functions (figure 1C). First, the LED power circuit is controlled by a MOSFET driver module whose gate is switched by a 980Hz pulse width modulation (PWM) digital output from the microcontroller board. This allows the light output from the LED array to be adjusted by altering the PWM duty cycle (the proportion of each ≈1.02 ms time period for which the LED is illuminated). Function F1 allows adjustment of the PWM duty cycle in integer steps between 0 (always off) and 255 (always on; each step is equivalent to an additional 3.99 μs of illumination per cycle at 980Hz). Figure 1D shows the monotonically increasing relationship between PWM duty cycle and light power (mW/cm²) at the stage for three different LED arrays. The rate of delivery of light energy for each LED is adjustable over a wide range; modest non-linearity is unimportant, as power is calibrated against an external power meter for each experiment, following which PWM is not adjusted. Function F2 provides a timer that illuminates the LED at the chosen PWM value for a predetermined length of time selected by the user. Function F3 allows the user to input the reading from a power meter (in mW/cm²) placed on the stage with the LED illuminated at the selected PWM value. Function F4 allows the user to specify a desired exposure energy (in J/cm²). The unit then calculates the exposure time from the calibration entered in function F3 and illuminates the LED at the selected PWM duty cycle value with precise timing to deliver the specified amount of energy. Figure 1E shows an example in which an exposure energy of 60 J/cm² was requested at a power calibration of 180 mW/cm²; the graph shows light power recorded continuously at the stage during the exposure. The exposure ran for 333.33 seconds, with the area under the time/power curve showing an actual exposure of 60.046 J/cm², giving an acceptable error of < 1%. Multiple different experimental replicates using different calibrated powers, requested energies, and LEDs with different light wavelengths, showed actual exposures within ± 2% of the requested value.

Figure 2:

Control of irradiance, time, and radiant exposure. A: Photograph of the controller unit. B: Diagram showing the external connections of the controller unit. C: The four functions provided by the controller: F1, adjustment of LED PWM to control mean irradiance (power) at the sample plane; F2, timed exposure; F3, input irradiance calibration in mW/cm² from external power meter; F4, automatic delivery of specified radiant exposure (energy) in J/cm² based on calibration input in F3. D: Relationship between pulse width modulation duty cycle (x-axis; adjusted using function F1) and power at stage (y-axis) for three different LED arrays, 398nm (purple), 450nm (blue), 661nm (red). Correlation coefficients are shown for each LED. E: Example automated energy exposure using a 398nm UV_A LED array. The power of 180 mW/cm² was measured at the stage using a power meter, entered into the controller unit using function F3, then an exposure of 60 J/cm² was requested using function F4. The graph shows the output from the power meter over time; the exposure was started 2 minutes into the recording. The area under the curve was calculated to compare the requested and actual light exposure.

Example applications

The ability to expose whole zebrafish larvae to predetermined amounts of narrow waveband light energy at calibrated power has multiple applications in zebrafish optogenetics. We chose three complementary examples to demonstrate how the light stand and controller unit can be used. The three example applications show optogenetic reagents that are activated at distinct light wavelengths, using different experimental configurations (supplemental figure 6), at different ages. These examples illustrate ways in which control of exposure energy and power can be important in practice.

1. Photoconversion of Kaede

Kaede, a protein derived from stony coral (Trachyphyllia geoffroyi), exhibits green fluorescence (peak excitation 509nm, emission 518nm) that is converted to red fluorescence (peak excitation 574nm, emission 582nm) by exposure to UV_A radiation in the range 350 – 400nm (figure 3A) (Ando et al., 2002). One application of this property in transgenic zebrafish is to mark all cells expressing Kaede under tissue-specific cis-regulatory sequences at a specific developmental stage, allowing their subsequent identification at later time points (McLean and Fetcho, 2009). This approach necessitates full-field illumination to ensure that all cells expressing the transgene are photoconverted at the initial time point.

Figure 3:

Photoconversion of Kaede in CNS neurons of live transgenic zebrafish by 398nm UV_A radiation. A: The coral protein Kaede (Japanese for ‘maple’) is converted from green to red fluorescence by exposure to UV_A light. B: Normalized spectrum for the UV_A LED array used in this experiment. The peak wavelength was 398 nm and half-width at half height was 5.9 nm. C: Maximum intensity confocal Z-plane projections of a 3 dpf Tg(elavl3:gal4-vp16); Tg(UAS:kaede) zebrafish (dorsolateral view) with green (excitation 488 nm; emission 499 – 544 nm) and red (excitation 561 nm; emission 571 – 750 nm) channels overlaid. The same zebrafish was imaged before (above) and after (below) exposure to 50 J/cm² 398 nm UV_A radiation at 150 mW/cm², to illustrate photoconversion of Kaede throughout the brain and spinal cord. D: Maximum intensity confocal Z-plane projections of the brains of five Tg(elavl3:gal4-vp16); Tg(UAS:kaede) zebrafish at 3dpf, viewed from the dorsal surface (the Z-stack extends from the brain surface to a depth of 200μm). The same acquisition settings were used throughout. The green (first column) and red (second column) channels are shown separately for the first zebrafish; ratiometric [red]/[green] images are shown for all 5 zebrafish in the remaining columns, colored according to the ratiometric scale shown to the right. The top row shows baseline images for each zebrafish; the subsequent rows show the same zebrafish after cumulative exposure to 10, 25, 50 or 100 J/cm² UV_A radiation (λ = 398 nm) at 150 mW/cm². Sequential exposure times resulting in these cumulative exposures are shown in blue to the left. E – G: Graphs showing the mean signal within the brain in the (E) green or (F) red channels, and (G) the mean [red]/[green] ratio, for each zebrafish shown in panel D. Each data point shows a single zebrafish, bars show mean ± SE. p < 0.0001****, 0.001***, 0.01**, 0.05* (repeated measures 1-way ANOVA with Geisser-Greenhouse correction for sphericity, and Šidák multiple comparisons test to compare values after each sequential exposure).

We combined the light stand (optical configuration #1; figure 1 and supplemental figure S6) and controller with a narrow bandwidth UV_A-emitting LED array (398 ± 5.9nm; figure 3B) to photoconvert Kaede expressed in a pan-neuronal pattern in transgenic Tg(elavl3:gal4); Tg(UAS:kaede) zebrafish at 3 days post-fertilization (dpf). 50 J/cm² UV_A radiation λ = 398nm @ 150 mW/cm² was sufficient to convert green to red fluorescence throughout the entire CNS of a larval zebrafish viewed from the side (figure 3C). Taking advantage of the ability to deliver energy accurately and at constant power, we investigated how exposure energy influences photoconversion (figure 3D, E, F, G; supplemental figure S7). Individual zebrafish were embedded in agarose, exposed to incremental amounts of UV_A energy, and imaged sequentially after each exposure (figure 3D; supplemental figure S7). At baseline (figure 3D, top row), bright green fluorescence signal was visible throughout the brain, but red fluorescence signal was barely detectable (figure 3D, left column). With increasing cumulative light exposure, signal in the green channel decreased while signal in the red channel increased (figure 3D, second column). Ratiometric images of [signal in red channel]/[signal in green channel] illustrated the progressive conversion of Kaede from green to red fluorescence with increasing exposure energy and showed remarkable reproducibility between different zebrafish (figure 3D, remaining columns). Measurement of the mean pixel value in each channel within the CNS for each zebrafish showed that signal in the green channel was barely detectable by 25 J/cm² exposure, and it continued to decline further with increasing exposure (figure 3E). Strong signal was seen in the red channel by 25 J/cm² exposure. Further exposure did not significantly increase CNS red signal overall (figure 3F), but isolated pixels became brighter. Analysis of the mean pixel-by-pixel red/green ratio within the CNS showed conversion from green predominance to red predominance by 10 J/cm² exposure, with no significant increase in mean red/green ratio above 25 J/cm² (figure 3G). These data suggest that 25 J/cm² may be adequate for most applications, but complete abrogation of green signal at this developmental point may require higher exposure energy.

2. Photodimerization of GAVPO

Gal4-UAS genetics provides a powerful method for expressing transgenes in the zebrafish CNS (Asakawa et al., 2008; Distel et al., 2009). Gal4 driver lines (constructed by transgenesis, gene trap or gene knock-in approaches) allow stringent spatial restriction of transgene expression, directed by tissue-specific cis-regulatory elements. The recent introduction of GAVPO (Mruk et al., 2020), a blue light activated Gal4 derivative, has further extended the utility of this system to add temporal control of transgene expression. Gal4 dimerizes to form a DNA binding domain that recognizes the UAS enhancer sequence in the responder cassette to activate transgene expression. In GAVPO, the endogenous Gal4 dimerization domain is replaced by the light-sensitive dimerization domain of VVD (Mruk et al., 2020; Zoltowski and Crane, 2008), a blue light photoreceptor from the filamentous fungus Neurospora crassa (Schwerdtfeger and Linden, 2003). Consequently, the transcriptional activity of GAVPO is dependent on illumination with blue light (figure 4A).

Figure 4:

Photodimerization of GAVPO using 450nm blue light to induce gene expression in CNS neurons of live transgenic zebrafish. A: GAVPO is a synthetic transcription factor including the DNA-binding domain of Gal4, an optimized version of the VVD blue light-activated dimerization domain, and a p65 transactivation domain. Exposure to blue light results in dimerization, enabling the Gal4 domain to bind to UAS (the cognate DNA sequence) and activate gene transcription. B: Normalized spectrum for the blue LED array used in this experiment. The peak wavelength was 450 nm and half-width at half height was 9.5 nm. C: Maximum intensity confocal Z-plane projections of Tg(elavl3:GAVPO); Tg(UAS:egfp) zebrafish at 5 dpf (dorsolateral view) showing GFP expression. Zebrafish were raised in darkness (upper image) or exposed to 100J/cm² 450 nm light at 1.7 mW/cm² for 16 hours Both images were captured using identical settings to illustrate transgene induction by light in neurons throughout the zebrafish CNS. D: Maximum intensity confocal Z-plane projections encompassing the most dorsal 200μm of the CNS in four different Tg(elavl3:GAVPO); Tg(UAS:egfp) zebrafish at 5dpf. The same acquisition settings were used to detect GFP signal throughout. The top left image shows a zebrafish raised in the dark, the remaining images show zebrafish exposed to a total of 100 J/cm² blue light at different powers between 1.7 mW/cm² and 120 mW/cm² as indicated. The main panels show grayscale images; the inset panels show the 12-bit pixel values in the raw images, encoded according to the color intensity scale shown. E: Graph showing the mean signal within the brain in the GFP channel for 5 groups each containing n = 6 zebrafish, exposed to 100 J/cm² blue light at different powers between 1.7 mW/cm² and 600 mW/cm² as indicated. Each data point shows a single zebrafish, bars show mean ± SE. p < 0.001***, 0.01** (Brown-Forsyth 1-way ANOVA, with Dunnett multiple comparisons test to compare each group to unexposed controls). Note the discontinuous y-axis to accommodate the substantial increase in fluorescence in the 1.7 mW/cm² group. F: Graph of mean GFP signal intensity as a function of exposure time for the data shown in panels D and E. A linear regression line and R² value are shown to illustrate the relationship.

We combined the light stand and controller with a narrow bandwidth blue light-emitting LED array (450 ± 9.5nm; figure 4B) to photodimerize GAVPO and induce pan-neuronal GFP expression in transgenic Tg(elavl3:gavpo); Tg(UAS:egfp) zebrafish at 3 – 4 dpf. 100 J/cm² of blue light delivered over 16 hours between 4 dpf and 5 dpf was adequate to activate GFP expression throughout the CNS of a larval zebrafish viewed in oblique lateral projection (figure 4C). We next exploited the ability to control the power of the light source, while exposing samples to a constant amount of light energy, to determine how the rate of light delivery influenced the activation of GAVPO (figure 4D, E, F; supplemental figure S8). For exposure at 12 – 600 mW/cm², groups of 6 free-swimming zebrafish were exposed simultaneously in a 250μL meniscus of E3 buffer over the 14mm coverslip of a glass bottom dish, using optical configuration #1. For the > 16-hour exposure necessary in the 1.7mW/cm² group, 6 free-swimming zebrafish were exposed in 2mL of E3 buffer in a 35mm dish, using optical configuration #2 to mitigate light fall-off across the larger arena. Exposures were completed at 4dpf and GFP expression was evaluated in the larval CNS at 5 dpf by intravital confocal microscopy (figure 4D; supplemental figure S8). Embryos raised in the dark showed minimal ‘leaky’ GFP expression, and as expected this was increased robustly by exposure to light at 450nm. However, the same amount of light energy (100 J/cm²) resulted in substantially different GFP expression levels, depending on its rate of delivery and consequently on the time period of light exposure (figure 4D, E). Quantification of mean GFP fluorescence signal in the CNS of 6 replicate zebrafish in each group showed a strong and statistically significant relationship between decreasing light power and transgene expression at the same total energy. Thus, exposure to 1.7 mW/cm² over 16h 20m gave rise to bright green fluorescence throughout the CNS, whereas exposure to 12 mW/cm² for 2h 20m activated transgene expression significantly less. Exposure to 120mW/cm² for 13m 50s, or 600mW/cm² for 2m 46s, did not induce transgene expression significantly compared with dark-only controls. Consistent with these observations, there was a near-linear relationship between exposure time and transgene expression (figure 4F) at the same total cumulative exposure energy. These data suggest that the slow kinetics of gene expression (and the accumulation GFP) are important determinants of light-induced GAVPO-directed transgene activation.

3. Photoactivation of singlet oxygen production

When the di-iodine modified fluorogen MG2I binds to the fluorogen activating protein dL5**, exposure to far red light at 660 nm causes formation of singlet oxygen ¹O₂ (He et al., 2016) (figure 5A). Since ¹O₂ is both highly reactive and short lived, oxidative modification of cellular components occurs in close proximity to its site of production. dL5** expression can be directed to specific subcellular compartments by fusion to localization sequences or to other proteins, allowing stringent spatial restriction of the resulting oxidative damage. Coupled with the precise temporal control of ¹O₂ production inherent in its activation by light, and the ability to control the amount of ¹O₂ produced by varying the light exposure energy, the system provides powerful experimental opportunities to determine the cellular consequences of localized oxidative damage. This approach been used to target both mitochondria (Qian et al., 2019) and telomeres (Fouquerel et al., 2019) in vitro, and we recently reported its application to target neuronal mitochondria in zebrafish in vivo (Xie et al., 2020). Tg(eno2:gal4); Tg(UAS:cox4-cox8-dL5**-mCer3) (abbreviated to mito-dL5**) zebrafish express dL5** fused to the fluorescent protein mCerulean3 in the mitochondria of large projection neurons throughout the CNS. After preincubation of mito-dL5** zebrafish with MG2I, exposure to 80 J/cm² far-red light @ 160 mW/cm² abrogated neuronal mitochondrial respiration, causing neuronal bioenergetic collapse and depolarization, and resulting in loss of neurological functions including the visual motor response (VMR) (Xie et al., 2020).

Figure 5:

Photoinduction of singlet oxygen formation by dL5**-MG2I complex in the CNS neurons of live transgenic zebrafish using 661 nm light. A: When bound to the fluorogen-activating protein dL5**, exposure of the fluorogen MG2I to far-red light results in formation of highly reactive and short-lived singlet oxygen (¹O₂). B: Normalized spectrum for the far-red LED array used in this experiment. The peak wavelength was 661 nm and half-width at half height was 9.0 nm. C: Tg(eno2:gal4ff); Tg(cox4-cox8-dL5**-mCer3) zebrafish express the fluorogen-activating protein dL5** fused to the fluorescent protein mCerulean3 within neuronal mitochondria. Cyan fluorescence is visible throughout the brain, spinal cord, and peripheral nerves of live transgenic zebrafish (main panels). At high magnification, individual mCer3-labeled mitochondria are seen within axons of the lateral line nerve (inset panels). The upper panels show baseline images, and the lower panels show the same zebrafish following activation of ¹O₂ production in mitochondria, by exposure to 80 J/cm² 661 nm light at 160 mW/cm². D: Swimming speed of mito-dL5** zebrafish in response to cyclic illumination. Each graph shows a different experimental group (n = 14 to 18 zebrafish per group): no MG2I/no light control (black); 661 nm light-only control (blue); MG2I-only control (green); MG2I + 661 nm light 80J/cm² delivered at different power (red). Motor responses were elicited using a green safelight that does not activate ¹O₂ production. Zebrafish were transferred to a 96-well plate for the assay. Movements were captured by infrared videography and measured using automated video tracking software. Responses for each individual zebrafish were averaged over five illumination cycles. For each experimental group, mean frame-to frame larval centroid displacement was scaled to show instantaneous speed (gray traces). The colored markers show group mean ± SE swimming speed within each 1-minute time bin over the 20-minute illumination cycle. E: Mean swimming speed during the dark phase of the visual motor response shown in panel D. Data points show individual zebrafish, bars show mean ± SE; p < 0.0001**** vs. no chemical or light control (1-way ANOVA with Dunnett multiple comparisons test to compare each group to transgene-only controls). Color key for experimental groups is the same as panel D.

We combined the light stand and controller with a narrow bandwidth far red light-emitting LED array (661 ± 9 nm; figure 5B) to investigate how light power influenced phenotypic outcomes in this model at the same total energy exposure (generation of experimental groups and controls for these experiments is shown in supplemental figure S9A). Transgenic mito-dL5** zebrafish were bred onto the transparent Casper background (White et al., 2008). Larvae were incubated in MG2I from 3 dpf and exposed to light at 5 dpf under 0.015% tricaine anesthesia. 14 – 18 zebrafish per experimental group were located in a 250μL meniscus of E3 buffer over the 14mm coverslip of a glass bottom dish and exposed to 661nm far-red light using optical configuration #1. Exposure to 80 J/cm² far-red light at 160 mW/cm² caused fragmentation of neuronal mitochondria in MG2I-treated dL5** zebrafish, visible as a striking transition from elongated mCer3-labeled axonal mitochondria in the lateral line nerve to smaller, rounder structures (figure 5C; shown also by electron microscopy and quantified in (Xie et al., 2020)). After 2 hours of post-exposure recovery, motor responses to changes in green ambient illumination (which does not activate ¹O₂ production) were measured using a video tracking system to evaluate the functional integrity of neurons mediating swimming movements (figure 5D, E; supplemental figure 9B). As expected, all three control groups (mito-dL5** zebrafish without light or MG2I exposure, black; mito-dL5** zebrafish without MG2I exposed to 80 J/cm² far-red light, blue; mito-dL5** zebrafish incubated with MG2I without light exposure, green) showed a robust VMR response, characterized by a significant increase in motility at an abrupt light-dark transition. In contrast, the response was severely attenuated in dL5** zebrafish incubated with MG2I and exposed to 80 J/cm² far-red light (figure 5D; red). The requirement of all three components of the optogenetic system to provoke this striking neurological phenotype confirms that the response is dependent on ¹O₂ production. Remarkably, the loss of VMR phenotype was quantitatively similar, regardless of the rate at which energy was delivered between 40 and 160 mW/cm² (figure 5E). In this setting, the ability to deliver light rapidly at high power may be advantageous, potentially allowing induction of oxidative damage with temporal precision while avoiding asynchronous engagement of downstream mechanisms. Delivering the same amount of light energy to each experimental group, repeatedly, and at constant power, is critical for this work. This is greatly facilitated by the accurate control of high-power LED sources afforded by the controller unit.

Together, these three examples illustrate the utility of apparatus to deliver whole-field illumination to zebrafish larvae for multiple optogenetics applications.

Discussion

Open-source optogenetics light stand and controller

We report a LED illuminator and controller that allows the use of LED chip arrays to drive optogenetic events throughout whole zebrafish larvae. The approach is cost-effective and the equipment is readily constructed from widely-available materials. The apparatus delivers a sufficiently flat illumination field for whole larval illumination without bath heating and is optimized to control exposure power and energy. We developed the system for zebrafish optogenetic models, as this is the primary focus of our work, and we anticipate it will be useful without modification for multiple applications in zebrafish optogenetics and chemoptogenetics. Minor changes to the apparatus, such as editing the controller firmware to add functions or adjust acceptable ranges, should be straightforward. Potentially, the drawings, circuit diagrams, and source code provided will allow more significant modifications, for example altering the light stand to accommodate other sample formats or experimental models.

Exposure energy for individual zebrafish

Compared with the output from the LED array, the amount of light reflected from the black shade cover is negligible, allowing us to estimate the rate at which light energy reached each unit area of the sample plane by measuring the irradiance of light passing orthogonally through the center of the stage from below (mW/cm²; ‘power’). Radiant exposure (J/cm²; ‘energy’) was then calculated by integrating irradiance over time to determine the amount of light energy that passed through each unit area of the stage during exposure. These parameters allow estimates of the light energy to which individual zebrafish were exposed. An anesthetized 3 – 5 dpf zebrafish in contact with the cover slip presents a silhouette of approximate maximum dimensions 3.5 – 4.0 mm × 0.5 mm, equivalent to an area < 0.02 cm². Thus, the rate of light energy exposure for an individual zebrafish of this size is approximately 2% of the measured irradiance in mW/cm² at the stage. The zebrafish in the Kaede photoconversion experiment shown in figure 3C were consequently exposed to a total of approximately 1 J/zebrafish (λ=398 nm) @ 3 mW/zebrafish. In contrast, the GAVPO photoconversion group with brightest transgene expression in figure 4C was exposed to approximately 2 J/zebrafish (λ=450 nm) @ 34 μW/zebrafish (although exposure in this case is likely somewhat lower than this value, as larvae were not anesthetized and were consequently swimming further from the LED than the plane of the stage and not directly over its center). In figure 5, mito-dL5** zebrafish were exposed to approximately 1.6 J/zebrafish (λ=661 nm) @ 0.8 – 3.2 mW/zebrafish.

Energy transfer to zebrafish

Of the incident light reaching each zebrafish, the majority passes through without attenuation, resulting in little energy dissipation and accounting for the transparent appearance of zebrafish larvae. The amount of energy that is transferred to the zebrafish likely varies by anatomical location (the head and viscera are thicker than the fins and correspondingly attenuate the incident light beam more) and is influenced by light wavelength (longer wavelengths are more tissue penetrant), transgene expression (optogenetic proteins absorb a small fraction of incident light at specific wavelengths), and pigmentation (melanin absorbs a high proportion of light over a broad waveband). We predict that energy transfer to the zebrafish from experimentally-relevant light exposures is modest, and results in minimal heating that is likely mitigated by the water bath. Full characterization of tolerated total exposure energy, across power and wavelengths, and through a range of developmental points, is outside the scope of this report, as are phenotypes resulting from excessive light energy exposure or power. However, the new apparatus reported here will facilitate studying these outcomes systematically in future work. For individual downstream applications, it is strongly recommended to define the lowest energy and power that yield the desired optogenetic outcome, and to include light-only controls to reveal confounding factors that might occur independent of acute heating and cause delayed phenotypes, such as DNA damage from UV radiation. For quantitative applications, it is also recommended to breed transgenic lines onto a Casper background (White et al., 2008), as the lack of pigment both mitigates heat transfer to the zebrafish and ensures uniform exposure of optogenetic gene products to light throughout the zebrafish.

Controlling the rate of light exposure using pulse width modulation

Pulse width modulation does not alter the amount of light generated by the light source when illuminated, but instead turns the LED on and off rapidly to modulate mean incident light flux during an integration period that exceeds the PWM cycle length of 1.02 ms. Using PWM to control sample plane irradiance is acceptable for modulating optogenetically-triggered biological events, provided that some aspect of either the optogenetic target or the downstream biology has a sufficiently damped response to act as an integrator of the pulsed light signal. For example, the neuronal consequences of ¹O₂ production in mitochondria likely arise from cumulative oxidative damage to the electron transport chain that results in ATP depletion and neuronal depolarization (Qian et al., 2019; Xie et al., 2020). Compared with the delivery of ¹O₂ pulses at 980Hz, the rate of respiratory chain protein repair or turnover is likely very slow and can be ignored. Consequently, ¹O₂ pulses lasting half of each PWM cycle cause accumulation of oxidative damage to respiratory chain components at approximately half the rate as production of ¹O₂ driven by the same light beam illuminated continuously. Similarly, the irreversible photoconversion of Kaede by UV light results in accumulation of red-converted fluorophore at a rate dictated by the PWM duty cycle. For GAVPO-directed gene transcription, the PWM-modulated incident light signal is likely integrated by the slow kinetics of VVD switch-off (Kawano et al., 2015), allowing establishment of an equilibrium between dissociated and dimerized forms that can be driven towards dimerization by longer PWM duty cycles.

Potentially, the digital approach of controlling power by PWM enhances linearity. For example, it is not necessarily the case that ¹O₂ production would be increased by the same factor as the irradiance from a continuous light source, for example if saturation of response occurred at higher light intensity, whereas this is not true of a PWM modulated source where ‘on’ irradiance is constant. We are not aware of situations in which the delivery of pulsatile light at 980Hz would be problematic. However, two options could be deployed if there was concern that rapid, cyclic activation of optogenetic proteins might alter the biological response. Setting the PWM duty cycle to 255 using controller function F1 configures the LED array to provide an uninterrupted light beam. This could be attenuated to the required power by adding neutral density filters to the light path, or by increasing the distance from the LED to the stage (exploiting the inverse square relationship between distance and power).

Biological nonlinearity

If the same amount of light energy provoked similar optogenetic effects in different individual zebrafish expressing the same transgene construct, there should be a reciprocal relationship between PWM-controlled light power and exposure time: since energy (J/cm²) = power (W/cm²) × time (s), increasing the power delivers the same light energy with a shorter exposure. In the singlet oxygen generation experiments shown in figure 5, the same total energy exposure of 80 J/cm² provoked similar neurological phenotypes in mito-dL5**/MG2I zebrafish at multiple time/power combinations (33m20s @ 40 mW/cm² to 8m20s @ 160mW/cm²), providing empirical support for the validity of the reciprocity assumption, at least within the range tested for this application. However, the reciprocal power-time relationship did not hold in GAVPO zebrafish, where the prolonged and delayed nature of the biological response introduced a significant nonlinearity. Since the endpoint in this assay is dependent on gene expression that itself occurs with kinetics lasting multiples of minutes, prolonged light exposures resulted in more GFP expression. In addition, the long half-life of GFP (26 hours in cultured cells (Corish and Tyler-Smith, 1999)) allows its accumulation under conditions of continuous expression. Consequently, delivery of light at low irradiance for a prolonged period of time strongly favored enhanced cellular GFP fluorescence and resulted in a significant departure from the reciprocal time/power relationship for driving the targeted biological effect. It is important to recognize nonlinearities such as this to avoid pitfalls in replication.

Calibrating and reporting power and energy

We showed three example situations in which light energy, power, and exposure time strongly influenced the results of optogenetic manipulations. Total light energy delivered to the sample dictated the observed outcomes for Kaede photoconversion (also shown for mito-dL5**/MG2I photoactivation in our previous work (Xie et al., 2020)). In addition, the rate of energy delivery, through its effect on the time course of gene expression, strongly influenced the results for GAVPO transcriptional induction by photodimerization. Both energy and power further influenced bath warming. These considerations highlight the importance of specifying both light power and energy for optogenetic studies, since experiments employing the same biological preparation and light source could potentially yield significantly different results if different exposure parameters are used. Given current discussion about the reproducibility of scientific studies, this seems an important consideration. The workflow inherent in operation of the controller unit encourages measuring, controlling, recording, and reporting light wavelength, power, time, and exposure energy, thereby facilitating precise reproduction of these parameters in subsequent experiments. This should facilitate standardization of experimental protocols and enhance the expectation that replicate experiments conducted by different research groups should produce similar outcomes.

Highlights.

• LED chip arrays activated optogenetic reagents throughout larval zebrafish

• A novel light stand provided flat illumination without bath warming

• Power calibration and energy delivery were automated by a controller

• Three example applications illustrate the utility of the approach

• Drawings and instructions are provided to build the stand and controller