Coupling optogenetic stimulation with NanoLuc-based luminescence (BRET) Ca++ sensing

Optogenetic techniques allow intracellular manipulation of Ca++ by illumination of light-absorbing probe molecules such as channelrhodopsins and melanopsins. The consequences of optogenetic stimulation would optimally be recorded by non-invasive optical methods. However, most current optical methods for monitoring Ca++ levels are based on fluorescence excitation that can cause unwanted stimulation of the optogenetic probe and other undesirable effects such as tissue autofluorescence. Luminescence is an alternate optical technology that avoids the problems associated with fluorescence. Using a new bright luciferase, we here develop a genetically encoded Ca++ sensor that is ratiometric by virtue of bioluminescence resonance energy transfer (BRET). This sensor has a large dynamic range and partners optimally with optogenetic probes. Ca++ fluxes that are elicited by brief pulses of light to cultured cells expressing melanopsin and to neurons-expressing channelrhodopsin are quantified and imaged with the BRET Ca++ sensor in darkness, thereby avoiding undesirable consequences of fluorescence irradiation.


Supplementary Tables
Supplementary Table 1. Dynamic ranges of the various BRET constructs. The dynamic range was calculated by the formula (R i -R 0 )/R 0 where R i = BRET ratio at high Ca ++ (50 µM Ca ++ ) and R 0 =BRET ratio at low Ca ++ (10 mM EGTA without any added calcium).
*The values for the original Twitch-2 and Twitch-3 FRET sensors were based on published values from purified protein samples 1 Fig. 1a), and was renamed CalfluxVTN for "CALcium FLUX composed of Venus (V), Troponin (T), and NanoLuc (N)." Using the EcoRI (upstream) and HindIII (downstream) restriction enzyme sites, CalfluxVTN was inserted into pRSETB, then was transfected into BL21 E. coli for expression and purification via an N-terminally linked, six-histidine-residue tag (His6). The T7 tag and the enterokinase sites were excised from pRSETB so that the CalfluxVTN sequence is immediately adjacent to the His6 sequence. The His6-tagged fusion protein was purified via TALON® Metal Affinity Co ++ Resin and the BRET signal, in response to varying [Ca ++ ], was measured in Ca ++ buffers from Molecular Probes ® (Life Technologies™) using a QuantaMaster™ (Photon Technology International Inc.) fluorescence spectrophotometer with the excitation off. BRET ratiometric values were calculated by dividing the light emitted at 525 nm by that emitted at 450 nm after a full scan of the spectrum from 400 to 600 nm. Similarly, the BRET signal of purified CalfluxVTN was also measured microscopically using a hemacytometer to ensure a uniform volume of the protein solution. The BRET ratio was measured using emission filter cube sets of 480/40 nm bandpass (NanoLuc luminescence peak) and 520 LP (Venus fluorescence peak) in the microscopic apparatus described below under "Microscopic Imaging." For all experiments, the NanoLuc substrate, furimazine, was added to a final concentration of 10 µM.

Cellular Expression of Ca ++ -Sensors and Recording
The plasmid VSFP Butterfly

Hippocampal viral injection and brain slice preparation
On the day of surgery, male C57BL/6J mice were anesthetized with isoflurane (3% initial, 1.5% for maintenance) and placed in a stereotaxic apparatus (my NeuroLab, Leica AngleTwo stereotaxic system; Leica Biosystems, BuffaloGrove, IL). Angle Two software was used for setting injection targets in the dorsal hippocampus (coordinates were 2.18 mm posterior to bregma, 2.78 mm lateral to the midline, and 1.73 mm below the skull surface). Other details about the surgical procedure have been described previously 12 . A 33-gauge needle of a 10 µl syringe (Hamilton Company, Reno, NV) was heat sterilized immediately before back filling with AAV-CalfluxVTN virus. 500 nL of the virus was injected bilaterally into the dorsal hippocampus at 50 nL/min using a UltraMicroPump II and Micro4-controller (World Precision Instruments, Sarasota, FL). Five minutes later, the syringe was withdrawn and the scalp wound was sutured. Postsurgical care included immediate subcutaneous saline (1.0 ml per 20 g of body weight) and analgesic (ketoprofen, 5 mg/kg, subcutaneously) followed by additional ketoprofen injections every 24 hours for 3 days. Animals were monitored for health concerns including loss of body weight >20%, signs of uncontrolled pain, stress or dehydration. No animals displayed these signs and therefore none were removed from further studies. All animal studies were performed under guidelines approved by Vanderbilt University's Institutional Animal Care and Use Committee (IACUC).

Flow-through microscopic imaging experiments of acute brain slices
The acute brain slices were prepared as described above and then kept in ACSF (119 mM NaCl, 26.2 mM NaHCO 3 , 2.5 mM KCl, 1 mM NaH 2 PO 4 , 1.3 mM MgCl 2 , 10 mM D-glucose, 2.5 mM CaCl 2 ; sterile filtered) at 37 ºC, while bubbled continuously with 95% O 2 -5% CO 2 . In an open flow chamber (Warner Instruments Inc.: RC-26G), a gravity flow system was set up where the inward flow (2 ml/min) was controlled by gravity but the outward flow was removed by a peristaltic pump (Pharmacia © : LKB-Pump P-1). The ACSF was bubbled with 95% O 2 -5% CO 2 before it flowed into the chamber. Because imaging occurred on an inverted microscope, a small nylon mesh was placed between the surface of the acute brain tissue and the glass coverslip that formed the bottom of the chamber, to allow for good oxygenation of both the top and bottom of the slice. A metal Harp (Warner Instruments Inc.: SHD-26GH/10) kept the slice in place during imaging.

Microscopic Imaging and Data Analysis
Primary neurons, acute tissue slices, and immortalized cells were imaged on an inverted Olympus IX-71 epifluorescence microscope inside a temperature-controlled, light-tight box. A liquid-cooled EB-CCD (Hamamatsu Photonics K.K., C7190-13W) was used to image the cells at a frame rate of 1-4 Hz. To capture and ratio the BRET images, filters for NanoLuc (EM 480/40 nm bandpass) and Venus (EM 520 longpass) were rotated with a motorized filter turret wheel within the microscope to alternately image the blue vs. yellow/green wavelengths. The speed of rotation of these filters was ~100 ms. For faster imaging (e.g., Fig. 9d), an EM-CCD camera (Hamamatsu ImagEM X2) was used with comparable results. This EM-CCD camera was coupled to a light splitter (Hamamatsu W-View Gemini) that used a dichroic filter (495 nm LP) to separate the blue and yellow light and projected them onto distinct regions of the camera's CCD chip. This allowed for simultaneous collection of blue and yellow wavelengths for optimal temporal resolution. During luminescence microscopy, all images were collected in complete darkness. To stimulate the Opn4 or the CheRiff optogenetic probes, ~470 nm (470/30 nm) light was used for the duration stated in the relevant figure legend. Neurons that were cultured on glass coverslips (ThermoFisher TM ) were inserted into a Chamlide magnetic chamber (Live Cell Instrument, Korea) for imaging.
The images were analyzed with ImageJ software (NIH) using background-subtracted images collected from the blue and yellow channels. After background subtraction, a simple division of the yellow wavelength intensities by the blue wavelengths produced the BRET Ratio values as depicted in all figures. The average light intensity of blue vs. yellow from regions of interest within the cells was compared pixel by pixel to obtain ratiometric BRET estimates of cytosolic Ca ++ over the image. To create the ratiometric photos and movies (e.g., Fig. 5b), both blue and yellow channels were background subtracted using the ImageJ function. Then the blue wavelength images were used as a template to create an image that separated the cells completely from the background (cell region was given a value of 255, and background was given 0). The blue and yellow images were then multiplied by this template image where the background was held at 0 and areas within cells were given the full bit value. The resulting images were converted from 16-bit (from camera) to a 32-bit to mitigate losing data after the multiplication step. To finally achieve the ratiometric image, the yellow emission needed to be divided by the blue emission, but to prevent dividing by 0, a gray value of 1 (negligible, since it is out of 4.3 x 10 9 ) was added to all the pixels in the blue images, so that dividing the yellow values by the blue values would not result in division by 0 at any pixel. When the final yellow images were divided by the final blue images, the ratio changes that are indicative of [Ca ++ ] were observed (as indicated by the calibration curve in Fig. 1b-c or in Supplementary Fig. 8). The ImageJ lookup table (LUT) named "fire" was applied to the images to better highlight the areas were [Ca ++ ] was changing.
Statistical significance was tested using the Data Analysis Tool Pack from Microsoft Excel (Microsoft © ). The Hill coefficient and K d values (Fig. 1c) were determined using OriginLab 6 software (OriginLab © ). The emission light spectrophotometric data measured with the QuantaMaster (Figs. 1 & 2) were collected across the 400-600 nm range in 1-nm intervals. The microscopic data were collected as the average pixel intensity over a user-defined region of interest (ROI). Where t-tests were done, the data fit a normal distribution (Fig. 2d, 3d, 4e-f) and for non-parametric data sets (Fig. 5d, 7d, 9c) ANOVAs were used to determine the significant difference between experimental groups. Except for the negative control neurons in Fig. 7d, statistical analyses were performed in cellular experiments that contained a minimum of 10 cells (in a single experiment) or a minimum of 3 independent experiments, with multiple cells from each group. Unhealthy cells were identified as those cells whose BRET ratio exceeded 3 before any experimental manipulation was done, and were excluded from further analyses.