Visual stimulation for calcium imaging

Custom-written software using C++ and OpenGL was used to generate visual stimuli. A LightCrafter (Texas Instruments, Dallas, TX, USA) running at a 100 Hz frame rate was used for the stimulus projection. Stimulus light travelled through a 482/18 band pass filter and a ND1.0 neutral density filter onto an 8 cm x 8 cm rear projection screen. The projection screen was positioned in front of the animal and reached 60° of the fly visual field in azimuth and 60° in elevation (Figure 11). The visual stimuli used are described below. All contrast values were calculated using the definition of Weber contrast.

When testing the L2 and L3 contrast and luminance sensitivity across different adaptation states, we first empirically tested how bright the stimulus could be without getting any bleed through while imaging. This intensity was 15,75 cd/m² measured from the projection screen on the same side where flies look at the screen. In order to alter the overall mean intensity of the stimulus we used neutral density (ND, Thorlabs) filters ND 0.6 (25% transmission) and ND 1.3 (5% transmission). Therefore, intensity of the stimulus was 15,75 cd/m2 without a filter, 3.84 cd/m² with the ND 0.6 and 0.91 cd/m² with the ND 1.3. The intensity of the stimulus used for all other experiments was 2.25 cd/m².

Full-field flashes: Periodic, alternating, 100% contrast light and dark flashes covering the whole screen were presented to the flies. Flashes were each lasting 2 s, 5 s or 60 s.

Each 2 s and 5s stimulus period was presented for ~10 trials, each 60 s period for ~2 trials.

Full-field flashes from an intermediate gray: Alternating light and dark flashes of full-contrast and lasting for 2 s were interleaved by an intermediate gray epoch lasting for 4 s or light and dark flashes of 5 s were interleaved by an intermediate gray epoch lasting for 10 s. Presentation of the ON or OFF flash from an intermediate gray was random.

Each stimulus presentation period was presented for ~5 trials.

Different contrast steps from intermediate gray: The stimulus consisted of a 10 s gray period, followed by 5 s ON or OFF flashes of 20%, 40%, 60%, 80% or 100% contrast increments or decrements, relative to the intermediate gray background.

Flashes of different intensities: The stimulus consisted of 10 s full-field flashes of 5 different intensities (ranging from 0 to 100% of maximal intensity in 25% steps). The

MATERIAL AND METHODS order between the periods was randomized, resulting in 20 different step combinations with varying Weber contrast.

Two OFF steps from adapted background stimulus: To distinguish contrast and luminance sensitivity, we designed a stimulus after (Oesch and Diamond, 2011). The stimulus consisted of a 30 s adapting period of 100% of max intensity, followed by two consecutive 3 s or 10 s OFF steps: the A and the B steps. The A step was one of 6 linear decreasing luminance values (Figure 13A), resulting in 6 different contrast steps (Figure 12B) relative to the adapting step. The luminance of the next OFF step, the B step, was one of 6 linear decreasing luminance values, depending on the previous A step, all of which resulted in 6 equally sized 25% contrast steps (Weber contrast) (Figure 13B). The order of the different A steps and their associated B steps was randomized.

Figure 13. Schematic of the contrast steps from adapted background stimulus

(A-B) Schematic of the stimulus, consisting of an adapting bright period, followed by two sequential OFF steps is shown. (A) Luminance changes during the course of the stimulus are shown. (B) Contrast changes during the course of the stimulus are shown. Stimulus designed by Marvin Seifert.

Visual stimulation at different wavelengths: To test the spectral efficiency of L2 and L3, we used a multispectral light synthesiser based on LEDs and a diffraction grating (LED synthesizer (Belušič et al., 2016). The stimulus was passed through a 1 mm diameter

MATERIAL AND METHODS light fiber and a collimating lens and was projected straight into the fly’s compound eye.

There, the stimulus subtended ~25°. We presented the flies with light coming from a series of LEDs with peak wavelengths spanning from UV to red light. The peak wavelengths of LEDs used were 367, 373, 385, 405, 420, 439, 451, 469, 485, 500, 518, 530, 554, 573, 590, 626, 651, 660, 672, and 682 nm (Figure 14A). The launching fiber has a limited aperture, thus the LED bandwidth was ~10 nm. LED intensity was calibrated with a spectroradiometer (Flame, calibrated with a radiometric light source DH-3P-CAL; both Ocean Optics, USA). LED intensity was adjusted using 12-bit pulse width modulation at 1 kHz to create an isoquantal spectral sequence (i.e. all LEDs had equal photon fluxes). Visual stimulation and PMT shutters were synchronized so that the PMT shutters were closed during the presentation of the light flash. We experimentally determined the shutters closing time to be 18 ms and shutters opening time to be 6 ms, by measuring stimulus bleedthrough. Therefore, we programmed the imaging protocol using The Mark Points dialog in the Prairie View program. The imaging protocol was set up in in a way that the imaging computer sent a signal to close the shutters 18 ms before presentation of the flash and it sent the signal to open the shutters 6 ms before the stimulus end (Figure 14B). Doing so, the shutters were closed during the entire presentation of the stimulus flash and opened right after the end of the flash. For example, when we presented the flies with 2 s long flashes, we set up the duration of the shutters being closed to 1994 ms. 1994 ms plus the opening time of the shutters resulted in exactly 2 s of shutters being closed. Thus, the full offset response was recorded and analyzed. Peak responses to each LED presentation were used to obtain spectral efficiency response curves.

MATERIAL AND METHODS

Figure 14. Stimulus setup of the LED synthesizer.

(A) List of LEDs with their peak wavelengths. (B) Schematic of the imaging protocol. Closing time of the shutters (dtc) is 18 ms, whereas opening time of the shutters (dto) is 6 ms. In blue, example trace of one L2 neuron is shown. Note that the shutters are closed during the presentation of the flash and they open immediately after the end of the flash due to the fact that we took the shutter opening and closing times into an account.