Supplementary Fig. 1: Light propagation simulation in the human retina. (a) Müller cell

Transcription

1 Supplementary Fig. 1: Light propagation simulation the human reta. (a) Müller cell refractive dex distribution (red) along the cell s length (130 µm), and the refractive profile of the surroundg area (blue). The cell s refractive dex is higher than that of the surroundg along the entire retal depth. (b) Simplified Müller cell structure and the correspondg refractive dex. The cell s diameter and its refractive dex are subject to small perturbations, which are cluded the algorithm. The put light distribution has a gaussian shape with a pla wave front. (c) A characteristic tensity distribution for 560 nm light at the bottom of the Müller cell, after full propagation side the cell and its surroundg. 1

2 Supplementary Fig. 2: Robustness analysis of the optical simulation. (a) Twenty human Müller cells realizations, possessg random bendg along the longitudal axis. (b) 5% random perturbations of the radius of one of the cells along its length (130 µm = 1000 numerical steps). (c) Perturbations of the refractive dices of surroundg neuronal layers and of Müller cells were added. (d) Spectra of light at the exit of all cells (N = 20) after illumated by an abberated pupil. (e) Mean ± s.d. (N = 20) of the spectra of light at the exit of the Müller cells. The peak transmission is the green, λ ~560 nm. 2

3 Supplementary Fig. 3: Light concentration by a Müller cell. (a) Calculatg the ratio R( ) between the output and put light tensity. The light tensity is summed before cidence on the cell (put tensity) and after light propagation (output tensity), before the photoreceptors. The ratio between the ner and outer radius for the human Müller cell is r / r out ~5-6. (b) Wavelength dependency of light concentration side Müller cell M ( ) obtaed by Eq. S9. At 560 nm there is ~10-fold enhancement of light impgg upon the cone receptive field as a result of Müller cell concentration. (c) The correspondg wavelength dependency of light leakg outside Müller cell S( ) obtaed by Eq. S12. (d) Transmission with the cell obtaed for normal cidence (blue curve) and averaged tilted field, up to 10 o (red curve). The peak transmission is located the same put wavelength. 3

4 Supplementary Fig. 4: dentification of photoreceptors outer segments (POS) layer. POS were identified without exogenous labelg usg the reflection mode of the LSM 510 meta-confocal microscope is demonstrated here for a rat retal slice, 12 µm thick. The slice was illumated with a 488 nm argon laser. (a) White channel represents the transmitted light, revealg retal morphology. (b) The slice was illumated with 488 nm argon laser and reflected light was measured, dicated by the yellow channel of the microscope (502±5 nm). (c) Overlay of both channels. (d) Profile of mean tensity of yellow channel along the retal slice, demonstratg a significantly higher reflected light tensity from POS, allowg their identification. 4

5 Supplementary Fig. 5: Spectral analysis of light transmission the isolated guea pig reta. (a) Spatial light distribution as recorded 50 µm above the layer of photoreceptors ner segment. Light is beg tunneled side distct pathways. (b) Müller cells light tunnelg areas were located by threshold determation, and the 10 cells of highest transmission were marked (red circles). The red circles mask was marked the 26 remag images. The 27-image stack corresponds to 27 distct visible wavelengths ( nm). 5

6 Supplementary Fig. 6: llustration of the normalization scheme. (a) 27 mages recorded by the microscope lambda mode, where the halogen light source was projected by the optical fiber, without a sample. (b) Sum (over all pixels) for each of the 27 images is the spectrum of the halogen and fiber. (c) mage set after the normalization scheme. (d) Three representative images of the 27 retal transmission images recorded by the microscope lambda mode at the distal end of the Müller cells the guea pig reta. The halogen light source was projected by the optical fiber on the retal surface. (e) The 27 images were normalized usg the transformation weights derived for the light spectrum recorded without tissue. (f) Mean (± sd) of the spectral tensity side Müller cells, studied the guea pig reta #1 after normalization. 6

7 Supplementary Fig. 7: Ga of light absorption the guea pig photoreceptors (a) Absorption spectra of the guea pig s S- and M-cones compared to the computational simulation of Müller cells spectral transmission onto cones. (b) Absorption spectra of the guea pig s rods compared to spectral leakage from Müller cells to the rods. (c) Photoreceptors ga factor of light absorption due to theoretical separation of wavelengths by Müller cells. (d, e, f): The same as for (a, b, c) but with the experimentally measured spectral transmission and leakage the guea pig s reta. 7

8 Supplementary Fig. 8: Spectral transmission other tissue. (a) mages of the mouse s small teste 27 wavelength bands. (b) Spectrum of light recorded after propagation through the tissue. The spectrum is obtaed with the high transmission areas of the tissue (bright patches), by the same scheme used for the reta. This spectrum is markedly different from the retal spectrum. 8

9 Supplementary Methods Simulated light concentration by Müller cells Due to the funnel-like shape of the Müller cell s endfoot the vitreo-retal junction, a Müller cell collects light from a large area and concentrates it to a smaller area the distal end of the cell, onto one coupled cone (Supplementary Fig. 3a). The ratio between the diameter of its distal part (r out ) and the diameter of the endfoot (r ) for the human Müller cell is ~1:5. n order to obta the factor by which light tensity is multiplied due to the cell s light guidg properties, we calculated the light exit from the cell (output) relative to the light tensity impgg upon the endfoot (put). Thus, on the first step of the simulation we calculated the cident tensity over the cell upper area ( r r, defed Supplementary Fig. 3a) ij ij{ rr}, which is given by a summation Light arrives at Müller cells from the eye s pupil a diffraction pattern, which is approximated by a uniform distribution space (40 m wide Gaussian, much wider than the endfoot). Accordgly we calculated the tensity out after light propagation the cell by a summation over the cell s outer area ( r r ) out out. ij ij{ rrout } We defe the ratio between output and put tensity (Supplementary Fig. 3a) for a given wavelength as R( ) ( ) out. Throughout the simulation process, the put tensity was constant for all wavelengths. Light density is determed by its tensity as well as by the area over which it is distributed. Therefore, the put light density is given by and the output light density is given by, 2 r 9

10 . out out 2 rout Thus for a given wavelength, the ga photon density as a result of Müller cell s light concentration, which we term as the concentration factor M ( ), is given by the ratio of output and put density of light (Supplementary Fig. 3b) M ( ) out out rout 10 ( ) r For the human Müller cell, r / r out 5, thus M( ) 25 R( ). Thus, for 560 nm light, M ~ 10, and there are 10 photons impgg on the cone receptive field as a result of Müller cell light concentration. Sce light absorption the neural layers of the reta is negligible there is conservation of energy and out out_surr. Here out_surr is the tensity of light leakg out of the Müller cell durg propagation, to the surroundg area, and fally beg cident on the rod photoreceptors. Now we defe, a similar manner, the light density the surroundg area. ( ) out out_surr 2 2 r rout Thus, for a given wavelength, the light concentration factor the Müller cell s surroundg area S( )(Supplementary Fig. 3c) is given by ri S( ) [1 ( )] 2 out_surr n R 2 2 r rout For 560 nm light S ~ 0.6, thus there are ~ 0.6 photons impgg the rod receptive fields as a result of Müller cell light concentration (a 40% reduction of light tensity for rods, less at shorter wave lengths). When light enters the pupil away from its center, it reaches the reta as a tilted wavefront, rather than perpendicularly. At night time, the pupil dilates up to 8 mm, and with an average eye length of 23 mm, the maximum cidence angle with respect to the reta is ~10 0. Therefore, we calculated also the average transmission of light the waveguide cells with an cident slant of up to 10 0, and found it to have the same wavelength for peak transmission (Supplementary Fig. 3c, red curve). The relative 2.

11 tensity is lower, as can be expected when the leakage creases as a result of higher cident angle. 11