The Dual Functional Reflecting Iris of the Zebrafish

Abstract Many marine organisms have evolved a reflective iris to prevent unfocused light from reaching the retina. The fish iris has a dual function, both to camouflage the eye and serving as a light barrier. Yet, the physical mechanism that enables this dual functionality and the benefits of using a reflective iris have remained unclear. Using synchrotron microfocused diffraction, cryo‐scanning electron microscopy imaging, and optical analyses on zebrafish at different stages of development, it is shown that the complex optical response of the iris is facilitated by the development of high‐order organization of multilayered guanine‐based crystal reflectors and pigments. It is further demonstrated how the efficient light reflector is established during development to allow the optical functionality of the eye, already at early developmental stages.

performed on a total of 5 whole, hydrated zebrafish eyes. The tomographic volumes were obtained by taking 1100 projections over 180 o at 30 KV and 150 μA, and the isotropic voxel size in the reconstructed volume was 2.2 μm.
Reflectivity measurements: The reflectivity of the zebrafish iris was measured on either fixed or recently removed fresh eyes, which were immersed in PBS and placed underneath a cover slip held in place with silicon grease. The eyes were positioned such that the iris was facing directly upwards. Impinging light was approximately normal to the objective lens. Reflectivity was measured as described in detail in [10] with the exception of using a Shamrock 303i spectrometer (Andor) equipped with an iDus 420 CCD camera and cooled to -70°C .
Briefly, we used a custom-built microscope consisting of a microspectrophotometer, two CCD cameras, and a high numerical aperture objective, which enabled imaging the iris while obtaining both the reflectance spectrum and the Fourier transform of the reflectance for the same location in the sample. The light source was a halogen lamp coupled to an optical fiber, which guided the light into the microscope.
Imaging was first used to determine the correct focal point for the light source. Then the reflectance spectrum was used to determine the reflectance intensity, which was normalized to the reflectance of a silver mirror. Light was then imaged through a beam splitter onto the back aperture of an objective (Olympus, UPLSAPO 60XW, NA 1.2). The objective was used both to illuminate a wide area (∼250 μm in width) and to collect the scattered light. The collected light was directed to one of three different paths by a set of folding mirrors. In the first path, the sample was imaged onto a CCD camera (Mintron, MTV 13 V5Hc). In the second path, the Fourier transform of the scattered light was captured by imaging the back aperture of the objective onto a similar CCD camera. In the third path, the light was collected and coupled into a fiber, which guided the light into a spectrometer. The sample was placed on top of translational stage and goniometer, such that both its position and orientation could be controlled.

Reflectivity simulations:
The reflectivity spectrum was simulated based on crystal thicknesses and spacing obtained from cryo-SEM images using a Monte Carlo transfer matrix calculation, as described in detail in the supporting information of [S1] . In brief, the percentage of reflectivity was calculated by averaging 500 runs, assuming normal incident light. Each layer was characterized by two variables: nj, a refractive index, and d j, which is the layer thickness randomly picked from the experimental distribution. Thus, for each layer we defined the following 2 × 2 matrix: The set of k double layers was characterized by an overall reflectivity 2 × 2 matrix: The reflectivity was extracted from the following equation: The refractive index of the guanine crystal plates was taken as 1.83, which was the refractive index in the direction of the impinging light. The weak dependence of the refractive index on wavelength was neglected, assuming that all the interfaces, i.e., inside a crystal stack and between stacks, were parallel. We also assumed no correlation between the crystal spacings within a single crystal stack.
Transmission electron microscopy (TEM) imaging: To extract the guanine crystals, eyes were embedded in 7% agar and cut into 100 μm horizontal sections using a vibratome. The sections were then homogenized and the crystals were concentrated using centrifugeation. A suspension of the crystals in DDW was then removed and a drop was applied to a glowdischarged carbon-coated, copper TEM grid. The suspension was allowed to settle for 30 seconds and were then blotted. The TEM grids were observed using an FEI Tecnai T12 TEM operated at 120 kV. Images and diffraction patterns were recorded on a Gatan OneView camera using imaging and diffraction modes respectively. The observed electron diffraction patterns of the crystals (in set in Fig. S5) correspond to anhydrous β-guanine [S2] .
Synchrotron XRD measurements: The scanning XRD measurements were carried out at beamline ID13 of the European Synchrotron Radiation Facility (ESRF) in Grenoble, France.
The beam was monochromatized using a Si(111) channel-cut monochromator to an energy of 13.9 keV. Compound refractive lenses were used as a focusing system that yielded a focal spot size of 2 × 2 µm^2. The sample was placed in the focus of the beam and the detector was positioned 149.25 mm behind the focal plane. The beam center, detector tilt angles and the sample-to-detector distance were determined using Al2O3 as calibrant. Diffraction patterns were collected at high frame rates using the single-photon counting Eiger 4 M detector (Dectris, Switzerland). We were able to acquire diffraction images with an exposure time of 50 ms. The samples were scanned in a continuous movement through the x-ray beam. The 'step size' corresponding to the interval traveled during a single exposure was 2 µm along the horizontal (y) and vertical (z) direction.
Analysis of XRD data: Data correction steps were necessary, before scattering data could be analyzed. First, invalid detector pixels were masked such that their respective value were not taken into account in the analysis. Secondly, the absorption of a semi-transparent capillary used as a beamstop holder was corrected for by pixel wise multiplication with a correction matrix. Thirdly, background was subtracted from each scattering pattern to base the PCA analysis solely on the scattered intensity due to crystalline reflections. Lastly, to reduce data load and to improve the speed of calculation, the PCA analysis was performed only on a range of q-values, centered on the required respective (100), (012) and (002) reflection with a width of nm -1 .
Following the approach described by Bernhardt et al. [13] the scattering distribution is treated as a probability density function for the distribution of photons. To retrieve the eigenvectors of the scattering distribution, the covariance matrix C of the distribution of the wave vector components qy and qz in the detection plane is diagonalized. C is defined as The principal direction of scattering is thereby given by the largest of the two corresponding eigenvectors v1 and v2. The length of the eigenvectors (the variance) is given by the eigenvalues λ1 and λ2. One can thereby define a dimensionless parameter, the anisotropy of the scattering: .
A value of 1 would thereby correspond to a single scattering direction, while a value of 0 would correspond to a perfectly isotropic scattering.
Based on our analysis, not all of crystals are aligned co-axially with respect to each other as suggested by the anisotropy of the scattering (< 0.2 for the (012) and (002) crystal planes; < 0.5 for the (100) crystal plane, see Fig. S5). The relativley low anisotropy of the scattering is probably due to the underlying disordered iridiphore layers, which reduces the levels of anisotropy . The anisotropy maps corresponding to Figs. 4A and 4D are shown in Figs. S5 C and D, respectively. Disregarding a particular reflection, one should keep in mind, that at any scan point from within the iris, at least one of the crystal planes meet the diffraction condition. This is shown in the map shown in Fig. S5 B which integrates the scattered intensity that is above a manually chosen threshold.
Several structural parameters can be extracted from a single scattering pattern. It is therefore informative to use multiple representations of the scattering data. First, in Fig. S5 A we show the scattered intensity integrated within the 10.5 nm -1 and 11.5 nm -1 . The q-range was chosen based on the fact that within this range, no reflections occur. We hereby obtain a dark field contrast that is solely based on the scattering of the isotropic sample matrix and solution. In this contrast, one can for example observe trapped air bubbles in the sample preparation. The overall distribution of crystals in the sample can be obtained by integrating the scattered intensity above a manually chosen threshold that discriminates between Bragg reflections and background scattering. An example is shown in Fig. S5B. Furthermore, the three Bragg reflections can be clearly seen in maximum intensity projections of the entire dataset, see Fig. S5E. To generate a maximum intensity projection, a given pixel with index (i,j) is assigned the maximum value of all pixels with index (i,j) in the entire data set. Clearly, all reflections can hereby be visualized while in an average scattering pattern, a single reflection would be averaged out and barely visible. For comparison, a single diffraction pattern is shown in Fig. S5F.
In addition to the contrasts presented so far, one can furthermore radially integrate a single scattering pattern within a given q-range to yield a one-dimensional representation of the intensity as a function of azimuthal angle I(phi). We have performed a radial integration on the (100) reflection for each scan point. We then calculated the normalized crosscorrelation of the radial intensity of each scan point with its next neighbors. We found that adjacent scan points are well correlated over a distance of 4 µm (distance between two scan points), especially, where the anisotropy was high. An overview over the entire sample as well as a zoom region is shown in Figs. S5 G and H, respectively.        (100) and (002) reflections. (E) Maximum intensity projection and (F) Isolated diffraction pattern. (G) Correlation of the radial intensity of the (100) reflection with data from its next neighbors. A zoom is shown on the right.