*2.8. Electron Microscopy*

Mouse eyes were processed for plastic embedment, ultramicrotomy, and EM analysis essentially as described in detail previously [23]. Immediately after sacrifice, eyes were orientated by marking the superior hemisphere along the vertical meridian at the limbus with a hot needle, before starting the dissection. A cut was made in the superior cornea and the eyes were fixed for 2 h at 4 ◦C in fresh 0.1 M sodium phosphate bu ffer (pH 7.4), containing 2.5% (v/v) glutaraldehyde, 2.0% formaldehyde and 0.025% CaCl2. After a 20–30 min primary fixation, the superior cornea and lens were removed, and fixation was continued overnight. The fixed eyes were then rinsed with 0.1 M sodium cacodylate bu ffer (pH 7.4) containing 0.025% CaCl2, and then post-fixed for 1 h in 1% osmium tetroxide in 0.1 M sodium cacodylate bu ffer. After post-fixation, the eyes were rinsed twice in 0.1 M sodium cacodylate bu ffer and once in distilled water, then dehydrated in graded ethanol series followed by propylene oxide and infiltration overnight in Spurr's resin. The eyes were then embedded in resin-filled BEEM ® capsules (Polysciences, Warrington, PA, USA) and allowed to polymerize in a 70 ◦C oven for 48 h. Tissue sections were obtained with a Reichert–Jung Ultracut E ® microtome using a diamond knife. Thin (60–80 nm thickness) sections were collected on copper 75/300 mesh grids and stained with 2% (v/v) uranyl acetate and Reynolds' lead citrate. Sections were viewed with a JEOL 100CX electron microscope at an accelerating voltage of 60 keV.

### *2.9. Serial Block-Face Scanning Electron Microscopy (SBF-SEM)*

Samples prepared for TEM as described above were further processed by Thermo Fisher Scientific (Waltham, MA, USA) using an Apreo VolumeScope ™ serial block-face scanning electron microscope (SBF-SEM). Excess resin was removed from the tissue using a Leica ultramicrotome. The trimmed blocks were then glued to a SEM stub (Agar Scientific, AGG1092450) using a two-component silver conductive epoxy, H20E EPO-TEK (Ted Pella, Inc.; Redding, CA, USA). To minimize charging of the block by the electron beam, the bottom and sides of the block were sputter-coated with a 30 nm thick gold film layer. The samples were then imaged on the VolumeScope ™ operating in low vacuum mode at 50 Pa and using a lens mounted backscattered detector. All of the data sets were imaged with an accelerating voltage of 2.2 kV and a beam current of 100 pA using 1-μs dwell time combined with two-line integration. Two regions of interest (ROIs) were acquired on the knock-out sample (KO-148). For ROI1, 738 sections were collected with the internal microtome set to a 40 nm cutting thickness (z resolution) with an area of 92.9 μm × 89.1 μm at 10 nm/pixel. For ROI2, 745 sections were collected with an area of 97.6 μm × 96.3 μm using the same imaging condition as ROI1. For the wild type sample (WT-146), an area of 92.9 μm × 89.1 μm was imaged at 10 nm/pixel using a cutting thickness (z resolution) of 40 nm. The acquired data sets were finally aligned and visualized using 3D volume rendering to highlight the RPE anomalies using Amira software (Thermo Fisher Scientific).
