Light sheet fluorescence microscopy of cleared human eyes … – Nature.com

Eye samples

Eye samples from donors were obtained through the Lions Gift of Sight (Saint Paul Minnesota, USA), operating under the regulations of the U.S. Food and Drug Administration (FDA) and the Eye Bank Association of America. Consent from the donor was obtained and the next of kin received no charge or monetary gain from the donation. Harvesting of the samples was done <24h post-mortem. The lens was extracted through a corneal incision and the samples were fixed in 4% paraformaldehyde (15710, ThermoFisher scientific) in PBS overnight at room temperature. The lens was removed because we often observed reopacification of cleared lenses over time. Importation in France was done under relevant regulations for transfer of human tissues (CODECOH DC-2015-2400). All ethical regulations relevant to human research participants were followed.

We developed a clearing protocol termed ClearEye (graphically depicted in Fig.1a) for intact human eyes (major ocular structures schematized in Fig.1b). The timing of the successive steps of the ClearEye procedure are schematized in Fig.1a. Whenever a shaker at 37 is mentioned, the apparatus used was an Incu-shaker mini (Benchmark scientific); at room temperature, the apparatus was a Ms Major Science Rocking Shaker, on a low rocking setting (1015 RPM). Samples were dehydrated in successive baths of phosphate buffered saline (PBS)/methanol (322415, Sigma Aldrich) (50%, 80%, and 100% methanol, 2h each). Bleaching was done with 11% hydrogen peroxide (23613.297, vWR) in methanol, under white light (A025416, Manutan, 11W) for 10 days with agitation.

To help with pigment elimination, the bleaching solution was renewed every day and a gentle flow toward the interior of the eye was generated using a pipette during each bleaching solution renewal. After bleaching, rehydration in PBS was achieved using successive baths (100% Methanol, 80%, 50%, 0% Methanol/PBS, 2h each). Four to six iterative 90min PBS washes were done over 1 day.

To improve penetration of antibodies, we used an immunolabeling protocol derived from the System Wide control of Interaction Time and Kinetics of CHemicals (SWITCH) protocol25, which achieves deeper penetration of antibodies by modulation of antibody-antigen binding using sodium dodecyl sulfate (SDS). Samples were first permeabilized and equilibrated for 4 days in PBSGT-Sw (PBS 1X containing 0.2% gelatin: 24350262, Prolabo, 5% Triton X-100: X100500ml, Merck Millipore Sigma-Aldrich, and 10mM SDS: L3771, Sigma Aldrich) under gentle agitation at 37C. We used the following primary antibodies: goat anti-Collagen IV (diluted 1/1000, 134001 Biorad), mouse anti-alpha-smooth muscle cell actin (SMA; diluted 1/2000, A2547 Merck Millipore Sigma Aldrich), and rabbit anti-Tubulin III (diluted 1/2000, T2200-200uL Merck). Antibodies were first incubated separately from samples in PBSGT-Sw for 2h. Samples were then incubated in this primary antibody solution for 20 days at 37C, with agitation of 70 RPM. After 20 days, samples were rinsed and left for 4 days in PBSGT without SDS, to allow primary antibodies to bind to their antigenic targets. After filtering the secondary antibodies mix using 0.22m Syringe Filter Rotilabo (ROTH- KH54.1), secondary antibodies (cross-absorbed donkey secondary antibodies conjugated to AlexaFluors from ThermoFisher, diluted 1/500) were then incubated in darkness for 48h in PBSGT at 37C with 70 RPM agitation, and washed three times for 2h in PBS. To mitigate tissue softening and damage during clearing, samples were fixed again in 4% paraformaldehyde for 45min.

For clearing, samples were dehydrated in successive PBS/methanol baths (20% methanol overnight, then 40%, 60%, 80%, 2100%, each 2h), then put overnight in a 2/3 dichloromethane (DCM) (270997-1L, Merck) 1/3 methanol solution. Samples were then immersed in 100% DCM for 45min, and then in BenZyl Ether (DBE) (108014, Merck). DBE baths (2h) were changed until no swirls were seen in the solution following mild agitation of the tube. Cleared samples were then stored in DBE in darkness, where transparency and immunofluorescence were stable for at least a year.

For imaging, cleared human eyes (macroscopic views shown in Fig.1c) were placed in an immersion quartz cuvette (3cm3cm4cm) filled with the imaging medium DBE. Imaging was performed using a custom made mesoscale selective plane-illumination microscope mesoSPIM21 system at the Wyss Center in Geneva, which allowed imaging of a human eye in its entirety (travel range 4444100mm) and provided near-isotropic resolution 3D datasets. Briefly, the sample is illuminated by one of the two digitally scanned light sheets coming from opposite directions. The excitation paths also contain galvo scanners for light-sheet generation and reduction of shadow artifacts due to absorption of the light-sheet. In addition, the beam waist is scanned using electrically tunable lenses synchronized with the rolling shutter of the sCMOS camera. This axially scanned light-sheet mode (ASLM) leads to a uniform axial resolution across the field-of-view (FOV). Emitted fluorescence is collected by Olympus MVX-10 zoom microscope with a 1 objective (Olympus MVPLAPO 1) and imaged on a digital camera (Hamamatsu ORCA-Flash 4.0) at 5 Frame Per Second (FPS). Excitation wavelength of the autofluorescence (Supplementary Fig.2b), 561 and 647nm labels were set at 488, 561 and 647nm, respectively with an emission 530/40nm bandpass filter, 593/40 bandpass and 663 LP filter respectively (BrightLine HC, AHF). All acquisitions were made from only one orientation of the human eye in a multi tile setup (22) using one of the two digitally scanned light sheets. A separate dual side illumination was used to increase image quality and the penetration of the light through the whole sample. To mitigate the constraints of imaging a large size sample, sequences of z-sub-stacks were therefore taken at different focal ranges. Z-stacks were acquired with a zoom set at 0.8X at 5m spacing, resulting in an in-plane pixel size of 8.28.2m (20482048 pixels). This resulted in an under-sampled 3D image according to the Nyquist sampling rule. Indeed, the effective detection NA of the MVX-10/MVPLAPO varies with zoom and can be estimated as NA=0.065 at 0.8X. Based on this value, the diffraction-limited lateral resolution would be 3.8m.

After data acquisition z-sub-stacks were concatenated using Image J and the tiles were stitched in 3D using the Grid Collection Stitching plugin tool in TeraStitcher (BMC Bioinformatics, Italy). The resulting HdF5 files were then converted to .ims format and analyzed using the ImarisBitplane software or the virtual reality software Syglass. Imaris software (versions 9.59.7, Bitplane) was used to render and study samples in 3D. To isolate target ocular structures and vascular beds, built-in tools of slicers or manual segmentation and 3D rendering are used (see Supplementary Fig.1b). Built-in Imaris animation and snapshot tools were used for video and image acquisitions. FIJIs plugin ScientiFig was used to create scientific figures and iMovie (v10.1.11) was used for video editing.

Supplementary Table details the labeling and imaging characteristics for all figures.

Further information on research design is available in theNature Portfolio Reporting Summary linked to this article.

See more here:
Light sheet fluorescence microscopy of cleared human eyes ... - Nature.com