Multi-immersion open-top light-sheet microscope for high-throughput imaging of cleared tissues

Recent advances in optical clearing and light-sheet microscopy have provided unprecedented access to structural and molecular information from intact tissues. However, current light-sheet microscopes have imposed constraints on the size, shape, number of specimens, and compatibility with various clearing protocols. Here we present a multi-immersion open-top light-sheet microscope that enables simple mounting of multiple specimens processed with a variety of clearing protocols, which will facilitate wide adoption by preclinical researchers and clinical laboratories. In particular, the open-top geometry provides unsurpassed versatility to interface with a wide range of accessory technologies in the future.

mostly of water and therefore have a refractive index close to that of pure water (n = 1.33). While there 120 are currently no glasses at this refractive index, fluoro-polymers (i.e., Teflons), including fluorinated 121 ethylene-propylene (FEP) and poly-tetra-fluoro-ethylene (PTFE) possess a compatible refractive index 122 (n = 1.34). These materials can be manufactured as thin sheets and stretched tight as "drumhead" 123 surfaces that are ideal holders for expanded specimens. Using a customized drumhead, we imaged a 4× 124 expanded 200 µm thick kidney section ( Fig. 3d and Supplementary Videos 7-9). After expansion, the 125 physical size of the tissue was 2.1×3.2×0.1 cm. Representative zoomed-in views provide a highly 126 detailed view of 3D structures such as glomeruli, renal tubules, and blood vessels. 127 We have developed and characterized a multi-immersion OTLS microscope that enables high-128 throughput automated imaging of optically cleared specimens with an ease of use that should facilitate 129 broader adoption of light-sheet-based 3D microscopy by both researchers and clinicians. Our system 130 imposes minimal constraints on specimen shape/size and allows for fast and convenient mounting of 131 multiple tissue specimens for automated imaging. The system provides an imaging depth of 0.5 cm over 132 a maximum lateral area of 10×10 cm at a speed of ~ 1 min/mm 3 , which can be tailored for specific 133 research applications in future designs (see Supplementary Figure 22). We have shown that our system 134 can interface with a variety of modular specimen holders tailored for specific tissue types and clearing 135 protocols, with the ability to isolate different specimens in individual wells. Due to its open-top 136 geometry, our system also provides unsurpassed versatility to interface with future tissue-based 137 protocols and a wide range of potential accessory technologies such as microfluidic devices, single-cell 138 electrophysiology, and micro-aspiration [27][28][29][30]. XY stage, motorized Z actuators, motorized filter wheel, scanning mirror, computer-controlled multi-152 wavelength fibre-coupled laser package, and sCMOS camera, all of which enable high-throughput 153 automated imaging of multiple specimens simply placed on a flat plate, or placed within a diverse 154 assortment of transparent holder designs (f). (g) Volumetric imaging is achieved by using a combination 155 of stage-scanning and lateral/vertical tiling. The scale bars in (c) denote 1 µm. 156 function of Δn×t is plotted, indicating that for diffraction-limited imaging, the condition that Δn×t < 162 0.002 should be maintained. Based on this condition, potential glass and monomer/polymer holder 163 materials are shown in (b). The color scale indicates the maximum material thickness, tmax, that is 164 allowed based upon the intrinsic mismatch, Δn, of those materials with published clearing protocols. 165

Main
Chemically incompatible combinations of materials and chemical reagents are also indicated. 166 successive frames, the de-skewing is quickly performed by simply shifting each plane of pixels in the 244 image strip by an integer pixel offset (Supplementary Figure 13). This operation is extremely fast 245 compared to alternative de-skewing approaches using computationally expensive affine transformations. 246 The data is then written from RAM to disk using the Hierarchical Data Format (HDF5) with the 247 metadata and XML file structured for subsequent analysis using BigStitcher [2]. A custom HDF5 248 compression filter (B3D) is used with default parameters to provide ~10× compression which is within 249 the noise limit of the sCMOS camera [3]. This pre-processing routine is applied to all DCIMG files, 250 ultimately resulting in a single HDF5/XML file for BigStitcher. The alignment of all image strips is 251 performed in BigStitcher, and finally fused to disk in either TIFF or HDF5 file formats. The resulting 252 TIFF and HDF5 files are then visualized using open-source and commercial packages, including 253 ImageJ, BigDataViewer, Aivia (DRVision), and Imaris (Bitplane) [4,5]. To optionally provide false-254 colored pseudo-H&E histology images, a Beer-Lambert coloring algorithm is applied using a Python 255 script [6]. The entire processing pipeline is shown in Supplementary Figure 14 and available as 256

Specimen holders 258
All holders were attached to the motorized XY stage using custom machined aluminum adapters plates 259 (HILLTOP21). For the mouse brain slices, a 1-mm thick fused silica window (Esco Optics) with a 260 10×10-cm cross-section was attached to a custom adapter plate using UV-curing glue (Supplementary 261 Figure 15). Mouse organs cleared using Ce3D were imaged on a customized 6-well plate. The bottom 262 of a conventional polystyrene 6-well plate (Cat:CLS3506, Sigma-Aldrich) was removed and replaced 263 with a 0.5 mm thick PMMA plate (Goodfellow USA) (Supplementary Figure 16). For the expanded 264 kidney specimen, a custom "drumhead" was fabricated and adapted for mounting to the microscope. The 265 drumhead tightens a 0.1 mm thick FEP film over an extruded opening, which is ideal for OTLS imaging 266 of expanded specimens (Supplementary Figure 17). To overcome the hydrophobic nature of the FEP 267 films (which cause drifting of expanded specimens), the upper surface of the FEP films were treated 268 with 0.1% (w/v) poly-lysine (Cat:P8920, Sigma-Aldrich) for charged-based adhesion of specimens to 269 the FEP surface. For the human prostate biopsies, HIVEX lens blanks (Conant Optical) were purchased 270 and custom machined using an in-house desktop mill (OtherMill, Bantam Tools) (Supplementary 271 Figure 18). The 1/8 inch, 1/16 inch, and 1/32 inch drill bits were used, and the feed rates and drill 272 speeds were optimized for the HIVEX material. CAD files for all sample holders are available as 273 Supplementary CAD Files. The system can also be used as a whole-slide scanner for conventional 274 fluorescently labeled histology slides using a commercially available slide holder (MLS203-P2, 275 Thorlabs) (Supplementary Figure 19). Dispersion curves for the various holder materials and clearing 276 reagent combinations are shown in Supplementary Figure 20.

Optical simulations 278
Optical simulations were performed using commercially available ray-tracing software (ZEMAX, LLC) 279 with a "blackbox" model of the multi-immersion objective (provided by the manufacturer, Special 280 Optics). For the simulations shown in Fig. 2, the base refractive index of the immersion medium and 281 specimen was assumed to be n = 1.45, and the optical path difference was varied. For all scenarios, the 282 imaging depth was set to 1 mm, and the PSF was measured at the center of the imaging field of view. 283 The same relationship between Strehl Ratio and optical path difference was observed for other base 284 refractive-indices and imaging depths, under the assumption that the optical properties of the immersion 285 medium and specimen were the same. The ZEMAX files for the OTLS system are available as 286

Supplementary ZEMAX Files. 287
Collection and processing of mouse brain slices 288 A mouse of line Sst-IRES-Cre;Ai139(TIT2L-GFP-ICL-TPT), characterized previously [7] was used for 289 imaging experiments. Genotyping confirmed expression of Cre and tdTomato for this individual. The 290 mouse was sacrificed at age P96 by trans-cardial perfusion with 4% paraformaldehyde. The brain was 291 dissected and post-fixed in 4% paraformaldehyde at room temperature for 3-6 hr followed by overnight 292 fixation at 4 deg. C. The brain was rinsed with 1× PBS and stored in 1× PBS with 0.1% sodium azide 293 prior (Cat:S2002, Sigma-Aldrich) prior to sectioning. 200-µm thick cortical sections were cut on a 294 vibratome and stored in 1× PBS. Prior to OTLS imaging, brain slices were incubated in a mixture of 295 68% 2,2′-thiodiethanol (TDE) (Cat:166782, Sigma-Aldrich) and 32% 1× PBS for clearing. The 296 refractive index of the solution (n ~ 1.46) was verified using a refractometer (PA202, Misco). 297 Procedures involving mice were approved by the Institutional Animal Care and Use Committee of the 298 Allen Institute for Brain Science in accordance with NIH guidelines. 299 Collection and processing of heart, lung, prostate, and lymph nodes 300 Lung, heart, prostate, and lymph nodes were collected from a CD11-YFP, Actin-dsRed expressing 301 mouse. Tissues were fixed for 24 hr at 4 deg. C in 1 part fixative (Cat:554655, BD Biosciences) and 2 parts 1× PBS and incubated in blocking buffer for 24 hr at 37 deg. C. The buffer consisted of 30 mL 303 Tris (Cat:252859, Sigma-Aldrich), 0.3 mL NMS (Cat:SML1128, Sigma-Aldrich), 0.3 mL BSA 304 (Cat:A2058, Sigma-Aldrich), and 0.09 mL TritonX100 (Cat:T8787, Sigma-Aldrich). Lymph nodes were 305 stained for 4 days at 37 deg. C in 400 μL blocking buffer, 2 μL CD3-BV421 (Cat: 100228, BioLegend), 306 and 2 μL B220-e660 (Cat: 50-0452-82, ThermoFisher). Lung tissue was stained for 3 days at 37 deg. C 307 in 500 μL blocking buffer and 2.5 μL ThermoFisher). Heart tissue was 308 stained for 1 day with 1 mM DRAQ5. Prostate tissue was incubated with fluorophore-conjugated anti- Committee of the University of Washington in accordance with NIH guidelines. 316 Collection and processing of expanded mouse kidney 317 4% PFA fixed mouse kidney was sliced to 200 μm and processed using a previously described protocol 318 [8]. The tissue was then incubated in blocking/permeabilization buffer for 6 hr at 4 deg. C. Primary 319 antibodies goat anti-podocalyxin (cat: AF1556, R&D Sys. Inc., 1:50) and rabbit anti-collagen IV (cat: 320 ab6586, abcam, 1:50) were diluted with blocking/permeabilization buffer and used to stain the tissue for 321 2 days at 4 deg. C. The tissue was then washed with 1× PBS three times at room temperature (1 hr 322 each). Fluorescently-labeled secondary antibodies, Alexa 488 conjugated WGA (cat: W11261, Thermo 323 Fisher Scientific, 1:25), and Hoechst 33342 were then diluted in blocking/permeabilization buffer to 324 stain the tissue for 2 days at 4 deg. C. The tissue was washed with 1xPBS three times at room 325 temperature (1 hr each) followed by incubating in 1 mM MA-NHS (cat:730300, Sigma-Aldrich) for 1 hr 326 at room temperature. The tissue was then incubated in monomer solution for 1 hr at 4 deg. C and then gelled in a humidified environment at 37 deg. C for 2 hr. Excess gel was removed and the specimen was 328 digested by proteinase K (cat: EO0491, Thermo Fisher Scientific) at 37 deg. C for two days and then 329 collagenase (cat; C7926, Sigma-Aldrich) at 37 deg. C for two days refreshing the solution daily. After 330 digestion, the specimen was incubated in DI water for at least 2 hr and the expansion factor was 331 determined through measuring the dimensions of the gel. The expanded specimen was mounted on poly-332 lysine coated film for imaging. 333 Collection and processing of human prostate biopsies 334 All specimens were obtained from an IRB-approved genitourinary biorepository with patient consent. 335 Core-needle biopsy specimens were obtained from fresh ex vivo prostatectomy specimens using an 18-336 gauge (approximately 1 mm inner diameter) needle biopsy device (Bard Max Core, Bard Biopsy). The 337 biopsy was immediately placed in 10% neutral buffered formalin, where it was maintained at room 338 temperature for 24 hrs. In contrast to mouse tissues, we found that human tissues require more 339 aggressive solvent-clearing approaches (Supplementary Figure 21). Due to its clearing efficacy and 340 non-toxic nature, we used ECi-clearing, which we observed does not interfere with downstream 341 histology or immunohistochemistry (Supplementary Figure 22). 342 Biopsies were then washed in 1× PBS with 0.1% Triton X-100 (Cat:T8787, Sigma-Aldrich), and 343 each biopsy was stained for 4 hr in a 1:2000 dilution of TO-PRO3 Iodide (Cat:T3605, Thermo-Fischer) 344 at room temperature with light shaking. Each biopsy was then dehydrated in ethanol for with 25/75, 345 50/50, 75/25, and 100/0 grades. The dehydration time for each grade was 1 hr, and the 100% ethanol 346 grade was performed twice to ensure removal of any excess water. Biopsies were then stained in 1:2000 347 dilution of Eosin-Y (Cat:3801615, Leica Biosystems) for 4 hr at room temperature with light shaking. 348 Finally, biopsies were cleared in ethyl-cinnamate (Cat:112372, Sigma-Aldrich) for 1 hr. Biopsy #12 was 349 stained with anti-CK8. The biopsy issue was incubated simultaneously with fluorophore-conjugated 350 anti-CK8-18 (Cat:MS743S0, Fisher) conjugated to Alexa-Fluor 488 (Cat:A20181, Invitrogen) (1:100 351 dilution) in PBS/1% non-fat dry milk/0.2% Triton X-100 at 37 deg. C for 7 days with gentle agitation.