Accelerated clearing and molecular labeling of large tissue samples using magnetohydrodynamic force

Here we report a strategy to efficiently render opaque biological tissues transparent and demonstrate that this approach can be modified to rapidly label intact samples with antibodies for large volume fluorescence microscopy. This strategy applies a magnetohydrodynamic (MHD) force to accelerate the removal of lipids from and the introduction of antibodies into tissue samples as large as an intact adult mouse brain. This strategy complements a growing array of tools that enable high-resolution 3-dimensional anatomical analyses in intact tissues using fluorescence microscopy. MHD-accelerated clearing and MHD-accelerated antibody labeling are fast, reliable, inexpensive, and compatible with existing strategies for high-quality fluorescence microscopy of intact tissues.

Many biological samples, including the brain of most vertebrate animals, are recalcitrant to microscopy 23 without first being cleared. Though chemical clearing has existed for over a century, these methods 24 tended to quench fluorescence, making them unsuitable for high resolution fluorescence microscopy 25 (Shultze, et al., 1897;Spalteholz, et al., 1914). More recent approaches to reduce opacity involve 26 removing lipids from the tissue sample to reduce light scattering at lipid water interfaces (Chung,et al.,27 2013). When combined with genetically encoded fluorophores, these approaches have proven powerful 28 for investigating anatomical relationships in a wide range of tissues and across a broad array of 29 resolutions (Hama, et  Immunohistochemistry is widely used because it is extremely effective at identifying molecules of 32 interest in thinly sectioned biological samples. However, diffusion of antibodies into larger tissues (e.g. 33 ~1 cm 3 ) takes inordinate amounts of time (weeks to months), depending on the size of the sample; even 34 then, antibodies often cannot penetrate deep enough to label the sample completely and uniformly. 35 Strategies have been developed to enhance the diffusion of antibodies into large tissue samples, 36 however, these strategies can have serious drawbacks. CLARITY-based approaches provide high 37 resolution imaging of endogenous fluorophores but can be difficult to implement with exogenous labels 38 (e.g., IHC) in large samples (Chung, et al., 2013;Kim, et al., 2015;Lee, et al., 2016). Other approaches 39 (e.g., iDISCO) use chemicals, microwaves, or electrophoresis to achieve optical transparency and allow 40 fluorescence microscopy of large samples labeled with antibodies, but can quench endogenous 41 fluorescent labels (Pan, et  Here, we describe a strategy that addresses both major barriers to high resolution fluorescent imaging 44 of large tissue samples. Magnetohydrodynamic (MHD) force, in combination with a conductive buffer 45 and detergent, efficiently removes lipids from hydrogel-infused tissue and produces transparent 46 samples excellent for fluorescence microscopy of genetically encoded fluorophores. The MHD force acts 47 directly on ions inside the tissue sample, which simultaneously propels lipids out of the tissue and 48 produces a constant flow of buffer through the tissue. This provides constant temperature regulation 49 through the entire tissue. With this method, transparency of an intact mouse brain can be achieved 50 within 2 days. MHD forces can subsequently be harnessed to drive antibodies into cleared tissues to 51 bind to antigens. MHD-accelerated clearing and labeling works in both vertebrate (shown for mouse and 52 zebrafish) and invertebrate (shown for the nudibranch mollusk Berghia stephanieae) species, providing a 53 generalizable method to render intact tissue transparent and accelerate immunohistochemical labeling 54 for fluorescence microscopy of intact tissues. 55

Results: 56
Effects of MHD force: MHD force produces a linear increase in flow velocity that is not observed with the 57 application of purely electrical force. To quantify the effects of the MHD force, we compared the 58 movement of sodium alginate spheres in response to purely electrical or MHD forces. The MHD 59 condition produced higher velocity flow over the electrical only condition for all tested non-zero 60 voltages ( the more than 50 cleared using this technique in multiple laboratories, all achieved transparency with 72 no visible damage ( Figure 2A). An intact adult mouse brain conditionally expressing GFP via EnvA-G-73 deleted rabies virus in aromatase expressing neurons was prepared using MHD-accelerated clearing and 74 imaged on a Zeiss Z1 lightsheet microscope (Watabe-Uchida et al., 2016; Yao, et al., 2017). Sparse GFP 75 cells were easily identified in the center of the brain ( Figure 3B, C, D). Whole brain images resolved 76 tissue architecture throughout the brain with subcellular resolution ( Figure 4C, D). Higher magnification 77 showed that fine processes, such as dendrites and axons, can be easily identified and analyzed several 78 mm from surface of the brain ( Figure 3D, Video 2). 79 An intact adult nudibranch (Berghia stephanieae) (medio-lateral: 1.3 mm, dorso-ventral: 1.5 mm, 82 anterio-posterior: 2 cm) that had been delipidated using the MHD-accelerated clearing device was 83 incubated with an anti-serotonin (5-HT; Immunostar; 1:500) antibody followed by a fluorescent 84 secondary (488 nm conjugated; ThermoFisher; 1:200). Passive incubation for 12 hours resulted in little 85 to no penetration into the brain ( Figure 4A), whereas MHD-accelerated antibody labeling for 12 hours 86 drove antibodies throughout the sample and revealed 5-HT expressing cell bodies and projections 87 ( Figure 4B). 88 Intact zebrafish brains (medio-lateral: 3 mm; dorso-ventral: 3 mm; anterio-posterior: 6 mm) were 89 passively delipidated in SDS for 7 days and then incubated with anti-acetylated tubulin antiserum 90 (Immunostar; 1:500; Piperno, et al., 1985) for 12 hours to identify neural fibers (Figure 4 C,D). Control 91 tissue samples (no MHD force applied) showed minimal antibody penetration along the outer edge of 92 the tissue with little fluorescence visible in the optic tectum ( Figure 4C). In contrast, MHD-accelerated 93 labeling for the same amount of time showed robust labeling of neural tracts throughout the brain 94 ( Figure 4D). 95 To test MHD-accelerated labeling in mammalian tissue, an anti-oxytocin (OT) antibody was applied to a 96 cube of mouse brain (medio-lateral: 6 mm, ventro-dorsal: 6 mm, antero-posterior: 6 mm) centered on 97 the periventricular nucleus of the hypothalamus (1:500 primary; 1:200 secondary). As above, antibodies 98 did not effectively penetrate the tissue sample in the absence of MHD force ( Figure 4E). In contrast, OT-99 expressing cells were clearly visible in the PVN, located deep within of tissue cube, using MHD-100 accelerated labeling ( Figure 4F). OT-expressing neuronal processes were easily resolved and were seen 101 to project towards the third ventricle and, consistent with OT neuron morphology. Accurate OT-labeling 102 was seen >1.8 mm from the nearest edge. The ability to visualize axonal varicosities and nuclei in  labeled neurons demonstrated that the MHD-accelerated labeling strategy can be used to resolve 104 subcellular structures ( Figure 4F). 105 To confirm the specificity of antibody binding is maintained in MHD-accelerated labeling, we used an 106 anti-vasopressin antibody in mice that expressed tdTomato in vasopressin-expressing neurons ( Additionally, MHD forces produce unidirectional buffer flow inside the tissue sample itself which helps 124 regulate temperature in the center of the sample and to pull unbound molecules into and out of the 125 sample. 126 The efficiency of MHD forces to rapidly drive charged molecules into and out of tissue is a consequence 127 of a fundamental difference in the way that MHD fields and electrical fields act on charged particles. 128 Electrophoresis drives cations and anions in opposite directions. In contrast, MHD-forces drive cations 129 and anions in the same direction along the third orthogonal axis (Qian, et al., 2009). Because the MHD 130 force acts on all ions is in the same direction, MHD creates a 'flow' in all charged molecules inside the 131 field ( Figure 1B). As a result, MHD forces generate rapid flow of buffer through the device and through a 132 tissue sample located within the MHD field (Video 1). Indeed, the velocity produced by MHD increases 133 linearly with voltage and far exceeds the nominal velocities produced in the same device when the 134 magnets are removed ( Figure 1). The induced flow of buffer dissipates local heating near the electrodes, 135 and in the core of the tissue sample, that could otherwise damage tissue. Because MHD forces are 136 cumulative to the maintained electrophoretic forces, the MHD approach described here maintains all 137 the advantages of electrophoretic tissue clearing and labeling and then adds a complementary force 138 that further propels lipids, antibodies, and buffer through the tissue. 139 140 MHD acts directly on ions at the intersection of electric and magnetic fields and both labeling and 141 clearing devices described here exploit this feature. By placing the tissue directly at the intersection of 142 these fields, a force is produced from inside the tissue ( Figure 1C). The buffer flow induced by this force 143 serves three purposes. First, in the case of tissue clearing it helps remove lipids from the tissue sample. 144 Second, in the case of antibody labeling it helps push antibodies into the tissue. While MHD acts directly 145 on antibodies, we believe that the strategy described here may be effective primarily because MHD 146 generates something akin to a river of buffer flowing in the same direction through the microchannels in 147 the hydrogel fixed tissue. Like twigs caught in the flow of a river, antibodies and lipids are pulled through 148 the tissue sample allowing rapid clearing and labeling. Third, MHD driven buffer flow facilitates efficient 149 thermal regulation of precious tissue samples. Instead of simply introducing cooler buffer solution to the 150 surface of the tissue, the cooler solution actively replaces the existing warmer buffer within the tissue. 151 This ensures that the center of the tissue will experience more similar conditions to the outer edges of 152 the tissue than would be possible with only an electrical field. As a result, an endogenous pump is no 153 longer required to maintain thermal equilibrium and a stronger force can be applied with less electrical 154 current (Qian, et al., 2009). Moreover, because the 'pumping' action of MHD is produced directly from 155 the electrical and magnetic fields without moving parts, it is virtually impossible for the pump to fail 156 during the clearing or labeling process. 157 This approach eliminates the need for solvents that are harmful to fluorophores (e.g., methanol), and 158 simplifies tissue clearing to the bare minimum components. Indeed, the only obligatory requirement is 159 that the tissue sample is held at the intersection of an electrical and a magnetic field. Thus, the strategy 160 outlined here is remarkably clean, efficient, and adaptable. The device itself can be 3D printed in plastic 161 ( Figure 5) making the device remarkably simple to build. Accordingly, with an investment of about $100 162 this device can be modified and specialized for the specific needs of a given experiment. 163 The MHD-based approaches described here will reliably allow rapid tissue clearing and antibody labeling 164 of large intact tissue samples, rendering them suitable for 3D fluorescent imaging. We demonstrate the 165 efficacy of a simple MHD device by clearing and labeling sea slug, zebrafish, and mouse tissue and using 166 multiple antibodies. Each of the antibody labeling procedures used here required no more than 4.5 mL 167 (1:200 concentration) of labeling solution which can be collected at the end of the procedure and 168 reused. Combined with the linear rate of antibody penetration observed with longer durations of active 169 labeling , we believe this system can be adapted for the fluorescent labeling of even larger samples, with 170 the limit ultimately constrained by the imaging system employed and not by the ability to produce high-171 quality tissue samples. 172

Methods: 173
Animal Use: All vertebrate animals were handled according to a protocol approved by the UMass 174 Amherst Institutional Animal Care and Use Committee (IACUC; protocol #2018-0014 and #2017-0060). 175 Measure of MHD-induced flow: A solution of sodium chloride was made in a small tank (2.5 L). Sodium 176 chloride was slowly added to the tank until the electric conductivity of the solution matched that of the 177 clearing solution. The clearing device was then submerged in the solution with a measured grid behind 178 the tank to provide scale. 0V, 10V, 20V, 30V, 40V, 50V, or 60V were then applied to the device and 179 sodium alginate spheres were introduced into the tank at a constant location (N = 7). The velocity of the 180 spheres through the device was measured. Velocity was calculated using a high-speed video taken over 181 a calibrated grid. This process was then repeated using only an electric field (without magnets). Paired-182 sample t-tests were performed between the MHD and electric-only conditions at each voltage using 183 MATLAB. The p-values were corrected for multiple comparisons using Bonferroni correction. Each 184 condition was fit to a linear model using MATLAB. (typically 16 hours for mouse brain tissue and 2 hours for intact zebrafish brains; Figure 5D). Refractive Index Matching and Light Sheet Microscopy: The tissue was transferred to "Optiview" (Isogai,215 et al., 2014) refractive index matching solution and incubated at 37 ˚C for at least 12 hours to achieve 216 optical clarity through RI matching ( Figure 5A; Isogai, et al., 2014). Samples were imaged at 5X or 20X 217 magnification with a lightsheet microscope adapted for a 1.45 RI imaging solution (Zeiss Z1). 218 Measures of Clearing Efficacy: 36 mouse brains were embedded in hydrogel, cleared using the MHD-219 accelerated clearing protocol, and assessed for transparency. The tissue was divided into two groups: 220 one that was pretreated by passively delipidating in SDS clearing solution for two days at 37 °C, and a 221 second that was placed in a 0.1 M borate buffered solution at 37 °C for the same time as the 222 pretreatment. Tissue samples from each condition (n=3) were then actively delipidated using the MHD-223 accelerated clearing system for 2, 6, 8, 12, or 15 hours. After washing in 0.01M PBS, the tissue was 224 equilibrated in Optiview (Isogai, et al., 2014) for 48 hours at 37 °C. 225 Transparency was determined by the percentage of light transmitted through the tissue and the 226 maximum depth from the external surface at which the morphology of neural processes (including 227 primary dendrites and axons) could be resolved. Light transmission was measured using a wide-228 spectrum light-source and calibrated photodiode. Data from each condition was fit with a saturating 229 exponential curve in MATLAB. 230 MHD-accelerated staining of fixed tissue with methylene blue: Penetration of methylene blue into a 1 231 cm 3 cube of homogeneous brain tissue under MHD force was tested over 1, 2, and 4 hours (N = 1). 232 Cubes of uncleared sheep brain tissue were equilibrated to the antibody labeling buffer solution for 12 233 hours. The tissue was then placed at the intersection of a strong magnetic and electric field (30V DC) and 234 submerged in a solution of methylene blue (0.1 M) buffered to pH 9.5 (37 °C). The orientation of the 235 electric field was reversed at 15-minute intervals for 3 minutes. Three samples were labeled using this 236 approach for 1, 2 or 4 hours. Following the stain, the tissue was bisected and imaged. A control sample 237 was incubated in the same solution (37 °C) for 4 hours without the application of any active force. This 238 sample was bisected and imaged as the others. 239

Comparative staining of methylene blue into agarose cubes as a result of various strengths of electrical 240
force conjugated to MHD force: 15 1 cm 3 of 3 % agarose were subjected to labeling methylene blue 241 labeling by MHD force for 0, 5, 10, 15, 30, 60, or 120 minutes at varied electrical field strengths. The 242 distance penetrated into the agarose cubes was measure after bisection and plotted against staining 243 time with 10, 20, or 30V in a constant magnetic field. 244 Optimization of MHD-Accelerated Immunohistochemistry: 0.22 cm 3 cubes of hydrogel were incubated 245 with FITC-conjugated antibodies (1:200; Jackson Immunoresearch) in a buffered solution (pH 9.5) with 246 MHD assistance, with an equivalent electrical field (30V DC), or passively for 1 hour at 37 °C (N = 1). 247 These cubes were incubated at the center of the intersection of the electrical and magnetic fields or the 248 center of the electrical field. After labeling samples were RI-matched using "Optiview" (Isogai,et al.,249 2017) and imaged using a Zeiss Z1 light-sheet microscope at 5x magnification. The farthest distance from 250 the edge of the tissue in the orientation of the MHD and electrical force where fluorescent antibody was 251 observed determined the penetration of the antibody. 252 Antibody Labeling: Delipidated tissue was placed inside of a 2-inch length of 0.25-inch diameter dialysis 253 tubing (6 -8 kDa); Spectrum). After equilibration, samples were incubated in an antibody solution inside 254 dialysis tubing at the center of intersecting electrical and magnetic fields where the MHD force was 255 strongest ( Figure 6). Confining the tissue sample inside dialysis tubing reduced the volume of antibody 256 required for labeling and protected the tissue sample and antibody solution from direct exposure to the 257 electrodes. Magnets (Applied magnets; NB057-6-N52) were placed on the top and bottom of the MHD 258 labeling device creating a central chamber Figure 6B). The ends of the dialysis tubing were connected to 259 9.5 mm diameter vinyl tubing (ThermoFisher: S504591) using 0.25-inch Leur lock barbs (Cole-Parmer; 260 UX-45501-20) to create a torus-shaped chamber allowing the antibody solution to circulate continuously 261 and provide an even and continuous source of antibody to the tissue sample ( Figure 6). Antibody 262 solution (4.5 mL; 0.1 M borate buffer titrated to pH 9.5 with 0.1 M LiOH, 1% heparin, 0.1% Triton X-100; 263 1:500 primary antibody) was transferred into the dialysis tubing using a 5 mL syringe. The labeling 264 chamber was submerged in a 1L tub containing 0.1M Borate Buffer pH 9.5/0.1% Triton X-100 solution. A 265 5 mL syringe filled with the buffer solution was attached to the circulation line to maintain constant 266 pressure inside of the dialysis tube. 60 volts DC (~0.2 Amps) was applied across the electrodes for 15 267 minutes, followed by 3 minutes of inactivity repeatedly for 12 hours to drive antibodies into the tissue 268 sample. The system was held at 37 ˚C and protected from ambient light to minimize bleaching of 269 fluorophores throughout the procedure. 270 Following each round of MHD-accelerated labeling, the antibody solution was replaced with a wash 271 solution (0.1 M borate buffer titrated to pH 9.5 with 0.1 M LiOH, 1% heparin, 0.1% Triton X-100) and the 272 tissue was exposed to 6-hours of "active washing" using the same voltage settings. Labeled tissue was 273 then washed in 0.01 M PBS for at least 12 hours. 274

Acknowledgements: 275
We thank D. Kelly, P. Sterling, and the members of the Bergan lab for helpful comments on this paper.  Transparency was measured as percentage wide-spectrum light penetration through the tissue (curve 351 fit with a saturating exponential). 352 transparency of brains that were actively cleared for 12 hours with and without passive treatment (N = 357 3). Transparency was measured as percentage wide-spectrum light penetration through the tissue 358 (curve fit with a saturating exponential). 359  hour of MHD-accelerated labeling, electrophoretic labeling, or passive labeling (N = 1) was measured 389 using light-sheet microscopy. The blue and orange bars represent the distance the antibody penetrated 390 the sample in the orientation of the electric (blue) or MHD (orange) force. 391