X-ray microscopy as an approach to increasing accuracy and efficiency of serial block-face imaging for correlated light and electron microscopy of biological specimens.

Sample Preparation

All animals were used according to a protocol approved by the Institutional Animal Use and Care Committee at the University of California, San Diego following AAALAC guidelines. Mice were anesthetized with ketamine / xylazine and transcardially perfused with Tyrode’s solution followed by 2.5% glutaraldehyde / 2% formaldehyde in 0.15 M cacodylate buffer containing 2 mM CaCl 2 . The brain was removed and postfixed on ice in the same fixative solution for 2 hours and then cut into 100 µm thick slices on a vibrating microtome. The slices were post-fixed at 4 °C overnight before further processing. For iontophoretic dye-filling and photooxidation, a mouse was perfused with 4% formaldehyde / 0.1% glutaraldehyde in 0.1M PBS. The brain was postfixed on ice for 2 hours and then cut into 100 µm thick slices. An astrocyte in the hippocampus was filled with 5% Lucifer yellow (LY) and then postfixed in 4% formaldehyde / 0.1% glutaraldehyde for 1 hour. LY fluorescence was imaged and the slice was bathed in 100 mM glycine-PBS buffer to quench excess aldehydes, washed with PBS, and then bathed for 5 minutes in PBS containing 0.15% 3,3’-diaminobenzidine tetrahydrochloride (DAB) and 0.1% potassium cyanide. The tissue was then illuminated with a LY filter set until LY fluorescence was replaced with light brown precipitate in transmitted light mode. The tissue was washed with PBS and then stained for SBEM imaging, as described below. For MiniSOG photooxidation, a mouse expressing MiniSOG and tdTomato in a subpopulation of retinal ganglion cells (RGCs) was transcardially perfused with 4% formaldehyde / 0.1% glutaraldehyde in 0.1M PBS. The retina was dissected and post-fixed with 4% formaldehyde in 0.1M PBS on ice for 2 hours and then cut into 100 µm thick vertical slices. The tdTomato expressed RGCs were imaged and the retina was fixed with 2.5% glutaraldehyde, 2.5mM CaCl 2 in 0.15 M cacodylate buffer (CB) pH 7.4, and the tissue was rinsed five times with ice-cold CB, and blocked for 30 min with 10 mM KCN, 20 mM aminotriazole, 50 mM glycine, and 0.01% hydrogen peroxide in CB. Freshly prepared DAB in CB was added to the retina, and RGCs were illuminated with 450–490 nm light from a xenon lamp for 10–15 minutes until a light brown reaction product was observed in place of the green fluorescence of MiniSOG. The tissue was then processed for SBEM.

SBEM Staining

Tissue was prepared for SBEM as previously described ( Deerinck, et al., 2010 ). Briefly, tissue was washed with buffer and then placed into 2% OsO 4 / 1.5% potassium ferrocyanide in either 0.15M CB containing 2 mM CaCl 2 or 0.1M PBS for LY photooxidized specimens. The slices were left for 30 minutes on ice and then 30 minutes a room temperature. After thorough washing in double distilled water, the slices were placed into 0.5% thiocarbohydrazide for 30 minutes. The slices were again washed and then stained with 2% aq. OsO 4 for 30 min. The slices were washed and then placed into 2% aq. uranyl acetate overnight at 4 °C. The slices were washed with water at room temp and then stained with en bloc lead aspartate for 30 minutes at 60 °C. The slices were washed with water and then dehydrated on ice in 70%, 90%, 100%, 100% ethanol solutions for 10 minutes at each step. The slices were then washed twice in dry acetone and then placed into 50:50 Durcupan ACM:acetone overnight. The slices were transferred to 100% Durcupan resin overnight in vacuum chamber. The slices were then flat embedded between glass slides coated with mould-release compound and left in oven at 60 °C for 48 hours.

XRM Sample Preparation

Generally, small blocks of tissue (maximum dimension < 5mm) were cut from glass slides and then mounted with cyanoacrylate glue to the top of aluminum rods. For some samples, a rhodium-flashed copper finder grid was affixed to one face of the specimen using 5-minute epoxy (Product #14250; Devcon, Danvers, MA, USA). Alternatively, one face of the specimen was covered with a small aliquot of poly-L-lysine solution and left for 2–3 minutes. The lysine solution was briefly washed off under a stream of water and then the specimen was allowed to air dry. Tungsten (IV) carbide powder (2 µm particles; Sigma-Aldrich Corp., St. Louis, MO, USA) was temporarily suspended in distilled water by vortexing 0.1 g powder in 10 mL water. A small drop was placed on the surface of the sample previously coated with lysine. The sample was allowed to rest for 2 minutes and then wash briefly rinsed several times under a stream of distilled water and allowed to dry.

Biolistic labeling

Nanophosphor particles (NaYSO 4 : Yb, Er) with a diameter of 1 µm were kindly provided by Intelligent Material Solutions, Inc. (San Diego, CA, USA). A 2 mg/mL solution of nanophosphor particles in 95% isopropanol/5% acetone was deposited within Tefzel tubing to create bullets for a Helios gene gun (Bio-Rad Laboratories, Hercules, CA, USA), as previously described ( O'Brien & Lummis, 2011 ). The nanophosphor particles were shot into slices of brain tissue at 160 psi with a 2 cm distance between the tip of the gun nozzle and the tissue slice. The slice was then washed briefly in buffer, imaged by light microscopy, and subsequently processed for SBEM staining.

Light Microscopy

Prior to photooxidation, LY-filled cells were imaged on a Leica SPEII using a 63× oil objective (NA 1.30). Brain slices containing nanophosphor particles were imaged on a Radiance 2000 microscope (Bio-Rad) equipped with a Ti-sapphire infrared laser (Spectra-Physics, Santa Clara, CA, USA). The slices were initially imaged with a 10× objective (Nikon, NA 0.3; Chiyoda, Tokyo, Japan) with pulsed-mode illumination (980 nm) and then continuous wave illumination (980 nm), in order to collect nanophosphor and autofluorescence signal, respectively. Sub-regions were then imaged with a 40× objective (Nikon, NA 1.30) using continuous wave illumination at 980 nm, simultaneously collecting fluorescence and transmitted light images. XRM Most XRM work was performed on a MicroXCT-200 instrument (Zeiss X-Ray Microscopy, Pleasanton, CA, USA). This architecture combines both geometric (as found in conventional microCT constructs) and optical magnification (lens-coupled post-sample magnification) as well as optimized scintillation technologies to yield high-contrast sub-micron 3D reconstructions of a sample. XRM tilt series were generally collected at 40 kV and 4W power (100 µA current), unless otherwise noted. A Zeiss Xradia 510 Versa (Zeiss X-Ray Microscopy, Pleasanton, CA, USA) was utilized for these studies as well, as indicated in Figure 3 . Collection parameters and acquisition times depend on sample geometry and resolution and are indicated in figures legends for each data set.

Serial Block-face Imaging

SBEM data was collected with a 3View unit (Gatan, Inc., Pleasanton, CA, USA) installed on a Merlin field emission SEM (Carl Zeiss Microscopy, Jena, Germany). The retinal volume was collected at 2.0 kV accelerating voltage, with a raster size of 19k × 39k and pixel dwell time of 0.5 µsec. The pixel size was 5.7 nm and section thickness was 70 nm. The astrocyte volume was collected at 2.0 kV, with a raster of 20k × 20k and pixel dwell time of 1.0 µsec. The pixel size was 4.5 nm and section thickness was 90 nm.

Image Analysis

XRM volumes were generated from XRM tilt series using XMReconstructor (Xradia). Scaling, rotation, transformation of images to find best alignment between features of interest was performed in Photoshop (Adobe Systems Inc., San Jose, CA, USA) or Imaris (Bitplane AG, Zurich, Switzerland). 2D affine transformation was performed using SciPy LinearNDInterpolator. In order to estimate the location of a target ROI on the SEM stage coordinates, fiducial markers were chosen near the ROI and their SEM stage coordinates were collected, as well as their coordinates in a previously acquired XRM volume. The two sets of coordinates were then used to calculate the SEM coordinates for the ROI based on the XRM coordinates of the ROI.