Novel electron microscopic staining method using traditional dye, hematoxylin

Uranyl acetate (UA) has been routinely used as a staining solution for ultrathin sections used in biological electron microscopy. As a radioactive nuclear material, UA is subject to strict international regulations. To develop an alternative and easy-to-use staining method for ultrathin sections, we examined various commercial light microscopic dyes. We found that Mayer’s hematoxylin followed by Reynold’s lead citrate solution showed staining results comparable to UA and Reynold’s lead citrate solution, and this method is therefore suggested as a reliable and promising alternative to UA staining.

Novel electron microscopic staining method using traditional dye, hematoxylin Hiroyuki Sasaki 1* , Hisako Arai 2 , Emi Kikuchi 2 , Hideki Saito 2 , Keiko Seki 2 & Takeshi Matsui 3* Uranyl acetate (UA) has been routinely used as a staining solution for ultrathin sections used in biological electron microscopy. As a radioactive nuclear material, UA is subject to strict international regulations. To develop an alternative and easy-to-use staining method for ultrathin sections, we examined various commercial light microscopic dyes. We found that Mayer's hematoxylin followed by Reynold's lead citrate solution showed staining results comparable to UA and Reynold's lead citrate solution, and this method is therefore suggested as a reliable and promising alternative to UA staining.
The electron microscopic staining method for biological specimens using uranyl acetate (UA) was reported in 1958 by Watson 1 . Since then, the double staining method of UA followed by lead solution has been used in electron microscopy facilities worldwide owing to its simplicity and optimal staining results 1-3 . Furthermore, array tomography in electron microscopy (EM) with serial section transmission electron microscopy (TEM) or scanning electron microscopy (SEM), serial block-face imaging SEM, and focused ion beam SEM have recently become increasingly popular for a wide range of applications in many biological science disciplines 4,5 . Array tomography offers more flexibility than serial block-face imaging SEM and focused ion beam SEM, because it preserves all sections 4 . Recent technical advances have enabled us to prepare 300-5000 serial ultrathin sections of specimens, stain them with UA, and acquire images by TEM, resulting in terabytes of data. During this procedure, a large amount of UA is required 4,5 . However, obtaining uranyl compounds has recently become difficult because of strict international regulations 6 . Furthermore, restrictions on their use as well as their availability, storage, and disposal, are also expected to become stricter worldwide, because they are used as nuclear materials for weapons.
Here, we establish a novel staining method, using a pre-dyeing agent that is easy to handle, as an alternative to double staining with UA and other heavy metals. We examined various basic staining solutions that are conventionally used in light microscopic methods to identify an alternative reagent, which can stain the conventionally prepared thin and semi-thin sections embedded in epoxy resin (Supplementary Fig. 1; Supplementary information).

Results and discussions
Double staining with various dyes (hematoxylins, Kernechtrot, basic fuchsin, methyl green, alizarin red, and Giemsa solution) followed by Reynold's lead citrate solution 18 (RPb) (Fig. 1a-h) indicated that the best ultrastructure and image contrast were obtained upon double staining with Mayer's hematoxylin (MH) for 10 min and RPb for 5 min (MH-RPb). The staining effects were generally good, although the image contrast was slightly softer compared to that obtained by double staining with UA for 5 min and RPb (UA-RPb) for 5 min (Fig. 1a). Staining times of 3, 5, 10, 15, and 20 min for MH were attempted and the staining results improved with staining time. The optimal staining time was 10 min because contamination was increased after 15 and 20 min of staining (data not shown). Among various commercial hematoxylin stains (Mayer's, Lilly Mayer's, Carracci and Gill No.3), MH showed the best contrast ( Fig. 1a- www.nature.com/scientificreports/ citric acid are only present in MH, suggesting that these are responsible for the better electron staining effect. Further investigation will be needed in the future. Ultrathin sections of mouse livers were stained solely with UA for 5 min, RPb for 5 min, MH for 3 min, MH for 5 min, or MH for 10 min. Cells were not clearly observed in the 3 min or 5 min MH single staining samples, but slight contrast enhancements of the hepatocytes were observed in the UA, RPb, and 10 min MH stained samples ( Fig. 1i-m). Lilly Mayer's and Carracci hematoxylins showed numerous section breakages, presumably because these stains contain glycerin (Table 1). With the Gill No. 3 stain, there was no structure-specific increase in contrast, and the entire section was stained darkly.
Digital images were subjected to gray value (i.e. contrast) line analysis using ImageJ image processing software to semi-quantitatively determine the image quality. Figure 2 shows the line profile displaying two-dimensional graphs of the intensities of pixels along black lines within the images. There was a significant difference in contrast between the upper panel (MH-RPb and UA-RPb, Fig. 2a  In MH-RPb, nuclear chromatin was well stained in the liver tissue as well as in various other cells and tissues (Fig. 3). Other organelles such as the cell membrane structures, ribosomes, glycogen, fat droplets, cell adhesion apparatus, and cytoskeletal systems also showed enhanced contrast ( Fig. 3a-o). The plasma membrane staining in all samples was also satisfactory. The Z-bands of the skeletal muscle seemed to stain with a lower density, whereas the actin and myosin fibers showed sufficient contrast (Fig. 3j,k). In bacteria, MH-RPb showed the cell walls of E. coli and Staphylococci with higher affinity and contrast compared to those obtained by double staining with UA-RPb. Additionally, the agar used for bacterial collection also showed staining with high contrast (Fig. 3l-m). Back scatter image of 200-nm semi-thin sections of mouse kidney observed with FE-SEM showed high quality images of the renal cortex (Fig. 3n) and the renal glomerulus (Fig. 3o). Although the entire areas on the slide grass were slightly stained blue, we were unable to observe the structures with a light microscope.
In these all cells and tissues, the roughness of the image quality was unnoticeable, even at a high magnification of approximately 20,000×. Furthermore, good staining results have also been obtained with insect flight muscle embedded in Araldite-506 epoxy resin as another example of a resin compatible for MH staining (Fig. 3k). However, in neural tissue, MH-RPb deposited high electron-dense granules only on the myelin sheath (Fig. 3i  inset). The cause of the low-density contrast in the Z-bands of the skeletal muscles are unknown. Electron-dense granular depositions in the myelin sheath may occur in phospholipid-rich structures. Further verification of these phenomena will be needed in the future.
UA and MH have an affinity to macromolecular structures containing nucleic acids, i.e., nucleoplasm and ribosomes 1,19 . However, lead ions of alkaline lead citrate bind to sulfhydryl and carboxyl groups in the tissues, resulting in enhanced contrast by interacting with proteins and glycogens 18 . MH, which was used to prepare tissue specimens in this experiment, was a sufficient alternative to UA. Although single staining of MH and MH-RPb are slightly inferior to that of UA and UA-RPb. The reason for the somewhat inferior single staining effect of MH compared with that of UA and lead may be that the metal it contains, which has electron scattering effects, is aluminum (atomic number = 13), which has a lower atomic number than those of uranium (atomic number = 92) and lead (atomic number = 82). These differences were thought to be compensated for the contrast by the lead used in the post-staining. Of the various kinds of commercial hematoxylin solutions, MH stains nuclei selectively as it contains citric acid. It is a useful staining method that does not require separation by hydrochloric acid in alcohol, or background co-staining 19 . The hematein produced by the oxidation of hematoxylin forms a complex with the aluminum part of the mordant and becomes positively charged. It then binds to the phosphate and/or carboxyl groups of the nucleic acid, which is negatively charged, resulting in staining 19 . It was therefore assumed that the contrast in the EM image was due to the high amount of aluminum-hematein in the MH (Supplementary Fig. 1; Supplementary information).
The uranyl-lead staining procedure dates to 1958. Currently (2022), transmission electron microscopy has evolved into an instrument with greatly improved contrast. Modern electron optics, variable acceleration voltages, variable apertures, high-contrast and high-resolution charge-coupled device (CCD) or complementary metal oxide semiconductor (CMOS) camera image recording, and high-performance image processing software will undoubtedly improve image quality, even for low-contrast specimens. However, it is likely that double staining with uranium acetate and lead will still be used extensively in many electron microscope facilities around the world. MH has advantages, such as a stable supply of commercially and economically available dye solutions and no need to worry about liquid waste (as it is widely used in staining paraffin sections of clinical samples for diagnosis). The staining time is 5-20 min, which is the same as that of UA. However, a disadvantage of MH is that it stains as a dark blue-violet color, which makes it difficult to see the grids during soaking. This can be overcome by staining the grid floating on the droplets of the MH solution.
The International Atomic Energy Agency's "International Basic Safety Standards for Protection against Ionizing Radiation and for the Safety of Radiation Sources" (BSS) has established specific exemption levels, and new regulations for radioactive materials are being developed internationally through legislation 6 . As described above,   The staining solutions were filtered through a Millipore filter (0.45 μm, Merck, Darmstadt, Germany) before use. Staining was performed with each staining solution for 3, 5, 10, 15 and 20 min, respectively, followed by RPb for 5 min. As a control, conventional double staining was performed using saturated UA solution for 5 min and RPb for 5 min. Furthermore, sections were solely stained MH, UA or RPb, respectively.
The samples were observed and recorded using a Hitachi H-7500 transmission electron microscope (Hitachi High Technologies, Tokyo, Japan) with a 20 µm objective lens aperture at an acceleration voltage of 100 kV. The images were digitized with a 1024-pixel (px) × 1024-px AMT Advantage 12HR slow scan CCD camera system (Advanced Microscopy Techniques, Woburn, MA) mounted on a TEM, which was recorded using AMT Image Capture Engine software ver. 5.4.2.40 (Advanced Microscopy Techniques, Woburn, MA). To determine the staining quality, the brightness and contrast was maintained during image acquisition with the AMT Image Capture Engine. The resulting digital images of each image were subjected to gray level (contrast) line analysis using ImageJ (National Institute of Mental Health, Bethesda, Maryland, USA) image processing software to determine the image quality. Additionally, 200-nm sections prepared with an EM UC-7 ultramicrotome (Leica Microsystems) attached to glass slides were stained with hematoxylin for 20 min and with RPb for 5 min, and backscattered electron images were obtained using a Hitachi Regulus 8240 field emission scanning electron microscope attached to an yttrium aluminum garnet (YAG) backscatter electron detector (Hitachi High Technologies, Tokyo, Japan) with 1280-px × 960-px at an acceleration voltage of 5 kV (Supplementary information).

Data availability
The datasets generated and/or analysed during the current study available from the corresponding authors on reasonable request.