Silver Carbonate Staining Reveals Mitochondrial Heterogeneity

Río-Hortega's Silver Carbonate Solution

As in many other silver methods, the silver solution must be carefully prepared in clean glassware and with high-quality distilled water. The silver carbonate obtained by precipitation of silver nitrate with sodium carbonate must be white or yellowish. If it is brown or black, both reagents must be discarded. The solution is stable for months in dark bottles and even in clear bottles if not contaminated by dust particles. For this solution, 30 ml of 10% silver nitrate is admixed with 90 ml of 5% sodium carbonate. After the silver carbonate precipitate has formed, 25% ammonia is added dropwise until precisely the point at which the precipitate disappears. Any excess of ammonia invalidates the reagent. Finally, 330 ml of distilled water is added (Río-Hortega 1918). This is the “débil” (diluted) solution that may be used directly or as a 50% aqueous dilution.

Polak's Method with Silver Carbonate

This method was performed as follows. (a) Tissues were fixed in 10% formol, phosphate-buffered formol, or 4% paraformaldehyde-PBS for 24 hr up to 2 weeks. It should be noted that glutaraldehyde- or formol-glutaraldehyde fixed tissues do not react in the same way. (b) Frozen or cryostat sections 5–15 μm were recovered, as above, in a Petri dish with slightly alkaline distilled water (1 drop of ammonia/50 ml). PBS is not recommended in this case. (c) For mordant, the specimens were washed in distilled water and incubated for 30–40 min at 37C in prewarmed 5% uranyl acetate (or overnight at RT). (d) Impregnation was carried out with specimens adhering to a bent and very thin glass rod. Each section was immersed for 15–60 sec in the silver solution with constant slow motion until the yellowish color disappeared. (e) For reduction, the section was placed in 1% high-quality freshly prepared formalin (neutral but not buffered) and agitated slowly until some brown clouds are liberated (2 min or more). (f) For gold toning, specimens were washed in distilled water and placed in 1:500 (w/v) gold chloride at RT for 30–60 min. (g) The samples were then fixed in 1:500 (w/v) sodium thiosulfate for 3–5 min and washed in three changes of distilled water. (h) Specimens were mounted on gelatinized slides, dehydrated quickly with alcohols and xylene or phenicated xylene (if available), and mounted in Merckoglas or DPX. Although thin sections are preferable, sections of up to 20-μm can be stained. Thicker specimens cannot be used because the impregnation of the central part is not complete even with prolonged impregnation. Furthermore, the nuclei start to impregnate and the staining is no longer selective for mitochondria.

The silver nitrate (S-0139), sodium carbonate (S-6139), and gold chloride (tetrachloroauric acid (III) (Cl₄AuH) yellow, G-4022, trihydrated) were obtained from Sigma (St Louis, MO). Uranyl acetate (dihydrate) (108473) and Merckoglas were from Merck (Darmstadt, Germany).

Optical Microscopy

These silver staining procedures stain mitochondria with unmatched contrast, selectivity, and low background. Using a ×100 oil objective and a narrow depth of field, the high diffraction angle from the metallized mitochondrial border permits blurring of normal-sized mitochondria (around 1 μm diameter in mouse hepatocytes) through minimal variations in z-axis and the successive focusing of several mitochondria on the same section (Figures 1 and 2). Three-dimensional counting was easily performed with an ocular grid in all tissues studied. Material from the central nervous system showed less mitochondrial density than expected in the neuropil, and in the myelinized areas the impregnation did not work. Given sections of standardized thickness, a statistically ran dom sampling on vertical columns is a more rapid method to estimate the distribution of mitochondrial density than stereological methods.

Merging the different digital images (Figure 3) taken at three to five focused z-levels with the algorithm used by the Microimage software changes the values of peripheral pixels for minute mitochondria (≤0.3 μm), but the central pixels maintain their gray values and this is related to the optical density (and staining density). Therefore, although the staining intensity could be digitally quantified, there is no correspondence between density and diameter (see below) and it is not a useful parameter. The detection limit for minute mitochondria in brightfield was very low and difficult to quantify but could be estimated from the distance that can be resolved between mitochondria. This limit appears to be exactly at the resolution limit for the optical microscopy (0.2 μm) (Figure 2, inset). This limit of detection is in agreement with further studies of the deposition pattern of metallic dots in TEM (see below). At the optical microscopic level, only mitochondria are detected but not all the mitochondria are stained, as revealed by TEM controls.

Electron Microscopy

Examination of ultrathin sections of impregnated tissue, not counterstained, revealed that the colloidal silver, converted to metallic silver by formol reduction, had been substituted by colloid-like gold deposits. These have little or no tendency to aggregate and vary in size from 5 to 20 nm (Figure 4; the small red dot scale is 20 nm). These deposits remain perfectly circumscribed by an invisible membrane with an otherwise absolutely clean adjacent background. Adjacent mitochondria of the same size and in the same area were differentially stained. The density of gold grains was not related to mitochondrial size in the medium range, but all the really large mitochondria were intensely stained, as were the smallest mitochondria. Approximately half of the heavily impregnated mitochondria had clumps of argyrophilic material apposed on their surface.

A continuum of impregnation density was evident until the mitochondria become small, dense, and aggregated deposits with irregular limits. To locate these deposits and to correlate them with hepatocyte ultrastructure, uranyl-lead stained grids were studied. The results indicated that not all the mitochondria were stained. In spite of the technical manipulation and the absence of osmium postfixation, ultrastructure was clearly recognizable, and in every field 10–25% of otherwise normal mitochondria were not stained at all. In some hepatocytes the percentage was low, and in others the staining was only moderate, perhaps reflecting the well-established differences in hepatocyte ultrastructure related to acinar position. Mitochondrial impregnation was heterogeneous and sporadic cytoplasmic foci were also detected. However, these did not arise from unspecific background impregnation or technical noise.

OPEN IN VIEWER

Figures 1–16

Silver carbonate-impregnated mitochondria in hepatocytes. Images at ×100 focused at four different z-levels (two are shown in Figures 1 and 2 ) were merged to a single image ( Figure 3 ) Red bar = 10 μm. Digital image from a 4 × 4-μm area was enlarged in both single and merged to evaluate the effect of the merging algorithm on image resolution (inset in Figure 2 ). Black bars = 1 and 0.2 μm. Non-counterstained ultrathin section ( Figure 4 ) shows the colloid-like gold dots bounded by the limiting membrane (not stained). Red dot scale = 20 nm. Uranyl and lead-counterstained ultrathin section permits ultrastructural location ( Figure 5 ). Heterogeneous impregnation can be graded in four stages (numbers). Free cytoplasmic but circumscribed deposits and non-stained mitochondria (m) can be detected. Steropair image ( Figure 6 ) of a small mitochondrion (blue bar = 100 nm) reveals a negative staining of slit-like spaces (cristae) (arrows) and the pattern of staining inside matrix (and not along membranes). Progressive accumulation obliterates all the matrix space ( Figures 7 and 8 ) Black bars = 500 nm. Surface material can be shown apposed to but not connected with internal content ( Figure 9 ). Three of the four images interpreted as fusion events reveal asymmetric distribution of matrix material (Figures 10–13) Blue bars = 200 nm. Advanced, “piknotic” and disintegration stages ( Figures 14 and 15 ) merge gradually with images of dispersed cytoplasmic material ( Figure 16 ).

Silver carbonate staining of mitochondria revealed not only heterogeneity but also what appears to be a graded morphological transition (Figure 5). This can be described as four different stages starting from very slight staining (Figure 5, number 1). The individual mitochondria varied from slightly and diffusely stained (Figure 5, number 1) to an intense accumulation in the matrix (Figure 5, number 2), to complete staining with an irregular surface (clumps of argyrophilic material appear on the external surface) (Figure 5, number 3) and, finally, small irregularly shaped mitochondria exhibiting very dense staining (Figure 5, number 4). Save for the most densely stained mitochondria, colloidal silver was not deposited along outer membranes but only within the mitochondrial matrix. In stereo images, the staining revealed slit-like plates corresponding to the negative images of cristae (Figures 6 and 7). Argyrophilic material appeared to form clumps on the mitochondrial surface only on the compacted-stage compartments (Figure 8) and, in these cases, was clearly located outside the limiting membrane (Figure 9). Some images could be attributed to mitochondrial fission or fragmentation (Figure 10), whereas other images suggested possible mitochondrial fusion (Figure 10, inset, and Figures 11–13). The most frequent were two-hemispheric images suggesting a lateral fusion and not elongation and fission (Figures 12 and 13). The cytoplasmic foci show also a gradual transition from small isolated and dispersed dots to well-defined clouds (200–300-nm diameter) with some dense clumps. This transition can also be interpreted in reverse, as the final dissolution of shrunken mitochondria, as suggested by the transition from dense broken masses, to dispersed material (Figures 14–16).