The hybrid freezing method
To achieve an adequate ultrastructural preservation, we developed a combined protocol of chemical fixation, post-fixation, cryofixation, and cryo-substitution, which we refer to as hybrid freezing method (HFM). The hybrid freezing method contained aldehyde fixation, post-fixation with osmium tetroxide, high pressure freezing (HPF), and cryo-substitution with three different cryo-substitution cocktails, and was compared to the standard method, an aldehyde fixation followed by dehydration at room temperature.
A first test comparing a sample that had been post fixed in osmium tetroxide prior to freezing with another sample without post fixation showed that post-fixation with osmium tetroxide is an important step towards the preservation of the myelinated axon in human brain tissue (Fig. 1). In the left panel of this figure, the individual layers of the myelin sheath can be recognized, and they remain compact compared to the sample that had not been post fixed (shown in the right panel).
Comparison of standard processing with hybrid freezing method (HFM) using three different cryo-substitution cocktails
At first, we examined the general ultrastructural preservation of frontal lobe samples taken from four patients with difference in PMIs. It showed that the myelin sheaths remained visible regardless of the post-mortem times, and thus the myelin sheaths appeared more resistant against autolysis (compare Fig. 2a-d with Fig. 2e-h) than other cell components. Therefore, we focused on the myelin sheaths during our detailed comparison of four different embedding procedures for brain samples. A detailed analysis was performed by two independent researchers in a double-blinded approach. This analysis showed that the samples prepared with substitution cocktail I, which contains osmium tetroxide and uranyl acetate, showed the highest contrast in each sample (compare Fig. 2b and f with all the other panels of this figure).
Preservation of myelin sheaths in the human frontal lobe
Focusing on the myelin sheaths, we tested if the novel hybrid freezing method (HFM) preserves the myelin sheaths better than standard embedding. An assessment of myelin sheath preservation was done with a point counting method [12], estimating the percentage of the myelin sheath volume that appeared well preserved, disbanded, or unspecified in each brain sample with each of the four processing procedures. Figure 3 shows that the PMI appeared to have an influence on the preservation of the myelin sheaths: the shorter the PMI, the higher the percentage of well-preserved parts of the myelin sheaths. Moreover, in three out of four cases, the HFM with cryo-substitution cocktail I showed a clear increase in well-preserved myelin sheath volume with respect to classical embedding (Fig. 3). This increase was largest in the sample with the lowest post-mortem time (case 1, from 15.4% in standard preparation to 23.2% with cocktail I). In case 2, cocktail 1 apparently neither increased nor decreased the percentage of well-preserved myelin sheaths with respect to classical embedding (18.2% of the myelin sheath volume appeared well preserved, both after standard embedding and after using substitution cocktail I, Fig. 3, second panel). Case 3 with a PMI of 20 h and case 4 (PMI of 24 h) showed a lower incidence of well-preserved myelin sheaths and a lower increase of well-preserved myelin sheaths when applying cocktail I (from 14.7 to 16.4% and from 7.4 to 11.7%, respectively, Fig. 3 right side). Conversely, the percentage of disbanded myelin sheaths decreased from 21.3% in the standard embedded sample to 16.4% in the sample processed with cocktail I. An exception was Case 2, in which the application of cocktail I apparently did not change the myelin sheath preservation (18.2% both after standard embedding and after using substitution cocktail I, Fig. 3, second panel). In contrast to cocktail I, no trend was visible when determining the percentage of well-preserved myelin sheaths after freeze substitution in cocktails II and III, both solutions either increased or decreased the percentage of well-preserved myelin sheaths (Fig. 3).
Influence of post-mortem interval on ultrastructure of human brain tissue
The natural autolysis of the brain samples dismantled the ultrastructure and could not be reversed with the HFM. However, those structures that had not been degraded by autolysis, such as the myelin sheaths, appeared to be better preserved after HFM and freeze substitution with cocktail I. Moreover, this cocktail also resulted in consistently improved myelin sheath preservation with respect to standard embedding. Whereas the myelin sheaths remained visible even after long post-mortem times, finer structures, such as microtubules in the axons, were increasingly degraded with an increasing PMI. These drastic losses of ultrastructural details in the tissue were due to autolytic processes after the death of the patient. Other effects of autolysis were the enlargement of the coarse ER and loss of its ribosomes, swelling of the mitochondria and the fact that the chromatin in the nuclei became increasingly coarse (Fig. 4), resulting in an “empty” appearance of the cytoplasm. (i.e. it appeared void of compartments). After long post-mortem times, we were not able to recognize many of these compartments any more, instead, vacuole-like structures became apparent.
Light microscopic signs of autolysis were dilations around the vessels and the glia cells. These increased dramatically in size with increasing PMI (Fig. 5). The samples taken from case 1 (PMI 16 h, frontal grey matter) showed a dense organization of the neuronal tissue with a high number of finer structures. In the samples taken from frontal white matter (Fig. 5c-e), with increasing PMI, the organization of the tissue loosened more and more, and vacuole-like structures (putative degraded cellular compartments that could not be recognized any more) dominated instead (Fig. 5 and Fig. 6). Samples with a long PMI (case 4) showed an increased number of vacuole-like structures and hardly any structural details in the surrounding matrix.
Nevertheless, the samples processed with the HFM with cocktail I appeared to have a denser organization of the tissue and better preservation of finer details than those samples processed with standard embedding (compare Fig. 6a-d with Fig. 6e-h).
Analysis of myelin sheath thickness (g-ratio) and density
G-ratios were measured for each sample showing that each individual had their own g-ratio for each sample. In some cases, the individual g-ratios differed significantly from each other. For example, the g-ratios of case 2 and case 3 showed a significant difference between all four ways of processing the tissue, Fig. 7. With standard resin embedding, the mean values ± standard deviation of the g-ratios were as follows: case 1: (PMI 16 h) 0.59 ± 0.12, case 2: (PMI 18 h) 0.62 ± 0.12, case 3: (PMI 20 h) 0.70 ± 0.15, and case 4 (PMI 24 h) 0.67 ± 0.12. With cryo-substitution cocktail I, case 1: 0.61 ± 0.12, case 2: 0.64 ± 0.10, case 3: 0.70 ± 0.11 and case 4: 0.65 ± 0.12. For cocktail II the g-ratios were for case 1: 0.65 ± 0.12, case 2: 0.62 ± 0.12, case 3: 0.73 ± 0.12, case 4: 0.68 ± 0.12. After applying cocktail III, the ratios were 0.63 ± 0.14 for case 1, 0.61 ± 0.13 for case 2, 0.69 ± 0.10 for case 3, and 0.67 ± 0.12 for case 4. There was no statistically significant difference between g-ratios when the different substitution cocktails are compared in the same case as determined by one-way ANOVA, case 1: (F (3,232) = 1.957, p = 0.121), case 2: (F (3,213) = 0.515, p = 0.673), case 3: (F (3,231) = 0.448, p = 0.719) and case 4: (F (3,226) = 0.418, p = 0.740).
A comparison of the densities of myelinated axona in sections of human frontal white matter revealed two facts. Firstly, these densities decreased with longer PMI, as can be seen in Fig. 8. Second, the samples prepared with the standard embedding approach showed significantly higher densities of myelinated axons than hybrid frozen samples (Fig. 8). In total, over 6000 myelinated axons were counted in ~ 6000 μm2 image area. A one-way ANOVA revealed significant differences between the two embedding methods regarding the density of the myelinated axons, standard embedding: (F (33,212), p = 0.000), cocktail I: (F (22,943), p = 0,000). With standard resin embedding, the mean values ± standard deviation of the axon density were as follows: case 2 (PMI 18 h): 1.59 ± 0.22, case 3 (PMI 20 h): 0.99 ± 0.06, case 4 (PMI 24 h): 0.80 ± 0.08. After preparation with cryo-substitution cocktail I, the axon density was (PMI 18 h): 1.20 ± 0.12 for case 2, (PMI 20 h): 0.80 ± 0.08 for case 3, and (PMI 24 h): 0.49 ± 0.11 for case 4.