Accomplishments of “Old-Fashioned” Electron Microscopy in the Period of Dominance of Immunofluorescent Methods

Abstract

The goal of this review is to bring to the attention of the scientific community the opportunities of transmission electron microscopy for analyses of biological subjects and resolving complicated cases of data interpretation. Although procedures for electron microscopy are in general more elaborate (particularly for simultaneous immunolabeling of multiple antigens) compared to fluorescent microscopy, they can help view cellular morpho-functional features undetectable using other methods. In this review, we consider several unexpected and serendipitous discoveries made in our laboratory and fulfilled using unique opportunities provided by electron microscopy of ultrathin sections. We are deliberating the following topics: interpretation of unusual results of immunolabeling; a novel method for in situ identification of cells undergoing mitochondrial disorder and necrosis-like death; the sequence of organelles’ reorganization in dying cells; simultaneous rupture of nuclear and plasma membranes in migrating neurons; and the role of cytoskeleton in lateral expansion of the cerebral cortex.

1. Introduction

Nowadays, if a general biologist is asked “What is the main benefit of transmission electron microscopy?”, they will likely answer: “It has very high magnification and resolution”. Such an answer, while correct, does not capture the entire picture. Indeed, the focus on high resolution elicits another question that does not have a proper answer: “For what purpose do we need extremely high resolution if that which fluorescent microscopy provides is sufficient?” The main benefit of electron microscopy is not its unbeatable magnification and resolution but the freedom to analyze all the cells and organelles regardless of immunolabeling or any selective staining procedure. This provides an important advantage in comparison with most types of light microscopy, including modern confocal, multiphoton, and super-resolution methods that visualize only stained cells or molecules. As an advantage of immunofluorescent microscopy, we must highlight the relative simplicity to perform multiple antigen labeling (triple and more), whereas double labeling for transmission electron microscopy is much more elaborate. Nevertheless, electron microscopy combined with immunolabeling provides not only an opportunity to analyze the labeled cells and organelles but also to gain context from the surrounding structures. Specifically, this can help with the correct interpretation of immunolabeling or to identify misleading artifacts. A combination of electron microscopy with light microscopy for analysis of the same sample, so-called correlative microscopy, can be extremely useful by harnessing the advantages of both methods. Light microscopy can analyze a large area of tissue, whereas electron microscopy lets the researcher investigate chosen cells or tissue segments in detail. Here we review unexpected and serendipitous discoveries made in our laboratory using unique opportunities provided by electron microscopy combined with immunolabeling, light microscopy and three-dimensional (3D) reconstruction from serial ultrathin sections. In particular, we consider the following aspects: (1) double selectivity of antibodies producing unusual immunolabeling of mitochondria that helps for in situ identification of cells undergoing mitochondrial disorder and necrosis-like cell death; (2) the sequence of morpho-functional reorganization of mitochondria and Golgi apparatus in anoxia-exposed mouse embryo brains; (3) a novel mechanism of cell pathology identified through the simultaneous rupture of nuclear and plasma membranes in neurons migrating through tightly packed brain tissue; and (4) the role of centrosomes—the main cytoskeleton organizing center—in prolonged lateral expansion of rhesus macaque’s cerebral cortex.

2. Interpretation of Anti-Cannabinoid Type 1 Receptor Sera Binding with Mitochondria

The use of antibodies for immunohistochemical identification of proteins has resulted in striking advances in cellular and molecular research of complex biological systems. Nevertheless, there are several technical issues that need to be taken into consideration. In particular, the molecular construction of the antigen’s antibody-binding site—the epitope, may be conformational, i.e., composed of discontinuous sections of the amino acid sequence that are brought together through a secondary or tertiary protein structure. In such cases, precise analysis of the epitope is an extremely complicated and time-intensive procedure that is impractical to perform and has not been performed for thousands of commercially available sera currently used for scientific research [1,2,3,4,5]. As a result, immunolabeling of conformational epitopes may depend on the experimental conditions, making serological identification of proteins uncertain and prone to misinterpretation. For example, in our study of cannabinoid signaling in the mammalian brain, we detected binding of anti-cannabinoid type 1 receptor (CB₁R) sera with mitochondria of two morpho-functional types, whereas many other mitochondria stay immunonegative (Figure 1A–F and [6]). One type of immunopositive mitochondria, designated as “type 1”, contained immunoreaction end products on the outer membrane and inside the cristae. The other type, designated “type 2”, showed immunolabeling in the mitochondrial matrix. (See next chapters for discussion of probable functionality of these mitochondrial types.) At first glance, this might be interpreted as evidence of this G-protein-coupled receptor’s functionality in a new location in the mitochondria [7,8]. Mitochondrial immunolabeling takes place in parallel with well-documented immunolabeling of CB₁R in the axonal plasma membrane and intracellular vesicles in neuronal cell bodies confirming selectivity of the applied serum for detection of CB₁R (reviewed in [9,10,11,12,13,14,15]). However, negative control showed equal labeling of mitochondria in wild-type and CB₁R-knock-out (CB₁R^−/−) mice, demonstrating non-CB₁R binding of the antibodies (Figure 1G–I). After thorough analysis, we identified that, in parallel with CB₁R, the applied anti-CB₁R serum also binds to a conformational epitope of the mitochondrial protein stomatin-like protein-2 (SLP2) [6,16]. In accordance, several analyzed agonists and antagonists of CB₁R produced irregular effects on the respiratory activity of the isolated mitochondria, suggesting their direct action to the electron transport chain complexes or to the lipid components of the mitochondrial membranes rather than to the suspected cannabinoid receptor in the mitochondria [17]. Thus, the visible location and functionality of CB₁R in mitochondria has an alternative explanation. Overall, extra caution should be taken when interpreting unusual types of immunolabeling. In such cases, a proper application of electron microscopy can help.

3. Anti-CB₁R Serum Detects Disordered Mitochondria and Damaged Cells in Mouse Brain

In our light and electron microscopy analyses of CB₁R expression in the developing and adult mouse brain, we often utilized a combination of two visualization methods: (i) a highly sensitive method of immunoperoxidase reaction with nickel-intensified 3,3′-diaminobenzidine-4HCl (DAB-Ni) as a chromogen; and (ii) pre-embedding an ultra-small gold immunolabeling procedure with silver amplification that is less sensitive but provides more precise antigen location. This enabled us to detect two patterns of anti-CB₁R sera binding to mitochondria that likely indicate SLP2 rather than CB₁R. The best results were obtained with polyclonal antibodies raised against the last 31 amino acids of the C-terminus (L31) of CB₁R [6]. One population of the immunopositive mitochondria, designated as type 1, contained DAB-Ni immunoreaction end products on the outer membrane and inside the cristae (Figure 1A,B,H). The location of the antigen on the outer surface of the mitochondrial membrane, but not in the cristae, was confirmed using immunogold labeling (Figure 1C,D). The other type of immunopositive mitochondria, designated as type 2, contained the antigen in the matrix—also confirmed using immunogold labeling (Figure 1E,F,I). The specificity of these immunolabeling patterns is supported by our data showing that pre-absorption of the anti-CB₁R sera with the fusion peptide (L31) abrogated binding of the antibodies [6].

Systematic analysis of serial ultrathin sections paired with 3D reconstruction of cells and organelles considerably increases the scientific value of electron microscopy study of biological subjects [18,19]. For example, modern automatic systems for volume electron microscopy analysis (e.g., FIB-SEM—Focused Ion Beam Scanning Electron Microscopy) are very effective but require expensive specialized equipment. In our studies, we performed 3D reconstructions using universally applicable transmission electron microscopy methods and publicly available Reconstruct software version 1.1.0.0 [20,21,22], which do not require additional expenses. Application of 3D reconstruction enabled us to demonstrate that type 1 immunopositive mitochondria had, on average, the same length and diameter as immunonegative mitochondria [16]. In contrast, the type 2 mitochondria demonstrated a variety of ultrastructural pathologies, ranging from a slightly increased diameter to distinct swelling and lysis of the inner compartment (Figure 2). Notably, emergence of the type 2 (but not the type 1) mitochondria is linked to disorder of the cells. In the normal developing mouse brain, we detected sporadic spots of neuropil containing immunopositive type 2 mitochondria, while surrounding tissue was immunonegative (Figure 2A,B). Electron microscopy analysis revealed that some of the type 2 mitochondria-containing cells demonstrated minor ultrastructural pathology, namely, slightly swollen mitochondria (regardless of their immunopositive or negative status) as well as swollen nuclear membranes and endoplasmic reticulum cisterns (Figure 2C,D). Other cells showed severe ultrastructural pathology, nearing complete degradation of cytoplasm (Figure 2E–H). Regardless of the degree of damage in the type 2 mitochondria-containing cells, adjacent cells containing only immunonegative mitochondria demonstrated an ultrastructure characteristic of normal developing neurons. Thus, the observed phenotype of the type 2 mitochondria provides a new target for in situ diagnosis of cellular degeneration.

4. What CB₁R-L31 Immunolabeling Reveals About Mitochondrial Function

Owing to its central role in energy supply and several other aspects of cellular physiology and regulated cell death, the mitochondria have been heavily studied (reviewed in [23,24,25,26]). This has led to many immunological and biochemical methods for assessing their functionality and dysregulation. However, most current methods are applicable only to cultured cells or require organelle fractionation, whereas there is a lack of histochemical approaches to detect disordered mitochondria in intact organs. Our findings on immunohistochemical detection of SLP2 may fill this void and offer new insights into morpho-functional alterations of mitochondria in situ. SLP2 is a constitutive mitochondrial protein that is incorporated in the inner mitochondrial membrane and plays a role in mitochondrial functionality interacting with phospholipids and prohibitins, or creating the hetero-oligomer complex with mitofusin 2 [27,28,29,30,31]. According to a recent publication, SLP2 is attached to the inner mitochondrial membrane facing the mitochondrial matrix. SLP2 plays a role in the anchoring and spatial organization of the protease complex composed of the rhomboid protease PARL and the i-AAA protease YME1L in the inner mitochondrial membrane. SLP2, together with the protease complex, regulates mitochondrial dynamics, quality control, and cell survival [32].

Because only one unique CB₁R-immunopositive band was visible in our Western blot analysis of mitochondrial fractions [6], we hypothesized that SLP2 is detectable in both type 1 and type 2 mitochondria. Although SLP2 is a constitutive mitochondrial protein and likely expressed in all mitochondria, CB₁R-L31 serum detected it in a relatively small number of mitochondria. We hypothesized that the previously demonstrated tight contacts between SLP2 and other molecules [27,28,29,32] block its epitope in the functional mitochondria from binding with CB₁R-L31 antibodies as well as with currently available commercial SLP2 antibodies. This explains why most mitochondria with normal morphology are immunonegative (Figure 3). Release of SLP2 from tight contacts with other molecules frees it up for interaction with CB₁R-L31 antibodies in type 1 and type 2 mitochondria. Type 1 mitochondria, which show the ultrastructure characteristic for normal functionality, contain immunolabeling end products on the outer membrane and inside the cristae, of which only the outer membrane location was confirmed through immunogold labeling (Figure 1A–D). Such an immunolabeling pattern likely indicates that new SLP2 molecules synthesized in cytoplasm are docked on the mitochondrion surface for subsequent transportation inside the organelle (Figure 3). DAB-Ni depositions may diffuse inside the crista through the pores or tiny perforations in the outer mitochondrial membrane. If so, DAB-Ni depositions in the crista represent an artifact and should be ignored. In general, type 1 SLP2-like immunopositive mitochondria do not differ from normal immunonegative mitochondria. In contrast, the anti-CB₁R-L31 serum labels matrix of the type 2 mitochondria that are either moderately or dramatically swollen (Figure 2 and Figure 3). This corresponds with the modern view on the location of SLP2 on the inner surface of the inner mitochondrial membrane [32]. In such a way, anti-CB₁R-L31 labeling may detect cells with early or advanced stages of pathology. Some mitochondria, although dramatically swollen, remain immunonegative—likely a result of SLP2’s denaturation destroying the epitope for CB₁R-L31 antibodies (Figure 2 and Figure 3). Although the hypothetical complex of SLP2 and other mitochondrial proteins is likely one of numerous incompletely characterized mitochondrial complexes, many of them might behave analogically in disordering conditions. Thus, immunolabeling of SLP2 released from its complex may serve a unique role as a marker of general mitochondrial dysfunction and cellular disorder.

5. Using Type 2 Mitochondria for Identification of Necrosis-like Cellular Ultrastructural Pathology in Mouse Brain

Immunohistochemical labeling of SLP2 with anti-CB₁R-L31 serum represents a convenient method for identification of disordered cells because the type 2 mitochondria are well detectable with light microscopy and can be a subject of quantitative analysis for characterizing cellular functionality in certain experimental conditions. For example, we sporadically observed the type 2 mitochondria in developing neurons and endothelial cells of generally healthy mouse embryo brains [6]. We also detected that emergence of the type 2 mitochondria in certain cells correlated with the necrosis-like ultrastructural pathology of these cells, while adjacent cells containing immunonegative mitochondria showed normal ultrastructure. Anoxic conditions applied to mouse embryo brains, as well as hypoxia–ischemia of adult mice provoked a marked increase in the type 2 mitochondria in the neuropil in parallel with massive necrosis-like cell death [16]. In particular, we analyzed CB₁R-L31 immunolabeling in isolated mouse embryo brains that were exposed to anoxic conditions through immersion in a saline solution. As a control, we used isolated mouse embryo brains immersed in the saline solution bubbled with oxygen. In anoxic conditions, a marked increase in the type 2 immunopositive mitochondria in all brain segments was observed after 4.5 and 6 h of incubation, whereas their numbers were reduced in the oxygenated group (Figure 4A). The low number of immunopositive mitochondria and generally normal cellular ultrastructure in ≤3 h of anoxia may reflect mammalian embryos’ high resistance to hypoxia [34,35,36]. We also detected that the numbers of type 2 mitochondria are increased at high temperatures (40 °C) and decreased at low temperature (33 °C) in comparison with the physiological temperature (37 °C) during anoxia (Figure 4B). Consistently, emergence of the type 2 mitochondria was unequivocally linked to the necrosis-like ultrastructural pathology of neurons (Figure 4C–K). Importantly, ultrastructural evidence of other cell death mechanisms, such as apoptosis or autophagy, was not encountered in the embryo brains exposed to anoxia. This indicates that the anoxia-induced necrosis-like cell death mechanism is virtually undetectable using biochemical or immunochemical methods such as terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay, caspase immunolabeling, or labeling of autophagy markers. Our findings also provide evidence that the mechanism of hypothermia therapy of neonatal hypoxic–ischemic encephalopathy [37,38,39,40] is based on suppression of necrosis rather than apoptosis. Apoptosis results in controlled cell shrinkage and fragmentation via caspase activity, as well as an anti-inflammatory cytokine release. In contrast, necrosis signals (e.g., RIP1 and RIPK1) lead to cell swelling, lysis, and a pro-inflammatory cytokine release that diffuses through the surrounding tissue [41,42]. Necrosis represents the most destructive cell-death pathway, whereas apoptosis locally eliminates damaged cells and may serve to protect the tissue and improve the survival of the organism. Thus, immunolabeling of the type 2 mitochondria serves as a target for detecting derangement of mitochondria and necrosis-like cell death in situ. Anti-CB₁R-L31 serum provides a unique opportunity for selective visualization of disordered mitochondria. We suggest using this serum because none of the currently available SLP2 antibodies provide immunohistochemical labeling of mitochondria.

6. Disorder of Golgi Apparatus Precedes Anoxia-Induced Pathology of Mitochondria

We applied electron microscopy with 3D reconstruction to quantitatively analyze the mitochondria and other organelles during the first hours of anoxia. Although our quantitative study showed an approximately 3 h long delay for type 2 mitochondria upregulation (Figure 4A), we detected reshaping of the Golgi apparatus (GA) after only 1 h of anoxia. Thus, GA disorder is the earliest morpho-functional reaction of mouse embryo brains to anoxic conditions. In particular, GA showed concentrical swirling of the cisternae forming a spheroid with the trans-cisterna in the center of the sphere that contrasted with the usual stack of cisternae in normally functioning GA (Figure 5 and [33]). Similar reshaping of GA was described in mammalian cells at several pathological conditions and named onion-like GA [43,44]. Such disturbance of GA likely interferes with its function for post-translational protein modification and secretory trafficking [45]. It is characteristic that the onion-like GA phenotype was observed before other organelles, including mitochondria, showed ultrastructural disorder. Analyses after 3 h of anoxia confirmed swirling in almost 100% of GA and swelling in the majority of mitochondria. At this time point, the length of the mitochondria was significantly decreased, and the diameter was increased so the mitochondria took on a nearly spherical shape [33].

We also found that the reaction of GA to anoxia may be reversible in the case of fast reoxygenation. After 1 h of anoxia with subsequent return of the embryo brains to the oxygenated liquid, we observed a recovery of normal ultrastructure of most GA. Namely, in the cerebral cortex, only 3 GA among a total of 71 identified showed the onion-like phenotype, and only 8 out of 50 GA were swirled in the lateral ganglionic eminence (Figure 6A). Other organelles, including mitochondria, also demonstrated normal ultrastructure, although we did not perform 3D and morphometric analysis after 1 h anoxia/reoxygenation. In contrast, reoxygenation after 3 h of anoxia does not reverse the damage of the cells. In particular, among 29 GA identified in the cerebral cortex, 27 showed the onion-like phenotype. Analogically, in the lateral ganglionic eminence, no normal GA were identified, while 29 GA showed the onion-like phenotype (Figure 6A). A morphometric analysis of mitochondria in the cerebral cortex after 3 h of anoxia with subsequent reoxygenation showed a minor, although statistically significant, increase in the average length and a decrease in the average diameter of the mitochondria in comparison with the 3 h anoxia group. At the same time, the ranges of the length and the diameter (measured as the standard deviations of the means) in the reoxygenation group increased (Figure 6B). This indicates heterogeneity in mitochondrial reactions. While some mitochondria return to healthy morpho-functional characteristics, others may be disrupted further. As a result, we conclude that a majority of GA can reestablish morpho-functional characteristics if the oxygen supply returns after 1 h of anoxia when the mitochondria show a normal ultrastructure. Reoxygenation does not allow the GA ultrastructure to recover after 3 h of anoxia when the mitochondrial disruption is significant and irreversible.

According to the observed normal ultrastructure of mitochondria after 1 h of acute anoxia, oxygen deficit is likely to reduce oxidative phosphorylation and ATP production without a morphologically detectable disruption in the energy-production machinery. Nevertheless, a lack of ATP apparently provokes the observed ultrastructural pathology and dysfunction of the GA. Such dysfunction may be reversible if the oxygen supply returns soon. In case the energy deficit continues long enough, the GA malfunction may damage other organelles, including mitochondria. In turn, damaged mitochondria may initiate numerous irreversible reactions, such as reactive oxygen species production, calcium leakage, cytochrome C leakage, cell apoptosis or necrosis, etc. [23,46,47]. Thus, the molecular architecture of mitochondria, at least in mouse embryo brains, is resistant to anoxia for approximately 1 h, whereas intracellular transportation and other functions executed by the GA are vulnerable to a shortage of ATP supply and likely participate in subsequent cellular disruption, including massive fission and swelling of mitochondria that were detected after 3 h of anoxia. During the next step of disorder, after 4.5 h of anoxia, we detected the exponential upregulation of the type 2 mitochondria, which likely indicates dissociation of mitochondrial protein complexes, making the mitochondria irreversibly unfunctional. Thus, a dramatic increase in the quantity of type 2 mitochondria likely signifies the point of no return for the necrosis-like death of cells in the mouse brain.

7. Type 2 Mitochondria Identify Necrosis-like Cell Death in the Liver and White Adipose Tissue

We investigated if CB₁R-L31 serum can be used for identification of disordered mitochondria in other tissues and organs. We chose mouse liver and white adipocytes because these cells are morpho-functionally distinct from brain cells. In the liver from postnatal mice, we very rarely observed SLP2-like type 2 immunopositive mitochondria in five analyzed young (11-day-old) mice, whereas they were numerous in two analyzed aged (16-month-old) mice. Our correlative light/electron microscopy investigation confirmed that immunopositive dots visible in light microscopy represent accumulations of DAB-Ni immunoperoxidase reaction end product in the mitochondrial matrix (Figure 7) as we observed in the brain [6,16]. It is characteristic that immunopositive mitochondria concentrate in distinct hepatocytes, while they are absent from adjacent cells. In the liver of the aged mice, like in the developing brain, the type 2 mitochondria are linked to mitochondrial fission and swelling paralleled with destruction of other organelles showing necrosis-like cellular ultrastructural pathology. Adjacent immunonegative cells contain long-branched mitochondria and preserve the normal ultrastructure of functional hepatocytes (Figure 7).

Using anti-CB₁R-L31 serum and correlative light/electron microscopy, we also detected SLP2-like immunopositive type 2 mitochondria in the abdominal white adipose tissue from adult mice (N = 6). Although the main body of the adipose tissue remains immunonegative, we detected immunopositive granules in the cellular periphery of many adipocytes. Electron microscopy confirmed that a DAB-Ni immunoreaction end product was accumulated in the matrix of spherically shaped swollen mitochondria (Figure 8). Like in the brain and liver, emergence of the type 2 mitochondria in the adipocytes was accompanied by fission and swelling of all the mitochondria and disorder of other organelles and cytoplasm. Thus, immunohistochemical labeling using CB₁R-L31 serum identifies a necrosis-like ultrastructural pathology of white adipocytes. At the same time, the immunonegative compartments of the adipose tissue preserve normal organelles and long-branching mitochondria characteristic of functional adipocytes. Distribution of the immunopositive adipocytes, their relative number and the reason for their appearance are currently unclear. Overall, our application of CB₁R-L31 serum for brain, liver and adipose tissue indicates that immunohistochemical detection of SLP2-like immunopositive type 2 mitochondria can serve as a marker for identification of a necrosis-like ultrastructural pathology and cell death in different tissues and organs.

8. Dissociation of Mitochondrial Protein Complexes Identifies Necrosis-like Cell Death

Identification of disordered mitochondria in different organs and cell types suggests the universal character of SLP2 immunolabeling and wide applicability of anti-CB₁R-L31 serum for identification of malfunctional mitochondria and the disordered cells that contain them. Because many mitochondrial protein complexes likely behave similarly in pathological conditions, SLP2 released from the complex with other mitochondrial proteins may serve as a novel immunohistochemical marker of mitochondrial dysfunction and cellular disorder. The mitochondrial and cellular disorders identified here occur sporadically in normal brains and other organs and ubiquitously in hypoxia-exposed developing and adult mouse brains. We propose using anti-CB₁R-L31 serum for in situ identification of dissociating mitochondrial protein complexes which likely indicate a point of no return during necrosis-like cell death (Figure 9). Although we do not know the epitope of antibody binding, our findings open up the possibility of using anti-CB₁R-L31 serum as a tool for immunohistochemical exploration of the role of SLP2 in mitochondria under normal and pathological conditions. In our hands, the method was effective for different types of cells and tissues, such as central nervous and vascular systems, hepatocytes and white adipocytes. The observed mitochondrial phenotype provides a new target for diagnosis of necrosis-like cell degeneration in different organs. Our original experimental approach partially covers the gap in the contemporary methodical arsenal complementing popular detection of nuclear and cytoplasmic reactions (e.g., TUNEL assay and active caspase immunolabeling) that reveal the apoptosis-like regulated cell death.

9. Piercing Nuclear Hernia Is a Novel Ultrastructural Pathology of Migrating Cells

Recently, we applied electron microscopy of ultrathin sections for complete 3D reconstructions of nuclei and cell bodies for analysis of neurons from CB₁R^−/− and wild-type mouse embryos exposed to CB₁R agonists. We documented local breakups of the inner and outer nuclear membranes accompanied by herniation of chromatin into the cytoplasm (Figure 10 and [53]). Similar cellular disorders named nuclear envelope (NE) ruptures were previously identified in developing neurons and metastatic cancer cells [54,55,56,57,58]. Surprisingly, in a fraction of migrating neurons, streams of herniated chromatin ruptured not only the NE but also the plasma membrane, exposing cytoplasm and nucleoplasm to the intercellular space (Figure 11). As it was the first description of a simultaneous rupture of the NE and the plasma membrane, we named this cellular pathology “piercing nuclear hernia” (PNH).

Rupture of the nuclear and plasma membranes in cases of PNH severely disrupts the barrier function of the membranes and obviously results in fast death of the cell that should be qualified as regulated cell death (RCD; see recommendations of the Nomenclature Committee on Cell Death [59]). Belonging to a divergent group of RCD mechanisms is defined as a consequence of the fall of adaptive responses to environmental perturbations [59]. Numerous mechanisms of RCD can be crudely classified into three complex groups: apoptosis-, autophagy-, and necrosis-type cell-death mechanisms [41,42,59,60,61,62,63,64,65]. In contrast to RCD, cells exposed to extreme physical, chemical, or mechanical stimuli may immediately lose their structural integrity and die in an uncontrollable manner termed “accidental cell death” [59]. Cases of accidental cell death are virtually undetectable using biochemical or immunochemical methods suitable for RCD, such as TUNEL assay, active caspase immunolabeling, or labeling of autophagy markers. Electron microscopy, on the other hand, visualizes organelles regardless of their functional conditions and can detect dead cells or cellular remnants. Accordingly, we found that the emergence of PNH in migrating neurons is often paralleled with an ultrastructural pathology of the organelles. In particular, cells with PNH demonstrated short mitochondria, while adjacent non-herniated cells showed normally sized mitochondria (Figure 12). A reduction in mitochondrial length—predominance of fission over fusion—may serve as evidence of disordered cellular energetics or other malfunctions [23,25,26,48,49,50,51,66]. Involvement of disordered endocannabinoid signaling in the PNH emergence (see below) indicates belonging to RCD mechanisms. On the other hand, the abrupt loss of the barrier function of the nuclear and plasma membranes may result in fast death of the cell, i.e., suggests the executive mechanism of cell death that resembles the accidental cell death. Thus, PNH represents a novel type of cell pathology that is likely lethal for the cell. After extensive study, the procedure of PNH upregulation may be useful for inducing breaks of the plasma membrane and death of hazardous cells, for example, migrating metastatic tumor cells (see [67] for discussion).

10. Cannabinoid Signaling Plays a Role in Cytoskeleton Functionality

We showed that the probability of chromatin herniation is increased in cases of disorder of the endocannabinoid system [53]. Indeed, numerous migrating neurons in CB₁R^−/− mouse embryos and wild-type embryos exposed to different agonists of CB₁R show NE ruptures or PNHs (Figure 13). This indicates that deviations from optimal functioning of the endocannabinoid system in under- or over-activity may trigger mechanisms increasing the membrane’s vulnerability and chromatin herniation.

The large volume of herniated chromatin [53] indicates that NE ruptures and PNHs in nuclei moving through the developing cerebrum are consequences of extremely high intranuclear pressure. In accord, the commonly accepted mechanism of NE rupture includes increased intranuclear pressure and cytoskeleton malfunction [54,56,68,69,70,71,72,73,74,75]. A link between the endocannabinoid system disorder and NE ruptures and PNHs provides evidence that optimal endocannabinoid signaling is crucial for cytoskeleton functionality in migrating neurons. Maintenance of the shape of mature neurons, as well as, their reorganization during migration are determined by numerous cytoskeleton components, such as microtubules, neurofilaments, actin, myosin II, centrosomes, and others [76,77,78,79,80,81,82,83,84,85,86]. We recently reviewed the possible role of the endocannabinoid system in actomyosin cytoskeleton functionality in developing and mature neurons from different species [53,67]. Overall, deviated expressions of the cytoskeletal components after stimulation or knock-out of CB₁R establish evidence of cannabinoid signaling participation in neuronal cytoskeleton consolidation. Our discovery of PNH in migrating neurons gives us an opportunity to study enigmatic molecular mechanisms of the endocannabinoids’ action in the developing brain.

11. Role of the Cytoskeleton in Lateral Expansion of the Cerebral Cortex

Initiating study of mitotic cells in the cerebral ventricular zone, we expected to find a role of cannabinoid signaling in neurons’ proliferation. However, during the investigation, we found no evidence of the cannabinoid receptors’ expression in dividing cells in mouse and rhesus macaque cerebral cortices [87]. Instead, we discovered another aspect of cytoskeleton function. We used the recognizable ultrastructure of centrosomes, which play a key role as a microtubule organizing center [82,84,86], for quantitative morpho-functional characterization of the cytoskeleton. We identified significant species-specific differences in the proportion of anchored versus free centrosomes in dividing apical neural progenitor cells (NPCs). Anchorage of the mother centriole to the apical membrane of NPCs is a fundamental part of repetitive circles of cell divisions in the apical region of the developing telencephalon [82,84,86,88,89,90]. In contrast, baso-lateral anchorage of the mother centriole predisposes the cell for delamination from the apical surface and its basal migration, after which the cell may continue proliferation as a basal NPC but does not participate in lateral expansion of the prospective cerebral cortex [91,92,93]. In particular, we analyzed positions of centrosomes in the apical NPCs shortly after mitosis [94]. We showed that mother centrioles may be anchored to the plasma membrane basally relative to the adherens junction belt (not to the apical segment of the plasma membrane) or stay free in cytoplasm (Figure 14 and Figure 15). Free post-mitotic centrosomes preserve the chance to anchor either baso-laterally or apically [95]. The apical position (below the adherens junction belt) provides a faster link between the daughter cell and the ventricular surface, thus increasing the probability of locating the cell at the ventricular surface as an apical NPC during the next cell cycle [91,96,97,98,99]. Quantification of anchored and free centrosomes showed characteristic differences between the species. In mice, 66.3% of centrosomes were anchored (namely, we identified 86 anchored and 57 free centrosomes), whereas in monkeys, only 31.8% centrosomes were anchored to the baso-lateral membrane (27 anchored centrosomes and 58 free ones) [94]. In mice, anchored centrosomes were more numerous than free ones in all embryonic age groups studied. The ratio of anchored versus free cytoplasmic centrosomes was much lower in the macaque embryos (Figure 16). The dynamics of centrosome anchorage might represent a mechanism that impacts transformation of apical NPCs into basal NPCs. In particular, the increased apical anchorage of NPCs may prolong apical neurogenesis and lateral expansion of the cerebral cortex. Thus, our findings suggest a role of centrosomes in the maintenance of ventricular neurogenesis that may increase the size of the cerebral cortex in rhesus macaques and other gyrencephalic animals. Our observations afford evidence strengthening the radial unit hypothesis of cerebral expansion [100,101,102,103,104,105,106,107]. The hypothesis proposes that an increased rate of ventricular proliferation produces a larger number of radial units, resulting in an expanded cellular sheet in the cerebrum, which later buckles and transforms a lissencephalic cerebral cortex into a gyrencephalic one. This is an extremely important feature because the size of the cerebral cortex is the basis of high cognitive abilities, culminating in humans’ capacity for thinking and language [108,109,110,111,112,113,114,115,116].

12. Conclusions

In this article, we reviewed selected studies from our laboratory which highlight unique opportunities presented by electron microscopy of ultrathin sections.

In particular, our comprehensive study of CB₁R immunolabeling led to the discovery of double selectivity of anti-CB₁R-L31 serum, which in addition to CB₁R, also recognized the mitochondrial protein SLP2. This finding turned out to be instrumental for in situ identification of mitochondrial protein complexes’ dissociation in cells undergoing necrosis-like death. Such dying cells are undetectable using currently available methods for identification of apoptosis and autophagy. Application of this novel method combined with transmission electron microscopy let us analyze the time course of the morphological disorder of organelles at pathological conditions. For example, we observed that Golgi apparatus cisternae swirling preceded anoxia-induced disorder of other organelles, including mitochondria, in mouse embryo brains. Such a Golgi apparatus disorder was reversible in the case of quick reoxygenation.

Using electron microscopy of serial ultrathin sections enabled our unequivocal identification of a new type of ultrastructural cell pathology. Namely, we identified chromatin streams that herniated from the nucleus and ruptured through both the nuclear envelope and the plasma membrane. We named this cellular pathology “piercing nuclear hernia”. Moreover, we showed that abnormalities of endocannabinoid signaling upregulated ruptures of the nuclear and plasma membranes in migrating cerebral neurons. This grants a new look at as-of-yet unknown mechanisms of marijuana’s effects on the human fetus brain.

High resolution of transmission electron microscopy enabled us to identify the characteristic locations of extremely small organelles such as centrosomes, which serve as the main cytoskeleton organizing center during and after cellular division. We found that lateral expansion of the developing cerebral cortex correlates with a species-specific ratio of free versus anchored centrosomes in the apical neural progenitor cells shortly after mitosis. In mice, centrosomes showed a high rate of anchorage to the plasma membrane. In rhesus macaques, in contrast, prolonged free cytoplasmic location of the centrosomes provides more opportunity for lateral expansion of the cerebral cortex—the critical basis of primates’ cognitive abilities.

Despite the development of new morphological methods and improved access to equipment for super-resolution fluorescent microscopy and volume electron microscopy, transmission electron microscopy of ultrathin sections provides opportunities for multiple unique experimental approaches. We hope that the above examples of successful studies will encourage young and experienced scientists to utilize classical methods of transmission electron microscopy.