The highly complex structure of the brain requires an approach that can unravel its connectivity. Using volume electron microscopy and a dedicated software we can trace and measure all nerve fibers present within different samples of brain tissue. With this software tool, individual dendrites and axons are traced, obtaining a simplified "skeleton" of each fiber, which is linked to its corresponding synaptic contacts. The result is an intricate meshwork of axons and dendrites interconnected by a cloud of synaptic junctions. To test this methodology, we apply it to the stratum radiatum of the hippocampus and layers 1 and 3 of the somatosensory cortex of the mouse. We find that nerve fibers are densely packed in the neuropil, reaching up to 9 kilometers per cubic mm. We obtain the number of synapses, the number and lengths of dendrites and axons, the linear densities of synapses established by dendrites and axons, and their location on dendritic spines and shafts. The quantitative data obtained through this method enable us to identify subtle traits and differences in the synaptic organization of the samples, which might have been overlooked in a qualitative analysis.

ABSTRACT Unraveling the highly complex connectivity at the synaptic level is critical for the understanding of brain function. However, investigating the structure of the brain is time consuming and technically demanding. We have developed and applied a tracing method that allows us to obtain a simplified or skeletonized version of all nerve fibers present within a given brain sample, without the need for full 3D reconstructions. Every tracing or “skeleton” is linked to its corresponding synaptic contacts, so the result is an intricate meshwork of axons and dendrites interconnected by a cloud of synaptic junctions. The tissue sample is obtained by volume electron microscopy (FIB-SEM) and tracing is performed by dedicated software. In addition, the software provides the length of each skeleton as well as the number and positions of its corresponding synapses. With this approach, we have identified clear as well as subtle quantitative differences between three brain regions (the stratum radiatum of the hippocampus and layers 1 and 3 of the somatosensory cortex). For example, we have found that axons and dendrites are densely packed in the neuropil of these three regions, reaching close to 8 kilometers of nerve fibers per cubic mm in layer 1. In all cases, the predominant fibers were excitatory axons, followed by spiny dendrites, inhibitory axons and smooth dendrites. Other examples include the fact that the excitatory axons in the stratum radiatum establish many more synapses per micron than the axons of the other regions, or the lack of correlation between the number of synapses established on the spines of a given dendrite and the number of synapses established on the dendritic shaft. Using this methodology, we obtain a set of quantitative parameters including the number and proportions of excitatory and inhibitory synapses; the numbers of dendrites and axons; the proportion of excitatory and inhibitory axons; the linear densities of synapses established on dendritic spines and dendritic shafts; the linear densities of synapses established by excitatory and inhibitory axons; and the lengths of the different types of dendrites and axons. Taken together, these data characterize each region, allowing us to compare them on a quantitative basis and to identify subtle traits and differences that would have escaped a qualitative analysis. With this strategy, we can achieve a better understanding of the structure and dynamics of the brain, and we are better equipped to explore how they change during both normal and pathological conditions.

Dravet syndrome caused by SCN1A variants is a severe developmental epileptic encephalopathy (DEE) characterized by pharmaco-resistant epileptic seizures and progressive neurodevelopmental decline with cognitive impairment and autism-spectrum-traits. Numerous preceding studies indicate that the initial pathophysiology due to impaired Na V 1.1 function mainly derives from reduced interneuron firing leading to a network hyperexcitability (Bender et al. 2012). However, little is known how epileptogenesis and generally disease pathogenesis progress from the inborn molecular defect to infantile seizure onset. We address this question in a Dravet mouse model by comprehensive single-cell RNA sequencing and selected downstream analysis via single-cell electrophysiology, histology, live cell imaging and electron microscopy. Our data reveal a continuum of early primary (preseizure) and secondary (post-seizure onset) transcriptomic changes in various cell populations in the hippocampus. Focusing on cornu ammonis , we find a number of transcriptional pathways that are dysregulated including synaptic transmembrane adhesion molecules of the neurexin superfamily and voltage-gated ion channels. Further investigations support an ultrastructural and functional axonopathy and synaptopathy of parvalbumin interneurons. These processes precede somatic firing impairment and seizures suggesting they underlie fundamental early-phase disease mechanisms. Taken together we provide a cellularly resolved transcriptomic resource of early disease phases of Dravet syndrome and demonstrate epileptogenesis beyond Na V 1.1 loss-of-function during an early developmental time window of CNS maturation. Altogether these data establish proof-of principle that the concept of epileptogenesis, originally devised for acquired forms of epilepsy, similarly applies to genetic epilepsies and DEEs.

The structural complexity of nervous tissue makes it very difficult to unravel the connectivity between neural elements at different scales. Numerous methods are available to trace long-range projections at the light microscopic level, and to identify the actual synaptic connections at the electron microscopic level. However, correlating mesoscopic and nanoscopic scales in the same cell, cell population or brain region is a problematic, laborious and technically demanding task. Here we present an effective method for the 3D reconstruction of labeled subcellular structures at the ultrastructural level, after single-neuron labeling in fixed tissue. The brain is fixed by intracardial perfusion of aldehydes and thick vibratome sections (250 μm) are obtained. Single cells in these vibratome sections are intracellularly injected with horseradish peroxidase (HRP), so that the cell body and its processes can be identified. The thick sections are later flat-embedded in epoxy resin and re-sectioned into a series of thinner (7 μm) sections. The sections containing the regions of interest of the labeled cells are then imaged with automated focused ion beam milling and scanning electron microscopy (FIB-SEM), acquiring long series of high-resolution images that can be reconstructed, visualized, and analyzed in 3D. With this methodology, we can accurately select any cellular segment at the light microscopic level (e.g., proximal, intermediate or distal dendrites, collateral branches, axonal segments, etc.) and analyze its synaptic connections at the electron microscopic level, along with other ultrastructural features. Thus, this method not only facilitates the mapping of the synaptic connectivity of single-labeled neurons, but also the analysis of the surrounding neuropil. Since the labeled processes can be located at different layers or subregions, this method can also be used to obtain data on the differences in local synaptic organization that may exist at different portions of the labeled neurons.

Determining the number of synapses that are present in different brain regions is crucial to understand brain connectivity as a whole. Membrane-associated guanylate kinases (MAGUKs) are a family of scaffolding proteins that are expressed in excitatory glutamatergic synapses. We used genetic labeling of two of these proteins (PSD95 and SAP102), and Spinning Disc confocal Microscopy (SDM), to estimate the number of fluorescent puncta in the CA1 area of the hippocampus. We also used FIB-SEM, a three-dimensional electron microscopy technique, to calculate the actual numbers of synapses in the same area. We then estimated the ratio between the three-dimensional densities obtained with FIB-SEM (synapses/µm3) and the bi-dimensional densities obtained with SDM (puncta/100 µm2). Given that it is impractical to use FIB-SEM brain-wide, we used previously available SDM data from other brain regions and we applied this ratio as a conversion factor to estimate the minimum density of synapses in those regions. We found the highest densities of synapses in the isocortex, olfactory areas, hippocampal formation and cortical subplate. Low densities were found in the pallidum, hypothalamus, brainstem and cerebellum. Finally, the striatum and thalamus showed a wide range of synapse densities.

Thalamocortical posterior nucleus (Po) axons innervating the vibrissal somatosensory (S1) and motor (MC) cortices are key links in the brain neuronal network that allows rodents to explore the environment whisking with their motile snout vibrissae. Here, using fine-scale high-end 3D electron microscopy, we demonstrate in adult male C57BL/6 wild-type mice marked differences between MC versus S1 Po synapses in (1) bouton and active zone size, (2) neurotransmitter vesicle pool size, (3) distribution of mitochondria around synapses, and (4) proportion of synapses established on dendritic spines and dendritic shafts. These differences are as large, or even more pronounced, than those between Po and ventro-posterior thalamic nucleus synapses in S1. Moreover, using single-axon transfection labeling, we demonstrate that the above differences actually occur on the MC versus the S1 branches of individual Po cell axons that innervate both areas. Along with recently-discovered divergences in efficacy and plasticity, the synaptic structure differences reported here thus reveal a new subcellular level of complexity. This is a finding that upends current models of thalamocortical circuitry, and that might as well illuminate the functional logic of other branched projection axon systems.SIGNIFICANCE STATEMENT Many long-distance brain connections depend on neurons whose branched axons target separate regions. Using 3D electron microscopy and single-cell transfection, we investigated the mouse Posterior thalamic nucleus (Po) cell axons that simultaneously innervate motor and sensory areas of the cerebral cortex involved in whisker movement control. We demonstrate significant differences in the size of the boutons made in each area by individual Po axons, as well as in functionally-relevant parameters in the composition of their synapses. In addition, we found similarly large differences between the synapses of Po versus ventral posteromedial thalamic nucleus axons in the whisker sensory cortex. Area-specific synapse structure in individual axons implies a new, unsuspected level of complexity in long-distance brain connections.

The correlation between dysfunction in the glutamatergic system and neuropsychiatric disorders, including schizophrenia and autism spectrum disorder, is undisputed. Both disorders are associated with molecular and ultrastructural alterations that affect synaptic plasticity and thus the molecular and physiological basis of learning and memory. Altered synaptic plasticity, accompanied by changes in protein synthesis and trafficking of postsynaptic proteins, as well as structural modifications of excitatory synapses, are critically involved in the postnatal development of the mammalian nervous system. In this review, we summarize glutamatergic alterations and ultrastructural changes in synapses in schizophrenia and autism spectrum disorder of genetic or drug-related origin, and briefly comment on the possible reversibility of these neuropsychiatric disorders in the light of findings in regular synaptic physiology.

Multivesicular bodies (MVBs) are membrane-bound organelles that belong to the endosomal pathway. They participate in the transport, sorting, storage, recycling, degradation, and release of multiple substances. They interchange cargo with other organelles and participate in their renovation and degradation. We have used focused ion beam milling and scanning electron microscopy (FIB-SEM) to obtain stacks of serial sections from the neuropil of the somatosensory cortex of the juvenile rat. Using dedicated software, we have 3D-reconstructed 1618 MVBs. The mean density of MVBs was 0.21 per cubic micron. They were unequally distributed between dendrites (39.14%), axons (18.16%), and nonsynaptic cell processes (42.70%). About one out of five MVBs (18.16%) were docked on mitochondria, representing the process by which the endosomal pathway participates in mitochondrial maintenance. Other features of MVBs, such as the presence of tubular protrusions (6.66%) or clathrin coats (19.74%) can also be interpreted in functional terms, since both are typical of early endosomes. The sizes of MVBs follow a lognormal distribution, with differences across cortical layers and cellular compartments. The mean volume of dendritic MVBs is more than twice as large as the volume of axonic MVBs. In layer I, they are smaller, on average, than in the other layers.

Knowing the proportions of asymmetric (excitatory) and symmetric (inhibitory) synapses in the neuropil is critical for understanding the design of cortical circuits. We used focused ion beam milling and scanning electron microscopy (FIB/SEM) to obtain stacks of serial sections from the six layers of the juvenile rat (postnatal day 14) somatosensory cortex (hindlimb representation). We segmented in three-dimensions 6184 synaptic junctions and determined whether they were established on dendritic spines or dendritic shafts. Of all these synapses, 87-94% were asymmetric and 6-13% were symmetric. Asymmetric synapses were preferentially located on dendritic spines in all layers (80-91%) while symmetric synapses were mainly located on dendritic shafts (62-86%). Furthermore, we found that less than 6% of the dendritic spines establish more than one synapse. The vast majority of axospinous synapses were established on the spine head. Synapses on the spine neck were scarce, although they were more common when the dendritic spine established multiple synapses. This study provides a new large quantitative dataset that may contribute not only to the knowledge of the ultrastructure of the cortex, but also towards defining the connectivity patterns through all cortical layers.

Thalamocortical synapses from "lemniscal" neurons of the dorsomedial portion of the rodent ventral posteromedial nucleus (VPMdm) are able to induce with remarkable efficacy, despite their relative low numbers, the firing of primary somatosensory cortex (S1) layer 4 (L4) neurons. To which extent this high efficacy depends on structural synaptic features remains unclear. Using both serial transmission (TEM) and focused ion beam milling scanning electron microscopy (FIB/SEM), we 3D-reconstructed and quantitatively analyzed anterogradely labeled VPMdm axons in L4 of adult mouse S1. All VPMdm synapses are asymmetric. Virtually all are established by axonal boutons, 53% of which contact multiple (2-4) elements (overall synapse/bouton ratio = 1.6). Most boutons are large (mean 0.47 μm3), and contain 1-3 mitochondria. Vesicle pools and postsynaptic density (PSD) surface areas are large compared to others in rodent cortex. Most PSDs are complex. Most synapses (83%) are established on dendritic spine heads. Furthermore, 15% of the postsynaptic spines receive a second, symmetric synapse. In addition, 13% of the spine heads have a large protrusion inserted into a membrane pouch of the VPMdm bouton. The unusual combination of structural features in VPMdm synapses is likely to contribute significantly to the high efficacy, strength, and plasticity of these thalamocortical synapses.

Mitochondria play a key role in energy production and calcium buffering, among many other functions. They provide most of the energy required by neurons, and they are transported along axons and dendrites to the regions of higher energy demands. We have used focused ion beam milling and scanning electron microscopy (FIB/SEM) to obtain stacks of serial sections from the somatosensory cortex of the juvenile rat. We have estimated the volume fraction occupied by mitochondria and their distribution between dendritic, axonal, and nonsynaptic processes. The volume fraction of mitochondria increased from layer I (4.59%) to reach its maximum in layer IV (7.74%) and decreased to its minimum in layer VI (4.03%). On average, 44% of mitochondrial volume was located in dendrites, 15% in axons and 41% in nonsynaptic elements. Given that dendrites, axons, and nonsynaptic elements occupied 38%, 23%, and 39% of the neuropil, respectively, it can be concluded that dendrites are proportionally richer in mitochondria with respect to axons, supporting the notion that most energy consumption takes place at the postsynaptic side. We also found a positive correlation between the volume fraction of mitochondria located in neuronal processes and the density of synapses.

Changes in the size of the synaptic junction are thought to have significant functional consequences. We used focused ion beam milling and scanning electron microscopy (FIB/SEM) to obtain stacks of serial sections from the six layers of the rat somatosensory cortex. We have segmented in 3D a large number of synapses (n = 6891) to analyze the size and shape of excitatory (asymmetric) and inhibitory (symmetric) synapses, using dedicated software. This study provided three main findings. Firstly, the mean synaptic sizes were smaller for asymmetric than for symmetric synapses in all cortical layers. In all cases, synaptic junction sizes followed a log-normal distribution. Secondly, most cortical synapses had disc-shaped postsynaptic densities (PSDs; 93%). A few were perforated (4.5%), while a smaller proportion (2.5%) showed a tortuous horseshoe-shaped perimeter. Thirdly, the curvature was larger for symmetric than for asymmetric synapses in all layers. However, there was no correlation between synaptic area and curvature.

Dendritic spines are thin protrusions that cover the dendritic surface of numerous neurons in the brain and whose function seems to play a key role in neural circuits. The correct segmentation of those structures is difficult due to their small size and the resulting spines can appear incomplete. This paper presents a four-step procedure for the complete reconstruction of dendritic spines. The haptically driven procedure is intended to work as an image processing stage before the automatic segmentation step giving the final representation of the dendritic spines. The procedure is designed to allow both the navigation and the volume image editing to be carried out using a haptic device. A use case employing our procedure together with a commercial software package for the segmentation stage is illustrated. Finally, the haptic editing is evaluated in two experiments; the first experiment concerns the benefits of the force feedback and the second checks the suitability of the use of a haptic device as input. In both cases, the results shows that the procedure improves the editing accuracy.