Volume Electron Microscopy Reveals Unique Laminar Synaptic Characteristics in the Human Entorhinal Cortex

  1. Sergio Plaza-Alonso
  2. Nicolás Cano-Astorga
  3. Javier DeFelipe
  4. Lidia Alonso-Nanclares

  • Curated by eLife

    eLife logo

    eLife assessment

    This study presents a useful examination of dense neuroanatomy in human postmortem medial entorhinal cortex, using a large number of small electron microscopy image volumes sampled from multiple cortical layers and individuals. The authors use solid experimental and annotation techniques, demonstrating the suitability of postmortem tissue reconstructions for analysis and presenting careful, detailed measurements of synapse properties and overall tissue composition. However, there is inadequate support connecting these findings to claims about general connectivity in medial entorhinal cortex, since factors affecting interpretability like noise, the spatial scales examined, and relationships between structural properties and connectivity are not characterized. With a more thorough contextualization, this work would be of interest for studies of cellular neuroanatomy or brain network organization.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

The entorhinal cortex (EC) plays a pivotal role in memory function and spatial navigation, connecting the hippocampus with the neocortex. The EC integrates a wide range of cortical and subcortical inputs, but its synaptic organization in the human brain is largely unknown. We used volume electron microscopy to perform a 3D analysis of the synapses in all layers of the medial EC (MEC) from the human brain. Using this technology, 12,974 synapses were fully 3D reconstructed at the ultrastructural level. The MEC presented a distinct set of synaptic features, differentiating this region from other human cortical areas. Furthermore, synaptic organization within the MEC was predominantly homogeneous, although layers I and VI exhibited several synaptic characteristics that were distinct from other layers. The present study constitutes an extensive description of the synaptic organization of the neuropil of all layers of the EC, a crucial step to better understand the connectivity of this cortical region, in both health and disease.

Significance Statement

Analysis of the synaptic characteristics provides crucial data on cortical organization. However, synaptic organization data for the normal human brain is virtually non-existent. The present study analyzes synaptic characteristics of the human entorhinal cortex at the ultrastructural level. This brain region is essential for memory processes and spatial navigation, acting as an interface between sensory areas and the hippocampus. Moreover, the entorhinal cortex is one of the first regions affected by Alzheimer’s disease. The present results provide a large quantitative ultrastructural dataset of synapses in all layers of the entorhinal cortex using 3D electron microscopy. Our findings show remarkable uniformity in the synaptic characteristics. This may seem surprising given the cytoarchitectonic and innervation complexity of the medial entorhinal cortex.

Article activity feed

  1. eLife

    eLife assessment

    This study presents a useful examination of dense neuroanatomy in human postmortem medial entorhinal cortex, using a large number of small electron microscopy image volumes sampled from multiple cortical layers and individuals. The authors use solid experimental and annotation techniques, demonstrating the suitability of postmortem tissue reconstructions for analysis and presenting careful, detailed measurements of synapse properties and overall tissue composition. However, there is inadequate support connecting these findings to claims about general connectivity in medial entorhinal cortex, since factors affecting interpretability like noise, the spatial scales examined, and relationships between structural properties and connectivity are not characterized. With a more thorough contextualization, this work would be of interest for studies of cellular neuroanatomy or brain network organization.

  2. eLife

    Reviewer #1 (Public Review):

    In this work, Plaza-Alonso et al. present a collection of volume electron microscopy (EM) reconstructions of human postmortem medial entorhinal cortex (MEC), and they measure properties of MEC cytoarchitecture and synapses as a function of neuroanatomical subdivision. The authors generate a sampling of 9 smaller (≲10 µm/side) EM reconstructions per subdivision to avoid prohibitively large (petabyte) EM volumes, using 3 reconstructions for each of 3 brain donors to control for inter-individual variability. Conducting in-depth analyses for 7 subdivisions (63 reconstructions total), the authors find little significant inter-subdivision variability in structural composition (volume fractions of cell bodies vs. neuropil vs. blood vessels) and multiple synapse properties (spatial distribution, density, area, shape, excitatory/inhibitory type, and postsynaptic cell compartment). They conclude that human MEC connectivity is largely homogeneous, with synapses arranged in a generally random spatial distribution and a large fraction of synapses being asymmetric (putatively excitatory). Their other findings include that asymmetric synapses are larger than symmetric/putatively inhibitory synapses; that asymmetric synapses prefer dendritic spines whereas symmetric synapses prefer dendritic shafts; and that a small fraction of synapses have larger, complex shapes that may suggest increased synaptic efficacy. They note that inhomogeneities may include inter-subdivision variation in asymmetric synapse area and complex-shaped synapse prevalence, and for some reconstructions (12/63), possible substructure in synapse distributions.

    Strengths:
    The authors have carefully conducted this work, using reasonable methods and comparing their findings with previous volume EM reconstructions where possible. It represents a substantial effort, given the challenges of producing and annotating volume EM data and of collecting human postmortem tissue. They have thus contributed a brain-region-specific characterization of human postmortem tissue with value as both a data resource and an examination of postmortem EM reconstruction quality, given that postmortem tissue is less-studied with volume EM but could be an important source of human brain samples (for example in regions that are surgically inaccessible). Further, some of the authors' measurements may be of added value, as they suggest functional correlates for less-studied synapse structures (such as the differing sizes of complex and simple "macular" synapses formed onto dendritic spines vs. shafts).

    Weaknesses:
    Despite these strengths, the analysis in this work may be impacted by multiple sources of experimental variability that may have contributed to the observed lack of structural variability, and the potential contributions of these should be addressed in making their claims.

    (1) The authors' approach to tissue sampling may have resulted in under-sampling, which may have reduced the detection power of their tests. More specifically, each reconstructed EM volume measured ~10 µm x 7 µm x 6 µm (360 - 502 µm^3) and contained ~300-400 synapses (Lines 211-212, 772-773). Per donor, this amounts to a sampling volume of ~1500 µm^3 for each MEC subdivision or ~1x104 µm^3 total. By contrast, the volume of the adult human MEC is ~1x10^12 µm^3, roughly 1x10^8 times larger [1]. Thus, while these EM reconstructions reflect a substantial effort, it is likely that they represent an under-sampling of MEC structure, especially since multiple excitatory and inhibitory neuron types are likely interspersed throughout (the authors also note this possibility in Lines 640-659).

    (2) The authors' measurements are combined across three donors who are biologically diverse (Table S11), including in terms of characteristics that themselves may impact neuronal connectivity. Without controlling for these variables, the possible reduction in stochastic, biological inter-individual variability that could be achieved by combining data across donors may be offset by increases in phenotype-related variability, which could reduce the detectability of true, conserved connectivity variations across MEC subdivisions. Specifically, these donors represent a mix of males and females; a mix of ages (40, 53, and 66 years) that suggest differing degrees of aging-related changes in neuronal connectivity (according to previous work, a majority of people >55 years of age are estimated to have Alzheimer's-associated neurofibrillary tangles, regardless of whether they have dementia symptomatology; see for instance [1]); and one death from metastatic cancer, indicating that for one donor cellular/neuronal abnormalities associated either with cancer itself or related therapies could be present.

    These two factors could substantially increase the dispersion of the authors' measurements in each MEC subdivision and lead to a situation with no detectable differences between subdivisions. It would be important to address these impacts when determining whether to interpret a lack of significant differences as true biological homogeneity for human MEC.

    One helpful approach would be to explicitly show the variance of each measurement obtained for each EM reconstruction. For example, error bars showing the interquartile range could be added to each data point in Fig. 3C, to show how much synapse areas vary per reconstruction and to allow some comparison across donors and MEC subdivisions.

    (3) A third potential source of variability relates to the authors' approach for synapse annotation. They appear to annotate active zones and postsynaptic densities by thresholding synapse images at some user-defined pixel intensity value, taking only pixels darker than that threshold as their annotations (Lines 806 - 812). This technique seems like it could be prone to producing noisy annotations, particularly since in the EM images provided (Figs. S11-16) the pixel intensities of active zones/postsynaptic densities and surrounding neuropil do not appear to be highly distinct.

    It would be important for the authors to support their findings by quantifying the variability that may be associated with this technique.

    [1] Price, C.C. et al., J. Int. Neuropsychol. Soc., (2010), doi: 10.1017/S135561771000072X.

  3. eLife

    Reviewer #2 (Public Review):

    Plaza-Alanso et al. characterize synaptic properties across human medial entorhinal cortex across layers and, importantly, across multiple individuals. Using an impressive collection of post-mortem autopsy samples, they generate high resolution 3d FIB-SEM volumes across layers and MEC subregions and measure features such as synapse density, spatial distribution, size, shape and target location. The use of volumes permits a richer local context to synaptic reconstructions, and the methods used to count and quantify synapses appear thorough and convincing, although with limited descriptions at times. The core findings suggest few differences in most properties across either layers or individuals, with some modest exceptions in layers 1 and 6. A particular strength of the dataset is the large number of high quality synaptic contact reconstructions.
    However, because the volumes have no specific labels and are too small to associate axons or dendrites with individual cells or cell types, it is not clear how to extrapolate these findings to new insights toward the stated goal of a better understanding of the networks and connectivity characteristics of the MEC. Broadly speaking, the paper would benefit from a better explanation of why these specific parameters were chosen and what the authors hoped to gain from them. It might be useful to think of what would need to be the case to see something substantially different. Many of the measures here reflect the properties of dendrites passing through a small volume, which depends on the number of cells of different cell types, the length and thickness of their dendritic arbors, synapse density distributions, local and long range afferents, and more. One interpretation of these results is that these neuropil volumes across layers and individuals are effectively fully packed with dendrites, with a similar ratio of excitatory and inhibitory neurons, dendrites with roughly similar thickness and synaptic input density and local E/I balance. Can the authors disentangle these cellular-scale contributions or constrain their inter-individual variability across individuals? The lack of variability is perhaps the main observation here, and understanding this more clearly could be useful for thinking about larger volumes where fewer replicates are currently possible.

  4. Version published to 10.1101/2023.11.20.567821v2 on bioRxiv
  5. Version published to 10.1101/2023.11.20.567821v1 on bioRxiv