Differential Structure of Hippocampal CA1 Pyramidal Neurons in the Human and Mouse

Abstract

Pyramidal neurons are the most common cell type and are considered the main output neuron in most mammalian forebrain structures. In terms of function, differences in the structure of the dendrites of these neurons appear to be crucial in determining how neurons integrate information. To further shed light on the structure of the human pyramidal neurons we investigated the geometry of pyramidal cells in the human and mouse CA1 region—one of the most evolutionary conserved archicortical regions, which is critically involved in the formation, consolidation, and retrieval of memory. We aimed to assess to what extent neurons corresponding to a homologous region in different species have parallel morphologies. Over 100 intracellularly injected and 3D-reconstructed cells across both species revealed that dendritic and axonal morphologies of human cells are not only larger but also have structural differences, when compared to mouse. The results show that human CA1 pyramidal cells are not a stretched version of mouse CA1 cells. These results indicate that there are some morphological parameters of the pyramidal cells that are conserved, whereas others are species-specific.

Comparing basal dendrite branches in human and mouse hippocampal CA1 pyramidal neurons with Bayesian networks

“…While the authors of Ref. 15 reported such observations (e.g., branch diameter decreased with increasing branch order in non-terminal branches), they did not test for correlation between variables and based such claims on the rejection of the hypothesis of equal median diameters across branch orders. As mentioned above, we focus on basal dendrites and thus leave apical dendrites for future work.We learn Bayesian networks from three different subsets of our data set: (a) from terminal branches alone; (b) from non-terminal branches alone; and (c) from both terminal and non-terminal branches.…”

“…The authors of Ref. 15 accounted for the determinants precisely by splitting the branches into terminal and non-terminal ones, and further down according to branch order, and then testing hypotheses of location difference (Kruskal-Wallis test) between pairs of obtained subsets of branches (e.g., mouse terminal branches of branch order two are as long as human terminal branches of the same branch order).Instead of splitting the branches according to the determinants and running multiple tests, we could specify a multivariate statistical model over the morphometrics and the morphological determinants. Models such as Bayesian networks (BNs) [16][17][18] can represent the probabilistic relationships among the variables of a domain, while algorithms for learning Bayesian networks from data can uncover such relationships.…”

“…This representation allows us to easily identify independencies and correlations among the morphometrics and determinants and thus identify whether a particular difference is due to intrinsic inter-species differences, due to differences in determinants, or due to a combination of both.In this paper, we use the hippocampus cells used in Ref. 15 (henceforth 'the previous study') yet focus on basal dendrites alone. In addition to branch length and average branch diameter, we consider three additional branch-level morphometrics: tortuosity, bifurcation angles and tilt angles.…”

“…An additional contribution, not present in Ref. 15 , is that we formally test whether a morphometric increases or decreases with the distance from the soma, by assessing linear correlation. While the authors of Ref.…”

“…For further details on tissue reconstruction, intracellular injections, immunocytochemistry, and cell reconstruction, see Ref. 15 .We only considered branches from branch order 1 to branch order 5, as only 4% of the branches corresponded to higher branching orders. We omitted from our analysis 23 branches that had a single child branch, as well as one trifurcating branch, 2/14…”