Quantitative morphometry of electrophysiologically identified CA3b interneurons reveals robust local geometry and distinct cell classes

Abstract

The morphological and electrophysiological diversity of inhibitory cells in hippocampal area CA3 may underlie specific computational roles and is not yet fully elucidated. In particular, interneurons with somata in strata radiatum (R) and lacunosum-moleculare (L-M) receive converging stimulation from the dentate gyrus and entorhinal cortex as well as within CA3. Although these cells express different forms of synaptic plasticity, their axonal trees and connectivity are still largely unknown. We investigated the branching and spatial patterns, plus the membrane and synaptic properties, of rat CA3b R and L-M interneurons digitally reconstructed after intracellular labeling. We found considerable variability within but no difference between the two layers, and no correlation between morphological and biophysical properties. Nevertheless, two cell types were identified based on the number of dendritic bifurcations, with significantly different anatomical and electrophysiological features. Axons generally branched an order of magnitude more than dendrites. However, interneurons on both sides of the R/L-M boundary revealed surprisingly modular axodendritic arborizations with consistently uniform local branch geometry. Both axons and dendrites followed a lamellar organization, and axons displayed a spatial preference toward the fissure. Moreover, only a small fraction of the axonal arbor extended to the outer portion of the invaded volume, and tended to return toward the proximal region. In contrast, dendritic trees demonstrated more limited but isotropic volume occupancy. These results suggest a role of predominantly local feedforward and lateral inhibitory control for both R and L-M interneurons. Such a role may be essential to balance the extensive recurrent excitation of area CA3 underlying hippocampal autoassociative memory function.

Figure 1

Electrophysiological characteristics of R and L-M interneurons in area CA3b. A, B. Examples of membrane responses and firing patterns elicited by low and high depolarizing current injections (240 pA, upper traces, and 510 pA, lower traces, respectively). The membrane potential was -70 mV. Adaptation ratio: 4.0 and 2.2 for the R and LM interneuron, respectively. C, D. Samples of average (10 sweeps) EPSPs and EPSCs evoked from three excitatory pathways innervating R and LM interneurons. Only mossy fiber responses show high sensitivity to the application of the group II mGluR agonist, DCG-IV (1 μM, red traces, DCG-IV sensitivity 79 ± 7 % for EPSPs and 82.9 ± 5% for EPSCs).

Figure 2

R interneurons show predominantly MF LTD. HFS (3 trains of 100 pulses each at 100 Hz, repeated every 10 sec) delivered to the mossy fibers induced robust synaptic depression in current-clamp recordings (-70 mV) at 35 min post-HFS (40.69 ± 3.89 % of the control EPSP amplitude; p< 0.0001, unpaired t test; N=7). Insets are average MF EPSP (10 sweeps) from a typical experiment.

Figure 3

Representative micrographs of CA3b interneurons typical of cells selected for quantitative morphometric analysis. Such cells exhibited dense labeling of both the somatodendritic compartment and axonal arbor in multiple serial sections. Panels A and B demonstrate the dense labeling of both dendrites and axons. The boxed areas in panel A illustrate the smooth fiber morphology characteristic of dendrites (lower left box) and the varicose nature of axons (two remaining boxes). Note the localized arborization of both classes of process with respect to the cell soma evident at both high and low magnification (inset: f, fimbria). This distinction is also evident in panel B where black arrows identify a long sparsely branching dendrite and varicose axons (gray arrows) branch prolifically within stratum pyramidale of CA3. All scale bars are 100 μm (panels A and B, and inset of A).

Figure 4

Cropped images around the somata and proximal processes of four illustrative cells used in the analysis (A-D), along with their digital reconstruction (E-H). R (panels A/E and C/G) and L-M (B/F and D/H) groups are defined by somatic location. These examples also represent the distinction between cells with high (panels A/E and B/F) and low (C/G and D/H) numbers of dendritic branches. Dendrites (red) have been thickened 5-fold to help distinguish them from axons (blue). All photomicrographs are of the same magnification; scale bar in A: 100 μm.

Figure 5

Analysis of axonal terminal tips. (A) The position of every axonal termination in each of the reconstructed cells after shrinkage correction was rank ordered in the depth of the slice from lowest (left) to highest (right). The locations of the soma and of the farthest point along the path are marked in each case. Significant truncation artifacts due to slice sectioning would result in a disproportioned number of terminals lying along one of the edges. No cells clearly exhibit this trend. Two interneurons (*) had their farthest point near the edge of the cell, and one (#) had more than 20% of its terminals in the extreme tenth of the slice, indicating the potential for truncation. None of these cells exhibited morphometric outliers compared to the other cells. (B) Histogram of the frequency of tips at several depths, with left and right ordinates representing absolute numbers and relative proportions of the total count.

Figure 6

Length distribution across path distance from soma. (A) Sholl-like plot of the amount of axonal length at subsequent 100 μm-wide path distance bins. Cell values are grouped and averaged by layer of somatic location and by internal (Int) and terminal branches (Term). (B) The axonal length distributions for two representative cells from both the L-M and R groups are plotted in subsequent 5%-wide bins, where path distance is measured as a proportion of the maximum path distance for each cell. (C) Comparison of relative axonal and dendritic length distributions by 5% path increments for all cells pooled together.

Figure 7

Dendritic branching in CA3b interneurons. (A) Scatter plot distribution of the number of dendritic bifurcations by somatic surface area. Dashed lines separate cells with high (HiDe) and low (LoDe) numbers of dendritic bifurcations. (B) Semi-log histogram of dendritic branching characteristics. Cells cluster into two groups by number of dendritic bifurcations, with a gap in between ≤10 and ≥18 dendritic bifurcations. R and L-M cells are found in both high and low dendritic groups. Mann-Whitney (M-W) test was used to compare number of dendritic bifurcations for LoDe and HiDe cell groups. Kolmogorov-Smirnov (K-S) Poisson and uniform distribution tests were performed on all cells grouped together.

Figure 8

Dendritic and axonal morphological distributions. (A) Branch path length, (B) contraction, (C) partition asymmetry, and (D) bifurcation amplitude angle are averaged for all cells and plotted across branch order (bars are standard errors). (E) Schematic diagram of the morphometrics reported in A-D. To the left and right are simplified illustrations of a dendritic and an axonal tree, respectively, where the similarity of the branches across branch order represents the actual similarities found for both axonal and dendritic branches. In the middle is a zoomed-in representation of an individual branch (thick gray).

Figure 9

Morphological properties across Euclidean distance from the soma. (A) Axonal (black) and dendritic (gray) path distance averaged for all cells and plotted against absolute Euclidean distance (25 μm bins). (B) Axonal surface area averaged for all cells and plotted in subsequent 5%-wide bins, where distance is measured as a proportion of the maximum Euclidean distance for each cell. (B: Inset) Path distance and surface area are averaged for all 13 cells in the “core” (“C”) and “shell” (“S”), respectively defined as the inner and outer halves of the smallest spherical volume enclosing the entire axonal tree. Bars represent standard errors in A and B, and standard deviations in the Inset.

Figure 10

Anatomical orientation of axonal/dendritic trees. (A) Axonal and (B) dendritic maps show the volumetric proportion for eight sectors, each corresponding to a square on a cube face. The soma of each cell would be located in the middle of the cube. The darkness of every square is proportional to the volumetric proportion of the given sector. Matching color stars indicate significantly different volumetric proportions. (C-D, left) Axonal and dendritic polar histograms of average length distribution relative to the soma for R (blue) and L-M (red) cell groups (bin size: π/8 rad). Values are normalized by dividing the length in each bin by the mean length over all bins. (C-D, right) Both axons and dendrites exhibit lamellar spread, where the arborization is more confined in the longitudinal direction compared to the transverse plane. The lamellar ratio between the longitudinal and transverse spreads of tracing points is averaged over all cells.