Unusual target selectivity of perisomatic inhibitory cells in the hilar region of the rat hippocampus

Abstract

Perisomatic inhibitory innervation of all neuron types profoundly affects their firing characteristics and vulnerability. In this study we examined the postsynaptic targets of perisomatic inhibitory cells in the hilar region of the dentate gyrus where the proportion of potential target cells (excitatory mossy cells and inhibitory interneurons) is approximately equal. Both cholecystokinin (CCK)- and parvalbumin-immunoreactive basket cells formed multiple contacts on the somata and proximal dendrites of mossy cells. Unexpectedly, however, perisomatic inhibitory terminals arriving from these cell types largely ignored hilar GABAergic cell populations. Eighty-ninety percent of various GABAergic neurons including other CCK-containing basket cells received no input from CCK-positive terminals. Parvalbumin-containing cells sometimes innervated each other but avoided 75% of other GABAergic cells. Overall, a single mossy cell received 40 times more CCK-immunoreactive terminals and 15 times more parvalbumin-positive terminals onto its soma than the cell body of an average hilar GABAergic cell. In contrast to the pronounced target selectivity in the hilar region, CCK- and parvalbumin-positive neurons innervated each other via collaterals in stratum granulosum and moleculare. Our observations indicate that the inhibitory control in the hilar region is qualitatively different from other cortical areas at both the network level and the level of single neurons. The paucity of perisomatic innervation of hilar interneurons should have profound consequences on their action potential generation and on their ensemble behavior. These findings may help explain the unique physiological patterns observed in the hilus and the selective vulnerability of the hilar cell population in various pathophysiological conditions.

Fig. 1.

Camera lucida drawings of the dendritic branching pattern (A) and the proximal axonal arbor (B) of CCK-positive interneurons in the dentate gyrus. The reconstructions were made from consecutive 60-μm-thick sections immunostained for CCK and SPR in A (see Materials and Methods) and for CCK in B.A, Hilar fusiform (cells 1 and3, black) and pyramidal-like (cell2, dark gray) CCK-positive interneurons are shown. Both cell types extend dendrites into str. moleculare as well as into the hilus. Note the prominent apical tuft of the pyramidal-like cell. Arrows point to dendrites that were used to establish that cell 1 is negative for VIP in “mirror” sections (see Materials and Methods).Arrowheads indicate axon initial segments. Scale bar, 40 μm. B, Proximal axon arbor of fusiform (cells1 and 2, black) and pyramidal-like (cells 2 and 4,dark gray) CCK-positive interneurons in the dentate gyrus. The axon terminals of both cell types innervate the supragranular str. moleculare as well as the hilus. The axon of cell1 reaches str. pyramidale of the CA3c region.Arrows point to thick, proximal axon segments that faded abruptly within the section, a sign of myelination.Arrowheads indicate axon initial segments. Scale bar, 50 μm.

Fig. 2.

Target selectivity of CCK-immunoreactive terminals in the hilus. A, High-power light micrograph of a double-immunostained section demonstrates that several CCK-positive terminals (arrows, blackchromogene) contact GluR2/3-positive mossy cells (brownchromogene) in a basket-like manner.B–E, In contrast, CCK-positive pericellular baskets (arrows) never surround the cell bodies (asterisks) of various interneuron classes visualized by parvalbumin (B, C), SPR (D), or mGluR1a (E). In these sections, multiple perisomatic contacts can always be seen around immunonegative, presumed mossy cells (S).Arrowhead in C indicates a single CCK-positive contact on a parvalbumin-immunoreactive cell. Scale bars, 10 μm.

Fig. 3.

CCK-immunoreactive terminals selectively innervate mossy cells in the hilar region. Camera lucida drawings are of CCK–GluR2/3 and CCK–SPR double-immunostained sections.A, Several CCK-immunoreactive terminals surround the somata and proximal dendrites of each GluR2/3-positive neuron.B, In contrast, none of the SPR-containing interneurons are contacted by multiple CCK-positive boutons. Arrowsin A and B indicate CCK-positive neurons that are avoided by CCK-immunoreactive terminals, similar to other GABAergic cells. Arrowhead in B points to a single CCK-positive contact on an SPR-expressing interneuron. Scale bar (shown in A for A andB): 50 μm.

Fig. 4.

Correlated light and electron micrographs of GluR2/3-immunoreactive mossy cells (S₁–S₃) receiving multiple contacts from CCK-positive boutons (b₁–b₆). The low-power (B) and high-power micrographs (C–H) verify that conventional symmetrical synapses are established on the somata (b₂,b₄, b₆) and proximal dendrites (b₁, b₃,b₅) of mossy cells (arrows). Note the long (D) or perforated (F,G) postsynaptic specializations of synapses formed by CCK-positive terminals. Scale bars: A, 6 μm;B, 3 μm; C, 0.4 μm.

Fig. 5.

Bar graphs showing the number of CCK-positive (A) and parvalbumin-positive (B) terminals around the somata of interneurons [INT (L), black bars] and mossy cells [MOSS(R), gray bars]. Note that most of the interneurons receive no CCK or parvalbumin contact, whereas most mossy cells are innervated by multiple contacts. In A,INT (L) comprises the pooled data of CCK–PV, CCK–mGluR1a, and CCK–CCK innervation, whereasMOSS (R) includes CCK–GluR2/3 and CCK–CGRP data. In B, INT(L) is based on parvalbumin–SPR immunostaining, and MOSS (R) consists of parvalbumin–GluR2/3 double-immunostaining data.

Fig. 6.

Interaction among parvalbumin-immunoreactive cells in the dentate gyrus.A, Parvalbumin-positive terminals contact (arrowheads) a parvalbumin-containing neuron in the hilus. Arrow points to an axon that could be traced back to the parvalbumin-positive plexus in str. granulosum.B, Similar multiple contacts were observed in the fimbria-fornix-lesioned animals, indicating that these terminals have local origin. C, Parvalbumin-immunoreactive terminals in str. granulosum densely cover the somata and apical dendrites of pyramidal-like parvalbumin-positive basket cells. Scale bars, 10 μm.

Fig. 7.

Innervation of mossy cells by parvalbumin-positive terminals. A, B, Correlated light and electron micrographs demonstrate that five parvalbumin-positive terminals (b₁–b₅) surround a GluR2/3-positive interneuron in this focal plane.C–E, Two of these terminals (b₃, b₄) are shown to establish symmetrical synapses on the somata of a mossy cell. These terminals rarely formed perforated synapses, e.g., seeb₃ shown in two neighboring sections. Scale bars: A, 20 μm; B, 2 μm; C, 0.5μm.

Fig. 8.

Interaction among CCK- and parvalbumin-positive neurons in the str. moleculare. Correlated light and electron microscopy demonstrate that the CCK-immunoreactive boutonb₁ establishes a symmetrical synapse on the apical dendritic shaft of a pyramidal-like parvalbumin-positive interneuron. Five additional contacts were identified on the same dendrite. Note that the dark electron-dense DAB-Ni precipitate inb₁ and b₂ is easily distinguishable from the light DAB end product in the dendrite. For a color version of A, see supplementary information. Scale bars: A, 10 μm; B, 1 μm; C, 0.5 μm.

Fig. 9.

Schematic drawing demonstrating the interaction among various hilar cell types in the hilus and in str. moleculare. The connectivity of CCK- and parvalbumin-positive cells is shown inblack and gray, respectively.Dotted line and an X mark indicate lack of connection, whereas continuous lines show the verified synaptic interactions. The thickness of the lines corresponds to the strength of the connection. Note that besides a weak interaction among parvalbumin-containing cells, interneurons avoid each other in the hilus but have mutual connections in str. moleculare and str. granulosum. For clarity, only a single dendrite of each basket cell is drawn that reaches str. moleculare. Mossy cells receive innervation from both CCK- and parvalbumin-positive basket cells, but hilar interneurons, many of which are somatostatin-containing, are avoided. The innervation of parvalbumin-positive cells by somatostatin-immunoreactive neurons was demonstrated by Katona et al. (1999).