Synaptic and dendritic architecture of different types of hippocampal somatostatin interneurons.

GABAergic inhibitory neurons fundamentally shape the activity and plasticity of cortical circuits. A major subset of these neurons contains somatostatin (SOM); these cells play crucial roles in neuroplasticity, learning, and memory in many brain areas including the hippocampus, and are implicated in several neuropsychiatric diseases and neurodegenerative disorders. Two main types of SOM-containing cells in area CA1 of the hippocampus are oriens-lacunosum-moleculare (OLM) cells and hippocampo-septal (HS) cells. These cell types show many similarities in their soma-dendritic architecture, but they have different axonal targets, display different activity patterns in vivo, and are thought to have distinct network functions. However, a complete understanding of the functional roles of these interneurons requires a precise description of their intrinsic computational properties and their synaptic interactions. In the current study we generated, analyzed, and make available several key data sets that enable a quantitative comparison of various anatomical and physiological properties of OLM and HS cells in mouse. The data set includes detailed scanning electron microscopy (SEM)-based 3D reconstructions of OLM and HS cells along with their excitatory and inhibitory synaptic inputs. Combining this core data set with other anatomical data, patch-clamp electrophysiology, and compartmental modeling, we examined the precise morphological structure, inputs, outputs, and basic physiological properties of these cells. Our results highlight key differences between OLM and HS cells, particularly regarding the density and distribution of their synaptic inputs and mitochondria. For example, we estimated that an OLM cell receives about 8,400, whereas an HS cell about 15,600 synaptic inputs, about 16% of which are GABAergic. Our data and models provide insight into the possible basis of the different functionality of OLM and HS cell types and supply essential information for more detailed functional models of these neurons and the hippocampal network.

Amyloid ß induces interneuron-specific changes in the hippocampus of APPNL-F mice.

Alzheimer's disease (AD) is a neurodegenerative disorder characterized by cognitive decline and amyloid-beta (Aß) depositions generated by the proteolysis of amyloid precursor protein (APP) in the brain. In APPNL-F mice, APP gene was humanized and contains two familial AD mutations, and APP-unlike other mouse models of AD-is driven by the endogenous mouse APP promoter. Similar to people without apparent cognitive dysfunction but with heavy Aß plaque load, we found no significant decline in the working memory of adult APPNL-F mice, but these mice showed decline in the expression of normal anxiety. Using immunohistochemistry and 3D block-face scanning electron microscopy, we found no changes in GABAA receptor positivity and size of somatic and dendritic synapses of hippocampal interneurons. We did not find alterations in the level of expression of perineuronal nets around parvalbumin (PV) interneurons or in the density of PV- or somatostatin-positive hippocampal interneurons. However, in contrast to other investigated cell types, PV interneuron axons were occasionally mildly dystrophic around Aß plaques, and the synapses of PV-positive axon initial segment (AIS)-targeting interneurons were significantly enlarged. Our results suggest that PV interneurons are highly resistant to amyloidosis in APPNL-F mice and amyloid-induced increase in hippocampal pyramidal cell excitability may be compensated by PV-positive AIS-targeting cells. Mechanisms that make PV neurons more resilient could therefore be exploited in the treatment of AD for mitigating Aß-related inflammatory effects on neurons.

Brainstem nucleus incertus controls contextual memory formation.

Hippocampal pyramidal cells encode memory engrams, which guide adaptive behavior. Selection of engram-forming cells is regulated by somatostatin-positive dendrite-targeting interneurons, which inhibit pyramidal cells that are not required for memory formation. Here, we found that γ-aminobutyric acid (GABA)-releasing neurons of the mouse nucleus incertus (NI) selectively inhibit somatostatin-positive interneurons in the hippocampus, both monosynaptically and indirectly through the inhibition of their subcortical excitatory inputs. We demonstrated that NI GABAergic neurons receive monosynaptic inputs from brain areas processing important environmental information, and their hippocampal projections are strongly activated by salient environmental inputs in vivo. Optogenetic manipulations of NI GABAergic neurons can shift hippocampal network state and bidirectionally modify the strength of contextual fear memory formation. Our results indicate that brainstem NI GABAergic cells are essential for controlling contextual memories.

Co-transmission of acetylcholine and GABA regulates hippocampal states.

The basal forebrain cholinergic system is widely assumed to control cortical functions via non-synaptic transmission of a single neurotransmitter. Yet, we find that mouse hippocampal cholinergic terminals invariably establish GABAergic synapses, and their cholinergic vesicles dock at those synapses only. We demonstrate that these synapses do not co-release but co-transmit GABA and acetylcholine via different vesicles, whose release is triggered by distinct calcium channels. This co-transmission evokes composite postsynaptic potentials, which are mutually cross-regulated by presynaptic autoreceptors. Although postsynaptic cholinergic receptor distribution cannot be investigated, their response latencies suggest a focal, intra- and/or peri-synaptic localisation, while GABAA receptors are detected intra-synaptically. The GABAergic component alone effectively suppresses hippocampal sharp wave-ripples and epileptiform activity. Therefore, the differentially regulated GABAergic and cholinergic co-transmission suggests a hitherto unrecognised level of control over cortical states. This novel model of hippocampal cholinergic neurotransmission may lead to alternative pharmacotherapies after cholinergic deinnervation seen in neurodegenerative disorders.

Median raphe region stimulation alone generates remote, but not recent fear memory traces.

The median raphe region (MRR) is believed to control the fear circuitry indirectly, by influencing the encoding and retrieval of fear memories by amygdala, hippocampus and prefrontal cortex. Here we show that in addition to this established role, MRR stimulation may alone elicit the emergence of remote but not recent fear memories. We substituted electric shocks with optic stimulation of MRR in C57BL/6N male mice in an optogenetic conditioning paradigm and found that stimulations produced agitation, but not fear, during the conditioning trial. Contextual fear, reflected by freezing was not present the next day, but appeared after a 7 days incubation. The optogenetic silencing of MRR during electric shocks ameliorated conditioned fear also seven, but not one day after conditioning. The optogenetic stimulation patterns (50Hz theta burst and 20Hz) used in our tests elicited serotonin release in vitro and lead to activation primarily in the periaqueductal gray examined by c-Fos immunohistochemistry. Earlier studies demonstrated that fear can be induced acutely by stimulation of several subcortical centers, which, however, do not generate persistent fear memories. Here we show that the MRR also elicits fear, but this develops slowly over time, likely by plastic changes induced by the area and its connections. These findings assign a specific role to the MRR in fear learning. Particularly, we suggest that this area is responsible for the durable sensitization of fear circuits towards aversive contexts, and by this, it contributes to the persistence of fear memories. This suggests the existence a bottom-up control of fear circuits by the MRR, which complements the top-down control exerted by the medial prefrontal cortex.

The ascending median raphe projections are mainly glutamatergic in the mouse forebrain.

The median raphe region (MRR) is thought to be serotonergic and plays an important role in the regulation of many cognitive functions. In the hippocampus (HIPP), the MRR exerts a fast excitatory control, partially through glutamatergic transmission, on a subpopulation of GABAergic interneurons that are key regulators of local network activity. However, not all receptors of this connection in the HIPP and in synapses established by MRR in other brain areas are known. Using combined anterograde tracing and immunogold methods, we show that the GluN2A subunit of the NMDA receptor is present in the synapses established by MRR not only in the HIPP, but also in the medial septum (MS) and in the medial prefrontal cortex (mPFC) of the mouse. We estimated similar amounts of NMDA receptors in these synapses established by the MRR and in local adjacent excitatory synapses. Using retrograde tracing and confocal laser scanning microscopy, we found that the majority of the projecting cells of the mouse MRR contain the vesicular glutamate transporter type 3 (vGluT3). Furthermore, using double retrograde tracing, we found that single cells of the MRR can innervate the HIPP and mPFC or the MS and mPFC simultaneously, and these double-projecting cells are also predominantly vGluT3-positive. Our results indicate that the majority of the output of the MRR is glutamatergic and acts through NMDA receptor-containing synapses. This suggests that key forebrain areas receive precisely targeted excitatory input from the MRR, which is able to synchronously modify activity in those regions via individual MRR cells with dual projections.

Quantitative ultrastructural analysis of basket and axo-axonic cell terminals in the mouse hippocampus.

Three functionally different populations of perisomatic interneurons establish GABAergic synapses on hippocampal pyramidal cells: parvalbumin (PV)-containing basket cells, type 1 cannabinoid receptor (CB1)-positive basket cells both of which target somata, and PV-positive axo-axonic cells that innervate axon initial segments. Using electron microscopic reconstructions, we estimated that a pyramidal cell body receives synapses from about 60 and 140 synaptic terminals in the CA1 and CA3 area, respectively. About 60 % of these terminals were PV positive, whereas 35-40 % of them were CB1 positive. Only about 1 % (CA1) and 4 % (CA3) of the somatic boutons were negative for both markers. Using fluorescent labeling, we showed that most of the CB1-positive terminals expressed vesicular glutamate transporter 3. Reconstruction of somatic boutons revealed that although their volumes are similar, CB1-positive boutons are more flat and the total volume of their mitochondria was smaller than that of PV-positive boutons. Both types of boutons contain dense-core vesicles and frequently formed multiple release sites on their targets and innervated an additional soma or dendrite as well. PV-positive boutons possessed small, macular synapses; whereas the total synaptic area of CB1-positive boutons was larger and formed multiple irregular-shaped synapses. Axo-axonic boutons were smaller than somatic boutons, had only one synapse and their ultrastructural parameters were closer to those of PV-positive somatic boutons. Our results represent the first quantitative measurement-using a highly reliable method-of the contribution of different cell types to the perisomatic innervation of pyramidal neurons, and may help to explain functional differences in their output properties.

Neuroligin 2 is expressed in synapses established by cholinergic cells in the mouse brain.

Neuroligin 2 is a postsynaptic protein that plays a critical role in the maturation and proper function of GABAergic synapses. Previous studies demonstrated that deletion of neuroligin 2 impaired GABAergic synaptic transmission, whereas its overexpression caused increased inhibition, which suggest that its presence strongly influences synaptic function. Interestingly, the overexpressing transgenic mouse line showed increased anxiety-like behavior and other behavioral phenotypes, not easily explained by an otherwise strengthened GABAergic transmission. This suggested that other, non-GABAergic synapses may also express neuroligin 2. Here, we tested the presence of neuroligin 2 at synapses established by cholinergic neurons in the mouse brain using serial electron microscopic sections double labeled for neuroligin 2 and choline acetyltransferase. We found that besides GABAergic synapses, neuroligin 2 is also present in the postsynaptic membrane of cholinergic synapses in all investigated brain areas (including dorsal hippocampus, somatosensory and medial prefrontal cortices, caudate putamen, basolateral amygdala, centrolateral thalamic nucleus, medial septum, vertical- and horizontal limbs of the diagonal band of Broca, substantia innominata and ventral pallidum). In the hippocampus, the density of neuroligin 2 labeling was similar in GABAergic and cholinergic synapses. Moreover, several cholinergic contact sites that were strongly labeled with neuroligin 2 did not resemble typical synapses, suggesting that cholinergic axons form more synaptic connections than it was recognized previously. We showed that cholinergic cells themselves also express neuroligin 2 in a subset of their input synapses. These data indicate that mutations in human neuroligin 2 gene and genetic manipulations of neuroligin 2 levels in rodents will potentially cause alterations in the cholinergic system as well, which may also have a profound effect on the functional properties of brain circuits and behavior.

Extrinsic and local glutamatergic inputs of the rat hippocampal CA1 area differentially innervate pyramidal cells and interneurons.

The two main glutamatergic pathways to the CA1 area, the Schaffer collateral/commissural input and the entorhinal fibers, as well as the local axons of CA1 pyramidal cells innervate both pyramidal cells and interneurons. To determine whether these inputs differ in their weights of activating GABAergic circuits, we have studied the relative proportion of pyramidal cells and interneurons among their postsynaptic targets in serial electron microscopic sections. Local axons of CA1 pyramidal cells, intracellularly labeled in vitro or in vivo, innervated a relatively high proportion of interneuronal postsynaptic targets (65.9 and 53.8%, in vitro and in vivo, respectively) in stratum (str.) oriens and alveus. In contrast, axons of in vitro labeled CA3 pyramidal cells in str. oriens and str. radiatum of the CA1 area made synaptic junctions predominantly with pyramidal cell spines (92.9%). The postsynaptic targets of anterogradely labeled medial entorhinal cortical boutons in CA1 str. lacunosum-moleculare were primarily pyramidal neuron dendritic spines and shafts (90.8%). The alvear group of the entorhinal afferents, traversing str. oriens, str. pyramidale, and str. radiatum showed a higher preference for innervating GABAergic cells (21.3%), particularly in str. oriens/alveus. These data demonstrate that different glutamatergic pathways innervate CA1 GABAergic cells to different extents. The results suggest that the numerically smaller CA1 local axonal inputs together with the alvear part of the entorhinal input preferentially act on GABAergic interneurons in contrast to the CA3, or the entorhinal input in str. lacunosum-moleculare. The results highlight differences in the postsynaptic target selection of the feed-forward versus recurrent glutamatergic inputs to the CA1 and CA3 areas.

New silver-gold intensification method of diaminobenzidine for double-labeling immunoelectron microscopy.

The available methods for double-labeling preembedding immunoelectron microscopy are highly limited because not only should the ultrastructure be preserved, but also the different antigens should be visualized by reaction end products that can be clearly distinguished in gray-scale images. In these procedures, one antigen is detected with 3,3'-diaminobenzidine (DAB) chromogen, resulting in a homogeneous deposit, whereas the other is labeled with either a gold-tagged immunoreagent, or DAB polymer, on the surface of which metallic silver is precipitated. The detection of the second antigen is usually impeded by the first, leading to false-negative results. The authors aimed to diminish this hindrance by a new silver intensification technique of DAB polymer, which converts the deposit from amorphous to granular. The method includes three major postdevelopmental steps: (1) treatment of nickel-enhanced DAB with sulfide, (2) silver deposition in the presence of hydroquinone under acidic conditions, and (3) precious metal replacement with gold thiocyanate. This new sulfide-silver-gold intensification of DAB (SSGI) allows a subsequent detection of other antigens using DAB. In conclusion, the new technique loads fine gold particles onto the DAB deposit at a very low background level, thereby allowing a reliable discernment between the elements stained for the two antigens at the ultrastructural level.

Distinctive properties of CXC chemokine receptor 4-expressing Cajal-Retzius cells versus GABAergic interneurons of the postnatal hippocampus.

The CXC chemokine receptor 4 (CXCR4) for the chemokine (C-X-C motif) ligand 12/stromal cell-derived factor-1 alpha (CXCL12/SDF-1 alpha) is highly expressed in the postnatal CA1 stratum lacunosum-moleculare. However, both the network events triggered by SDF-1 alpha in this microcircuit and the cellular targets of this chemokine remain virtually unexplored. Here, we have studied SDF-1 alpha-mediated neuromodulation of the stratum lacunosum-moleculare by directly comparing the properties of CXCR4-expressing Cajal-Retzius cells vs. CXCR4-non-expressing interneurons, and by recording the electrophysiological effects caused by application of SDF-1 alpha on either cell type. We demonstrate that SDF-1 alpha dramatically reduces spontaneous firing in Cajal-Retzius cells via hyerpolarization, and that cessation of firing is prevented by the CXCR4-specific antagonist AMD3100. In contrast, no effects on the excitability of interneurons of the same layer were observed following exposure to the chemokine. We also provide evidence that, despite the expression of functional glutamate receptors, Cajal-Retzius cells are integrated in the synaptic network of the stratum lacunosum-moleculare via excitatory GABAergic input. Furthermore, we show that the axons of Cajal-Retzius cells target specifically the stratum lacunosum-moleculare and the dentate gyrus, but lack postsynaptic specializations opposite to their axonal varicosities. These results, taken together with our observation that SDF-1 alpha reduces evoked field responses at the entorhinal cortex-CA1 synapse, suggest that Cajal-Retzius cells produce a diffuse output that may impact information processing of stratum lacunosum-moleculare. We propose that pathological alterations of local levels of SDF-1 alpha or CXCR4 expression may affect the functions of an important hippocampal microcircuit.

Types and synaptic connections of hippocampal inhibitory neurons reciprocally connected with the medial septum.

The morphological properties and connectivity of gamma-aminobutyric acid (GABA)ergic hippocampal cells projecting to the medial septum (HS cells) were examined in the rat. Two types of HS cells are located in different layers of the hippocampus: sparsely-spiny cells are in CA1-3 str. oriens and CA3 str. radiatum, where recurrent axons of pyramidal cells arborize. Densely-spiny HS cells with spiny somata are located in the termination zone of granule cell axons. In the hilus, intermediate morphologies can also be found. HS cells receive GABAergic medial septal afferents in all layers where they occur, thus the connectivity of the septum and the hippocampus is reciprocal at cell level. HS cells receive extremely dense innervation, sparsely-spiny cells are innervated by approximately 19,000 excitatory inputs, while densely-spiny cells get an even larger number (approximately 37,000). While 14% of the inputs are inhibitory for the sparsely-spiny cells, it is only 2.3% in the case of densely-spiny cells. Because a high proportion (up to 54.5% on somata and 27.5% on dendrites) of their GABAergic inputs derived from labelled septal terminals, their predominant inhibitory input probably arises from the medial septum. CA1 area HS cells possessed myelinated projecting axons, as well as local collaterals, which targeted mostly pyramidal cell dendrites and spines in str. oriens and radiatum. The synaptic organization suggests that by sampling the activity of large populations of principal cells HS cells can reliably broadcast hippocampal activity level to the medial septum.