Ca2+ mechanisms of synaptic integration and plasticity
in inhibitory interneurons
Doctorat en biochimie
Philosophiæ doctor (Ph. D.)
mechanisms of synaptic integration and
plasticity in inhibitory interneurons
Sous la direction de :
La signalisation calcique dendritique joue un rôle important dans la régulation de mécanismes neuronaux, tels que la plasticité synaptique et l’intégration de l’information transmise. Bien compris chez les neurones principaux, ce processus de régulation est moins étudié chez les divers types d’interneurones GABAergiques qui modulent l’acquisition et l’envoi de signaux neuronaux. Chez les interneurones à décharge rapide, un type d’interneurone commun dans les circuits corticaux, il a été démontré qu’il y a absence de rétropropagation des potentiels d’action dans les dendrites distales (Hu et al., 2010). Cette découverte a des implications fonctionnelles, car la rétropropagation des potentiels d’action est un signal important pour l’induction des formes de plasticité synaptique hebbiennes. Par contre, il a été suggéré que l’activité dendritique locale pourrait compenser pour l’absence de rétropropagation des potentiels d’action. En conséquence, ce travail porte sur l’étude des évènements calciques dans les dendrites distales des interneurones à décharge rapide. Nous avons cherché à déterminer s’il est possible de générer ces signaux calciques par stimulation dendritique locale, à étudier les mécanismes responsables de ces signaux et à déterminer si ces signaux jouent un rôle dans la régulation de la plasticité synaptique à ces synapses. Pour atteindre ces objectifs, nous avons utilisé une combinaison de méthodes électrophysiologiqes (patch-clamp en mode cellule entière), d’imagerie calcique deux-photons et de modélisation computationnelle. Nous avons pu établir qu’il est possible de générer des évènements calciques postsynaptiques supralinéaires dans les synapses excitatrices étudiées par stimulation électrique locale. Ces signaux sont médiés par l’influx calcique provenant de l’activation des récepteurs AMPA perméables au Ca2+, qui déclenche
à son tour le relâchement de Ca2+ par les récepteurs ryanodine présents sur
réserves calciques intracellulaires. Ces signaux comprennent aussi une contribution calcique mineure des récepteurs NMDA, et ils restent locaux (pas de propagation dans l’arbre dendritique). De plus, nous avons déterminé que ces évènements calciques supralinéaires produisent un revirement de la plasticité synaptique, car ils induisent la dépression à long-terme dans les synapses étudiées, alors que les signaux calciques de basse amplitude induisent la potentiation à long-terme. Nous avons aussi examiné si ces évènements calciques supralinéaires étaient générés de façon équivalente dans les dendrites apicales et basales, qui reçoivent des synapses de différentes sources. Nous avons observé que les signaux des dendrites apicales avaient une plus grande amplitude et étaient associés à un plus haut niveau de dépolarisation. À partir de la modélisation, nous avons pu prédire le nombre de synapses nécessaires à la génération de ces signaux et la contribution potentielle des mécanismes d’extrusion du Ca2+. Finalement, nous avons étudié la spécificité cellulaire des
mécanismes d’intégration dendritique en combinant l’imagerie calcique et la modélisation dans un type différent d’interneurone, les interneurones spécifiques aux interneurones type III. En conclusion, nous avons prouvé qu’il existe dans certains interneurones des mécanismes alternatifs, médiés par des hausses de Ca2+ locales, permettant la régulation de la plasticité aux synapses excitatrices.
Dendritic Ca2+ signaling plays an important role in the regulation of neuronal
processes, such as synaptic plasticity and input integration. Well-studied in principal neurons, this form of regulation is not well understood in the various types of GABAergic interneurons that modulate activity in neuronal networks. In fast-spiking (FS) interneurons, a common interneuron type in cortical circuits, it has been shown that there is a lack of action potential (AP) backpropagation in distal dendrites (Hu et al., 2010). This discovery has functional implications, AP backpropagation is an important signal for the induction of Hebbian forms of synaptic plasticity. However, it has been suggested that local dendritic activity could compensate for the absence of AP backpropagation. Consequently, this work focuses on the study of Ca2+ transients in distal dendrites of FS interneurons.
We sought to determine whether it is possible to generate supralinear Ca2+
transients through local dendritic stimulation, to study the mechanisms responsible for those transients and to determine whether those signals play a role in the regulation of synaptic plasticity at those synapses. To reach those objectives, we used a combination of electrophysiological methods (whole-cell patch-clamp recordings), two-photon Ca2+ imaging and of computational modeling. We were
able to establish that supralinear postsynaptic Ca2+ transients can be generated
through local electrical stimulation of excitatory synapses in distal dendrites. These Ca2+ transients were mediated by Ca2+ influx from the activation of Ca2+-permeable
AMPA receptors, which triggers Ca2+ release through ryanodine receptors present
on intracellular Ca2+ stores (Ca2+-induced Ca2+ release). These Ca2+ signals also
contain a minor contribution from NMDA receptors, and stay localized (no significant propagation in the dendritic arbor). In addition, we determined that these supralinear Ca2+ signals constitute a switch in the expression of synaptic plasticity,
as they induce long-term depression in local synapses, while low-amplitude Ca2+ signals induced synaptic long-term potentiation. We also examined whether these supralinear Ca2+ transients were generated in both apical and basal dendrites,
which receive synaptic contacts from different sources (Schaffer collaterals vs local collaterals). We observed that Ca2+ transients in apical dendrites had a higher
amplitude and were associated with a higher level of somatic depolarization. We were also able to predict, through computational modeling, the number of synapses necessary to the generation of those signals and the potential contribution of Ca2+ extrusion mechanisms. Finally, we studied the cell-specificity
of dendritic integration mechanisms by combining Ca2+ imaging and modeling in a
different interneuron type, interneuron-specific interneurons type III. In conclusion, we were able to prove that certain interneurons possess alternative mechanisms, mediated through local Ca2+ transients, that allow for the regulation of plasticity at
Table of contents
Résumé ... ii
Abstract ... iii
Table of contents ... iv
List of figures and tables ... vi
List of abbreviations and acronyms ... viii
Acknowledgments ... xii
Preface ... xiii
Hypothesis and objectives ... 47
Chapter 1 Two-photon Calcium Imaging in Neuronal Dendrites in Brain Slices ... 48
1.1 Résumé ... 48 1.2 Abstract ... 48 1.3 Introduction ... 49 1.4 Protocol ... 52 1.5 Representative Results ... 60 1.6 Discussion ... 60 1.7 Figures ... 63 1.8 References ... 67
Chapter 2 Dendritic calcium nonlinearities switch the direction of synaptic plasticity in fast-spiking interneurons ... 70
2.1 Résumé ... 70
2.2 Abstract ... 70
2.3 Introduction ... 71
2.4 Materials and Methods ... 72
2.5 Results ... 77
2.6 Discussion ... 87
2.7 Figures ... 93
2.8 References ... 103
Chapter 3 Mechanisms of supralinear calcium integration in dendrites of hippocampal CA1 fast-spiking cells ... 113
3.1 Résumé ... 113
3.2 Abstract ... 113
3.4: Materials and Methods ... 116
3.5: Results ... 125
3.6: Discussion ... 133
3.7: Figures ... 138
3.8: References ... 147
Chapter 4 Using A Semi-Automated Strategy To Develop Multi-Compartment Models That Predict Biophysical Properties Of Interneuron Specific 3 (IS3) Cells In Hippocampus ... 158
4.1 Résumé ... 158
4.2 Abstract ... 158
4.3 Introduction ... 159
4.4 Materials and Methods ... 161
4.5 Results ... 170
4.6 Discussion ... 184
4.7 Figures and Tables ... 194
4.8 References ... 223 Conclusion... 231 Bibliographie ... 248 Annexe A ... 290 Annexe B ... 311 Annexe C... 333
List of figures and tables
Figure 1: The hippocampal formation ... 3
Figure 2: Sharp-wave ripple in the CA1 region of the hippocampus ... 6
Figure 3: Four types of fundamental spatial cell. ... 10
Figure 4: Classes of interneuron in the hippocampal CA1 area. ... 14
Figure 5: Confocally targeted recording from basket cells. ... 19
Figure 6: Ionotropic glutamate receptors and Ca2+ influx at different types of glutamatergic synapse. ... 33
Figure 7: Dendritic properties of PV+ interneurons ... 36
Figure 8: Localization of excitation by two-photon excitation ... 40
Figure 9: Neuronal Calcium signaling ... 43
Figure 1-1: Representative example of AP-evoked Ca2+ transients. ... 63
Figure 1-2: Postsynaptic Ca2+ transients evoked by a burst of electrical stimulation (3 stimuli at 100 Hz). ... 64
Figure 2-1: Postsynaptic Ca2+ mechanisms in distal dendrites of hippocampal CA1 FS interneurons. ... 93
Figure 2-2: Supralinear Ca2+ signals in distal dendrites of FS interneurons. ... 95
Figure 2-3: Supralinear Ca2+ signals are mediated by Ca2+-permeable AMPA receptors, with a minor contribution from NMDARs. ... 96
Figure 2-4: Ca2+ nonlinearities in distal dendrites do not involve voltage-gated Ca2+ and Na+ channels. ... 97
Figure 2-5: Calcium dependence of Ca2+ nonlinearities. ... 99
Figure 2-6: Ca2+ nonlinearities are produced by Ca2+-induced Ca2+ release. ... 100
Figure 2-7: Spatial profile of Ca2+ nonlinearities. ... 101
Figure 2-8: Ca2+ nonlinearities control the direction of plasticity. ... 102
Figure 3-1: Two-photon imaging of dendritic Ca2+ transients in hippocampal CA1 FS cells... 138
Figure 3-2: Supralinear summation of postsynaptic Ca2+ transients in apical and basal dendrites of FS cells. ... 140
Figure 3-3: Postsynaptic Ca2+ sources involved in the generation of the supralinear
Ca2+ events in apical vs basal dendrites of FS cells. ... 142
Figure 3-4: Computational model of the supralinear Ca2+ events in dendrites of FS interneurons. ... 144
Figure 3-5: Computational simulations of the supralinear Ca2+ signals in dendrites of FS interneurons. ... 145
Figure 4-1: Morphological, membrane and dendritic properties of IS3 cells. ... 194
Figure 4-2: Semi-automated strategy. ... 196
Figure 4-3: Experimental measurement histograms. ... 198
Figure 4-4: Semi-automated strategy and model databases. ... 199
Figure 4-5: Example adjustments made in between parameter refinement cycles for SDprox.1 and SDprox.2 model databases. ... 202
Figure 4-6: Kv3.1 and Kv2.1 expression in putative IS3 cells. ... 203
Figure 4-7: VGC distributions in Proximal Dendrites ... 205
Figure 4-8: Electrotonic Analysis of the M2 Morphology. ... 207
Figure 4-9: Threshold weight with distance from soma in top models. ... 209
Table 1-1: Solution recipes. ... 65
Table 1-2: Troubleshooting table. ... 66
Table 4-1: Model passive properties. ... 210
Table 4-2: IS3 cell signature features, CIPs, measurements, and model pre-processing criteria ... 212
Table 4-3: Statistical Tests. ... 213
Table 4-4: Voltage-gated channel equations ... 216
Table 4-5: Summary of channel type combinations and spatial distribution profiles across the morphology of the model. ... 217
Table 4-6: Summary of the starting and final conductance ranges found using the semi-automated strategy for S.1, S.2 and SD. ... 219
Table 4-7: Summary of starting hand-tuned conductance values and top model conductance values from SDprox.1 and SDprox.2. ... 221
List of abbreviations and acronymsAA : axo-axonic cell
AM : acetoxymethyl
AMPA : α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid AP : action potential
ATP : adenosine triphophate
bAP : back-propagating action potential
BAPTA : 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid BC : basket cell
BIS : bistratified cell
BK : Big potassium channel CA1 : Cornu Ammonis 1
CA2 : Cornu Ammonis 2
CA3 : Cornu Ammonis 3
CaNL: Ca2+ non-linearity
CaT : Ca2+ transient
CCK : cholecystokinin
CCKBC : cholecystokinin-expressing basket cell CGE : caudal ganglionic eminence
CI-AMPAR : calcium-impermeable AMPA receptor CICR : Ca2+-induced Ca2+ release
CNS: central nervous system
CP-AMPAR : calcium-permeable AMPA receptor CR : calretinin
DG : dentate gyrus EC : entorhinal cortex
EEG : electroencephalography
EPSP : excitatory postsynaptic potential ER : endoplasmic reticulum
GECI : genetically-encoded calcium indicator
GIRK : G protein-coupled inwardly-rectifying potassium channel IP3R : inositol triphosphate receptor
IR : infrared
ISI : interneuron-specific interneuron KAR : kainate receptor
LM : stratum lacunosum-moleculare
LTD : long-term depression LTP : long-term potentiation
mAChR : muscarinic acetylcholine receptor M2R : muscarinic acetylcholine receptor 2 MGE : medial ganglionic eminence
mGluR : metabotropic glutamate receptor nAChR : nicotinic acetylcholine receptor NCX : Na+/Ca2+ exchanger
NMDA : N-methyl-D-aspartate NMDAR : NMDA receptor
nNOS : neuronal nitric oxide synthase NPY : neuropeptide Y
O/A : stratum oriens/alveus
O-LM : oriens-lacunosum-moleculare PMCA : plasma membrane Ca2+ ATPase
PV : parvalbumin
PVBC : parvalbumin-expressing basket cell PYR : stratum pyramidale
RAD : stratum radiatum
REM : rapid eye movement RYR : ryanodine receptor
SCAC: Schaffer collateral-associated cell
SERCA : sarco/endoplasmic reticulum Ca2+-ATPase
SO : stratum oriens
STDP : spike-timing-dependent plasticity STSP : synapse-type-specific plasticity STIM : stromal-interacting molecule SWR : sharp-wave ripple
TPE : two-photon excitation
TRP : transient receptor potential channel VGCC : voltage-gated calcium channel VIP : vasointestinal peptide
List of units and symbolsCa : calcium
Cm : specific membrane conductance
Hz : hertz K : potassium kHz : kilohertz Mg : magnesium ms : millisecond mM : millimolar Na : sodium nA : nanoampere nM : nanomolar Rm : membrane resistivity Ri : axial resistivity s : second µM : micromolar µm : micrometer/micron
I would like to thank my supervisor, Lisa Topolnik. I could not have made it here without your constant advice, support and encouragement.
I would like to thank my colleagues and labmates for their help, warmth and positivity. Thank you to Sona, Étienne, Ruggiero, Xiao, Vincent, Linda, Einer, Élise, Émilie, Stéphanie, Audrée, Maelle, Katya, Dmitry, Marie-Andrée, Simon and to the many others that made our lab a great environment to work in.
I would like to thank my collaborators, particularly Alex Guet-McCreight, Ivan Lazarevich and Frances Skinner, for their feedback and for being willing to share their immense expertise with me.
I would like to thank NSERC for providing funding for my work.
And finally, I would like to thank my parents, Marc and Lucie, and my brother, Laurent, without whom none of this would have been possible. Merci!
This work contains 4 articles, presented as separate chapters, detailing my work over the course of my PhD project.
The first article is titled “Two-photon Calcium Imaging in Neuronal Dendrites in Brain Slices”.
Publication status: Published, Journal of Visualized Experiments, March 15th 2018 Author status: First author
Role in the preparation of the article: I performed the experiments presented in the video article and prepared most of the written article (text and figures).
Coauthors: Lisa Topolnik
The second article is titled “Dendritic calcium nonlinearities switch the direction of synaptic plasticity in fast-spiking interneurons”.
Publication status: Published, Journal of Neuroscience, March 12th 2014 Author status: First author
Role in the preparation of the article: I performed most of the experiments included in the article, did most of the analysis and participated to the preparation of the written article (text and figures).
Coauthors: Lisa Topolnik
The third article is titled “Mechanisms of supralinear calcium integration in dendrites of hippocampal CA1 fast-spiking cells”.
Publication status: Published, Frontiers in Synaptic Neuroscience, December 11th 2018
Author status: First author
Role in the preparation of the article: I performed most of the experiments included in the article, did most of the analysis and participated to the preparation of the written article (text and figures).
Coauthors: Ivan Lazarevich, Tommy Gilbert, Lisa Topolnik
The fourth article is titled “Using a Semi-Automated Strategy to Develop Multi-Compartment Models That Predict Biophysical Properties of Interneuron-Specific 3 (IS3) Cells in Hippocampus”.
Role in the preparation of the article: I performed experimental manipulations, participated to the analysis and participated to the preparation of the article (text and figures).
Coauthors: Alexandre Guet-McCreight*, Lisa Topolnik, Frances K. Skinner
Three review articles that we published on Ca2+ signaling in interneurons are also
1. The hippocampus
1.1 Anatomy of the hippocampus
The hippocampus is a cortical region located in the brain’s medial temporal lobe. Initially identified by Italian anatomist Giulio Cesare Aranzio (1587), its relatively simple laminar organization and its apparent role in memory consolidation made it an important object of study in multiple disciplines, including neuroanatomy, physiology and psychology. Its internal organization is typically described by using the terminology of Rafael Lorente de Nó (Lorente De Nó, 1934), which subdivided the hippocampus into five anatomically distinct subregions: the dentate gyrus (DG), the Cornu Ammonis subfields (CA1, CA2 and CA3) and the subiculum. It is a part of the hippocampal formation, a region that groups the hippocampus with adjacent cortical structures to which it is functionally associated: the parasubiculum, the presubiculum and the entorhinal cortex.
While multiple characteristics differentiate the hippocampus from other cortical areas, the most striking from a neuroanatomical standpoint is the seemingly unidirectional orientation of its excitatory inputs. While major excitatory connections between cortical areas are typically reciprocal, principal cells in the hippocampal formation mostly send their axonal projections in a given direction: from the entorhinal cortex (EC) to the dentate gyrus (DG), from the DG to the CA3 and from the CA3 to the CA1 (Amaral and Lavenex, 2007). This led early neuroanatomists to consider the hippocampus a relatively simple model of a cortical network. The region has since become one of the most widely studied regions in the brain. It should be noted that most of our detailed anatomical knowledge of the hippocampus comes from studies in rodents and that while the structure of the region is largely analogous in all mammals, there are multiple species-specific differences in terms of its organization and connectivity (Amaral and Lavenex, 2007).
Information, arriving via excitatory inputs, largely enters the hippocampus proper though the perforant path. The perforant path comprises a set of excitatory projections that arise mostly from layer II-III of the EC and innervates principal cells in all hippocampal subfields. This pathway can be divided based on its origin: fibers that come from the EC layer II mostly terminate in the molecular layer of the dentate gyrus and in the lacunosum-moleculare layer of the CA3/CA2 area, while fibers that come from the EC layer III innervate the lacunosum-moleculare layer of the CA1 and the subiculum (this set of projections is often called the temporoammonic pathway) (Witter et al., 2000).
Granule cells (principal cells of the dentate gyrus) send their axonal projections to the CA3, where they innervate a layer called stratum lucidum. These projections were named "mossy fibers", because the large terminals along the axons gave them a mossy appearance (Ramon y Cajal, 1911). CA3 pyramidal cells then send their own projections, the Schaffer collaterals, to the CA1 area, where they terminate in the radiatum layer. CA1 pyramidal cells, for their part, project mostly to the subiculum and the entorhinal cortex.
Of course, this representation of information flow in the hippocampus, often called the trisynaptic circuit, is simplified, as the region connects to multiple other areas, notably the amygdala, the septum, the prefrontal cortex and the contralateral hippocampal formation (Van Groen and Wyss, 1990; Pitkanen et al., 2000; Thierry
et al., 2000; Unal et al., 2015). It also excludes CA2, an area that resembles CA3
but features a distinct connectivity with other hippocampal areas (Reviewed in Robert et al. 2018). But it informed the common view of the hippocampus as a loop of information processing starting and ending in the entorhinal cortex.
Figure 1: The hippocampal formation
A. Schematic representation of the hippocampal formation showing connectivity between the
various regions of the hippocampus proper and the EC. B. Projections along the transverse axis of the hippocampal formation; the DG is located proximally and the EC distally. From Amaral and Lavenex, 2007
The CA1 area
Of all hippocampal subfields, the CA1 is the most widely studied, largely because of discoveries regarding its role in memory (see section 1.2), its association with
area. Like the other CA areas, it is highly laminar, containing 5 layers or strata: the alveus, stratum oriens, stratum radiatum and stratum lacunosum-moleculare. The alveus contains myelinated axons from pyramidal cells that project outside the hippocampus through the fimbria. The stratum oriens (SO) contains various types of GABAergic interneurons, the basal dendrites of CA1 pyramidal cells, projections from other hippocampal areas (notably CA2, which preferentially innervates SO; Shinohara et al., 2012) and projections from the septum and amygdala. It is frequently grouped together with the alveus (O/A). The stratum pyramidale (PYR) is the cell-rich layer containing the vast majority of pyramidal cells and, by extension, the majority of cells in the area. The stratum radiatum (RAD) contains GABAergic interneurons, apical dendrites of CA1 pyramidal cells and Schaffer collateral inputs from CA3 pyramidal cells. Finally, the stratum
lacunosum-moleculare (LM) contains other GABAergic interneurons, the distal portion of
pyramidal apical dendrites and the terminals of perforant path inputs, coming mostly from the EC with some other extrinsic inputs (such as thalamic inputs from the nucleus reuniens and inputs from the basolateral nucleus in the amygdala; Krettek and Price, 1977; Kemppainen, Jolkkonen, and Pitkänen, 2002).
Pyramidal cells form the vast majority of the neuron population in the CA1 area. Their cell body is situated in the PYR and their dendrites span all CA1 layers. These dendrites are highly studded with spines, protrusions that form the postsynaptic portion of excitatory synapses. They are also modulated by different classes of inhibitory interneurons which selectively target different subcellular compartments of pyramidal cells (see section 2). CA1 pyramidal cells have a single axon that heads into the SO and alveus before leaving the area. While in SO, local axonal collaterals contact local GABAergic interneurons, but form relatively few associational connections with neighboring pyramidal cells (unlike CA3 pyramidal cells which form these connections at a high rate). After leaving the CA1, the main target of these axons is the subiculum where they arborize extensively, followed by the EC, where they settle principally in layers III to V. Subsets of pyramidal cells also send projections outside the hippocampal formation: to the retrosplenial
cortex, to the prefrontal cortex and to the amygdala, among others (Spruston and McBain, 2007).
CA1 pyramidal cells were long thought to be a homogeneous cell class, but recent studies have shown that different subtypes can be distinguished based on neurochemical, physiological and morphological differences. For example, deep pyramidal cells, located closer to SO, and superficial pyramidal cells, located closer to RAD and identified through their expression of calbindin, show significant differences in their physiological properties (deep cells fire at higher rates and burst more frequently; Mizuseki et al. 2011). They also had different patterns of connectivity and were modulated differentially during specific brain states, such as those brought on by sleep or goal-oriented learning (Mizuseki et al., 2011; Lee et
al., 2014; Danielson et al., 2016). Other studies find a gradual differentiation of
pyramidal cells along the dorso-ventral axis of the CA1 region (Cembrowski et al., 2016). Overall, this suggests that CA1 pyramidal cells should be considered a diverse heterogeneous population rather than a discrete cell type.
1.2 Function of the hippocampus: oscillatory activity
As information in the brain is conveyed though electrical activity, electroencephalography (EEG) has been a common method to investigate brain function since the late 19th century. Since the activity of individual neurons could
not be studied at that time, early investigators, through larger-scale recordings, sought to distinguish distinct patterns of electrical activity in specific areas in order to understand the mechanisms of brain function. Identified oscillatory rhythms were classified based on their frequency range: theta (6-12 Hz), beta (12-30 Hz), gamma (30-150 Hz) and ripples (100-200 Hz).
In the hippocampus, early EEG studies revealed a prominent slow oscillation in the theta frequency band that was associated with specific animal behaviors. Initially discovered in the rabbit (Jung and Kornmüller, 1938), this theta rhythm was then found to be present in all mammals studied, including rodents (Vanderwolf, 1969)
and humans (Arnolds et al., 1980). While there are slight differences in theta-associated behaviors between different animals, the hippocampal theta rhythm is mostly present during rapid eye movement (REM) sleep (Jouvet, 1969) and during exploratory activity (Vanderwolf, 1969). Further investigations revealed that hippocampal theta frequency and amplitude correlate with the animal’s running speed (Whishaw and Vanderwolf, 1973) and that hippocampal theta is involved in learning and memory (Berry and Thompson, 1978; Seager et al., 2002).
Figure 2: Sharp-wave ripple in the CA1 region of the hippocampus
Simultaneous recordings from the CA1 pyramidal layer (electrode 1) and stratum radiatum (electrode 2). Uppermost trace of electrode 1 is the wide-band recording (1 Hz to 10 kHz). Second and third traces are digitally filtered derivatives of the wide-band trace (unit activity 500 Hz to 10 kHz and fast field oscillation 100 to 400 Hz). Note the co-occurrence of ripples in PYR (third trace) and the sharp wave in RAD (fourth trace). From Buzsáki et al., 1992
While theta is predominant during movement and exploration, sharp-wave ripples are the primary oscillatory activity seen in the hippocampus during immobility, slow-wave sleep, eating and drinking (Buzsáki, Leung and Vanderwolf, 1983). Thought to be an intrinsic hippocampal activity, they occur when a burst of activity in CA3 pyramidal cell ensembles leads to a depolarization of CA1 apical dendrites (seen as a sharp wave in the local field potential centered on the SR). These sharp waves are associated with short (50-100 ms) oscillations in the 110-200 Hz range with maximum amplitude in the CA1 PYR (Buzsáki et al., 1992). While they appear to occur synchronously in multiple cell ensembles, these events are ordered, with neurons firing in the same sequence as they do during the preceding theta activity (Skaggs and McNaughton, 1996; Lee and Wilson, 2002). Hence sharp-wave ripples might represent the state of the hippocampus during memory consolidation (see section 1.3).
The hippocampus features other types of rhythmic activity, notably beta and gamma waves. Gamma waves are oscillations in the 30-150 Hz range which can co-occur with both theta rhythm and sharp-wave ripples. The gamma rhythm can be split between three subtypes: slow gamma (30-60 Hz), mid-frequency gamma (60-90 Hz) and fast gamma (90-150 Hz), which are thought to originate in the CA3, EC and CA1 respectively (Bragin et al., 1995; Colgin et al., 2009; Schomburg et
al., 2014). The specific function of these rhythms is still unclear, but hippocampal
gamma may be involved in the retrieval of hippocampus-dependent memories (Montgomery and Buzsaki, 2007; Shirvalkar, Rapp and Shapiro, 2010; Yamamoto
et al., 2014). The neocortical gamma rhythm has also been linked to attention
(Fries et al., 2001; Gregoriou et al., 2009). Beta oscillations are more rare, but are consistently elicited in the dentate gyrus as a response to olfactory stimulation (Vanderwolf, 1992).
1.3 Function of the hippocampus: behavior
Research into hippocampal function has been split for much of the last half-century. On one hand, studies in the region uncovered a role for the hippocampus
in declarative memory, which involves the conscious recollection of facts and events. On the other hand, past research (particularly studies performed on rodents) indicated that the region was involved in spatial navigation and cognitive mapping. Here I will briefly describe both theories and the recent attempts to reconcile them.
The hippocampus and declarative memory
Initial hints on the purpose of the hippocampal area came from clinical observations of brain-damaged patients. The most famous of these, patient H.M., suffered from temporal lobe epilepsy and underwent a bilateral resection of the medial temporal lobe to treat it. After the operation, H.M. suffered from severe anterograde amnesia (an inability to form new memories) and limited retrograde amnesia (an inability to recall memories from 1-2 years before the operation) (Scoville and Milner, 1957). The results of that surgery, and the various cases of other similar patients, led to the conclusion that damage to the hippocampus impaired the retention of memories. However, the effect on memory was not universal: patients forgot facts and events (items that could be verbally expressed, or “declared”) but retained skills, habits and facets of behavior even when those were acquired during the forgotten period. From this, Cohen and Squire elaborated a theory separating memory into separate types, including “declarative” and “nondeclarative”, that could be localized to separate areas in the central nervous system (CNS) (Cohen and Squire, 1980). The hippocampal formation was grouped together with the surrounding perirhinal and parahippocampal cortices into a discrete memory structure dedicated to processing declarative memory: the medial temporal lobe system.
As patients like H.M. did not suffer from complete retrograde amnesia, researchers concluded that this structure could not constitute the long-term storage site of declarative memory. As it was not involved in acquisition or long-term storage of information, studies on hippocampal function started focusing on memory consolidation (Glickman, 1961). Initially proposed early in the 20th century, memory
consolidation describes a process of reorganization and stabilization of information guided by the hippocampus. Memories are distributed in strongly interconnected neocortical areas and their recall becomes gradually independent of hippocampal activity (Squire et al., 2015). Consolidation is thought to occur in particular during slow-wave sleep, through mechanisms such as ‘‘replay’’, the reactivation of previous neuronal firing sequences (Wilson and McNaughton, 1994) and ‘‘downscaling’’, the homeostatic regulation of synaptic weight (Tononi and Cirelli, 2014). Essentially, this suggests that the medial temporal lobe system acts as a memory guide, pointing to the relevant pieces of information in the neocortex and pushing them to link together to form stable memory traces.
Since this theory was elaborated, a number of studies have highlighted the role of the hippocampus in various types of memory: spatial memory (Burgess, Maguire and O’Keefe, 2002), social memory (Hitti and Siegelbaum, 2014; Okuyama et al., 2016) and contextual fear memory (Amaral et al., 2007; Liu et al., 2012). With that said, there is evidence supporting the idea of functional segregation along the hippocampal dorsoventral axis in regards to memory. Differences in gene expression and connectivity suggest that the dorsal hippocampus may be more heavily involved in cognitive function while the ventral hippocampus is more implicated in emotion and affect (Moser and Moser, 1998; Fanselow and Dong, 2010). This hints that we should be careful not to treat the hippocampus as a unitary structure.
The hippocampus and spatial navigation
The idea of the hippocampus being involved in spatial navigation was spurred by the discovery of “place cells” in the rat CA1 area. Those neurons are principal cells whose activity correlates closely with the presence of the animal in a particular location, called “place field”. O’Keefe and Nadel used the discovery to elaborate the cognitive mapping theory of hippocampal function, based on the idea that spatial information is encoded into hippocampal neurons and that the collective firing of place cells can form a cognitive map of the animal‘s environment (O’Keefe
and Nadel, 1978). This theory was since bolstered by the discovery of other spatially-modulated cell types in the hippocampal formation. These include head-direction cells in the presubiculum, cells that fire when the animal’s head has a given orientation (Taube, Muller and Ranck, 1990), and grid cells in the medial entorhinal cortex, neurons that are active in multiple locations that, when put together, form a regular grid-like pattern (Hafting et al., 2005). There are also boundary cells, cells whose firing field parallels environmental obstacles like walls or drop edges, which are found in the subiculum, in the medial entorhinal cortex, in the presubiculum and the parasubiculum (Lever et al., 2009; Boccara et al., 2010).
Figure 3: Four types of fundamental spatial cell.
Figure shows one example of each type of fundamental spatial cell: (a) place cell; (b) head-direction cell; (c) grid cell; (d) boundary cell. For each cell: left-hand column shows locational firing ratemap (a,c,d) or directional firing polar plot (b), with peak firing rate in hertz shown top left; right-hand column depicts path taken over whole trial (black line), on which are plotted the locations at which spikes were recorded (green squares). In firing rate maps, one of five colours in locational bin indicates spatially smoothed firing rate in that bin (autoscaled to firing rate peak; dark blue, 0–20%; light blue, 20–40%; green, 40–60%; yellow, 60–80%; red, 80– 100%). From Hartley et al., 2013
Alternatively, encoding of spatial information is not done purely through the activity rate of neurons. Hippocampal cells ensembles are temporally coordinated to fire in internally-generated sequences (Dragoi and Buzsáki, 2006; Pastalkova et al., 2008), sequences that may be necessary to plan trajectories (Frank, Brown and Wilson, 2000). What’s more, it was found that place cells fire during progressively earlier phases of the hippocampal theta rhythm as they move further into the cell’s place field, a phenomenon called phase precession (O’Keefe and Recce, 1993). Hence the animal’s position within a trajectory and within a given place field are both encoded temporally.
The fact that so many hippocampal cells respond to specific spatial cues suggests that the hippocampal formation could be the center of a spatial representation system that can function to support navigation. Spatial navigation works through two distinct mechanisms: egocentric navigation, also called path integration, which relies on self-motion cues and knowledge of previous positions, and allocentric navigation, which relies on landmarks and a map-based reference frame (Buzsáki and Moser, 2013). The spatial and temporal ordering of principal cell activity may allow the hippocampus to participate in both egocentric and allocentric navigation, by providing a spatial reference frame and the internal temporal sequencing needed for path integration. Both mechanisms may be necessary for navigation depending on the availability of landmarks and other external cues.
The hippocampus and spatiotemporal processing
Setting aside the possibility that the hippocampus might be dedicated to distinct functions in different animals, investigators have advanced theories seeking to unite the role of the hippocampus in declarative memory and spatial navigation. One such view that is growing in prominence is the relational processing theory, which holds that the hippocampus organizes events, objects and actions by relating them to their spatial and nonspatial context, thus making it both essential to memory and dependent on spatial information (Eichenbaum, 2017). This theory was bolstered by the discovery of time cells in the rat CA1, principal cells that fire
at a specific point in a temporal sequence of events (MacDonald et al., 2011). Consequently, a purely spatial view of hippocampal function may be too narrow, as the region processes both spatial and temporal information in parallel. According to the relational processing theory, an episodic memory unfolding in a set temporal order could be encoded in a similar way as the encoding of a route through place cell firing, linking the two main theories of hippocampal function together. Interestingly, some studies suggest that spatial and temporal information may converge in the CA1 area, making it a major region of interest for the study of how the brain integrates contextual information (Kesner, Hunsaker and Gilbert, 2005; Mankin et al., 2012).
2. Interneurons of the hippocampus 2.1 A heterogeneous cell type
As stated in section 1, the majority of neurons in the hippocampus are excitatory pyramidal cells; the rest (10-15%) are interneurons (Bezaire and Soltesz, 2013). These cells mainly innervate local neurons and release the inhibitory neurotransmitter GABA (γ-Aminobutyric acid). The most striking feature of this population is its heterogeneity: as noticed by Ramon y Cajal (1911) and his student Lorente de No (1934), nonpyramidal cells in the hippocampus showcase a large variety of anatomical characteristics (size, shape, orientation, etc). This diversity, which also extends to their connectivity and physiological properties (Pelkey et al., 2017), has led researchers to classify hippocampal interneurons into multiple subtypes, with as many as 21 in the CA1 alone (Klausberger and Somogyi, 2008a). This classification relies on different factors: cell anatomy, cell neurochemistry (expression of different markers like parvalbumin or somatostatin), cell physiology (characteristic firing patterns, passive membrane properties) and cell connectivity (such as innervation of specific cell types or cell domains). It should be noted that these cell types are not all unique to the hippocampus; many have homologs in other cortical brain regions.
Why are there so many different types of interneurons in the hippocampus? Interneuron diversity relates to functional diversity. Through gamma-aminobutyric acid (GABA)-mediated inhibition, interneurons are involved in all facets of hippocampal function, acting as dynamic regulators of pyramidal cell activity. As discussed in section 1, pyramidal cells integrate multiple spatially-segregated inputs to fire in coordination within large cell ensembles. Interneuron inhibition provides the precise spatial and temporal control that makes such activity possible. Hence, interneuron diversity represents specialization, with each subtype having different regulatory functions. Those range from regulating action potential generation (basket cells) to regulating dendritic integration and dendritic spike initiation (bistratified cells) and even inhibiting other interneurons (interneuron-specific cells). Hence, specialization leads to a division of labor, and allows different interneuron functions to have separate input/output relationships within cortical networks.
The first level of interneuron specialization occurs early in development. The vast majority of cortical interneurons are derived from the ganglionic eminences, fetal brain structures located in the telencephalon (Anderson et al., 1997). More specifically, they originate in the medial ganglionic eminence (MGE) and in the caudal ganglionic eminence (CGE). Researchers later determined that their point of origin influences their specialization: parvalbumin- (PV). somatostatin- (SST) and neuronal nitric oxide synthase- (nNOS) expressing cells mostly come from the MGE (Butt et al., 2005; Fogarty et al., 2007; Tricoire et al., 2010) while cells that express reelin, calretinin (CR) and the vasoactive intestinal peptide (VIP) mostly come from the CGE (Lee et al., 2010; Miyoshi et al., 2010). Development and maturation of interneurons from their progenitor cells requires precise molecular signaling which involves a multitude of cell-type specific transcription factors, including notably Nkx2.1 for MGE-derived interneurons (Butt et al., 2008) and Prox1 for CGE-derived interneurons (Miyoshi et al. 2015; for further review, see DeBoer and Anderson 2017).
Interneurons cell types often have highly preserved properties and functions across different brain structures and can be viewed as homologous (such as PV-expressing fast-spiking cells in the hippocampus and neocortex). Regardless, they are classified in a region-specific manner, as their function ultimately depends on the nature of the circuit they are integrated in. Of all hippocampal areas, the CA1 is the one where interneurons have been most studied and arguably the one where their role is best understood (more than 1500 articles in Pubmed as of 2018). Consequently, I will focus on describing CA1 interneuron types.
Figure 4: Classes of interneuron in the hippocampal CA1 area.
Scheme showing the soma, dendrites, axons and main synaptic terminations of 21 CA1 interneuron subtypes. The main termination of five glutamatergic inputs are indicated on the left. From Klausberger and Somogyi, 2008
The CA1 contains 3 main subtypes of PV-expressing interneurons: axo-axonic cells, basket cells and bistratified cells. Axo-axonic cells, also called “chandelier cells” because of the characteristic shape of their axon terminal field, specifically target the axon initial segment of CA1 pyramidal cells (Somogyi et al., 1983; Li et
al., 1992). Located in O/A or PYR, these interneurons can form axo-axonic
contacts on as many as 1200 different pyramidal cells, and their dendritic arbor extends to all CA1 layers, enabling them to receive both entorhinal and CA3 excitatory inputs (Buhl et al., 1994).
Bistratified cells are named as such because their axonal arborization is localized within two layers: O/A and RAD, an arborization that largely overlaps with the location of CA3 excitatory inputs (Buhl, Halasy and Somogyi, 1994). They form inhibitory synapses on proximal dendrites of pyramidal cells and their dendrites extend to O/A and RAD, generally avoiding LM (Maccaferri et al., 2000). They are also known to be connected to other interneurons through synaptic contacts (Pawelzik, Hughes and Thomson, 2003) and through gap junctions (Baude et al., 2007).
PV-expressing basket cells are interneurons specialized to innervate the soma of pyramidal cells. Their axon forms a dense field localized mainly within PYR, with occasional collaterals in O/A and RAD (enabling them to also form proximal dendritic contacts, to a variable extent; Pawelzik, Hughes, and Thomson 2002)). Their dendrites mostly reside in the O/A and RAD layers, and rarely branch into LM. Aside from pyramidal neurons, they also innervate other PV+ basket cells (Cobb et al., 1997) and are interconnected through gap junctions (Fukuda and Kosaka, 2000). Their anatomical characteristics are similar to those of axo-axonic cells, making them hard to distinguish without observation of terminals through electronic microscopy.
oriens-lacunosum-moleculare (O-LM) cells also express PV, albeit at a lower level than the three previously described cell types (Ferraguti et al., 2004). They will be discussed in the following segment. PV-expressing interneurons will be discussed further in section 2.2.
There are three different cell types known to express SST in the CA1 area: bistratified cells (described in the preceding section), O-LM cells and hippocampo-septal cells.
O-LM cells are named based on their anatomy: their cell body and horizontally-oriented dendrites are located in the O/A, while their axon forms a dense terminal cloud in LM (Mcbain, Dichiara and Kauer, 1994). They mainly contact the distal dendrites of pyramidal cells (Maccaferri et al., 2000) and make a small number of contacts on other inhibitory interneurons (Katona, Acsády and Freund, 1999; Kogo
et al., 2004). Another population of hippocampal interneurons send axonal
projections to the medial septum (MS) (Gulyás et al., 2003). These SST-expressing cells (Jinno and Kosaka, 2002) are located in O/A and target other interneurons, both locally and in the MS (Gulyás et al., 2003).
Four different types of interneurons express cholecystokinin (CCK): CCK-positive basket cells, Schaffer collateral-associated cells, apical dendrite-innervating cells and perforant path-associated cells (Klausberger, 2009).
Like their PV-expressing counterparts, CCK-expressing basket cells (CCKBCs) contact pyramidal cells perisomatically through a large axonal field limited to the PYR (Harris, Marshall and Landis, 1985). Most CCKBCs reside in RAD, and can coexpress VIP (vasointestinal peptide; Acsády, Görcs, & Freund, 1996). The other CCK-expressing interneurons in the CA1 mostly target the dendrites of pyramidal neurons. Schaffer collateral-associated cells (SCACs) reside in RAD, where their
axon targets the oblique and basal dendrites of pyramidal neurons (Vida et al., 1998). Apical dendrite-targeting interneurons are similar to SCACs, but preferentially innervate the apical dendritic shaft of pyramidal cells (Klausberger et
al., 2005). Perforant path-associated cells are located at the RAD/LM border and
their axon mostly innervates LM, overlapping with excitatory inputs from the entorhinal cortex (Vida et al., 1998).
Neuropeptide Y-expressing interneurons
Neuropeptide Y (NPY) is expressed by bistratified cells, but also by two other interneuron populations: neurogliaform cells and ivy cells. These two populations can be distinguished by their location: neurogliaform cells are located in LM (Khazipov, Congar and Ben-Ari, 1995; Vida et al., 1998), while ivy cells mostly reside in the PYR (Fuentealba et al., 2008). Both cell types possess a remarkably dense axonal arborization which doesn’t extend far from the soma. Interestingly, ivy cells are thought to be the most numerous interneuron cell type in the CA1, representing almost a quarter of interneurons in the CA1 (Bezaire and Soltesz, 2013).
While the previously described cell types mainly innervate principal cells, interneuron-specific interneurons (ISIs), as their name suggests, have been found to specifically target other interneurons (Acsády, Arabadzisz and Freund, 1996; Acsády, Görcs and Freund, 1996; Gulyás, Hájos and Freund, 1996). They have so far been classified into 3 different groups. Type 1 ISIs express calretinin, are generally located in RAD and contact other RAD interneurons. Type 2 ISIs express VIP, are located at the border of RAD and L/M and contact mostly CCK/VIP expressing cells in RAD (Acsády, Arabadzisz and Freund, 1996; Acsády, Görcs and Freund, 1996). Type 3 ISIs express both calretinin and VIP, are located in RAD and PYR and send their axon to O/A, where they preferentially contact dendrite-targeting interneurons, such as O-LMs (Acsády, Arabadzisz and Freund, 1996; Acsády, Görcs and Freund, 1996; Tyan et al., 2014). Another recently
discovered cell type also fits this category: VIP-expressing long-range-projecting interneurons (VIP-LRPs). These cells innervate CA1 interneurons and both subicular pyramidal cells and subicular interneurons.
Some identified interneuron types do not fit within the previously described groups. Trilaminar cells are located in O/A and have an axon that innervates O/A, PYR and RAD, as well as the subiculum. They also express the muscarinic acetylcholine receptor 2 (M2R) and metabotropic glutamate receptor 8 (mGluR8; Sik et al., 1995; Ferraguti et al., 2005). Back-projecting cells are named for their axonal arborization, which innervates both CA1 and CA3 (Sik et al., 1994). They are also time-locked to theta oscillations, suggesting they may be involved in the coordination of the two regions (Jinno et al., 2007).
This description of interneuron diversity in the CA1 region is not meant to be exhaustive. Indeed, many rarer cell types have not been extensively characterized, and likely others have yet to be discovered. Still, it shows how the complex nature of inhibitory circuits can help them control and coordinate a number of different functions in a single hippocampal area.
2.2 Fast-spiking cells
Of all the interneuron types mentioned previously, PV-expressing interneurons have attracted the most attention. Consisting of axo-axonic cells (AAs), bistratified cells (BISs) and PV-expressing basket cells (PVBCs), they are also called fast-spiking (FS) cells, because of their non-accommodating, high frequency (>150 Hz) firing pattern (Freund and Buzsáki, 1996). They are uniquely well-positioned to control the output of their targets, particularly in the case of BCs and AAs, as they inhibit pyramidal cells perisomatically.
Figure 5: Confocally targeted recording from basket cells.
A. Confocal image (pseudocolor representation) of a patched BC filled with Alexa Fluor 488. B. Infrared differential interference contrast videoimage of the apical dendrite of the same cell as in (A). C. Light micrograph of a BC filled with biocytin during recording and labeled by using 3,3′-diaminobenzidine. 10-μm stack projection. Arrows indicate the axonal arbor, forming “baskets” around granule cell somata. D. Train of APs evoked by a 1-s, 0.75-nA current pulse applied at the soma (top) and AP frequency (f) – current (I) relation (bottom). Same cell as in (C). Bottom right graph shows mean maximal AP frequency (bar) and data from individual cells (points). E. (Left) Confocal micrograph of a BC filled with biocytin and stained with fluorescein isothiocyanate (FITC)– conjugated avidin; (center) parvalbumin immunoreactivity of the same BC; (right) overlay. 40-μm stack projection. From Hu et al., 2010a
FS cells receive the majority of their inputs from principal neurons: in the CA1 region more than 90% of synapses made on PV+ cells are excitatory (Gulyás et al., 1999). Their dendrites span all CA1 layers, but are localized mainly within O/A and
RAD, suggesting they are driven largely by collaterals from CA3 and local CA1 pyramidal cells. As for inhibitory inputs, studies on neocortical FS cells suggest that they come from other PV cells, with minor contributions from SOM+ and VIP+ interneurons (Hioki et al., 2013; Pfeffer et al., 2013). This PV+ interconnectivity extends to electrical synapses: some FS cells are coupled to each other through gap junctions (Galarreta and Hestrin, 1999; Bartos et al., 2001). This may help synchronize the FS population and enhance dendritic excitatory inputs.
FS cells also receive inputs beyond glutamatergic excitation and GABAergic inhibition. They express M1 muscarinic acetylcholine receptors (mAChRs) postsynaptically, and M2 mAChRs presynaptically (Hájos et al., 1998; Cea-del Rio
et al., 2010), allowing them to react to cholinergic inputs from the medial septum in
various ways (Bell, Bell and Mcquiston, 2015). Subsets of PV-expressing hippocampal neurons were found to express the D2 and D4 dopaminergic receptors (Andersson et al., 2012; Puighermanal et al., 2015), although it is still unclear whether FS cells receive a significant dopaminergic drive. They are also known to be modulated by opioids, though differentially: BCs and AAs express µ-opioid receptors, while BISs express δ-opioid receptors (Svoboda, Adams and Lupica, 1999; Drake and Milner, 2002).
As previously mentioned, FS cells mainly contact pyramidal cells, with their subcellular target depending on their cell type. PVBCs target the soma and proximal dendrites, BISs target dendrites and AAs target the axon initial segment. FS cells do not necessarily target pyramidal cells uniformly: a 2014 study proposed that PVBCs inhibit deep pyramidal cells more strongly, while receiving stronger excitatory input from superficial pyramidal cells (Lee et al., 2014). Accordingly, while these cells may at first glance be positioned to provide simply feedback inhibition, they may in fact form complex microcircuits organizing activity in different principal cell ensembles. A minority of synapses formed by FS cells are made on other interneurons, specifically other PV+ interneurons (Sik et al., 1995). The fact that FS cells connect mostly with each other suggests that the PV+ inhibitory
microcircuit may function with a large degree of independence from other local inhibitory neurons. FS cells also form autapses, synaptic connections a neuron makes onto itself (Cobb et al., 1997). It is still unclear whether autapses are vestigial or whether they have a functional role, but their presence is thought to help promote network synchronization (Connelly, 2014; Fan et al., 2018).
FS cells, as their name suggests, are optimized for fast signaling. This is due to several electrophysiological properties that set them apart from pyramidal cells and other interneurons. Their high frequency firing pattern is due to the elevated expression of voltage-gated K+ channels from the Kv3 family (Weiser et al., 1995;
Du et al., 1996). These channels are characterized by their fast activation and deactivation, allowing for a short repolarization period between spikes (Rudy and McBain, 2001). Their dendrites express a low density of voltage-gated Na+
channels, and a high density of Kv3 K+ channels (Hu, Martina and Jonas, 2010).
The high number of fast K+ channels quickens the decay of excitatory postsynaptic
potentials (EPSPs) and so shortens the EPSP summation window, suggesting that FS cells are coincidence detectors, cells activated only by temporally precise inputs. The axons of FS cells contain a very high density of Na+ channels (Hu and
Jonas, 2014), mostly Nav1.1 and Nav1.6 channels (Ogiwara et al., 2007; Lorincz
and Nusser, 2008). This ensures fast action potential (AP) propagation and high firing frequency. FS axons also exclusively express presynaptic Ca2+ channels of
the P/Q subtype (Hefft and Jonas, 2005), which have high efficacy, fast gating and are selectively activated by short APs (Li, Bischofberger and Jonas, 2007). All these properties suggest that FS cells are adapted to receive, integrate and send out fast, temporally precise signals.
FS cells also stand out from other interneurons from a bioenergetic point of view. Indeed, their activity is particularly energy-intensive: multiple studies have noted that FS cells contain a particularly high number of mitochondria (Ribak and Anderson, 1980; Gulyás et al., 2006; Fitzgerald et al., 2012; Takács et al., 2015). This is in-line with the expression of the FS phenotype: multiple aspects of action
potential firing consume ATP (neurotransmitter release, restoration of membrane ion gradients, etc.). Hence it is logical that cells specialized to fire at a high frequency would possess an enhanced energy infrastructure.
Regarding studies focused on PV+/FS cells, it is worth noting that, because of the rarity of AAs compared to the other FS cell types (Bezaire and Soltesz, 2013) and the fact that BISs may not uniformly express PV (Ferraguti et al., 2004), the properties of BCs may be over-represented in the findings cited here, and may hide small differences in how the three cell types function. This is a natural limitation of the current marker expression-based interneuron classification, and may require further investigation.
2.3 Function of fast-spiking cells
Knowing the properties of FS cells gives us a superficial understanding of what they do in the CA1: they provide a strong, temporally precise inhibitory drive to pyramidal neurons when activated through either local collaterals (feedback inhibition) or Schaffer collaterals (feedforward inhibition). But to truly understand their function, we need to know their contribution to network activity, which is represented by brain oscillations.
As mentioned in chapter 1, the prominent hippocampal oscillation during movement, exploration and REM sleep is theta. FS cells are known to be phase-locked to theta oscillations, with AAs most active at the theta peak, BCs slightly after and BISs most active at the trough, when pyramidal cells are most active (Klausberger et al., 2003, 2004). Both in vitro and in vivo studies have shown that PV+ interneurons can entrain pyramidal cells to fire at theta frequency, and play a role in the generation of the intrinsic CA1 theta rhythm (Cobb et al., 1995; Stark et
al., 2013; Amilhon et al., 2015). The cell-type-specific spike timing suggests they
might play different roles in that process, with AAs and BCs, firing when pyramidal cells are least active, being more likely to drive theta synchrony.
Cortical rhythmic activity in the gamma frequency range is associated with attention and working memory (Fries et al., 2001; Pesaran et al., 2002). Activity of FS cells has been associated with the generation of gamma oscillations in cortical areas (Bartos, Vida and Jonas, 2007; Sohal et al., 2009). In in vitro studies, FS cells in CA3 were found to be strongly phase-locked to gamma oscillations (Hajos
et al., 2004; Gloveli et al., 2005), and crucial to the generation of the slow gamma
rhythm in CA3 (Mann et al., 2005; Oren, Hájos and Paulsen, 2010). However, in the CA1, FS BCs are only weakly modulated by gamma and their activity is not essential to generation of the oscillation (Tukker et al., 2007; Craig and McBain, 2015). Of all interneuron subtypes, the one most strongly modulated by gamma in CA1 was the BIS type. This may be because BISs gate the integration of the Schaffer collateral inputs that bring the slow gamma oscillation to the CA1. Consequently, they would have a tight association with CA3 inputs. Overall, while FS cells drive gamma oscillations in cortical areas, CA1 FS cells may have different levels of involvement in generating this rhythm locally.
Sharp-wave ripples (SWR) occur mostly during immobility and slow-wave sleep. They represent a burst of principal cell activity, but some interneurons also fire during SWRs (Buzsáki, 1986). In vitro studies showed PVBCs and BISs are very active during ripples, while AAs are silent (Klausberger et al., 2003, 2004, 2005). This was later confirmed in vivo (Klausberger and Somogyi, 2008a). CA1 FS cells do not generate SWRs, but they can pace the spiking of pyramidal cell ensembles (Stark et al., 2014).
This shows that even just within the CA1 area, maintaining brain oscillations requires complex patterns of interneuron spiking. Therefore, it seems clear that FS cells play an essential part in the synchronization of pyramidal cell activity that gives rise to oscillations.
Interest in brain oscillations has been largely driven by their correlation with specific behavioral states. However, advances in optogenetics and in vivo
experimental tools have allowed teams to directly investigate the effect of FS cell activity on behavior and hippocampus-associated forms of cognition, such as memory and spatial orientation. Such a study showed that inhibition of CA1 PV+ interneurons affected spatial working memory, but not reference memory (Murray
et al., 2011). It also did not affect contextual fear conditioning, which was
dependent on SST+ interneuron activity (Lovett-Barron et al., 2014). In vivo research also touched on the relationship between interneurons and place cell activity, especially given that hippocampal interneurons can have their own place fields (Wilent and Nitz, 2007). An in vivo study demonstrated that while PV+ interneurons did not shape the size of pyramidal place fields, inhibiting them affected phase precession, leading to less precise spike timing of place field firing (Royer et al., 2012). Such a role involving temporal control of principal firing would be well-suited to FS cells, given their characteristic fast firing properties.
Overall, it is still difficult to parse out a common function of CA1 FS interneurons. The varied oscillatory responses of each subtype suggest they have different roles to suit their different targets. Given the difficulty of targeting subtypes individually in vivo, these differences may stay hidden at the behavioral level for now. However, it is clear that their temporally precise firing helps them synchronize local pyramidal cell activity in a way that is crucial for spatial cognition.
2.4 FS cells and disease
Given the role that interneurons play in forming and maintaining neuronal networks, we can expect that interneuron dysfunction may lead to pathological conditions. Indeed, as one of the most common groups of interneurons, abnormalities in FS cells have been linked to multiple disorders.
Epilepsy is a term used to describe a number of conditions that feature recurrent seizures, which are generally caused by excessive neuronal activity in cortical regions. The hippocampus, and the CA1 area in particular, is thought to be
especially vulnerable to damage in epilepsy (Freund et al., 1992). A study using animal models found that epileptic mice suffered a loss of PV+ interneurons in CA1 (Dinocourt et al., 2003). It was also shown that reducing FS BC excitability through deletion of the ErbB4 receptor could make the animal more vulnerable to the disease, suggesting a possible role for those cells in epileptogenesis (Li et al., 2011).
Schizophrenia is a complex psychiatric condition characterized by a number of social and cognitive deficits as well as manifestations of psychosis. Its causes are still poorly understood (and beyond the scope of this work), but many pieces of evidence point to interneuron dysfunction as an important part of the pathology (Lewis, Hashimoto and Volk, 2005). What’s more, multiple studies specifically link FS interneuron dysfunction with the disorder. A postmortem study of schizophrenic patients showed a significant loss of PV neurons in the hippocampus (Zhang and Reynolds, 2002). Another hint to FS interneuron involvement comes from the ErbB4 receptor, which is specifically expressed by PVBCs. Genetic variants of this protein, along with its ligand, neuregulin, have been conclusively linked to an elevated risk of schizophrenia (Mei and Nave, 2014). Finally, schizophrenic patients consistently show abnormally low levels of gamma oscillatory activity (Williams and Boksa, 2010). As mentioned in chapter 2.3, these oscillations are closely associated with FS interneuron activity in cortical regions. While the role of FS interneurons in schizophrenia is far from settled, they are likely to remain a prominent target of investigation going forward.
Autism spectrum disorder
Autism spectrum disorder (ASD) is a developmental disorder that affects social behavior and learning. Its causes are still unclear, but are thought to be both genetic and environmental in nature. In animal models of autism, mutations to the SHANK1 and CNTNAP2 genes both led to a loss of PV+ interneurons (Peñagarikano et al., 2011; Mao et al., 2015). In addition, studies of Rett