The presubiculum: Anatomy, function, microcircuit
The presubicular cortex
The presubiculum is a cortical region of the hippocampal-parahippocampal forma-tion in brain’s temporal lobe. Hippocampus proper is subdivided into the Dentate Gyrus, Ammon’s Horn (CA3, CA2, CA1) and Subiculum; The parahippocampal area is its continuation, composed of presubiculum, parasubiculum, entorhinal (me-dial and lateral parts), peri- and postrhinal cortices (Fig. 1; van Strien et al., 2009).
The presubiculum corresponds to Brodmann’s area 27 and 48 (Brodmann, 1909), following the temporoventral-to-septodorsal hippocampal axis in rodents (Fig. 1). The most dorsal part, corresponding to Brodmann’s area 48 is also called « Post-subiculum » (Brodmann, 1909; Rose and Woolsey, 1948; Blackstad, 1956). In the proximo-distal axis of the hippocampus (from dentate gyrus to subiculum), the pre-subiculum is located just next to subiculum and is then followed by the parasubicu-lum; these 3 areas being classically grouped together into the « subicular complex ». Eventually, in its retrodorsal part, the presubiculum is bordered by retrosplenial cortex.
The presubiculum is easily distinguishable from its neighboring areas regarding anatomical features such as cytoarchitecture and topography of aﬀerent fibers (Fig. 2, Ramon y Cajal, 1899; Brodmann, 1909; Rose and Woolsey, 1948; Blackstad, 1956).
Presubiculum is a 6-layered cortex. Layers were already described by Ramon y Cajal according to their neuronal content and density, from the pial surface to the white matter (Fig. 2C; Ramon y Cajal, 1899). This description of layer holds true for non-human primates and rodent presubiculum.
Layer 1, the molecular layer, is almost empty and contains only few putative interneurons, (Cajal’s « short axon cells ») and glial cells. Layer 2 is a thin layer of (B) views of the rat brain. Hippocampus contains dentate gyrus (DG), Ammon’s horn (CA1 to CA3) and subiculum (sub). Parahippocampal region is subdivided into pre-subiculum (PrS), parasubiculum (PaS), medial and lateral entorhinal areas (MEA and LEA) peri- and postrhinal cortices (PER and POR). Retrosplenial cortex is subdivided here in A29ab, A29c and A30 (Brodmann’s nomenclature). Hippocampus, PrS and PaS follow a dorsoseptal-to-ventrotemporal axis; Entorhinal cortices follow a dorsolateral-to-ventromedial (dl, vm); PER and POR are defined along a rostro-caudal axis. The dashed vertical (a, b) and horizontal (c, d) lines indicate levels of coronal and horizontal sections depicted in C. rf: rhinal fissure; cc: corpus callosum; f: fibria. Adapted from Sugar et al. (2011) and van Strien et al. (2009) approximately the same thickness as layer 1 and contains densely packed pyramidal and fusiform cells. In the most dorsal part of mouse presubiculum, layer 2 cell bodies tend to form clusters separated by fiber stripes (Slomianka and Geneser, 1991). Layer 3 is larger than layer 2 with a much lower neuronal density and is composed of pyramidal neurons. Layer 4 is also named « lamina dissecans » because it was described as a neuron free layer, containing only fibers and glial cells (Rose, 1926; Lorente De Nó, 1933). It is a convenient marker separating superficial layers (1, 2 and 3) from deep layers (5 and 6). Layer 5 is a layer with large to medium sized pyramidal cells whereas layer 6 contains smaller fusiform and pyramidal cells. In primate presubiculum, deep layers are separated in 3 sub-layers (5, 6, 7). This laminar organization has been observed with specific in situ hybridization stainings and is less clear in rodents (Ding, 2013).
The laminar organization of the presubiculum marks an abrupt transition with the adjacent subiculum, organized more like a cloud (even if subiculum may also be subdivided in diﬀerent layers, O’Mara et al., 2001). An « extremely dense plexus formed by [ ] many aﬀerent axons » in superficial layers of presubiculum distinguishes it from its neighbors, subiculum, parasubiculum and retrosplenial cortex (Ramon y Cajal, 1899; Blackstad, 1956). These terminals are more numerous in the dorsal part of presubiculum (area 48; Rose and Woolsey, 1948; Blackstad, 1956). The presubiculo-parasubiculum transition is marked by the absence of the densely packed layer 2 in parasubiculum, the cellular density of its superficial layers being more homogeneous. This transition is clearly visible with a specific marker of presubicular layer 2, calbindin (Boccara et al., 2010).
The dense presubicular layer 2 is also remarkably avoided by the characteristic plexus targeting the presubiculum (Fig. 2A and B, Ramon y Cajal, 1899). These dense aﬀerent fibers define very well the limits of presubiculum, especially dorsally, where their interruption marks the border of presubiculum with retrosplenial cortex (Ramon y Cajal, 1899; Blackstad, 1956).
What kind of cortex?
During development, radial migration of neuronal progenitors from the ventricu-lar zone shapes laminar compartments (Angevine and Sidman, 1961; Rakic, 1974). Then, subsequent change may occur to generate the adult cortical organization. Cortical areas may be classified according to the development of their laminar or-ganization and their aspect in the adult stage (Lorente De Nó, 1933; Filimonoﬀ, 1947). These historic classifications can be criticized because they are based only on anatomy, but they are still of interest for defining diﬀerent parts of the cortex horizontal section through the rat hippocampal formation. DG: dentate gyrus; S: subicu-lum; PrS: Presubiculum; PaS: Parasubiculum; EC : Enthorinal cortex; Note the obvious separation of superficial (1,2,3) and deep (5,6) layers by lamina dissecans (layer 4) in the presubiculum. Note that layer 2 is more dense than layer 3, and that presubicular deep layers appear as a continuation of the subiculum and entorhinal cortex deep layers. Adapted from Amaral and Witter, 1989. B: Drawing of a horizontal section correspond-ing to A, but using a 15 day old mouse, stained with the Golgi method. Adapted from Ramon y Cajal (1899). Note the dense « plexus » of aﬀerent fibers in the presubiculum that partially avoid layer 2. C: Laminar organization of the human presubiculum. Nissl method, from Ramon y Cajal (1899). Cajal’s nomenclature (my interpretation): A, plexi-form layer (layer 1); B, small pyramidal and fusiform cell layer (layer 2); C, deep plexiform layer (layer 3); D, large to medium size pyramidal cell layer (layer 4 and 5); E, fusiform and triangular cell layer (layer 6).
The Isocortex (or Cortex Completus) comprise 6 layers whereas the Allocortex (or Cortex Incompletus) displays an incomplete structure (less than 6 layers) in develop-mental and adult stages. The Periallocortex (or Cortex Intermedius) physically lies between the two others and its structure changes between developmental and adult stage. Neocortex is Isocortex; hippocampus and subiculum constitute the Archicor-tex, which is part of the Allocortex; the presubiculum was lumped together with the entorhinal area and termed Periarchicortex, which is part of the Periallocortex (Lorente De Nó, 1933; Filimonoﬀ, 1947).
More recent findings (Bayer, 1980) have shown that embryogenesis is actually diﬀerent between presubiculum and entorhinal cortex. First, neurogenesis occurs later in presubiculum. Second, deep layers are formed before superficial layers (like the classical cortical development) with a strong neurogenetic gradient. Indeed, deep layers appear at E15-18 whereas superficial layers appear at E17-20. A small gradi-ent also exists in entorhinal cortex but it occurs a little earlier (finished at E17 in deep layers and E18 in superficial layers). Another intriguing fact is that neurogene-sis timelines of presubiculum and subiculum are the same for deep layers but not for superficial layers. In the adult, it is interesting to look at the presubiculo-subiculum transition in horizontal slices (Fig 2A) to see that presubicular deep layers really appear to be a continuation of subiculum. From his studies on Marsupials, Brod-mann (Brodmann, 1909) even described this transition as an « abrupt interruption of layer II-V at the beginning of the subiculum, with only layer I and VI continuing into Ammon’s horn in greatly widened form ».
All these developmental data showed that the six layers of presubiculum appear in a very specific and unique manner. However, functional consequences of this specific development, compared to neocortex or entorhinal cortex remain unknown.
To survive, mammals rely on their sense of orientation to get water, food, and mate, or to escape predators. This requires the innate ability to learn features of a novel environment as it is explored. This is spatial orientation and it uses two diﬀerent cognitive processes: path integration and landmark navigation. Path integration uses a self derived representation of space using vestibular, proprioceptive and motor inputs; landmark navigation represents space using external cues such as visual, olfactory, auditory and somatosensory information. Among all the brain areas involved in these processes, the presubiculum encodes the head direction, one critical information for spatial cognition (Wiener and Taube, 2005; Taube, 2007).
Head direction cells of the presubiculum
Extracellular recordings in freely moving rats have shown that 50-60 % of neurons in dorsal presubiculum (postsubiculum) are discharging as a function of animal’s directional heading (Ranck, 1984; Taube et al., 1990a; Taube, 2007; Boccara et al., 2010). These neurons are called Head Direction Cells. Each head direction cell is characterized by a specific tuning curve of its firing rate as a function of the animal’s head direction (Fig. 3A).The cell’s preferred direction is defined as the one leading to the maximum firing rate. Basal firing rate is close to zero and increases only for directional ranges varying from 60◦ to 150◦ (average 90◦) with a triangular or Gaussian distribution of frequencies around the preferred direction (Blair and Sharp, 1995; Taube, 1995). One cell is accurately tuned to only one head direction and the whole population allows a complete representation of orientation. Each neuron has a very stable tuning curve but the peak firing rate varies among presubicular neurons (from 5 to 115 Hz). Last, but not least, discharge persists without adaptation as long as the preferred direction is maintained (Taube et al., 1990a).
Head direction cells are not sensitive to the geomagnetic field but to environ-mental visual landmarks. Rotation of the major polarizing visual cue within the environment leads to a corresponding shift of the preferred direction (Fig. 3D). Head direction cell firing does not change in the absence of visual landmarks, but preferred direction can drift over time (Fig. 3E). Visual cues are used to control but not to generate the head direction signal. Furthermore, visual inputs exert a higher degree of control than other senses such as auditory or olfactory inputs (Goodridge et al., 1998). Motor activity seems to improve signal quality but is not necessary for its generation because preventing an animal from moving reduces peak firing rate but does not abolish head direction cell activity ((Fig. 3B), Taube et al., 1990b) features of head direction cells (adapted from Taube, 1995): background firing rate is close to zero but increases within the directional firing range to reach the peak firing rate for the preferred direction. B. Stability of head direction cell firing across two recording sessions, one (dashed line) recorded 15 days after the other (solid line). In standard condition, a prominent cue card is disposed as a polarizing cue on one side of the open field wall. C. Carrying the animal by hand and moving it around in the arena (dashed line) only decreased peak firing rate compared to standard condition (solid line). D. Cue card rotation causes a corresponding shift in preferred direction. Here, the same head direction was recorded in standard condition (1, solid line), after a 180◦ clockwise rotation of the cue card (2, dash-dot line) and after the equivalent counter rotation putting the card in its initial position (3, line with 2 short dashes). Animal has been returned to his home cage as environmental modifications were made. E. Drift of preferred direction following card removal. The same head direction was recorded in standard condition (1, solid line), after cue card removal (2, dash-dot line) and after cue card return to its initial position (3, line with 2 short dashes). Experimental results were adapted from Taube et al. (1990a,b).
Properties of presubicular head direction cells show that an animal primarily uses path integration to keep track of changes in head direction but also landmark navigation to stabilize and correct the signal. The sense of head direction is com-puted, not only in the presubiculum, but through a head direction macrocircuit containing several interconnected brain areas.
Head Direction Circuit
Areas containing head direction cells
The head direction circuit that generates and maintains the directional heading signal includes the dorsal tegmental nucleus (DTN) (Sharp et al., 2001b), lateral mammillary nucleus (LMN) (Stackman and Taube, 1998), anterior dorsal thala-mic nucleus (ADN) (Taube, 1995), lateral dorsal thalamus (LDN) (Mizumori and Williams, 1993), retrosplenial cortex (both granular and agranular regions) (Chen and Johnston, 2004; Cho and Sharp, 2001), entorhinal cortex (Sargolini et al., 2006) and the presubiculum. All these interconnected areas (Fig. 4; Table 1.1) contain head direction cells that diﬀer in their specific tuning properties. One remarkable parameter is the directional range that is narrower for presubiculum and retrosple-nial cortex compared to ADN, LMN and DTN (Tuning curves, Fig. 4). In addition, subcortical head direction cells anticipate future head direction, that is, ADN and LMN tuning curves slightly vary between clockwise and counterclockwise head ro-tations (Fig. 4). Cortical neurons appear to be the most accurate in signaling head direction. This is explained by the hierarchy in the head direction circuitry, which was established mainly by doing lesioning of one area and looking at the conse-quences in others (Clark and Taube, 2012 for review). These studies have drawn attention to a sub-cortical generator using self-movement information; cortical areas may bring sensory information to increase stability and precision.
Subcortical origin of head direction signals
Head direction cell activity requires information generated by the vestibular labyrinth. The labyrinth is composed of the semicircular canal and the otolith organ that detect angular and linear acceleration respectively. Semicircular canal function is necessary for generating head direction cell activity in ADN (Muir et al., 2009) whereas the otolith organ is involved in signal robustness and stability (Yoder and Taube, 2009). The vestibular signal is carried by angular head velocity cells, that fire as a function of head rotation speed and direction. These neurons are found all along the inte-grative pathway, from the vestibular organ to the Dorsal Tegmental Nucleus (DTN) and Lateral Mammilary Nucleus (LMN). These two last areas also contain head di-rection cells. Many experimental and modeling studies suggest that the DTN-LMN interactions would constitute the head direction cell generative circuit, converting angular velocity information in head direction information (Bassett et al., 2007; Clark and Taube, 2012).
The head direction signal is thought to be generated according to continuous attractor dynamics (see Fig. 5 ; Skaggs et al., 1995; Redish et al., 1996; McNaughton et al., 2006) and diﬀerent versions exists for the head direction circuit (e.g. Sharp et al., 2001a versus Boucheny et al., 2005). Recent experimental findings reinforces the validity of these models in the generation of stable activity states (Schmidt-Hieber and Häusser, 2013; Domnisoru et al., 2013), such as the head direction signal.
From LMN, the head direction signal is then relayed via the anterodorsal tha-lamus (ADN) (Fig. 4) that sends projections to cortical areas such as retrosplenial cortex (van Groen and Wyss, 1990a) and presubiculum (van Groen and Wyss, 1990c) driving cortical head direction cells. Functionally, ADN is a critical relay in the head direction circuit, its lesion disrupting head direction cells in cortical areas, including presubiculum (Goodridge and Taube, 1997), parasubiculum and entorhinal cortex (Clark and Taube, 2012).
If head direction signal in ADN is not abolished by lesions of presubiculum (Goodridge and Taube, 1997), this last one plays a significant feedback control in refining the signal with visual information.
Visual landmark control of the head direction signal by the presubiculum Presubiculum is one entry point of visual information into the head direction sys-tem (Fig. 4). It receives direct projections of primary and secondary visual cortices (Vogt and Miller, 1983) and projections from retrosplenial cortex, relaying infor-mation from visual cortex (Vogt and Miller, 1983; van Groen and Wyss, 1990a; Jones and Witter, 2007) and from associative visual cortical areas, such as posterior parietal and postrhinal cortices (Yoder et al., 2011). Visual information might also come from the laterodorsal thalamus (LDN) that sends direct projections to pre-subiculum (van Groen and Wyss, 1992b). LDN receives visual inputs from pretectal areas and superior colliculus but it has no functional impact onto visual landmark dependent activity in presubiculum (Golob et al., 1998). LDN seems also to be associated with somatosensory inputs (Bezdudnaya and Keller, 2008), but head di-rection signal dependence upon somatosensory inputs has never been shown. By its direct projections to ADN (van Groen and Wyss, 1990c; Ishizuka, 2001; Yoder and Taube, 2011) and LMN (Allen and Hopkins, 1989; Gonzalo-Ruiz et al., 1992; Yoder and Taube, 2011), the presubiculum appears like an ideal relay for carrying visual landmark information into subcortical generators of head direction signal. Indeed, presubiculum lesion impairs visual landmark control of a cell’s preferred direction in ADN (Goodridge and Taube, 1997) and LMN (Yoder et al., 2011). In other words, without the presubiculum, Head direction cells’ preferred directions in ADN and LMN are much less influenced by visual cues (Fig. 3D). This feedback visual control might be exerted in a larger extent in the whole head direction circuit, the presubiculum projecting also to the retrosplenial cortex (Wyss and van Groen, 1992), LDN (van Groen and Wyss, 1990b,c) or medial entorhinal cortex (Honda et al., 2008). Moreover, visual information transmitted via the presubiculum is also critical for the activity in the downstream hippocampus.
The presubiculum is a major contributor of spatial representation and memory
The first evidence for the representation of space in the brain was the discovery of « place cells » in the hippocampus by O’Keefe and Dostrovsky (1971). Place cells fire as a function of the animal’s position within space, and they are believed to be the neuronal substrate of a spatial cognitive map. Since, spatial information processing has been shown to occur at the level of the whole hippocampal-parahippocampal area, especially through dialogue between the hippocampus and the medial entorhi-nal cortex.
Interconnectivity within the hippocampal, parahippocampal and entorhinal cortices is depicted in figure 6. Entorhinal cortex sends many diﬀerent projections to the hip-pocampus. Layer 2 neurons project to the dentate gyrus and also directly to CA3 (perforant path). Dentate gyrus granule cells excite CA3 pyramidal cells, which then contact CA1 pyramidal cells (Amaral and Witter, 1989) and also other CA3 pyramidal cells (Le Duigou et al., 2014). Entorhinal layer 3 cells also make direct contacts onto CA1 (Amaral and Witter, 1989; Kohara et al., 2013), Subiculum, and CA2 receives strong inputs from superficial entorhinal neurons; the originat-ing layer(s) being debated: layer 2/3 (Chevaleyre and Siegelbaum, 2010) or solely layer 2 (Kohara et al., 2013). CA1 projects to subiculum. Both close the loop by projecting back to entorhinal cortex (Amaral and Witter, 1989). Subiculum is also interconnected with pre- and parasubiculum (Amaral and Witter, 1989; Kim and Spruston, 2011), CA1 projections to the dorsal part of the presubiculum have been described (van Groen and Wyss, 1990c), but contradicted thereafter by another study (Cenquizca and Swanson, 2007).
Table of contents :
1 The presubiculum: Anatomy, function, microcircuit
1.1 The presubicular cortex
1.1.2 What kind of cortex?
1.2 Presubiculum and spatial orientation
1.2.1 Head direction cells of the presubiculum
1.2.2 Head Direction Circuit
1.2.3 The presubiculum is a major contributor of spatial representation and memory
1.3 Information processing in the presubicular microcircuit
1.3.1 Anatomy and intrinsic excitability of presubicular neurons
1.3.2 Interlaminar, intralaminar and modular organization
1.3.3 Input and output relays in the presubicular microcircuit
2 How does a microcircuit work?
2.1 Many integrative levels in neuronal networks
2.2 Neuronal intrinsic excitability
2.2.1 Resting membrane potential
2.2.2 Neuronal passive properties
2.2.3 Action potentials
2.2.4 Firing properties
2.3 Wiring a network: axonal conduction and regulation of information
2.3.1 Axonal conduction velocity
2.3.2 Analog information encoding in the axon
2.4 Synaptic transfer and modulation of information in the presynaptic terminal
2.4.1 Basic mechanism of neurotransmitter release
2.4.2 Synchronous versus asynchronous release of neurotransmitter
2.4.3 Short term presynaptic plasticity
2.4.4 Voltage dependent regulation of synaptic activation
2.4.5 Regulation of presynaptic function by extrinsic factors
ARTICLE 1. Cellular neuroanatomy of rat presubiculum
ARTICLE 2. Properties of presubicular neurons that project to lateral mammillary nucleus or anterodorsal thalamus
ARTICLE 3. A continuum of diversity of Parvalbumin or Somatostatin expressing interneurons in mouse presubiculum
ARTICLE 4. Memory of past activity determines the recruitment of a Martinotti cell-mediated inhibitory feedback loop in mouse presubiculum
1 Building blocks of the presubiculum
1.1 Did we correctly addressed the whole diversity of principal neurons?
1.2 Interneuron diversity
2 Perspective: from neuronal diversity to function
3 Neurons that project to lateral mammillary (LMN) and anterodorsal thalamus (ADN): implication for the visual update of the head direction signal
4 Memory of past activity at the pyramidal cell-to-Martinotti cell synapse: properties and mechanisms
4.1 Better define the dynamics of the plasticity, its specificity and variability
4.2 Mechanisms of activity dependent synaptic transfer at the pyramidal cell to Martinotti cell synapse?
4.2.1 Activity dependent action potential broadening
4.2.2 Modulation at the synapse
4.2.3 The transfer rate increase may results from a synergistic mechanism
V General conclusion
ARTICLE. Cellular anatomy, physiology and epileptiform activity in the CA3 region of Dcx knockout mice: a neuronal lamination defect and its consequences
List of figures