The entorhinal cortex is commonly perceived as a major input and output structure of the hippocampal formation, entertaining the role of the nodal point of cortico-hippocampal circuits. Superficial layers receive convergent cortical information, which is relayed to structures in the hippocampus, and hippocampal output reaches deep layers of entorhinal cortex, that project back to the cortex. The finding of the grid cells in all layers and reports on interactions between deep and superficial layers indicate that this rather simplistic perception may be at fault. Therefore, an integrative approach on the entorhinal cortex, that takes into account recent additions to our knowledge database on entorhinal connectivity, is timely. We argue that layers in entorhinal cortex show different functional characteristics most likely not on the basis of strikingly different inputs or outputs, but much more likely on the basis of differences in intrinsic organization, combined with very specific sets of inputs. Here, we aim to summarize recent anatomical data supporting the notion that the traditional description of the entorhinal cortex as a layered input-output structure for the hippocampal formation does not give the deserved credit to what this structure might be contributing to the overall functions of cortico-hippocampal networks.
The
entorhinal cortex (Brodman area 28) derives its name from the fact that it is
partially enclosed by the rhinal (olfactory) sulcus. This feature is
particularly striking in nonprimate mammalian species, but also in primates at
least the anterior part of the entorhinal cortex is bordered laterally by a
rhinal sulcus. Interest in the entorhinal cortex arose around the turn of the 20th
century when Ramon y Cajal, in his seminal studies on the anatomy of the
nervous system, described a peculiar part of the posterior temporal cortex
which is strongly connected to the hippocampus with fibers that merge in the
angular bundle and perforate the subiculum. Cajal was so struck by this massive
connection that he suggested that the physiological significance of the
hippocampus had to be related to that of the entorhinal cortex. At that time,
he assumed that the entorhinal cortex was part of the olfactory cortex and so
was, therefore, the hippocampus. He even stated that if this part of the
posterior temporal cortex, which he called the sphenoidal cortex/angular
ganglion, would be visual, so would be the hippocampus [
Our current insights into the functional relevance of the hippocampal formation,
and how its anatomy is related to function, are much more detailed than what we
know about the entorhinal cortex. It therefore seems attractive to turn the
argument of Cajal around by stating that in view of the findings that the
hippocampus is crucially involved in conscious, declarative memory processes so
should be the entorhinal cortex. This conjecture is apparently supported by
available functional studies. Although the specific functional contributions of
the entorhinal cortex to memory remain to be established, they are most likely
different from, but complementary to, those of the hippocampus [
A cortical area can be defined in
many different ways, using a variety of different criteria, such as location,
connectivity, cyto- and chemoarchitectonics. For the entorhinal cortex, all these
approaches have been applied, resulting in a confusing variety of borders,
subdivisions, and description of layers. A good lead, since it has withstood
over a century of arguments, is the definition of the entorhinal cortex on the
basis of its connectivity with the hippocampus as originally suggested by Cajal [
The entorhinal cortex is surrounded by a number of cortical
areas. Anteriorly, it meets with olfactory and amygdaloid cortices, such as the
piriform (olfactory) cortex laterally, and medially it is bordered by the
periamygdaloid cortex and the posterior cortical nucleus of the amygdala. On
its medial side, the entorhinal cortex merges with structures that belong
either to the hippocampal formation or the parahippocampal region, such as the
amygdalo-hippocampal transition, and the parasubiculum. The lateral and
posterior borders are with the other two major constituents of the
parahippocampal region, the perirhinal cortex laterally and the parahippocampal
cortex (in nonprimate species generally referred to as postrhinal cortex)
posteriorly. The lateral and posterior borders are quite easy to establish on
the basis of a variety of cytoarchitectonic and chemoarchitectonic features.
The most prominent features are that the fairly large-sized cells of layer II
in the entorhinal cortex are replaced by much smaller neurons in the perirhinal
and postrhinal cortices, the lamina dissecans disappears, and these changes
coincide with similarly striking changes in the density of parvalbumin-positive
neuropil, high in entorhinal cortex, virtually absent in perirhinal and
parahippocampal areas. The mirror-image pattern appears when staining for heavy
metals (Timm stain) or the calcium binding protein calbindin. All additional
criteria that have been described seem to coincide with these borders. The
anterior and medial borders, in contrast, are somewhat harder to establish.
They apparently coincide with a rather striking change in the ease with which
layers II and III can be differentiated from each other as well as with a loss
of differentiation between the deep layers (medial border) or even complete
disappearance of the deep layers (anterior border). Combined with subtle
changes in chemoarchitectonic features and connectional differences, an overall
consensus has now been reached (for further details see [
Attempts to subdivide the entorhinal cortex have, likewise, been numerous (see [
A good start to subdivide the
entorhinal cortex is to use the entorhinal-to-dentate projection, which has
been documented in extensive detail in a variety of species. On the basis of the
terminal distribution of this projection in the rat and the mouse, it seems
plausible to divide the entorhinal cortex into two subareas, generally referred
to as the lateral and medial entorhinal cortices (LEC and MEC, resp.). These
areas roughly correspond to the description of Brodmann’s areas 28a and b, respectively
[
A note of caution should be added
here: the choice for the terms lateral and medial entorhinal cortex is not
simply related to a particular anatomical position of these areas in relation
to the hippocampal formation and the rhinal fissure. In general, the lateral
area occupies a more rostrolateral position versus a more caudomedial position
for the medial area (see Figure
Schematic representation of the overall organization of the entorhinal cortex and its connectivity. (a) Position of the entorhinal cortex and surrounding cortices and hippocampus in the rat left hemisphere. Indicated are the dorsoventral extent of the hippocampus, positions of LEC and MEC, and the approximate position of a representative horizontal section, illustrated in (b). (b) Horizontal section illustrating entorhinal-hippocampal connectivity (see text for more details). (c) and (d) Representation of the topographical arrangement of entorhinal-hippocampal reciprocal connections. A dorsolateral band of entorhinal cortex (magenta) is preferentially connected to the dorsal hippocampus. Increasingly, more ventral and medial bands of entorhinal cortex (purple to blue) are connected to increasingly more ventral levels of the hippocampus. Yellow line in (c) indicates the border between LEC and MEC. (e) Enlarged entorhinal cortex, taken from (c), indicating the main connectivity of different portions of entorhinal cortex. Brain areas preferentially connected to LEC are printed in green, those connected to MEC are in magenta. The color of the arrows indicates preferential connectivity to the dorsolateral-toventromedial bands of entorhinal cortex (magenta or blue, resp.) or that no preferential gradient is present (green).
The lamination of the entorhinal
cortex generally is considered the prototype of the transition between the
three-layered allocortex and the six-layered neocortex [
Entorhinal connections with the hippocampal formation in the rat have been comprehensively
described and reviewed in a number of recently published papers and reviews to
which the reader is referred for further details [
The CA1-subicular projections are topographically organized along the transverse or
proximodistal axis as well, such that parts of CA1 and subiculum that receive
comparable inputs that are either from LEC or MEC are connected to each other
[
The
most comprehensive systematic series of studies on entorhinal connectivity in
the rat is from the Burwell lab [
Although
we will address the layered organization of the entorhinal cortex in more
detail below, it is relevant to point out that entorhinal-cortical projections
largely arise from deep layers, primarily from layer V pyramidal neurons. Possible
exceptions are the entorhinal-infralimbic and entorhinal-olfactory projections,
which appear to arise in layers II and III as well [
Studies
conducted in multiple species indicate extensive subcortical connectivity for
the entorhinal cortex. Although differences exist with respect to the detail of
the information, it is safe to conclude that the entorhinal cortex has
connections with the basal forebrain, claustrum, amygdala, basal ganglia,
thalamus, hypothalamus, and brainstem (for review see [
Entorhinal-amygdala connectivity has been
studied in rather detail in both monkey and rat. For recent reviews, the reader
is referred to McDonald [
The
entorhinal cortex is connected with thalamic and hypothalamic structures. Major
thalamic input arises in midline nuclei, particularly the reuniens, paratenial,
and periventricular nuclei [
Our understanding of the entorhinal cortex is
still rather premature, and to a large extent, influenced by our current
functional concept for MEC. The generally accepted division of the entorhinal
cortex into at least two functionally different domains stresses the need for
an answer to the questions whether or not they differ with respect to their intrinsic
wiring and neuronal makeup, in addition to their gross differences with respect
to cortical and subcortical connectivity summarized above. The entorhinal
network,
Compared to the details known for the
hippocampal formation and some parts of the neocortex, such as the visual or
barrel cortices in rodents, our understanding of the entorhinal cortex is
rather in its infancy. The first detailed description of the morphology of
entorhinal neurons, based on Golgi impregnated material, was published in 1933
by Lorente de Nó [
Summary diagram of the morphology of main cell types in LEC and MEC. (a) Cells in superficial layers I-III. (b) Cells in deep layers IV-VI. See text for more details.
In layer I throughout the entorhinal cortex, Lorente de Nó
described two cell types; horizontal cells and short axis cylinder cells,
nowadays known as multipolar neurons (MPNs). This latter category constitutes
the majority of cells in layer I, and generally, they are non- or sparsely spiny.
MPNs are quite often positive for calretinin (CR) and GABAergic, and two types
have been described. Small CR positive MPNs are more often located just deep to
layer Ia [
Horizontal cells
are located in the transitional zone between layers I and II [
Layer II is mainly made up of densely packed, large and
medium sized pyramidal and stellate cells. The most abundant cell type
throughout layer II in MEC is the stellate cell, with their preferred location
within superficial and middle layer II [
Stellate cells are less common in LEC than in MEC. In LEC, stellate cells are most
likely replaced by a comparable cell type, called fan cells [
Aside from the stellate-like principal neurons, layer II contains a number of
pyramidal-like cells that have medium sized triangular or ovoid shaped soma
with a perpendicular elongation with respect to the pial surface. Most are
located in the deep portion of layer II [
Another pyramidal
cell type described in LEC has a very thick and sparsely or nonspiny apical
dendrite, which branches in layer II. Thin apical dendritic tufts reach layer
I. The apical dendrite is not as frequently tilted as in MEC pyramidal neurons
[
Interneurons within layer II are described as MPNs, bipolar, basket, and chandelier cells.
MPNs have polygonal, fusiform, or round cell bodies with multiple, sparsely
spiny dendrites, extending in all directions, reaching layer I and deep into
layer III. It has been described that the axons of MPNs travel to the white
matter but also form local synapses within layer II [
Sparsely
spiny horizontal bipolar cells although considered to be local/interneurons
project to the hippocampus [
Fast
spiking basket-like cells have small spherical cell bodies with sparsely spiny
dendrites that often ramify into layer I. The extensive axonal arbor is mainly
confined to layer II. They form basket-like complexes mainly around the soma of
other cells, preferably forming symmetric, inhibitory synapses with stellate or pyramidal cells [
Chandelier
or axo-axonic cells are characterized by vertical
aggregations of axonal boutons, called candles, which preferably are located
superficial to the cell body. The somata of chandelier cells are medium sized
with different shapes. The almost nonspiny, poorly ramifying dendrites
originate from the basal and apical poles of the somata, displaying a bipolar
or bitufted arbor that often stays within layer II/III. Vertical chandelier
cells that are restricted to MEC issue a vertically oriented axonal tree that
is around 200–300
MEC and LEC layer III
pyramidal neurons have comparable morphological as well as electrophysiological
characteristics [
Nonspiny pyramidal
cells (NSPCs), also called type 2 cells [
Layer III also
contains stellate cells, in particular in the upper part of the layer. The somata
of these neurons are elongated, polygonal, or spherical. Cells belonging to the
latter subgroup sometimes have evenly distributed spiny dendrites around the
somata, whereas others have one or two spiny basal dendrites and a variety of
ascending dendrites that branch in layer I. The axons reach the white matter,
and collaterals are formed in layer III and the lamina dissecans [
Also located
within layer III are principal MPN somata. These MPNs are either small and spherical,
with laterally extending dendrites, or they are large. The largest MPNs are
located in the outer half of layer III of the LEC with a conspicuous spatial
lateral separation (500
Multipolar local
circuit neurons, mainly described in MEC, are characterized by wide-ranging
apical dendrites that reach the cortical surface, multiple compact basal
dendrites, and a prominent axonal arborization. The axon reaches layers I to
III but rarely extends into the lamina dissecans or superficial layer V [
Interneurons
resembling pyramidal cells, the so-called pyramidal looking interneurons (PLIs)
have also been described as Type 3-(Gloveli) or Type 1-(Kumar) cells [
Bipolar
cells have been described in layer III of MEC and LEC. They have a spindle-like
perikaryon with one ascending and one descending smooth, thin and sometimes
long dendrite. The ascending dendritic collaterals traverse throughout layer
II, reaching layer I. The extent of the descending dendrites has not been
described yet. The axon arises from the primary descending dendrite and extends
into layer III and the lamina dissecans, deep to the parent cell body [
Occasionally, pyramidal-shaped neurons are located in the lamina dissecans, at the borders to layers III and V. These neurons have the morphological and physiological properties of either layer III or layer V pyramidal neurons, respectively (own unpublished data).
Furthermore, bipolar
cells, whose dendrites grow horizontally instead of vertically to the pial
surface, with axonal collaterals that can travel towards superficial layer III
and deep layers, have been found in the lamina dissecans (unpublished data). It
has been shown that bipolar cells might contain VIP, CCK, and CRF [
There is no difference between layer V principal neurons
in LEC and MEC [
A second
principal cell type described in layer V is generally referred to as a type of horizontal
cell [
A third type of
principal neurons is polymorphic MPNs [
Fusiform cells
that project to the hippocampus were found in superficial layer V [
Superficial
layer V further harbours bipolar cells with a spindle-like soma having an
average diameter along the short axis of around 12
The multilaminated layer VI borders the white matter.
MPNs are located throughout layer VI. They have a spherical soma with a
diameter of approximately 14
The somata of
classical pyramidal cells in the MEC are medium sized. Pyramidal cells in the
LEC have not been described yet. The difference compared to layer V or III
pyramidal cells is that the dominant dendrite does not always travel radially
towards superficial layers but also either horizontally within layers VI and V
or descends towards the angular bundle and the subiculum (own unpublished
data). The basal dendrites and the widely spreading collaterals spread within
layers VI and V. The axons of pyramidal cells travel towards the angular bundle
and subiculum as well as towards superficial layers. Their axon collaterals are
located within layers V, VI, the angular bundle, and the subiculum [
In conclusion, there are
differences between cell types and the distribution of cell types in LEC and
MEC (see Figure
The entorhinal cortex contains a substantial
system of associational connections that are best described at two different
levels. The first is that in all species studied, intraentorhinal fibers are
organized in a limited number (generally three) of rostrocaudally oriented
bands. Connections that link different transverse (or mediolateral) regions of
the entorhinal cortex, thus providing connectivity between these bands, are
rather sparse [
The overall organization of the longitudinal
intrinsic connections is best considered in relation to the organization of the
reciprocal entorhinal connections with the hippocampal formation. Interconnected
portions of the LEC and MEC close to the rhinal fissure, in rats referred to as
the dorsolateral band of entorhinal cortex, are connected to the dorsal
(nonprimate) or posterior (primate) part of the hippocampal formation (see
Figures
The second organizational level deals with
the local connectivity within and among layers of more restricted portions of
the entorhinal cortex. As we know from the studies summarized above, neurons in
different layers have very different inter- and intralaminar connectional
patterns that include axon collaterals confined to the parent cell layer or
spanning several layers. But not only the axonal distribution is of importance,
the dendritic trees may also play an essential role in that they either span
several layers or are more restricted to the parent cell layer. Although
detailed information for quite a few of neuronal types in the entorhinal cortex
is still lacking, it is safe to say that the entorhinal network, on the basis
of its neuronal composition alone, cannot be properly described in terms of
superficial and deep layers as more or less independent layers. All this may
not come as a surprise since comparable concepts have been described with
respect to the organization of the neocortex [
Schematic representation of laminar distribution and synaptic interactions between inputs and principle cells of the entorhinal cortex. Different inputs are represented by color-coded arrows; position of the arrows indicates the main laminar distribution. Circles indicate putative synaptic contacts between inputs and principle cells. Main output connectivity of principle cells is indicated as well. The figure emphasizes the integrative capacity of layer V cells.
One important anatomical observation already
reported by Cajal [
As illustrated in Figure
What
then is the functional relevance of inputs from for example the medial
prefrontal, cingular, and retrosplenial areas? Afferents from these areas
preferentially, and in some instances even exclusively, terminate in the deep
layers of the entorhinal cortex. Note that in the monkey, however, it has
recently been reported that projections from the retrosplenial cortex densely
innervate entorhinal layer I [
The functional relevance of the
organization of networks in the brain is often interpreted on the basis of a surprisingly
restricted point of view. Debates on the functional organization of the
hippocampal formation have been strongly influenced by the idea that the
prevailing hippocampal circuitry is unidirectional. With regards to the entorhinal
cortex, the breakthrough discovery, that deep entorhinal layers receive
hippocampal output from CA1 and the subiculum on the one hand, and that these
same layers are the origin of strong cortical projections, has biased our view towards
the rather simple concept that the deep layers mediate hippocampal-to-cortical
connectivity, similar to superficial layers providing the way in for cortical
inputs to the hippocampal formation. If the entorhinal cortex is such an
important hub, similar to the central station of a large city, and that is what
all data seem to converge on, it is quite likely that it serves yet another
role. In addition to serving simply to get into the city or leave the city, the
station also provides the powerful potential for new interactions between and
among incoming and outgoing people. This potential for “new” interactions has been
grossly neglected in case of the entorhinal cortex. The potential of the
entorhinal cortex to act as an interactive hub, contributing essentially to the
functions of the cortico-hippocampal system instead of just transferring
information, has been underscored not only by the recent finding of the unique
spatial firing properties of grid cells in the entorhinal cortex [
Similar to the yet unresolved
mystery of the relevance of cortical inputs to deep layers of the entorhinal
cortex, it remains to be established what the functional relevance is of LEC.
The data summarized above indicate that with the exception of neurons in layer
II, it is likely that both LEC and MEC are largely similar with respect to
their intrinsic wiring, both in terms of neuronal elements that comprise the nodal
points of the network as well as with respect to how these are wired together
(see Figures
The preparation of this paper and the original research on entorhinal neurons are supported by the Kavli Foundation and a Centre of Excellence grant from the Norwegian Research Council. Cathrin B. Canto thanks the Department of Anatomy and Neurosciences, VU University medical center for the generous hospitality and warm atmosphere.