Tải bản đầy đủ - 0trang
1 Introduction: The Evolvement of the Concept of the Ventral Striatopallidal System
H.J. Groenewegen et al.
systems, a dorsal and a ventral one, that via their distinctive relay nuclei in the
thalamus have an inﬂuence on, respectively, the somatomotor and the associative,
prefrontal cortical regions in the frontal lobe (Heimer and Wilson 1975). They
emphasized that, whereas the dorsal and ventral striatum receive functionally different inputs, carrying either somatomotor or limbic/associative information, the
basic cellular, chemoarchitectonic and connectional organization in these two parallel circuits appears to be very similar. Therefore, somatomotor and limbic cortical
information, via separate dorsal and ventral striatopallidal channels, in which comparable neuronal mechanisms play a role, lead to transfer of these streams of information via the thalamus back to different somatomotor or limbic-associated parts of
the frontal lobe. It was further stipulated in these early days that, at the level of the
striatum, the transfer of information is modulated by dopamine from the nigrostriatal and mesolimbic systems originating in substantia nigra pars compacta and the
ventral tegmental area (VTA), respectively.
Although the parallel nature of the somatomotor and limbic cortical-basal ganglia circuits was emphasized at the time, Nauta and colleagues also showed that a
major output of the nucleus accumbens reaches the VTA and substantia nigra pars
compacta, in which the dopaminergic neurons project back to both the ventral and
dorsal striatum (Nauta et al. 1978). That led them to hypothesize that there exists a
dopaminergic feed-forward circuit for the integration of the ventral and dorsal circuitries, i.e., a means to integrate limbic and motor functions. Together with the
pioneering electrophysiological work of Gordon Mogenson and his colleagues
(Mogenson et al. 1980) on the ventral striatum as a key structure involved in the
translation of ‘motivation into action’, these ideas now form the cornerstones for
our understanding of the role of the cortical-basal ganglia circuits in motor, cognitive, and emotional/motivational behaviors and their dysfunctioning in neurological
and psychiatric disorders (e.g. Voorn et al. 2004; Humphries and Prescott 2010;
Haber and Behrens 2014; Everitt and Robbins 2015; Floresco 2015).
The concept of a parallel organization of corticostriatopallidal projections in the
sensorimotor and limbic realms, as put forward by Heimer and Nauta and co-workers,
was further developed by Alexander, DeLong, and colleagues in their seminal review
on the organization of basal ganglia-thalamocortical circuits (Alexander et al. 1986;
DeLong 1990; also DeLong and Georgopoulos 1981). In primates, Alexander et al.
(1986) proposed the existence of ﬁve parallel, functionally segregated basal gangliathalamocortical circuits: a somatomotor, an oculomotor, and three complex circuits,
one of which was designated as the ‘limbic circuit’. These circuits have their origin in
distinct (pre)frontal cortical regions from which the corticostriatal projections originate and further include topographically organized striatopallidal/striatonigral and
pallido/nigrothalamic projections to distinct medial and ventral nuclei of the thalamic
complex. The different thalamic nuclei targeted by the internal globus pallidus, ventral pallidum, and substantia nigra pars reticulata project back to the (pre)frontal cortical areas of origin of the individual circuits, in this way forming closed loops
(Alexander et al. 1986; Groenewegen et al. 1990). At the level of the striatum, projections from other more posterior cortical areas, i.e., from the parietal, occipital, and
temporal cortices, converge with their associated and mutually interacting frontal
cortical areas (Yeterian and Van Hoesen 1978; Selemon and Goldman-Rakic 1985) in
Limbic-Basal Ganglia Circuits Parallel and Integrative Aspects
order to subserve the integration of information from these sensory association
cortices with higher order cognitive areas of the prefrontal-striatal system (e.g. Cavada
and Goldman-Rakic 1989; Flaherty and Graybiel 1991). Recent studies show the
functional signiﬁcance of such integration at the level of the caudate nucleus in value
processing and decision making within local microcircuitries to be discussed later.
Furthermore, the corticostriatal circuitry involving the rostral head of the caudate
nucleus is important for ﬂexible (short-term) values; the caudal tail of the caudate
plays a role in stable (long-term) values and the behavioral decisions made on that
basis (Kim and Hikosaka 2013, 2015).
Parts of the output of the basal ganglia-thalamocortical circuits, primarily originating in the different (pre)frontal cortical regions, are directed at motor areas in the
brainstem, such as the superior colliculus, the midbrain extrapyramidal area, the
pedunculopontine nucleus, and the reticular formation, as well as the spinal cord.
The limbic-related parts of the cortical-basal ganglia system project, in addition, to
hypothalamic and brainstem areas that are involved in various types of emotional
and incentive, cue-directed motor behavior (Mogenson et al. 1980) and the regulation of eating/drinking, autonomic, and endocrine functions (Kelley 1999; Richard
et al. 2013; Castro et al. 2015).
As already indicated above, within the ﬁve initially identiﬁed basal gangliathalamocortical circuits (Alexander et al. 1986), various functionally different subcircuits have been identiﬁed (e.g., somatomotor functions: Alexander et al. 1990;
decision making: Kim and Hikosaka 2013; incentive behavior: Richard et al. 2013).
Without ignoring this multiplicity of the circuits between the (pre)frontal cortex and
the basal ganglia, a classiﬁcation into three larger ‘families’ of circuits within the
basal ganglia-thalamocortical system is nowadays most frequently being adopted: a
collection of somatomotor circuits, a group of complex or associative circuits, and
a ‘family’ of limbic, emotional, and motivational circuits (cf. Parent and Hazrati
1995a; Humphries and Prescott 2010). In line with a partitioning of the striatum and
the pallidum into somatomotor, associative, and limbic parts, also in the subthalamic nucleus, these three functionally different regions have been identiﬁed based
on their different afferent striatopallidal and frontal cortical inputs (Groenewegen
and Berendse 1990; Parent and Hazrati 1995b; review: Temel et al. 2005).
Whereas Nauta and Heimer and their colleagues, based on their experimental
work, concentrated primarily on the organization of corticostriatopallidal circuits in
rodents, Alexander and colleagues based their ideas about the basal gangliathalamocortical circuits primarily on electrophysiological and neuroanatomical results
in primates. This facilitated the extrapolation of the neuronal relationships between
the basal ganglia and the thalamocortical system from the rodent and primate brain to
the human situation. In a recent study, special attention was paid at the homologies
between the cortical-basal ganglia systems in rodents and primates (Heilbronner et al.
2016). The above-mentioned early conceptual papers have inspired many researchers
in the last decades to investigate in more detail the various different subdivisions of the
basal ganglia not only structurally and functionally, but also with respect to their putative roles in neurological and psychiatric disorders. It has thus been hypothesized
already in the seventies and eighties of the last century that speciﬁc dysfunctions exist
in particular basal ganglia-thalamocortical circuits in neurological disorders, such as
H.J. Groenewegen et al.
Parkinson’s disease and Huntington’s disease (Alexander et al. 1986; Delong 1990;
Albin et al. 1989), and in psychiatric disorders, such as schizophrenia, obsessivecompulsive disorders, Tourette syndrome, drug addiction, and depression (e.g.,
Stevens 1973; Cummings 1993; Mega and Cummings 1994; Mink 1996; Humphries
and Prescott 2010; Willner et al. 2013; Tremblay et al. 2015). In the last two decades,
with the great advent of modern neuroimaging techniques, the functional–anatomical
relationships of the cerebral cortex, basal ganglia, and thalamus have been extensively
studied in humans (e.g., Lehericy et al. 2004; Barnes et al. 2010; Jeon et al. 2014). The
results of these studies conﬁrm and extend the existence of multiple, functionally
segregated, as well as interacting circuits between these structures also in the human
forebrain (e.g., Postuma and Dagher 2006; Jung et al. 2014; Kotz et al. 2014; Haber
and Behrens 2014). This further opens the way to explore the dysfunctional circuitry
in neurological and psychiatric disorders.
In the following part of this chapter, we will review the main input–output relationships of the ‘limbic’, ventral striatopallidal system, primarily based on ﬁndings in rats
with some reference to primates. Whereas it is already generally acknowledged that
the striatum, as the input structure of the basal ganglia, is an important site for the
integration of information from multiple and different sources, recent data show that
there is even more overlap between corticostriatal projections than has long been
assumed. This extends our understanding of the architecture of the parallel basal ganglia-thalamocortical loops in providing rich and speciﬁc possibilities for interactions
between these parallel loops with functionally different roles. This may be the basis
for the ﬂexibility in behavioral and cognitive functioning in animals and man. With
respect to the outputs, the ventral striatopallidal system parallels the projections of its
dorsal counterpart in that there are strong projections to the mediodorsal thalamus, but
it is unique in that it has also projections to the dopaminergic cell groups in the ventral
mesencephalon. These projections provide the possibility for the ventral striatopallidal system to modulate the dopamine input to the dorsal striatum (Nauta et al. 1978;
Haber et al. 2000; Voorn et al. 2004; Belin and Everitt 2008) (see also Fig. 2.5).
Interestingly, in recent years there has been renewed interest in the projections from
the habenula, part of the epithalamus, to the mesencephalon. Thus, several studies
have shown that the lateral habenula has a direct and an indirect inﬂuence on the dopaminergic cells of the VTA, namely via the mesencephalic GABAergic rostromedial
tegmental nucleus (RMTg; also indicated as the ‘tail part’ of the VTA) (e.g. Yetnikoff
et al. 2015). Since the ventral pallidum, like the internal segment of the globus pallidus, consistently projects to the lateral habenula, there exists yet another pathway for
the modulation of the dopaminergic systems by the limbic part of the basal ganglia.
What Is the “Limbic” Ventral Striatum?
Since the inclusion of the nucleus accumbens and parts of the olfactory tubercle as
‘true’ parts of the striatum was based on cytoarchitectonic criteria (Heimer and
Wilson 1975), a clear distinction with the classical dorsal striatum (caudate nucleus
Limbic-Basal Ganglia Circuits Parallel and Integrative Aspects
and putamen) cannot be based on cellular characteristics. Histochemical or
immunohistochemical characteristics provide in some cases differences and in other
instances great similarities between dorsal and ventral parts of the striatum. For
example, the distribution of the acetylcholine metabolizing enzyme acetylcholinesterase (AChE) is quite homogeneous throughout the striatum and its distribution
was considered as supporting the inclusion of the nucleus accumbens and olfactory
tubercle in the family of striatal nuclei (Heimer and Wilson 1975). By contrast, the
calcium-binding protein calbindin D28K, present in striatal GABAergic mediumsized spiny neurons, is quite unevenly distributed over the striatum with a low density in its ventromedial part, deﬁning the shell of the nucleus accumbens, and with
higher densities in the core and in large parts of the caudate-putamen, but again with
low density in its dorsolateral (somatomotor) part (Zahm and Brog 1992). Dopamine
is distributed over the entire striatum, showing areas with higher or lower concentration throughout (Voorn et al. 1986, 2004). By contrast, neurotransmitters like
serotonin and noradrenalin are concentrated primarily in the ventromedial parts of
the striatum, noradrenalin even conﬁned to the most ventromedial region of the
nucleus accumbens, i.e., the medial shell (Delfs et al. 1998; human: Tong et al.
2005). The serotonin innervation extends into the medial and ventral parts of the
caudate-putamen complex and, of quite some clinical interest, serotonin ﬁbers in
the medial shell are different in that they lack the serotonin transporter (Brown and
Molliver 2000). Thus, as has been concluded previously, there appears to be no
clear boundary between the dorsal and the ventral striatum on the basis of cytoarchitecture, myeloarchitecture, or chemoarchitecture (Voorn et al. 2004). However,
as will be discussed in the next paragraphs in more detail, the organization of inputs
and outputs presents a somewhat different distinction within the striatum as a whole,
namely a dorsolateral-to-ventromedial orientation of striatal zones that are reached
by afferents from different (pre)frontal cortical areas and their associated subnuclei
of the intralaminar and midline thalamus as well as from distinctive amygdala and
hippocampal areas. In that way, a distinction between striatal zones, respectively,
innervated by (1) cortical sensorimotor ﬁbers, (2) higher order association cortical
ﬁbers, and (3) limbic and visceral cortical and subcortical structures can be distinguished. This provides support for a dorsolateral-to-ventromedial-oriented functional organization of the striatum into three functionally different zones, which
appears to be quite universal for different species, including rodent, non-human
primates, and humans (Voorn et al. 2004; Haber et al. 2000; Stoessl et al. 2014)
(Fig. 2.1A). Interestingly, in the human brain, the vascularization of the striatal
complex follows this three-partition and its orientation (Feekes and Cassell 2006)
(Fig. 2.1B). The striatal area innervated by limbic structures like the hippocampus,
amygdala, and ventromedial prefrontal and anterior agranular insular cortical areas
in this way includes the nucleus accumbens and striatal elements of the olfactory
tubercle, as well as ventromedial parts of the caudate nucleus and ventral parts of
the putamen. It is now generally accepted that this ventromedial region of the striatum is the ‘limbic’ striatum.1
The term ‘limbic’ deserves some attention since it is being widely used in the literature, but often
in different ways. We should still keep in mind the words of A. Brodal (1981, page 690), namely
H.J. Groenewegen et al.
Fig. 2.1 Three-partitioning of the striatum based on cortical afferents and vascularization. (A)
Schematic representation of the topographical organization of the projections from functionally
different cortical areas to the striatum. Note that the functional subdivision of the striatum, related
to the corticostriatal topography, does not follow the boundaries between caudate nucleus and
putamen: there exists a dorsolateral-to-ventromedial gradient rather than a functional division
between the caudate nucleus and the putamen. Boundaries between the different functional areas
are not sharply deﬁned but merely consist of transition zones. (B) Three vascular territories shown
in calbindin-immunostained section of the human striatum largely corresponding with the functional three-partitioning shown in (A). The lateral lenticulostriate artery (black arrowhead) supplies the dorsolateral part of the striatum, the medial lenticulostriate artery (white arrowhead)
vascularizes the intermediate striatal zone, and the recurrent artery of Heubner (arrow) supplies the
ventromedial striatum including the nucleus accumbens. From: Feekes and Cassell (2006), ﬁgure
4; Courtesy Martin Cassell and with permission from Oxford University Press. Acb nucleus
accumbens, Cd caudate nucleus, ic internal capsule, Pu putamen
It should be noted that by far the most studies on the striatum, whether anatomical,
electrophysiological or behavioral, concentrate on the rostral parts of the striatal
complex. However, the caudal part of the striatum also contains extensive areas,
including the amygdalostriatal transition zone, which receive inputs from limbic
structures, such as the hippocampus and posterior insular areas. This caudal part of
that the term looses its meaning when the structural and functional deﬁnitions do not coincide and
become so diffuse that ﬁnally the entire brain can be considered to belong to the ‘limbic system’
(cf. also Nauta 1986; Nieuwenhuys 1996). However, whereas the term ‘limbic’ cannot be discarded nowadays, it remains very important to deﬁne what is exactly meant with the term and
which brain areas are considered to be part of the ‘limbic system’. Even though these structures
may have quite diverse functions, we consider the amygdala, hippocampus and hypothalamus as
the ‘core structures’ of the limbic system. Brain regions that are directly inﬂuenced by these core
limbic structures are considered also to belong to the limbic system, i.e., in rodents the ventromedial and insular parts of the prefrontal cortex, midline thalamic nuclei and structures along the
pathway of the medial forebrain bundle (preoptic, hypothalamic and medial midbrain structures).
As indicated in the text, the region of the striatum innervated by ‘limbic’ brain structures mentioned here is considered the ‘limbic striatum’. Nevertheless, the borders between ‘limbic’ and
‘associative/cognitive’ related parts of the striatum remain diffuse.
Limbic-Basal Ganglia Circuits Parallel and Integrative Aspects
the striatum may also be considered part of the ‘limbic’ striatum (e.g., Groenewegen
et al. 1987; Fudge et al. 2004; Heimer et al. 1999). This striatal region still is a relatively unexplored area of the basal ganglia.
The delineation of a ‘limbic’ striatum becomes even more complex when we take
into account the compartmental nature of the striatum. Both dorsal and ventral striatum
contain characteristic inhomogeneities that are primarily visible using neurochemical or
immunohistochemical staining techniques. In the dorsal striatum (caudate nucleus and
putamen) using such methods, patch-matrix (rats) or striosome-matrix compartments
(primates, cats) can be recognized (as originally described by Graybiel and Ragsdale
1978; reviews: Graybiel 1990; Gerfen 1992; Dudman and Gerfen 2015). Within the
ventral striatum, the nucleus accumbens can be subdivided into an outer shell subregion
and an inner core subregion, among others on the basis of the distribution of several
neuropeptides (cholecystokinin, substance P, enkephalin), opioid receptors, and calbindin D28K (e.g. Záborszky et al. 1985; Voorn et al. 1989; Zahm and Brog 1992;
Groenewegen et al. 1999a). Like the dorsal striatal patch and matrix compartments, the
distinction of shell and core subregions of the nucleus accumbens is primarily based
upon differential neurochemical and immunohistochemical characteristics and this differentiation is also supported by numerous behavioral studies (see: Dalley et al. 2004,
2011; Humphries and Prescott 2010; Floresco 2015; Haber and Behrens 2014).
The dorsal part of the core of the nucleus accumbens contains patches like the
dorsally adjacent caudate-putamen complex. As will be brieﬂy touched upon below,
patch and matrix compartments in caudate-putamen as well as shell and core in the
nucleus accumbens have different inputs and outputs (e.g. Graybiel 1990; Gerfen
1992; Groenewegen et al. 1999a). Based on such differential inputs, patches in the
dorsal striatum may represent ‘limbic’ striatal islands in a striatal matrix that on the
basis of its cortical inputs must be considered to belong to the associative part of the
striatum (Gerfen 1992; Berendse et al. 1992; Eblen and Graybiel 1995). This unique
intermingling of limbic and associative striatal elements forms the basis of its established role in the integration of emotional and higher cognitive behavioral functions
(e.g. Friedman et al. 2015). In short, the latter authors showed that the limbic-innervated striosomes located in the associative part of the caudate-putamen inﬂuence
decision-making for choices with cost-beneﬁt tradeoffs that are processed in the
matrix compartment in which these striosomes are embedded.
Afferent Connections of the “Limbic” Striatum
Hippocampal and Amygdaloid Inputs
The hippocampal formation (hippocampus proper and subiculum) and the amygdala
reach primarily the ventromedial parts of the striatum, in particular the nucleus accumbens, the striatal elements of the olfactory tubercle, and the ventromedial parts of the
caudate-putamen (Groenewegen et al. 1987, 1999b; Wright et al. 1996). The hippocampal afferents originate predominantly in the subiculum and to a lesser extent in the
CA1 region and they are mostly restricted to the shell region of the nucleus accumbens,
H.J. Groenewegen et al.
although the medial and rostroventral core also receives hippocampal projections. In
rats, there is a clear topographical arrangement in the hippocampal-striatal projections
in that the ventral hippocampus projects primarily to the medial shell, while progressively more dorsal parts of the hippocampus2 project to successively more rostrolateral
parts of the shell and adjacent core of the nucleus accumbens (Groenewegen et al.
1987; monkey: Friedman et al. 2002) (Fig. 2.2). As will be brieﬂy discussed below, the
excitatory hippocampal inputs interact with thalamic, amygdaloid, and prefrontal cortical inputs. Convergence of these inputs onto individual medium-sized spiny neurons
can either lead to an additive or a competitive effect on these neurons. In the ﬁrst circumstance, the medium-sized output neurons are brought in an ‘upstate’ by one input
and in a state of ﬁring action potentials by the second input. In the second, i.e., the
competitive circumstance, one input prevents the second from bringing the output neurons to become active. In other words, convergence of excitatory inputs may lead to
either opening or closing the gate for striatal output (Calhoon and O’Donnell 2013).
The character of interactions between hippocampal and amygdaloid inputs is dependent on the rostrocaudal level in the nucleus accumbens (Gill and Grace 2011).
The entorhinal cortex, which is closely associated with the hippocampal formation, projects primarily to the lateral core and lateral shell of the nucleus accumbens
and more sparsely to a medial rim of the caudate-putamen, bordering the lateral
ventricle. The entorhinal ﬁbers originate both in the medial and lateral entorhinal
cortex with a slight topographical arrangement, the medial entorhinal ﬁbers terminating more rostrally than the lateral entorhinal ﬁbers (Totterdell and Meredith 1997).
The amygdaloid inputs to the striatum, originating primarily in different subnuclei of the basal amygdaloid complex, have a more widespread distribution and
these projections are likewise topographically organized (Wright et al. 1996).
Caudal parts of the basal amygdaloid complex project to the medial shell and core
of the nucleus accumbens, with a dominance for the caudal part of the nucleus.
Intermediate and rostral parts of the basal amygdala project to, respectively, more
lateral and dorsal parts of shell and core of the nucleus accumbens and the adjacent
caudate-putamen complex (Wright et al. 1996). As can be appreciated from Fig. 2.2,
In primates the posterior-to-anterior axis in the hippocampal formation corresponds to the dorsalto-ventral axis in rodents.
Fig. 2.2 (continued) and amygdaloid ﬁbers in the medial and lateral parts of the nucleus accumbens. Details of the hippocampal and amygdaloid projections can be found in Groenewegen et al.
(1987) and Wright et al. (1996). The distribution of retrogradely labeled neurons in the nucleus
accumbens following injections in the ventromedial and ventrolateral parts of the ventral pallidum
(C) shows a mediolateral topographical organization. A similar conclusion can be drawn for the
organization of the ventral striatal projections to the ventral mesencephalon as shown in (D).
Combining the various patterns of afferents from the hippocampus and amygdala and efferents to
the ventral pallidum and the ventral mesencephalon shows a rich spectrum of input-output channels through the ventral striatum. ac anterior commissure, AcbC core of the nucleus accumbens,
AcbSh shell of the nucleus accumbens, BST bed nucleus of the stria terminalis, cBmg caudal part
of the magnocellular basal amygdaloid nucleus, cBpc caudal part of the parvicellular basal amygdaloid nucleus, CP caudate-putamen, CA1 cornu Ammonis area 1, LPO lateral preoptica area,
lVTA lateral part of the ventral tegmental area, mSN medial part of the substantia nigra, OT olfactory tubercle, rBmg rostral part of the magnocellular basal amygdaloid nucleus, VP ventral pallidum, VPd dorsal subcommissural part of VP, VPm medial part of VP, VPvm ventromedial part of
VP, VPv ventral part of VP, VPvl ventrolateral part of VP, VTA ventral tegmental area
Fig. 2.2 Schematic representation of the distribution of anterograde and retrograde labeling in the
nucleus accumbens and olfactory tubercle following injections in different parts of the hippocampal formation (A), the basal amygdaloid complex (B), the ventral pallidum (C), and the ventral
mesencephalon (D). The drawings of the ventral striatum are based on sections immunostained for
the calcium-binding protein calbindin D28K, showing the shell (AcbSh) and core (AcbC) subregions, as well as the patches in the core and the ventral caudate-putamen (CP). Fibers and terminals
originating from the ventral subiculum (A) and those from the caudal part of the parvicellular basal
amygdaloid nucleus (B) are represented in blue (primarily in the medial shell), those from the
intermediate part of the subiculum (All-IV) and from the mid-rostrocaudal amygdala (BII-IV) are
shown in red (predominantly in the intermediate shell), and the ﬁbers and terminals from the dorsal
subiculum (A) and the rostral part of the magnocellular basal amygdaloid nucleus (B) are depicted
in green (mostly in the lateral shell). Note the varying degrees of overlap between hippocampal
H.J. Groenewegen et al.
the hippocampal and amygdaloid inputs form a complex mosaic of overlapping and
interdigitating projections, in part related to cellular and immunohistochemical heterogeneities in the receiving striatal tissue (Pennartz et al. 1994; Groenewegen et al.
1999b; Voorn et al. 2004).
Inputs from the prefrontal cortex are derived from all cortical subareas and display a
clear topographical relationship with different parts of the striatal complex (Fig. 2.3).
The most ventral areas of the medial prefrontal cortex, i.e., the medial orbital and
infralimbic areas, project to the ventromedial parts of the striatal complex, including
the medial shell of the nucleus accumbens, the medial parts of the olfactory tubercle,
and the ventromedial part of the caudate-putamen complex at the inferior tip of the
lateral ventricle. The medial orbital and infralimbic cortices form the cortical node of
the endocrine parallel corticostriatal loop by way of its efferents to various areas in
the lateral hypothalamus (Kelley 1999). The prelimbic cortex and the more dorsally
situated ventral and dorsal anterior cingulate areas project to successively more lateral and dorsal zones of the caudate-putamen, extending into the dorsal core of the
nucleus accumbens (e.g. Berendse et al. 1992; Voorn et al. 2004) (Fig. 2.3). The laterally located ventral and dorsal agranular insular areas of the prefrontal cortex project to laterally situated regions in the ventral part of the striatal complex. The ventral
agranular insular area reaches the lateral part of the olfactory tubercle and the lateral
shell of the nucleus accumbens; the dorsal agranular insular area entertains the more
dorsally located core of the nucleus accumbens and the adjacent ventral caudateputamen (Berendse et al. 1992) (Fig. 2.3). The orbital cortical areas located in the
ventral part of the frontal lobe likewise show a medial-to-lateral topography in their
projections to the striatum, and these projections extensively overlap with the projections from the medial and lateral prefrontal areas mentioned above (Schilman et al.
2008; Groenewegen and Uylings 2010) (Fig. 2.3). The most dorsolateral part of the
caudate-putamen is not reached by prefrontal afferents, but is innervated by the
somatosensory cortices. The functionally different striatal zones, deﬁned on the basis
of the topography of the projections from functionally different (pre)frontal cortical
areas (Fig. 2.3), give rise to topographically organized striatopallidal and striatonigral projections, which via their associated thalamic nuclei lead back to functionally
related parts of the (pre)frontal cortex, thus forming the well-known parallel basal
Fig. 2.3 (continued) projection area is represented in E and F with a stippled line. The ventrolateral
and lateral orbital areas both project quite strongly to the most lateral part of the shell of the nucleus
accumbens. In C and E, shell and core are delineated with stippled lines (black in C and white in E).
ac anterior commissure, ACd dorsal anterior cingulate cortex, AId dorsal agranular insular cortex, AIv
ventral agranular insular cortex, DLO dorsolateral orbital cortex, FR2 frontal area 2, IL infralimbic
cortex, LO lateral orbital cortex, MO medial orbital cortex, PFC prefrontal cortex, PLd dorsal prelimbic cortex, PLv ventral prelimbic cortex, VLO ventrolateral orbital cortex, VO ventral orbital cortex
Fig. 2.3 Schematic drawing summarizing the topographical arrangement of the cortico-striatal projections originating in the orbital prefrontal cortex (A, C, D) and the medial and lateral prefrontal
cortices (A, B, E, F). The prefrontal cortical areas and their connectionally related striatal targets are
coded in the same color. Since there is a considerable overlap between the orbital projections on the
one hand (C, D) and the medial and lateral prefrontal projections on the other hand (E, F), these
projections are represented in two different sets of a rostral and a caudal striatal level. As shown in
(E) and (F), the dorsolateral striatum receives somatotopically organized inputs from the sensorimotor cortices (light blue), the most ventromedial part of the striatum collects inputs from the infralimbic and ventral prelimbic areas (red and purple). Striatal areas intermediate between these extremes
receive projections from the dorsal prelimbic, anterior cingulate, and Fr2 areas. The ventral agranular
insular area projects to the lateral shell and adjacent olfactory tubercle, while the dorsal agranular
insular area sends ﬁbers to the core (E) and a broad mediolateral zone in the ventral caudate-putamen
more caudally (F). Although the global relationships between the projection areas from different
medial and lateral prefrontal cortices are maintained from rostral to caudal, the relative space occupied by the projections from various cortical areas changes from rostral to caudal (compare E and F).
The orbital prefrontal projection areas in the striatum show a medial-to-lateral topographical organization (C, D). The medial and ventral orbital areas overlap considerably in the medial part of the
striatum, while the lateral and dorsolateral orbital areas overlap quite extensively in the lateral part of
the striatum (stippled lines). In an intermediate position, in the central part of the caudate-putamen,
lies the projection area of the ventrolateral orbital area. To show the overlap of the ventrolateral
orbital projection with the projections of the medial and lateral prefrontal areas, the ventrolateral
H.J. Groenewegen et al.
It must be realized that the representation of the topographical organization of
the prefrontal corticostriatal projections as represented in Fig. 2.3 is a schematic
one, aimed to present an ‘easy’ understandable overview, emphasizing the
ventromedial-to-dorsolateral organizational aspect in the limbic-to-cognitive-tomotor corticostriatal afferent systems. Yet, overlap between the projections from
(pre)frontal cortical and parietal, occipital, and temporal cortical areas at the level
of the striatum also provides the anatomical substrate for interactions and integration of information between and within these circuits. With respect to prefrontal and
orbitofrontal corticostriatal projections, it has recently been shown that there is
more extensive and also speciﬁc overlap between individual prefrontal corticostriatal projections than previously assumed (Mailly et al. 2013) (Fig. 2.4 dealt with
below). This is primarily based on the fact that two patterns of corticostriatal projections can be distinguished. Thus, all prefrontal cortical areas have a primary, ‘focal’
striatal target area in which dense projections from that particular cortical area terminate, as well as a quite extensive more ‘diffuse’ terminal ﬁeld that is distributed
along the borders of the focal terminal ﬁeld in the striatum, expanding the striatal
area that is reached by a particular cortical area. It is important to note that the relative extension of the dense and diffuse areas is not related to the extent of the injection sites in the cortex, but is primarily related to the identity of the injected cortical
area. Such dual mode of corticostriatal innervation observed following the tracing
of the projections of deﬁned cortical regions, involving large collections of neurons
in different layers, may be reminiscent of the patterns of axonal arborizations
observed at a single cell level (Kincaid and Wilson 1996; Zheng and Wilson 2002).
Thus, some corticostriatal neurons provide a high density of terminals in a small
striatal volume, whereas others have a low background innervation. These two
modes of corticostriatal innervation appear to originate from distinct cortical layers
(Kincaid and Wilson 1996; Zheng and Wilson 2002; cf also Wright et al. 1999).
Whether this is indeed the case for the dense and diffuse projections originating in
the prefrontal cortex needs to be further established and would be of functional
interest since different layers of the cortex receive functionally different inputs
(Calzavara et al. 2007; Haber and Calzavara 2009; Mailly et al. 2013).
Fig. 2.4 (continued) projection areas are represented. Both the degree of overlap between the projections from different cortical areas as well as their ‘private’ areas can be appreciated and is quantitatively shown in (C). (C) Using the methodology described in Mailly et al. (2010), the volumes
of overlap between the focal projections of the nine different cortical areas have been calculated.
The pie charts give a quantitative representation of these volumes of overlap. In each case, the portions of the pie chart represent the amount of the focal projection ﬁeld from a given cortical area that
is overlapped by the focal projection ﬁelds from the other cortical areas. The portion of the pie chart
represented with the color code of the reference area indicates the ‘private’ part of the focal projection ﬁeld remaining segregated from the focal projection ﬁelds of all other areas. Notable is the lack
of complete overlap for the focal projections (B) and the rather limited portion of the projection
remaining segregated from other cortical focal projection areas (C). Color codes: DLO (purple),
VLO (green), PL (yellow), MOVO (red), ACd (orange), IL (brown), ACv (cyan), AID (magenta),
and FR2 (blue). ACd dorsal anterior cingulate cortex, ACv ventral anterior cingulate cortex, AID
dorsal agranular insular cortex, cc corpus callosum, DLO dorsolateral orbital cortex, FR2 frontal
area 2, IL infralimbic cortex, MOVO medial orbital and ventral orbital cortices, PL prelimbic cortex,
VLO ventrolateral orbital cortex. Adapted from ﬁgures 1 and 6 in Mailly et al. (2013)