Introduction
The ventrobasal forebrain nuclei, including the nucleus accumbens (Ac), bed nucleus of stria terminalis, lateral part (BSTL) and other components of extended amygdala (EA), ventral pallidum (VP) and cholinergic cell groups (such as the basal nucleus of Meynert) have been implicated in the initiation and reinforcement of movements, motivation and emotion, reward and aversion (Alheid et al.
1995; Alheid and Heimer
1988; de Olmos et al.
2004; Li and Sakaguchi
1997). These regions are extensively connected with the amygdala, whose involvement in emotional responses is also well established (Phelps and LeDoux
2005; Swanson
2000). While the majority of relevant studies focused on mammalian species, our laboratory has been active in revealing a similar role of relevant systems in birds, in which the anatomical structures and connectivity of ventrobasal forebrain nuclei show extensive homologies with their mammalian counterparts (Csillag
1999; Csillag et al.
1997,
2008; Csillag and Montagnese
2005; Jarvis et al.
2005; Kuenzel et al.
2011; Reiner et al.
2004). As a model system, young domestic chicks offer a unique opportunity for studying, learning and motivation because of their early maturation (precocial development of a nidifugous species). For example, one-trial passive avoidance training is a simple and reproducible way of investigating early adaptive learning processes (Rose
2000). It has been established that, of all telencephalic regions, the basal ganglia show the highest degree of homology between birds and mammals. This may serve as justification for a comparative approach in the investigation of neural mechanisms, such as motivation of elementary actions, which have been conserved throughout vertebral evolution in both mammals and Sauropsida (diverging over 200 million years ago).
Earlier results from our laboratory, based on electron microscopic immunocytochemistry, have indicated the presence of
l-Asp and
l-Glu in excitatory axodendritic boutons in the striatum/accumbens region of rats and chickens (Adam and Csillag
2006; Hanics et al.
2012). The source of these terminals proved to be the basolateral amygdala (BLA) in the rat, whereas in chicks the source region was located in the amygdaloid arcopallium (Hanics et al.
2012). Further, to investigate the specificity of
l-Asp containing input pathways in the domestic chicken, it was necessary to carry out a detailed pathway tracing study, combined with immunohistochemistry, relevant to the chemical nature of source neurons and of potential target areas.
Description of projection patterns between select brain areas critically relies on domain-specific tract tracing approach. To project the arcopallial output onto the avian ventrobasal forebrain regions including the Ac, EA, BSTL, we applied high precision region-specific in vivo anterograde and retrograde tracing experiments, including dual tracing by simultaneously using different dyes. The analysis of serial high power magnification multi-tile section reconstructions allowed us to map a hitherto uncharacterized amygdalosubpallial pathway.
Discussion
Retrograde tracing with choleratoxin B subunit (CTb), injected into the Ac, yielded labeled perikarya in a ring-shaped area of arcopallium, including the amygdalar dorsal region of Puelles et al. (
2007) (ADo), corresponding to dorsal arcopallium (AD), according to Atoji et al. (
2006); the amygdalohippocampal area (AHi) of Puelles et al. (
2007), largely corresponding to the nucleus taeniae (TnA) of Atoji et al. (
2006); the hilar amygdalar region (AHil) of Puelles et al. (
2007), approximate correlate of medial arcopallium, parvocellular part (AMp), according to Atoji et al. (
2006). A wedge-shaped node of dense accumulation of retrogradely labeled cells was observed in a laterodorsal subunit termed amygdalopiriform area (APir, Puelles et al. (
2007)). The latter region largely coincided with the fields designated by Atoji et al. (
2006) as the caudal ventrolateral nidopallium (NCVl), subnidopallium (SuN) and the posterior nucleus of arcopallium, compact division (PoAc), in the pigeon. Further retrogradely labeled cells were found in the posterolateral amygdala (APL) of Puelles et al. (
2007), similar to the regions designated by Atoji et al. (
2006) as basal posterior arcopallium (PoAb). Injections spreading into the BSTL led to similar distribution of labeled cells in the arcopallium. This is in agreement with the observation of Atoji et al. (
2006) concerning the source region of BSTL projections. However, these authors restricted their analysis to the EA (including BSTL), not considering the Ac (corresponding to the generally accepted notion about the position of Ac back then). Later studies (Balint and Csillag
2007; Balint et al.
2011; Husband and Shimizu
2011) have led to a reappraisal of the position of Ac subregions in the domestic chicken (in the vicinity of BSTL throughout the rostrocaudal extent A8.8–A10.6 of ventrobasal forebrain, according to the coordinates of Kuenzel and Masson (
1988)). Thus, overlapping simultaneous projections from the arcopallium to both BSTL and Ac have become a distinct possibility.
Notably, the ATn was largely devoid of retrogradely labeled neurons unlike the adjacent AHil region, which contained abundant CTb+ cells. In addition to arcopallial sources, retrogradely labeled neurons were also seen in the EA, particularly its border region with the arcopallium.
The position of source neurons for the arcopallial-accumbens pathway was verified also by anterograde pathway tracing. The results of more refined analysis, based on discrete subregional injections, show that the fibers arising from the APir are likely to reach the Ac (in addition to BSTL), whereas those arising from the dorsal and medial arcopallial subdivisions mainly innervate the BSTL and EA only. Overall, the projection to any ventrobasal target area was more dense in those cases, where the tracer had been deposited in the laterodorsal (APir) area of arcopallium, in agreement with a greater density of source neurons there (as detected by retrograde tracing).
The study enabled precise topographic description of the course of the arcopalliofugal pathway (essentially corresponding to the amydalofugal pathway in question). It derives from two main output fiber streams, also mentioned by Atoji et al. (
2006): one along the dorsal border of arcopallium, presumably corresponding to the stria terminalis of mammals, and another ventral tract along the ventral pallial border (putative equivalent of the ansa peduncularis of mammals). Further course of the pathway can be traced in our material as follows. The fibers arising from caudal levels follow a dorsal course and enter the vaf. Then, having bypassed the ATn, they traverse the subpallial (extended) amygdala (with profuse terminal fields), and the BSTL (also terminating there in large numbers) before invading the shell and the core of Ac. The fibers arising from levels that are more rostral mainly follow a ventral course, passing through the ventrobasal part of EA, and then invading the nucleus basalis and olfactory tubercle. Efferents were also observed in the ventral pallidum, lateral septum and diagonal band. It has to be noted that the nucleus taeniae (ATn) of Puelles et al. (
2007), adjacent to the vaf, is not identical with the nucleus termed TnA by Atoji et al. (
2006), which is placed at some distance from the vaf. We reconstructed the course of the amygdalofugal pathway in a pseudo-3D (movie) format (Electronic Supplementary Material 2).
The presence of DARPP-32 has been well established, also in avian brain regions (Durstewitz et al.
1998; Roberts et al.
2002). This protein is an important signaling molecule present in dopaminoceptive neurons (Hemmings et al.
1987). In agreement with previous observations (Schnabel et al.
1997), in our study DARPP-32 immunoreactivity was present in all striatal regions (including the Ac), but the BSTL was largely devoid of DARPP-32. DARPP-32 in ‘NST’ (an earlier name variant for BSTL) has been reported poor staining by Reiner et al. (
1998), together with a low density of substance P (SP). DARPP-32 labeling was found to be prominently weaker in the BSTL than in the surrounding ventral striatum (identified as the rostral pole of Ac) (Balint and Csillag
2007). Thus, immunoreactivity for DARPP-32 could be used as a marker distinguishing adjacent Ac and EA regions. Interestingly, Ac-bound arcopallial neurons were devoid of DARPP-32, except for a few cells in the lateral nidopallium bordering the dorsal arcopallium. Apparently, a ring of DARPP-32 containing cells surrounds the arcopallial source region of the amygdalofugal pathway, without considerable overlap. Massive labeling against DARPP-32 in the caudolateral nidopallium and piriform cortex, adjacent to the APir (but not in central arcopallial fields) has been observed also by Schnabel et al. (
1997). In the same study, the amount of TH labeling was found to be very high in the dorsal and laterodorsal arcopallium (overlapping the source regions of our present study), which otherwise showed weaker labeling to DARPP-32. In most cases, absence of this signaling molecule does not involve a similar lack of dopaminergic innervation. DARPP-32 labeling was found to be low in the ‘Ac’, despite a dense staining of TH fibers (Schnabel et al.
1997). It has to be noted that the region defined by these authors as Ac was later renamed BSTL (Reiner et al.
2004). Distribution of DARPP-32 labeling, if overall similar, was by no means an exact match of the distribution of dopamin D1 receptors (Ball et al.
1995), in the quail (Absil et al.
2001).
An important finding is that the source cells of the amygdalofugal tract specified in the present study are devoid of calbindin, calretinin and parvalbumin, albeit these calcium-binding proteins do occur in many neighboring cells, profusely intermingling with the retrogradely traced neurons. All three calcium-binding proteins are known to be widely distributed in various subregions of mammalian amygdala (Pitkanen and Kemppainen
2002). The presence of calcium binding proteins has been typically exploited for the identification of functionally distinct neuronal subsets also in the avian brain (Gati et al.
2014; Husband and Shimizu
2011; Pfeiffer and Britto
1997; Roberts et al.
2002; Suarez et al.
2006). The arcopallium harbors subsets of parvalbumin
+, calbindin
+ and calretinin
+ neurons (Cornez et al.
2015; Roberts et al.
2002). The observed lack of calcium binding proteins is in harmony with our previous finding that at least a contingent of the source neurons of the amygdalofugal pathway are excitatory based on the presence of glutamate and aspartate in asymmetrical synaptic terminals (of excitatory morphological type), deriving from amygdalofugal axons (Hanics et al.
2012). The calcium binding proteins calbindin D28K, calretinin and parvalbumin tend to occur in smooth non-pyramidal interneurons (and some pyramidal neurons) of mammalian cortex (for review: DeFelipe (
1997)). At least in cortical fields, these calcium-binding proteins mark specific classes of inhibitory interneurons (Hof et al.
1999).
Of the arcopallial subregions yielding the densest projections to BSTL and Ac, the dorsolateral arcopallium and neighboring caudal nidopallial regions have been considered to be lateral pallial derivatives and homologous to the basolateral amygdala of mammals or reptiles (Guirado et al.
2000; Lanuza et al.
1998; Martinez-Garcia et al.
2002,
2008; Redies et al.
2001). However, this has been disputed by other authors, categorizing the regions rather as ventral pallial derivatives (Medina et al.
2004; Puelles et al.
2000), though maintaining the possibility of part of basolateral amygdala being ventral pallial (Abellan et al.
2009). Thus, the main source region for projections directed to BSTL/Ac may still be categorized, as equivalent of mammalian BLA, since a lateral pallial origin, at least in part, has not been ruled out. This interpretation is in agreement with Moreno and Gonzalez (
2006), placing the APir and caudolateral nidopallium (NCL) into a lateral pallial zone, while other, less dense source regions (ADo, AHi, AHil) would already belong in the ventral pallial field. The ATn (amygdaloid taenial nucleus, designated as pallial medial amygdalar nucleus, PMA, by Abellan et al. (
2009) was devoid of retrogradely labeled cells, at least at the sectional levels of its largest extension. This region did not contain anterogradely labeled fibers either; the fibers seem to pass by the nucleus without termination. In addition to pallial amygdalar sources of the pathway, retrogradely labeled cells were also observed en route in the extended amygdala, especially in the region adjacent to the arcopallium (including the capsular central amygdala, intercalated cell patches, peri-INP island field and the oval central amygdalar nucleus, according to the categories by Vicario et al. (
2014).
Convergent amygdalar input to the EA and BSTL, as well as to Ac subregions likely transmits contextual fear and aggression-related signals to both viscerolimbic (EA) and learned reward- and motivation-related (Ac) ventrobasal forebrain regions. Fear responses have been attributed to either the central amygdala or the pallial laterobasal amygdala (Davis and Whalen
2001). According to previous observations, the BSTL is primarily involved in contextual fear (Duvarci et al.
2009; Phelps and LeDoux
2005; Walker and Davis
2008), whereas the central amygdala is more involved in lasting fear responses, similar to anxiety (Duvarci et al.
2009; Walker and Davis
2008; Walker et al.
2003,
2009). Based on evidence from the previous (Hanics et al.
2012) and present observations, this pathway is excitatory, with potential cotransmission of Glu and Asp. Dopaminergic input to the source neurons of this pathway is unlikely to involve DARPP-32 as the main signal transducer, as evidenced by the present study.
There appears to be a certain degree of overlap between the numerous subregions of EA and those of the Ac (in particular the shell). The EA can be envisaged as a network of neurons of multiple origins, extending from selected nuclei of the amygdala to specific areas of the ventrobasal forebrain. In the course of development, these neurons originating from subpallial (medial ganglionic eminence, lateral ganglionic eminence, preoptic) and pallial (esp. ventral pallial) primordia followed specific migratory routes or cell subcorridors (Vicario et al.
2015), e.g., the stria terminalis. Because of migration, cellular clusters originating from one domain may invade the territories of other domains, rendering the borders ‘fuzzy’. This may well be the case with the border between the Ac and BSTL or, even more so, the border between Ac and rostral extended amygdala (SpAr). Yet certain cellular characteristics may be preserved in spite of overlapping migration. For example, a recently described calcium binding protein, secretagogin, known to occur in EA regions of mammals (Mulder et al.
2010) labels clusters of selected neurons also in the subpallial amygdala, including BSTL, of domestic chickens, whereas the Ac (together with the striatal complex) are impoverished in secretagogin label (Gati et al.
2014). Based on a detailed study on specific transcription factors and cell tracking in chicken (Vicario et al.
2015), the BSTLd (as defined in the paper) contains neurons of both pallidal and striatal origin. While BSTL develops in the pallidal domain of the forebrain, Ac is a ventral striatal derivative. According to Vicario et al. (
2015), the amount of Islet1 (taken as a marker for the ventral striatal domain) was significant, mainly in the dorsomedial BSTL but also visible in the dorsolateral subregion, interspersed with Pax6 and Nkx.2.1., pointing to a potential ‘overflow’ from Ac. The area defined SpAr may also encroach upon territories of Ac shell [Alheid et al.
1995; de Olmos et al.
2004), see further discussion of the question by Alba Vicario, doctoral thesis (2015)].
Of the two known subregions of EA (central and medial) (Martinez-Garcia et al.
2008), the central EA includes the BSTL, and is implicated in fear and aggression-related behaviors. Thus, the BSTL, relevant central amygdalar components and specific neural groups, scattered along the path of the stria terminalis and the associated vaf, are likely involved in mediating these modalities to other viscerolimbic centers (diagonal band nucleus, septum, and hypothalamus and forebrain cholinergic system). However, the very same information is salient also to the processing of reward, aversion and memory formation for these modalities, as well as the initiation of locomotor response and cognitive functions based thereupon, all considered to be typical for the Ac. This dichotomy of viscerolimbic-related and reward-related amygdalar input may be represented in the described pathway of the domestic chicken, terminating in both EA and Ac regions.