Characterization of factors regulating lamina‐specific growth of thalamocortical axons

During development, most thalamocortical axons extend through the deep layers to terminate in layer 4 of neocortex. To elucidate the molecular mechanisms that underlie the formation of layer‐specific thalamocortical projections, axon outgrowth from embryonic rat thalamus onto postnatal neocortical slices which had been fixed chemically was used as an experimental model system. When the thalamic explant was juxtaposed to the lateral edge of fixed cortical slice, thalamic axons extended farther in the deep layers than the upper layers. Correspondingly, thalamic axons entering from the ventricular side extended farther than those from the pial side. In contrast, axons from cortical explants cultured next to fixed cortical slices tended to grow nearly as well in the upper as in the deep layers. Biochemical aspects of lamina‐specific thalamic axon growth were studied by applying several enzymatic treatments to the cortical slices prior to culturing. Phosphatidylinositol phospholipase C treatment increased elongation of thalamic axons in the upper layers without influencing growth in the deep layers. Neither chondroitinase, heparitinase, nor neuraminidase treatment influenced the overall projection pattern, although neuraminidase slightly decreased axonal elongation in the deep layers. These findings suggest that glycosylphosphatidylinositol‐linked molecules in the cortex may contribute to the laminar specificity of thalamocortical projections by suppressing thalamic axon growth in the upper cortical layers. © 2000 John Wiley & Sons, Inc. J Neurobiol 42: 56–68, 2000     

Ephrins regulate the formation of terminal axonal arbors during the development of thalamocortical projections

The development of connections between thalamic afferents and their cortical target cells occurs in a highly precise manner. Thalamic axons enter the cortex through deep cortical layers, then stop their growth in layer 4 and elaborate terminal arbors specifically within this layer. The mechanisms that underlie target layer recognition for thalamocortical projections are not known. We compared the growth pattern of thalamic explants cultured on membrane substrates purified from cortical layer 4, the main recipient layer for thalamic axons, and cortical layer 5, a non-target layer. Thalamic axons exhibited a reduced growth rate and an increased branching density on their appropriate target membranes compared with non-target substrate. When confronted with alternating stripes of both membrane substrates, thalamic axons grew preferentially on their target membrane stripes. Enzymatic treatment of cortical membranes revealed that growth, branching and guidance of thalamic axons are independently regulated by attractive and repulsive cues differentially expressed in distinct cortical layers. These results indicate that multiple membrane-associated molecules collectively contribute to the laminar targeting of thalamic afferents. Furthermore, we found that interfering with the function of Eph tyrosine kinase receptors and their ligands, ephrins, abolished the preferential branching of thalamic axons on their target membranes, and that recombinant ephrin-A5 ligand elicited a branch-promoting activity on thalamic axons. We conclude that interactions between Eph receptors and ephrins mediate branch formation of thalamic axons and thereby may play a role in the establishment of layer-specific thalamocortical connections.     

Acetylcholine influences growth cone motility and morphology of developing thalamic axons

The neurotransmitter acetylcholine (ACh) is expressed in the developing telencephalon at the time when thalamic axons project to the cortex, long before synapses are being formed. Since previous studies demonstrated an influence of ACh on neurite extension we used different in vitro assays to examine possible effects of ACh on the growth of thalamic axons. In explant cultures, application of ACh reduced the length of thalamic axons in a dose dependent manner, an effect that could also be evoked by selective muscarinic and nicotinic agonists. Time-lapse imaging of thalamic axons exposed to microscopic gradients of ACh revealed that growth cones no longer advanced, but maintained high filopodial activity. This growth cone pausing was not accompanied by axon retraction or growth cone collapse. It could at least partially be blocked by muscarinic and nicotinic antagonists, indicating that both types of ACh receptors contribute to mediate these effects on thalamic axons. Finally, we also found that ACh changed the morphology of growth cones; they became larger and extended more filopodia. Since such changes in the structure and motility of growth cones are observed at decision regions along the path of many fiber populations including thalamic axons, we suggest that ACh plays a role during the elaboration of thalamocortical projections.Key words: cortical development, thalamocortical projections, neurotransmitter, acetylcholine, growth cone, axonal guidance, wiring molecules     

Multiple roles of ephrins during the formation of thalamocortical projections: maps and more

The functional architecture of the cerebral cortex is based on intrinsic connections that precisely link neurons from distinct cortical laminae as well as layer-specific afferent and efferent projections. Experimental strategies using in vitro assays originally developed by Friedrich Bonhoeffer have suggested that positional cues confined to individual layers regulate the assembly of local cortical circuits and the formation of thalamocortical projections. One of these wiring molecules is ephrinA5, a ligand for Eph receptor tyrosine kinases. EphrinA5 and Eph receptors exhibit highly dynamic expression patterns in distinct regions of the cortex and thalamus during early and late stages of thalamocortical and cortical circuit formation. In vitro assays suggest that ephrinA5 is a multifunctional wiring molecule for different populations of cortical and thalamic axons. Additionally, the expression patterns of ephrinA5 during cortical development are consistent with this molecule regulating, in alternative ways, specific components of thalamic and cortical connectivity. To test this directly, the organization of thalamocortical projections was examined in mice lacking ephrinA5 gene expression. The anatomical studies in ephrinA5 knockout animals revealed a miswiring of limbic thalamic projections and changes in neocortical circuits that were predicted from the expression pattern and the in vitro analysis of ephrinA5 function.     

Effects of the thalamus on the development of cerebral cortical efferents in vitro

The cerebral cortex is a multilayered tissue, with each layer differing in its cellular composition and connections. Axons from deep layer neurons project subcortically, many to the thalamus, whereas superficial layer neurons target other cortical areas. The mechanisms that regulate the development of this pattern of connections are not fully understood. Our experiments examined the potential of the thalamus to attract and/or select neurites from appropriate cortical layers. First, we cocultured murine cortical slices in close proximity to thalamic explants in collagen gels. The amount of neurite outgrowth from deep layer cells was enhanced by, but not attracted to, the thalamic explants. Second, we cocultured cortical slices in contact with thalamic or cortical explants to test for laminar specificity of connections. Specificity was apparent after culture for about a week, in that deep cortical layers contained the highest proportions of corticothalamic cells and superficial cortical layers contained the highest proportions of corticocortical cells. After shorter culture of only a few days, however, specificity was not apparent and there were larger numbers of corticothalamic projections from the superficial layers than after a week. To study how the early nonspecific pattern of corticothalamic connections was transformed into the more specific pattern, we labeled corticothalamic cells early, after 2 days, but let the cultures survive for 8 days. On day 8, the nonspecific pattern of early-labeled cells was still seen. We conclude that although the thalamus does not block the initial entry of inappropriate axons from the superficial layers, many of these axons are subsequently lost. This suggests that contact-mediated interactions between cortical axons and the thalamus allow cortical efferents from appropriate layers to be distinguished from those arising in inappropriate layers. This may contribute to the development of layer-specific cortical connections in vivo.     

Area specificity and topography of thalamocortical projections are controlled by ephrin/Eph genes

The mechanisms generating precise connections between specific thalamic nuclei and cortical areas remain poorly understood. Using axon tracing analysis of ephrin/Eph mutant mice, we provide in vivo evidence that Eph receptors in the thalamus and ephrins in the cortex control intra-areal topographic mapping of thalamocortical (TC) axons. In addition, we show that the same ephrin/Eph genes unexpectedly control the inter-areal specificity of TC projections through the early topographic sorting of TC axons in an intermediate target, the ventral telencephalon. Our results constitute the first identification of guidance cues involved in inter-areal specificity of TC projections and demonstrate that the same set of mapping labels is used differentially for the generation of topographic specificity of TC projections between and within individual cortical areas.     

The early topography of thalamocortical projections is shifted in Ebf1 and Dlx1/2 mutant mice

The prevailing model to explain the formation of topographic projections in the nervous system stipulates that this process is governed by information located within the projecting and targeted structures. In mammals, different thalamic nuclei establish highly ordered projections with specific neocortical domains and the mechanisms controlling the initial topography of these projections remain to be characterized. To address this issue, we examined Ebf1(-/-) embryos in which a subset of thalamic axons does not reach the neocortex. We show that the projections that do form between thalamic nuclei and neocortical domains have a shifted topography, in the absence of regionalization defects in the thalamus or neocortex. This shift is first detected inside the basal ganglia, a structure on the path of thalamic axons, and which develops abnormally in Ebf1(-/-) embryos. A similar shift in the topography of thalamocortical axons inside the basal ganglia and neocortex was observed in Dlx1/2(-/-) embryos, which also have an abnormal basal ganglia development. Furthermore, Dlx1 and Dlx2 are not expressed in the dorsal thalamus or in cortical projections neurons. Thus, our study shows that: (1) different thalamic nuclei do not establish projections independently of each other; (2) a shift in thalamocortical topography can occur in the absence of major regionalization defects in the dorsal thalamus and neocortex; and (3) the basal ganglia may contain decision points for thalamic axons' pathfinding and topographic organization. These observations suggest that the topography of thalamocortical projections is not strictly determined by cues located within the neocortex and may be regulated by the relative positioning of thalamic axons inside the basal ganglia.     

Topography of thalamic projections requires attractive and repulsive functions of Netrin-1 in the ventral telencephalon

Recent studies have demonstrated that the topography of thalamocortical (TC) axon projections is initiated before they reach the cortex, in the ventral telencephalon (VTel). However, at this point, the molecular mechanisms patterning the topography of TC projections in the VTel remains poorly understood. Here, we show that a long-range, high-rostral to low-caudal gradient of Netrin-1 in the VTel is required in vivo for the topographic sorting of TC axons to distinct cortical domains. We demonstrate that Netrin-1 is a chemoattractant for rostral thalamic axons but functions as a chemorepulsive cue for caudal thalamic axons. In accordance with this model, DCC is expressed in a high-rostromedial to low-caudolateral gradient in the dorsal thalamus (DTh), whereas three Unc5 receptors (Unc5A–C) show graded expression in the reverse orientation. Finally, we show that DCC is required for the attraction of rostromedial thalamic axons to the Netrin-1–rich, anterior part of the VTel, whereas DCC and Unc5A/C receptors are required for the repulsion of caudolateral TC axons from the same Netrin-1–rich region of the VTel. Our results demonstrate that a long-range gradient of Netrin-1 acts as a counteracting force from ephrin-A5 to control the topography of TC projections before they enter the cortex.     

Development and plasticity of cortical areas and networks

The development of cortical layers, areas and networks is mediated by a combination of factors that are present in the cortex and are influenced by thalamic input. Electrical activity of thalamocortical afferents has a progressive role in shaping cortex. For early thalamic innervation and patterning, the presence of activity might be sufficient; for features that develop later, such as intracortical networks that mediate emergent responses of cortex, the spatiotemporal pattern of activity often has an instructive role. Experiments that route projections from the retina to the auditory pathway alter the pattern of activity in auditory thalamocortical afferents at a very early stage and reveal the progressive influence of activity on cortical development. Thus, cortical features such as layers and thalamocortical innervation are unaffected, whereas features that develop later, such as intracortical connections, are affected significantly. Surprisingly, the behavioural role of 'rewired' cortex is also influenced profoundly, indicating the importance of patterned activity for this key aspect of cortical function.     

Early thalamocortical tract guidance and topographic sorting of thalamic projections requires LIM-homeodomain gene Lhx2

The thalamocortical tract is the primary source of sensory information to the cerebral cortex, but the mechanisms regulating its pathfinding are not completely understood. LIM-homeodomain (LIM-HD) gene Lhx2 has been proposed to participate in a combinatorial "code" to regulate dorsal thalamic patterning and also the topography of thalamocortical projections. Here, we report that Lhx2-/- embryos exhibit a gross disruption in the early development of the thalamocortical tract, such that thalamic axons are unable to enter the ventral telencephalon. A possible cause for this deficit is a severe reduction of "pioneer" cells in the mutant ventral telencephalon that constitutes a putative mechanism for guiding the entry of the thalamocortical tract into this structure in vivo. However, in vitro, the thalamocortical tract is able to enter the ventral telencephalon, and this permitted an examination of whether thalamocortical topography is normal in the Lhx2 mutant. Contrary to hypotheses that proposed a cell-autonomous role for Lhx2 in the thalamus, Lhx2-/- thalamic explants generate a normal topography of projections in control ventral telencephalic preparations. This is consistent with our findings of normal patterning of the Lhx2 mutant dorsal thalamus using a wide array of markers. In the reverse experiment, however, control thalamic explants display aberrant topography in Lhx2-/- telencephalic preparations. This perturbation is restricted to projections from caudal thalamic explants, while rostral and middle explants project normally. Thus Lhx2 is required for multiple steps in thalamocortical tract pathfinding, but these functions appear localized in the ventral telencephalon rather than in the dorsal thalamic neurons. Furthermore, the absence of Lhx2 in the ventral telencephalon selectively disrupts a subset of thalamic axon topography, indicating a specific rather than a general perturbation of cues in this structure.     

Evidence that descending cortical axons are essential for thalamocortical axons to cross the pallial-subpallial boundary in the embryonic forebrain

Developing thalamocortical axons traverse the subpallium to reach the cortex located in the pallium. We tested the hypothesis that descending corticofugal axons are important for guiding thalamocortical axons across the pallial-subpallial boundary, using conditional mutagenesis to assess the effects of blocking corticofugal axonal development without disrupting thalamus, subpallium or the pallial-subpallial boundary. We found that thalamic axons still traversed the subpallium in topographic order but did not cross the pallial-subpallial boundary. Co-culture experiments indicated that the inability of thalamic axons to cross the boundary was not explained by mutant cortex developing a long-range chemorepulsive action on thalamic axons. On the contrary, cortex from conditional mutants retained its thalamic axonal growth-promoting activity and continued to express Nrg-1, which is responsible for this stimulatory effect. When mutant cortex was replaced with control cortex, corticofugal efferents were restored and thalamic axons from conditional mutants associated with them and crossed the pallial-subpallial boundary. Our study provides the most compelling evidence to date that cortical efferents are required to guide thalamocortical axons across the pallial-subpallial boundary, which is otherwise hostile to thalamic axons. These results support the hypothesis that thalamic axons grow from subpallium to cortex guided by cortical efferents, with stimulation from diffusible cortical growth-promoting factors.     

Hypocretin (orexin) induces calcium transients in single spines postsynaptic to identified thalamocortical boutons in prefrontal slice

In vivo, thalamocortical axons are susceptible to the generation of terminal spikes which antidromically promote bursting in the thalamus. Although neurotransmitters could elicit such ectopic action potentials at thalamocortical boutons, this hypothesis has never been confirmed. Prefrontal cortex is the cortical area most implicated in arousal and is innervated by thalamic neurons that are unusual since they burst rhythmically during waking. We show that a neurotransmitter critical for alertness, hypocretin (orexin), directly excites prefrontal thalamocortical synapses in acute slice. This TTX-sensitive activation of thalamic axons was demonstrated electrophysiologically and by two-photon sampling of calcium transients at single spines in apposition to thalamic boutons anterogradely labeled in vivo. Spines receiving these long-range projections constituted a unique population in terms of the presynaptic excitatory action of hypocretin. By this mechanism, the hypocretin projection to prefrontal cortex may play a larger role in prefrontal or "executive" aspects of alertness and attention than previously anticipated.     

Cooperative activity of multiple upper layer proteins for thalamocortical axon growth

During development, sensory thalamocortical (TC) axons grow into the neocortex and terminate primarily in layer 4. To study the molecular mechanism that underlies lamina-specific TC axon termination, we investigated the responsiveness of TC axons to ephrin-A5, semaphorin-7A (Sema7A) and kit ligand (KL), which are expressed in the upper layers of the developing cortex. Dissociated cells of the dorsal thalamus from embryonic rat brain were cultured on dishes that were coated with preclustered Fc-tagged extracellular domains of these molecules. Each protein was found to promote TC axon growth in a dose-dependent fashion of a bell-shaped curve. Any combination of the three proteins showed a cooperative effect in lower concentrations but not in higher concentrations, suggesting that their growth-promoting activities act in a common pathway. The effect of spatial distributions of these proteins was further tested on a filter membrane, in which these proteins were printed at a size that recapitulates the scale of laminar thickness in vivo, using a novel protein-printing technique, Simple-To-mAke Micropore Protein-Printing (STAMP2) method. The results demonstrated that TC axons grew massively on the laminin-coated region but were prevented from invading the adjacent ephrin-A5-printed region, suggesting that TC axons detect relative differences in the growth effect between these regions. Moreover, the inhibitory action of ephrin-A5 was enhanced by copresence with KL and Sema7A. Together, these results suggest that the lamina-specific TC axon targeting mechanism involves growth-inhibitory activity by multiple molecules in the upper layers and detection in the molecular environments between the upper and deep layers.     

Specificity and Plasticity of Thalamocortical Connections in Sema6A Mutant Mice

The establishment of connectivity between specific thalamic nuclei and cortical areas involves a dynamic interplay between the guidance of thalamocortical axons and the elaboration of cortical areas in response to appropriate innervation. We show here that Sema6A mutants provide a unique model to test current ideas on the interactions between subcortical and cortical guidance mechanisms and cortical regionalization. In these mutants, axons from the dorsal lateral geniculate nucleus (dLGN) are misrouted in the ventral telencephalon. This leads to invasion of presumptive visual cortex by somatosensory thalamic axons at embryonic stages. Remarkably, the misrouted dLGN axons are able to find their way to the visual cortex via alternate routes at postnatal stages and reestablish a normal pattern of thalamocortical connectivity. These findings emphasize the importance and specificity of cortical cues in establishing thalamocortical connectivity and the spectacular capacity of the early postnatal cortex for remapping initial sensory representations.     

Patterning of the basal telencephalon and hypothalamus is essential for guidance of cortical projections

We have investigated the mechanisms that control the guidance of corticofugal projections as they extend along different subdivisions of the forebrain. To this aim, we analyzed the development of cortical projections in mice that lack Nkx2-1, a homeobox gene whose expression is restricted to two domains within the forebrain: the basal telencephalon and the hypothalamus. Molecular respecification of the basal telencephalon and hypothalamus in Nkx2-1-deficient mice causes a severe defect in the guidance of layer 5 cortical projections and ascending fibers of the cerebral peduncle. These axon tracts take an abnormal path when coursing through both the basal telencephalon and hypothalamus. By contrast, loss of Nkx2-1 function does not impair guidance of corticothalamic or thalamocortical axons. In vitro experiments demonstrate that the basal telencephalon and the hypothalamus contain an activity that repels the growth of cortical axons, suggesting that loss of this activity is the cause of the defects observed in Nkx2-1 mutants. Furthermore, analysis of the expression of candidate molecules in the basal telencephalon and hypothalamus of Nkx2-1 mutants suggests that Slit2 contributes to this activity.     

GAP-43 is critical for normal targeting of thalamocortical and corticothalamic, but not trigeminothalamic axons in the whisker barrel system

Mice lacking the growth-associated protein GAP-43 (KO) show disrupted cortical topography and no barrels. Whisker-related patterns of cells are normal in the KO brainstem trigeminal complex (BSTC), while the pattern in KO ventrobasal thalamus (VB) is somewhat compromised. To better understand the basis for VB and cortical abnormalities, we used small placements of DiI to trace axonal projections between BSTC, VB, and barrel cortex in wildtype (WT) and GAP-43 KO mice. The trigeminothalamic (TT) pathway consists of axons from cells in the Nucleus Prinicipalis that project to the contralateral VB thalamus. DiI-labeled KO TT axons crossed the midline from BSTC and projected to contralateral VB normally, consistent with normal BSTC cytoarchitecture. By contrast, the KO thalamocortical axons (TCA) projection was highly abnormal. KO TCAs showed delays of 1-2 days in initial ingrowth to cortex. Postnatally, KO TCAs showed multiple pathfinding errors near intermediate targets, and were abnormally fasciculated within the internal capsule (IC). Interestingly, most individually labeled KO TCAs terminated in deep layers instead of in layer IV as in WT. This misprojection is consistent with birthdating analysis in KO mice, which revealed that neurons normally destined for layer IV remain in deep cortical layers. Early outgrowth of KO corticofugal (CF) axons was similar for both genotypes. However, at P7 KO CF fibers remained bundled as they entered the IC, and exhibited few terminal branches in VB. Thus, the establishment of axonal projections between thalamus and cortex are disrupted in GAP-43 KO mice.     

Interdigitated paralemniscal and lemniscal pathways in the mouse barrel cortex

Primary sensory cortical areas receive information through multiple thalamic channels. In the rodent whisker system, lemniscal and paralemniscal thalamocortical projections, from the ventral posteromedial nucleus (VPM) and posterior medial nucleus (POm) respectively, carry distinct types of sensory information to cortex. Little is known about how these separate streams of activity are parsed and integrated within the neocortical microcircuit. We used quantitative laser scanning photostimulation to probe the organization of functional thalamocortical and ascending intracortical projections in the mouse barrel cortex. To map the thalamocortical projections, we recorded from neocortical excitatory neurons while stimulating VPM or POm. Neurons in layers (L)4, L5, and L6A received dense input from thalamus (L4, L5B, and L6A from VPM; and L5A from POm), whereas L2/3 neurons rarely received thalamic input. We further mapped the lemniscal and paralemniscal circuits from L4 and L5A to L2/3. Lemniscal L4 neurons targeted L3 within a column. Paralemniscal L5A neurons targeted a superficial band (thickness, 60 μm) of neurons immediately below L1, defining a functionally distinct L2 in the mouse barrel cortex. L2 neurons received input from lemniscal L3 cells and paralemniscal L5A cells spread over multiple columns. Our data indicate that lemniscal and paralemniscal information is segregated into interdigitated cortical layers.     

Robo1 regulates the development of major axon tracts and interneuron migration in the forebrain

The Slit genes encode secreted ligands that regulate axon branching, commissural axon pathfinding and neuronal migration. The principal identified receptor for Slit is Robo (Roundabout in Drosophila). To investigate Slit signalling in forebrain development, we generated Robo1 knockout mice by targeted deletion of exon 5 of the Robo1 gene. Homozygote knockout mice died at birth, but prenatally displayed major defects in axon pathfinding and cortical interneuron migration. Axon pathfinding defects included dysgenesis of the corpus callosum and hippocampal commissure, and abnormalities in corticothalamic and thalamocortical targeting. Slit2 and Slit1/2 double mutants display malformations in callosal development, and in corticothalamic and thalamocortical targeting, as well as optic tract defects. In these animals, corticothalamic axons form large fasciculated bundles that aberrantly cross the midline at the level of the hippocampal and anterior commissures, and more caudally at the medial preoptic area. Such phenotypes of corticothalamic targeting were not observed in Robo1 knockout mice but, instead, both corticothalamic and thalamocortical axons aberrantly arrived at their respective targets at least 1 day earlier than controls. By contrast, in Slit mutants, fewer thalamic axons actually arrive in the cortex during development. Finally, significantly more interneurons (up to twice as many at E12.5 and E15.5) migrated into the cortex of Robo1 knockout mice, particularly in both rostral and parietal regions, but not caudal cortex. These results indicate that Robo1 mutants have distinct phenotypes, some of which are different from those described in Slit mutants, suggesting that additional ligands, receptors or receptor partners are likely to be involved in Slit/Robo signalling.     

Postnatal development of thalamic reticular nucleus projections to the anterior thalamic nuclei in rats

The thalamic reticular nucleus (TRN) projects inhibitory signals to the thalamus, thereby controlling thalamocortical connections. Few studies have examined the development of TRN projections to the anterior thalamic nuclei with regard to axon course and the axon terminal distributions. In the present study, we used parvalbumin (PV) immunostaining to investigate inhibitory projections from the TRN to the thalamus in postnatal (P) 2- to 5-week-old rats (P14–35). The distribution of PV-positive (+) nerve fibers and nerve terminals markedly differed among the anterior thalamic nuclei at P14. Small, beaded nerve terminals were more distributed throughout the anterodorsal nucleus (AD) than in the anteroventral nucleus (AV) and anteromedial nucleus (AM). PV+ fibers traveling from the TRN to the AD were observed in the AV and AM. Nodular nerve terminals, spindle or en passant terminals, were identified on the axons passing through the AV and AM. At P21, axon bundles traveling without nodular terminals were observed, and nerve terminals were distributed throughout the AV and AM similar to the AD. At P28 and P35, the nerve terminals were evenly distributed throughout each nucleus. In addition, DiI tracer injections into the retrosplenial cortex revealed retrogradely-labeled projection neurons in the 3 nuclei at P14. At P14, the AD received abundant projections from the TRN and then projected to the retrosplenial cortex. The AV and AM seem to receive projections with distinct nodular nerve terminals from the TRN and project to the retrosplenial cortex. The projections from TRN to the AV and AM with nodular nerve terminals at P14 are probably developmental-period specific. In comparison, the TRN projections to the AD at P14 might be related to the development of spatial navigation as part of the head orientation system.Key words: Thalamic reticular nucleus, parvalbumin, axon terminal, development, anterior thalamic nucleus, rat