Developmental interactions between thalamus and cortex: a true love reciprocal story

Abstract

The developmental programs that control the specification of cortical and thalamic territories are maintained largely as independent processes. However, bulk of evidence demonstrates the requirement of the reciprocal interactions between cortical and thalamic neurons as key for the correct development of functional thalamocortical circuits. This reciprocal loop of connections is essential for sensory processing as well as for the execution of complex sensory-motor tasks. Here, we review recent advances in our understanding of how mutual collaborations between both brain regions define area patterning and cell differentiation in the thalamus and cortex.

Introduction

The cerebral cortex contains billions of neurons organized in delimited areas that process particular aspects of sensation, movement and cognition. The mechanisms that allow these neurons to integrate into specific functional circuits during development have received much attention in the last decade. Given that thalamic and cortical developmental programs concur in time and overlap in space (Figure 1), both structures influence each other during brain ontogeny. Hence, the incipient and long-lasting relationship between the cortex and the thalamus can be understood as a love story.

Figure 1. Development of corticothalamic and thalamocortical circuits.

Developmental time line of thalamocortical and corticothalamic circuits, notice that the upper part of the scheme represents the cortical development timeline (A) and the lower the thalamic one (B). The central core of the scheme is focused in summarizing the CTAs and TCAs navigation. Key developmental time-points and special features are highlighted in the scheme. (A) Cortical development of excitatory neurons takes place from E11.5 to E16.5, at the same time, GABAergic interneurons invade the cortical plate. (B) Thalamic glutamatergic neurons are born between E10.5 and E16.5 at the ventricular zone of the caudal pre-thalamus. GABAergic thalamic interneurons, which origin remains controversial, invade the thalamus during the first postnatal week.

In sensory systems, the thalamus receives information from peripheral organs which is reliably processed and transmitted to cortical territories. Corticothalamic feedback from layer 6 is sent back to the corresponding first-order thalamic sensory nuclei. This thalamo-cortico-thalamic loop is built at late prenatal-early postnatal stages, and has been demonstrated to be fundamental to regulate developmental processes, for instance, the specification of cortical and thalamic territories [1–6].

The interaction between thalamus and cortex has deserved much attention in the last four decades. It was in the late 80’s when two influential and opposing views emerged on how cortical areas are developmentally specified; called the Protomap and the Protocortex theories. The Protomap hypothesis postulated that cortical cells are already “pre-specified” at birth, being their area of generation key for the acquisition of identity features [7]. Indeed, neuronal precursors from different cortical areas show variations in cell-cycle kinetics, division mechanisms and cell fate specification [8–10]. Hence, areal differences are set by intrinsic developmental programs that unfold before thalamic fibers arrival. In accordance with this hypothesis, early cortical expression of guidance molecules and transcription factors guide thalamic axons to their final destination [11,12]. However, thalamic axons arrive to the cortex while corticogenesis is still ongoing, influencing the rate of proliferation across the germinal zone and, therefore, determining cytoarchitectural features [13].

On the opposite hand, the Protocortex hypothesis proposes that the cortex is originated as a tabula rasa where all neurons are born equal and multipotent. In this case, afferent axons (mostly thalamic input) impose cortical areal identity through activity-dependent mechanisms [13]. Various studies in primate visual system supported the Protocortex hypothesis [14,15]. Following this line, prenatal enucleation experiments in macaque showed that the depletion of thalamocortical axons (TCAs) during early corticogenesis changes the specification and area size of the visual cortex [16,17]. For several years, even decades, the Protomap and Protocortex hypotheses developed as independent and contradictory models, both supported by a separated amount of studies [18–20]. In the last few years, a similar discussion is applied to the thalamus, where the influence of intrinsic versus extrinsic programs (cortical and peripheral) is being disassembled. The astonishing advances in molecular biology, genetics, imaging, and electrophysiology have led to new evidences and opened novel avenues to understand cortical and thalamic development as processes where intrinsic and extrinsic mechanisms cooperate. The aim of this review is not to give an extensive recapitulation of what is known about the origin and principles of thalamocortical organization; instead, we stress out the present understanding of how thalamus and cortex influence each other’s development and function.

The first interactions: Development of the thalamocortical connectivity

Thalamocortical and corticothalamic axons (CTAs) must pursue a long journey passing through several territories before reaching their main targets, layer 4 cortical neurons and neurons of the thalamic nuclei, respectively (Figure 1). Every step of this trip needs to be tightly regulated to ensure a functional precise circuit. Thalamic neurons are generated from the alar plate of prosomere 2, between embryonic day (E)10.5 and E16.5 in mice [21,22]. Subsequently, thalamic axons are extended towards the prethalamus guided by prethalamic and ventral telencephalic axons [23–26]. Slit-Robo mediated repulsions avoid the entrance of TCAs into the hypothalamus, and force them to turn into the internal capsule, headed to the diencephalic-telencephalic border [27–30]. Once in the ventral telencephalon, corridor cells pave the way to traverse the mantle of the medial ganglionic eminence, based on ErbB4-Neuregulin1 interactions [31]. Also in this region, distinct TCAs get topographically sorted based on molecular interactions [32–37] regulated by spontaneous calcium activity [38,39] and other factors. TCAs must then cross the pallium-subpallium boundary to reach the cortex.

Cortical neurons located in the pallial ventricular zone generate progressively between E11.5 and E16.5 (Figure 1). The earliest generated neurons migrate radially to occupy the deepest layers in the cortical plate, while the ones born at later stages migrate through to form upper layers [40,41]. Early-born subplate neurons send their axons towards the thalamus around E12.5. According to the “handshake hypothesis”, subplate axons meet TCAs at the pallial subpallium boundary, serving them as a scaffold. This close interaction has been demonstrated to be crucial for the arrival of TCAs to the cortex [24,42,43]. In contrast, CTAs grow towards the internal capsule and the thalamic reticular nucleus, where they wait for a day before invading the thalamus around E17.5 [44,45]. At the time TCAs arrive to the cortical primordium, around E15.5 in mice, layer 2-4 neurons are not born yet [44,46], and thus accumulate at the subplate. During this waiting period, TCAs engage in activitydependent interactions with subplate cells, leading to their realignment before entering the cortical plate [47,48]. Shortly before birth, TCAs invade the cortical plate, form branches and synaptic contacts with layer 4 neurons principally. Cortical cells may influence this process by sending “stop” and “branch” signals [49]. Although cortical regionalization is initially created by graded expression of various cortical genes, TCA input can influence the size and identity of specific cortical areas [2–4,50].

Cortical influence over thalamocortical development

As aforementioned, the arrangement of cortical maps is initially controlled by intrinsic factors and particular genetic developmental programs [20,51]. These factors not only influence the positioning of sensory maps, but are also important for guiding TCAs towards their final destination. For instance, the fibroblast growth factor 8 (FGF8), an essential anterior telencephalic morphogene, promotes an early developmental cascade that specifies distinct cortical areas [52]. Duplications and areal shifts of sensory cortical maps are generated when FGF8 is expressed ectopically [5,53] (Figure 2). Therefore, FGF8 acts as an indirect regulator of TCA innervation by providing early positional information of cortical guidance cues. Similarly, other transcription factors such as Emx2 [54], Pax6 [55], Sp8 [56] and COUP-TF1 [6,56,57] are important to specify positional information in the cortex [20,58]. For example, Pax6 miss-expression in cortical progenitors, which preferentially determines sensorimotor areas, prompts to an aberrant and miniaturized body map representation in the principal somatosensory area (S1) [1]. This significant reduction of cortical territories engages a topdown plasticity process that generates an aberrant body representation also in the ventro-posteromedial nucleus (VPM) of the thalamus (Figure 2) [1]. Postmitotic COUP-TF1 expression is essential to repress frontal/motor area features in parietal and occipital cortex, thus controlling occipital-related genes such as Rorb [57]. When COUP-TF1 is eliminated specifically from the cortex, primary sensory areas suffer a caudal compression, leading to a shift of TCA connectivity [6] (Figure 2). All these evidences highlight the influence of cortical genetic patterns to the final position of TCAs.

Figure 2. Cortical influence over thalamic development.

Schemes representing the interaction between cortex and thalamus at early developmental stages. (A) In the wild type, a sensory representation for each modality (visual, somatosensory and auditory) is represented in both the cortex and thalamus. TCAs are topographically organized in a sensory modality fashion. (B) Modifications of sensory cortical areas promote thalamic rearrangements, either by TCA rewiring or by changing thalamic structures. Ectopic expression of FGF8 at the caudal pole of the cortex generates a duplication of S1, both S1 and S1’ are innervated by TCAs from the VPM nucleus. Reduction of Pax6 expression in the cortex generates a miniaturize S1 cortical area, engaging a reduction of the VPM nucleus. Ablation of COUP-TF1 expression in the cortex produces a shift of cortical areas towards de occipital pole of the cortex, TCAs rewire to properly connect with the shifted cortical territories.

POm, Posterior Medial nucleus; LP, Lateral Posterior nucleus; dLG, Dorsal Lateral Geniculate nucleus; MGv, ventral division of the Medial Geniculate body; VPM, Ventral Posterolateral nucleus; TCAs, Thalamocortical axons; S1, primary somatosensory cortex; S2, secondary somatosensory cortex; V1, primary visual cortex; V2, secondary visual cortex; A1, primary auditory cortex.

Additional cortical factors may also influence TCAs development. For instance, early neuronal activity patterns emerge during the development of the cortex, starting from early asynchronous activity to correlated firing, local burst, and ultimately, experience dependent modifications [59]. This neuronal activity regulates key developmental features, such as, axonal navigation and refinement of topographic maps [60,61]. Disruption of cortical neuronal activity provokes defects in the somatosensory map formation. Studies with specific cortical knock-outs for NMDAR1, AC1 and mGluR5 show deficiencies in barrel wall formation, disturbing neuronal reorganization. These cortico-specific animal models also show slight TCA defects, developing smaller and blurred barrels, while barrelloids in the thalamus remain intact [62–64].

Thalamic influence over cortical development: arealization and circuitry

As aforementioned, the cortex modulates distinct aspects of thalamic development and maturation, but it also receives an influence from the thalamus in a reciprocal fashion. Thalamic input impacts on several aspects of cortical development such as cortical proliferation, CTAs connectivity, cortical area specification, interneuron maturation and circuits assembly. However, bulk of evidence shows that several features of cortical development can proceed with little thalamic influence, at least embryonically. For instance, a shifted topography of thalamic connectivity during embryonic life only weakly affects cortical regionalization [65]. Moreover, mice lacking TCAs and corticofugal axons, and thus not receiving subcortical influence, develops a grossly normal cortex [66], further supporting the existence of intrinsic cortical mechanisms.

Anyhow, the influence of thalamic input into cortical developmental programs is doubtless. Early on, TCAs release a diffusible factor (related to bFGF signaling) that exerts a mitogenic effect on upper layer cortical progenitors, increasing the number of proliferative divisions and influencing the early developmental features of glutamatergic cortical neurons [67]. Therefore, TCAs might influence cortical arealization. In fact, recent studies have demonstrated that TCAs modulate the correct topographical navigation of their counterparts, the CTAs. Total abolishment of TCAs provokes CTAs to take an abnormal path, and that they acquire a corticospinal-like trajectory at the subpallium [68]. Thus, TCAs direct CTAs into the corridor region [31]. Moreover, TCAs are not only essential for guiding CTAs to their final targets, genetic manipulations of thalamic structures confirmed the importance of the thalamic input as an extrinsic factor, imperative for the correct development and establishment of the right genetic profile of the primary [2] versus high-order [3] cortical sensory areas (Figure 3). TCAs from first-order sensory nuclei provide specific information to instruct target cortical region. In the absence of these TCAs cortical areas will remain in the default mode and engage in a high-order fate [2–4].

Figure 3. Bottom-up plasticity. Thalamic influence over cortical arealization and circuitry.

Schemes representing the interaction between cortex and thalamus at early developmental stages. (A) In the wild type, a sensory representation for each modality (visual, somatosensory and auditory) is represented in both the cortex and thalamus. Primary and secondary areas are represented in both structures. (B) Ablation or reduction of first-order sensory nuclei of the thalamus engage a bottom-up plasticity, where corresponding primary cortical areas acquire properties of secondary areas. Cross-modal plastic changes are directed by embryonic thalamic activity before sensory onset. Disruption of spontaneous calcium activity in the auditory thalamic nucleus engages an increment of calcium waves in the VPM, producing an increment of the S1 size by augmenting TCA arborisation.

POm, Posterior Medial nucleus; LP, Lateral Posterior nucleus; dLG, Dorsal Lateral Geniculate nucleus; MGv, ventral division of the Medial Geniculate body; VPM, Ventral Posterolateral nucleus; S1, primary somatosensory cortex; S2, secondary somatosensory cortex; V1, primary visual cortex; V2, secondary visual cortex; A1, primary auditory cortex.

A correct topographical axonal organization is crucial to generate cortical maps, being a common feature in different sensory modalities and systems [69,70]. This organization is shaped by axon guidance molecules and axonal competition, that drive the correct axon pathfinding along their route. In the thalamocortical system, the precise point-to-point topographic organization of TCAs plays an important role in the appropriate development of cortical maps [12]. Disrupting the correct organization of TCAs along their pathway leads to a blurry somatosensory map, which lacks TCAs clustering [71]. This study showed for the first time the importance of the fine ordering of TCAs at the subpallium for the correct arrangement of the functional whisker pad representation in S1. Nevertheless, topographical miss-organization of TCAs subcortically could also affect the barrel map formation indirectly, by influencing the development of the thalamic barrelloids or by top-down mechanisms.

Moreover, thalamic input influences diverse aspects of cortical interneuron ontogeny [72]. Recently, several studies have pointed out the influence of thalamic activity-dependent mechanisms over the process of interneurons identity acquisition and function. For instance, thalamocortical input preferentially activates NR2B receptors expressed by Reelin interneuron population. Thus, manipulating thalamic input by the restricted expression of tetanus toxin in the thalamus prevents thalamic glutamate release, and influences the axonal and dendritic development of Reelin expressing cortical interneurons [73]. Moreover, other types of interneurons are also affected by the lack of TCA input, such as PV and SST neurons [74]. It has been shown that thalamic input innervates transiently somatostatin interneurons during the first postnatal week, before sensory onset. This early synaptic connectivity is important to modulate thalamic input to pyramidal and PV neurons in layer 4, contributing to the correct development of cortical circuits [73,75–77]. More than that, TCA input directly influences the correct development of layer 4 spiny stellate neurons. Blockade of thalamic presynaptic release provokes significant morphological changes in layer 4 cells that fail to develop the proper segregation and polarization needed to form the barrel walls around the TCA clusters [78,79]. In addition, the interference of thalamic activity triggers layer 4 spiny stellate neurons to develop an apical process, which is normally present in pyramidal neurons in non-granular layers [78].

Although the neuronal location is a process mainly controlled by intrinsic cortical mechanisms, the postmitotic identity of these cortical cells could be influenced by subcortical inputs. In the absence of a first-order thalamic nucleus, the corresponding primary cortical area acquires properties of associative secondary areas [2–4] (Figure 3). The ablation of the VPM not only causes layer 4 neurons of S1 to molecularly resemble of layer 4 neurons from secondary somatosensory areas (S2), but also, that presynaptic terminals from the high-order posteromedial thalamic nucleus (POm) aberrantly target layer 4 neurons in S1. This connection (POm-S1 layer 4) acquires the ability to respond to noxious stimuli, a function carried out by S2 layer 4 neurons in wild-type animals [4]. Altogether, these results strongly suggest that thalamocortical inputs exert a control of the molecular identity and function of postsynaptic layer 4 neurons and their upstream circuits.

The thalamus and the cortex are electrophysiologically interconnected from very early developmental stages. This interconnection is based on spatiotemporal synchronization of electrophysiological patterns from both structures. Early studies have shown that during the first postnatal week early gamma oscillations implement thalamocortical synchronization between a single thalamic barreloid and its equivalent cortical barrel [80]. More recently, it has been well demonstrated that this thalamic neuronal activity also influences the formation of sensory maps [81,82] and it has been recorded in vivo in the postnatal cortex [83]. Disruption of the presynaptic glutamatergic communication between somatosensory TCAs and the cortex disturbs, to some extent, barrel map formation. Ablation of NMDAR in the VPM perturbs profoundly the formation of barrelloids and consequently, the patterning of the posteromedial barrel subfield in S1 prevails blurred, meanwhile, barrellettes in the brainstem remain intact [84]. Manipulations of the presynaptic release in the TCA system also reproduce defects in barrel map formation: RIM1/RIM2 thalamic specific knock-out do not present defects in TC clusterization, although layer 4 spiny neurons acquire significant structural changes [85]. More severe phenotypes can be noticed when vesicular release is blocked from TCAs. Specific thalamic depletion of Vglut2 in a total Vglut1 knock-out background, generates a barrel less phenotype that lacks the somatosensory cortical map. Thus, a thalamic presynaptic vesicular release seems to be important for the cortical map formation [78], although the lack of the postsynaptic cortical Vglut1 may also contribute to the phenotype observed. Presynaptic AC1 is also critical for barrel formation, as thalamic knock-out presents disrupted barrels [86]. Nevertheless, it is not clear whether this effect is produced by a dysregulation of neurotransmitter release or acting over axon guidance molecules and spontaneous activity, as it has been already shown in the retinotectal projection [87].

It is clear that cortex and thalamus influence each other throughout development within the same sensory modality. But, do they also interact for triggering mechanisms of plasticity as those observed after sensory manipulations? Are cross-modal cortical modifications generated and controlled by subcortical input? The expansion of the posteriomedial barrel subfield in S1 has been described after postnatal visual deprivation [88,89], before sensory-experience onset [90]. A recent study locates this experienceindependent cross-modal expansion as a phenomenon that is led by the thalamus, where intercommunication among distinct sensory modalities takes place [50]. This intercommunication occurs by means of spontaneous calcium waves, which control thalamic gene expression and the arborisation of TCAs in sensory cortical areas (Figure 3) [50]. Genetic disruption of prenatal thalamic calcium waves leads to changes in the pattern of thalamic activity in the remaining nuclei. This novel mechanism serves to promote plasticity among developing sensory systems and places the thalamus, for the very first time, as a main input regulator of cross-modal cortical readjustments before sensory onset.

Future perspectives.

As highlighted in this review, the interaction between cortex and thalamus has always gained much attention, as the thalamocortical circuit is established from early embryonic stages at the time both structures are developing key features. The generation of tools to specifically target and modify single brain structures has been fundamental for advancing in our knowledge of the thalamo-cortical reciprocal modulation. However, further studies will be necessary to determine many aspects of this cortical and thalamic relationship. For instance, to what extent the acquisition of topographical and electrophysiological functional properties of cortical and thalamic neurons depends on their interaction. Moreover, peripheral input must also be included into the equation, as it has been shown that a correct thalamocortical circuit and interaction depends on it. Finally, the recent findings on the involvement of thalamic spontaneous activity in modulating the developmental aspects of cortical features constitute a fascinating topic for the coming years that will contribute in further understanding the role of neural activity in thalamocortical wiring.