Mesencephalic GABA neuronal development: no more on the other side of oblivion

Introduction

The midbrain – an important relay center for sensory inputs and motor outputs networking with the forebrain, hindbrain and spinal cord – is a hub of constant activity (1–10). Gamma-aminobutyric acid (GABA) neurons are key players in midbrain neuronal circuitry, robustly networking with glutamatergic and dopaminergic (DA) neurons to shape multiple aspects of behavior. Midbrain GABA neurons function not only as local inhibitory interneurons but also as projection neurons with targets in various brain regions. Along their developmental path, midbrain GABA neurons acquire molecular and functional diversity and can be divided into three categories based on anatomy and function: (1) dorsally located GABA neurons in the superior colliculus (SC) and inferior colliculus (IC); (2) medially located GABA neurons in the midbrain reticular formation (MRF) and periaqueductal gray (PAG); (3) ventrally located GABA neurons in substantia nigra (SN), ventral tegmental area (VTA) and retrorubral field (RRF), (Figure 1A).

Figure 1

(A) Schema of location of mature midbrain GABA neuronal populations in the superior colliculus (SC), inferior colliculus (IC), midbrain reticular formation (MRF), periaqueductal gray (PAG), substantia nigra (SN) and ventral tegmental area (VTA). Midbrain GABA neuronal populations regulate several brain functions, for instance, they are involved in the SC and IC for sensory integration, in the MRF for sleep, PAG for nociception and defensive behavior, SN for motor activity and in the VTA for motivated behavior. (B) Coronal schematic section of the embryonic mesencephalon depicting its patterning along the dorso-ventral axis into roof plate (RP), alar plate (AP), basal plate (BP) and floor plate (FP) on the left side along with dorso-ventral subdivisions on the right (m1–m7). Aq, aqueduct.

The functional diversity of midbrain GABA neurons based on their location is fascinating. Dorsal midbrain GABA neurons together with glutamatergic neurons are involved in processing and incorporation of sensorimotor, visual, auditory, and defensive behavior (11–16). Although GABA neurons are present in all of the seven layered SC, an abundance of GABA neurons is found in the superficial layers when compared to deep and intermediate layers (17–20). The SC is noteworthy for its high GABA content next only to the SN, globus pallidus and hypothalamus in the central nervous system (CNS) (21, 22). Medial midbrain GABA neurons of PAG regulate vocalization, endogenous control of pain sensation, fear, anxiety and aggression (23–25) while those in the MRF have been implicated with sleep-wake state control (26, 27). Ventral midbrain GABA neurons regulate DA neuron activity in the SN and VTA (6, 7, 28–31) and have projection targets similar to DA neurons, to the prefrontal cortex, basal ganglia and other limbic areas (6–10). Additionally, a considerable number of ascending projections from midbrain dopaminergic nuclei are GABAergic in nature (7). Ventral midbrain GABA neurons are thus critical for the function of dopaminergic pathways; have important functional roles in control of voluntary movements, emotion, mood, motivation, processing of appetite and aversive stimuli, addiction and reward behaviors (6, 29, 32–35) and may be involved in the etiology of several neurological and psychiatric disorders including schizophrenia, depression, mood disorders, addiction and Parkinson’s disease. Midbrain GABA neurons can therefore serve as important targets for treatment of neuropsychiatric disorders and for drugs of abuse. Recent evidence reveals that midbrain SN and VTA DA neurons co-release GABA although they do not synthesize it, by using GABA reuptake transporters (mGATs) to recycle extracellular GABA for release (36). Another study examining mechanisms of nicotine addiction shows that activation of midbrain GABA neurons in the VTA controls nicotine elicited burst firing of DA neurons and points to a concerted role for GABA neurons and DA neurons in mediating nicotine reinforcement (35). Thus, mechanisms adopted by DA neurons to locally regulate GABAergic transmission (36) or GABAergic regulation of DA neuron activity (35) illustrates the complexity of midbrain GABA-DA neuron interactions.

Despite the functional significance of GABA neurons in the adult midbrain, progress made in identifying the mechanisms underlying GABA neuron development has been relatively slow. Compared with the embryonic forebrain (37–40) and developing dorsal spinal cord (41–46) in which tremendous advances have been made in understanding GABA neuron development, in the embryonic midbrain, we are a decade behind. In the mesencephalon, DA neurons have been in the spotlight due to discovery of consequences of midbrain DA neurons degeneration in Parkinson’s disease (PD) and dopamine related neuropsychiatric disorders (47–54). Therefore the molecular mechanisms that define DA neuron development – generation, specification and differentiation of DA neurons have been extensively studied to generate cell replacement and pharmacological intervention strategies to alleviate some of the disease symptoms. However, proper migration of GABA neurons to their final location in the ventral mesencephalon during embryonic development seems to be dependent on the complete DA neuron architecture in the ventral mesencephalon, signifying important interactions between them for final location and connectivity (55). Thus the embryonic developmental period offers a favorable milieu when mesencephalic GABA-DA neuron interactions form and establish. Clearly it is a very exciting time to bring mesencephalic GABA neuron development for study onto a common platform with DA neuron development. In this mini-review we discuss current knowledge about origin of mesencephalic GABA neurons and molecular mechanisms involved in their specification, proliferation, differentiation and migration and emphasize missing links where more work is needed. Generation of detailed maps of key regulators of mesencephalic GABA neuron fate and function will initiate new studies to screen for their failure in mouse and human pathology and help develop new therapeutic strategies that focus on co-ordinate rescue of GABA-DA neuron interactions and novel pharmacological intervention paradigms.

Origin of mesencephalic GABA neurons

The embryonic mesencephalon is patterned along the dorso-ventral axis into roof plate (RP), alar plate (AP), basal plate (BP) and floor plate (FP) by BMP/Wnt signals from the RP and Shh signals from the FP/BP (56–59) and further divided into seven dorso-ventral subdivisions (m1–m7) with specific gene expression codes (60–62) to ensure cellular diversity (Figure 1B). Domains m1–m3 are structured into layers and domains m4–m7 are organized into distinct nuclei. For spatial patterning at dorso-ventral levels of the mesencephalon, several GABAergic progenitor domains have been identified, defined by expression of transcription factors such as Nkx6-1, Nkx2-2 and Pax3/7 (63). GAD expression, starting at E10.5 has been reported in BP and AP (64, 65). Both ventro-lateral and dorsal regions of the mesencephalon produce GABA neurons at E10.5–E14.5 (61, 66) and laminar organization develops during E13–E17 (67). The ventral mesencephalon at E13 is still completely devoid of GABA neurons, but by E17, GABA neurons are found robustly intermingled with DA neurons (55, 68).

GABA neurons originate from five of the seven mesencephalic progenitor domains (m1, m2, m3, dorsal half of m4 and m5). GABA neurons of the dorsal midbrain arise from the m1 and m2 domains. Domains m3–m5 may give rise to GABA neurons of the medial midbrain, MRF and PAG. GABA neurons in the ventral midbrain seem to have two developmental origins: 1) outside the midbrain and 2) from BP region. An early study has reported that all SN neurons in the rat brain originate from the midbrain-hindbrain border known as the isthmus (rostral most rhombomere 1) (69). More recent fate mapping studies have shown that midbrain VTA and SN pars reticulata (SNpr) GABA neurons originate from rhombomere 1 (r1), whereas the most anterior part of SNpr has a distinct origin outside the midbrain, possibly in the diencephalon (68). We have observed many GFP-positive GABA neurons in GAD65-GFP mesencephalon oriented ventrally in BP region at E13 and a stream of these neurons coursing from BP region to the ventral mesencephalon by E17. BrdU birthdating experiments additionally revealed that many E11-labeled neuronal progenitors migrated to contribute to GABA neurons of the ventral mesencephalon by E17 (55). The origin of these migrating cells remains to be better clarified.

Specification of GABAergic neuronal fate in the mesencephalon

Combinations of multiple transcription factors act as first selectors of neuronal fate (Figures 2 and 3) and contribute to neuronal diversity. In the mesencephalon, GABA neuronal fate determination is primarily associated with two basic helix-loop-helix (bHLH) genes: Helt (also known as Megane or Heslike) and Ascl1 (also known as Mash1) (61, 65, 70–72).

Figure 2

Overview of the molecular codes essential for generation of mesencephalic GABA neurons in relation to their specific position in the ventricular zone (VZ), intermediate zone (IZ) and mantle zone (MZ).

Beginning with inductive signals, a specific cascade of molecular instruction follows in the VZ, IZ and MZ to ensure GABAergic fate specification, neurogenesis, fate maintenance and differentiation into mature GABA neuronal populations.

Figure 3

Summary of the signaling pathway of key molecular players identified up until now and their interplay that is essential for efficient generation of mesencephalic GABA neurons.

Helt is co-expressed with Ascl1 in the ventricular zone of the alar-basal mesencephalic boundary from E9.5 onwards. Subsequently the Helt- Ascl1 co-expression domain expands dorsally as well. Gain of function studies in which transgenic mice were generated mis-expressing Helt from the nestin promoter-enhancer resulted in ectopic GABA neurons in the mantle layer outside the Helt/Ascl1⁺ zone (70). As the effect of mis-expression of Helt was specific to the mesencephalon and no ectopic cells were observed in other brain regions (dorsal telencephalon and thalamus) which do not express Ascl1, Helt was initially believed to specify the GABA neuronal fate only when Ascl1 was co-expressed (70). Interestingly, although ectopic Helt expression induced GABA neurons, it also suppressed the generation of glutamatergic neurons (61). Loss of function studies revealed that in homozygous Helt-null mice GABAergic progenitors were generated (61, 71) but they failed to become GABA neurons specifically in the dorsal mesencephalon. Ventral GABA neurons formed only from the m5 domain in Helt-null mesencephalon whereas in the dorsal domains there was a complete loss of GABA neurons and instead, glutamatergic neurons, induced by neurogenin (Ngn) genes emerged (61). Helt is now recognized as a key determinant of the GABAergic fate in dorsal mesencephalon by direct repression of Ngn genes (Ngn1 and Ngn2) in GABAergic progenitors, and induction of Lim1/2 and Gata2 (60, 61). Helt is required for Gata2 expression in the embryonic mesencephalon except in the ventro-lateral m5 domain (60). Loss of function of Helt did not affect dopaminergic and cholinergic neurons. Thus the essential role of Helt for regulating GABA neuron fate specification in the dorsal midbrain and for development of the SC was confirmed.

Ascl1, expressed by neural progenitors in all mesencephalic progenitor domains is another GABAergic fate determinant that is not regulated by Helt or other proneural genes such as Ngns. Ascl1 was unaffected in the Helt-null mice and may have also compensated for loss of Helt in ventral but not dorsal mesencephalon (70). In Ascl1-null mice, virtually no GABA neurons formed in the mesencephalon up until E11.5 while Helt continued to be expressed and other neurons were generated. From E13.5 onwards, GABA neurons were produced in the ventromedial mesencephalon (m3–m5), but not dorsally (m1–m2). By E18.5, GABA neurons were completely lost in Ascl1-null dorsal mesencephalon including SC, IC and dorsal PAG. Although GABA neurons were reduced in medial mesencephalon including ventral PAG and MRF, GABA neurons of the ventral mesencephalon in SN and VTA were not affected and do not seem to require Ascl1 (72).

Thus ventral mesencephalic GABA neurons represent an intriguing population of GABA neurons that develop independently of regulation by Helt and Ascl1.

Mesencephalic GABA neurogenesis

GABAergic neurogenesis is completed first in the MRF by E12.5, next in the SNpr by E13.5, followed by SC at E14.5 (68) and has been associated mostly with transcription factor combinations Helt and Ascl1 (Figures 2 and 3). The Helt-Ascl1 co-expression domain in the mesencephalon is co-ordinate with GABA neurogenesis, decreasing rapidly as neuronal differentiation proceeds (70). However, Ascl1’s requirement for neurogenesis seems to be highly region-specific. In Ascl1-null mesencephalon, a loss of neurogenic and neural stem cell specific expression with markers Delta1 (Notch ligand) and Hes5 (Notch target) was observed at early embryonic stages. This delay at the start of neurogenesis is believed to cause reduced GABA neuron numbers in the ventromedial midbrain of Ascl1-null mice (72). Ascl1 has been reported to promote GABAergic neurogenesis in in vitro cultured mesencephalic neural progenitors (73). Co-expression of both Ascl1 and Helt seems to significantly promote GABA neurogenesis in neural precursor cell cultures (70).

Mesencephalic GABA neuron differentiation and fate maintenance

After the cell cycle exit, the GABA neuron precursors activate genes typical for functional GABA neuron precursors. This role is assigned to C4 zinc-finger transcription factor Gata2, bHLH transcription factors Tal1/2, paired-like homeodomain transcription factor Pitx2 and homeobox transcription factor Lhx1 (Lim1) (Figures 2 and 3). Gata2 is expressed in m1–m5 domains in the embryonic mesencephalon and expression is activated in GABAergic progenitors as they exit the cell cycle, become postmitotic precursor cells and begin to differentiate (60). Selective loss of Gata2 in the mouse midbrain and rhombomere 1 (Gata2^cko mutant) resulted in a specific loss of GABA neuron precursors at early embryonic stages and transformation from GABAergic to glutamatergic phenotype in all mesencephalic GABA neuron subpopulations in the Gata2^cko mutant during embryonic development except for the GABA neurons associated with DA neurons in the SNpr and VTA that were unaffected (60, 63). Tal1/2 are expressed in m1–m4 dorsal half and m5 mesencephalic domains and co-operatively activate genes necessary for GABA neuronal differentiation (74). Tal2 is co-expressed with Gata2 in the ventricular zone (VZ) and intermediate zone (IZ) and regulates selection of GABAergic over glutamatergic neuronal fate. An ectopic upregulation of glutamatergic gene expression was observed in Tal2^cko mutants in m3 and m4 dorsal domains (74). Similar to Gata2^cko mutant, in Tal2^cko mutant, GABAergic markers were completely lost in the dorsal mesencephalon including SC. Again, both Gata2 and Tal2 are dispensable for ventral mesencephalic GABA neurons (74).

Pitx2, is expressed in m1–m4 and m6 domains with highest expression in the intermediate layer of the SC (75) and works downstream of Gata2 in the transcription factor cascade (60, 75). Loss of Pitx2 disrupts GABAergic neuronal differentiation and axonal outgrowth in the dorsal mesencephalon, specifically SC (76). In the ventral mesencephalon Pitx2 lineage neurons are mostly glutamatergic (75). Lhx1 (or Lim1) is expressed in postmitotic precursors of all mesencephalic GABA neurons in domains m1–m5 (60) and is an important marker of GABAergic differentiation.

Migration of mesencephalic GABA neurons

Neuronal migration is a key event during brain development as neurons and/or neuronal progenitors originating in VZs navigate along diverse courses to eventually find their destination and integrate into specific brain circuits. In the pallial telencephalon, pyramidal neurons follow radial migratory routes using radial glial guides to the cortical plate (40, 77). GABA neurons of the subpallial telencephalon, conversely, take long tangential paths into the cortex (78, 79) along vascular guides (80) to become intermingled with excitatory neurons. Unlike in the embryonic telencephalon where neuronal migration has been well elucidated, this vital event has been little explored in the embryonic mesencephalon. In the dorsal mesencephalon, majority of GABA neurons have been reported to migrate radially from the dorsal VZ to their final location in superficial layers through deep and intermediate layers in an inside-out manner (81) followed by slight tangential dispersion within the superficial layer (Figure 4A). Few cells here (81) show direct tangential migration from VZ to the superficial layer of the SC (Figure 4A). Three types of inhibitory neurons – stellate cells, pyriform cells and horizontal cells with tangentially oriented dendrites – thus come to reside in the multilayered SC (20).

Figure 4

(A) Coronal view of embryonic mesencephalon showing different patterns of neuronal migration. Black arrows depict radial migration of GABA neurons from the VZ to their final location. Green arrows indicate slight tangential dispersion within the superficial layer of the SC. Red arrow depicts that few cells show direct tangential migration from VZ to the superficial layer of the SC. Blue arrows highlight perpendicular migration of GABA neurons from basal plate region to ventral mesencephalon. (B) Sagittal view of embryonic mesencephalon depicting migration of postmitotic GABA neuronal precursors that cross the specific mesencephalon-rhombomere1 (r1) boundary to enter from r1 into the ventral mesencephalon (vm). The mode of migration of these neurons with origins outside the mesencephalon is currently unknown. tel, Telencephalon; mes, Mesencephalon; SC, superior colliculus.

In the ventral mesencephalon, GABA neuronal migration seems to be more complex with contributions from both outside and within. Recent genetic fate mapping studies have uncovered a separate origin of VTA and SNpr GABA neurons outside the mesencephalon (68). The findings unveil a new migratory pathway of GABA neurons originating in r1 compartment and crossing the r1-mesencephlon boundary to migrate to the VTA and SNpr as postmitotic neuronal precursors at E14.5–E15.5 (Figure 4B). Guidance mechanisms for this form of migration remain to be elucidated. The anterior SNpr GABA neurons in the diencephalon however do not arrive from the r1 region; they are speculated to come from the diencephalon although contribution from the mesencephalon is not excluded (68). These results were further supported by analysis of Tal1 mutant mice in which GABAergic precursors are normal in the mesencephalon, but failure in GABA neuron production in r1 correlated with loss of mature VTA and SNpr GABA neurons (68). Additionally, as Gata2 regulates GABA neuronal differentiation in the mesencephalon but not r1, in Gata2 mutants, although mesencephalic precursors failed to activate GABAergic neuron-specific gene expression, the VTA and SNpr GABA neurons appeared largely unaffected (60), supporting their outside origin and border migration across r1-mesencephalon compartments.

Our studies have shown that both DA and GABA neurons occupy ventral mesencephalon in a temporally sequential manner during embryonic development and depicted GABA neuronal silhouette oriented from BP region to ventral mesencephalon in GAD65-GFP embryos (55). It emphasized the importance of perpendicular migration, a novel mode of neuronal migration that seems to be unique to the mesencephalon (Figure 4A) and essential for proper set up of the anatomical architecture of ventral mesencephalic structures (55). BrdU birthdating experiments revealed that E10 and E11 labeled neuronal progenitors migrated ventrally (from FP) as well as perpendicular (from BP) to the aqueduct to form the distinct anatomical architecture of the boat shaped SN and VTA by E17 (55). Pitx3 represents a critical regulator of mesencephalic DA neuron development (49, 51, 82–85) and there is selective and early loss of A9 DA neurons in the SN of ak/ak mice (49, 84–86). The cells affected in the ak/ak mice are known to be very sensitive to neurotoxins such as 6-OHDA and MPTP (54, 87) and are the same cells that degenerate first in Parkinson’s disease (PD) patients (83–85). Our findings have revealed that loss of SN DA neurons in ak/ak mesencephalon is a result of defective perpendicular migration that resulted in cells which were stuck or trailing in the middle of their migratory trajectory and distributed abnormally in the red nucleus area. Interestingly, coordinate with this loss of DA neurons there was also a significant loss of ventral mesencephalic GABA neurons that persisted in the adult ak/ak midbrain (55). Pre-existing DA neurons seem to modulate the migration of BP GABA neurons to ventral mesencephalon along perpendicular migration routes suggestive of important interactions between both neuronal populations for final location and connectivity and the earliest signs of inter-dependence that arises during embryonic development (55). Identification of the anatomical/cellular substrate for perpendicular neuronal migration will help model future investigations to induce this intriguing population of GABA neurons to migrate successfully in cell transplantation experiments coordinate with DA neurons.

At a mechanistic level, neuronal migration overall in the mesencephalon suffers greatly from lack of insights. Mesencephalic axon guidance and/or ventral migration of DA neurons has been linked only to a small number of molecules [netrin 1 (88) and its receptor DCC (89), reelin (90), proteoglycan phosphacan 6B4 (91), L1CAM (92), neural cell adhesion molecule (93) and polysialic acid (93)]. Transcription factors, primary mechanisms controlling neuronal migration, remain unidentified in the embryonic mesencephalon. For perpendicular migration to occur from the BP to the ventral mesencephalon, neurons need to perfectly synchronize multiple actions and respond appropriately to guidance cues encountered during their trajectory. Transcriptional regulation is inevitably a key determinant of this process. In the vertebrate spinal cord, for instance, expression of specific combinations of transcription factors in postmitotic motorneurons encodes axon trajectories (94–96) and similar transcription factors carry on different functions depending on the cellular context (97–99). Another example is that of postmitotic Nkx2.1 expression, which has to be down-regulated for interneurons to migrate to the cortex, whereas Nkx2.1 expression is required for interneuron migration to the striatum (100). Similar possibilities lie in the embryonic mesencephalon and much work lies ahead with respect to elucidating mechanisms of neuronal migration.

Conclusion and perspectives

The generation of several mouse models with abnormal development and function of cortical GABA interneurons, which recapitulated defective behavior similar to those seen in conditions like schizophrenia, autism, epilepsy, mood and anxiety disorders, was a major step that associated forebrain GABA neurons with the pathobiology of several neuropsychiatric illnesses (101–107). Thus, the identification of the molecular components involved in forebrain GABA neuron development in both mouse and human triggered efficient generation of GABA neuron populations based on ES cell engineering. ES-derived GABA interneurons today have remarkable potential – they are functional, survive well, migrate and integrate into desired regions in both embryonic and adult brains post-transplantation and are attractive candidates for cell-based therapies (108–112). In the midbrain, ES cell technology today focuses selectively to generate DA neurons (113–116). This is possible because the molecular cascade of DA neuron development has been well studied. A missing link in the midbrain field is our current incapability for efficient generation of GABA neurons from ES cells, as the molecular mechanisms underlying mesencephalic GABA neuron fate and function are just beginning to be elucidated and furthermore the transcriptional machinery identified here, except for Ascl1 is different from that of the forebrain.

The quest to unravel the transcriptional cascade in the embryonic midbrain is however no easy task. It is particularly daunting given the intricate diversity of GABA neurons found in the mesencephalon based on location and function and transcriptional factor networks, with overlapping function for GABAergic vs. glutamatergic fate, or GABAergic vs. serotonergic/dopaminergic fate depending on their domain. Table 1 summarizes expression stages, domains of expression and function of key molecular players regulating mesencephalic GABA neuron development. As most of these candidates are broadly expressed in several brain regions as well as other organs, systemic knockouts die prenatally or at birth (Table 1). There is a great need for gain and loss of function assays, including generation of more mouse mutants with deletion of GABA neuron-specific genes selectively in the embryonic mesencephalon and subsequent analysis of brain development and postnatal behavior to be able to fully understand the developmental significance of mesencephalic GABA neurons in the CNS. Helt, Gata2 and Tal1/2 are transcription factors that have been best studied in this regard (Table 1). Given Helt’s unique expression pattern in the mesencephalon and prosomere1-2 in the diencephalon, isolation of Helt served as a vital key to unlock the mysteries of mesencephalic GABA neuronal development. Helt deletion resulted in loss of weight and postnatal lethality in mice around 3–5 weeks of age with behavioral defects such as fore/hindlimb clenching and seizure-like phenotype from P14 onward (71). Elegant work demonstrating conditional inactivation of Gata2 and Tal1/2 in mouse midbrain and r1 have provided several new insights into the specific role of these transcription factors for GABA neuron development in the embryonic midbrain, although postnatal phenotypes have not yet been reported (60, 74).

The VTA and SNpr GABA neurons are particularly distinct and interesting because they develop independently of the known regulators of midbrain GABAergic fate specification and maintenance: Helt, Ascl1, Gata2 and Tal2 (60, 61, 65, 71, 72, 74). Given the functional importance of GABA-DA neuron associations in the adult midbrain, it has become critical to decode transcriptional regulation of ventral mesencephalic GABA neurons as well as delineate development of GABA-DA neuron interactions and connectivity. Additionally, as GABA neurons were also affected in addition to DA neurons as a result of loss of Pitx3 (55), a crucial transcriptional regulator of DA neuron development, there lies the possibility of overlapping molecular pathways and shared mechanisms mediating GABA-DA neuron development in the ventral mesencephalon. Many vital goals remain to be achieved before we can explore and combine the distinctive features of both GABA neurons and DA neurons so as to understand complex neurological and psychiatric illnesses and to test novel therapies that can ultimately become a clinical reality. Clearly, the field of mesencephalic GABA neuron development is poised for several exciting discoveries in the future.