Abstract
The striatum controls behaviors via the activity of direct and indirect pathway projection neurons (dSPN and iSPN) that are intermingled in all compartments. While such mosaic ensures the balanced activity of the two pathways, how it emerges remains largely unknown. Here, we show that both SPN populations are specified embryonically and progressively intermix through multidirectional iSPN migration. Using conditional mutants of the dSPN-specific transcription factor Ebf1, we found that inactivating this gene impaired selective dSPN properties, including axon pathfinding, whereas molecular and functional features of iSPN were preserved. Remarkably, Ebf1 mutation disrupted iSPN/dSPN intermixing, resulting in an uneven distribution. Such architectural defect was selective of the matrix compartment, revealing that intermixing is a parallel process to compartment formation. Our study reveals that, while iSPN/dSPN specification is largely independent, their intermingling emerges from an active migration of iSPN, thereby providing a novel framework for the building of striatal architecture.
Introduction
The dorsal striatum controls major brain functions, such as motor behaviors and habit formation, through the coordinated activation of descending direct and indirect pathways. Consistently, loss or damage of striatal projection neurons (SPN) is associated with a spectrum of pathologies that include Parkinson’s disease, Huntington’s disease and obsessive-compulsive disorders. SPN are medium-spiny GABAergic neurons which account for 90-95% of striatal neurons and are divided into two subtypes: i) direct SPN (dSPN) project to the substantia nigra (SN) and express the D1 dopamine receptor and the neuropeptide Substance P; (ii) indirect SPN (iSPN) send axons to the globus pallidus (GP) and express D2 dopamine receptor and the opioid peptide Enkephalin. Functionally, it is proposed that dSPN activation promotes action selection and positive reinforcement, whereas iSPN are important to suppress unwanted motor sequences1–7. dSPN and iSPN are completely intermixed within the striatum, thereby forming a relatively uniform mosaic8–10. Building on this relatively simple cellular organization, the dorsal striatum is split into at least two major compartments, cellular islands known as striosomes and a surrounding matrix, which form immunohistologically and functionally distinct modules differing by their input and output11–17. In such precise three-dimensional organization, the intermingling of dSPN and iSPN ensures the balanced activation of direct and indirect pathways.
While functional studies revealed distinct roles for dSPN/iSPN and striosomes/matrix, much less is known on how they develop. SPN are generated in the embryonic lateral ganglionic eminence (LGE) and migrate radially to form the striatum18–24. Both SPN subtypes are present in the entire striatum, except for a restricted caudal domain8, and several transcription factors have been involved in their generation and differentiation including Ctip2, FoxP1 and FoxP2 25–40. In addition, distinct transcriptomic programs have been involved in the specification of either dSPN or iSPN, including Islet1 and Ebf1 for dSPN or Sp9 and Six3 for iSPN23, 41–46. In particular, the conditional inactivation of Islet1 impairs the differentiation of early-born dSPN41, 45 and the full inactivation of Ebf1 leads to a defective development of late-born dSPN, disorganized projections to the SN and altered survival of matrix dSPN at postnatal stages42–44. Regardless of their subtype, early progenitors give rise to striosome SPNs whereas later-derived progenitors generate matrix SPN, which form the two compartments potentially via migration and selective cell-sorting17, 24, 47–51. In contrast, how dSPN and iSPN are specified and integrated into striatal circuits remains largely to be characterized.
Here we show that formation of the striatal mosaic relies on the early specification of SPN subtypes combined with a dSPN-dependent tangential migration of iSPN. We found that dSPN and iSPN form two molecularly-defined populations throughout development, which only progressively intersperse. Using time-lapse imaging, we observed that iSPN undergo multidirectional migration, thereby intermixing with dSPN. Furthermore, we took advantage of a unique combination of genetic tools, including embryonic in vivo fate map and conditional knockout models, to examine the role of Ebf1 in the differentiation of dSPN and striatal mosaic formation. We found that Ebf1 deletion in dSPN (i) perturbed specific aspects of direct neurons differentiation, leading to defective axogenesis and integration in cortico-striatal circuits, without affecting iSPN cardinal properties (ii) impaired intermixing of dSPN and iSPN in the matrix. These results establish Ebf1 as a master regulator of dSPN connectivity and intermixing with iSPN in the matrix compartment. Furthermore, our study reveals that the active intermingling of early-specified dSPN and iSPN is required to assemble the striatal mosaic.
Results
Defining a molecular fingerprint of embryonic dorsal dSPN
To investigate how dSPN and iSPN are specified, positioned and assembled in the striatum, we searched for specific markers of these two neuronal populations during embryogenesis. To this aim, we took advantage of the BAC transgenic mouse line Drd2-EGFP52, which has been widely used to label iSPN in neonates and is expressed embryonically53, 54. In addition, we crossed R26mt/mt reporter mice55 with Islet1Cre transgenic mice, which mediate Cre-recombination in dSPN from embryonic stages41, 45. We first confirmed that these two distinct transgenic lines respectively labeled iSPN and dSPN at postnatal day (P) 5, when the two populations can be unambiguously defined53, 56. In Islet1Cre;R26mt/+;Drd2-EGFP mice, we found that tdTomato-positive (tom+) cells constituted approximately half of all Ctip2+ SPN25 in both striosome and matrix compartments (as defined in Fig. S1) and were largely unlabeled by Drd2-EGFP (Figs. 1a-d). Using these two mouse lines, we found that iSPN and dSPN lineages are non-overlapping throughout embryogenesis, consistently with previous studies focusing on SPN subtypes specification 23, 41, 43–46. We furthermore identified high expression of transcription factors FoxP2 and Ebf1 as specific of dSPN at E13.5 and E17.5 (Figs. 1e-h). These results identify a dSPN developmental fingerprint which, combined to the iSPN-specific Drd2-EGFP expression, delineates two segregated populations of Ctip2+/Foxp1+ SPN throughout development (Fig. 1i). Our findings thus indicate that dSPN and iSPN exhibit distinct molecular identities from the earliest steps of striatogenesis and provides us with tools to follow their integration in the striatal architecture.
iSPN progressively intermix with dSPN via intrastriatal migration
We took advantage of the early molecular fingerprints to monitor the dynamics of SPN distribution and thereby extend our comprehension of how the two populations are specified and assembled (Fig. 2). Previous studies showed that all SPN are generated in the LGE over an extended period of time and migrate radially to form the striatum18–22, 24. Moreover, birthdating experiments indicate that a majority of earliest-born SPN are dSPN24, 48. Consistently, we found that the striatum at E12.5 contained mostly dSPN neurons and few scattered GFP+ iSPN (Figs. 2a-a”’). GFP+ iSPN were detected in larger numbers at E13.5 and E14.5, albeit concentrated in the lateral striatum (Figs. 2b-c”), which contains early-born neurons that will mostly contribute to striosomes20, 47. From E15.5 onwards, we detected a progressively increasing number of GFP+ iSPN in the initially dSPN-dense part of the medial striatum (Figs. 2d-e”). Thus iSPN generated in the LGE proliferative zones intersperse into dSPN-dense territories, first in the lateral striatum and then medially, in a gradual process that spans several days (Fig. 2f). This phenomenon was temporally and spatially overlapping with the formation of striosome and matrix compartments20, 21, 24, 49.
Such progressive mosaic formation could be explained by: i) delayed expression of Drd2-EGFP compared to dSPN markers; ii) tangential dispersion of iSPN in the striatal mantle. To discriminate between these possibilities, we first examined whether immature iSPN in the LGE subventricular zone (SVZ) expressed Drd2-EGFP and whether timed EdU injections showed preferential labelling of either iSPN or dSPN. We found that immature iSPN already expressed Drd2-EGFP in the SVZ (Figs. S2a-b’) and that these neurons were produced continuously from E11 onwards (Figs. S2c-e), consistently with previous studies24, 54. iSPN however constitute a minority of the early-born SPN, as revealed by EdU staining (Figs. S2c-e) as well as labeling with the dSPN-specific Foxp2 immunostaining and Islet1Cre driven recombination (Figs. S2f-h). Taken together, our results indicate that Drd2-EGFP labels iSPN as they differentiate and enter the striatum primordium.
It is thus unlikely that mosaic formation is solely due to either a delay in Drd2-EGFP expression or a sequential production of SPN subtypes. To examine whether iSPN disperse tangentially inside the striatum, we performed two-photon time-lapse imaging on Drd2-EGFP embryonic slices57. We found that GFP+ iSPN show multidirectional saltatory migration within the striatal mantle at E15.5 (Figs. 3a-e; Movie S1). By performing cell behavior analyses over several time-lapse acquisitions, we found that iSPN have a global displacement away from the SVZ but present multidirectional trajectories inside the striatum (Figs. 3f-i and S3). In particular, iSPN changed their migration direction over time (Movie S1). Importantly, we confirmed that Drd2-GFP+ cells in slices are Ctip2+ SPN and that the processes of these neurons also present multiple orientations in fixed embryonic tissue (Fig. S3). Thus, in contrast to the assumption that SPN only migrate radially, iSPN undergo a tangential migration within the striatum and intermix with dSPN.
Taken together, our work shows that dSPN and iSPN form segregated embryonic populations that intermix over time, first in the lateral early-born proto-striosomes and then in the medial region. This process occurs at least in part via a multidirectional migration of iSPN thereby raising questions about the underlying mechanisms and possible interactions between the two populations.
Ebf1 conditional deletion affects specific aspects of dSPN differentiation
To address these issues, we searched for mouse models selectively perturbing the development of one of the SPN subtypes. Transcription factors specifically involved in either dSPN or iSPN differentiation include Islet1 and Ebf1 or Sp9 and Six3, respectively23, 41–46. In contrast to Islet1, Six3 or Sp9 deletions, which induce cell death in the targeted population23, 41, 45, 46, Ebf1 inactivation has been previously proposed to impair dSPN wiring in the striatal matrix42–44. We thus generated and compared two different conditional Ebf1 mutants (cKO), using the dSPN-restricted Islet1Cre line and the Dlx5/6::Cre. Indeed, Islet1Cre drives recombination in dSPN and cholinergic interneurons47, 51, 41, 45, 58–60 and Dlx5/6::Cre recombines in all SPN and all interneurons61–65. Importantly, Ebf1 is not expressed in striatal interneurons (Figs. S1a-c) and the two cre lines are nonoverlapping outside of the ventral telencephalon58, 60, 61, 65–68. Thus comparing the phenotypes shared in these two conditional mutants (cKO) allowed us to determine the deficits due to Ebf1 inactivation in dSPN. As expected, Ebf1 was absent from the entire striatal mantle at E17.5 in both cKOs (Figs. 4b-d) and striatal size was slightly reduced, as in Ebf1−/− mice42. However, all the generic early SPN markers we examined, including Ctip2 (Figs. 4e-g), Foxp1, gad1 (not shown), DARPP-32 (Figs. 8j-l) indicated that the striatum was still formed by GABAergic neurons with a SPN-like identity. In order to compare the phenotypes in the two cKOs and gain insights in the molecular programs controlled by Ebf1, we performed RNA-sequencing (RNA-seq) in striatal tissue of control, Islet1Cre;Ebf1fl/− and Dlx5/6::Cre;Ebf1fl/− embryos at E17.5 (Figs. 4h-k and Tables S1, S2). Importantly, most genes were commonly deregulated in the two cKOs (Figs. 4h-k and Tables S1, S2) suggesting that inactivation of Ebf1 in dSPN or all SPN lead to similar phenotypes. To compare the datasets, we examined genes that were previously identified as representative of the core SPN generic identity, versus specific of dSPN or iSPN identity41, 43, 45, 46, 69. We found that genes associated with a generic SPN identity or with iSPN identity were not significantly altered in both mutants (Figs. 4h-i and Tables S1, S2). In contrast, the expression of genes associated with dSPN were either preserved (Drd1, Foxp2, Islet1, Pdyn), or severely deregulated in both mutant models (Figs. 4j-k and Tables S1, S2). The preserved expression of Foxp2 and Islet1 shows that these transcriptional regulators must act in a partially distinct pathway. Deregulated genes encoded selective transcriptional regulators such as Zfp521, Mef2c31, 44, 70–72, transmembrane receptors such as PlexinD1, which is involved in synaptogenesis27, and intracellular effectors such as Slc35d3, which regulates D1 receptor trafficking73. These results were validated by in situ hybridization on selected key genes (Fig. S4). Finally, Gene Ontology analyses revealed specific alterations of factors regulating axon development, cell-cell adhesion properties and synaptogenesis in both cKOs (Figs. S4m-o). Taken together, our findings indicate that, while dispensable for the acquisition of a generic SPN transcriptomic profile, Ebf1 regulates the expression of genes involved in selective aspects of dSPN differentiation.
Ebf1 conditional deletion selectively impairs dSPN differentiation
Since our transcriptomic analysis revealed that Ebf1 controls the prenatal expression of major regulators of axon development and synaptogenesis (Figs. 4 and S4), we examined striatonigral and striatopallidal projections in both cKOs (Fig. 5). In P5 controls, DARPP-32+ labels SPN axons including the ones of dSPN which crossed the globus pallidus (GP) and the entopeduncular nucleus (EP), joined the cerebral peduncle (CP) and reached the substantia nigra (SN) (Figs. 5a). In both cKOs, DARPP-32+ projections extended to the EP but only a small fraction of them reached the SN (Figs. 5a-c), with axonal density in the CP showing an approximatively 60% reduction compared to controls (Figs. 5d-g). In contrast, when we examined striatopallidal projections by measuring the density of Enkephalin signal in the GP we found no major differences between cKOs and controls (Figs. 5h-k). Thus, Ebf1 inactivation selectively impaired the capacity of dSPN to form the direct pathway.
In contrast to Ebf1−/−42–44, cKOs survive up to adult age, allowing us to examine the long-term impact on SPN properties and pathway formation. We observed in cKOs that dtTomato-labeled dSPN were reduced in number and size at P45 in cKO mice (Figs. 6a-g), indicating possible functional impairment. To further analyze the impact of Ebf1 inactivation on dSPN connectivity, we recorded glutamatergic miniature excitatory postsynaptic currents (mEPSCs) in the wholecell configuration from adult control and cKO mice. We used either dtTomato or Drd2-EGFP expression as selective SPN subtype markers (Figs. 6h-j). Moreover, neurons were filled with biocytin during recording for later morphology reconstructions. Consistent with dtTomato labelling, reconstruction of mutant biocytin-filled neurons showed overall reduced dSPN morphology, in contrast to iSPN (Figs. 6k-m). In addition, in both cKOs, dSPN mEPSC frequency was severely reduced, with most cells presenting no detectable spontaneous synaptic activity. In contrast, iSPN did not show significant differences from controls (Figs. 6g-l and Fig. S5a-f). We further examined the functional impact on direct and indirect pathways using a pharmacological approach in vivo, by testing motor responses to injections of either D1 or D2 receptor agonists41, 74. In baseline conditions, we found that neither cKOs showed major locomotion deficits, Dlx5/6::Cre; Ebf1fl/− mutants exhibiting only a mild hyperactivity phenotype (Fig. S5g). Such phenotype, which is likely due to extra-striatal defects, precludes a full interpretation of pharmacological challenges (Fig. S5). We thus focused on Islet1Cre; Ebf1fl/− mutants and first activated the direct pathway by injection of the D1 receptor agonist SKF82958 (Fig. 6n). In contrast to control mice, D1 receptor agonist injections in Ebf1 cKOs did not lead to significant locomotion increases, indicating important functional disruptions of the direct pathway in line with the anatomical defects observed (Figs. 5a-f). Conversely, activation of the indirect pathway through injections of the D2 receptor agonist quinpirole (Fig. 6o) caused strong, statistically similar decreases in the mobility in both control and cKOs, confirming that direct pathway functions are selectively affected by Ebf1 inactivation. Notably, we found that the indirect pathway is also functionally preserved in Dlx5/6::Cre; Ebf1fl/− mutants (Fig. S5h).
Our results reveal that development and functioning of the direct pathway is selectively impaired in adult Ebf1 cKOs, while cardinal properties of iSPN, including circuit integration and electrophysiological properties, are preserved. Thus, iSPN differentiation and maturation are largely independent of dSPN.
Ebf1 deletion affects iSPN migration and intermix in the striatal matrix
Next, we examined whether, in addition to its role in dSPN differentiation, Ebf1 could play a role in SPN intermix and striatal mosaic formation. As is the case for other dSPN markers such as Islet1 and Drd1 (Figs. 4 and S4), the transcriptional levels of FoxP2 were preserved in both early Ebf1 cKOs (Figs. 4 and S4). Therefore, we used its expression to analyze the distribution of dSPN in the embryonic striatum. We consistently observed non-overlapping populations of Ctip2+ FoxP2+ dSPN and Ctip2+ Drd2-EGFP+ iSPN in mutant striata (Fig. 7). However, we found that dSPN and iSPN distribution was strikingly different in the striatum of controls and of the two cKOs at E17.5 (Fig. 7). Indeed, on one side the density of dSPN in both cKOs was reduced in the dorsal striatum and increased in the lateral region in comparison to controls (Figs. 7a-e). Conversely, the density of Drd2-EGFP+ iSPN was enhanced dorsally and reduced laterally (Figs. 7f-i). Thus, although the overall respective proportions of dSPN and iSPN were preserved, Ebf1 inactivation perturbed the distribution of the two SPN populations.
The dorsal accumulation of iSPN near progenitor zones raised the possibility that their migration might be indirectly impaired in Ebf1 cKO. To test this hypothesis, we first performed timed EdU injections to follow SPN progression into the striatal mantle of Ebf1 mutants, using Drd2-EGFP and either Foxp2 or dtTomato as respective markers of iSPN and dSPN (Fig. S6 and data not shown). We observed that EdU-stained iSPN abnormally accumulated in the dorsolateral part of the striatum in cKOs (Fig. S6), suggesting that Ebf1 non cell-autonomously regulates their migration. To directly monitor iSPN migration, we performed two-photon time-lapse imaging in E15.5 Islet1Cre;Ebf1fl/−;Drd2-EGFP and Dlx5/6::Cre;Ebf1fl/−;Drd2-EGFP slices. We found that in both genetic backgrounds, mutant iSPN still harbored pattern of multidirectional migration as in controls (Fig. 7j and Movies S2-4). However, analyses over several time-lapse movies showed that iSPN average speed was significantly decreased (Fig. 7k), thereby revealing that Ebf1 non cell-autonomously controls the migration efficiency of iSPN. Thus Ebf1 inactivation in dSPN impairs iSPN migration and their process of intermixing. Since Ebf1 and Islet1 both control dSPN differentiation, we investigated the specificity of such intermixing defect by comparing the distribution of D2 receptor (Drd2) transcripts in Ebf1 and Islet1 cKOs (Fig. S7). In contrast to Ebf1 cKO, we found that iSPN distribution was not drastically altered in Islet1 cKO (Fig. S7). Therefore, iSPN intermixing specifically relies on Ebf1 expression in dSPN, indicating a novel and specific role for this transcription factor in the acquisition of dSPN properties.
In addition to changes in dorso-ventral distribution, we observed that iSPN presented a higher degree of segregation, especially in the lateral region (Figs. 7f-h”). This observation raised questions about possible disruptions of the striosomes/matrix organization. Using early striosome markers such as DARPP-329, 75 and Fig. S1), we found that the two compartments could be unambiguously identified in both cKOs (Figs. 8a-c). However, we observed that the matrix contained Enkephalinhigh domains (Figs. 8d-f) which segregated from cellular islands containing matrix dSPN (Fig. S8) in both cKOs, suggesting an intermixing deficit between iSPN and dSPN. To further test this possibility, we examined Drd2-EGFP controls and cKOs at E17.5, when compartment segregation is largely engaged20, 21, 47–51 and proto-striosomes can be identified by the expression of SubstanceP (not shown) and DARPP-32 (Figs. 8j-l), a maturation marker, which is turned on in the matrix only at later stages. We observed that the deficit in iSPN and dSPN intermixing was more pronounced in the matrix (Figs. 8k-l’), consistently with previous reports showing that Ebf1 inactivation more specifically affects matrix dSPN44. Collectively, our findings reveal that the intermixing of iSPN and dSPN in the matrix specifically relies on Ebf1 expression in dSPN (Fig. 8m-n). This major step in the emergence of a balanced striatum parallels the formation of compartments, thereby revealing that multiple migratory processes regulate the assembly of the striatal mosaic.
Discussion
Striatal functioning depends on the balanced activation of intermixed direct and indirect pathway neurons76. It is thus essential to understand how the two subtypes are specified and how they assemble into a mosaic striatal architecture. Here, we show that while dSPN and iSPN differentiation is largely independent, their intermixing relies on a developmental crosstalk, which is crucial to organize striatal architecture and is largely independent from striosome and matrix compartment formation.
While both SPN derive from the lgE19, 22–24, 49, producing fully differentiated dSPN or iSPN in vitro77–81 has remained a challenge. While there was increasing evidence that distinct transcriptional programs underlie the specification of both subtypes23, 41, 45, 46, in vitro studies raised questions regarding the mechanisms controlling SPN differentiation. Our results confirm that dSPN and iSPN possess distinct genetic fingerprints from early embryonic stages, reinforcing the idea that the two cell types are already specified when they enter the striatal mantle.
Furthermore, iSPN progressively invade dSPN-rich areas through dynamic multidirectional migration. Intriguingly, iSPN are initially detected in the lateralmost part of the striatum, consistently with the observation that transcription factors regulating iSPN development, Six3 and Sp9, are located in the dorsal part of the LGE23. Tangential migration is a well-known mechanism for increasing neuronal diversity in different brain structures. In the telencephalon, tangential migration has mostly been associated with interneuron behavior82, 83, but has also been involved in the positioning of other LGE-derived projection neurons, including arkypallidal neurons of the globus pallidus84, 85, corridor neurons86–88 and dSPN-like neurons of the central89 and extended amygdala90. Further investigation will be required to determine the cellular and molecular mechanisms governing the migration and interaction of different LGE-generated populations. Nevertheless, tangential migration of projection neurons is emerging as a shared mechanism to build and increase cellular diversity in basal ganglia nuclei, in contrast to the laminated cerebral cortex where projection neurons migrate radially from the underlying proliferative neuroepithelium91.
Regarding the transcriptional cascades governing SPN identity, our work indicates that Ebf1 plays a major role in dSPN differentiation, independently of the other known regulators Islet1 and Foxp2. Ebf1 is a member of the COE family that plays a key role in the specification of B cells and the differentiation of neuronal subtypes79, 92–100. Ebf1 acts both as a direct regulator of downstream target genes92, 97, 98, 100 and as a chromatin remodeler, poising or regulating the accessibility of enhancers101, 102. Ebf1 is expressed in the developing striatum42 and analyses of Ebf1 full knockout mice showed normal proliferation in the LGE but defective survival of matrix dSPN and disorganized projections to the SN43, 44. By combining analysis of two distinct cKOs, we first showed that Ebf1 is dispensable for the fate choice between dSPN and iSPN identity and the core aspects of differentiation into GABAergic SPN neurons. Indeed, Ebf1 inactivation leads to an abnormal differentiation of dSPN-like neurons that nonetheless retain their GABAergic identity and expression of DARPP-32, Ctip2, FoxP1, Drd1, Islet1 and FoxP2. Notably, deregulated genes included Zfp521 and Mef2c, which are directly bound by Ebf1 in pro-B cells and thus likely direct targets92, 97. Second, Ebf1 is not expressed in the ventral striatum and its inactivation selectively perturbs the matrix compartment of the dorsal striatum, potentially revealing a subdivision of dSPN neurons. Third, Ebf1 regulates expression of genes controlling axon and synaptogenesis, thereby highlighting a major role in circuit integration. Importantly, Ebf1 and Islet1 both regulate direct pathway formation and the expression of PlexinD127, 41, 42, 45, suggesting that the two genes might cooperatively regulate specific features of dSPN neurons.
Finally, we found that Ebf1 non cell-autonomously regulates iSPN intermixing with dSPN in the matrix compartment. In Ebf1 cKOs, iSPN are correctly specified and functional, but their progression and intermixing with dSPN is impaired. This intermixing deficit occurs independently of the formation of striosomes/matrix and is more pronounced in the matrix compartment. Remarkably, this phenotype is absent in Islet1 cKO, indicating a novel and specific role for Ebf1 in the building of striatal functional organization. The underlying mechanisms remain to be deciphered and could include the production of membrane-bound or secreted factors acting on iSPN. Our study also reveals that the intermix of dSPN/iSPN and the formation of striosomes/matrix compartments constitute parallel processes. Indeed, it has been shown that early-born and late-born SPN interact in late embryogenesis to respectively form the striosomes and the matrix via migration and ephrinA/EphA4-dependant cell-sorting15, 20, 21, 24, 47, 49–51, 103. In Ebf1 cKOs, striosomes do form but the intermix of matrix neurons is defective. Our study thus reveals that distinct steps of migration and reorganization control the emergence of two compartments comprising intermixed dSPN/ iSPN (Fig. S9), an architecture which is essential for striatal functioning.
Taken together, our work shows that while iSPN and dSPN specification is largely independent, the assembly of striatal mosaic emerges from a dSPN-dependent tangential migration of iSPN. This study establishes a novel framework for the formation of the striatum, a major structure associated with developmental disorders, and provides key insights on how migration controls the wiring of neural circuits.
Methods
Mouse lines
For fate mapping studies, Islet1Cre/+ animals63 were crossed with either R26mt/mt or R26mt/mt;Drd2-EGFP mice52. To obtain Islet1Cre/+;Ebf1fl/−, Islet1Cre/+;Ebf1fl/− R26mt/+ and Islet1Cre/+:Ebf1fl/−;Drd2-EGFP conditional mutants, we initially crossed Islet1Cre/+ mice with Ebf1+/−104 to generate Islet1Cre/+;Ebf1fl/− animals. These were in turn backcrossed with either Ebf1fl/fl97, 105, Ebf1fl/fl;R26mt/mt, or Ebf1fl/fl;Drd2-EGFP mice, respectively. Similarly, Dlx5/6::Cre;Eb1fl/− and Dlx5/6::Cre;Ebf1fl/−;Drd2-EGFP mice were obtained by crossing Dlx5/6::Cre65 mice with Ebf1+/− to generate Dlx5/6::Cre;Ebf+/− animals, which were in turn backcrossed with Ebf1fl/fl or Ebf1fl/f;Drd2-EGFP mice. All transgenic lines were maintained on a C57/Bl6 background, with the exception of Islet1Cre/+ and Drd2-EGFP lines that remained on a B6D2F1/J genetic background. Heterozygous embryos did not show any phenotype and were used as controls. Nkx2.1Cre/+ mice provided by S. Anderson’s laboratory106. 12-μm-thick cryosections from Dlx1Cre/+;Islet1fl/fl embryos provided by K. Campbell’s laboratory41. The day of vaginal plug was considered as embryonic day (E) 0.5 and day of birth as postnatal day (P) 0. Animals were bred under French and EU regulations, following recommendations of the local ethics committee.
In situ hybridization and immunohistochemistry
For in situ hybridization, brains were fixed overnight in 4% paraformaldehyde in PBS (PFA) at 4°C. 80 to 100μm-thick free-floating vibratome sections (Leica S1000) were hybridized as described105; Dlx1Cre/+;Islet1fl/fl and respective control cryosections were hybridized as described107. For immunohistochemistry, mice were perfused with 4% PFA. Brains were dissected and post fixed overnight at 4°C. Immunohistochemistry was performed on 60μm-thick free-floating vibratome sections. Slices were incubated 1h at room temperature (RT) in a blocking solution containing 0,25% Triton X-100 (Sigma), 0,02% Gelatine in PBS, and incubated in the same blocking solution with primary antibodies overnight at 4°C. Primary antibodies were used at the following concentrations: rat anti-CTIP2 1/500 (Abcam), mouse anti-DARPP-32 1/100 (Santa Cruz), rabbit anti-DARPP-32 1/1000 (Millipore), rabbit anti-DsRed 1/500 (Living colors), rabbit antiEnkephalin 1/500 (Millipore), rabbit anti-Ebf1 1/250 (Abcam), rabbit anti-FoxP1 1/200 (Abcam), goat anti-FoxP2 1/200 (Santa Cruz), chicken anti-GFP 1/1000 (Aves), rabbit anti-Slc35d3 1/250 (Novusbio), rat anti-SubstanceP 1/400 (Millipore), rabbit anti-Tyrosine Hydroxylase 1/1000 (Abcam), guinea pig anti-vGlut1 1/10000 (Millipore), guinea pig anti-vGlut2 1/10000 (Millipore). Sections were rinsed several times in PBS and incubated from 2h to overnight at 4°C with the appropriate fluorescent secondary antibodies: A488-conjugated donkey antirabbit, anti-rat or anti-chicken, Cy3-conjugated donkey anti-rat, anti-goat, antimouse, Cy5-conjugated donkey anti-rabbit or anti-rat, DyLight488-conjugated donkey anti-guinea pig (1/400, all antibodies from Jackson ImmunoResearch). Hoechst (Sigma) was used for fluorescent nuclear counterstaining and sections were mounted in Mowiol or Vectashield (Vector).
Birthdating
Pregnant dams were injected intraperitoneally at the appropriate gestation day with a solution containing 5-Ethynyl-2’-deoxyuridine (EdU, Thermo Fisher). 60-100μm-thick free-floating vibratome sections were processed following manufacturer instructions (Click-iT EdU Alexa Fluor 488 Imaging kit, Life Technologies) for 30 minutes at RT. Sections were rinsed three times in 3% BSA and then in PBS. Hoechst staining was performed for 30 min at RT before pursuing the immunohistochemistry protocol as described above.
Image acquisition, analysis and quantification
Images were acquired on a fluorescence microscope (Leica MZ16 F), a fluorescent microscope (Leica DM5000 B) or a confocal microscope (Leica TCS SP5). Images were then processed with ImageJ and Adobe Photoshop software. For cell density and colocalization analysis, three different rostro-caudal striatal levels of the striatum were initially defined, using the anterior commissure as an anatomical landmark. Single-plane confocal images were taken at each level in three different animals for each condition. For cell distribution analysis in Fig. 7, a high-definition tilescan of the whole striatum at medial level was imaged on a single confocal plane in at least three different animals for each condition. The tilescans were later subdivided in lateral and dorsal regions (Fig. 7d) for quantification. Cell counting was performed semi-automatically using built-in functions in ImageJ. For each experiment, sample images were manually counted to double-check the quality of semi-automated counting.
Timelapse imaging and analysis
E13.5 and E15.5 Drd2-EGFP and Dlx5/6::Cre;Ebf1fl/−;Drd2-EGFP embryos were dissected and kept in an ice-cold solution consisting of L-15 Medium (SIGMA-Aldrich) supplemented with 3% Glucose. Brains were included in 3,5% low-melting agarose (Promega) in L-15 Glucose+ solution and 350 μm-thick coronal slices were cut on a vibratome. Slices were imaged using a multi-photon microscope (DIMM, IBENS Imaging Platform) over a period of up to 240’, while being constantly perfused with a L-15 Glucose+ solution at 37°C and bubbled with O2/CO2 (95% / 5%). 100 μm-thick z-stacks were imaged every 6’ for each single frame. For time-lapse movie analysis, eventual drift in the three dimensions was first corrected using IMARIS software (Bitplane, Oxford); subsequently, z-stacks were flattened on a two-dimensional plane. The high number of cells present in each single movie rendered automated analysis of cell displacement unadvisable. Therefore, single cell displacement was measured manually frame-by-frame using the “Manual Tracking” plugin in ImageJ. Cells included in the analysis responded to two criteria: being readily identifiable from the first to the last frame of the movie and showing a total net displacement of more than 50 μm. In order to examine directionality and migratory pattern of iSPN migration, stationary (total net displacement less than 50 μm) cells were not included in the analysis.
RNA-sequencing
E17.5 Dlx5/6::Cre;Ebf1fl/−, Islet1Cre/+;Ebf1fl/− and control embryos were dissected in RNAse-free conditions on ice. Brains were conserved in RNAlater stabilization reagent (Qiagen) solution. Following genotype identification via PCR, messenger RNA was obtained from n=3 brains from each condition using RNeasy mini kit (Qiagen). Library preparation and Illumina sequencing were performed at the Ecole Normale Supérieure Genomic Platform (Paris, France). Messenger (polyA+) RNAs were purified from 400ng of total RNA using oligo(dT). Libraries were prepared using the strand specific RNA-Seq library preparation TruSeq Stranded mRNA kit (Illumina). Libraries were multiplexed by 9 on run Nextseq 500. A 75bp read sequencing was performed on a Nextseq 500 device (Illumina). A mean of 37,5±8,95 million passing Illumina quality filter reads was obtained for each of the 9 samples. Analyses were performed using the Eoulsan pipeline, including read filtering, mapping, alignment filtering, read quantification, normalisation and differential analysis. Before mapping, poly N read tails were trimmed, reads ≤40 bases were removed, and reads with quality mean ≤30 were discarded. Reads were then aligned against the Mus musculus genome from Ensembl version 81 using STAR (version 2.4.0k). Alignments from reads matching more than once on the reference genome were removed using Java version of samtools. To compute gene expression, Mus musculus GFF3 genome annotation version 81 from Ensembl database was used. All overlapping regions between alignments and referenced genes were counted using HTSeq-count 0.5.3. Sample counts were normalized using DESeq 1.8.3. Statistical treatments and differential analyses were also performed using DESeq 1.8.3108.
Slice preparation and electrophysiological recordings
In vitro electrophysiological recordings were performed in coronal slices from the dorsal striatum of control animals (either Islet1Cre/+;R26mt/+ or Drd2-EGFP mice), Dlx5/6::Cre;Ebf1fl/−;Drd2-EGFP and Islet1Cre/+;Ebf1fl/−;R26mt/+ mice. Mice were anesthetized with isofluorane before decapitation. After isolation, the portion of the brain containing the striatum was placed in bicarbonate-buffered saline (BBS) at 2–5°C for a few minutes. Slices (300 μm) were then cut using a 7000smz-2 vibratome (Campden Instruments Ltd.). The slicing procedure was performed in an ice-cold solution containing (in mM): 130 potassium gluconate, 15 KCl, 0.05 EGTA, 20 Hepes, 25 glucose, 1 CaCl2 and 6 MgCl2. Slices were then briefly transferred to a solution containing (in mM): 225 D-mannitol, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 25 glucose, 1 CaCl2 and 6 MgCl2, and finally stored for the rest of the experimental day at 33 °C in oxygenated BBS, containing: 115 NaCl, 2.5 KCl, 1.6 CaCl2, 1.5 MgCb, 1.25 NaH2PO4, 26 NaHCO3 and 30 glucose (pH 7.4 after equilibration with 95% O2 and 5% CO2). For all recordings, slices were continuously superfused with oxygenated BBS, supplemented with the GABAa receptor blocker SR95531 (Gabazine; 2μM) and with Tetrodotoxin (TTX; 500nM), at 32–34°C. Electrophysiological recordings were performed from either dSPN-dTomato positive, iSPN-dTomato negative, dSPN-GFP negative or iSPN-GFP positive dorsal striatal cells. Cells were patched in the transmitted deep red light with which slices were visualized using a CoolSnap HQ CCD camera (Photometrics) run by Metamorph software (Universal Imaging) and mounted on either a Slicescope (Scientifica), or a BX51 (Olympus) microscope. Before patching, Tomato or GFP positive/negative cells were identified by the presence/absence of somatic fluorescence using LEDs of the corresponding excitation wavelengths (Thorlabs) coupled to the slice chamber via the epifluorescence pathway of the microscope.
Pharmacology and Behavior analyses
The motor responses of Dlx5/6::Cre;Ebf1fl/− and Islet1Cre/+;Ebf1fl/− mice and their corresponding controls were examined in a round open field arena (diameter: 38 cm) following injection of pharmacological agents. Minimum two weeks separated tests performed with different drugs on the same cohort of mice. For each condition, animals were acclimated to the experimental luminosity conditions (27–35 lux) of the test room for 1 hr. Following adaptation, baseline motor activity in the open field of each mouse was then measured for 8 minutes before subcutaneous injection with one of the following solutions: 0,9% NaCl (control injection); 2mg/Kg D1 receptor agonist SKF82958 (Sigma); and 1mg/Kg D2 receptor agonist Quinpirole (Sigma). Drugs were administered in a volume of 10 ml/Kg of body weight. Following the injection, animals were left in their cage for 40 minutes. They were then re-introduced into the arena for recording their motor activity over another 8 min-long timeperiod. Post-hoc analysis of the total distance travelled was performed using Ethovision software (Noldus).
Statistical analyses
All data are presented as mean ± SD or SEM (detailed in each figure legend). Two-tailed non-parametric Mann-Whitney U test was used to compare two distributions in all experiments, with the exception of RNA-seq analysis (see above). All statistical analyses were performed using GraphPad Prism software. P-values are as follows: * p<0.05, **p<0.01, ***p<0.0001.
Additional information
Supplementary information includes 9 supplementary figures, 4 movies and 2 Tables/Spreadsheets.
Author contributions
AT and SG designed the study and wrote the manuscript. MAD performed and analyzed electrophysiological recordings. MAD, FM, DP and AT performed motor behavior experiments and analyses. FC and SL performed the RNAseq and transcriptomic analyses. BM provided expertise for two-photon imaging. AT, LL, CM and MK performed all other experiments and analyzed data. IG, PMS, KC and RG provided mouse models. All the authors edited the manuscript. SG supervised the project.
Competing Financial Interests
The authors declare no competing financial interests.
Acknowledgements
We thank members of the S.G. lab, M. Casado, J.A. Girault, C. Goridis, D. Hervé, C. Lena and E. Valjent for stimulating discussions. We are grateful to C. Auger, D. Souchet, A. Boudjouher and C. Le Moal for excellent technical assistance. We thank E. Thierion and L. Danglot for scientific input. We are grateful to the IBENS Imaging Facility (France BioImaging, supported by ANR-10-INBS-04, ANR-10-LABX-54 MEMO LIFE and ANR-11-IDEX-000-02 PSL* Reseach University, “Investments for the future”; NERF N°2011-45; FRM DGE 20111123023; and FRC Rotary International France), the IBENS Genomics Facility (supported by the France Génomique national infrastructure, funded as part of the “Investissements d’Avenir” program managed by the Agence Nationale de la Recherche, ANR-10-INBS-09) and the IBENS acute transgenesis facility. AT is part of the ENP doctoral program. AT was supported by a doctoral fellowship from Boehringer Ingelheim Fonds (BIF), the ENP program and the Labex Memolife « Investements for the future » (ANR-10-LABX-54 MEMO LIFE and ANR-11-IDEX-0001-02 PSL* Research University). This work was supported by grants from INSERM, CNRS, the ANR program (ANR-12-BVS4-0010-01) and the ERC NImO (616080) to S.G, from the ANR program (ANR-12-JSV4-004) to D.P and from NIH (R01 MH090740) to KC.
References
- 1.↵
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.
- 13.
- 14.
- 15.↵
- 16.
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.
- 27.↵
- 28.
- 29.
- 30.
- 31.↵
- 32.
- 33.
- 34.
- 35.
- 36.
- 37.
- 38.
- 39.
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.
- 60.↵
- 61.↵
- 62.
- 63.↵
- 64.
- 65.↵
- 66.
- 67.
- 68.↵
- 69.↵
- 70.↵
- 71.
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.
- 79.↵
- 80.
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.
- 94.
- 95.
- 96.
- 97.↵
- 98.↵
- 99.
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵