Brain-wide spatiotemporal mapping of glutamate/GABA dual-transmitter neurons during postnatal development in mice

Abstract

Emerging evidence suggests that individual neurons can utilize multiple neurotransmitters, challenging traditional views of neuronal identity. However, as most studies to date have focused on the adult stage, a fundamental question remains: does this dual-transmitter phenotype emerge only in adulthood, or does it have specific developmental origins? To address this, we performed a comprehensive whole-brain screen across various postnatal stages (P0–P60) in mice. Using a dual Flp/Cre-mediated intersectional labeling strategy to drive fluorescent protein expression, we genetically marked glutamatergic (Vglut2⁺) and GABAergic (Vgat⁺) neurons. Our results reveal that dual-transmitter neurons are not exclusive to the adult brain but are already prevalent during development. Notably, the emergence of these neurons is asynchronous across brain regions, with some appearing at P0 and others gradually emerging at later time points. Furthermore, direct mRNA in situ detection confirmed the concomitant expression of both markers in selected regions, establishing that these dual-transmitter neurons arise through co-expression rather than a developmental switch between neurotransmitters, at least within the brain regions we examined. Overall, our study deepens the understanding of dual-transmitter phenomena by demonstrating that this phenotype is established during development and follows region-specific temporal trajectories.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13578-026-01591-2.

Introduction

For decades, the fundamental understanding of neurotransmission was guided by a simplified version of “Dale’s Principle,” which posits that a single neuron releases one type of neurotransmitter. However, accumulating evidence has demonstrated the co-existence and co-release of excitatory (glutamate) and inhibitory (GABA) neurotransmitters in different neuronal populations of the brain [1–7]. Notably, recent studies across the brain have suggested this type of dual-transmitter phenotype far more widespread than previously recognized [8, 9].

Advances in intersectional genetic strategies have enabled the identification of approximately 30 brain regions containing co-expressing neurons in the adult mouse [9]. Key populations have been localized in the supramammillary nucleus (SUM), ventral tegmental area (VTA), and globus pallidus internal segment (GPi) [3, 10], each exhibiting distinctive synaptic architectures [11]. Functionally, these neurons are critical for modulating motivated behaviors and enhancing spatial memory [4, 5].

While the distribution and functions of these neurons have been characterized in the adult brain, their developmental trajectory remains largely unexplored [9]. Consequently, a fundamental question remains: are these dual-transmitter identities formed during development, or are they recruited during late stages? Furthermore, do different brain regions share the same or individualized temporal properties regarding the appearance of dual-transmitter phenotype?.

In the present study, we employed an intersectional genetic approach using mice to map neurons with a history of dual-transmitter expression across postnatal stages (P0, P7, P14, P21, and adult). Our findings reveal that the dual-transmitter phenomenon is not an exclusive feature of the adult brain but emerges dynamically during development. Crucially, we identified significant regional heterogeneity in these temporal patterns. While regions like the GPi exhibit dual-transmitter markers as early as P0, others, such as the medial amygdalar nucleus, posteroventral part (MeApv), show a notable developmental lag, with co-expression only becoming apparent around P7. mRNA analyses suggest that, at least in certain brain regions, these neurons likely undergo neurotransmitter co-existence rather than a developmental switch during maturation. Our results provide a comprehensive dataset characterizing the distribution and temporal evolution of potential neurotransmitter co-existence during development of the brain.

Results

Widespread distribution of Vglut2/Vgat dual-phenotype neurons in adult and P21 mouse brains

To establish a brain-wide baseline and determine whether this dual-transmitter phenotype is acquired developmentally or represents an adult-specific feature, we utilized an intersectional genetic strategy. The Vglut2-Flp mice were crossed with Vgat-Cre and Ai65 mice to generate the Vglut2-Flp;Vgat-Cre;Ai65 reporter mice, in which tdTomato labels the dual-transmitter neurons (Fig. 1A).

Fig. 1.

Brain-wide distribution of Vglut2/Vgat co-expressing neurons in adult and P21 mice. A Schematic illustration of the intersectional genetic strategy used to generate Vglut2-Flp;Vgat-Cre;Ai65 reporter mice. B Schematic coronal sections illustrating the brain-wide distribution of Vglut2/Vgat dual-transmitter neurons in adult mice. Red dots represent the mapped coordinates of co-expressing somata. C, D Representative fluorescence images of tdTomato-labeled dual-transmitter neurons across various specific brain regions in adult mice. E Quantification of tdTomato-positive neurons in the indicated brain regions of adult mice (n = 3 mice). F Schematic coronal sections illustrating the distribution of Vglut2/Vgat neurons in P21 mice. G, H Representative fluorescence images of tdTomato expression in corresponding brain regions of P21 mice. I Quantification of tdTomato-positive neurons in P21 mice (n = 3 mice). Scale bars: 2 mm (C, D, G macroscopic views); 1 mm (H macroscopic view); 100 μm (all high-magnification insets)

In adulthood, we observed widespread co-expression across multiple brain regions (Fig. 1B, C, and Supplementary Fig. 1). Crucially, our model successfully recapitulated previously reported dual-transmitter populations in regions such as the GPi, SUM, VTA, and dorsal raphe (DR) [5, 10, 12, 13], validating the fidelity of our labeling strategy. Beyond these known hubs, we also observed previously undocumented co-expression in a series of nuclei of the brainstem and cerebellum, such as the superior central nucleus raphe medial part (CSm), pontine reticular nucleus, caudal part (PRNC), and intermediate reticular nucleus (IRN) (Fig. 1D).

Importantly, our strategy illuminated dense axonal projections alongside somata. For instance, while the lateral habenula (LH) lacked positive somata, we observed intense terminal labeling within this region. This anatomical feature provides support for previous studies on functional circuits, which demonstrated that the dual neurotransmitter neurons in the GPi of the basal forebrain and VTA of the midbrain provided a strong glutamate/GABA projection to the LH [10, 13–15] (Fig. 1B).

To assess whether this extensive network exists in immature brains, we analyzed P21 mice—a stage marking the end of the critical period for many sensory and motor circuits [16]. The spatial distribution at P21 was virtually identical to adults (Fig. 1F–H). However, the total number of labeled somata was smaller (Fig. 1E, I, and Table 1). The observed variations in soma counts may reflect a genuine increase in Vglut2/Vgat co-expressing neurons, or alternatively, a maturation lag in the genetic reporter system. These findings indicate that the dual-transmitter phenotype—characterized by the expression of both glutamatergic and GABAergic molecular markers—emerges before maturity, even if their simultaneous presence within the cell is not yet fully confirmed.

Table 1.

Quantification of tdTomato⁺ neurons of different brain regions across postnatal development. Statistical significance between each developmental stage (P0, P7, P14, and P21) and the Adult group was determined using unpaired Student’s t-tests. For all tests, P < 0.05 was considered statistically significant. “NA” in the table indicates that cell counting was not performed due to sub-optimal tissue quality

	Mean		SD		n	Mean	SD		n	Mean	SD		n		Mean	SD	n
	MeA					AHNP				GPi					VTA
adult	378.3		3.512		3	446.7	13.01		3	878.3	23		3		493.7	11.06	3
P21	280.7		15.95		3	377	12.12		3	819.3	23.54		3		453.7	32.52	3
P14	191		14.11		3	156	16.09		3	728.3	3.055		3		343.7	14.57	3
P7	105.3		8.327		3	67.33	5.686		3	688.3	6.658		3		259.7	16.26	3
P0	0		0		3	0	0		3	535.7	21.39		3		190.7	23.71	3
	P value					P value				P value					P value
	P21 vs adult		P < 0.001			P21 vs adult	P = 0.0025			P21 vs adult	P = 0.0371				P21 vs adult	P = 0.1139
	P14 vs adult		P < 0.001			P14 vs adult	P < 0.001			P14 vs adult	P < 0.001				P14 vs adult	P = 0.001
	P7 vs adult		P < 0.001			P7 vs adult	P < 0.001			P7 vs adult	P < 0.001				P7 vs adult	P < 0.001
	P0 vs adult		P < 0.001			P0 vs adult	P < 0.001			P0 vs adult	P < 0.001				P0 vs adult	P < 0.001
	SUM					EW/RL				NI					PRNC
adult	970.3		15.63		3	468	31		3	558.7	21		3		373.7	8.145	3
P21	843		16.52		3	423.3	8.963		3	260	25.24		3		261.3	11.72	3
P14	770		18.5		3	401.9	26.3		3	202.7	17.62		3		215	14.18	3
P7	383		8.185		3	NA	NA		3	140.7	11.24		3		177.7	10.02	3
P0	318.7		20.6		3	335.3	20.01		3	NA	NA		NA		NA	NA	NA
	P value					P value				P value					P value
	P21 vs adult		P < 0.001			P21 vs adult	P = 0.0746			P21 vs adult	P < 0.001				P21 vs adult	P < 0.001
	P14 vs adult		P < 0.001			P14 vs adult	P = 0.0018			P14 vs adult	P < 0.001				P14 vs adult	P < 0.001
	P7 vs adult		P < 0.001			P7 vs adult	NA			P7 vs adult	P < 0.001				P7 vs adult	P < 0.001
	P0 vs adult		P < 0.001			P0 vs adult	P < 0.001			P0 vs adult	NA				P0 vs adult	NA
	SCs/SCop/SCig/SCiw					mlf				CENT3gr					Crus1
adult	798	23.64		3		243.3	13.43	3		3102		72.67		3	3728	57.55	3
P21	522.3	5.686		3		181	18.36	3		2869		123.5		3	3357	128.9	3
P14	507.3	23.01		3		111.3	9.609	3		1360		127.8		3	2445	13.75	3
P7	405.3	8.327		3		97.33	12.5	3		497		12.49		3	74	6.245	3
P0	375.7	10.69		3		NA	NA	NA		NA		NA		NA	NA	NA	NA
	P value					P value				P value					P value
	P21 vs adult	P < 0.001				P21 vs adult	P = 0.009			P21 vs adult		P = 0.0478			P21 vs adult	P = 0.0104
	P14 vs adult	P < 0.001				P14 vs adult	P < 0.001			P14 vs adult		P < 0.001			P14 vs adult	P < 0.001
	P7 vs adult	P < 0.001				P7 vs adult	P < 0.001			P7 vs adult		P < 0.001			P7 vs adult	P < 0.001
	P0 vs adult	P < 0.001				P0 vs adult	NA			P0 vs adult		NA			P0 vs adult	NA
	IRN					CSm				MDRNd					UvUmo
adult	567	14.11		3		468	31	3		579.3		10.6		3	2734	75.78	3
P21	442.7	29.57		3		423.3	8.963	3		360		9.644		3	2115	140.2	3
P14	335	5		3		401.9	26.3	3		257		15.39		3	1392	115.1	3
P7	229.7	3.786		3		NA	NA	3		135		14.18		3	1110	47.08	3
P0	NA	NA		NA		335.3	20.01	3		113		6		3	NA	NA	NA
	P value					P value				P value					P value
	P21 vs adult	P = 0.0028				P21 vs adult	P = 0.0019			P21 vs adult		P < 0.001			P21 vs adult	P = 0.0025
	P14 vs adult	P < 0.001				P14 vs adult	P < 0.001			P14 vs adult		P < 0.001			P14 vs adult	P < 0.001
	P7 vs adult	P < 0.001				P7 vs adult	P < 0.001			P7 vs adult		P < 0.001			P7 vs adult	P < 0.001
	P0 vs adult	NA				P0 vs adult	P < 0.001			P0 vs adult		P < 0.001			P0 vs adult	NA

Early postnatal brains exhibit profound regional heterogeneity in the developmental dynamics of the dual-phenotype

Given the reduction at P21 compared to adults, we hypothesized that the acquisition of the dual-transmitter phenotype might parallel early postnatal circuit wiring [17]. We therefore tracked the expression backward to P14 and P7. At P14, the macroscopic spatial framework of Vglut2/Vgat co-expressing neurons closely mirrored the adult pattern (Fig. 2A, B), albeit with a further decline in somatic density (Fig. 2C, D and Supplementary Fig. 2A). Notably, extensive dual-transmitter populations were already prominent in the brainstem and cerebellum (Fig. 2C). Extending this analysis to P7, we observed an even more marked reduction in the number of tdTomato-labeled somata (Table 1). Of particular note, somatic counts in the cerebellum and brainstem at P7 were drastically lower than in subsequent stages, even though the fundamental spatial organization had already been established (Fig. 2G, H).

Fig. 2.

Spatiotemporal distribution of Vglut2/Vgat co-expressing neurons at early postnatal stages (P14 and P7). A Schematic coronal sections illustrating the spatial distribution of Vglut2/Vgat neurons (red dots) in P14 mice. B, C Representative fluorescence images of tdTomato-labeled neurons in specific brain regions of P14 mice. D Quantification of tdTomato-positive neurons in the corresponding brain regions at P14 (n = 3 mice). E Schematic coronal sections illustrating the spatial distribution in P7 mice. F, G) Representative fluorescence images of tdTomato expression in P7 mice. H Quantification of tdTomato-positive neurons at P7 (n = 3 mice). Scale bars: 2 mm (B, F macroscopic views); 1 mm (C, G macroscopic views); 100 μm (all high-magnification insets)

To pinpoint the precise developmental onset of these populations, we systematically analyzed the brains of neonates at P0. Intriguingly, the majority of brain regions that exhibit tdTomato-positive neurons in adulthood already possessed labeled cells at this earliest postnatal stage (Fig. 1, and Supplementary 3). However, regional specificities were also observed. Specifically, in forebrain regions such as the medial amygdala (MeA) and the anterior hypothalamic nucleus, posterior part (AHNP), Vglut2/Vgat co-expressing neurons were completely absent at P0 (Fig. 3A, B). In sharp contrast, the GPi—representing one of the regions with early-onset labeling—exhibited stable and abundant dual-transmitter neuron expression at P0 into adulthood (Fig. 3C, D).

Fig. 3.

Region-specific developmental trajectories of dual-transmitter neurons. A Schematics illustrating the spatiotemporal presence of Vglut2/Vgat neurons (red dots) in the MeA and AHNP across distinct developmental stages. B Representative fluorescence images and high-magnification views of the MeA and AHNP at different developmental stages. C Schematics illustrating the stable spatiotemporal presence of Vglut2/Vgat neurons in the GPi from birth to adulthood. D Representative fluorescence images and high-magnification views of the GPi across development. E Heatmap showing the number of Vglut2/Vgat dual-positive neurons in the MeA, AHNP, and GPi during development. The color scale (0–800 +) represents the average cell count per biological replicate (n = 3 mice). Scale bars: 2 mm (macroscopic views); 100 μm (high-magnification insets)

This pronounced regional heterogeneity could be developmentally significant. The GPi is deeply involved in fundamental basal ganglia output pathways that require early functional establishment, whereas the MeA processes complex olfactory and social cues that largely mature during late postnatal life [18]. Our quantitative tracking suggests that while early-maturing regions (like the GPi) achieve a stable dual-transmitter population size at birth, late-maturing regions (like the MeA) might undergo a progressive, postnatal acquisition of this phenotype (Fig. 3E, and Table 1).

RNAscope confirms real-time transcriptional co-expression in examined brain regions

A limitation of intersectional genetic fate-mapping is its developmental accrual nature; it cannot distinguish whether dual-positive signals represent the concurrent, real-time expression of both transporters, or a historical “neurotransmitter switching” during development [19, 20]. To unequivocally resolve this at the single-cell level, we employed RNAscope in situ hybridization to directly detect Vglut2 and Vgat mRNA transcripts.

We first validated the correspondence between these two labeling approaches. Using the MeApv—a region reliably labeled by the intersectional strategy in adulthood—as an example, we found that tdTomato RNA-positive cells were also dual-positive for Vglut2 and Vgat mRNA (Supplementary Fig. 4). This confirms that the results from both labeling strategies are consistent and can be used for comparative analysis.

To confirm the reliability of our RNAscope performance, we first examined adult brain regions such as the SUM, RL, and VTA, which are known to harbor neurons that co-release glutamate and GABA. Consistent with previous findings, we identified a substantial population of Vglut2 and Vgat mRNA double-positive neurons in these nuclei (Supplementary Fig. 5). In contrast, no co-expression of Vglut2 and Vgat was observed in the spinal cord, a region where this dual-transmitter phenotype has not yet been documented (Supplementary Fig. 5E).

We then utilized the MeApv as a representative late-onset region to track real-time developmental dynamics. At P60, a single-cell colocalization confirmed dual-transmitter identity (Fig. 4A). Crucially, reverse developmental tracking revealed a progressive decline in co-expressing neurons at P21 and P7, culminating in a complete absence of colocalization at P0 (Fig. 4B–E). This transcriptomic evidence is consistent with our reporter mouse data, proving that the dual-phenotype in the MeApv is genuinely acquired postnatally, not an artifact of delayed reporter expression. Conversely, in the early-onset GPi, clear Vglut2/Vgat colocalization was robustly detected at P0 (Fig. 4F, G) and persisted through P60 (Fig. 4H, I). Together, these direct transcriptional validations demonstrate that within individual neurons, the co-expression of Vglut2 and Vgat during development occurs in at least some brain regions. Furthermore, the timing of its emergence follows distinct, region-specific trajectories.

Fig. 4.

Single-cell transcriptomic validation of concurrent Vglut2 and Vgat expression in the MeApv and GPi. A–D Representative confocal images of multiplex RNAscope in situ hybridization and corresponding schematics for the MeApv at P60 (A), P21 (B), P7 (C), and P0 (D). Green signals indicate Vglut2 mRNA, red signals indicate Vgat mRNA. Red dots in schematics represent colocalized cells. E Quantification of dual-positive cells in the MeApv across development. Each dot represents one slice (13–16 slices from 3 mice per group). F, G Representative RNAscope confocal images (F) and schematic mapping (G) of Vglut2 (green) and Vgat (red) co-expression in the GPi at P0. H, I Representative RNAscope confocal images (H) and schematic mapping (I) of the GPi at P60. Scale bars: A 200 μm, 100 μm, 50 μm; B 100 μm, 20 μm; C, D 50 μm, 5 μm; G, I 50 μm, 10 μm

Discussion

This study investigates the brain-wide distribution and spatiotemporal heterogeneity of Vglut2/Vgat dual-transmitter neurons throughout mouse postnatal development. By integrating intersectional genetic fate-mapping with single-cell transcriptional validation, our findings provide robust molecular evidence for the dual-transmitter phenotype. Furthermore, we demonstrate that the emergence of the dual-transmitter signature is not a monolithic, synchronized developmental event, but manifests as a heterogeneous mosaic of region-specific trajectories.

Our longitudinal analysis from early postnatal stages (P7/P14) to adulthood suggests a principle in the development of dual-transmitter systems: the establishment of the spatial blueprint precedes population scaling. We observed that the macroscopic spatial framework—particularly in the hindbrain, cerebellum, and basal ganglia—is already recognizable by P7, yet the observed somatic counts continue to increase through adulthood. This aligns with the concept of protracted neurodevelopmental maturation [21]. These findings indicate that while the anatomical coordinates for dual-transmitter circuits are genetically hardwired early, the full acquisition of the dual-phenotype across the broader neuronal population may be driven by subsequent, potentially activity-dependent, postnatal processes [22].

A central novelty of our study is the discovery of profound heterochrony in the developmental onset of the dual-phenotype. We identified regions like the GPi that host stable dual-transmitter populations at birth (P0), in stark contrast to regions like the MeApv where this phenotype is acquired progressively postnatally. This anatomical divergence likely reflects the distinct ecological and functional demands of these neural circuits. The GPi is a critical output node of the basal ganglia, regulating fundamental motor and survival-related behaviors that require early functional competence [23]. Conversely, the MeApv is primarily involved in processing complex olfactory, reproductive, and social cues—behaviors that mature significantly later during juvenile and peripubertal development [18]. Thus, the temporal onset of the Vglut2/Vgat phenotype might correlate with the maturation of its resident circuit.

The phenomenon of “neurotransmitter switching”—where neurons transiently express one transmitter before switching to another in response to activity—has been demonstrated in the developing brainstem and spinal cord [20]. This raised the critical question of whether our genetic labeling merely captured a historical transit state. By employing RNAscope to validate concurrent transcription at multiple developmental stages, we explored the temporal co-expression in a limited number of brain regions. In the adult spinal cord, we observed no Vglut2/Vgat colocalization, suggesting complete temporal switching or segregation. However, in the GPi and MeApv across different ages, we detected robust, real-time colocalization. These findings indicate that, at least in certain brain regions, dual-transmitter identities during development represent stable, terminal developmental destinations rather than transient developmental intermediates.

The early anatomical presence and projection of these neurons, such as the innervation of the LH by P7, carry profound physiological implications. During early postnatal development, the high intracellular chloride concentration in immature neurons renders GABAergic transmission depolarizing rather than inhibitory [24]. In this context, the co-release of glutamate and GABA may provide a synergistic, amplified excitatory signal essential for robust synaptogenesis and the stabilization of early network wiring [25]. As the circuits mature and GABA shifts toward an inhibitory role, this dual-transmitter input would then provide a critical inhibitory brake. Notably, the LH is characterized by a striking scarcity of intrinsic inhibitory interneurons, making it highly dependent on extrinsic GABAergic inputs for its functional regulation [10, 26].

While our brain-wide mapping provides a critical structural foundation, certain limitations must be acknowledged. First, we observed an increase in the number of labeled neurons over time. However, it remains uncertain whether this reflects a true expansion of Vglut2/Vgat dual-positive neurons. Two factors complicate this interpretation: (1) a maturation lag in fluorescence expression could lead to cumulative signal at later time points, making neurons more easily detectable; (2) the observed increase may simply parallel the general developmental rise in neuronal numbers, making it difficult to determine whether dual-transmitter phenotypes undergo substantive expansion during development. Second, our RNAscope analyses were conducted only in a subset of the genetically labeled brain regions. Expanding these analyses in additional brain regions will be essential to determine the prevalence of neurotransmitter co-existence versus developmental switching. Third, to clarify whether dual-transmitter identities represent stable phenotypes or transitional states, studies employing high-throughput single-cell RNA sequencing (scRNA-seq) combined with RNA velocity analysis will be vital to reveal the machinery underlying the dual-transmitter phenotype [27]. Fourth, while co-localization in adults has been shown to generally indicate co-release, it remains to be determined whether molecular co-localization during development also translates to functional co-release of both neurotransmitters [4–6, 11]. Electrophysiological recordings of postsynaptic responses will be necessary to confirm whether these neurons release both glutamate and GABA in vivo. Fifth, resolving the precise synaptic physiology—specifically, whether glutamate and GABA are co-packaged in the same vesicles or released from spatially segregated terminals will be essential to decode the functional grammar of these dual-transmitter circuits across development. Finally, the sample size in our study was limited to three animals per group, which may have precluded the detection of subtle or moderate effects. Future investigations with larger cohorts will be necessary to enhance statistical power and reveal more nuanced developmental changes.

In summary, our study provides a comprehensive whole-brain atlas illustrating that during development, individual neurons express molecular components associated with both excitatory (Vglut2) and inhibitory (Vgat) neurotransmitter release. We also present evidence for the temporal coexistence of such dual-transmitter phenotypes in the specific brain regions we examined. Furthermore, we reveal that the onset or occurrence of this dual-phenotype expression varies across different brain areas. These findings contribute to a more complete understanding of the organizational principles and functional implications of neurotransmitter coexistence in the brain.

Materials and methods

Animals and genetic crosses

To selectively label and map dual glutamatergic-GABAergic neurons, we generated a triple-transgenic reporter mouse line using an intersectional genetic strategy. Specifically, Slc17a6-IRES2-FlpO-D (Vglut2-IRES2-FLPo-D)(JAX:030212) and B6;129S-Gt(ROSA)26Sor^tm65.1(CAG-tdTomato)Hze/J (Ai65-D) (JAX:021875) and Slc32a1^tm2(cre)Lowl/J (Vgat-ires-cre) (JAX:016962). Driver mice were crossed with the dual-recombinase dependent fluorescent reporter line Ai65 (B6;129S-Gt(ROSA)26Sor^tm65.1(CAG-tdTomato)Hze^/J) to generate Vglut2-Flp;Vgat-Cre;Ai65 offspring. All experimental offspring were heterozygous for the transgenes, and were derived from heterozygous intercrosses. All transgenic lines were maintained on a C57BL/6 genetic background. Both male and female mice were used in all experiments. Animals were group-housed in a specific pathogen-free (SPF) facility under a standard 12-h light/12-h dark cycle, with ad libitum access to food and water. All animal care, breeding, and experimental procedures were rigorously conducted in accordance with the Guide for the Care and Use of Laboratory Animals and were strictly approved by the Institutional Animal Care and Use Committees (IACUC) at Anhui Medical University and the Songjiang Research Institute of Shanghai Jiao Tong University School of Medicine.

Tissue preparation and fluorescence imaging

For the morphological mapping of tdTomato-expressing cells across different developmental stages, stage-specific tissue preparation protocols were employed. Adult, postnatal day 21 (P21), and P14 mice were deeply anesthetized and transcardially perfused with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in PBS. Extracted brains were post-fixed overnight in 4% PFA at 4 °C and subsequently cryoprotected in 30% sucrose/PBS at 4 °C for 2–3 days until they sank. For early postnatal stages (P0 and P7), mice were transcardially perfused with 5–10 mL of filtered 4% PFA. The delicate brains were carefully dissected, post-fixed overnight in 4% PFA at 4 °C, and cryoprotected in 30% sucrose/PBS at 4 °C for 1–2 days. Cryoprotected brain tissues were embedded in optimal cutting temperature (OCT) compound and sectioned into 50–60 μm thick coronal slices using a cryostat. Whole-brain overview images were acquired using an Olympus VS200 slide scanner equipped with a 10 × objective to capture macroscopic distribution patterns. High-resolution anatomical details were subsequently captured using an Olympus spinning disk confocal microscope.

Single-molecule fluorescent in situ hybridization (RNAscope)

To rigorously validate the concurrent transcription of excitatory and inhibitory markers within individual cells, single-molecule fluorescent in situ hybridization was performed using the RNAscope Multiplex Fluorescent v2 Assay (Advanced Cell Diagnostics, ACD). To maximize RNA integrity, a fresh-frozen tissue protocol was utilized without prior transcardial perfusion. Mice were rapidly euthanized, and the brains were immediately dissected. The brains were placed in embedding molds and completely covered with optimal cutting temperature (OCT) compound. Samples were rapidly frozen by immersion in dry ice–precooled isopentane to preserve tissue morphology and molecular integrity. The frozen tissues were stored at − 80 °C until further processing. Coronal cryosections (14 μm thickness) were mounted onto SuperFrost Plus slides (Fisher Scientific) and stored at − 80 °C until use. Prior to hybridization, sections were immediately fixed in pre-chilled 4% PFA for 15 min at 4 °C, followed by a sequential dehydration series in 50%, 70%, and 100% ethanol (5 min each at room temperature). Treatment with RNAscope Protease was skipped. Subsequently, sections were hybridized with target-specific probes against mouse Slc17a6 (Vglut2; ACD probe target region) and Slc32a1 (Vgat; ACD probe target region) and tdTomato (Cat No. 317041) in a HybEZ oven for 2 h at 40 °C. Following a series of sequential signal amplification steps (Amp 1–3), specific transcript signals were developed utilizing Opal 647 and Opal 520 and Opal 570 fluorophores (Akoya Biosciences). High-resolution images of the RNA transcripts were acquired using a confocal microscope.

Brain region alignment and neuronal quantification

Regional alignment was performed using the second edition of the mouse brain atlas, together with developmental brain atlases at postnatal day 0 (P0) and postnatal day 6 (P6) [28, 29]. The Allen Brain Atlas was then employed to identify and define the corresponding brain regions. Coronal sections were aligned to the brain atlas, with their anterior–posterior positions recorded as millimeter coordinates relative to a bregma point. Within this standardized framework, double-positive neurons in specific brain regions were manually quantified.

Statistical analyses

A sample size of three mice per group was used. In the fluorescence labeling experiments, for each brain region at a given time point, four coronal sections from each mouse were examined. To ensure data stability and representative sampling of the entire nucleus, the four sections were anatomically matched across animals, and were evenly spaced along the anterior–posterior axis. For RNAscope experiments, neurons were quantified from 3 mice, with 4–6 representative sections per animal. Marker-positive neurons were manually counted across these sections and averaged to provide a single value per biological replicate. Data are presented as mean ± standard deviation (SD), with individual data points in the figures representing values from individual mice or slices (see figure legends for specific details). Statistical analyses were conducted using MATLAB (version R2025b). For the developmental quantification data presented in Table 1, statistical significance between each developmental stage (P0, P7, P14, and P21) and the adult group was determined using unpaired Student’s t-tests. For all tests, P < 0.05 was considered statistically significant. Samples marked as “NA” were excluded from analysis due to technical issues with tissue quality.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MeA

Medial amygdalar nucleus

AHNP

Anterior hypothalamic nucleus, posterior part

GPi

Globus pallidus internus

LH

Lateral habenular

VTA

Ventral tegmental area

SUM

Supramammillary nucleus

EW

Edinger-Westphal nucleus

RL

Rostral linear nucleus raphe

SCs

Superior colliculus, sensory related

SCop

Superior colliculus,optic layer

SCig

Superior colliculus,motor relater intermediate

SCiw

Superior colliculus, motor relater intermediate, white layer

CSm

Superior central nucleus raphe medial part

NI

Nucleus incentus

PRNC

Pontine reticular nucleus, caudal part

Mlf

Medial longitudinal fascicle

CENT3gr

LoublueШ, granular layer

Crus1

Ancr1

IRN

Intermediate reticular nucleus

MDRNd

Medullary reticular nucleus, dorsal part

UvUmo

UvUla(Ⅸ) molecular layer

PH

Posterior hypothalamic nucleus

RCH

Retrochiasmatic area

DMH

Dorsomedial nucleus of the hypothalamus

ICe

Inferior colliculus,external nucleus

LGv

Ventral part of the lateral geniculate complex

PAG

Periaqueductal gray

LDG

Dorsal part of the lateral geniculate complex

VTD

Ventral tegmental decussation

VMH

Ventromedial hypothalamic nucleus

MRN

Midbrain reticular nucleus

PRNr

Pontine reticular nucleus

ZI

Zona incerta

InCo

Intercollicular nucleus

CIC

Inferior colliculus commissure

PBLV

Parabrachial nucleus lateral division ventral lateral part

Crus2

Ancr2

SUV

Superior vestibular nucleus

LHA

Lateral habenular area

ML

Medial lemniscus

DR

Dorsal nucleus raphe

VCO

Ventral cochleal nucleus

PYR

Pyramus molecular layer

MDRNd

Medullary reticular

RO

Nucleus raphe obscurus

MEApV

Medial amygdalar nucleus,posteroventral part

Author contributions

H.W. and Y.B. devised the project, the main conceptual ideas, and the proof outline. H.W., C.Z., S.C., and H.Z. designed and performed the experiments, and collected data. X.Z. performed experiments. H.W., Y.B., and C.Z. wrote the manuscript. All authors discussed the results and provided feedback on the manuscript.

Funding

This work was supported by the Startup Fund for Scientific Research of Songjiang Research Institute of Shanghai Jiao Tong University School of Medicine to H.W.; National Natural Science Foundation Incubation Program of The Second Affiliated Hospital of Anhui Medical University (2021GQFY02) to Y.B.

Data availability

The data that support the findings of this study are available on request to the corresponding author H.W. (whshsj@shsmu.edu.cn).

Declarations

Competing interests

The authors declare no competing interests.