Molecular and cellular evolution of the amygdala across species analyzed by single-nucleus transcriptome profiling

Yu, Bin; Zhang, Qianqian; Lin, Lin; Zhou, Xin; Ma, Wenji; Wen, Shaonan; Li, Chunyue; Wang, Wei; Wu, Qian; Wang, Xiaoqun; Li, Xiao-Ming

doi:10.1038/s41421-022-00506-y

Download PDF

Article
Open access
Published: 14 February 2023

Molecular and cellular evolution of the amygdala across species analyzed by single-nucleus transcriptome profiling

Bin Yu^1,2^na1,
Qianqian Zhang^3,4^na1,
Lin Lin^1,2^na1,
Xin Zhou⁵,
Wenji Ma³,
Shaonan Wen⁵,
Chunyue Li^1,2,
Wei Wang³,
Qian Wu^5,6,
Xiaoqun Wang ORCID: orcid.org/0000-0003-3440-2617^3,4,5,6,7 &
…
Xiao-Ming Li ORCID: orcid.org/0000-0002-8617-1702^1,2,8

Cell Discovery volume 9, Article number: 19 (2023) Cite this article

9981 Accesses
5 Citations
30 Altmetric
Metrics details

Subjects

Abstract

The amygdala, or an amygdala-like structure, is found in the brains of all vertebrates and plays a critical role in survival and reproduction. However, the cellular architecture of the amygdala and how it has evolved remain elusive. Here, we generated single-nucleus RNA-sequencing data for more than 200,000 cells in the amygdala of humans, macaques, mice, and chickens. Abundant neuronal cell types from different amygdala subnuclei were identified in all datasets. Cross-species analysis revealed that inhibitory neurons and inhibitory neuron-enriched subnuclei of the amygdala were well-conserved in cellular composition and marker gene expression, whereas excitatory neuron-enriched subnuclei were relatively divergent. Furthermore, LAMP5⁺ interneurons were much more abundant in primates, while DRD2⁺ inhibitory neurons and LAMP5⁺SATB2⁺ excitatory neurons were dominant in the human central amygdalar nucleus (CEA) and basolateral amygdalar complex (BLA), respectively. We also identified CEA-like neurons and their species-specific distribution patterns in chickens. This study highlights the extreme cell-type diversity in the amygdala and reveals the conservation and divergence of cell types and gene expression patterns across species that may contribute to species-specific adaptations.

Cellular development and evolution of the mammalian cerebellum

Article Open access 29 November 2023

Mari Sepp, Kevin Leiss, … Henrik Kaessmann

A lamprey neural cell type atlas illuminates the origins of the vertebrate brain

Article Open access 14 September 2023

Francesco Lamanna, Francisca Hervas-Sotomayor, … Henrik Kaessmann

Comparative single-cell transcriptomic analysis of primate brains highlights human-specific regulatory evolution

Article Open access 04 September 2023

Hamsini Suresh, Megan Crow, … Jesse Gillis

Introduction

The amygdala is responsible for animal survival due to its involvement in reproduction and fear/escape responses^1,2,3,4. It also plays a critical role in emotional valence, memory encoding, feeding, social behaviors, and behavioral state assessment^5,6. Damages or functional disruptions of the amygdala have been implicated in various neurological diseases, especially neuropsychiatric disorders, such as schizophrenia, anxiety, and bipolar disorder^7,8,9. The amygdala is a heterogeneous complex that includes multiple subnuclei and neuronal cell types originating from cortical and subcortical territories¹⁰. In a landmark study¹¹, Johnston suggested that the amygdala was composed of cortical-like and striatal-like components. Studies over the last few years have demonstrated that the amygdala is far more complex than previously thought. In addition to cortical and striatal (lateral ganglionic eminence; LGE) derivatives, the amygdala also contains neuronal cell types derived from the medial ganglionic eminence (MGE), caudal ganglionic eminence (CGE), preoptic area (POA), and even supraoptoparaventricular hypothalamic domain (SPV)^12,13,14. Given its multiple embryonic origins, the amygdala is a mosaic-like structure with an abundance of neuronal cell types. However, our understanding of the cellular composition and gene expression patterns of the amygdala remains limited, though two hundred years have passed since¹⁵ first described this almond-shaped mass of gray matter. Notably, the lack of cell-type-specific studies has been a major obstacle to dissecting amygdala function and understanding amygdala-related disorders.

Although the amygdala is highly complex in terms of neuronal subtypes, molecular profiles, and connections, growing evidence suggests that primary amygdala subnuclei and basic circuit connections and functions are conserved across species, especially in mammals and sauropsids^14,16,17,18. They share a common pattern of organization and the same basic combinations of developmental regulatory genes that orchestrate similar functions, although species differences do exist. The relative size and complexity of the amygdala subnuclei differ among species^19,20,21. The paralaminar nucleus (PL) of the amygdala is relatively enlarged in humans and non-human primates compared to other mammals²². Localization of amygdala subnuclei in sauropsids remains controversial¹⁷. However, the extent to which evolution has changed cell types and gene expression patterns of the amygdala across species is still unclear.

Recent advances in single-cell and single-nucleus RNA-sequencing (scRNA-seq/snRNA-seq) technologies have facilitated the molecular characterization of diverse cell types in different brain regions across species^{23,24,25,26,27}. Although single-cell profiling of the amygdala in different species has been reported, the number of neurons previously captured in the human amygdala was too small to allow effective clustering and in-depth analysis of neurons²⁸, amygdala subnuclei, which varied greatly in cell types and physiological functions were not distinguished in scRNA-seq/snRNA-seq in primates^28,29,30, and inconsistent dissection of amygdala regions also hindered cross-species comparison^30,31. To ensure the consistency of dissection regions, obtain high quality and unbiased sequencing data, and explore the evolutionary conservation and divergence of the amygdala, we applied single-nucleus transcriptomic methods for profiling human, macaque, mouse, and chicken amygdala homologs.

Results

Taxonomy of cell types in human, macaque, and mouse amygdala

To characterize cell types of the amygdala in mammals and sauropsids (the most diverse and successful terrestrial vertebrates) at the transcriptomic level, we applied snRNA-seq assays (Chromium v3) using human, macaque, mouse, and chicken samples (Fig. 1a). Human amygdala samples were obtained from postmortem donors within a very short postmortem interval (within 6 h), and macaque, mouse, and chicken amygdala samples were obtained from freshly dissected specimens. We profiled a total of 17 specimens, including three human amygdala samples, three macaque amygdala samples, four mouse amygdala samples, three mouse BLA (containing lateral (LA), basal (BA), basomedial (BM), and posterior (PA) nuclei) samples, two mouse CEA samples, and two chicken amygdala homolog samples (Supplementary Table S1). After quality control and filtering, we obtained 203,971 high-quality single nuclei from the four species. 91,699 nuclei with a median of 9417 unique molecular identifiers (UMIs) and 3617 genes were from the human dataset, 45,626 nuclei with a median of 4631 UMIs and 2373 genes were from the macaque dataset, 54,186 nuclei with a median of 5020 UMIs and 2316 genes were from the mouse dataset, and 12,460 nuclei with a median of 3660 UMIs and 1716 genes were from the chicken dataset (Supplementary Fig. S1a, b and Table S1).

**Fig. 1: Profiling of amygdala cell types across mammals by snRNA-seq.**

To evaluate conservation and divergence across species, we first integrated amygdala RNA-seq data from mammals using Seurat (v4.0.5). Integration revealed broad conservation of cell types across mammals, as nuclei from different species were well mixed (Fig. 1b). Seven cell classes were recognized based on the expression of known cell class markers: i.e., excitatory neurons (ExN), inhibitory neurons (InN), oligodendrocyte precursor cells (OPC), oligodendrocytes (Oligo), astrocytes (Astro), endothelial cells/pericytes (Endo/Peri), and microglia (Micro) (Fig. 1b, c; Supplementary Fig. S1c). Analysis of differentially expressed genes (DEGs) in each cell class revealed shared DEGs across species, although many showed enriched expression in only one species (Supplementary Fig. S1d). Interestingly, ExN shared a lower percentage of DEGs across species than InN, and non-neuronal cells shared a greater percentage of DEGs across species than neuronal cells (Supplementary Fig. S1d), implying that neurons, especially ExN, were evolutionarily more diverse. We also identified many other ExN and InN markers in the mammalian amygdala in addition to SLC17A7, GAD1, and GAD2, such as C1QL3 for ExN, and ERBB4 and IGF1 for InN (Supplementary Fig. S1e). Remarkably, more marker genes were identified in the InN that were conserved among species than in the ExN, again suggesting higher ExN diversity in the amygdala (Supplementary Fig. S1e).

An unprecedented number of mammalian amygdala neuronal and non-neuronal cells were captured in our datasets. Strikingly, the vast majority were neuronal cells (52,650 nuclei in the human dataset, 32,525 nuclei in the macaque dataset, and 43,654 nuclei in the mouse dataset), with ExN and InN accounting for about 50% each (Fig. 1d; Supplementary Fig. S1f). In contrast, InN constitute ~15%–30% of neuronal cells in the neocortex of different species²⁶, consistent with the diversity of InN in the amygdala. Notably, primates possessed slightly more ExN, while mice possessed more InN, in agreement with the expanded ExN-enriched BLA structure in primates²¹. In addition, consistent with white matter expansion during primate evolution³², we found that primates, especially humans, contained more oligodendrocyte precursor cells and oligodendrocytes (Supplementary Fig. S1f).

Unsupervised clustering identified 45, 38, and 39 distinct transcriptomic cell types in the human, macaque, and mouse amygdala, respectively (Fig. 1e–g; Supplementary Fig. S2 and Table S2). We assigned each cluster a name based on the marker gene showing the most specific expression in that cluster (Fig. 1e–g; Supplementary Fig. S2). In most clusters, two genes were used to define a cell type. Some cluster names were manually adjusted based on knowledge for consistency of nomenclature across species. We found that some widely reported cell types in the mouse amygdala were captured in our mouse dataset. For example, the Prkcd, Sst Tac2, and Drd2 clusters, which likely originated from the CEA^33,34,35, and the Tshz1 clusters, which likely originated from the intercalated amygdalar nucleus (IA)³⁶, occupied the same branches on the dendrogram (Supplementary Fig. S2). The Npsr1 Rspo2 and Hgf Grp clusters probably correspond to the Rspo2⁺ and Grp⁺ neurons reported previously^37,38. Notably, assumed homologous cell types, e.g., the PRKCD, DRD2, and TSHZ1 clusters, were also identified in the human and macaque datasets and were clustered together on the dendrogram (Supplementary Fig. S2), indicating general conservation across species.

Evolutionary conservation and divergence of amygdala subnuclei in mammals

The amygdala consists of multiple subnuclei with various neuronal cell types originating from different cortical and subcortical territories. Anatomically, the amygdala subnuclei are well defined in mammals (Fig. 2a). To determine subnuclear localization of each neuronal cell type, we first applied in situ hybridization (ISH) data from the Allen Brian Atlas (ABA)³⁹ and performed RNAscope multiplex fluorescent ISH (mFISH) to validate the expression patterns of the most specific marker genes of each cluster in the mouse amygdala (Fig. 2b; Supplementary Fig. S3a). Furthermore, using coordinated information and expression levels of genes in the ABA, we further analyzed the mean expression level and patterns of the top five marker genes in each cluster in the mouse amygdala (Fig. 2c; Supplementary Fig. S3b). Finally, we preformed enrichment analysis to identify BLA- or CEA-enriched clusters based on the anatomical origin of each neuronal nucleus (Supplementary Fig. S3c). Analysis clearly showed that neurons in each cluster exhibited spatial distribution specificity and were mainly confined in one subnucleus of the amygdala. Based on the expression patterns of marker genes, cluster dendrograms, and knowledge, we divided the mouse neuronal cell types into six groups: BLA-derived excitatory neurons (BLA group), CEA-derived inhibitory neurons (CEA group), IA-derived inhibitory neurons (IA group), cortical amygdalar area/medial amygdalar nucleus-derived excitatory and inhibitory neurons (COA/MEA group), amygdala cortical interneuron-like inhibitory neurons (Amygdala CI-like group), and nucleus of the lateral olfactory tract-derived excitatory neurons (NLOT group). To assign cell group information to human and macaque neuronal cell types, we used Seurat label transfer to predict the group origin of every neuronal cell based on transcriptomic similarity with the mouse dataset (Supplementary Fig. S3d, e). Except for the NLOT group neurons in the macaque dataset, we captured all subnucleus-derived cell types in the mammalian amygdala after prediction and manual adjustment, including primate-specific PL-derived cell types (PL group), which show high expression of immature marker genes²⁹. As NLOT group neurons were not captured in our macaque dataset and given the continued debate over whether the NLOT is part of the amygdala, the NLOT was not included in further analyses. Remarkably, nuclei from the same cell group were clustered together in the uniform manifold approximation and projection (UMAP) plots, indicating similarity in gene expression between neurons from the same cell group (Fig. 2d; Supplementary Fig. S3d, e).

Hierarchical clustering of cell groups and cell types revealed general conservation of the amygdala across mammals, although the COA/MEA group neurons showed relatively high divergence (Fig. 2e; Supplementary Fig. S4b). The CEA group, IA group, and Amygdala CI-like group neurons showed similar proportions and diversity in the amygdala across species. However, the proportion of BLA group neurons was approximately two-fold higher in the primate amygdala than in the mouse amygdala, and more cell types were captured in primates. Conversely, mice exhibited a higher proportion and diversity of COA/MEA group neurons (Fig. 2f; Supplementary Fig. S4a). These results are consistent with the different relative sizes and cell densities of the BLA and COA/MEA across species, which may reflect species-specific adaptations²¹. We next identified cell group-level markers. Several marker genes were conserved across species, including some transcription factors and ion channel-encoding genes, but most markers were enriched in only one species (Fig. 2g; Supplementary Fig. S4c and Table S3). Consistent with the hierarchical clustering results, the COA/MEA group had fewer conserved marker genes (Fig. 2g). Furthermore, specifically activated regulons in each cell group were identified, which may be important for subnuclear specialization during development (Fig. 2h). For example, NEUROD2 is a neuronal basic helix-loop-helix transcription factor that is necessary for neurogenic differentiation and synaptic maturation^40,41. Previous studies have shown that Neurod2 is essential for LA and BA formation during development, as well as normal amygdala function such as emotional learning⁴². Our results indicated that NEUROD2 regulon activity was enriched in the BLA excitatory neurons, further underscoring the importance of NEUROD2 and related regulatory genes for BLA formation and function. Many major regulon genes, such as LHX2, TBR1, PAX6, FOXP1, MEIS1, MEIS2, and FOXP2, have been implicated in subnuclear specialization and normal physiological function of the amygdala^14,43,44, thus demonstrating the reliability of our regulon analysis. Importantly, we identified many novel major genes and regulons that may play critical roles in the specialization of specific cell groups (Fig. 2h).

To evaluate differences in evolutionary conservation of amygdala subnuclei, we performed unsupervised MetaNeighbor analysis. Results showed that the IA and CEA group neurons were the most evolutionarily conserved, while the BLA and COA/MEA group neurons were relatively evolutionarily divergent (Fig. 2i). Consistent with the MetaNeighbor results, analysis of the expression patterns of orthologous genes across species demonstrated that the BLA and COA/MEA groups had the greatest divergent expression (Supplementary Fig. S4d). In addition, fewer divergent genes were identified between primates than between humans and mice, suggesting that amygdala homology decreases with evolutionary distance (Supplementary Fig. S4e).

It should be noted that humans and macaques contained a neuronal cluster with high expression of immature markers, such as SOX11 and BCL2, which were likely derived from the PL^22,29. The PL is expanded in primates, containing numerous immature neurons even in adults, and is believed to play an important role in neuroplasticity in the adult brain²². PAGA developmental trajectory analysis indicated that the maturation trajectory of the PL-derived cluster in humans and macaques may be towards the VGLL3 and TFAP2C clusters (Fig. 2j, k; Supplementary Fig. S4f–i).

Homology of inhibitory neurons in mammals

Abundant inhibitory neurons were enriched in the CEA, IA, and COA/MEA (mainly MEA), while CI-like inhibitory neurons were sparsely distributed in the amygdala. To further assess neuronal homology in the amygdala, we integrated the inhibitory neuron datasets of the mammals. Inhibitory neurons were well mixed across species and same cell group-derived nuclei were clustered together in the UMAP plots (Fig. 3a; Supplementary Fig. S5a). We identified 12 major cell types that were well aligned across species, except for one major cell type whose nuclei were predominately derived from the mouse MEA (named Mix-cluster) (Fig. 3b). Consistent with the motor cortex²⁶, the expression of several marker genes in each major cell type was conserved across species, but most marker genes were enriched in only one species (Fig. 3c; Supplementary Table S4). We next identified 22 homologous cell types at appropriate resolution (Fig. 3b). Cluster membership shared among species, hierarchical clustering of homologous cell types, and MetaNeighbor analysis of cell types from humans, macaques, and mice further revealed cross-species conservation of inhibitory neurons, although differences across species did exist (Fig. 3d, e; Supplementary Fig. S5b, c). Inhibitory neuron homology between humans and macaques was higher than that between humans and mice, reflecting differences in evolutionary distance (Supplementary Fig. S5d).

CEA- and IA-derived inhibitory neurons are interesting because their LGE origin is distinct from other amygdala inhibitory neurons and cortical interneurons, which are mainly derived from the MGE and CGE, constituting the amygdala’s main output and inhibitory control center^14,45. Previous studies have illustrated that the CEA and IA represent the most primitive and evolutionarily conserved entities in the amygdala, and likely existed in the last common ancestor of tetrapods^14,17. The evolutionary conservation of these subnuclei is crucial for animal survival because they are involved in mediating many autonomic and endocrine responses in fear and anxiety behaviors^5,46. Our snRNA-seq datasets confirmed the conservation of CEA and IA subnuclei at the cell-type level in mammals. Notably, the IA subnuclei were highly conserved, and all mammals in our datasets contained two types of TSHZ1⁺ neurons, which matched one-to-one between species (Fig. 3d). Studies have shown that IA-derived inhibitory neurons highly express DRD1 and are densely innervated by dopaminergic afferents, providing feedforward inhibition of the BLA and CEA^47,48. Interestingly, in our datasets, only one type of TSHZ1 cluster highly expressed DRD1, which was consistent across species (Fig. 3f). Therefore, we speculated that there were two types of neurons in the IA, i.e., Tshz1⁺ Drd1⁺ and Tshz1⁺ Drd1⁻ neurons. This speculation was confirmed by mFISH analysis, which showed that both types of neurons were present in the main and paracapsular islands of the IA, and Tshz1⁺ Drd1⁻ neurons tended to be distributed in the IA periphery. Moreover, Tshz1⁺ Drd1⁻ neurons primarily resided in the main island of the IA, while Tshz1⁺ Drd1⁺ neurons dominated in the medial paracapsular island of the IA (Fig. 3g, h). Overall, mFISH results showed two IA cell types in mice, one expressing Drd1 and the other not, which were probably well-conserved in mammals.

Despite broad conservation, differences among species were also observed. LAMP5⁺ neurons were classified into the CI-like group and likely shared the same derivation as cortical LAMP5⁺ interneurons. In accordance with recent observations that LAMP5⁺ interneurons are expanded throughout the neocortex in primates^25,45, the proportion of LAMP5⁺ inhibitory neurons was approximately two to five times higher in the primate amygdala than in the mouse amygdala (Supplementary Fig. S5e–g). The expansion of LAMP5⁺ inhibitory neurons may be a common feature of the cerebral cortex and cerebral nuclei, and its evolutionary implications need to be further explored. DRD2⁺ neurons were much more abundant in humans than in macaques and mice and were segregated into two cell types (Supplementary Fig. S5e–g and Table S5) expressing PAX6 and ISL1, respectively (Supplementary Fig. S5h). The former is a marker of the dorsal LGE, and the latter is a marker of the ventral LGE⁴⁹. Gene Ontology (GO) analysis implied that the two types of DRD2⁺ neurons may regulate different physiological functions and behaviors through different signaling pathways (Supplementary Fig. S5i). Given the importance of DRD2 and the dopamine system in human sociality and psychiatric disorders^50,51 and specific increased abundance of DRD2⁺ neurons in the human amygdala, it is important to study the function of DRD2⁺ neurons in the human amygdala directly.

To identify both evolutionarily conserved and species-specific sets of co-regulated genes across cell types, we performed Coordinated Gene Activity in Pattern Sets (CoGAPS) and transfer-learning analyses (projectR). CoGAPS analysis allowed unsupervised identification of patterns of gene sets (Patterns) that represented common features across cell types, and projectR was used to project the snRNA-seq expression matrix into the CoGAPS pattern amplitude matrix. In general, broadly similar cellular expression patterns in similar cell types were identified across species (Supplementary Fig. S6a and Table S6). For example, “Pattern-66” was strongly correlated with IA (TSHZ1) clusters in all datasets (Supplementary Fig. S6b), and many IA regulators and marker genes were among the top 30 weighted genes in “Pattern-66”, including FOXP2, NRG1, PBX3, MEIS2, and ERBB4 (Supplementary Fig. S6c)^44,52. However, many Patterns showed high correlations among incongruent cell types across species, which may contribute to species-specific functions.

Shared and distinct signatures of BLA excitatory neurons in mammals

The BLA is the largest subnucleus in the amygdala, contains the most excitatory neurons, and is extensively connected to cortical and subcortical areas, especially primary sensory areas. Hence, the BLA is thought to receive information inputs and transmit decoded information to output subnuclei of the amygdala or other downstream brain areas⁵. Evolutionarily, the relative enlargement of the BLA compared with other subnuclei suggests it may have undergone the most significant evolutionary changes^19,21. To assess BLA homology, we integrated the mammalian BLA-derived excitatory neurons and identified 14 cross-species cell types (Fig. 4a, b; Supplementary Fig. S7a). The BLA excitatory neurons showed higher interspecies heterogeneity than the inhibitory neurons, as reflected by the integrated data having more undefinable cell types and more species-dominant cell types (Fig. 4c; Supplementary Fig. S7b).

Despite variation among species, homologous major cell types were also identified in the BLA excitatory neurons. MetaNeighbor analysis and proportion of overlapping nuclei between clusters in the integrated datasets indicated that SATB2, HGF, and VGLL3 neuronal cell types were highly conserved in mammals (Fig. 4d; Supplementary Fig. S7c). Among them, VGLL3 cells were the most conserved BLA excitatory neurons, with at least one corresponding cell type found in each species (Fig. 4d; Supplementary Fig. S7c). CoGAPS analysis further identified highly correlated gene set Patterns in homologous cell types between species (Supplementary Fig. S7d, e and Table S7). Pattern-83 was strongly correlated with VGLL3⁺ cell types in all three mammalian datasets (Supplementary Fig. S7e). We also detected average expression of the top 30 weighted genes in Pattern-83 in human, macaque, and mouse BLA excitatory neuronal cell types, and found that gene expression was indeed enriched in the VGLL3⁺ neuronal cell types in each species (Supplementary Fig. S7f). Notably, ESR1 was one of the most influential genes in Pattern-83 (Supplementary Fig. S7f). Esr1 is highly expressed in the posterior amygdalar nucleus (PA) in the mouse brain, consistent with the expression pattern of the Vgll3 Car12 cluster marker gene Car12 (Supplementary Fig. S7g). Esr1⁺ neurons in the PA play key roles in both male aggression and reproduction by targeting the medial preoptic nucleus (MPN) and ventrolateral part of the ventromedial hypothalamus (VMHvl), respectively⁵³. Esr1 encodes estrogen receptor alpha, and estrogen signaling in other brain regions mediates instinctive social behaviors such as aggression, defense, and maternal behaviors^54,55,56. Therefore, we speculate that VGLL3⁺ neurons are important for instinctive behaviors and conservation of these high ESR1-expressing clusters across species may have been critical for animal survival and reproduction during evolution.

Broad evolutionary differences were observed in our integrated BLA excitatory neuronal datasets compared to the conserved subnuclei enriched by inhibitory neurons. Here, LAMP5⁺ and SATB2⁺ excitatory neurons were much more abundant in the human BLA (Fig. 4e, f). LAMP5⁺ excitatory neurons have also been captured in the human neocortex⁴⁵. Furthermore, as a marker of superficial layer excitatory neurons, SATB2 is also widely expressed in the neocortex⁵⁷. Interestingly, the LAMP5⁺ and SATB2⁺ excitatory neurons showed high expression of the lateral and dorsal pallial marker EMX1 but not of the ventral pallial marker LHX9 (Fig. 4g). To confirm the presence and spatial distribution of LAMP5⁺ and SATB2⁺ excitatory neurons in the amygdala, we performed RNAscope mFISH analysis. Results showed that a large number of SATB2⁺ neurons resided in the LA portion of the human BLA, but few Satb2⁺ neurons were observed in the mouse LA (Fig. 4h, i), consistent with our integrated sequencing data showing that SATB2-2 neurons were mainly derived from humans (Fig. 4f). On the other hand, LAMP5⁺ excitatory neurons (SCL17A7⁺) were widely distributed in both the human and mouse BLA, including the LA, BA, and BM subregions. In humans, however, the proportion of excitatory neurons expressing LAMP5 was higher in the LA, but similar in the BA and BM (Fig. 4j, k; Supplementary Fig. S8). These results highlight the differences in the LA across mammals, consistent with the evolutionary increase in LA neuronal populations in humans²⁰. Thus, we hypothesize that increased LAMP5⁺ and SATB2⁺ excitatory neurons in the human BLA, especially the LA, may be due to the substantial increase in the size of the neocortex in the human brain as well as the increase in lateral or dorsal pallial-derived excitatory neuronal (EMX1 positive) migration to the amygdala.

To explore the possible physiological functions of SATB2⁺ and LAMP5⁺ excitatory neurons in the human BLA, we examined differences in gene expression levels across species in these cell types. First, we found that many Patterns in CoGAPS analysis were strongly correlated with LAMP5 or SATB2 cell types in the integrated datasets (Supplementary Fig. S7h, i). Next, we detected the expression patterns of these Pattern genes, many of which were highly and specifically expressed in human cells due to the dominance of human cells in these integrated cell types. Interestingly, many Pattern genes correlated with the two neuronal clusters were associated with psychiatric disorders (Fig. 4l). For example, ADAMTSL3 and VAV3 are considered candidate genes for schizophrenia^58,59, and LTBP1 and FKBP5 are associated with depressive features in humans^60,61. The high expression of genes associated with psychiatric disorders implies the potential role of these two clusters in emotional processing in humans. Moreover, we checked the significant biological enrichment GO terms for pattern genes in C0 (LAMP5) and C8 (SATB2-2). As expected, some of the significant GO terms were related to learning or memory, cognition, and long-term depression (Supplementary Fig. S7j, k), suggesting these human-dominant clusters were relevant to the human superior intelligence such as cognition and learning.

Diversity and evolution of chicken amygdala cell types

The mammalian amygdala is generally conserved in structure and cellular composition. However, the evolutionary conservation and differences between the mammalian amygdala and that of the evolutionarily distant sauropsids are still unclear. The precise identification of amygdala subnuclei in sauropsids remains controversial due to the obvious differences in brain organization between mammals and sauropsids. Recent studies suggest that the caudal part of the dorsal ventricular ridge (DVR), which is present in all sauropsids, and the surrounding subpallial regions are homologous counterparts of the mammalian amygdala^17,62,63. To characterize sauropsid amygdala homologs at the single-cell level and compare evolutionary conservation and divergence with the mammalian amygdala, we dissected the chicken caudal DVR (including the arcopallium (APall) and caudal nidopallium (NPall)) and extended amygdala (EA) for snRNA-seq (Fig. 5a; Supplementary Fig. S9a)⁶⁴. Seven classes and 46 cell types were defined in the chicken amygdala homologs (Fig. 5b, c; Supplementary Fig. S9b, c). Hierarchical clustering revealed general conservation of cell classes across species (Fig. 5d). However, obvious differences were observed when all cell types from the mouse and chicken datasets were analyzed, reflected in the tendency of cell types derived from the same species to cluster on the same hierarchical tree branch (Supplementary Fig. S9d). We then performed cross-species expression correlation analysis and found that expression was more consistent between mammals than between mammals and chickens (Fig. 5e), emphasizing the differences among animal classes. On the other hand, chicken cell types showed more similarity with mice than with humans and macaques (Fig. 5f).

**Fig. 5: Broad distribution of LGE-derived inhibitory neurons in pallial amygdala in chicken.**

We also explored the possible subnuclear origin of chicken cell types based on transcriptomic similarity. Seurat and SingleR methods were applied, showing consistent subnuclear origin predictions for inhibitory neurons, but relatively divergent predictions for excitatory neurons (Fig. 5g). Remarkably, the assumed CEA and IA chicken cell types expressed similar marker genes as those in mice. For example, DRD2 and PRKCD were highly expressed in predicted CEA cells and TSHZ1 was highly expressed in predicted IA cells. In addition, the LHX8 CALCR cluster likely originated from the MEA of chicken amygdala (Fig. 5c, g). These results strongly suggest that inhibitory neuron-enriched subnuclei are highly conserved in amniotes. Recent genetic and migration studies have ascertained that the EA in chickens corresponds to the capsular part of the CEA in mammals^65,66. To verify this, we detected PRKCD and DRD2 expression patterns in chicken amygdala homologs. The mFISH analysis results showed that numerous PRKCD⁺ DRD2⁺ and PRKCD⁺ DRD2⁻ neurons resided in the EA area, consistent with the expression patterns in our snRNA-seq data (Fig. 5h). Our findings confirmed that the EA is at least part of the CEA at the single-cell level and contains cell types with similar gene expression patterns as mammals. However, PRKCD⁺ neurons were not only located in the EA, but also broadly distributed in the APall and NPall (Fig. 5h). Given that the PRKCD and DRD2 clusters were LGE-derived neurons showing high expression of LGE marker genes (Supplementary Fig. S9e), this observation is consistent with recent research showing that LGE-derived interneurons are broadly distributed in bird brains⁶². Furthermore, our results showed that chickens contained more LGE-derived neurons, while mammals possessed more MGE/CGE-derived neurons in the amygdala (Fig. 5i, j). Based on the above observations, we hypothesize that amygdala subnuclear specialization in chickens may be incomplete due to the widespread distribution of LGE-derived interneurons. On the other hand, the dramatic increase in MGE/CGE-derived neurons in mammals may be a driving force for specialization of amygdala subnuclei, replacing the function of LGE-derived neurons in the pallial amygdala during evolution (Fig. 5k). Although the distribution of PRKCD and DRD2 neurons differed between mammals and chickens, we found that chicken PRKCD and DRD2 neurons also expressed most genes that were conserved in corresponding mammalian clusters (Fig. 3c; Supplementary Fig. S9f and Table S4). This suggests that PRKCD and DRD2 neurons are well-conserved at the gene expression level, even in distantly related species such as the chicken. To explore possible functional differences in these neurons between mammals and chickens, we compared all DEGs and found many genes that were up- and down-regulated in mammalian cells (Supplementary Fig. S9g and Table S8). Interestingly, many of the up-regulated genes are involved in mental conditions and higher cognitive functions, including the schizophrenia- and autism-associated gene SYN2 and cognition- and intelligence-associated genes FMN2, CTNND2, and CACNA1A (Supplementary Fig. S9g)^{67,68,69,70,71}. These up-regulated genes imply that, due to the abundance of MGE/CGE-derived neurons in the mammalian amygdala, LGE-derived PRKCD and DRD2 neurons may be confined to the CEA region and specialized for new functions such as higher cognition and complex emotional regulation in mammals.

To explore potential subnuclear origins of chicken excitatory neurons, correlation analysis between expression profiles was performed, showing no perfect one-to-one correspondence between the mouse and chicken excitatory neurons. However, three populations were clearly visible (Supplementary Fig. S10a). The first corresponded to mouse clusters with high SLC17A7 expression, the second corresponded to mouse clusters with high SLC17A6 expression, and the third corresponded to mouse NLOT neurons with high STRIP2 expression. SLC17A7⁺ neurons were mainly located in the BLA and SLC17A6⁺ neurons were mainly located in the LA, BM, COA, and MEA, corresponding to lateral or dorsal pallial-derived neurons and ventral pallial-derived neurons, respectively^14,17. Chicken excitatory neurons were primarily from the caudal DVR, traditionally considered as the ventral pallium in chickens. These results suggest that the caudal DVR may contain many lateral or dorsal pallial-derived neurons, or that convergent evolution may have occurred in the chicken DVR.

Another striking difference between the chickens and mammals was the extreme diversity of non-neuronal cell types in chickens (Fig. 5c; Supplementary Fig. S10b). We captured many rare non-neuronal cell types not captured in mammals, including astrocyte progenitors and developing oligodendrocytes (Supplementary Fig. S10c–f). High non-neuronal activity in the chicken amygdala may be important for neuroplasticity in behavior, but its physiological significance should be further explored.

Discussion

The amygdala is of particular interest in comparative analysis given its involvement in emotional processing, higher cognitive functions, and various mental disorders^2,72,73. Here, we established a comprehensive single-nucleus transcriptomic atlas of the amygdala in humans, macaques, mice, and chickens. Our data revealed general evolutionary conservation of the amygdala across species. The CEA and IA were particularly conserved in mammals, with similar structures and cell types even identified in the chicken. The BLA and COA/MEA were relatively divergent during evolution. We also identified several primate-dominant and human-dominant cell types. LAMP5⁺ interneurons were much more abundant in primates, and DRD2⁺ inhibitory neurons and LAMP5⁺/SATB2⁺ excitatory neurons were enriched in the human CEA and BLA, respectively.

Inhibitory neuronal cell types were much more conserved than excitatory neuronal cell types in the datasets, consistent with that found in other brain regions^23,62. Many clusters of inhibitory neuronal cells were matched one-to-one between species. The IA showed the highest conservation, and all mammals contained two types of TSHZ1⁺ neurons, i.e., DRD1⁺ and DRD1⁻. IA neurons are densely innervated by dopaminergic afferents and have been implicated in fear and anxiety behaviors⁴⁶. Recent studies have shown that distinct clusters of intercalated neurons exist in the IA, which exert diametrically opposed roles through mutual synaptic inhibition and access functionally distinct downstream targets^48,74, similar to fear on and fear off neurons in the CEA^33,75. Thus, it would be interesting to investigate whether functionally distinct clusters of intercalated neurons correspond to two types of IA neurons in our transcriptomic data in the future. Notably, our data showed that chickens contained numerous TSHZ1⁺ clusters with high FOXP2 expression. The Seurat and SingleR algorithm predicted one TSHZ1⁺ cluster as CEA neurons and four clusters as IA neurons. Due to the high expression of FOXP2 in the avian basal ganglia⁷⁶, we could not rule out the possibility that these TSHZ1⁺ clusters were of basal ganglia origin, and thus deeper characterization is required. In addition to TSHZ1⁺ clusters, the DRD2⁺ clusters also showed conservation across species, although were much more abundant in the human amygdala. Remarkably, the DRD2⁺ neurons could also be subclustered into two subtypes, which may have originated from different parts of the LGE based on PAX6 and ISL1 expression (Supplementary Fig. S6d)¹⁴. However, further studies are required to confirm this hypothesis and determine the specific role of each subtype.

Our results support LGE-derived CEA and IA as the most ancient and conserved subnuclei in the amygdala^14,17. Despite their strong conservation, species–specific specializations were also observed. The mouse CEA contained a cluster, which we named Isl1, that was not captured in humans or macaques. Considering the abundance of the cluster in mice, this result is unlikely to be a sampling issue. In addition to the CEA, mice showed higher cell-type diversity in the COA/MEA, which receive olfactory and vomeronasal inputs, consistent with the relatively larger volume of the COA/MEA and stronger olfactory/vomeronasal system in rodents²¹. These results reflect species-specific adaptations to the environment. The reduction in relative volume and cell-type diversity in the primate COA/MEA suggests cell-type loss during evolution.

Our results showed that the BLA contained most of the excitatory neurons in the amygdala and exhibited extensive evolutionary differences across species, including significantly higher number and diversity of BLA excitatory neuronal clusters in primates. The BLA is part of the pallial amygdala and is proposed to originate from the ventral and lateral pallial sectors based on transcription factor expression and fate mapping⁷⁷. Recent evidence suggests that the dorsal pallium, which contains the neocortex in mammals, also contributes to the cellular composition of the amygdala⁷⁸. Notably, in our study, the SATB2⁺ and LAMP5⁺ excitatory clusters, which were dominant in the human BLA, demonstrated high expression of the dorsal and lateral pallium marker EMX1. SATB2 is also a marker of neocortical excitatory neurons⁵⁷, and LAMP5⁺ excitatory neuronal clusters are reported in the human neocortex²⁶. Thus, we propose that more and more neocortical-derived neurons migrated to the amygdala following expansion of the neocortex during higher primate evolution, and this evolutionary gain may play an important role in the integration of signal inputs and higher cognitive functions.

Although abundant cell types were identified in the amygdala of humans, macaques, mice, and chickens, we anticipate that more cell types will be found with deeper sampling, especially in humans and macaques, given the absolute volume of the amygdala across species. Indeed, we identified a Sst⁺ cluster in the mouse CEA through snRNA-seq profiling of the CEA directly (Supplementary Fig. S4a, b), which may correspond to previously reported Sst⁺ neurons in the mouse CEA³⁴. However, no corresponding SST⁺ neurons from the CEA were captured in the human or macaque datasets, indicating that capture of rare cell types may require additional sampling. Moreover, the CEA Sst⁺ cluster could be segregated into three subclusters (Supplementary Fig. S11), indicating that cell types in our datasets could be further subclustered and more cell types could be identified.

Single-cell transcriptome analysis, especially in cross-species analysis, relies on efficient data integration. In our study, we used the BBKNN algorithm method, which can effectively combat intra-species batch effects, as shown in Supplementary Fig. S1b. In the process of integrating datasets across species, we chose widely used Seurat workflow⁷⁹ with some adjustments. As shown in Supplementary Fig. S5a and Fig. 4a, most cells were well mixed across species except for a few species-dominant cell types. However, different datasets may require different batch-corrected methods, since too strong batch-corrected methods may eliminate biologically meaningful inter-/intra-species differences. Extensive comparison of the impact of different batch-corrected methods on the integrated datasets will facilitate the discovery of more intra-species cell types and more conserved or species-specific cell types across species in the future.

Materials and methods

Sample collection and ethical compliance

Postmortem adult human brain tissue samples were obtained from the National Human Brain Bank for Health and Disease in Hangzhou (NHBBHD), China. Written informed consent for the use of the tissues in research was obtained from donors or their families. All postmortem brain samples were from donors with no obvious brain disease based on pathological diagnosis by the NHBBHD. Postmortem tissue collection was approved by the ethics committees of NHBBHD and Zhejiang University. Three fresh human amygdala tissues (Supplementary Table S1) were collected for snRNA-seq, and one 10% formalin-fixed human amygdala tissue sample was collected for RNAscope multiplex fluorescence in situ hybridization (mFISH).

Rhesus macaque (Macaca mulatta) experiments were conducted in accordance with the Principles for the Ethical Treatment of Non-Human Primates and were approved by the Institutional Animal Care and Use Committee of the Institute of Biophysics, Chinese Academy of Sciences, and the Animal Advisory Committee at Zhejiang University. For snRNA-seq, three adult and senile macaques (Supplementary Table S1) were deeply anesthetized with ketamine (10 mg/kg) and perfused with cold artificial cerebrospinal fluid (ACSF) after the pedal withdrawal reflex was eliminated. Whole brains were rapidly extracted and placed on a precooled metal plate. The macaque amygdala was dissected within 30 min according to anatomical landmarks, such as optic tract, hippocampus, and ventrolateral temporal neocortex. Fresh tissues were snap-frozen in liquid nitrogen and stored at −80 °C until use.

All domestic chicken (Gallus gallus) experiments were reviewed and approved by the Animal Advisory Committee at Zhejiang University. For snRNA-seq, two adult chickens (Supplementary Table S1) were sacrificed by CO₂ asphyxiation, and whole brains were rapidly extracted into oxygenated ice-cold sucrose-based ACSF containing: 25 mM sucrose, 10 mM D-glucose, 24 mM NaHCO₃, 3 mM KCl, 5 mM MgCl₂, 2 mM CaCl₂, 1.25 mM NaH₂PO₄, and 2 mM MgSO₄. Coronal slice sectioning was performed as reported previously⁸⁰. Briefly, the forebrain was mounted on a steel plate, and 300-μm-thick coronal sections were cut in the cold, sucrose-based ACSF with a vibratome slicer (VT1200s, Leica Biosystems, Nussloch, Germany). Carbogen (95% O₂, 5% CO₂) was bubbled continuously during sectioning. Amygdala homologs in the caudal ventrolateral part of the forebrain, including the arcopallium, most of the caudal nidopallium, and extended amygdala, were dissected under a dissection microscope (SMZ645, Nikon). Dissected tissue was snap-frozen in liquid nitrogen and stored at −80 °C before further processing. For RNAscope mFISH analysis, one adult male chicken was sacrificed by CO₂ asphyxiation, and whole brains were rapidly extracted into 4% paraformaldehyde (PFA) for fixation.

All procedures involving C57BL/6 J mice (Jackson Labs, #000664) were approved by the Animal Advisory Committee at Zhejiang University. For snRNA-seq, eight-week-old male and female mice (Supplementary Table S1) were deeply anesthetized with 1% sodium pentobarbital (0.1 g/kg body weight) and transcardially perfused with ice-cold oxygenated N-methyl-D-glucamine (NMDG)-based ACSF containing: 93 mM NMDG, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 30 mM NaHCO₃, 20 mM HEPES, 25 mM glucose, 5 mM sodium ascorbate, 2 mM thiourea, 3 mM sodium pyruvate, 10 mM MgSO₄.7H₂O, and 0.5 mM CaCl₂.2H₂O. Mice were rapidly decapitated and whole brains were extracted into ice-cold oxygenated NMDG-based ACSF, with 300-μm-thick coronal sections then cut in the cold, NMDG-based ACSF with a vibratome slicer (VT1200s, Leica Biosystems). Carbogen (95% O₂, 5% CO₂) was bubbled continuously during sectioning. The whole amygdala, basolateral amygdalar complex (BLA), and central amygdalar nucleus (CEA) were rapidly dissected under a dissection microscope from different sections, snap-frozen in liquid nitrogen, and stored at −80 °C. For RNAscope mFISH, adult male C57BL/6 J mice were anesthetized with 1% sodium pentobarbital (0.1 g/kg body weight) and transcardially perfused with 1× phosphate-buffered saline (PBS), followed by 4% PFA. Brains were extracted into 4% PFA for post fixation.

Isolation and purification of nuclei

For human specimens, frozen tissues were placed on a precooled metal plate and cut into 1–2-mm-thick coronal sections using a cold scalpel according to spatial coordinates. For donor CBB052, each section was used for nuclei isolation. For donors CBB037 and CBB038, only one typical intermediate coronal section was used for nuclei isolation. Entire macaque amygdala samples and dissected mouse and chicken samples were utilized for nuclei isolation. Nuclei were isolated and purified following the 10× Genomics protocols (CG000393 • Rev A), with some modification. Briefly, the frozen tissues were homogenized on ice in lysis buffer containing: 10 mM Tris-HCl, 10 mM NaCl, 3 mM MgCl₂, 0.01% NP-40 (ThermoFisher), 1 mM β-mercaptoethanol, and 0.2 U/μL RNase inhibitor (Promega) by a glass homogenizer. Tissues were dounced until there were no visible clumps. Homogenates were incubated on ice for 5 min, and an equal volume of HEB medium (Hibernate E/B27/GlutaMAX, Gibco) was added to stop lysis. Cell debris and large clumps were removed using 30-μm (mouse and chicken) and 40-μm (human and macaque) filters (pluriSelect), and the supernatant was centrifuged at 500× g and 4 °C for 5 min. To further remove debris, Myelin Removal Beads II (Miltenyi Biotec, mouse tissues) and Debris Removal Solution (Miltenyi Biotec, human, macaque, and chicken tissues) were applied according to the manufacturer’s instructions. Nuclei were washed 1–2 times and resuspended in wash and resuspension buffer containing 1× PBS, 1% bovine serum albumin (BSA, Sigma), and 0.2 U/μL RNase inhibitor, then collected by centrifugation at 500× g and 4 °C for 5 min. DAPI DNA dye (ThermoFisher) was added and nuclei were counted manually under a fluorescence microscope (BX53, Olympus). The nuclear suspension was diluted to a concentration of 500–1000 nuclei/μL for loading onto the 10× Chromium instrument.

SnRNA-seq library preparation and sequencing

Nuclear suspension loading onto the Chromium Next GEM Chip, nuclear barcoding, cDNA amplification, and library construction were carried out according to the 10× Genomics protocols (CG000204 Rev D). Libraries were sequenced on the Illumina NovaSeq 6000 System at LC-Bio Technology Co., Ltd. in Hangzhou, China.

SnRNA-seq data processing

All raw snRNA-seq data are accessible from the Gene Expression Omnibus database (GEO: GSE195445). The snRNA-seq data constructed using 10× Genomics were processed through the Cell Ranger Single-Cell Software Suite (v3.1.0)⁸¹, which performed sequencing read alignment and gene expression quantification. The sequencing reads from humans were aligned to the GRCh38 human reference genome, while the mouse, macaque, and chicken sequencing reads were aligned to the mm10-3.0.0, Mmul_10.102, and GRCg6a species reference genomes, respectively. We used the CellRanger aggr pipeline to merge cells from the same species under different samples, with the parameter “normalize = none”. We applied quality control procedures to cells in each species step by step, first computing the highest density interval for the number of genes detected (“n_genes_by_counts”) and the percentage of mitochondrial genes expressed (“pct_counts_mt”) using the hdi function in the HDInterval R package with the parameter “credMass = 0.98”. Cells outside the confidence interval were filtered. We then removed potential doublets reported by Scrublet⁸², with default parameters. After strict quality control analysis, we retained 91,699 nuclei with a median of 9417 unique molecular identifiers (UMIs) and 3617 genes in the human dataset; 45,626 nuclei with a median of 4631 UMIs and 2373 genes in the macaque dataset; 54,186 nuclei with a median of 5020 UMIs and 2316 genes in the mouse dataset; and 12,460 nuclei with a median of 3660 UMIs and 1716 genes in the chicken dataset. We then used the normalize_total and log1p functions in SCANPY (v1.8.2)⁸³ to normalize (library-size correct) and logarithmize the raw count matrix with default parameters. Next, we detected highly variable genes (HVGs) using the scanpy.pp.highly_varible_genes function with the parameters “min_mean = 0.0125, max_mean = 3, min_disp = 0.5” and selected all HVGs for downstream analysis. Finally, we used the scanpy.pp.regress_out function to regress the effects of total counts and mitochondrial gene expression percentage.

Dimensionality reduction and unsupervised clustering of single-species datasets

We employed the SCANPY⁸³ workflow to reduce dimensionality and embed the neighborhood graph on the gene expression matrix. We first performed principal component analysis (PCA) using the scanpy.tl.pca (svd_solver = ‘arpack’, n_comps = 100) function to obtain a PC matrix with 100 components. We then used the bbknn function with the parameters “batch_key = ‘sample’, n_pcs = 75” to align the batch effects from samples and computed the neighborhood graph. For embedding the graph in two dimensions, we performed UMAP analysis using the scanpy.tl.umap function. Unsupervised clustering was realized using the sc.tl.leiden function with appropriate resolution parameters adapted to the diverse datasets. The parameters were set to: “resolution = 1.5” for mice, “resolution = 1.5” for humans, “resolution = 2.8” for macaques, and “resolution = 2.2” for chickens. Cluster-specific DEGs were identified using the scanpy.tl.rank_genes_groups function with the Wilcoxon test. For clarity, we unified the naming of cell populations in the diverse conditions as follows: clusters from unsupervised clustering were named “Cell Clusters”, cell populations containing multiple clusters were named “Cell Classes”, and amygdala subnuclear distributions were named “Cell Groups”. Cell classes in each species dataset were annotated using canonical markers of cell types, as shown in Fig. 1c.

Integrating snRNA-seq datasets across species

To identify homologous cell types across species, we performed three rounds of dataset integration: Mammal Datasets (all nuclei from mammals), Inhibitory Neuronal Datasets (inhibitory neurons from mammals), and BLA Excitatory Neuronal Datasets (nuclei from BLA cell group in mammals). We employed the Seurat (v4.0.5)⁷⁹ workflow to integrate datasets across diverse species. The vignette of SeuratDisk (https://github.com/mojaveazure/seurat-disk) was used to convert AnnData objects to Seurat objects. The BiomaRt⁸⁴ package was applied to reduce raw matrices to only orthologous genes (13,478 genes) with one-to-one orthologues defined in humans, macaques, and mice. The SelectIntegrationFeatures function in Seurat was used to select repeatedly variable features across datasets. We then applied the FindIntegrationAnchors function to identify cross-dataset pairs of anchors, and the IntegrateData function was used to create an integrated assay using pre-computed anchors. The integration procedures used default parameters, except “dim=1:50”. The integrated matrix (batch-corrected) was used for dimensionality reduction and unsupervised clustering with the RunUMAP and FindClusters functions, respectively. We set parameters to “resolution = 0.35” and “resolution = 0.25” when the FindClusters function was used for the Inhibitory Neuronal Datasets and BLA Excitatory Neuronal Datasets, respectively. For Mammal Datasets and BLA Excitatory Neuronal Datasets integration, each Seurat object from different species was normalized using the SCTransform function in Seurat. To integrate the Inhibitory Neuronal Datasets, we combined the top 100 DEGs from each cluster within each species as features for the FindIntegrationAnchors function. All adjustments were attempts to eliminate species-specific batch effects.

Spatial mapping of amygdala subnuclei

We mapped the mouse snRNA-seq gene expression profile to the mouse ISH atlas of the Allen Institute for Brain Science (http://mouse.brain-map.org) at the cluster level⁸⁵. The ISH data of each slice were summarized into voxel dataset, providing the gene expression profile of each voxel (Gene expression value X voxel). The brain structures file of each slice showed the anatomical annotation of each voxel (voxel X structures). First, we generated whole brain plots with only amygdala subnuclei annotation in each slice using brain structures file to select the proper coronal slice as reference image. We chose slice 37 to plot reference image, where we divided the amygdala into five subnuclei: BLA (Basolateral amygdalar complex), IA (Intercalated amygdalar nucleus), CEA (Central amygdalar nucleus), MEA (Medial amygdalar nucleus), and COA (Cortical amygdalar area). Then, we downloaded the ISH data of slice 37 to map gene expression on the reference image. We calculated the average expression value of the top 5 DEGs in each cluster and then, projected it on the reference image to infer the subnuclei distribution of clusters. (Only DEGs that appeared in the ISH coronal gene expression profile were retained).

Cell group prediction

Seurat label transfer

We first annotated clusters (Cell Group labels) in mice based on ISH data from the Allen Brain Atlas (https://portal.brain-map.org/) and Ro/e (ratio of observed to expected number of nuclei) distribution analysis, as shown in Supplementary Fig. S3. We quantified the fraction of clusters from each amygdala cell group based on transcriptomic similarity with mouse datasets. Seurat supports the projection of a list of reference labels (i.e., Cell Group labels in mice) onto a query object (i.e., human/macaque/chicken objects). We used the FindIntegrationAnchors function with the parameter “dims = 1:50” to identify anchors between the reference and query datasets, then used these anchors with the IntegrateData function to classify query cells based on the reference labels. We added “predicted.id” to the metadata of the Seurat object for data visualization and showed cell group ID on the UMAP plot. We calculated the proportion of cells from different cell groups in each cluster and annotated clusters with the highest proportion.

SingleR analysis

SingleR⁸⁶ is an effective tool to infer the development origin of each cell. Here, we used SingleR to transfer mouse cell group labels to the human/macaque/chicken datasets. We used the pairwiseTTests and getTopMarkers functions in the R package scran to identify the top markers for each pairwise group of cells⁸⁷. We performed SingleR procedures with default parameters and visualized “predicted.id” on the UMAP plot.

Cluster dendrograms

For clusters of each species, cluster dendrograms were established based on hierarchical clustering across excitatory neurons, inhibitory neurons, and non-neuronal cells. Average expression of HVGs was calculated for each cluster, and hierarchical clustering was performed using the hclust R function. The dendrogram branches were reordered manually without changing the tree structure.

Cell class DEG expression patterns across species

We selected the top 500 DEGs of each cell class across species. Venn diagrams were generated using the VennDiagram R package to visualize DEG intersections across species for each given Cell Class cluster.

To show the consistency of expression patterns in neurons across species, we combined the top 100 DEGs of excitatory and inhibitory neurons from every two species (human-macaque, human-mouse) and calculated the log₂FC value for each gene between classes of excitatory and inhibitory neurons. We defined genes with log2FC > 1 in mice or macaques and log2FC > 1 in humans as “ExN_sig”, and log2FC < (−1) in mice or macaques and log2FC < (−1) in humans as “InN_sig”. Expression patterns of neuronal DEGs were presented in a scatter plot.

MetaNeighbor analysis

To compare the conservation and divergence of co-expression in randomly defined gene sets, we performed unsupervised MetaNeighbor analysis on cross-species integrated datasets⁸⁸. We first used the get_variable_genes function to select the top quartile of variable genes. Cross-species comparisons at different levels (Cluster/Cell Group/Cell Class levels) were performed using the run_MetaNeighbor function. The output area under the receiver operator characteristic curve (AUROC) scores of each pair of labels was plotted as a heat map. We removed cells from the nucleus of the lateral olfactory tract (NLOT) subnuclei to analyze amygdala cell group characteristics as their definition was not clear enough in the amygdala.

We calculated the total AUROC scores in the box plot using the AUROC scores from the original clusters of all species in Supplementary Fig. S4b, which showed the accumulation of species-dependent expression differences in the amygdala cell groups.

To visualize the relationships between homologous excitatory neuronal clusters across chickens and mice, the AUROC scores of selected clusters were taken and loaded into Cytoscape⁸⁹ to draw a network graph, with the connection edge between two cell-type clusters displayed when AUROC > 0.75. We performed the same analysis on inhibitory neuronal clusters across all four species (humans, macaques, mice, and chickens) and visualized the network using the R package igraph (edges displayed when AUROC > 0.8).

Calculating DEGs in integrated datasets

To identify conserved and species-enriched DEGs in different cell types across species, we used the FindAllMarkers function in Seurat for all cell types in each species. Positive DEGs were reported using the ROC test with default parameters. To avoid computational memory issues, we down-sampled the Seurat object to a small number of cells (200 cells/species cell group in the Mammal Datasets and 80 cells/species cluster in the Inhibitory Neuronal Datasets). We first calculated the DEGs of cell types separately in each species. In DEGs of each cell type, conserved DEGs were defined as DEGs present in both species, and species-enriched DEGs were defined as DEGs only expressed in one species. The DoHeatmap function in Seurat was used to plot the heat map of DEGs with small volume cells (as mentioned above). When we performed DEG analysis of the Mammal Datasets across cell groups, we removed the “PL (paralaminar nucleus)” and “NLOT” groups as these two group-derived cells were only captured in two species.

Cell group expression divergence

For each pair of homologous cell groups between humans and macaques or humans and mice, we calculated the average expression of each orthologous gene, with scatter plot visualization after log-transformation. We calculated divergent genes in each cell group between pairs of species (human-mouse/human-macaque) using the limma R package⁹⁰. We identified significant divergent genes based on log₂FC > 1 and adjusted P-value < 0.01 across paired species. Pearson correlations were calculated to evaluate the relationships between humans and macaques or humans and mice using the cor.test function. Average expression of all divergent genes was calculated and visualized in a heat map with the parameters “cluster_columns = T, cluster_rows = T”.

SCENIC analysis

We used the SCENIC⁹¹ workflow to analyze activated regulons in each amygdala cell group. Raw count matrices from the single species were input to infer a co-expression network using the GRNBoost2 function. The RcisTarget R package was used to identify transcription factor binding motifs, and the AUCell R package was used to quantify regulon activity in single cells. We used the wilcox.test function to determine the differentially activated regulons in each cell group and the p.adjust function was used to correct multiple hypotheses. We selected and plotted (heat map) the top five regulons intersecting the three species.

PAGA developmental trajectory analysis

We reconstructed the PL (paralaminar nucleus) nucleus-derived cell development trajectories using the PAGA package⁸³ for the human and macaque datasets. The scanpy.tl.paga function was used for PAGA graph computation and non-significant edges were filtered out based on connectivity > 0.25. Edge connectivity between the C36 cluster (in macaques) and C24 cluster (in humans) with all other cluster nodes was visualized in the bar plot. We recomputed embedding based on PAGA initialization and calculated diffusion pseudotime using the scanpy.tl.dpt function (root cell set as a cell in macaque C36 cluster or human C24 cluster). We showed the PAGA trajectories using a set of genes.

Homologous cell types across species

We defined consensus clusters by comparing the overlap of species-original clusters in cell types of integrated Seurat objects (reported by cca algorithm) between humans and macaques and between humans and mice, as described previously⁴⁵. For each cluster pair, we calculated overlap values, i.e., the sum of the minimum proportion of nuclei in each original cluster that overlapped with each cell type. Cluster overlap values were scaled to 0–1 and visualized in a heat map.

Ro/e analysis

To quantify species distribution of clusters, we calculated the ratio of observed to expected number of nuclei (Ro/e) for each cell type cluster across species using the chi-square test^92,93, with higher Ro/e values indicating a greater degree of enrichment in a species.

We also used Ro/e analysis to quantify cell group distribution of clusters in the Inhibitory Neuronal Datasets, and tissue distribution of clusters in the mouse dataset. The Ro/e ratio for each cluster was calculated across Cell Group or dissection origin.

Enrichment analysis

We applied the enricher function in the clusterProfiler R package⁹⁴ for gene set enrichment analysis. Gene sets of “c5 category” from the MsigDB dataset (http://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) were loaded using the R package msigdf. Adjusted P-values and gene counts for enriched pathways were plotted (bar plot) and networks of the selected gene sets were visualized using the cneplot function in the R package enrichplot. To quantify gene set scores across clusters, we calculated the average expression of genes from a given gene set in a single cell.

We used the GSVA R package⁹⁵ to analyze gene set variation and the limma package⁹⁰ to calculate the T-value of differentially enriched pathways.

CoGAPS analysis

We used CoGAPS⁹⁶ to construct a latent space learning graph based on single-cell gene expression matrices, following previous research^97,98. As a nonnegative matrix factorization (NMF) algorithm, CoGAPS factorizes the gene expression matrix into two matrices: i.e., Pattern (cell × pattern) and Amplitude matrices (genes × pattern). Here, we performed NMF for each species step by step. First, we split the normalized gene expression of each species from integrated analysis and reduced the expression matrix into a matrix containing only 1000 HVGs (reported using the FindVariableFeatures function in Seurat). We then used the CoGAPS function with the parameters “nPatterns = 150, nIterations = 500, sparseOptimization = TRUE, seed = 77, nSets = 15” to obtain a CogapsResult object.

To determine the biological performance of patterns across different species, we used the R package projectR (https://github.com/genesofeve/projectR/) to project a new gene expression matrix onto the CoGAPS Pattern matrices. Human CoGAPS amplitude matrices were input as source data, while macaque and mouse gene expression matrices were input as target data. We calculated correlation of the Pattern matrix with the global features matrix (i.e., cluster labels) using the cor function and visualized these pattern correlations in a dot plot (for clarity, only patterns with correlation values > 0.1 were plotted). We selected the top 30 most heavily weighted genes from a given pattern and visualized scaled average expression across clusters in each species on the heat map.

Topic modeling

We used the FitGoM function in the CountClust R package to fit latent Dirichlet allocation topic models to the raw count matrix^99,100. Specifically, we filtered out genes with zero expression from the raw matrix prior to fitting the topic models. We then ran the FitGoM function with the parameter “k = 25” to obtain an object containing two matrices: Factors and Loadings. We added the loading weights of each cell to the metadata of the Seurat object and drew the topic loading weights in the UMAP coordinates. Finally, to interpret the biological processes of the selected topics, we used the diff_count_analysis function with default parameters to analyze DEGs among topics. We performed enrichment analysis of selected topic DEGs (P-value < 0.01, log₂FC > 2) and gene sets from the MsigDB datasets “c8 category” were used as source data. The log₂FC values (reported by volcano_plot function) of genes from selected pathways were plotted in a bar plot.

RNAscope mFISH analysis

Formalin-fixed human amygdala tissue and PFA-fixed chicken and mouse brains were embedded in optimal cutting temperature medium (OCT, Sakura), frozen with dry ice, and stored at −80 °C. The frozen tissues were sectioned (15–20 μm) onto Superfrost Plus glass slides (Epredia). Sections were dried at −20 °C and stored at −80 °C until use. RNAscope mFISH experiments were performed as manufacturer’s instruction for fixed-frozen tissue samples (Advanced Cell Diagnostics), but with the following modifications: (1) protease treatment was lengthened to 40 min when using human tissues; (2) TrueBlack (Biotium) was used to quench autofluorescence in human and chicken tissues after the RNAscope experiment. An Olympus FV3000 fluorescent microscope was used for imaging (10×, 20×, and 40× air lenses and 60× oil immersion lens), and image processing was carried out using Image J. Positive cells were counted manually, containing ≥ five RNA spots for that gene. Lipofuscin autofluorescence was distinguished based on the broad fluorescence spectrum of lipofuscin.

Quantification and statistical analysis

Details of the statistical analyses used and sample size (n) for each analysis were present within the figure legends. Statistical significance was determined using GraphPad Prism and R software, unless otherwise noted. For statistical comparisons, P-values below 0.05 were considered statistically significant.

Data availability

The datasets generated during this study have been deposited in the Gene Expression Omnibus public database under accession number GSE195445.

References

LeDoux, J. E. Emotion: clues from the brain. Annu. Rev. Psychol. 46, 209–235 (1995).
Article CAS PubMed Google Scholar
LeDoux, J. E. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184 (2000).
Article CAS PubMed Google Scholar
Risold, P. Y., Thompson, R. H. & Swanson, L. W. The structural organization of connections between hypothalamus and cerebral cortex. Brain Res. Rev. 24, 197–254 (1997).
Article CAS PubMed Google Scholar
Choi, G. B. et al. Lhx6 delineates a pathway mediating innate reproductive behaviors from the amygdala to the hypothalamus. Neuron 46, 647–660 (2005).
Article CAS PubMed Google Scholar
Janak, P. H. & Tye, K. M. From circuits to behaviour in the amygdala. Nature 517, 284–292 (2015).
Article CAS PubMed PubMed Central Google Scholar
Grundemann, J. et al. Amygdala ensembles encode behavioral states. Science 364, eaav8736 (2019).
Article PubMed Google Scholar
Anand, A. & Shekhar, A. Brain imaging studies in mood and anxiety disorders - Special emphasis on the amygdala. Ann. N. Y. Acad. Sci. 985, 370–388 (2003).
Aleman, A. & Kahn, R. S. Strange feelings: Do amygdala abnormalities dysregulate the emotional brain in schizophrenia? Prog. Neurobiol. 77, 283–298 (2005).
PubMed Google Scholar
Bigot, M. et al. An emotional-response model of bipolar disorders integrating recent findings on amygdala circuits. Neurosci. Biobehav. Rev. 118, 358–366 (2020).
Article PubMed Google Scholar
Swanson, L. W. & Petrovich, G. D. What is the amygdala? Trends Neurosci. 21, 323–331 (1998).
Article CAS PubMed Google Scholar
Johnston, J. B. Further contributions to the study of the evolution of the forebrain. J. Comp. Neurol. 35, 337–481 (1923).
Article Google Scholar
Puelles, L. et al. Pallial and subpallial derivatives in the embryonic chick and mouse telencephalon, traced by the expression of the genes Dlx-2, Emx-1, Nkx-2.1, Pax-6, and Tbr-1. J. Comp. Neurol. 424, 409–438 (2000).
Article CAS PubMed Google Scholar
Nery, S., Fishell, G. & Corbin, J. G. The caudal ganglionic eminence is a source of distinct cortical and subcortical cell populations. Nat. Neurosci. 5, 1279–1287 (2002).
Article CAS PubMed Google Scholar
Medina, L., Bupesh, M. & Abellan, A. Contribution of genoarchitecture to understanding forebrain evolution and development, with particular emphasis on the amygdala. Brain Behav. Evol. 78, 216–236 (2011).
Article PubMed Google Scholar
Burdach K. F. Vom Baue Und Leben Des Gehirns, Volume 3. German Edn (Nabu Press, 1923).
Martinez-Garcia, F., Martinez-Marcos, A. & Lanuza, E. The pallial amygdala of amniote vertebrates: evolution of the concept, evolution of the structure. Brain Res. Bull. 57, 463–469 (2002).
Article PubMed Google Scholar
Moreno, N. & Gonzalez, A. Evolution of the amygdaloid complex in vertebrates, with special reference to the anamnio-amniotic transition. J. Anat. 211, 151–163 (2007).
Article PubMed PubMed Central Google Scholar
Martinez-Garcia, F., Novejarque, A. & Lanuza, E. Two interconnected functional systems in the amygdala of amniote vertebrates. Brain Res. Bull. 75, 206–213 (2008).
Article PubMed Google Scholar
Chareyron, L. J., Banta Lavenex, P., Amaral, D. G. & Lavenex, P. Stereological analysis of the rat and monkey amygdala. J. Comp. Neurol. 519, 3218–3239 (2011).
Article PubMed PubMed Central Google Scholar
Barger, N. et al. Neuronal populations in the basolateral nuclei of the amygdala are differentially increased in humans compared with apes: A stereological study. J. Comp. Neurol. 520, 3035–3054 (2012).
Article PubMed PubMed Central Google Scholar
Rowniak, M. & Bogus-Nowakowska, K. The amygdala of the common shrew, guinea pig, rabbit, fox and pig: five flavours of the mammalian amygdala as a consequence of clade-specific mosaic-like evolution. J. Anat. 236, 891–905 (2020).
Article PubMed PubMed Central Google Scholar
deCampo, D. M. & Fudge, J. L. Where and what is the paralaminar nucleus? A review on a unique and frequently overlooked area of the primate amygdala. Neurosci. Biobehav. Rev. 36, 520–535 (2012).
Article PubMed Google Scholar
Tosches, M. A. et al. Evolution of pallium, hippocampus, and cortical cell types revealed by single-cell transcriptomics in reptiles. Science 360, 881–888 (2018).
Article CAS PubMed Google Scholar
Kebschull, J. M. et al. Cerebellar nuclei evolved by repeatedly duplicating a conserved cell-type set. Science 370, eabd5059 (2020).
Krienen, F. M. et al. Innovations present in the primate interneuron repertoire. Nature 586, 262–269 (2020).
Article CAS PubMed PubMed Central Google Scholar
Bakken, T. E. et al. Comparative cellular analysis of motor cortex in human, marmoset and mouse. Nature 598, 111 (2021).
Article CAS PubMed PubMed Central Google Scholar
Franjic, D. et al. Transcriptomic taxonomy and neurogenic trajectories of adult human, macaque, and pig hippocampal and entorhinal cells. Neuron 110, 452–469 (2021).
Tran, M. N. et al. Single-nucleus transcriptome analysis reveals cell-type-specific molecular signatures across reward circuitry in the human brain. Neuron 109, 3088–3103 (2021).
Article CAS PubMed PubMed Central Google Scholar
Sorrells, S. F. et al. Immature excitatory neurons develop during adolescence in the human amygdala. Nat. Commun. 10, 2748 (2019).
Article PubMed PubMed Central Google Scholar
Zhang, L. et al. Molecular taxonomy of the primate amygdala via single-nucleus RNA sequencing analysis. Sci. Bull. 66, 1379–1383 (2021).
Article CAS Google Scholar
Wu, Y. E., Pan, L., Zuo, Y. N., Li, X. M. & Hong, W. Z. Detecting activated cell populations using single-cell RNA-seq. Neuron 96, 313–329 (2017).
Article CAS PubMed Google Scholar
Huang, W. et al. Origins and proliferative states of human oligodendrocyte precursor Cells. Cell 182, 594–608 (2020).
Article CAS PubMed PubMed Central Google Scholar
Haubensak, W. et al. Genetic dissection of an amygdala microcircuit that gates conditioned fear. Nature 468, 270–276 (2010).
Article CAS PubMed PubMed Central Google Scholar
Li, H. et al. Experience-dependent modification of a central amygdala fear circuit. Nat. Neurosci. 16, 332–339 (2013).
Article CAS PubMed PubMed Central Google Scholar
McCullough, K. M., Daskalakis, N. P., Gafford, G., Morrison, F. G. & Ressler, K. J. Cell-type-specific interrogation of CeA Drd2 neurons to identify targets for pharmacological modulation of fear extinction. Transl. Psychiatry 8, 164 (2018).
Article PubMed PubMed Central Google Scholar
Kuerbitz, J. et al. Loss of intercalated cells (ITCs) in the mouse amygdala of Tshz1 mutants correlates with fear, depression, and social interaction phenotypes. J. Neurosci. 38, 1160–1177 (2018).
Article CAS PubMed PubMed Central Google Scholar
Shumyatsky, G. P. et al. Identification of a signaling network in lateral nucleus of amygdala important for inhibiting memory specifically related to learned fear. Cell 111, 905–918 (2002).
Article CAS PubMed Google Scholar
Kim, J., Pignatelli, M., Xu, S., Itohara, S. & Tonegawa, S. Antagonistic negative and positive neurons of the basolateral amygdala. Nat. Neurosci. 19, 1636–1646 (2016).
Article CAS PubMed PubMed Central Google Scholar
Lein, E. S. et al. Genome-wide atlas of gene expression in the adult mouse brain. Nature 445, 168–176 (2007).
Article CAS PubMed Google Scholar
Olson, J. M. et al. NeuroD2 is necessary for development and survival of central nervous system neurons. Dev. Biol. 234, 174–187 (2001).
Article CAS PubMed Google Scholar
Ince-Dunn, G. et al. Regulation of thalamocortical patterning and synaptic maturation by NeuroD2. Neuron 49, 683–695 (2006).
Article CAS PubMed Google Scholar
Lin, C. H. et al. The dosage of the neuroD2 transcription factor regulates amygdala development and emotional learning. Proc. Natl. Acad. Sci. USA 102, 14877–14882 (2005).
Davidson, S., Miller, K. A., Dowell, A., Gildea, A. & MacKenzie, A. A remote and highly conserved enhancer supports amygdala specific expression of the gene encoding the anxiogenic neuropeptide substance-P. Mol. Psychiatry 11, 410–421 (2006).
Article CAS Google Scholar
Kaoru, T. et al. Molecular characterization of the intercalated cell masses of the amygdala: implications for the relationship with the striatum. Neuroscience 166, 220–230 (2010).
Article CAS PubMed Google Scholar
Hodge, R. D. et al. Conserved cell types with divergent features in human versus mouse cortex. Nature 573, 61–68 (2019).
Article CAS PubMed PubMed Central Google Scholar
de la Mora, M. P., Gallegos-Cari, A., Arizmendi-Garcia, Y., Marcellino, D. & Fuxe, K. Role of dopamine receptor mechanisms in the amygdaloid modulation of fear and anxiety: Structural and functional analysis. Prog. Neurobiol. 90, 198–216 (2010).
Article PubMed Google Scholar
Marowsky, A., Yanagawa, Y., Obata, K. & Vogt, K. E. A specialized subclass of interneurons mediates dopaminergic facilitation of amygdala function. Neuron 48, 1025–1037 (2005).
Article CAS PubMed Google Scholar
Aksoy-Aksel, A., Gall, A., Seewald, A., Ferraguti, F. & Ehrlich, I. Midbrain dopaminergic inputs gate amygdala intercalated cell clusters by distinct and cooperative mechanisms in male mice. Elife 10, e63708 (2021).
Bupesh, M., Abellan, A. & Medina, L. Genetic and experimental evidence supports the continuum of the central extended amygdala and a mutiple embryonic origin of its principal neurons. J. Comp. Neurol. 519, 3507–3531 (2011).
Article PubMed Google Scholar
Pearce, E., Wlodarski, R., Machin, A. & Dunbar, R. I. M. Variation in the beta-endorphin, oxytocin, and dopamine receptor genes is associated with different dimensions of human sociality. Proc. Natl. Acad. Sci. USA 114, 5300–5305 (2017).
Ripke, S. et al. Biological insights from 108 schizophrenia-associated genetic loci. Nature 511, 421–427 (2014).
Article CAS PubMed Central Google Scholar
Chen, M., Li, Y., Liu, Y., Xu, H. B. & Bi, L. L. Neuregulin-1-dependent control of amygdala microcircuits is critical for fear extinction. Neuropharmacology 201, 108842 (2021).
Yamaguchi, T. et al. Posterior amygdala regulates sexual and aggressive behaviors in male mice. Nat. Neurosci. 23, 1111–1124 (2020).
Article CAS PubMed PubMed Central Google Scholar
Hashikawa, K. et al. Esr1(+) cells in the ventromedial hypothalamus control female aggression. Nat. Neurosci. 20, 1580–1590 (2017).
Article CAS PubMed PubMed Central Google Scholar
Fang, Y. Y., Yamaguchi, T., Song, S. C., Tritsch, N. X. & Lin, D. Y. A Hypothalamic midbrain pathway essential for driving maternal behaviors. Neuron 98, 192–207 (2018).
Article CAS PubMed PubMed Central Google Scholar
Wang, L. et al. Hypothalamic control of conspecific self-defense. Cell Rep. 26, 1747–1758 (2019).
Zhong, S. J. et al. A single-cell RNA-seq survey of the developmental landscape of the human prefrontal cortex. Nature 555, 524–528 (2018).
Article CAS PubMed Google Scholar
Dow, D. J. et al. ADAMTSL3 as a candidate gene for schizophrenia: Gene sequencing and ultra-high density association analysis by imputation. Schizophr. Res. 127, 28–34 (2011).
Article PubMed Google Scholar
Aleksic, B. et al. Analysis of the VAV3 as candidate gene for schizophrenia: evidences from voxel-based morphometry and mutation screening. Schizophr. Bull. 39, 720–728 (2013).
Article PubMed Google Scholar
Calabro, M. et al. FKBP5 gene variants may modulate depressive features in bipolar disorder. Neuropsychobiology 78, 104–111 (2019).
Article CAS PubMed Google Scholar
Fu, X. et al. LTBP1 plays a potential bridge between depressive disorder and glioblastoma. J. Transl. Med. 18, 391 (2020).
Article CAS PubMed PubMed Central Google Scholar
Colquitt, B. M., Merullo, D. P., Konopka, G., Roberts, T. F. & Brainard, M. S. Cellular transcriptomics reveals evolutionary identities of songbird vocal circuits. Science 371, eabd9704 (2021).
Article CAS PubMed PubMed Central Google Scholar
Pessoa, L., Medina, L., Hof, P. R. & Desfilis, E. Neural architecture of the vertebrate brain: implications for the interaction between emotion and cognition. Neurosci. Biobehav. Rev. 107, 296–312 (2019).
Article PubMed PubMed Central Google Scholar
Puelles, L. et al. eds. The Chick Brain in Stereotaxic Coordinates and Alternate Stains. 2nd edn (Academic Press, 2018).
Vicario, A., Abellan, A., Desfilis, E. & Medina, L. Genetic identification of the central nucleus and other components of the central extended amygdala in chicken during development. Front. Neuroanat. 8, 90 (2014).
Article PubMed PubMed Central Google Scholar
Vicario, A., Abelian, A. & Medina, L. Embryonic origin of the Islet1 and Pax6 neurons of the chicken central extended amygdala using cell migration assays and relation to different neuropeptide-containing cells. Brain Behav. Evol. 85, 139–169 (2015).
Article PubMed Google Scholar
Sardina, J. M., Walters, A. R., Singh, K. E., Owen, R. X. & Kimonis, V. E. Amelioration of the typical cognitive phenotype in a patient with the 5pter deletion associated with Cri-du-chat syndrome in addition to a partial duplication of CTNND2. Am. J. Med. Genet. A. 164, 1761–1764 (2014).
Article CAS Google Scholar
Law, R. et al. Biallelic truncating mutations in FMN2, encoding the actin-regulatory protein formin 2, cause nonsyndromic autosomal-recessive intellectual disability. Am. J. Hum. Genet. 95, 721–728 (2014).
Article CAS PubMed PubMed Central Google Scholar
Damaj, L. et al. CACNA1A haploinsufficiency causes cognitive impairment, autism and epileptic encephalopathy with mild cerebellar symptoms. Eur. J. Hum. Genet. 23, 1505–1512 (2015).
Article CAS PubMed PubMed Central Google Scholar
Lee, H. J. et al. Association study of polymorphisms in synaptic vesicle-associated genes, SYN2 and CPLX2, with schizophrenia. Behav. Brain Funct. 1, 15 (2005).
Corradi, A. et al. SYN2 is an autism predisposing gene: loss-of-function mutations alter synaptic vesicle cycling and axon outgrowth. Hum. Mol. Genet. 23, 90–103 (2014).
Article CAS PubMed Google Scholar
Salzman, C. D. & Fusi, S. Emotion, cognition, and mental state representation in amygdala and prefrontal cortex. Annu. Rev. Neurosci. 33, 173–202 (2010).
Article CAS PubMed PubMed Central Google Scholar
Herry, C. & Johansen, J. P. Encoding of fear learning and memory in distributed neuronal circuits. Nat. Neurosci. 17, 1644–1654 (2014).
Article CAS PubMed Google Scholar
Hagihara, K. M. et al. Intercalated amygdala clusters orchestrate a switch in fear state. Nature 594, 403–407 (2021).
Article CAS PubMed PubMed Central Google Scholar
Ciocchi, S. et al. Encoding of conditioned fear in central amygdala inhibitory circuits. Nature 468, 277–282 (2010).
Article CAS PubMed Google Scholar
Xiao, L. et al. Expression of FoxP2 in the basal ganglia regulates vocal motor sequences in the adult songbird. Nat. Commun. 12, 2617 (2021).
Article CAS PubMed PubMed Central Google Scholar
Medina, L. et al. Expression of Dbx1, neurogenin 2, semaphorin 5a, cadherin 8, and Emx1 distinguish ventral and lateral pallial histogenetic divisions in the developing mouse claustroamygdaloid complex. J. Comp. Neurol. 474, 504–523 (2004).
Article PubMed Google Scholar
Remedios, R. et al. A stream of cells migrating from the caudal telencephalon reveals a link between the amygdala and neocortex. Nat. Neurosci. 10, 1141–1150 (2007).
Article CAS PubMed Google Scholar
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).
Article CAS PubMed PubMed Central Google Scholar
Dheerendra, P., Lynch, N. M., Crutwell, J., Cunningham, M. O. & Smulders, T. V. In vitro characterization of gamma oscillations in the hippocampal formation of the domestic chick. Eur. J. Neurosci. 48, 2807–2815 (2018).
Article PubMed PubMed Central Google Scholar
Zheng, G. X. Y. et al. Massively parallel digital transcriptional profiling of single cells. Nat. Commun. 8, 14049 (2017).
Article CAS PubMed PubMed Central Google Scholar
Wolock, S. L., Lopez, R. & Klein, A. M. Scrublet: computational identification of cell doublets in single-cell transcriptomic data. Cell Syst. 8, 281–291 e289 (2019).
Article CAS PubMed PubMed Central Google Scholar
Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19, 15 (2018).
Article PubMed PubMed Central Google Scholar
Durinck, S., Spellman, P. T., Birney, E. & Huber, W. Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nat. Protoc. 4, 1184–1191 (2009).
Article CAS PubMed PubMed Central Google Scholar
Zeisel, A. et al. Molecular architecture of the mouse nervous system. Cell 174, 999–1014 (2018).
Article CAS PubMed PubMed Central Google Scholar
Aran, D. et al. Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage. Nat. Immunol. 20, 163–172 (2019).
Article CAS PubMed PubMed Central Google Scholar
Lun, A. T. L., McCarthy, D. J. & Marioni, J. C. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Res. 5, 2122 (2016).
PubMed PubMed Central Google Scholar
Crow, M., Paul, A., Ballouz, S., Huang, Z. J. & Gillis, J. Characterizing the replicability of cell types defined by single cell RNA-sequencing data using MetaNeighbor. Nat. Commun. 9, 1–12 (2018).
Article CAS Google Scholar
Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).
Article CAS PubMed PubMed Central Google Scholar
Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47 (2015).
Article PubMed PubMed Central Google Scholar
Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering. Nat. Methods 14, 1083–1086 (2017).
Article CAS PubMed PubMed Central Google Scholar
Cheng, S. et al. A pan-cancer single-cell transcriptional atlas of tumor infiltrating myeloid cells. Cell 184, 792–809 (2021).
Article CAS PubMed Google Scholar
Guo, X. et al. Global characterization of T cells in non-small-cell lung cancer by single-cell sequencing. Nat. Med. 24, 978–985 (2018).
Article CAS PubMed Google Scholar
Yu, G., Wang, L.-G., Han, Y. & He, Q.-Y. clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters. OMICS 16, 284–287 (2012).
Article CAS PubMed PubMed Central Google Scholar
Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinforma. 14, 7 (2013).
Article Google Scholar
Fertig, E. J., Ding, J., Favorov, A. V., Parmigiani, G. & Ochs, M. F. CoGAPS: an R/C++ package to identify patterns and biological process activity in transcriptomic data. Bioinformatics 26, 2792–2793 (2010).
Article CAS PubMed PubMed Central Google Scholar
Clark, B. S. et al. Single-cell rna-seq analysis of retinal development identifies NFI factors as regulating mitotic exit and late-born cell specification. Neuron 102, 1111–1126 (2019).
Article CAS PubMed PubMed Central Google Scholar
Lu, Y. et al. Single-cell analysis of human retina identifies evolutionarily conserved and species-specific mechanisms controlling development. Dev. Cell 53, 473–491 (2020).
Article CAS PubMed PubMed Central Google Scholar
Dani, N. et al. A cellular and spatial map of the choroid plexus across brain ventricles and ages. Cell 184, 3056–3074 (2021).
Article CAS PubMed PubMed Central Google Scholar
Dey, K. K., Hsiao, C. J. & Stephens, M. Visualizing the structure of RNA-seq expression data using grade of membership models. PLoS Genet. 13, e1006599 (2017).
Article PubMed PubMed Central Google Scholar
Saleem K. & Logothetis N. eds. A Combined MRI and Histology Atlas of the Rhesus Monkey Brain in Stereotaxic Coordinates. 2nd edn (Academic Press, 2012).

Download references

Acknowledgements

We thank G.-J. Zhang (Zhejiang University, China), and all members of X.-M. Li’s Lab for discussions and suggestions on experimental design and manuscript preparation. We thank the Core Facilities of the Zhejiang University Institute of Neuroscience for technical support and National Resource Center for Non-Human Primates. This work was supported by Science and Technology Innovation 2030 - “Brain Science and Brain-Inspired Technology” Major Project (2021ZD0202702, 2021ZD0202700, 2021ZD0200102), Major Program of the National Natural Science Foundation of China (NSFC) (82090031, 32192411, 32122037, 82090030), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA16020601, XDB32010100), Key-Area Research and Development Program of Guangdong Province (2018B030334001, 2019B030335001), Key R&D Program of Zhejiang Province (2020C03009), Fundamental Research Funds for the Central Universities (2021FZZX001-37), and CAMS Innovation Fund for Medical Sciences (2019-I2M-5-057). NSFC (32000714), Fellowship of China Postdoctoral Science Foundation (2020M671694), and CAS Project for Young Scientists in Basic Research (YSBR-013).

Author information

These authors contributed equally: Bin Yu, Qianqian Zhang, Lin Lin

Authors and Affiliations

Department of Neurobiology and Department of Neurology of the Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang, China
Bin Yu, Lin Lin, Chunyue Li & Xiao-Ming Li
NHC and CAMS Key Laboratory of Medical Neurobiology, Ministry of Education Frontier Science Center for Brain Research and Brain-Machine Integration, School of Brain Science and Brain Medicine, Zhejiang University School of Medicine, Hangzhou, Zhejiang, China
Bin Yu, Lin Lin, Chunyue Li & Xiao-Ming Li
State Key Laboratory of Brain and Cognitive Science, CAS Center for Excellence in Brain Science and Intelligence Technology, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China
Qianqian Zhang, Wenji Ma, Wei Wang & Xiaoqun Wang
University of Chinese Academy of Sciences, Beijing, China
Qianqian Zhang & Xiaoqun Wang
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing, China
Xin Zhou, Shaonan Wen, Qian Wu & Xiaoqun Wang
IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, China
Qian Wu & Xiaoqun Wang
Yunnan Key Laboratory of Primate Biomedical Research, Institute of Primate Translational Medicine, Kunming University of Science and Technology, Kunming, Yunnan, China
Xiaoqun Wang
Research Units for Emotion and Emotion disorders, Chinese Academy of Medical Sciences, Beijing, China
Xiao-Ming Li

Authors

Bin Yu
View author publications
You can also search for this author in PubMed Google Scholar
Qianqian Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Lin Lin
View author publications
You can also search for this author in PubMed Google Scholar
Xin Zhou
View author publications
You can also search for this author in PubMed Google Scholar
Wenji Ma
View author publications
You can also search for this author in PubMed Google Scholar
Shaonan Wen
View author publications
You can also search for this author in PubMed Google Scholar
Chunyue Li
View author publications
You can also search for this author in PubMed Google Scholar
Wei Wang
View author publications
You can also search for this author in PubMed Google Scholar
Qian Wu
View author publications
You can also search for this author in PubMed Google Scholar
Xiaoqun Wang
View author publications
You can also search for this author in PubMed Google Scholar
Xiao-Ming Li
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

B.Y., X.-M.L., X.W., and Q.W. conceptualized and designed the experiments. B.Y., Q. Z., W.M., and S. W. performed the snRNA-seq experiments and data analyses. B.Y. and L.L. performed the tissue section, RNAscope experiments, and data analyses. X.Z. and C.L. performed the RNAscope data analyses and image processing. W.W. prepared the macaque sample. B.Y. and Q.Z. wrote the manuscript, and all authors edited and proofed the manuscript. Q.W., X.W., and X.-M.L. supervised all phases of the project.

Corresponding authors

Correspondence to Qian Wu, Xiaoqun Wang or Xiao-Ming Li.

Ethics declarations

Conflict of interest

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Table S4

Supplementary Table S5

Supplementary Table S6

Supplementary Table S7

Supplementary Table S8

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Yu, B., Zhang, Q., Lin, L. et al. Molecular and cellular evolution of the amygdala across species analyzed by single-nucleus transcriptome profiling. Cell Discov 9, 19 (2023). https://doi.org/10.1038/s41421-022-00506-y

Download citation

Received: 18 July 2022
Accepted: 24 November 2022
Published: 14 February 2023
DOI: https://doi.org/10.1038/s41421-022-00506-y