Chimeric brain models to study human glial-neuronal and macroglial-microglial interactions

SUMMARY

Chimeric brain models generated by transplanting human pluripotent stem cell (hPSC)-derived neural cells are valuable for studying the development and function of human neural cells in vivo. To explore glial-neuronal and glial-glial interactions, we co-engraft hPSC-derived primitive neural progenitor cells and primitive macrophage progenitors into neonatal mouse brains, generating chimeric brains containing human microglia, macroglia, and neurons. Using super-resolution imaging and 3D reconstruction, we observe human microglia pruning synapses and engulfing neurons. Single-cell RNA sequencing reveals human glial progenitor populations and dynamic stages of astroglial development resembling those in the human brain. Cell-cell communication analysis identifies strong human neural interactions, including NRXN-NLGN3 signaling between neurons and astrocytes and SPP1- and PTN-MK-mediated communication between microglia and astroglia. This co-transplantation model provides a powerful approach to study complex human glial-neuronal interactions and mechanisms underlying neurological diseases.

In brief

Jin et al. develop chimeric mouse brains containing human microglia, astroglia, oligodendroglia, and neurons, enabling direct interrogation of human neuron-glia interactions in vivo. These models recapitulate human signaling pathways that drive synaptic development and glial maturation, providing a powerful platform for dissecting human brain development and disease mechanisms.

Graphical Abstract

INTRODUCTION

While animal models have been invaluable in advancing our understanding of human brain development, aging, and the pathogenic mechanisms of neurological diseases,^1–3 they may not fully recapitulate the cellular and molecular changes observed in the human brain due to significant species-specific differences between humans and other model organisms. Limited availability of functional brain tissue from healthy individuals or patients with specific neurological disorders impedes our understanding of human brain development and aging, particularly in the context of disease. The advancement of human pluripotent stem cell (hPSC) technologies offers promising new opportunities to overcome these challenges.⁴ Expanding on neural differentiation of hiPSCs, including human induced pluripotent stem cells (hiPSCs) and human embryonic stem cells (hESCs), in vitro two-dimensional (2D) human neural cell culture,⁵ and three-dimensional (3D) brain organoid models⁶ have aided in uncovering mechanisms underlying human brain development and diseases.^7,8 Nevertheless, hPSC-derived neural cells cultured in 2D or 3D organoids often exhibit limited cellular and functional maturation and cellular heterogeneity.^9,10 Transplanting hPSC-derived neural cells, such as dissociated human glial or neuronal cells,^11–20 or intact cerebral organoids,^21–25 into the brains of immunodeficient rodents to create chimeric brains can facilitate the improvement of cellular functionality and heterogeneity.²⁶

Cell-cell interactions, such as glial-neuronal and glial-glial interactions, are crucial for the proper functioning of the central nervous system (CNS). In the brain, microglia and macroglia, including astroglia and oligodendroglia, were historically perceived as support cells for neurons. However, recent research has illuminated their essential roles in regulating neuronal development and activity, synaptic transmission, and overall brain function.²⁷ Microglia, the brain-resident immune cells engage in immune surveillance, inflammation, phagocytosis, regulating neurogenesis, and synaptic pruning—the removal of unnecessary synapses during development or in response to injury or disease.²⁸ Astroglia, the most abundant glial type, closely interacts with neurons and synapses, modulating neuronal maturation and synaptic transmission.²⁹ Oligodendroglial cells are responsible for myelinating axons in the CNS, which greatly increases the speed and efficiency of neuronal communication. This myelination process is essential for proper neural signaling and is crucial for the function of the brain.³⁰ Understanding glial-neuronal and glial-glial interactions is vital for unraveling the complexities of brain function and developing therapeutic treatments for neurological disorders. In contrast to neurons, glial cells exhibit lower evolutionary conservation, as evidenced by the limited ability of rodent glia to replicate the characteristics of human glia.^12,31–36 This underscores the necessity for employing species-specific tools to augment our understanding of human glial development, glial modulation of neuronal development and function, and glial-glial interactions. Various types of cells, such as human glial progenitor cells (GPCs), oligodendroglia progenitor cells (OPCs), immature astroglial cells, microglial progenitors (also referred to as primitive macrophage progenitors [PMPs]), and neuronal progenitor cells, have been utilized for single-cell population transplants to establish human-mouse macroglial, microglial, or neuronal chimeric brains.^11,26 Human-mouse chimeric brains incorporate live human glia or neurons into the mouse brain, where they structurally and functionally integrate. However, the capacity of human-mouse chimeric brain models to explore human glial-neuronal and glial-glial interactions remains insufficiently explored.^17,24,25

In this report, we present an innovative transplantation approach where primitive neural progenitor cells (pNPCs) and PMPs are co-transplanted into the brains of neonatal immunodeficient mice. Our findings reveal that the resulting chimeric mouse brains, which incorporate human microglia, astroglia, oligodendroglia, and neurons, provide a valuable model for studying the interactions between human neurons and glial cells. When brain organoids containing PMPs and ventralized neural progenitor cells (NPCs) were cultured under conditions conducive to glial formation, it was observed that human microglia could survive and display phagocytic activity without the need for human colony-stimulating factor 1 (hCSF1) supplementation. This observation was corroborated by co-transplanting PMPs and ventralized NPCs into recipient mice without the requirement for a hCSF1 knockin. Furthermore, single-cell RNA sequencing (scRNA-seq) of these co-transplanted chimeric brains identified diverse human cell types, various stages of astroglia development, and the presence of human forebrain GPC populations, closely resembling those found in the human brain.

RESULTS

Microglial-neuronal interactions in a chimeric brain model containing human microglia, astroglia, and neurons

We aimed to establish a chimeric brain model to investigate human glial-neuronal and glial-glial interactions. Specifically, PMPs and pNPCs derived from ND2.0/UTY1/C5 hiPSC and H1/H9 hESC lines were co-transplanted at a 1:1 ratio into the hippocampus and corpus callosum of postnatal day 0 (P0) Rag2^−/− IL2rγ^−/− hCSF1^KI immunodeficient mice (Figure 1A), enabling the study of interactions between human neural cells and microglia in vivo. The PMPs and pNPCs were derived from control hiPSC and hESC lines that were fully characterized in our previous studies.^8,14,37 Most of the PMP cells expressed hematopoietic progenitor cell markers CD45³⁸ (Figures 1B and 1C), CD235, and CD43 (Figures S1A and S1B), and nearly all pNPCs expressed cell surface markers of early neural commitment, including CD133,³⁹ CD15,³⁹ CD90⁴⁰ (THY1), and A2B5^41,42 (Figures 1D and 1E), along with the transcription factors SOX2 and PAX6 (Figures S1A and S1B). Notably, we observed that very few cells (about 0.06%) expressed the pluripotency marker SSEA-4³⁹ (Figures 1D and 1E). Next, to further evaluate the identities of pNPCs and PMPs as generated in vitro, we conducted RNA sequencing (RNA-seq). The results revealed distinct gene-expression profiles for each cell type. Specifically, pNPCs exhibited high expression of NPC-associated genes such as SOX2, NES, and EOMES, while PMPs showed elevated expression of genes such as CSF1R, SPI1, and C1QA, as illustrated in Figure S1C. Moreover, the RNA-seq data for these pNPCs and PMPs showed no expression of pluripotent genes, such as NANOG and POU5F1 (also known as Octamer-binding transcription factor 4 [OCT4]) (Figure S1D), and pNPCs showed strong expression of dorsal markers (MSX1, ZIC1, BMP4, PAX7) and low expression of ventral markers (SHH, NKX2.2, OLIG2, FOXA1), indicating their dorsal regional identity (Figure S1E).

Figure 1. Generation and characterization of hMAN chimeric mice.

(A) A schematic diagram showing that hiPSC- and hESC-derived PMPs and pNPCs are engrafted into the brains of P0 Rag2^−/− IL2rγ^−/− hCSF1^KI mice.

(B and C) Representative flow cytometry plot (B) and quantification (C) of CD45-positive PMP cells.

(D and E) Representative flow cytometry plots (D) and quantification (E) of pNPC cells expressing CD90, CD133, A2B5, SSEA-4, and CD15.

(F) Sagittal brain dot maps showing hN⁺ cell distribution in the Rag2^−/− IL2rγ^−/− hCSF1^KI mouse brain 7 months post transplantation.

(G) Representative image from sagittal brain sections showing the distribution of hN⁺ xenografted cells at 7 months post transplantation. Scale bars: 500 μm, 100 μm, and 50 μm in the original and enlarged images, respectively.

(H and I) Representative images (H) and quantification (I) of NeuN-, OLIG2-, GFAP-, and IBA-expressing hN⁺ cells (n = 4 mice). Arrows: NeuN⁺, OLIG2⁺, GFAP⁺, and IBA1⁺ cells colocalized with hN⁺-human cells. Scale bars: 20 μm and 5 μm in the original and enlarged images, respectively.

(J) Representative images of hMAP2⁺ and hTMEM119⁺ cells. Scale bars: 100 μm.

(K) Representative raw fluorescent super-resolution and 3D surface-rendered images showing colocalization of hTMEM119⁺/P2RY12⁺ and hN⁺ staining. Scale bars: 20 μm and 10 μm in the original and enlarged images, respectively.

(L) Representative raw fluorescent super-resolution and 3D surface-rendered images showing colocalization of hTMEM119⁺ and hSynaptophysin⁺ staining. Scale bars: 5 μm.

(M) Representative raw fluorescent super-resolution and 3D surface-rendered images showing colocalization of hCD45⁺ and TBR2⁺ staining. Scale bars: 5 μm.

(N) Representative fluorescence image showing a GFP-labeled human neuron targeted for patch-clamp recording (white arrow).

(O) Current-clamp recordings showing action potentials in response to depolarizing steps.

(P) Voltage-clamp recordings and a zoomed-in image showing sodium and potassium currents.

(Q) Representative images of hCD44⁺ and GFAP⁺ cells. Scale bars: 20 μm.

At 7 months post transplantation, xenografted cells labeled with the human-specific anti-nuclei antibody (hN) were primarily located in the cerebral cortex and hippocampus. Additionally, some of these cells had migrated and integrated with mouse cells at the periphery of the cluster of engrafted human cells (Figures 1F, 1G, and S1F). To determine the identity of hN⁺ cells, we concentrated on examining the cells within the areas of the clusters of human cells. We stained the brain sections prepared from 2- and 7-month-old mice with a neuronal marker NeuN, an astroglial marker GFAP, a microglia/macrophage marker IBA-1, and OLIG2, a marker for GPCs and oligodendroglial lineage cells (Figures 1H and S1G). In 2-month-old chimeric brains, 57.5% of hN⁺ cells expressed NeuN, 5.8% expressed IBA-1, 26.2% expressed OLIG2, and 10.3% expressed GFAP (Figure S1H). We found that about 55.5% of the hN⁺ cells expressed NeuN, 6.0% expressed IBA-1, 18.8% expressed OLIG2, and 19.3% of hN⁺ cells expressed GFAP in 7-month-old chimeric brains (Figures 1I and S1H), indicating the proportions of different cell lineages remain stable after long-term engraftment. Notably, quantitative analysis of IBA1⁺/hN⁺ human microglia revealed that these cells were evenly distributed throughout the transplanted regions of the mouse brain rather than forming localized clusters (Figure S1I). To further assess spatial organization, we calculated the nearest-neighbor distance (NND) of human microglia in the cortex and hippocampus at multiple time points (Figure S1J). During the early post-transplantation stages, human microglia exhibited a sparser distribution within the hippocampus than in the cerebral cortex. Over time, as the transplanted human microglia integrated and proliferated, their spatial distribution gradually became more uniform across brain regions. By late stages of engraftment, NND values in both regions converged to approximately 50 μm, a distance previously reported as indicative of regular microglial tiling,^43–45 supporting the conclusion that human microglia progressively establish an even and stable spatial organization during long-term maturation in the chimeric brain. To further assess the specific subtypes of neurons, we co-stained human-specific MAP2 (hMAP2) with glutaminase (GLS) or γ-aminobutyric acid (GABA), markers of excitatory and inhibitory neurons, respectively⁴⁶ (Figure S1K). As shown in Figure S1L, there were 88.6% hMAP2⁺/GLS⁺ cells and 10.8% hMAP2⁺/GABA⁺ cells among the total hMAP2⁺ human neurons, indicating that the engrafted human cells primarily differentiated into excitatory neurons.

To explore microglial-neuronal interactions, we initially stained brain sections with human-specific antibodies recognizing TMEM119 (hTMEM119) and MAP2 (hMAP2) and observed that hTMEM119⁺ microglia were positioned adjacent to hMAP2⁺ neurons (Figure 1J). Additionally, all hTMEM119⁺ and P2RY12⁺ microglia were positive for hN⁺, confirming the presence of homeostatic human microglia in this co-transplantation model and remaining homeostatic after long-term integration (Figures 1K, S2A, and S2B). Microglia have been demonstrated to shape synaptic development through engulfing and eliminating synapses.⁴⁷ Therefore, to investigate the function of synaptic pruning, we double-stained hTMEM119 with the human-specific presynaptic vesicle protein synaptophysin (hSyn) in brain sections prepared from 7-month-old chimeric mice (Figure 1L). The 3D reconstruction images demonstrated that hSyn⁺ puncta were colocalized within hTMEM119⁺ microglia, indicating that these human synaptic proteins are phagocytosed by human microglia (Figure 1L). In addition, to examine the phagocytosis of immature neurons by microglia, we co-transplanted pNPCs derived from a GFP-labeled hESC line and hESC-derived PMPs at a 1-to-1 ratio into the hippocampus and corpus callosum of the P0 Rag2^−/− IL2rγ^−/− hCSF1^KI immunodeficient mouse brain. Then, we stained the brain sections prepared from 2-month-old chimeric mice with human-specific CD45 (hCD45) and intermediate progenitor stage marker TBR2 (Figure 1M). Our findings suggested that hCD45⁺ microglia phagocytized developing GFP⁺TBR2⁺ human neurons. We next assessed the expression of c-Fos, a well-established activity-dependent immediate-early gene, to map neuronal activation following depolarization.⁴⁸ As shown in Figure S2C, we observed c-Fos/hN double-positive nuclei, indicating that these human neurons were active within the mouse brain. Previously, we reported that xenografted human neurons reached electrophysiological maturity in 6-month-old chimeric brains.^37,49 In the present study, we extended this analysis by performing whole-cell patch-clamp recordings on 10 GFP-labeled pNPC-derived neurons in 3-month-old mice transplanted at P0 with GFP-expressing CAGG-pNPCs. Even at this earlier stage of human pNPC development in vivo, these neurons exhibited robust inward sodium currents and generated action potentials. These findings demonstrate that neurons derived from pNPCs can achieve functional maturation and display neuronal activity at early developmental stages in vivo (Figures 1N and 1P).

We examined the human astroglial population in the chimeric brain using the human glial marker CD44,^50,51 along with the astrocyte progenitor marker SOX9 and astroglial markers GFAP and S100B. Our findings revealed that the majority of hCD44⁺ cells co-expressed SOX9, indicating that these CD44⁺ glial cells represent astroglial progenitor cells (Figure S2D). The human astroglial cells labeled by human-specific CD44 (hCD44) developed complex structures with extended, long, and unbranched processes, as previously reported³⁵ (Figure 1Q). As shown in Figures S2E and S2F, we also observed that subpopulations of hCD44⁺ cells that did not express GFAP (38.0% GFAP⁺/hCD44⁺ and 58.0% GFAP⁻/hCD44⁺ among the total hCD44⁺ cells) or S100B (59.8% S100B⁺/hCD44⁺ and 35.2% S100B⁻/hCD44⁺ among the total hCD44⁺ cells). Since CD44 is also a marker for astroglial precursor cells,⁵⁰ this suggests that these hCD44⁺/GFAP⁻ or hCD44⁺/S100B⁻ human astroglia were likely at immature stages. Additionally, in our 7-month-old chimeric brains, we did not find any hN⁺ cells that expressed OCT4, a pluripotent stem cell marker and a tumor-initiating cell marker (Figure S2G).^52,53 In line with previous reports of prolonged proliferation of human progenitor cells in chimeric brains,^35,54 we observed that few hN⁺ cells expressed the neural progenitor marker SOX2 and the proliferation marker Ki67 (Figures S2H and S2I). The proportion of these cells decreased with longer integration periods (Figures S1Q and S1R). Collectively, these results indicate that, in the context of glial differentiation in a myelin-wild-type brain, the engrafted human pNPCs predominantly differentiated into astroglial cells, consistent with previous studies.^55,56 These findings demonstrate that the co-transplantation approach using hiPSC-derived pNPCs and PMPs effectively generates chimeric brains containing phagocytic human microglia, macroglia, and functional neurons. This model is referred to as hMAN mice in this study.

Oligodendroglial-neuronal interactions in chimeric shi/shi × Rag2^−/− mice containing human microglia, oligodendroglia, and neurons

Shiverer mice, characterized by a mutation in the myelin basic protein (Mbp) gene,^57,58 are commonly employed to investigate the myelination potential of transplanted cells.^59,60 The hypomyelinated brain environment of shiverer mice facilitates the differentiation of transplanted cells into oligodendroglia.^54,61,62 Furthermore, since shiverer mouse brains lack endogenous MBP expression, any detected MBP production is solely attributed to the xenografted cells,⁶⁰ making this a valuable model for investigating the development and maturation of xenografted human oligodendroglia. Building on the shiverer model and employing the same co-transplantation approach, we next set out to generate chimeric mouse brains containing human oligodendroglia, microglia, and neurons. We co-transplanted H9 hESC and C5 hiPSC-derived PMPs and pNPCs derived from GFP-labeled H9 hESC and C5 hiPSC line at a 1-to-1 ratio into the hippocampus and corpus callosum of P0 shiverer (shi/shi × Rag2^−/−) immunodeficient mice (Figure 2A). As shown in Figures 2B, 2C, and S3A, the donor-derived hN⁺ cells were distributed both in the cerebral cortex and hippocampus at 3 months post transplantation. In our 4-month-old chimeric mice, we did not detect any hN⁺ cells that expressed OCT4 (Figure S3B), while approximately 17.3% of hN⁺ cells expressed the proliferation marker Ki67 (Figures S3C and S3D). To further explore the engrafted cell identity in shiverer mice, we also focused on examining the cells within the areas of the clusters of human cells (Figure 2C2). We stained the mouse brain sections with NeuN, OLIG2, GFAP, IBA-1, and hN (Figures 2D and S3E). We observed that approximately 57.6% of the hN⁺ cells expressed NeuN; 5.3% expressed IBA-1; 10.3% expressed GFAP; and a second large proportion, 25.7%, expressed OLIG2 (Figure 2E). It is important to note that the shiverer mice were not humanized for colony-stimulating factor-1 (CSF1), suggesting that the co-transplantation approach supported the survival of human microglia. To further validate whether this approach could support human microglia survival in different mouse strains, we transplanted the same combination of pNPCs and PMPs into Rag1^−/− mice. We stained the mouse brain sections and observed the presence of NeuN⁺/hN⁺, GFAP⁺/hN⁺, OLIG2⁺/hN⁺ cells, and particularly IBA-1⁺/hN⁺ human microglia (Figure S3F). The presence of human-origin microglia in Rag1^−/− mice was further confirmed by the detection of cells positive for human-specific CD45 (hCD45) (Figure S3G).

Figure 2. Generation and characterization of hMON chimeric mice.

(A) A schematic diagram showing that hiPSC and hESC-derived PMPs and pNPCs are engrafted into the brains of P0 shi/shi x Rag2^−/− mice.

(B) Sagittal brain dot maps showing hN⁺ cell distribution in the shi/shi x Rag2^−/− mouse brain 3 months post transplantation.

(C) Representative image from sagittal brain sections showing the distribution of hN⁺ xenografted cells at 3 months post transplantation. Scale bars: 500 μm, 100 μm, and 100 μm in the original and enlarged images, respectively.

(D and E) Representative images (D) and quantification (E) of NeuN⁺, OLIG2⁺, GFAP⁺, and IBA⁺ cells (n = 3 mice). Arrows: NeuN⁺, OLIG2⁺, GFAP⁺, and IBA1⁺ cells colocalized with hN⁺-human cells. Scale bars: 20 μm and 5 μm in the original and enlarged images, respectively.

(F) Representative images of NG2⁺ and hN⁺ cells. Scale bars: 5 μm.

(G) Representative images of MBP⁺ and hN⁺ cells. Scale bars: 5 μm.

To further examine the oligodendroglial population, we co-stained hN with NG2, a marker for OPCs, or MBP, a marker for mature oligodendrocytes (Figures 2F and 2G). We found a large number of NG2⁺ human OPCs and a few MBP⁺ cells in the human microglia, oligodendroglia, and neurons (hMON) chimeric mice at 3 months post transplantation (Figures 2F and 2G; Figure S3H). As shown in Figure 2G, the 3D reconstruction images demonstrated soma and cytoplasmic processes of MBP⁺ oligodendrocytes. This suggests that engrafted human pNPCs differentiated into NG2⁺ OPCs that began to mature into myelinating oligodendrocytes at 3 months post transplantation. By examining the staining from Rag2^−/− IL2rγ^−/− hCSF1^KI and shiverer mice with human-specific NG2 (hNG2), we observed a higher abundance of hNG2⁺ OPCs in shiverer mice compared to Rag2^−/− IL2rγ^−/− hCSF1^KI mice (Figures S3H and S3I). This is consistent with prior findings that, compared to the myelin-intact mouse brain, the myelin-deficient environment of the Shiverer brain promotes differentiation of xenotransplanted human progenitor cells into oligodendroglia.^54,61,62 In addition, to further explore the specific subtypes of neurons, we co-stained hMAP2 with GABA and observed the presence of hMAP2⁺GABA⁺ inhibitory neurons (Figure S3J). Taken together, these findings demonstrate that the co-transplantation approach enables the creation of chimeric brains containing hMON, without requiring a human CSF1 knockin in the host transgenic mice.

Exposure to gliogenic conditions promotes microglial survival in brain organoids and chimeric brains

Microglia rely on signals from the CSF1 receptor (CSF1R) for survival, proliferation, and differentiation, which are activated by either CSF1 or interleukin-34 (IL-34).⁶³ Previous studies have shown that microglia cannot survive in dorsal forebrain organoids without the addition of these survival factors.²⁴ In the developing brain, CSF1 is mainly produced by glial cells, especially astroglial cells.^64,65 Therefore, we hypothesized that culturing organoids under gliogenic conditions might obviate adding CSF1 or IL-34 to the culture. We first obtained partially ventralized NPCs (v-NPCs) by treating pNPCs with purmorphamine, a small-molecule agonist of the Sonic Hedgehog signaling pathway, for 1 week.⁶⁶ After 1 week of patterning, 30.2% of NPCs expressed the dorsal brain marker PAX6, 35.1% expressed the ventral prosencephalic progenitor marker NKX2.1, 29.12% expressed the ventralization marker Meis2, and 7.5% were OLIG2⁺ (Figures S3K–S3L). As shown in Figure 3A, we then co-cultured these v-NPCs and PMPs at a ratio of 1:1 in 3D to generate microglia-containing brain organoids using a procedure described in our previous study.⁸ Following the recent report,²⁵ we cultured the organoids in a medium containing the gliogenic agent PDGF-AA^67,68 from day 4 to 18, after which we switched to a neuronal differentiation medium for continued culture till day 32. Throughout the culture period, no CSF1 or IL-34 was supplemented (Figure 3A).

Figure 3. Generation and characterization of co-transplantation of PMPs and ventralized NPCs into mouse brains.

(A) A schematic procedure for deriving microglia-containing brain organoids by co-culture of ventralized NPCs and PMPs under 3D conditions. Scale bars: 400 μm.

(B and C) Representative images (B) and quantification (C) of PU.1⁺/CD45⁺ cells in day 9 brain organoids (n = 3 from three independent experiments). Scale bars: 20 μm.

(D) Representative images of CD45⁺ cells in 1-month organoids. Scale bar: 20 μm.

(E) Representative raw fluorescent super-resolution and 3D surface-rendered images showing colocalization of SOX2⁺ and hTMEM119⁺ cells in 1-month organoids. Scale bars: 20 μm and 5 μm in the original and enlarged images, respectively.

(F) Representative images of TUJ-1⁺ and GABA⁺ cells in 1-month organoids. Scale bars: 10 μm.

(G) Representative images of S100B⁺ cells in 1-month organoids. Scale bars: 20 μm.

(H) Quantification of SOX2⁺, TUJ-1⁺, GABA⁺, S100B⁺, and hTMEM119⁺ cells in 1-month organoids (n = 3 from three independent experiments).

(I) qPCR analysis of CSF1, IL-34, and IBA-1 mRNA in 1-month organoids (n = 4 from four independent experiments). Student’s t test, *p < 0.05.

(J) Normalized expression (TPM) of IL-34 gene from bulk RNA-seq data across astrocytes isolated from NPC organoids, pNPCs, and PMPs. Student’s t test, **p < 0.01.

(K) Normalized expression (TPM) of CSF1 gene from bulk RNA-seq data across astrocytes isolated from NPC organoids, pNPCs, and PMPs. Student’s t test, *p < 0.05.

(L and M) Representative images (L) and quantification (M) of CD44⁺, CSF1⁺, and IL34⁺ cells in 1-month organoids (n = 11 and n = 6 organoids were analyzed from independent experiments for CD44/CSF1 and IL34/CSF1 staining, respectively).

(N) A schematic diagram showing that hiPSC and hESC-derived PMP and v-NPCs are engrafted into the brains of P0 Rag2^−/− mice.

(O and P) Representative images (O) and quantification (P) of NeuN⁺, OLIG2⁺, GFAP⁺, and IBA-expressing cells in the brains of Rag2^−/− mouse (n = 3 mice). Arrows: NeuN⁺, OLIG2⁺, GFAP⁺, and IBA1⁺ cells colocalized with hN⁺-human cells. Scale bars: 20 μm.

To explore the survival and maturation of PMPs into microglia in these brain organoids, we examined the expression of microglial markers at day 9 and day 30. At day 9, 4.2% of cells expressed CD45 and 2.8% expressed PU.1, a transcription factor essential for microglial differentiation (Figures 3B and 3C).⁶⁹ Next, we examined the identity of microglia at the neuronal differentiation stage using the homeostatic microglia marker TMEM119. At day 30, there were cells expressing CD45 (Figure 3D), and about 13.9% of cells expressed hTMEM119 in the organoids (Figures 3E–3H). To investigate neuronal and astroglial differentiation in the brain organoids, we stained them with GABA, TUJ-1, and S100B. Due to the patterning of pNPCs to NKX2.1⁺ ventral neural progenitors, we observed 10.3% GABA⁺/TUJ-1⁺ inhibitory neurons (Figures 3F and H). S100B⁺ astroglia accounted for 12.3% of the total cells (Figures 3G and 3H). High levels of CSF1, IL34, and IBA-1 gene expression were detected in our day 14 organoids, whereas these genes could not be detected in undifferentiated pNPCs (Figure 3I). To identify the primary source of the cytokines IL34 and CSF1 in our organoids, we compared gene-expression profiles of astrocytes isolated from 3D-cultured NPC-derived brain organoids⁷⁰ with those of pNPCs and microglia. The results revealed that astrocytes expressed significantly higher levels of both IL-34 and CSF1 than organoids and the other two cell types (Figures 3J and 3K). To further confirm that astrocytes were the primary source of key microglial survival factors, we first performed immunostaining for CSF1 and CD44 in 1-month-old ventral organoids and observed strong colocalization of the CSF1 and CD44⁺ astrocytes. We then stained for CSF1 and IL-34 in those organoids and noticed that the expression of those two cytokines was highly colocalized (Figures 3L–3M). Interestingly, we found that the CSF1 was almost exclusively expressed in CD44⁺ astroglial lineage cells (Figure 3M), which further supports that astroglia are the main source of these cytokines within the organoids, thereby supporting microglial survival (Figures 3L–3M). Furthermore, microglia have been shown to engulf neural progenitors in organoids.^8,71 The 3D reconstruction of hTMEM119 with SOX2 staining images demonstrated that SOX2⁺ cells were colocalized within hTMEM119⁺ microglia, indicating that these microglia were phagocytosing NPCs (Figure 3E). These observations confirm that PMPs were likely supported by CSF1 and IL-34 produced by astrocytes in the organoids and further differentiated into microglia.

To further explore the differentiation of human PMPs into microglia independently of human CSF1 knockin in vivo, we co-transplanted v-NPCs and PMPs derived from H9 hESCs and ND2.0 iPSC at a 1-to1 ratio into the hippocampus and corpus callosum regions of Rag2^−/− immunodeficient mice at P0 (Figure 3N). We observed the hN⁺ cells expressing IBA-1, along with hN⁺/GFAP⁺, hN⁺/OLIG2⁺, hN⁺/NeuN⁺(Figures 3O and 3P), and hMAP2⁺/GABA⁺ cells (Figure S3M), in the 4-month-old Rag2^−/− mouse brain. These findings collectively confirm that PMPs differentiate into microglia both in our brain organoids without the addition of CSF1 or IL-34 and in chimeric mouse brains independently of CSF1 knockin in the host mice.

scRNA-seq of hMAN chimeric mouse brains reveals heterogeneous glial populations

To explore the transcriptional characteristics of co-transplanted cells, we conducted scRNA-seq on the brains of 3-month-old Rag2^−/− IL2rγ^−/− hCSF1^KI mice that had been co-transplanted with GFP-labeled hESC-derived pNPCs and hESC-derived PMPs at P0. The 3-month time point was selected for scRNA-seq because it represents a developmental window in which human cells had successfully engrafted and were actively undergoing lineage specification. At this stage, the cells were mature enough to exhibit diverse transcriptional profiles while remaining developmentally plastic, allowing us to resolve progenitor dynamics, lineage diversification, and early-stage cell-cell interactions. We collected 23,994 high-quality human cells from the chimeric mouse brain after quality control and filtration (Figures 4A and 4B; S4A). Subsequent dimensional reduction and cluster analysis revealed 19 clusters (Figures S4B and S4C) representing eight major human cell types: radial glia, NPCs, excitatory neurons, inhibitory neurons, GPCs, astrocytes, microglia, and vascular leptomeningeal cells (VLMCs). Cell-cycle regression analysis identified three radial glial clusters and one NPC cluster with a high proportion of cells in the cell cycle (Figure S4D). These cells express proliferation markers but lack pluripotency markers (Figure S4E). Additionally, copy number variation (CNV) analysis revealed no genomic alterations in the transplanted human cells (Figure S4F). Gene-expression-level-based similarity analysis with human pre- and postnatal brain samples⁷² indicated that the human cells approximated a developmental stage matching the late second trimester of human brain tissue 3 months post transplantation, and the transplanted cells exhibited a similar predicted developmental age to that of human brain organoids transplanted into rodent brains for 5 and 8 months^23,25 (Figure S4G).

Figure 4. scRNA-seq analyses of hMAN chimeric brain.

(A) Uniform manifold approximation and projection (UMAP) plot of scRNA-seq data (n = 23,994 human cells) from 3-month-old Rag2^−/− IL2rγ^−/− hCSF1^KI hMAN chimeric brains.

(B) Dot plot showing expression of key marker genes of each annotated human cell type. NPC, neural progenitor cell; GPC, glial progenitor cell; Astro, astrocyte; VLMC, vascular leptomeningeal cell.

(C) Dot plot showing expression of marker genes of macroglia human cell types. OPC, oligodendrocyte precursor cell; Oligo., oligodendrocyte.

(D) Heatmap showing the significant overlap of marker genes between human cortical macroglia⁷³ and this study. Jaccard score represents the percentage of pairwise overlapping genes. Fisher’s exact test is used to evaluate the significance of overlapping. The significance levels are represented by asterisks. N.S., not significant.

(E) Representative images of hN⁺ and ETV4⁺ cells from 2-month-old Rag2^−/− IL2rγ^−/− hCSF1^KI hMAN chimeric brains. Arrows, ETV4⁺/hNu⁺ human GPCs. Scale bar: 20 μm.

(F) UMAP plot of astrocyte subclusters (n = 11,100 cells).

(G) UMAP plot of astrocyte subclusters; each cell is colored by its pseudotime trajectory assignment.

(H) Dot plot showing expression of marker genes of each astrocyte subcluster.

(I) Heatmap showing the significant overlap of marker genes between astrocyte subtypes of human hippocampus astrocytes⁷⁵ and this study. Jaccard score represents the percentage of pairwise overlapping genes. Fisher’s exact test is used to evaluate the significance of overlapping. The significance levels are represented by asterisks (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).

(J) Chord diagram showing inferred cell-cell interaction weights/strengths between cell types. The thickness of the line represents the weight of ligand-receptor pairs scaled by cell-type population.

(K) Bubble plot showing the ligand-receptor interactions between neurons and glia.

(L) Heatmap showing the interaction weight between cell types of the APP signaling pathway.

(M) Chord diagram showing the interaction weight between cell types of the NRXN signaling pathway.

(N) Bubble plot showing the ligand-receptor interactions between glial cell types.

(O) Heatmap showing the centrality scores/importance of cell groups as senders, receivers, mediators, and influencers in the PTN signaling pathway and MK signaling pathway.

(P) Chord diagram showing the significant interaction pairs involved in the NOTCH signaling pathways.

We identified a distinct cell population characterized by high expression of EGFR, OLIG1, and OLIG2 genes, but with limited expression of OPC markers (PDGFRA, CSPG4, PCDH15), oligodendrocyte markers (MBP, CTNNA3, FOXO4), and astrocyte markers (GFAP, S100B, SLC6A11) (Figure 4C). Jaccard similarity analysis suggested that these cells share transcriptomic signatures with glial progenitors identified in cortical macroglial cells from human brain tissue and show low similarity with astroglia and oligodendroglia⁷³ (Figure 4D). Consequently, we classified these cells as human GPCs. To further validate this identity, we performed immunostaining for ETV4, a recently identified GPC marker,⁷⁴ and observed the presence of both hN⁺/ETV4⁺ cells (Figure 4E). In addition, we also found hNG2⁺/ETV4⁺(Figure S4H), aligning with the scRNA-seq findings (Figure 4C). However, we did not observe a high THY1 expression in our GPC population, in contrast to the results reported by Liu et al⁷⁴ (Figure S4I). In addition, we dissected the astroglial populations, identifying five distinct stages of astroglial development (Figure 4F). Pseudotime analysis suggested that clusters Astro 2 and Astro 3 represented more advanced stages of astroglial maturity (Figure 4G), with Astro 3 characterized as a protoplasmic astroglial subtype (GFAP⁻/SLC1A3⁺) and Astro 2 as a mature fibrous astroglial subtype (GFAP⁺/SLC1A3⁻, Figure 4H).⁷³ Moreover, Astro 1 and Astro 4 clusters shared gene-expression signatures with astroglial progenitor cells from the human hippocampus⁷⁵ (Figure 4I). Notably, these astroglial stages did not align with signatures associated with cell death (AST3 in Figure 4I), further indicating their alignment with early developmental stages of the human brain.

To investigate the transcriptional states of transplanted microglia, we performed subclustering analysis and identified four distinct microglial subclusters (Figure S4J). Micro 0 and Micro 1 represented homeostatic microglia, characterized by high expression of canonical homeostatic markers such as TMEM119, P2RY12, and SALL1 (Figure S4K) and together accounted for over 95% of the total microglial population (Figure S4L). This is consistent with our immunostaining data showing that the majority of co-transplanted microglia remain in a homeostatic state (Figure 1K). Micro 3 was composed of proliferative microglia marked by genes such as MKI67, TOP2A, and PCNA (Figure S4K). One distinct cluster, Micro 2, displayed a transcriptional profile consistent with border-associated macrophages (BAMs), marked by expressing MRC1, CD163, and LYVE1.^76,77 Overall, this analysis reveals that the co-transplanted microglial population is largely homeostatic but exhibits notable heterogeneity. Interestingly, we detected MBP expression in the microglial population (Figure 4C), which is primarily contributed by the Micro 2 subcluster (Figure S4M). This suggests a functional role for Micro 2 in phagocytosing MBP⁺ oligodendrocytes or myelin debris. Supporting this, Micro 2 also showed elevated expression of phagocytosis-related genes (e.g., ITGAM, AXL, RAC2, SQSTM1, and LAMP1/2) (Figure S4M). Consistent with our findings, MBP⁺ microglia have also been observed in human brain tissue datasets. Analysis of the human microglia (HuMicA)⁷⁸ revealed enrichment of MBP-expressing microglia in white matter regions, where other oligodendrocyte lineage genes (e.g., OLIG1, OLIG2, MAG, MOG) are also expressed (Figure S4N), suggesting a specialized microglial state associated with myelinated regions. To compare the in vivo co-transplantation model with in vitro culture microglia-containing organoids, we projected publicly available scRNA-seq datasets from other microglia-containing organoid studies^71,79–81 onto our co-transplantation reference atlas (Figure S5A). These analyses revealed that none of the examined in vitro models were able to fully recapitulate the cellular diversity and glial complexity observed in our co-transplanted chimeric brains, particularly in terms of astrocyte subtypes and forebrain GPCs (Figure S5B). Notably, the organoid models exhibited a higher proportion of neural progenitor populations (e.g., radial glia and NPCs) and a lower abundance of glial cell types, even in the 3-month-old organoids that were timing-matched.

Cell-cell communication analysis in chimeric brains reveals robust human glial-neuronal and macroglial-microglial interactions

The presence of both human neurons and diverse glial populations in chimeric brains offers a fresh avenue for scrutinizing their specific ligand-receptor interactions via cell-cell communication (CCC) analysis of scRNA-seq data (Figure 4J). Regarding neuronal-glial interactions, inferred ligand-receptor pairs (Figures 4K and S5C) could be sorted into distinct signaling pathways (Figure 4K). This analysis pinpointed several crucial pathways implicated in neuronal-glial interactions, including the amyloid precursor protein (APP), pleiotrophin (PTN), midkine (MK), and neurexin (NRXN) signaling pathways. Notably, the APP signaling pathway exhibited the highest signaling probability between microglia and neurons (Figure 4L). These findings hold high relevance, as the APP pathway is recognized for its role in regulating microglia phenotypes.⁸² Communication within the NRXN signaling pathway was notably enriched between neurons and astroglia (Figure 4M). Interactions between NRXN-calsyntenin (CLSTN) play pivotal roles in excitatory and inhibitory synaptogenesis during development,^83–85 and NRXNs and neuroligins (NLGNs) constitute a critical pair of synaptic adhesion molecules crucial for synapse development and function.⁸⁶ These findings underscore that the hMAN model faithfully mirrors human neuroglial interactions.

Next, we explored glial-glial interactions (Figures 4N and S5D). The secreted phosphoprotein 1 (SPP1; osteopontin) signaling pathway was detected within the microglial population and between microglia and astroglia (Figure 4N). This aligns with previous studies indicating that SPP1 signaling fosters microglial growth and synaptic pruning during development.⁸⁷ Additionally, it facilitates interactions between microglia and astroglia, promoting astrogliosis and migration.⁸⁷ We also noted the presence of the pleiotrophin (PTN) signaling pathway within and between astroglial and GPC interactions (Figure 4O). Astroglia were found to have significant roles in the PTN signaling pathway and serve as key recipients in the MK signaling pathway (Figure 4O), both of which are heparin-binding cytokines associated with differentiation and growth.^88–90 The identification of the interaction between PTN and the receptor protein tyrosine phosphatase-β/ζ (PTPRZ1) within GPCs, as well as between astroglial cells and GPCs (Figure 4N), reinforces previous findings that both autocrine and paracrine sources of PTN contribute to the sustained self-renewal and homeostatic maintenance of human GPCs and OPCs.⁹¹ PTN also acts as an angiogenesis factor,⁹² elucidating the observed correlation with angiogenic astroglial subtypes in the human brain (Figure 4I) and robust PTN signaling within astroglia (Figure 4O). Furthermore, PTN/MK signaling pathways between astrocytes and neurons (Figure 4K) suggest crucial roles of astroglia in neurogenesis and neurite growth.^93,94 A recent preprint has also underscored the importance of astrocyte-secreted PTN in maintaining neuronal dendrite morphology and synaptic circuitry during normal and Down syndrome brain development.⁹⁵ GPCs play a predominant role in the NOTCH signaling pathway among all other cell types (Figure 4P). This pathway is crucial for gliogenesis and specifying cell fate.⁹⁶ We observed that the delta (DLL) signals from neurons interact with NOTCH receptors from both GPCs and astrocytes. Similarly, GPC-secreted DLL or Jagged (JAG) signals interact with astrocytes and other GPCs through the NOTCH signaling pathway. This finding aligns with previous knowledge that Delta-NOTCH signaling directs neuron and glial differentiation.⁹⁷ Furthermore, we found that macroglial NOTCH activation by JAG ligands from all glial cell types underscores the importance of JAG-NOTCH signaling in maintaining a proliferative glial pool.⁹⁸

Notably, we observed specific NRXN1-NLGN3 interactions within the astroglial population, highlighting the importance of this pathway in controlling astroglial morphogenesis, synaptogenesis, and synaptic functions.⁹⁹ Our CCC analysis also identified NLGN3 as the only neuroligin family member uniquely expressed as a receptor on astrocytes, mediating their interactions with other cell types via NRXN molecules (Figures 4K and 4N). This finding aligns with recent transcriptomic profiling demonstrating that NLGN3 is the most abundant astrocyte-enriched neuroligin isoform,¹⁰⁰ suggesting its involvement in astroglial development by mediating intercellular communication.

To validate that the cell-cell interactions from the co-transplanted mouse brain reflect human brain development, we performed CCC analysis using a human fetal brain dataset.⁷² Although the cell-type proportions differ between the datasets, likely due to technical differences between their single-nucleus RNA-seq (snRNA-seq) dataset and our scRNA-seq dataset, both successfully capture the major cell types involved in human brain development (Figure S5E). To enable a more accurate comparison, we used transcriptomic age prediction (Figure S4G) to match our dataset with a human gestational week (GW) 24 dataset. This analysis revealed that key signaling pathways identified in the co-transplantation model (e.g., SPP1, PTN, NRXN, NOTCH, and APP pathways) were also present in the human dataset, supporting the relevance of our model (Figure S5F).

DISCUSSION

Previous studies have shown that mouse brains possess the ability to incorporate live human glial cells or neurons. These human cells can structurally and functionally integrate into the mouse brain, where they widely disperse, establish connections, and interact dynamically with the host murine brain cells, creating human-mouse chimeric brains.¹¹ In this study, we further demonstrate that by co-transplanting hPSC-derived PMPs and pNPCs, we can generate human-mouse chimeric brains that incorporate human microglia, macroglia, and neurons. Earlier research has indicated that engrafted PDGFRα-expressing human GPCs differentiate into astroglia or oligodendroglia, depending on the brain environment. In a hypo-myelinated context, using immunodeficient shiverer mice as hosts, human GPCs predominantly differentiate toward oligodendroglia.^54,61,62 Conversely, in myelin-intact immunodeficient mice, GPCs tend to differentiate more toward astroglia.^17,35,54 Consistent with these findings, we also observe that engrafted human pNPCs respond to the brain environment, primarily generating astroglial lineage cells in myelin-intact brains while developing into oligodendroglial lineage cells in myelin-deficient brains. The distribution of human cells in the hMAN and hMON models differs from that in glial chimeric mice, where donor-derived human macroglia^35,61 or microglia^13,101–103 uniformly and widely colonize the host tissue. This difference is likely attributable to the varying migration capabilities between neurons and glial cells in the postnatal brain,¹⁰⁴ as well as the specific injection sites we targeted.^11,26 Post transplantation, these pNPCs efficiently differentiate into neurons, which typically exhibit less migratory tendencies compared to GPCs or microglia. As shown in Figures 1F and 2B, we observed that the engrafted cells demonstrated better migration and a wider dispersion in immunodeficient shiverer mice compared to myelin-wildtype mice, despite being deposited into the same hippocampal sites. This suggests that the shiverer mouse brain environment may facilitate the differentiation of pNPCs toward the glial lineage, particularly GPCs (Figures S2H–S2I), which exhibited enhanced migration. Additionally, targeting the hippocampal formation or the cerebral cortex^19,105 may result in less dispersal compared to targeting locations such as the lateral ventricle.^18,37,49 While some cells migrate and integrate with mouse cells at the periphery of the human cell clusters, the majority of the cells form structures resembling implant organoids in appearance.^23–25,106 Importantly, the human microglia differentiated from co-transplanted PMPs colocalize with human neurons and macroglia. These human glia and neurons actively interact with each other in the hMAN and hMON chimeric mice. Notably, we observed that the MBP expression was in a subset of human microglia, which likely reflects phagocytosis of MBP-expressing oligodendrocytes rather than endogenous transcription. Previous studies have shown that microglia can retain transcripts from engulfed cells, including myelin-associated genes such as MBP and PLP1 in MS brains¹⁰⁷ and neuronal transcripts during development.^108,109 In our dataset, MBP expression was enriched in the Micro 2 subcluster, which also expressed high levels of phagocytic genes, supporting a phagocytic origin. HuMicA data further confirmed that MBP⁺ microglia are localized to white matter, where we also observed co-expression of OLIG1, OLIG2, MAG, and MOG, indicating a specialized microglial state in myelinated regions. These findings highlight the potential of our model for studying microglia-oligodendrocyte interactions. Table 1 summarizes the key characteristics of each chimeric model, including the host strain, graft composition, human PSC line combinations, major lineage outcomes, and recommended research applications.

Table 1.

A summary of co-transplantation models

			hPSC combinations
Name	Host strain	Graft composition	PMP	pNPC/vNPC	Expected lineage outcomes	Recommended applications
hMAN	Rag2−/−IL2rγ−/−hCSF1KI (immunodeficient with hCSF1 knockin to facilitate microglia survival)	pNPCs + PMPs (1:1)	ND2.0 iPSC	ND2.0 iPSC	glutamatergic neurons, macroglia (astroglia biased), microglia	studying glial-neuronal and glial-glial interactions, modeling neurodevelopmental disorders (e.g., ASD linked to NLGN3 R451C mutation)
			C5 iPSC	C5 iPSC
			ND2.0 iPSC	C5 iPSC
			UTY1 iPSC	ND2.0 iPSC
			ND2.0 iPSC	H9 ESC
			H9 ESC	H1 ESC
hMON	shi/shi × Rag2−/− (immunodeficient and hypo-myelinated)		C5 iPSC	C5 iPSC	glutamatergic neurons, macroglia (oligodendroglia biased), microglia	investigating microglia-oligodendrocyte interactions, myelination studies, and modeling demyelinating disease
			H9 ESC	C5 iPSC
			H9 ESC	H1 ESC
hMAN (w/o hCSF1KI)	Rag2−/−	ventral NPCs + PMPs (1:1)	ND2.0 iPSC	ND2.0 iPSC	GABAergic neurons, macroglia, microglia	exploring interneuron-microglia interactions
			H9 ESC	H9 ESC
			ND2.0 iPSC	H9 ESC
			H9 ESC	ND2.0 iPSC

In addition to creating chimeric mouse brains with various human neural cell types, another significant benefit of conducting co-transplantation is avoiding the necessity of using human CSF1 or IL-34 knockin mice.^101,110 The sustained survival of hPSC-derived microglia within the mouse brain requires transgenic expression of human CSF1 or IL-34 because previous studies have demonstrated that no human microglia survive when human CSF1 is replaced with murine Csf1.¹⁰¹ A recent study showed that transplanting intact cerebral organoids containing microglial progenitors can lead to the development of human microglia in vivo in host mice that are not humanized for CSF1, owing to the release of human CSF1 and IL-34 by the astrocyte in the organoids.²⁴ Consistently, we show that, when PMPs are co-transplanted with dissociated pNPCs, these PMPs also survive and differentiate into human microglia in chimeric mouse brains. Notably, in in vitro culture, the neural lineage cells in dorsal forebrain organoids are unable to release sufficient human CSF1 and IL-34 to support human microglia differentiation.²⁴ To further improve microglia-containing brain organoid models, we discovered that generating organoids with ventralized NPCs and further culturing under gliogenic culture conditions can eliminate the need for adding IL-34 or CSF1 to the culture. Previous studies have demonstrated the production of IL-34 and CSF1 from glial cells, particularly astroglial cells.^64,65 We have also observed robust expression of IL-34 and CSF1 in these organoids, which may explain why PMPs within these organoids can differentiate into microglia with phagocytic functions successfully. This new organoid model has the potential to significantly reduce the high cost associated with the long-term culture of microglia-containing organoids. As anticipated, when PMPs and ventralized NPCs are co-transplanted into Rag2^−/− immunodeficient mice, we observe human microglia, macroglia, and neurons, especially GABA⁺ inhibitory neurons in chimeric brains, providing opportunities to explore the interactions of human microglia with human interneurons.

In our hMAN mice, we present the following evidence demonstrating the generation of forebrain human GPCs from engrafted PAX6⁺ dorsal forebrain pNPCs: (1) scRNA-seq data revealed a glial population characterized by high expression of EGFR, OLIG1, and OLIG2, but low levels of astroglial and oligodendroglial markers; (2) comparative analysis with macroglial cells isolated from human brain tissue indicated that this cluster exhibits high transcriptomic similarity to human glial progenitors; and (3) immunostaining of the hMAN chimeric brain confirmed the presence of ETV4, a marker for GPCs, in human cells. It is important to note that, in contrast to a previous report regarding markers of human GPCs,⁷⁴ our analyses did not reproduce these findings, as we observed minimal THY1 expression within the human GPC population. Importantly, in comparison to the transplantation of intact organoids into the mouse brain,²⁵ engrafting dissociated human neural cells may also promote the diversification of human astroglia. In our hMAN brain, we observed significant heterogeneity among human astroglia. We identified five distinct astroglial subclusters exhibiting dynamic transcriptomic profiles, including GFAP⁺ fibrous astrocytes and GLAST⁺ protoplasmic astrocytes. Additionally, we identified a subpopulation of astrocyte progenitor cells by comparing them with astrocytes isolated from human brains at early developmental stages. Some subpopulations of astrocytes exhibit angiogenic features similar to those of human astrocytes, which correlates with the pro-angiogenic effects of the PTN signaling pathway predicted from our cell-cell interaction analysis.

The chimeric models present new avenues for investigating interactions between human neurons and glia in vivo under neurodevelopmental, neurodegenerative, and human-specific viral infection conditions. Our scRNA-seq analysis of hMAN mouse brains uncovers the intricate ligand-receptor interactions between human neuron-astroglia and astroglia-astroglia, highlighting the significant role of NRXN-NLGN signaling pathways. This aligns with recent reports analyzing RNA-seq datasets from GPCs isolated from human brain tissue,¹¹¹ and single-cell datasets from the developing human fetal cortex, and examining astroglial differentiation in cerebral organoids.¹¹² The communication between neurons and astrocytes is predominantly facilitated by secreted factors and cell adhesion molecules, with NRXN- and NLGN-mediated signaling playing key roles in this crosstalk.^113–115 NRXN-NLGN has been shown to serve as a crucial ligand-receptor pair driving human astroglia development.¹¹² Additionally, studies in mice indicate that NLGN and NRXN play essential roles in neuronal spinogenesis, synaptic formation, and astrocyte morphogenesis.^86,99 There are multiple members in the NRXN and NLGN families.^116,117 Our cell-cell interaction analysis reveals the crucial involvement of NRXN1-NLGN3 in both neuron-astroglia and astroglia-astroglia interactions. As such, our co-transplantation chimeric brain models open unparalleled avenues for exploring the intricate interactions between human neurons and astrocytes, as well as the dynamics within each cell type. Since human neural cells frequently display species-specific characteristics,^{19,33–36,118} these chimeric models hold great potential to deepen our insights into human neuronal-glial interplay, offering a more accurate reflection of the complexities at play in the human brain. Notably, the NLGN3 R451C mutation has been associated with susceptibility to autism and Asperger syndrome.¹¹⁹ Previous studies using transgenic mice and hiPSCs have predominantly focused on the impact of NLGN3 R451C on neuronal functions and show gain-of-function effects, such as enhanced excitatory synaptic strengths in neurons with NLGN3 R451C.^120,121 Our findings of the involvement of NRXN1-NLGN3 in astroglia-astroglia interactions underscore the importance of exploring not only neuronal functions but also astroglia-astroglia interactions. Given the high heterogeneity of human astroglia and robust cell-cell connections in hMAN mice, this model provides a valuable tool for studying how changes in cell adhesion molecules, such as the autism-linked NLGN3 R451C, impact neuronal-astroglial and astroglial-astroglial interactions. In addition, recent research has highlighted that glial dysfunction plays a significant role in the progression of Alzheimer’s disease (AD).^122,123 Co-transplantation of PMPs and pNPCs from early-onset AD iPSCs, such as Down syndrome iPSCs^124,125 or Tau 4R iPSCs,¹²⁶ capable of generating Aβ and/or tau pathologies, to develop hMAN and hMON mice could expose human microglia and macroglia to a brain environment with both intracellular and extracellular pathological Aβ and/or hyperphosphorylated tau proteins. Moreover, these chimeric models will also provide important tools for studying human-specific viral infections that result in neurological deficits, such as HIV-associated neurocognitive disorders (HANDs). With human neuronal-glial and glial-glial interactions detected in the chimeric brains, these models offer new opportunities to better understand the mechanisms underlying the pathogenesis of a variety of neurological diseases.

Limitations of the study

While our study mainly focuses on human glial-neuronal and glial-glial interactions, it is important to recognize that cross-species effects in the chimeric brain could modulate cellular behavior. Specifically, murine microglia in the host environment may influence the development and function of transplanted human cells. Future studies using murine-specific markers and spatial omics approaches will be essential to systematically dissect these cross-species interactions and refine the biological interpretation of human-mouse chimeric models. It will also be important to explore how various transplantation paradigms impact the development and functional maturation of xenotransplanted human brain cells in vivo. Several prior studies have shown that the duration of post-engraftment periods significantly influences xenotransplanted human brain cells both functionally and transcriptionally.^55,111 Therefore, applying functional characterization, scRNA-seq, spatial transcriptomics, and epigenomic profiling across developmental stages will be essential to more precisely characterize cell-type heterogeneity, maturation status, and model-specific transcriptional programs and to determine the differences in cell-cell interactions.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Peng Jiang (peng.jiang@rutgers.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• The RNA-seq and scRNA-seq datasets generated in this study have been deposited at NCBI GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• This paper does not report original code. Any additional information required to reanalyze the data reported in this work is available from the lead contact upon request.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-human nuclei	Millipore	RRID: AB_827439
Rabbit anti-NeuN (Neuronal nuclei)	Millipore	RRID: AB_2298772
Rabbit anti-OLIG2 (Oligodendrocyte transcription factor 2)	Phosphosolutions	RRID: AB_2492193
Rabbit anti-GFAP (Glial fibrillary acidic protein)	Millipore	RRID: AB_2109645
Rabbit anti-IBA1 (Ionized calcium-binding adapter molecule 1)	Wako	RRID: AB_839504
Mouse anti-MAP2 (Microtubule-associated protein 2)	Santa Cruz Biotechnologies	RRID: AB_1126219
Rabbit anti-GLS (Glutaminase)	Abcam	RRID: AB_2721038
Rabbit anti-GABA (Gamma-aminobutyric acid)	Millipore	RRID: AB_477652
Rabbit anti-TMEM119 (Transmembrane protein 119)	Invitrogen	RRID: AB_2648507
Mouse anti-CD44 (HCAM, Homing cell adhesion molecule)	Abcam	RRID: AB_305297
Rabbit anti-S100B (S100 calcium-binding protein B)	Sigma	RRID: AB_477499
Mouse anti-CD45 (PTPRC, protein tyrosine phosphatase receptor type C)	Invitrogen	RRID: AB_11063696
Rabbit anti-TBR2 (T-box brain protein 2)	Abcam	RRID: AB_778267
Mouse anti-CD43 (Cluster of differentiation 43)	Invitrogen	RRID: AB_763493
Rabbit anti-CD235 (Cluster of differentiation 235)	Invitrogen	RRID: AB_90757
Mouse anti-SOX2 (SRY-box transcription factor 2)	Santa Cruz Biotechnologies	RRID: AB_10842165
Rabbit anti-PAX6 (Paired box 6)	GeneTex	RRID: AB_1951119
Rabbit anti-NG2 (Neuron-glial antigen 2)	Millipore	RRID: AB_91789
Rat anti-MBP (Myelin basic protein)	Millipore	RRID: AB_94975
Rabbit anti-PU.1 (SPI1, Spi-1 proto-oncogene)	Invitrogen	RRID: AB_11150963
Mouse anti-bIIItubulin	Millipore	RRID: AB_2210524
Rabbit anti-NKX2.1(TTF1) (Thyroid transcription factor 1)	Abcam	RRID: AB_1310784
Mouse-NKX2.1(TTF1) (Thyroid transcription factor 1)	Sigma	RRID: AB_571072
Rabbit anti-ETV4 (ETS translocation variant 4)	Abcam	RRID: AB_3095819
Mouse anti-NG2 (Neuron-glial antigen 2)	Millipore	RRID: AB_94509
Rabbit anti-P2RY12 (Purinergic receptor P2Y, G-protein coupled, 12 protein)	Sigma	RRID: AB_2669027
Rabbit anti-Ki-67 (Marker of proliferation Kiel 67)	Thermo Fisher Scientific	RRID: AB_10979488
Rabbit anti-OCT4 (Octamer-binding transcription factor 4)	Cell Signaling Technology	RRID: AB_2167691
Rabbit anti-SOX9 (SRY-box transcription factor 9)	Cell Signaling Technology	RRID: AB_2665492
Mouse anti- MEIS2 (Meis homeobox 2)	Developmental Studies Hybridoma Bank	RRID: AB_2618843
Goat anti-M-CSF (Macrophage Colony Stimulating Factor)	R&D Systems	RRID: AB_355351
Mouse anti-IL-34 (Interleukin 34)	Abcam	RRID: AB_10711208
Rabbit anti-cFOS	Santa Cruz	RRID:AB_2106783
Guinea pig anti-IBA1 (Ionized calcium-binding adapter molecule 1)	Synaptic Systems	RRID:AB_2924932
Goat Anti-mouse 488	Thermo Fisher Scientific	RRID: AB_2534088
Goat Anti-mouse 594	Thermo Fisher Scientific	RRID: AB_2534079
Goat Anti-mouse 647	Thermo Fisher Scientific	RRID: AB_2535804
Goat Anti-rabbit 488	Thermo Fisher Scientific	RRID: AB_143165
Goat Anti-rabbit 594	Thermo Fisher Scientific	RRID: AB_2534073
Goat Anti-rabbit 647	Thermo Fisher Scientific	RRID: AB_2535813
Goat Anti-rat 594	Thermo Fisher Scientific	RRID: AB_10561522
Goat Anti-rat 594	Thermo Fisher Scientific	RRID: AB_10561522
Goat Anti-Guinea Pig 647	Thermo Fisher Scientific	RRID: AB_2535867
Alexa Fluor® 647 anti-mouse/human A2B5 Antibody	BioLegend	RRID: AB_2566394
Alexa Fluor® 488 anti-human SSEA-4 Antibody	BioLegend	RRID: AB_1089199
Brilliant Violet 421™ anti-human CD90 (Thy1) Antibody	BioLegend	RRID: AB_10933261
PE/Cyanine7 anti-human CD15 (SSEA-1) Antibody	BioLegend	RRID: AB_2783154
PE anti-human CD133 Antibody	BioLegend	RRID: AB_2734477
Rabbit anti-NeuN (Neuronal nuclei)	Millipore	RRID: AB_2298772
Rabbit anti-OLIG2 (Oligodendrocyte transcription factor 2)	Phosphosolutions	RRID: AB_2492193
Rabbit anti-GFAP (Glial fibrillary acidic protein)	Millipore	N/A
Rabbit anti-IBA1 (Ionized calcium-binding adapter molecule 1)	Wako	RRID: AB_ 839504
Mouse anti-MAP2 (Microtubule-associated protein 2)	Santa Cruz Biotechnologies	RRID: AB_1126219
Rabbit anti-GLS (Glutaminase)	Abcam	RRID: AB_2721038
Rabbit anti-GABA (Gamma-aminobutyric acid)	Millipore	RRID: AB_477652
Rabbit anti-TMEM119 (Transmembrane protein 119)	Invitrogen	RRID: AB_2648507
Mouse anti-CD44 (HCAM, Homing cell adhesion molecule)	Abcam	RRID: AB_305297
Rabbit anti-S100B (S100 calcium-binding protein B)	Sigma	RRID: AB_477499
Mouse anti-CD45 (PTPRC, protein tyrosine phosphatase receptor type C)	Invitrogen	RRID: AB_11063696
Rabbit anti-TBR2 (T-box brain protein 2)	Abcam	RRID: AB_778267
Mouse anti-CD43 (Cluster of differentiation 43)	Invitrogen	RRID: AB_763493
Rabbit anti-CD235 (Cluster of differentiation 235)	Invitrogen	RRID: AB_90757
Mouse anti-SOX2 (SRY-box transcription factor 2)	Santa Cruz Biotechnologies	RRID: AB_10842165
Rabbit anti-PAX6 (Paired box 6)	GeneTex	RRID: AB_1951119
Rabbit anti-NG2 (Neuron-glial antigen 2)	Millipore	RRID: AB_91789
Rat anti-MBP (Myelin basic protein)	Millipore	RRID: AB_94975
Rabbit anti-PU.1 (SPI1, Spi-1 proto-oncogene)	Invitrogen	RRID: AB_11150963
Mouse anti-bIIItubulin	Millipore	RRID: AB_2210524
Rabbit anti-NKX2.1(TTF1) (Thyroid transcription factor 1)	Abcam	RRID: AB_1310784
Mouse-NKX2.1(TTF1) (Thyroid transcription factor 1)	Sigma	RRID: AB_571072
Rabbit anti-ETV4 (ETS translocation variant 4)	Abcam	RRID: AB_3095819
Mouse anti-NG2 (Neuron-glial antigen 2)	Millipore	RRID: AB_94509
Rabbit anti-P2RY12 (Purinergic receptor P2Y, G-protein coupled, 12 protein)	Sigma	RRID: AB_2669027
Rabbit anti-Ki-67 (Marker of proliferation Kiel 67)	Thermo Fisher Scientific	RRID: AB_10979488
Rabbit anti-OCT4 (Octamer-binding transcription factor 4)	Cell Signaling Technology	RRID: AB_2167691
Rabbit anti-SOX9 (SRY-box transcription factor 9)	Cell Signaling Technology	RRID: AB_2665492
Mouse anti- MEIS2 (Meis homeobox 2)	Developmental Studies Hybridoma Bank	RRID: AB_2618843
Goat anti-M-CSF (Macrophage Colony Stimulating Factor)	R&D Systems	RRID: AB_355351
Mouse anti-IL-34 (Interleukin 34)	Abcam	RRID: AB_10711208
Rabbit anti-cFOS	Santa Cruz	RRID:AB_2106783
Guinea pig anti-IBA1 (Ionized calcium-binding adapter molecule 1)	Synaptic Systems	RRID:AB_2924932
Goat Anti-mouse 488	Thermo Fisher Scientific	RRID: AB_2534088
Goat Anti-mouse 594	Thermo Fisher Scientific	RRID: AB_2534079
Goat Anti-mouse 647	Thermo Fisher Scientific	RRID: AB_2535804
Goat Anti-rabbit 488	Thermo Fisher Scientific	RRID: AB_143165
Goat Anti-rabbit 594	Thermo Fisher Scientific	RRID: AB_2534073
Goat Anti-rabbit 647	Thermo Fisher Scientific	RRID: AB_2535813
Goat Anti-rat 594	Thermo Fisher Scientific	RRID: AB_10561522
Goat Anti-rat 594	Thermo Fisher Scientific	RRID: AB_10561522
Goat Anti-Guinea Pig 647	Thermo Fisher Scientific	RRID: AB_2535867
Alexa Fluor® 647 anti-mouse/human A2B5 Antibody	BioLegend	RRID: AB_2566394
Alexa Fluor® 488 anti-human SSEA-4 Antibody	BioLegend	RRID: AB_1089199
Brilliant Violet 421™ anti-human CD90 (Thy1) Antibody	BioLegend	RRID: AB_10933261
PE/Cyanine7 anti-human CD15 (SSEA-1) Antibody	BioLegend	RRID: AB_2783154
PE anti-human CD133 Antibody	BioLegend	RRID: AB_2734477
Critical commercial assays
Chromium Next GEM Single Cell 3' GEM, Library & Gel Bead Kit v3.1, 16 rxns	10x Genomics	PN-1000121
Chromium Next GEM Chip G Single Cell Kit, 48 rxns	10x Genomics	PN-1000120
Dual Index Kit TT Set A	10x Genomics	PN-1000215
RNeasy Mini Kit	Qiagen	74104
SuperScript™ IV VILO™ Master Mix	Thermo Fisher Scientific	11756050
TaqMan Fast Advanced Master Mix	Thermo Fisher Scientific	4444557
TruSeq RNA Library Prep Kit v2	Illumina	RS-122-2001
Deposited data
Bulk RNA-seq	Jin et al.14	GEO: GSE189227
Bulk RNA-seq	Dang et al.70	GEO: GSE283484
Bulk RNA-seq	This study	GEO: GSE280909
scRNA-seq	This study	GEO: GSE280910
Experimental models: Cell lines
ND2.0 iPSC	National Institute of Health (NIH)	hPSCreg cell line ID: NHLBIi003-A
C5 iPSC	Di-DS hiPSCs, Coriell Institute for Medical Research	Coriell’s cat. no. AG06872
UTY1 iPSC	hiPSC from Dr. Ying Liu	N/A
H1 ESC	Coriell Institute for Medical Research	WA01
H9 ESC	Coriell Institute for Medical Research	WA09
Experimental models: Organisms/strains
Mouse: Rag2−/−IL2rγ−/−hCSF1KI		N/A
Mouse: shi/shi × Rag2−/−		N/A
Mouse: Rag2−/−		N/A
Oligonucleotides
IL-34 (Interleukin-34)	Thermo Fisher Scientific	Hs01050926_m1
AIF1 (IBA1, Allograft Inflammatory Factor 1)	Thermo Fisher Scientific	Hs00610419_g1
CSF1 (Colony stimulating factor 1)	Thermo Fisher Scientific	Hs00174164_m1
GAPDH (human specific) (Glyceraldehyde-3-Phosphate Dehydrogenase)	Thermo Fisher Scientific	Hs04420697_g1
AIF1 (IBA1, Allograft Inflammatory Factor 1)	Thermo Fisher Scientific	Hs00610419_g1
CSF1 (Colony stimulating factor 1)	Thermo Fisher Scientific	Hs00174164_m1
GAPDH (human specific) (Glyceraldehyde-3-Phosphate Dehydrogenase)	Thermo Fisher Scientific	Hs04420697_g1
Software and algorithms
Adobe Photoshop	Adobe	N/A
Adobe Illustrator	Adobe	N/A
R	https://www.r-project.org	v4.3.1
FastQC	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/	v0.12.1
Trim Galore!	https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/	v0.6.10
STAR	https://github.com/alexdobin/STAR	v2.7.11a
RSEM	https://github.com/deweylab/RSEM	v1.3.1
DESeq2	https://bioconductor.org/packages/release/bioc/html/DESeq2.html	v1.42.0
Cell Ranger	https://www.10xgenomics.com/support/software/cell-ranger/downloads/previous-versions	v7.1.0
Seurat	https://satijalab.org/seurat	v5.0.3
GeneOverlap	https://bioconductor.org/packages/release/bioc/html/GeneOverlap.html	v1.38.0
monocle3	https://cole-trapnell-lab.github.io/monocle3/	v1.3.4
CellChat	https://github.com/jinworks/CellChat	v2.1.1
clusterProfiler	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html	v4.10.1

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human iPSC and hESC lines generation and culture

Five hPSC lines were used in this study, including three control hiPSC lines—ND2.0 (male), UTY1 (male), and C5 (female) (corresponding to control-1, control-2, and Di-DS hiPSCs, respectively, as reported in our previous studies^14,37)—and two hESC lines, H1 (male) and H9 (female) (Table 1). These lines were fully characterized through karyotyping, gene expression profiling, and PluriTest analysis (www.PluriTest.org)—a robust, open-access bioinformatic tool for assessing pluripotency in human cells based on gene expression profiles¹²⁷ as described in our previous studies.^14,37,56 The hiPSCs and hESCs were cultured on dishes coated with hESC-qualified Matrigel (Corning) in mTeSR Plus medium (STEMCELL Technologies) under feeder-free conditions, and tested free mycoplasma contamination. Both hiPSCs and hESCs were passaged once per week using the ReLeSR medium (STEMCELL Technologies).

Animals

All animal work was conducted without gender bias with the approval of the Rutgers University Institutional Animal Care and Use Committee. The animals used in this study were as follows: B6.129S6-Rag2^−/− (The Jackson Laboratory; 008449), B6.129S7-Rag1^−/− (The Jackson Laboratory; 002216), Rag2^−/− IL2rγ^−/− hCSF1^KI (The Jackson Laboratory; 017708), and shi/shi x Rag2^−/− (generated by Dr. Steven Goldman group at University of Rochester).

METHOD DETAILS

Differentiation and culture of PMPs, pNPCs, and NPCs

PMPs were generated from the control hiPSC and hESC lines using a previously established protocol.^{13,14,128,129} Yolk sac embryoid bodies (YS-EBs) were generated by treating them with mTeSR 1 medium (STEMCELL Technologies) supplemented with bone morphogenetic protein 4 (BMP4, 50 ng/mL), vascular endothelial growth factor (VEGF, 50 ng/mL), and stem cell factor (SCF, 20 ng/mL) for 6 days. To induce myeloid differentiation, the YS-EBs were plated on dishes with X-VIVO 15 medium (Lonza) supplemented with interleukin-3 (IL-3, 25 ng/mL) and macrophage colony stimulating factor (M-CSF, 100 ng/mL). Four to six weeks after plating, human PMPs emerged in the supernatant and continued to be produced for more than three months.

Human pNPCs were generated from hiPSCs and hESCs.⁵⁵ The pNPCs were cultured in a medium consisting of a 1:1 mixture of Neurobasal (Thermo Fisher Scientific) plus GlutaMax (Gibco) and DMEM/F12 (Hyclone), supplemented with 1× N2, 1× B27-RA (Thermo Fisher Scientific), FGF2 (20 ng/mL, Peprotech), CHIR99021 (3 μM, Biogems), human leukemia inhibitory factor (hLIF, 10 ng/mL, Millipore), SB431542 (2 μM), and ROCK inhibitor Y-27632 (10 μM, Tocris). The pNPCs were passaged with TrypLE Express (Thermo Fisher Scientific) once per week. pNPCs within 6 passages were used for organoid generation.

To obtain ventralized NPCs, the expanded pNPCs were dissociated into single cells using TrypLE Express (Thermo Fisher Scientific). Next, the pNPCs were cultured in suspension in the presence of ROCK inhibitor Y-27632 (10 μM) at the first day in ultra-low-attachment 6-well plates. To pattern these neurospheres towards a ventral forebrain fate, we treated them with purmorphamine (1 μM, Cayman Chem), an agonist of sonic hedgehog signal pathway for one week (Figure 1A). The media were replenished every day. After one week of patterning, the neurospheres were dissociated into single cells using TrypLE Express. Then, the NPCs were cultured in a medium consisting of a 1:1 mixture of Neurobasal (Thermo Fisher Scientific) and DMEM/F12 (Hyclone), supplemented with 1× N2, 1× B27-RA (Thermo Fisher Scientific), FGF2 (20 ng/mL, Peprotech), and ROCK inhibitor Y-27632.

Brain organoid culture

Brain organoids were generated by co-seeding 10,000 total cells (5,000 NPCs and 5,000 PMPs) per well into ultra-low-attachment 96-well plates in the presence of 10 μM Y-27632 (Tocris). Cells were cultured for 4 days in an aggregation medium consisting of a 1:1 mixture of PMP medium (X-VIVO 15 (Lonza) supplemented with 100 ng/mL M-CSF (Invitrogen), 25 ng/mL IL-3 (R&D Systems), 2 mM GlutaMAX (Invitrogen), and 0.055 mM β-mercaptoethanol) and NPC medium (1:1 Neurobasal and DMEM/F12 supplemented with 1× N2 (Thermo Fisher Scientific), 1× B27 w/o RA (Thermo Fisher Scientific), and 20 ng/mL FGF2). On day 4, organoids were transferred to ultra-low-attachment 6-well plates and cultured on an orbital shaker at 85 rpm in gliogenic medium composed of DMEM/F12, 1× N2, 10 ng/mL FGF2 (Peprotech), 10 ng/mL PDGF-AA (Peprotech), 10 ng/mL IGF1 (Peprotech), and 1 μM dibutyryl-cyclic AMP (cAMP, Sigma) for 14 days, with media changes every other day. From day 18 onward, organoids were maintained in neuronal differentiation medium consisting of a 1:1 mixture of Neurobasal and DMEM/F12 supplemented with 1× N2 (Thermo Fisher Scientific), 10 ng/mL BDNF (Peprotech), 10 ng/mL GDNF (Peprotech), 1 μM cAMP (Sigma), and 200 nM ascorbic acid (Sigma), with media changes every other day. All cultures were maintained at 37°C in a humidified incubator with 5% CO₂.

Cell transplantation

In our previous studies,^8,14 we tested different pNPC-to-PMP ratios for organoid development and found no significant differences in cell survival across these conditions. A 7:3 pNPC-to-PMP ratio was used to generate organoids that contained approximately 5% microglia after long-term culture, reflecting the differing proliferative capacities of pNPCs and PMPs. In the current study, we increased the PMP proportion to a 1:1 ratio to promote more robust human microglia representation within the chimeric brain. Below is a step-by-step protocol for neonatal brain transplantation of human PMPs and pNPCs.

1. Human PMPs and pNPCs/v-NPCs were counted and resuspended at a 1:1 ratio (PMP:NPC) in sterile PBS to achieve a final concentration of 100,000 cells/μL. The cell suspension was gently mixed to ensure uniform distribution and kept on ice until transplantation.

2. Transplantations were performed using a digital stereotaxic device (David Kopf Instruments) fitted with a neonatal mouse adapter (Stoelting).

3. Neonatal (P0) immunodeficient mouse pups (B6.129S7-Rag1^−/−, B6.129S6-Rag2^−/−, Rag2^−/− IL2rγ^−/− hCSF1^KI, and shi/shi x Rag2^−/−) were kept on ice for 4–5 min to induce brief hypothermic anesthesia prior to injection.

4. Each pup received bilateral injections at the following coordinates relative to bregma: Medio-lateral (ML): ±1.0 mm from mid-line; Anterior-posterior (AP): −2.0 mm from bregma; Dorsoventral (DV): −1.5 mm and −1.2 mm depths.

5. At each site, 0.5 μL of the cell suspension was slowly injected to minimize reflux and tissue disruption. A total of four injection sites (two per hemisphere) were used per animal.

6. After transplantation, pups were gently rewarmed on a heating pad until normal activity resumed, then returned to their home cage with the dam.

Tissue immunostaining, image acquisition, and analysis

Mouse brains were fixed in 4% paraformaldehyde and subsequently dehydrated by immersion in 20% and then 30% sucrose solutions. Brain organoids were fixed in 4% paraformaldehyde for 2 h then dehydrated by immersion in a 25% sucrose solution. Following dehydration, brain tissues and organoids were embedded in the OCT compound and frozen for sectioning. Cryosections with a thickness of 30 μm were obtained from mouse brains and with a thickness of 12 μm from brain organoids for immunofluorescence staining. For immunostaining, tissue sections were first blocked with a solution containing 5% goat serum in PBS with 0.8% Triton X-100 at room temperature for 1 h; brain organoids were blocked with a solution containing 5% goat serum in PBS with 0.2% Triton X-100. Primary antibodies (listed in the key resources table) were diluted in the blocking solution and incubated with the tissues overnight at 4°C. After primary antibody incubation, sections were washed with PBS and then incubated with secondary antibodies for 1 h at room temperature. After immunostaining, 7-month-old brain sections were performed with TrueBlack treatment as previously described.¹²⁸ Following another round of PBS washing, slides were mounted with anti-fade Fluoromount-G medium containing DAPI (1,4,6-diamidino-2-phenylindole dihydrochloride) from Southern Biotechnology. Fluorescent images were captured using a Zeiss 800 confocal microscope and the all-in-one fluorescence microscope BZ-x800 (Keyence).

To obtain a 3D reconstruction, images were processed by the Zen software (Zeiss). To visualize phagocytic function, super-resolution images in Figures 1 and 3 were acquired by Zeiss Airyscan super-resolution microscope at 63× with 0.2mm z-steps. To generate 3D-surface rendered images, super-resolution images were processed by Imaris software (Bitplane 9.9).

Low power images in Figures 1G, S1E, 2C, and S2A were obtained by tile scan by the Keyence BZ-x800 microscope and stitched by the Keyence analyzer software. To create the human cell distribution dot map, whole-brain montages of 12 equidistantly spaced sections, each 360 μm apart, were imaged using a Keyence microscope. Human cells (hN⁺) were identified and mapped within outlined brain sections. Faint, sporadic cells outside the core injection areas were manually enhanced for a more accurate representation of the engraftment.

Nearest neighbor distance (NND) analysis

The spatial coordinates of IBA-1⁺/hN⁺ human microglia were identified from immunostained brain sections using QuPath (version 0.6.0). Cell centroids were exported as x- and y-coordinates (in μm) for each region of interest. The NND for each human microglia was then computed using the nndist() function from the spatstat R package, which calculates the Euclidean distance from each cell to its closest neighboring human microglia. The resulting NND values were used to quantify microglial spatial distribution and tiling across different brain regions and timepoints.

Electrophysiology

Recordings were performed on 3-month-old mice that had been transplanted at P0 with GFP-expressing CAGG-pNPCs. Detailed procedures for acute hippocampal slice recordings have been described previously.^130,131 Briefly, mouse brains were rapidly extracted, and 360-μm sagittal hippocampal slices were prepared using a vibratome (Leica VT1200S) in ice-cold cutting solution containing (in mM): 206 sucrose, 11 D-glucose, 2.5 KCl, 1 NaH₂PO₄, 10 MgCl₂, 2 CaCl₂, and 26 NaHCO₃, saturated with 95% O₂/5% CO₂. Slices were recovered in artificial cerebrospinal fluid (ACSF) containing (in mM): 120 NaCl, 3.0 KCl, 1.2 MgSO₂, 1.0 NaH₂PO₄, 26 NaHCO₃, 2.0 CaCl₂, and 11 D-glucose, bubbled with 95% O₂/5% CO₂ at 33°C for 30 min, then maintained at room temperature for at least 1 h before recording. Recordings were conducted in a chamber perfused with oxygenated ACSF. GFP-positive neurons were identified using an Olympus microscope equipped with appropriate filters and LED illumination. Whole-cell patch-clamp recordings were obtained using glass pipettes (5–8 MΩ) filled with intracellular solution containing (in mM): 126 K-gluconate, 4 KCl, 10 HEPES, 0.05 EGTA, 4 Mg-ATP, 0.3 Na-GTP, and 10 phosphocreatine. Current-clamp recordings involved step current injections ranging from −50 to +130 pA in 10 pA increments. Voltage-clamp recordings of sodium and potassium currents were performed using step depolarizations from −70 mV to +40 mV in 10 mV increments (250 ms duration per step). Data acquisition and analysis were conducted using pCLAMP 10.7 (Molecular Devices).

RNA isolation and quantitative reverse transcription PCR

RNA extraction was done by the Qiagen Mini kit (Qiagen: 74104) and Qiagen shredder (Qiagen: 79656). Reverse transcription was also done with SuperScript IV VILO Master Mix (Thermo Fisher Scientific: 11756050). Real-time PCR was performed on the ABI 7500 Real-Time PCR System using the TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific). All primers are listed in the key resources table. The 2^−ΔΔCt method was used to calculate relative gene expression after normalization to the GAPDH internal control.

Bulk RNA sequencing library preparation and data analysis

Libraries for bulk RNA-seq were prepared using the Illumina TruSeqV2 kit (Illumina, San Diego, CA) following the manufacturer’s protocol. Sequencing was performed on a Novaseq X Plus with 150 bp paired-end reads. The sequenced reads were quality-checked with FastQC (v0.12.1) and trimmed for adaptor and low-quality sequences using Trim Galore (v0.6.10). Trimmed reads were mapped to the GRCh38 reference genome and GENCODE v44 primary assembly using STAR¹³² (v2.7.11a). Read count extraction and normalization were performed with RSEM¹³³ (v1.3.1), and differential expression analysis (DEG) was conducted using the DESeq2¹³⁴ R package (v1.42.0). Heatmaps were generated using the pheatmap R package (v1.0.12).

Single-cell RNA sequencing library preparation

Two hMAN Rag2^−/− IL2rγ^−/− hCSF1^KI chimeric mice were anesthetized and perfused with cold 1× Dulbecco’s Phosphate-Buffered Saline (DPBS). Brains were then dissected and dissociated using the Adult Brain Dissociation Kit (Miltenyi Biotec) according to the manufacturer’s instructions. Human cells were isolated using the Mouse Cell Depletion Kit (Miltenyi Biotec), following the manufacturer’s protocols. The isolated cell suspension from the two mice were subsequently loaded into the Chromium Controller respectively for single-cell RNA sequencing (scRNA-seq) library preparation, utilizing the Chromium Next GEM Single Cell 3^’ Kit v3.1 and Dual Index Kit TT Set A, in accordance with the manufacturer’s guidelines. Quality control of the scRNA-seq libraries was performed using an Agilent 2100 Bioanalyzer, and the libraries were sequenced on a NovaSeq 6000 system using S4 flow cells.

Single-cell RNA sequencing data analysis

Raw reads were mapped to the human (GRCh38, Ensembl 98, GENCODE v32) and mouse (mm10, Ensembl 98, GENCODE vM23) reference genomes using Cell Ranger (v7.1.0), with the EGFP sequence included. Cells were classified as human if more than 95% of the reads mapped to the human reference genome. Filtered human cells were retained for further analysis if they contained between 1,000 and 5,000 genes, had fewer than 30,000 unique molecular identifiers (UMIs), and exhibited mitochondrial content below 15%. Doublets and multiplets were removed using the scDblFinder R package.¹³⁵

Downstream analysis was conducted using the Seurat R package (v5.0.3).¹²⁶ Raw gene count matrices were normalized by regularized negative binomial regression using the SCTransform() function (vst.flavor = “v2”), which also identified the top 3,000 highly variable genes using default parameters. Dimensionality reduction was performed using principal component analysis (PCA) on the top variable genes. Clusters of cells were identified in PCA space through shared nearest-neighbor graph construction and modularity detection, implemented by the FindNeighbors() and FindClusters() functions, using a dataset dimension of 50 and a resolution parameter set to 0.5. Astrocyte subclusters were identified by first isolating the astrocyte cells (n = 11,100) from the complete dataset. These cells were then processed using the SCTransform method, followed by dimensionality reduction (npcs = 50, ndims = 30). Clustering was subsequently performed with a resolution parameter set to 0.2.

For age prediction, we employed Seurat’s reference mapping method, which infers a predicted transcriptomic age based exclusively on gene expression similarity, by projecting our data and 10-week-old and 5-month-old in vitro cultured and 5, 6, and 8-month-old in vivo transplanted glia-enriched cortical organoid data from Wang et al., 2024²⁵ onto the human brain reference dataset across various brain development ages.⁷² The predicted transcriptomic age distribution before birth among different groups was compared using the paired Wilcoxon rank-sum test after randomly subsetting 100 cells per group. The Jaccard similarity index was calculated using the GeneOverlap R package (v1.38.0), with similarity significance calculated from Fisher’s exact test. Pseudotime analysis was performed using the monocle3 R package (v1.3.4)^136–138 with default parameters. The root cell type for the start of the trajectory was chosen manually by finding an endpoint in the astrocyte progenitor population in cluster Astro 1. Reference mapping and cell-type prediction of the microglia-containing organoids onto our dataset were performed using Seurat’s FindTransferAnchors() and TransferData() functions. Unimodal UMAP projection was conducted with MapQuery() and visualized using the ggplot2 R package. To study cell-cell communication between transplanted human cell types, we employed the CellChat R package (v2.1.1)^139,140 to infer ligand-receptor pairs between cell types. In brief, we subset only signaling genes from the expression matrix, identified over-expressed ligands and receptors, and determined over-expressed ligand-receptor interactions. Biologically significant cell-cell communications were inferred using the statistically robust triMean method. We then computed the communication probability at the signaling pathway level by summarizing the communication probabilities of all ligand-receptor interactions associated with each pathway. GO over-representation analysis and gene set enrichment analysis were performed by the clusterProfiler R package (v4.10.1).¹⁴¹

QUANTIFICATION AND STATISTICAL ANALYSIS

All data are represented as mean ± SEM. When only two independent groups were compared, significance was determined by using a two-tailed unpaired t test with Welch’s correction. A pp-value of <0.05 was considered significant. All statistical information, including sample sizes (n) and definitions of statistical significance, is reported in the figure legends. Descriptions of software and complete experimental procedures are provided in the STAR Methods section. All the analyses were done in GraphPad Prism v.9.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116794.

Highlights.

• Co-engrafting hPSC-derived neural and microglial progenitors forms chimeric brains

• Human microglia exhibit homeostatic features and prune human synapses in vivo

• Single-cell RNA-seq identifies diverse astroglial and glial progenitor populations

• NRXN-NLGN3, SPP1, and PTN-MK pathways mediate human neuron-glia communication