Multimodal Single-Cell Analysis Reveals Physiological Maturation in the Developing Human Neocortex

Abstract

In the developing human neocortex, progenitor cells generate diverse cell types prenatally. Progenitor cells and newborn neurons respond to signaling cues, including neurotransmitters. While single-cell RNA sequencing has revealed cellular diversity, physiological heterogeneity has yet to be mapped onto these developing and diverse cell types. By combining measurements of intracellular Ca²⁺ elevations in response to neurotransmitter receptor agonists and RNA sequencing of the same single cells, we show that Ca²⁺ responses are cell-type-specific and change dynamically with lineage progression. Physiological response properties predict molecular cell identity and additionally reveal diversity not captured by single-cell transcriptomics. We find that the serotonin receptor HTR2A selectively activates radial glia cells in the developing human, but not mouse, neocortex, and inhibiting HTR2A receptors in human radial glia disrupts the radial glial scaffold. We show highly specific neurotransmitter signaling during neurogenesis in the developing human neocortex and highlight evolutionarily divergent mechanisms of physiological signaling.

Keywords: calcium imaging; differentiation; human neocortical development; intermediate progenitor cells; neurogenesis; neurotransmitter; radial glia; radial glia scaffold; serotonin; single-cell RNA sequencing.

Figure 1.. Neurotransmitter Receptor Expression during Neurogenesis in the Developing Human Neocortex

(A–F) Pseudoage analysis of the excitatory neuronal lineage (A) Marker gene expression of radial glia cells (SLC1A3, orange shading), intermediate progenitor cells (EOMES, yellow shading), and excitatory neurons (SLA, green shading). Expression trajectories of NMDA receptor (NMDAR) subunits (B), and GABA_A receptor subunits (C), the purinergic receptor P2RY1 (D), AMPA receptor (AMPAR) subunits (E), and the serotonergic receptor HTR2A (F, left). Correlation of HTR2A expression values with age (PCW) in radial glia cells (F, right). (G–J) Enrichment of serotonergic receptor HTR2A in specific progenitor cell types in the developing human neocortex using single molecule in situ hybridization (smFISH). (G) smFISH for HTR2A was combined with immunohistochemistry against CRYAB (H), HOPX (I), and PPP1R17 and Ki67 (J). Arrows indicate colocalization, large arrowheads lack of colocalization. (K) Schematic showing HTR2A pseudoage expression trajectory in human and mouse. See also Figure S1 and Tables S1 and S2.

Figure 2.. Patch-Clamp Recordings in Cortical Slices Show In Situ Receptor Functionality

(A) Schematic diagram shows whole-cell patch clamp recordings with local application of agonists in slices. (B) Cells were loaded with biocytin after recording and immunostained with cell-type-specific markers (arrows). Note that the vRG is coupled to both a vRG and an oRG cell (arrowheads, middle), while the oRG is coupled to another oRG cell (arrowheads, right). (C) Example traces of currents in response to agonists. Neurons (red) were held at −60 mV except for NMDA + glycine (held at +40 mV). All oRGs (black) and vRGs (green) were recorded at a holding potential of −70 mV. See also Figure S2 and Table S2.

Figure 3.. Ca²⁺ Imaging Reveals Lamina-Specific Responses to Neurotransmitter Receptor Agonists

(A) Schematic of the experimental workflow for Ca²⁺ imaging of dissociated cells. (B) Images of [Ca²⁺]_i (pseudocolored) pre and post agonist application in cells dissociated from the GZ. Arrowheads indicate Ca²⁺ elevations of one representative cell in response to specific agonists. (C) Traces of [Ca²⁺]_i changes in response to agonist dosing. (D) Summarized results of the percentage of cells showing Ca²⁺ elevation upon agonist stimulation in GZ and CP and SP, respectively (PCW14–22, binary quantification). N = 10, n = 2,613 cells. (E–G) Summarized results of Ca²⁺ responses to the panel of neurotransmitter receptor agonists at PCW8–12 (N = 3, n = 608 cells) (E), and in the GZ (F) and in the CP and SP (G) by age range in midfetal development. No response indicates cells that only responded to ionomycin. A significant difference in GABA responses between PCW14–16 and PCW19–22 in CP and SP was found using multiple t tests (Holm-Sidak method). Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, dots indicate biological replicates. (H) Single-cell Ca²⁺ responses to different agonists at different stages of development were co-clustered to reveal 11 distinct physiological clusters (n = 3,221 cells). GZ, germinal zone. SP, subplate. See also Figure S3 and Table S2.

Figure 4.. Mapping Ca²⁺ Responses onto Cell Types at the Single-Cell Level

(A) Schematic of experimental workflow using the Polaris microfluidic system. (B) Representative images of [Ca²⁺]_i changes (pseudocolored) pre and post agonist application. Arrows indicate cell responding to agonist with Ca²⁺ elevation. (C) Representative traces of Ca²⁺ changes in response to agonists. (D) Unbiased clustering of cells into 10 molecular clusters based on their transcriptomes represented in t-distributed stochastic neighbor embedding (tSNE) space. (E) Feature plots show responses to agonists (pink). (F) Feature plots showing co-clustering of cells captured on Polaris or C1 microfluidic chips revealing 24 transcriptomic clusters. (G) Heatmap shows the percentage of cells in each C1-Polaris transcriptomic cluster (row, cluster number indicated in brackets) responding to agonists (columns). Only clusters with more than 5 Polaris cells were included. RG, radial glia; L, layer; CGE, caudal ganglionic eminence; MGE, medial ganglionic eminence; UL, upper layer; DL, deep layer; OPC, oligodendrocyte progenitor cell; EN, excitatory neuron; IN, interneuron. See also Figure S4 and Tables S3 and S4.

Figure 5.. Signaling through HTR2A Regulates Radial Glia Morphology

(A) BrdU incorporation and Ki67-positive proliferating cells are not changed after HTR2A receptor activation or inhibition in cortical slices. (B) Quantification of proliferation assay shown in (A). (C) Treatment of cortical slices with HTR2A antagonist EMD 281014 leads to loss of radial fiber organization as visualized by vimentin staining and GFP-staining of cells infected with a CMV-GFP adenovirus. (D) Higher magnification of inserts shown in (C) highlighting the lack of radial vimentin fiber organization after antagonist treatment. (E) CMV-GFP-positive fibers in OSVZ are oRGs based on co-staining with Hopx. (F and G) Quantification of fiber length reveals a significant reduction in average fiber length normalized by slice length (F) and fiber length normalized by slice length (G) in the EMD 281014-treated cortical slices compared to the control using the Wilcoxon rank-sum test. Data are represented as mean ± SEM. *p < 0.05, ****p < 0.0001. See also Videos S1 and S2.

Figure 6.. Physiological Response Patterns Predict Cell Identity

(A) The frequency of responses to agonist application varies as a function of pseudoage (top). See detection frequencies of marker genes (bottom). Major cell types are highlighted by shading: radial glia cells, orange; intermediate progenitor cells, yellow; excitatory neurons, green. (B) Transcriptomically defined cell types, excluding clusters “other,” which are outliers. (C) Predicted cell-type assignments of a classifier trained on the physiological response patterns in a multimodal dataset, relative to their transcriptomically defined cell types (B). (D) Predicted cell type based on responses to agonists of cells analyzed in bulk imaging experiments (Figures 3E and 3F). Percentage of cells relative to the total number of cells responding to at least one agonist is shown. CTX, neocortex. See also Figure S5 and Table S5.

Figure 7.. Physiological Heterogeneity Exists within Transcriptomically Defined Cell Types

(A) Single-cell Ca²⁺ responses to different agonists analyzed on microfluidic chips and on physiological rig were co-clustered to reveal physiological types (P). (B) Heatmap showing the relative responses to each neurotransmitter (columns) in the different physiological types (P, rows). Numbers in heatmap indicate number of cells in each cluster analyzed on microfluidic chips at the single-cell level. (C) Heatmap highlighting representative genes that are expressed in all mature upper layer excitatory neurons or enriched in physiological cluster P-1 or P-2. (D–F) Differential gene expression between distinct physiological types within one cell type. Top Gene Ontology terms enriched in differentially expressed genes between physiological types P-1 and P-2 in mature upper layer excitatory neurons (D), between physiological types P-5 and P-7 in dividing progenitors (E), and between physiological types P-6 and P-10 in intermediate progenitor cells (F). See also Figure S6 and Table S6.