Generation of FEV reporter hPSC

To identify premature human SN, an FEV-EGFP reporter cassette was introduced to the coding region immediately upstream of the stop codon of the endogenous FEV gene by CRISPR/Cas9-mediated homologous recombination (Fig. 1A). Edited hPSC clones were screened by PCR assay. As shown in Fig. 1B, clones (2, 3, 5, 7, 8, 11, 12, 14, 15, 16) exhibited two 2000 bp-band (5FR and 3FR), indicating the in-frame integration of the reporter cassette into the FEV locus. Clone 7 showed a 5000 bp-band (FR), indicating homozygous integration of the FEV reporter cassette. The top 10 potential off-target sites (Additional file 1: Table S3) were evaluated in the homozygous Clone 7 by Sanger sequencing, and no insertion/deletion (indel) mutation was detected in these sites, suggesting the high specificity of this gene editing system (Additional file 1: Figure S1). The homozygote reporter cell line (H9-Clone#7) was then selected for the entire study unless specially indicated. The Sanger sequencing showed the sequences of 5’-end junction, the T2A-EGFP junction and the 3’-end junction between the original FEV DNA sequence and the inserted sequence (Fig. 1C), indicating the in-frame integration of the reporter cassette. The FEV-EGFP reporter hPSC expressed pluripotent markers (Fig. 1D) and was able to form teratoma (Fig. 1E), indicating its pluripotency properties. Additionally, the engineered reporter hPSC showed normal karyotype (Fig. 1F). These data suggested that T2A-EGFP-CMV-puro cassette was correctly inserted into the endogenous FEV gene of the hPSC with high efficiency and specificity, and the FEV-EGFP reporter hPSC maintained pluripotency with the normal karyotype.

Fig. 1

Generation of FEV reporter hPSC. A Schematic diagram for strategy of CRISPR-Cas9-mediated knockin of reporter cassette into human FEV locus. B Genotyping of the edited hPSC (H9) clones with the primer sets (5FR, 3FR, FR) to identify the edited homozygous clone. (The 5FR and 3FR primer sets spanned the nucleotides outside the homologous arm to the internal reporter/selection cassette, aiming to verify the integration of the insertion cassette into the FEV locus; the FR primer set, flanking approximately 1500 bp of the stop codon of WT FEV gene, was designed to distinguish between homozygous and heterozygous clones) C Representative Sanger sequencing chromatograms for the homozygous Clone#7. D Immunofluorescence staining of the pluripotency markers (OCT4, NANOG, SSEA4, Tra-1-60, Tra-1-81) for the FEV-EGFP reporter cell line (H9-Clone#7). E H&E staining for the FEV-EGFP reporter cell line (H9-Clone#7)-derived teratoma. F Karyotyping analysis for the FEV-EGFP reporter cell line (H9-Clone#7). Scale bar for D, E 100 μm

Monitoring FEV expression during human SN differentiation via FEV reporter system

In order to determine the timing of FEV expression, the FEV-EGFP reporter hPSC were differentiated into SN according to our established protocol [5] (Fig. 2A). Co-activation of WNT and SHH signaling pathways induced the reporter hPSC to differentiate into serotonergic neural progenitors (SNP) expressing SOX1, GATA2 and NKX2.2 (Fig. 2B). Subsequently, SNP continued to differentiate into SN. As shown in Fig. 2C–E, FEV expression initiated at day 28 of differentiation towards SN, preceding the emergence of TPH2 (a specific marker for mature SN) by about one week. Hence, we defined the FEV + /TPH2- cells as immature SN. The proportion of EGFP + cells was positively corelated to the duration of differentiation (Fig. 2C, D), suggesting a progressive upregulation of FEV expression accompanying the maturation of SN over time. At day 56, all of the EGFP + cells were co-labeled with TPH2 (Fig. 2C, E), indicating that FEV-EGFP specifically labeled the early-stage SN. Meanwhile, the EGFP + cells were also co-labeled with other typical serotonergic markers, such as GATA3 (Fig. 2F) and 5-HT (Fig. 2G), validating the serotonergic identity of FEV + cells. Additionally, the engineered reporter hPSC showed similar differentiation potential with the parental hPSC (H9 cell line, Additional file 1: Figure S2), indicating that the differentiation potential of the reporter hPSC was not compromised by the gene-editing process. Therefore, these data demonstrated that the onset of FEV expression occurred 7 days prior to the expression of TPH2, and FEV expression gradually increased with the maturation process of SN over time.

Fig. 2

Investigation of the expression timing of FEV during SN differentiation. A Schematic diagram of the strategy to derive SN from the FEV reporter hPSC. B Immunofluorescence staining of SOX1, GATA2 and NKX2.2 to indicate reporter cell line (H9-Clone#7)-derived SNP at day 21 of differentiation. C. Immunofluorescence staining of EGFP and TPH2 to validate the expression time of FEV and TPH2. D, E Time course of FEV (D) and TPH2 (E) expression during SN differentiation process. F Immunofluorescence staining of GATA3 at day 42 of differentiation. G Immunofluorescence staining of EGFP and 5-HT at day 56 of differentiation. Scale bar for B, C, F, G 100 μm. n = 3 independent experiments. All data are presented as means ± SEM

Validation of the role of FEV as the specific marker for immature SN

Since the FEV-EGFP reporter system could faithfully indicate FEV expression in early-stage SN (Fig. 2), we next analyzed the FACS-purified EGFP + cells. With the increasing duration of differentiation, an increasing proportion of FEV-EGFP-positive cells could be obtained through FACS (Fig. 3A). The purified cells at day 35 of differentiation were reseeded onto poly-L-ornithine (PO)/laminin/Matrigel-coated coverslips, and exhibited rapid attainment of a healthy state and the extension of nerve fibers within 16 h following reseeding (Fig. 3B). Moreover, the percentage of EGFP + cells (95.63% ± 1.252%, n = 10) in the purified population was significantly higher than that (33.64% ± 0.796%, n = 10) in the mixed cell population (Fig. 3C, D), indicating that the FEV-EGFP reporter enabled us to purify early-stages SN from heterogeneous differentiation cultures in vitro. 3 weeks after reseeding (day 56), all of the EGFP + neurons were co-stained with 5-HT (Fig. 3E, F), indicating the role of FEV as a terminal selector for SN. Subsequent functional analysis was conducted on the purified FEV-EGFP-labeled mature SN (at day 56). These cells released 5-HT in response to escitalopram oxalate (EO, a selective serotonin reuptake inhibitor) (Fig. 3G, H). Additionally, when subjected to a 40-ms voltage step ranging from − 40 mV to + 30 mV, the FEV-EGFP-labeled SN (at day 56) exhibited characteristic inward Na + and outward K + currents (Fig 3I, J), indicating their functional properties as mature SN. Therefore, the FEV-EGFP reporter enabled us to obtain highly purified immature SN, which underwent maturation within 3 weeks and were capable of showing typical features of mature SN.

Fig. 3

Purification and validation of FEV-EGFP reporter system-derived SN. A FACS profiles of FEV-EGFP reporter cell (H9-Clone#7)-derived cells during differentiation from day 28 to day 42. B Representative brightfield image of the purified EGFP + cells 16 h after FACS-based purification and reseeding. C, D Immunofluorescence staining (C) and quantification (D) of EGFP and 5-HT for unsorted and sorted SN at day 42 of differentiation. E, F Immunofluorescence staining (E) and quantification (F) of EGFP and 5-HT for sorted SN at day 56 of differentiation. G Extracellular 5-HT released by unsorted or sorted cells before and after treatment with EO. H Evaluation of the sensitivity of the pharmacological responses to EO by unsorted or sorted SN. I Input voltage program (protocol) and current traces evoked by 40-ms depolarizing voltage stepped from − 40 to + 30 mV in 5-mV increments. J Current–voltage curves for voltage-gated sodium and potassium currents (n = 6). Scale bar for B: 100 μm; scale bar for C, E: 50 μm. n = 3 independent experiments. All data are presented as means ± SEM. Statistical comparisons were performed by Student’s t test. *P < 0.05; **P < 0.01; ***P < 0.001, n.s.: no significance

Alterations of transcriptomic profiles during human SN differentiation process

To comprehensively reveal the dynamic changes of transcriptomic profiles during human SN differentiation, we investigated 4 key stages of the differentiation process by RNA-seq analysis (Fig. 4A). Cells at Stage 3 and 4 were purified by FACS via FEV-EGFP reporter and TPH2-EGFP reporter respectively. Principal-component analysis (PCA) remarkably separated the samples into 4 clusters (Fig. 4B). The expression of the key differentiation stage genes, including "pluripotency genes", "neurogenesis genes", "synapse genes", “axon guidance genes” and "serotonergic genes", were indicative of their expression patterns during specific developmental stages (Fig. 4C–G). Volcano plots showed thousands of differentially expressed genes (DEGs) expressed in the individual key stage during differentiation (Fig. 4H). Kyoto Encyclopedia of Genes and Genomes pathway (KEGG) enrichment analysis was performed to identify functional categories of the DEGs in the individual key stage of differentiation. The upregulated DEGs at day 28 and day 42 were enriched in synapse- and serotonergic-associated pathways, indicating the determination of serotonergic fate with FEV expression at day 28 and progressive neuronal maturation at day 42 (Fig. 4I, J). These data revealed dynamic changes of transcriptomic profiles during human SN differentiation.

Fig. 4

Analysis of transcriptomic profiles during human SN differentiation process. A The schematic of sample collection from four stages during SN differentiation for smart RNA-seq (Day0: hPSC stage; Day21: SNP stage; Day28: immature SN stage; Day42: mature SN stage). B Principal component analysis (PCA) score plot for the samples. C–H Gene expressions of pluripotency (C), neurogenesis (D), synapse (E), axon guidance (F) and serotonergic genes (G) at 4 key stages during SN differentiation. H. Volcano plots of the DEGs. The up-regulated, down-regulated and unchanged metabolites are presented by red, blue, and gray dots, respectively. I, J The functional classification of the DEGs (I: day 28 vs. day 21; J day 42 vs. day 28) based on Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis

Identification of the potential key molecules crucial for SN fate specification

While we demonstrated that the cells committed to a serotonergic fate as early as day 28 of the differentiation process (Figs. 3, 4), the mechanisms responsible for initiating serotonergic fate determination before this time point remain elusive. Therefore, the K-means clustering algorithm was employed to analyze the DEGs expressed at each key differentiation stage (Fig. 5A): we identified that immature SN (at day 28) expressed several key transcription factors, including NKX2.2, NKX6.1, LMX1B, GATA2, INSM1 and ASCL1; in addition to the typical mature serotonergic markers, SLC18A2, MAOB and HTR2B, mature SN (at day 42) expressed high level of transcription factor FEV, indicating a potential role for FEV in facilitating and sustaining the maturation of human SN. GO enrichment analysis was performed for functional analysis of the marker genes shown in Fig. 5A. DEGs of cluster1 (day 0) exhibited significant enrichment in “ribosome biogenesis” pathway, indicating an increased demand for protein synthesis essential for cellular growth and differentiation (Fig. 5B). Compared to cluster2 (day 21), cluster3 (day 28) showed an augmented enrichment in the “pattern specification process” term. Moreover, this stage exhibited enrichment in terms such as neural precursor cell proliferation, cell fate commitment, regulation of synapse structure or activity, and regulation of neuron projection development (Fig. 5C, D). These findings suggested that the initiation of serotonergic fate specification might occur as early as day 21. Finally, the DEGs of cluster4 were highly enriched in “neuron/cell projection” and “synapse”-associated pathways (Fig. 5E), suggesting an active process of establishing synaptic connections and facilitating communication within the neighboring nervous system, which is a characteristic hallmark of neuronal maturation. To elucidate the crucial molecular events occurring during SN differentiation, we constructed interaction networks for DEGs enriched in the critical GO terms as presented in Fig. 5C–E, utilizing a protein–protein interactions (PPIs) degree analysis approach. As shown in Fig. 5F, the top 15 hub genes, ranked based on degree (the number of connections or a node possesses), were listed in the inner circle; and the genes of interests or those previously identified as critical for SN development were highlighted by yellow circle. This network emphasized the importance of transcription factors, including ASCL1, NKX2.2, LMX1B, GATA2 and INSM1, for the maturation of SN; furthermore, ZIC1, HOXA2 and MSX2 exhibited remarkable interactions with the SN-associated genes, indicating their pivotal roles in the differentiation process spanning from day21 to day28 (Fig. 5F, G). Then qPCR was performed to validate the expression levels of the hub genes and SN-associated genes: The qPCR data (Additional file 1: Figure S3) generally aligned with the RNA-seq data (Fig. 5A, F), although there were some slight discrepancies.

Fig. 5

Identification of the potential molecules responsible for serotonergic fate determination. A Heatmap of the DEGs exclusively expressed at each key differentiation stage. B–E Gene ontology (GO)-based functional analysis of the marker genes exclusively expressed by (B) cluster1, (C) cluster2, (D) cluster3 and (E) cluster4. F Protein-to-protein interaction (PPI) networks of the DEGs enriched in GO terms shown in B–E. G PPI networks involving the DEGs (day 21) having interactions with SN-associated genes expressed at day 28 of differentiation

Identification of the crucial genes and signaling pathways based on weighted gene co-expression network analysis (WGCNA)

Instead of a narrow focus on DEGs, WGCNA efficiently clusters the highly interconnected genes into diverse modules, enabling a comprehensive exploration of correlations between the modules and traits of interest. WGCNA was employed to construct the co-expression network among DEGs, adhering to the scale-free topology criteria: the soft-thresholding power β was set at 17 to reach the criteria (scale-independence > 0.85, average connectivity < 100) for the establishment of a scale-fee co-expression network (Additional file 1: Figure S4A, B). 20 modules were identified (Additional file 1: Figure S4B) and the heatmap plot of topological overlap matrix (TOM) of all genes was shown in Fig. 6A. Modules colored in red, purple, brown and pink showed the strongest correlations with the 4 key differentiation stages, respectively (Fig. 6B). To validate the accurate identification of modules, these modules were investigated again by calculating the mean absolute gene significance (GS) value for genes within each module. Notably, the GS-MM scatterplot for the brown and pink modules showed that the GS value was remarkably associated with the MM value (brown module: cor = 0.92, p < 1e−200; pink module: cor = 0.95, p < 1e−200) (Additional file 1: Figure S5), indicating the strong correlation of the two modules with the pivotal differentiation stages of SN. The gene set exclusively expressed on either day 28 or day 42 was intersected with the gene set of the brown or pink module respectively; then PPI analysis was conducted on the resulting two intersected gene sets. The hub genes of the two differentiation stages (day 28 and day 42) were identified to be associated with the cAMP and MAPK signaling pathways (Fig. 6C, D), underscoring the pivotal roles of these signaling pathways during the later stage of SN differentiation. To delineate the key molecules and signaling pathway networks involved in the differentiation of SN, an analysis was conducted using our previously published human SN single-cell RNA sequencing (scRNA) dataset [6] (GEO dataset: accession number GSE232698). Except for the well-known genes associated with serotonin synthesis and metabolism, including TPH2, FEV, MAOB, DDC, GCH1, QDPR and GCHFR, it was identified that 93% of mature human SN expressed GNG3 (Fig. 6E), which was also highly expressed in immature SN at day 28 of differentiation (Fig. 6C). Based on the above findings, the regulatory molecular network governing SN differentiation was exposed for the first time (Fig. 6F).

Fig. 6

Exploration of the key signaling pathways during SN differentiation. A Heatmap plot of topological overlap matrix (TOM) in the gene network. Rows and columns correspond to single genes, light color represents low overlap and progressively darker red color represents higher overlap. Blocks of darker colors along the diagonal are the modules. B The correlation of DEGs in modules across key stages of SN differentiation. C PPI analysis of the overlapping DEGs exclusively expressed on day 28 and in module brown. D PPI analysis of the overlapping DEGs exclusively expressed on day 42 and in module pink. E Expression patterns of the serotonergic-associated genes across individual serotonin neurons presented as UMAP plots. F Schematic of the signaling pathways and molecules that paly essential roles during SN differentiation