Abstract
Regional specificity of stem cell-derived astrocytes is believed to be an important prerequisite for their applications in disease modelling and cell-based therapies. The regional identity of these astrocytes is often defined by the positional characteristics of their antecedent, stem cell-derived neural progenitors patterned to a fate of interest, with the assumption that the positional specification is to be preserved by the derived astrocytes. Using a human induced pluripotent stem cell line designed for tracing midbrain floor plate derivatives, here we show that lineage composition of the derived astrocytes is not a faithful recapitulation of the founder progenitor population, as demonstrated by the loss of floor plate differentiated progeny in the final astrocyte products. Using deep single cell RNA sequencing, we identified distinct transcriptomic signatures of midbrain floor plate-derived astrocytes. Our study highlights the need for rigorous characterisation of pluripotent stem cell-derived regional astrocytes and provides a valuable resource for assessing midbrain floor plate-derived human astrocytes.
eLife assessment
The manuscript by Li and coworkers analyzed astrocytic differentiation of midbrain floor plate-patterned neural cells originating from human iPS cells, with a LMX1A reporter. This valuable work identifies transcriptomic differences at the single-cell level, between astrocytes generated from LMX1A reporter positive or negative cells, as well as non-patterned astrocytes and neurons. The evidence is solid, but the paper can be strengthened by further analyses of the transcriptomic data, and astrocytic morphology; also, searching for some of the differentially expressed genes by immunohistochemistry in different regions of the mammalian brain, or in human specimens, would be very informative.
Significance of findings
valuable: Findings that have theoretical or practical implications for a subfield
- landmark
- fundamental
- important
- valuable
- useful
Strength of evidence
solid: Methods, data and analyses broadly support the claims with only minor weaknesses
- exceptional
- compelling
- convincing
- solid
- incomplete
- inadequate
During the peer-review process the editor and reviewers write an eLife assessment that summarises the significance of the findings reported in the article (on a scale ranging from landmark to useful) and the strength of the evidence (on a scale ranging from exceptional to inadequate). Learn more about eLife assessments
Introduction
Astrocytes are the most abundant cells in the brain. They play important roles in the central nervous system in supporting neuronal survival and synaptic activities, including the regulation of ionic homeostasis, providing energetic support, elimination of oxidative stress, and neurotransmitter removal and recycle (Verkhratsky and Nedergaard 2018). Abnormalities in astrocytes have been linked to various neurodegenerative and neurodevelopmental disorders, such as Parkinson’s disease, Alzheimer’s disease, Huntington’s disease, autism spectrum disorders and Alexander’s disease (Molofsky et al. 2012; Phatnani and Maniatis 2015; Booth et al. 2017). There is therefore a growing interest in using human pluripotent stem cell (PSC)-derived astrocytes for disease modelling in vitro (Chandrasekaran et al. 2016).
Contrary to the widely held belief that astrocytes in the brain are largely identical, recent studies revealed diversity in their transcriptomic profile, physiological properties and function (Oberheim et al. 2009; Schober et al. 2022). For example, single cell and spatial transcriptomic studies have identified several astrocyte subpopulations in the mouse cortex (Zhu et al. 2018; Batiuk et al. 2020; Bayraktar et al. 2020). In human, although astrocyte heterogeneity remains largely elusive, heterogeneity in radial glia across brain regions and within midbrain has been reported (La Manno et al. 2016; Li et al. 2023). Furthermore, different molecular and physiological features and distinct responses to stimuli were also observed in astrocytes from different mouse brain regions (Takata and Hirase 2008; Chai et al. 2017; Morel et al. 2017; Itoh et al. 2018; Kostuk et al. 2019; Makarava et al. 2019; Xin et al. 2019; Lozzi et al. 2020). Indeed, astrocyte heterogeneity has been suggested to underly the regional susceptibility to human diseases (Schober et al. 2022). Therefore, recapitulating astrocyte regional specificity in PSC-derived astrocytes is generally accepted as an important prerequisite.
Several reports have described the generation of regional astrocytes from human embryonic stem cells or induced pluripotent stem cells (iPSCs) which include forebrain (Krencik et al. 2011; Zhou et al. 2016; Tcw et al. 2017; Lin et al. 2018; Bradley et al. 2019; Hedegaard et al. 2020; Peteri et al. 2021), ventral midbrain (Booth et al. 2019; Barbuti et al. 2020; Crompton et al. 2021; de Rus Jacquet et al. 2021), hindbrain and spinal cord (Roybon et al. 2013; Serio et al. 2013; Holmqvist et al. 2015; Bradley et al. 2019; di Domenico et al. 2019; Yun et al. 2019). The regional identity of these astrocytes is typically evaluated at the stage of early neural progenitors generated via cell type- or region-directed neural patterning protocols, with the assumption that the positional characteristics will be faithfully preserved in the final astrocyte products. However, astrocyte production in vitro involves an extended period of astrocytic fate induction and progenitor expansion using FGF and EGF while substantial literature reported alterations in region-specific gene expression and/or neurogenic competence in expanded neural progenitors (Jain et al. 2003; Sun et al. 2008; Koch et al. 2009; Falk et al. 2012). Therefore, better characterisations of PSC-derived astrocytes and their lineage-specific features are needed to advance our knowledge about the molecular heterogeneity of human astrocytes.
Using a human iPSC line that allows the tracing of LMX1A expressing midbrain floor plate neural progenitors and their differentiated progeny (Cardo et al. 2023), we discovered an unexpected gradual depletion of LMX1A+ progenitor progeny during astrocyte induction from a bulk population of ventral midbrain patterned progenitors despite LMX1A+ progenitors being the predominant starting population. LMX1A+ progenitor derived astrocytes can however be generated if astrocytic induction is initiated from purified LMX1A+ progenitors, indicating that the positional constituents of the founding cell population may not be preserved faithfully in the derived astrocytes. Single cell RNA sequencing (scRNAseq) of astrocytes derived from both parental populations identified distinct transcriptomic signatures, providing a useful resource for assessment of human PSC-derived midbrain astrocytes.
Results
Depletion of LMX1A+ progenitors and/or derivatives in ventral midbrain patterned neural progenitor cultures during astrogenic induction
To investigate whether regionally patterned neural progenitors retain their lineage identity during astrogenic induction and glial progenitor expansion, we made use of the LMX1A-Cre/AAVS1-BFP iPSCs tracer line, which enables the tracking of LMX1A+ midbrain floor plate progenitors and their differentiated progeny(Cardo et al. 2023). We differentiated the LMX1A-Cre/AAVS1-BFP iPSCs towards the ventral midbrain fate following a modified protocol based on Jaeger et al. and Nolbrant et al.(Jaeger et al. 2011; Nolbrant et al. 2017) (Figure 1A). Immunocytochemistry of day (d)19 cultures confirmed a high proportion of cells expressing BFP (96.01±0.42%) and ventral midbrain progenitor markers LMX1A (92.94±0.91%), FOXA2 (94.76±0.57%) and OTX2 (97.82±0.28%; Figure 1B-D with the original images shown in Figure S1A). Most cells (91.26±1.64%) co-expressed LMX1A and FOXA2 (Figure 1B and 1D). At this stage, all BFP+ cells also stained positive to a pan neural progenitor marker NESTIN (Figure 1C). We detected a small proportion of cells expressing midbrain basal plate marker NKX6.1 (2.03±0.47%, Figure 1B-C) whose expression domain in the early developing ventral midbrain partially overlap with that of (Andersson et al. 2006)(Andersson et al. 2006). However, few PAX6+ cells were present (Figure 1C), which marks the forebrain and (Duan et al. 2013)(Duan et al. 2013). Immunocytochemical analysis of FACS-sorted BFP+ cells confirmed highly enriched expression of LMX1A (95.51±0.09%) and FOXA2 (95.35±0.14%), and co-expression of both markers (94.68±0.10%; Figure 1B and 1D). In contrast, only a small number of LMX1A+ cells (4.96±0.70%) were present in the sorted BFP- population (Figure 1B and1D). These findings provide further support that BFP expression faithfully recapitulate LMX1A expression and that LMX1A+ ventral midbrain progenitors represent the major cell population in d19(Cardo et al. 2023)(Cardo et al. 2023).
The d19 cells were then induced to undergo astrogenic switch in media containing FGF2 and EGF with the BFP content monitored by flow cytometry at each weekly passaging (Figure 1A, representative gating strategy is shown in Figure S1B-E). We found, unexpectedly, a dramatic decrease of BFP+ cell proportion from the starting point of 88.20±1.10% to only 27.59±8.28% at d43 and to nearly absent as the differentiation continued (Figure 1E). We did not observe evident cell death during culture and replating; thus, the absence of BFP could be either due to the silencing of BFP expression in the derivatives of LMX1A+ progenitors or the loss of these cells through growth competition. To address this question, we performed progenitor expansion and astrogenic induction under the same culture condition with purified d19 BFP+ progenitors isolated using fluorescence-activated cell sorting (FACS). Interestingly, the sorted BFP+ cells exhibited similar population growth rate to that of unsorted cultures, and the proportion of BFP+ cells remained at approximately 90% throughout the astrogenic induction and glial progenitor expansion period (Figure 1E). This observation demonstrates that BFP expression can be maintained in the derivatives of LMX1A+ midbrain floorplate progenitors, and that the loss of BFP+ cells in the unsorted culture is likely due to their growth disadvantage compared to the derivatives of LMX1A- progenitors.
Our finding is unexpected and demonstrates that the regional or lineage identity of PSC-derived astrocytic cells should not be assumed merely based on the dominant regional property of their cellular origin, given that no in vitro fate specification paradigm is 100% efficient.
Astrogenic switch occurred earlier in derivatives of LMX1A+ midbrain progenitors
Since sorted BFP+ (LMX1A+ or their derivatives) and unsorted astrocytic cultures differ distinctively in BFP expression soon after the initiation of astrogenic induction, for simplicity, these cultures are hereafter referred to as the BFP+ and BFP- cultures, respectively. Wondering whether the two cell populations behave differently in the process of astrogenic switch, we examined astrocytic marker expression in these cultures at d45 and d98 by immunocytochemistry. SOX9 and NFIA are transcription factors crucial for the initiation of astrogenesis and acquisition of astrogenic competence in the developing central nervous system (Stolt et al. 2003; Deneen et al. 2006), while CD44 identifies astrocyte-restricted precursors (Liu et al. 2004). We found that all these markers were more abundantly detected in the BFP+ cultures (NFIA: 65.89±2.81%; SOX9: 57.19±4.25%) than the BFP- cultures (NFIA: 4.26±1.28%; SOX9: 8.88±1.82%) at d45 (Two-way ANOVA with post-hoc Turkey test, NFIA: p=2.52×10-5, SOX9: p=9.21×10-6; Figure 2A-B with the original images shown in Figure S2). Although the number of NFIA+ and SOX9+ cells significantly increased in the BFP- cultures by d98 (NFIA: 44.07±4.56% on d98, p=3.58×10-4; SOX9: 44.28±2.84% on d98, p=1.21×10-4), the BFP+ cultures still contained more cells expressing NFIA (65.71±4.25%; p=2.06×10-2) and SOX9 (73.25±2.12%; p=5.51×10-4) than BFP-cultures (Figure 2A and 2C).
To investigate whether the temporal difference in astrocytic switch between the BFP+ and BFP- cultures affects maturation and functionality of the derived astrocytes, we initiated astrocyte terminal differentiation by exposing the BFP+ and BFP- astrocyte precursors to CNTF and BMP4 (Krencik et al. 2011; Bradley et al. 2019) from d87 (referred to as early astrocytes) and d136 (late astrocytes). Both BFP+ and BFP-cultures exhibited a similar expression profile of classic astrocyte markers, including AQP4, EAAT2, and S100B, but few GFAP+ cells at both time points (Figure 2B and 2D with the original images shown in Figure S3).
As a reference, we also generated neural progenitors without employing any patterning cues and induced astrogenic switch and astrocyte differentiation from these non-patterned neural progenitors (Figure S4A-B). We found that, while the astrocyte cultures derived from the non-patterned progenitors contained a similar proportion of cells expressing AQP4, EAAT2 and S100B compared to the BFP+ and BFP- astrocyte cultures, there are more GFAP+ cells in the non-patterned astrocyte preparations (17.89±5.4%, Figure S4A-B).
Functional astrocytes exhibit transient calcium (Ca2+) spikes upon chemical stimulation, such as ATP (Zhang et al. 2016). Using a FLIPR Ca2+ release assay, we observed a sharp increase in the intracellular Ca2+ concentration upon ATP administration in both the early and late BFP+ astrocyte populations (early BFP+: p=5.37×10-12; late BFP+: p<2.2×10-16; Figure 2E-F). ATP induced Ca2+ release is partially mediated by inositol trisphosphate. Indeed, addition of an inositol trisphosphate receptor antagonist 2-aminoethoxydiphenylborate (2-APB) reduced the amplitude (early BFP+: p=5.88×10-5; late BFP+: p=9.22×10-10; Figure 2F) and rise time (early BFP+: p=2.63×10-5; late BFP+: p=3.59×10-4; late BFP-: p=0.028; Figure 2G) of ATP-induced Ca2+ response in both the early and late BFP+ astrocytes. Interestingly, the early BFP+ astrocytes had significantly lower peak amplitude than that observed in late BFP+ astrocytes (p=6.92×10-9; Figure 2F), despite their similar level of astrocyte marker expression, suggesting a difference in maturity at the functional level. Early and late BFP- astrocytes exhibited a similar profile of time-dependent increase in the amplitude of ATP-induced Ca2+ response but did not reach statistical significance (Figure 2E-F). However, late BFP- astrocytes showed a significantly lower peak amplitude than late BFP+ astrocytes (p<2.2×10-16; Figure 2E-F).
Taken together, our data demonstrate that astrogenesis occurred earlier in BFP+ cultures than BFP- cells. This temporal difference is also reflected in functional maturity of the derived astrocytes, despite a similar expression profile of classic astroglial markers.
Single cell RNA sequencing confirms the authenticity of PSC-derived astrocytes
To further characterize the PSC-derived astrocytes, we performed full-length scRNAseq on early and late BFP+ and BFP- astrocytes using the iCELL8 platform and SMART-seq technology, with non-patterned astrocytes derived from the LMX1A-Cre/AAVS1-BFP tracer line as a control. A sample of iPSC-derived neurons was included to facilitate downstream cell type identification. We profiled FACS purified astrocytes expressing CD49f as well as unsorted cultures for comparison of all three astrocyte populations (Barbar et al. 2020); Figure S4C-H). After stringent filtering (Figure S5A-C, see Methods and Materials on filtering), we obtained a total of 17478 protein-coding genes in 1786 qualifying cells, with an average of 6326 protein-coding genes detected per cell.
Unsupervised Louvain clustering identified 12 cell clusters (Figure 3A). Cells clustered mainly based on sample type (astrocytes and neurons; Figure S6A) and the estimated cell cycle phase (Figure S6B), while sorted CD49f+ and unsorted astrocytes were clustered together (Figure S6A). TagBFP was detected at a higher level in BFP+ astrocyte samples than in BFP- and NP samples, while TagBFP expression was negligible in the neuronal samples derived from an iPSC line without BFP transgene (Figure S6C-D). Using a set of known astrocyte and neuronal signature genes (Figure 3B), we identified cells in clusters 0, 1, and 5-11 as astrocytes (Figure 3B), which were enriched in expression of SOX9, NFIA, NFIB, HES1, S100A13, S100A16, EGFR, CD44 and GJA1 (Figure 3B). These transcripts were also detected at high levels in clusters 2 and 4, which were mostly estimated to be in cell cycle phases G2, M and S (Figure S6B). In addition, clusters 2 and 4 showed high levels of proliferation-related transcripts, such as TOP2A, MKI67, CDK1 and AURKA (Figure 3B), and are thus defined as astrocyte precursors. In contrast, Cluster 3 contains mostly cells from the neuronal sample (Figure 3A and Figure S6A) and indeed expressed high levels of genes closely related to neuronal structure and function (such as STMN2, SYT1, DCX, MAPT, and SNAP25; Figure 3B). We did not detect transcripts indicative of endoderm (GATA4), mesoderm (TBXT and TBX6), and oligodendrocyte progenitors (SOX10 and PDGFRA) in any of these clusters (Figure S6C).
To determine the authenticity of these PSC-derived astrocytes, we mapped our data to published scRNAseq datasets obtained from five foetal, an adult human brain and a PSC-derived astrocyte study using Seurat integration (La Manno et al. 2016; Sloan et al. 2017; Zhong et al. 2018; Polioudakis et al. 2019; Agarwal et al. 2020; Fan et al. 2020; Bhaduri et al. 2021; Eze et al. 2021). We found that cells in clusters annotated as astrocytes (clusters 0, 1, and 5-11) were indeed predominantly mapped to the reference astrocyte or astrocyte precursor populations with high confidence (prediction score over 0.5; Figure 3C and Figure S7). In contrast, neuronal cluster 2 was mapped to neurons of the foetal reference datasets, while the astrocyte precursor clusters (2 and 4) were mapped to progenitor populations in the foetal reference datasets (Figure 3C and Figure S7). These findings demonstrate that our iPSC-derived astrocytes closely resemble those in the human brains.
Distinct transcriptome fingerprints of LMX1A+ midbrain floor plate-derived astrocytes
Significant advance has been made recently in understanding the molecular profiles of midbrain dopamine neurons. However, our knowledge about midbrain astrocytes in this regard remains limited and does not inform anatomic or lineage origin of the cells. In this regard, the BFP+ astrocytes provide a unique resource to determine the transcriptomic characteristics of human midbrain floor plate derived astrocytes. By performing pairwise differential gene expression (details described in Methods and Materials), we identified 1153 genes differentially expressed (DEGs; adjusted p values less than 0.05 and log2 fold change over 0.25) in BFP+ astrocytes when compared to either BFP- or non-patterned astrocyte populations (Supplementary Data 1). Of these, 159 were unique to BFP+ astrocytes (BFP+ enriched, Figure 4A), which include genes associated with midbrain dopamine neuron development such as SULF1, LMO3, NELL2, and RCAN2 (Figure 4B) (Strelau et al. 2000; La Manno et al. 2016; Bifsha et al. 2017; Ahmed et al. 2021). Interestingly, LMX1A and FOXA2, which was used to evaluate PSC-derived midbrain astrocytes in previous studies, were not detected in BFP+ astrocytes (Figure S6D). We also identified 530 DEGs enriched in BFP- astrocytes only (BFP- enriched, Figure 4A and Supplementary Data 1), which include those known to be expressed in the ventrolateral - dorsal domain of the midbrain and hindbrain, such as IRX3, IRX5, PAX3, and PAX7 (Figure 4B)(Houweling et al. 2001; Matsunaga et al. 2001). This transcription profile supports the notion that the BFP- astrocytes are descendants of the initial minor populations of lateral midbrain progenitors. Moreover, 72 DEGs were shared by BFP+ and BFP- astrocytes against the non-patterned astrocytes (midbrain enriched, Figure 4A and Supplementary Data 1). This set of genes includes NR2F1, NR2F2, ZEB2, KCNJ6 and SRPX (Figure 4B), which have been reported to be signatures of mouse midbrain astrocytes (Endo et al. 2022). Together, our findings provide a new entry to transcriptomic characteristics of midbrain astrocytes and specifically a gene expression map of midbrain floor plate derived human astrocyte lineage.
Gene ontology (GO) enrichment analysis was performed on the 1153 DEGs enriched in BFP+ astrocytes (Supplementary Data 2). The significantly enriched GO terms were mainly related to various aspects of metabolism, stress response, biosynthesis, lysosomal activity, and cellular respiration (Figure 4C and Supplementary Data 3). These biological processes have previously been shown to be disrupted by several mutations causing familial Parkinson’s disease (di Domenico et al. 2019; Barbuti et al. 2020; Sonninen et al. 2020). In contrast, the GO terms associated with 530 BFP- astrocyte enriched DEGs were mostly related to formation of extracellular matrix and tissue development (Figure 4D and Supplementary Data 2-3). The differential enrichment of GO terms implicates functional difference between the BFP+ and BFP- astrocytes and supports the need for generating regional specific astrocytes for disease modelling.
Discussion
Despite the general belief that recapitulating astrocyte lineage heterogeneity is necessary for stem cell-based disease modelling and cell transplantation, the extent of astrocyte heterogeneity in different brain regions and subregions remains largely elusive. By harnessing an LMX1A based lineage tracing human iPSC line and cutting edge scRNAseq technology, we show that astrocytes derived from the LMX1A+ midbrain floor plate progenitors, the same cells giving rise to midbrain dopaminergic neurons, possess distinct transcriptional landmarks from those derived from non-midbrain patterned neural progenitors as well as other midbrain patterned progenitors. Moreover, we discovered unexpected negative selection against derivatives of LMX1A+ progenitors during astrocyte induction and progenitor expansion. Our study highlights the need for careful characterisation of PSC-derived astrocytes and provides a transcriptomic fingerprint for midbrain floor plate derived astrocytes.
Using a popular astrocyte in vitro differentiation paradigm, we found that astrocytes descended from the LMX1A+ midbrain progenitors could only be obtained from purified progenitors. In contrast, astrocytes derived from bulk midbrain patterned progenitors exhibits transcriptomic profiles of the lateral-dorsal midbrain, despite LMX1A+ midbrain progenitors being the predominant starting population. Our findings demonstrate that the lineage composition of the parent progenitors may not be faithfully preserved during astrocyte induction and progenitor expansion. FGF is the most popular inductive molecule used for astrocyte differentiation from stem cells (Chandrasekaran et al. 2016). It is evident however that FGF expanded neural progenitors, originated either from the brain or neutralized PSCs, exhibit restricted regional competence and positional gene expression. For example, bulk expanded human ventral midbrain neural progenitors (Jain et al. 2003), and fetal forebrain or spinal cord derived neural stem (NS) cells only give rise to GABAergic neurons (Sun et al. 2008); likewise, lt-NES cells display anterior hindbrain-like positional profile (Falk et al. 2012), while their antecedents, PSC-derived neural rosettes and early passage derivatives, express anterior forebrain markers (Koch et al. 2009). It is not clear whether this is due to deregulation of the original patterning at the level of gene expression or the loss of associated cell population (Gabay et al. 2003). In this study, since BFP+ astrocytes can be generated under the same culture condition with purified LMX1A+ progenitors, we reasoned that the loss of their derivatives in unsorted cultures was possibly due to differential growth capacity.
Our study highlights the need for careful assessment of astrocyte positional identity. A common practice in this regard is to confirm the regional characteristics of the founder progenitors following fate directed neural induction, with the assumption that the dominant positional features will be maintained by the astrocyte progeny (Krencik et al. 2011). This strategy is at least partly dictated by our limited knowledge in gene expression signatures of regional- and/or lineage-specific astrocytes. Hence, an end point evaluation of PSC-derived astrocytes often relies on region-specific markers defined in the developmental brain during the neurogenic period. For example, LMX1A and FOXA2 expression were used as criteria for midbrain astrocytes in previous studies (Barbuti et al. 2020; Crompton et al. 2023). However, scRNAseq of human fetal ventral midbrain and adult substantia nigra revealed negligible expression of these transcripts in astrocytes (La Manno et al. 2016; Agarwal et al. 2020; Kamath et al. 2022). Consistent with these findings, we did not detect LMX1A or FOXA2 in neither the BFP+ nor BFP- astrocytes. Our analysis however identified new positive and negative markers that could be applied to confirm the regional identity of ventral midbrain astrocytes.
In addition to distinct transcriptomic profile, BFP+ and BFP- astrocytes may also be functionally different. Astrocytes generated from progenitors broadly patterned to the dorsal forebrain, ventral forebrain and spinal cord have been shown to exhibiting different GO enrichment profile as well as different physiological and functional properties (Bradley et al. 2019). In comparison to the BFP- and non-patterned astrocytes, the current study revealed that GO terms enriched in BFP+ astrocytes, which originated from the same progenitor giving rise to midbrain dopaminergic neurons, were closely related to various biological processes disrupted in astrocytes carrying familial Parkinson’s disease mutations (di Domenico et al. 2019; Barbuti et al. 2020; Sonninen et al. 2020). Such a distinct enrichment profile implicates BFP+ astrocytes being functionally adapted to supporting midbrain dopaminergic neurons compared to BFP- and non-patterned astrocytes.
In conclusion, this study lends further support on regional diversity of astrocytes and identified a set of midbrain enriched genes. Crucially, the transcriptomic fingerprint of human midbrain floor plate-derived astrocytes described here offers a much-needed resource for assessing the authenticity of stem cell derived astrocytes in studies associated with Parkinson’s disease.
Methods and Materials
Stem cell culture and astrocyte differentiation
KOLF2 human iPSCs were maintained in E8 flex media (ThermoFisher) and manually dissociated using Gentle Cell Dissociation Reagent (STEMCELL Technologies) as previously described (Cardo et al. 2023). Astrocytes were differentiated using a three-stage stepwise strategy consisting neural induction and regional patterning, astrogenic switch and progenitor expansion, and astrocyte terminal differentiation. Midbrain floor plate progenitors were generated as previously described (Cardo et al. 2023). At day 19, cells were replated as single cells onto poly-D-lysine-laminin-coated plates at 1×106 cells/cm2 for astrogenic switch and progenitor expansion in N2B27 media supplemented with 10 ng/mL FGF2 (Peprotech) and 10 ng/mL Human EGF (Peprotech) and replated every 6-8 days. For astrocyte terminal differentiation, expanded neural progenitors were re-plated at a density of 3×104 cells/cm2 in expansion media and 24 hours later switched to N2B27 supplemented with 10 ng/mL human recombinant CNTF (Peprotech) and 10 ng/mL human recombinant BMP4 (Peprotech) for 7 days followed by media containing CNTF alone for another 13 days. 10 µM Y-27632 was used for 24 hours before and after each replating. The protocol for generating non-patterned astrocytes was the same as for floor plate-derived astrocyte except the neural progenitors were derived with duo-SMAD inhibitors only without ventral patterning reagents.
Flow cytometry analysis and cell isolation
Cells were dissociated in Accutase as described above and washed twice with DPBS by centrifugation for 5 minutes at 200 rcf. For evaluating BFP expression, dissociated cells were resuspended in 0.5 mM EDTA in DPBS (Sigma-Aldrich) and analysed on a BD LSRFortessa cell analyser (BD Biosciences). For purifying BFP+ cells, dissociated cells were resuspended in the same cell culture media. Background autofluorescence was compensated for using KOLF2 parental cell line at a similar stage of differentiation to define BFP- gating. For purifying CD49f+ astrocytes, dissociated cells were stained with Alexa Fluor 647-conjugated rat anti-CD49f antibody (5% v/v in a 100 µL reaction; BD Biosciences) for 25 minutes at 37°C on an orbital shaker at 200 rcf and resuspended in DPBS containing 0.5% bovine serum albumin and 50 units/mL DNase I (Sigma Aldrich). Background autofluorescence was compensated for using KOLF2 parental cell line at a similar stage of differentiation to define BFP- gating and unstained astrocytes to define CD49f- gating. Cell sorting was performed on a BD FACSAria III (BD Biosciences) using an 80 µm nozzle. Sorted cells were collected in the same media as for resuspension. Flow cytometry data were analysed in FlowJo v10.8.1 (BD Biosciences) as shown in Figure S1B-E. Briefly, non-debris events were selected using the eclipse gates on dot graphs of SSC-A versus FSC-A. Singlet events were sequentially gated using polygonal gates on dot graphs of FSC-H versus FSC-A and SSC-H versus SSC-A by selecting the events in the diagonal region. The positive and negative gates in the fluorescence channel were set as bifurcate gates at a minimum of 99.9% percentile (usually at 99.99% percentile) on the histogram of the fluorescence intensity of the negative control sample of the same flow cytometry experiment and applied to all samples of the same flow cytometry experiment.
Immunocytochemistry
Cultures were fixed with 3.7% PFA for 15-20 min at 4 °C. For nuclear antigen detection, an additional fixation with methanol gradient was performed, which include 5 mins each in 33% and 66% methanol at room temperature followed by 100% methanol for 20 min at -20°C. Cultures were then returned to PBST via inverse gradient and were then permeabilized with three 10-minute washes in 0.3% Triton-X-100 in PBS (PBS-T) and then blocked in PBS-T containing 1% BSA and 3% donkey serum. Cells were incubated with primary antibodies in blocking solution overnight at 4°C. Following three PBS-T washes, Alexa-Fluor secondary antibodies (Thermo Fisher Scientific) were added at 1:1000 PBS-T for 1 hour at ambient temperature in the dark. Three PBS-T washes were then performed that included once with DAPI (Molecular Probes). Images were taken on a Leica DMI6000B inverted microscope. Quantification was carried out in Cell Profiler (Stirling et al. 2021) or manually using ImageJ (Schindelin et al. 2012) by examining at least four randomly selected fields from three independent experiments. The antibodies used are provided in the Supplementary Table 1. Representative images shown in main figures were cropped by randomly selecting the region of interest in the DAPI-stained channel only, with the original unedited images shown in Figure S1A, S2 and S3.
Single-cell RNA-sequencing
Cells were dissociated with Accutase with 10 units/mL of papain (Sigma-Aldrich) for 10 minutes at 37°C and resuspended in 0.5% bovine serum albumin (Sigma Aldrich) with 50 units/mL DNase I in DPBS without calcium or magnesium (Gibco) and passed through a cell strainer with 35 µm mesh. Cells were stained with 1 µM SYTO16 (Invitrogen) and 0.08% (v/v) propidium iodide (part of the Invitrogen ReadyProbes™ Cell Viability Imaging Kit, Blue/Red) for 20 minutes on ice and then dispensed into the nano-well plates of the ICELL8® cx Single-Cell System (Takara). Wells containing single viable cells were automatically selected using the ICELL8 cx CellSelect v2.5 Software (Takara) with the green and not red logic. Manual triage was performed to recover additional candidate wells that contain viable single cells. The library for sequencing was then prepared using the SMART-Seq ICELL8 application kit (Takara) following the manufacturer’s recommended protocol. Next-generation sequencing was performed using the NovaSeq6000 and the Xp Workflow on a S4 flow cell for 200 cycles of pair-end sequencing.
Single-cell RNA-sequencing analysis
Using the Cogent NGS Analysis Pipeline software v 1.5.1 (Takara), FASTQ files containing all indices for each chip were demultiplexed to FASTQ files containing one index per file. Adaptor sequences were removed using cutadapt 3.2 with the following settings: -m 15 --trim-n --max-n 0.7 -q 20. Trimmed FASTQ files were aligned to the Homo sapiens GRCh38.106 primary assembly with the BFP reporter gene attached to the end of the genome, using STAR 2.7.9a (Dobin et al. 2013) with the following settings: --outSAMtype BAM Unsorted --quantMode TranscriptomeSAM -- outReadsUnmapped Fastx --outSAMstrandField intronMotif --chimSegmentMin 12 -- chimJunctionOverhangMin 8 --chimOutJunctionFormat 1 --alignSJDBoverhangMin 10 --alignMatesGapMax 100000 --alignIntronMax 100000 --alignSJstitchMismatchNmax 5 -1 5 5 --chimMultimapScoreRange 3 --chimScoreJunctionNonGTAG -4 -- chimMultimapNmax 20 --chimNonchimScoreDropMin 10 --peOverlapNbasesMin 12 - -peOverlapMMp 0.1 –alignInsertionFlush Right -- alignSplicedMateMapLminOverLmate 0 --alignSplicedMateMapLmin 30. After alignment, gene-level quantification was performed using featureCounts from subread 2.0.0 (Liao et al. 2013) with the following settings: -t exon --primary -R CORE -F GTF -Q 0 -B -g gene_id. The count matrix of each index was combined in R 4.2.0 (Team 2023).
All downstream analysis was performed in R 4.3.0 using Seurat 4.3.0 (Stuart et al. 2019). Gene-level filtering was applied by including only protein-coding genes with at least five total counts across all cells and being expressed in at least 1% of all cells. Poor quality cells were then identified using the is.outlier function from the scater 1.28.0 (McCarthy et al. 2017). Poor quality cells were defined as having a high percentage of mitochondrial gene count, or high or low the total number of genes detected, or high total gene counts. The thresholds of each metrics for each sample were determined as twice the median absolute deviation of the sample. Raw gene counts were log normalised with a scale.factor setting of 1×105. Data from the two batches of experiments were integrated using the FindIntegrationAnchors and IntegrateData based on the common top 2000 highly variable genes and the first 30 dimensions of principal components. The percentage of mitochondrial gene count and total gene count were regressed out using the ScaleData function. Principal component analysis (PCA) was performed on the top 2000 high variable genes and the number of principal components used for uniform manifold approximation and projection (UMAP) was determined using the JackStraw method (Chung and Storey 2015). UMAP and unbiased Louvain clustering was performed on the first 33 principal components. Pairwise differential gene expression analysis was performed using the MAST method (Finak et al. 2015) with “Chip” as the latent variable. Gene ontology enrichment analysis was performed using enrichGO function in the clusterProfiler 4.10.0 package (Stirling et al. 2021) with all genes in the filtered dataset as the background. GO term database was downloaded using the org.Hs.eg.db 3.18.0 package (Carlson 2019). Revigo v1.8.1 was used to group the representative GO terms based on semantic similarity using a size setting of 0.5, Homo sapiens database, and SimRel method for semantic similarity (Schlicker et al. 2006; Supek et al. 2011).
Published datasets were downloaded from NCBI’s Gene Expression Omnibus (Clough and Barrett 2016) and processed in R 4.2.0. Gene level filtering was performed by retaining only protein-coding genes with more than five total counts across all cells. Gene counts were normalised using the NormalizeData function (scale.factor settings are listed in Supplementary Table 2). PCA was performed based on the top 2000 highly variable genes to obtain the first 50 PCs. Visual inspection of the elbow plot was used to determine the number of PCs for downstream analysis. Batch effect between subjects was evaluated on the two-dimensional PC2∼PC1 plot. Where inter-subject batch effect was observed, Harmony integration was performed based on the PCs selected in the previous step (Supplementary Table 2). UMAP was performed based on either the PCA or Harmony reduction (using the top 30 dimensions), and Louvain clustering was performed (settings shown in Supplementary Table 2). Cluster identities were verified against the reported annotation where possible. For datasets without detailed annotation published or astrocyte lineage reported (Supplementary Table 2), reannotation was performed based on the expression of known markers (Figure S7A-E). Reference mapping was performed using FindTransferAnchors and TransferData function in Seurat based on the first 30 dimensions of either the PCA or Harmony loadings of the reference dataset.
Statistical analyses
All data were collected from at least three independent experiments and presented as mean ± standard error of means unless otherwise specified. Data were tested for normality with the Shapiro-Wilk test and for equal variance with Levene test before performing statistical analyses by two-way ANOVA with post-hoc Tukey test for multiple comparisons where relevant. Kruskal-Wallis test with post-hoc Dunn’s test for pairwise comparison was used where parametric test was not suitable. Effect size was calculated as Cohen’s f for ANOVA or eta squared based on the H-statistic for Kruskal-Wallis test. All statistical tests were performed in R4.3.0.
Supplementary Information
Acknowledgements
We would like to thank Mark Bishop and Joanne Morgan for conducting FACS and next generation sequencing, respectively. We thank Kathryn Peall and Laura Abram for providing iPSC-derived neurons for scRNAseq. We also thank the support of the Supercomputing Wales project, which is part-funded by the European Regional Development Fund (ERDF) via the Welsh Government.
Funding
This work was supported by the UK Dementia Research Institute, jointly funded by the UK Medical Research Council, Alzheimer’s Society and Alzheimer’s Research UK, to C.W. (MC_PC_17112) and Z.L. (DRI-TRA2021-02), and a seed corn fund to Z.L. from the Neuroscience and Mental Health Innovation Institute, Cardiff University. Z.L. was funded by a UK Dementia Research Institute PhD studentship.
Data Availability
The sequencing data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (Clough and Barrett 2016) and are accessible through a GEO Series accession number which will be made available upon final publication.
Declaration of interests
The authors declare no competing interests.
References
- A single-cell atlas of the human substantia nigra reveals cell-specific pathways associated with neurological disordersNat Commun 11https://doi.org/10.1038/s41467-020-17876-0
- Combinatorial ECM Arrays Identify Cooperative Roles for Matricellular Proteins in Enhancing the Generation of TH+ Neurons From Human Pluripotent CellsFront Cell Dev Biol 9https://doi.org/10.3389/fcell.2021.755406
- Identification of intrinsic determinants of midbrain dopamine neuronsCell 124:393–405https://doi.org/10.1016/j.cell.2005.10.037
- CD49f Is a Novel Marker of Functional and Reactive Human iPSC-Derived AstrocytesNeuron 107:436–453https://doi.org/10.1016/j.neuron.2020.05.014
- IPSC-derived midbrain astrocytes from Parkinson’s disease patients carrying pathogenic SNCA mutations exhibit alpha-synuclein aggregation, mitochondrial fragmentation and excess calcium releasebioRxiv https://doi.org/10.1101/2020.04.27.053470
- Identification of region-specific astrocyte subtypes at single cell resolutionNature Communications 11https://doi.org/10.1038/s41467-019-14198-8
- Astrocyte layers in the mammalian cerebral cortex revealed by a single-cell in situ transcriptomic mapNature Neuroscience 23:500–509https://doi.org/10.1038/s41593-020-0602-1
- An atlas of cortical arealization identifies dynamic molecular signaturesNature 598:200–204https://doi.org/10.1038/s41586-021-03910-8
- Specificity of Pitx3-Dependent Gene Regulatory Networks in Subsets of Midbrain Dopamine NeuronsMolecular Neurobiology 54:4921–4935https://doi.org/10.1007/s12035-016-0040-y
- The Role of Astrocyte Dysfunction in Parkinson’s Disease PathogenesisTrends in neurosciences 40:358–370https://doi.org/10.1016/j.tins.2017.04.001
- RNA sequencing reveals MMP2 and TGFB1 downregulation in LRRK2 G2019S Parkinson’s iPSC-derived astrocytesNeurobiol Dis 129:56–66https://doi.org/10.1016/j.nbd.2019.05.006
- Regionally specified human pluripotent stem cell-derived astrocytes exhibit different molecular signatures and functional propertiesDevelopment 146https://doi.org/10.1242/dev.170910
- Single-Cell Transcriptomics and In Vitro Lineage Tracing Reveals Differential Susceptibility of Human iPSC-Derived Midbrain Dopaminergic Neurons in a Cellular Model of Parkinson’s DiseaseCells 12https://doi.org/10.3390/cells12242860
- org.Hs.eg.db: Genome wide annotation for HumanR package version
- Neural Circuit-Specialized Astrocytes: Transcriptomic, Proteomic, Morphological, and Functional EvidenceNeuron 95:531–549https://doi.org/10.1016/j.neuron.2017.06.029
- Astrocyte Differentiation of Human Pluripotent Stem Cells: New Tools for Neurological Disorder ResearchFrontiers in cellular neuroscience 10:215–215https://doi.org/10.3389/fncel.2016.00215
- Statistical significance of variables driving systematic variation in high-dimensional dataBioinformatics 31:545–554https://doi.org/10.1093/bioinformatics/btu674
- The Gene Expression Omnibus DatabaseMethods Mol Biol 1418:93–110https://doi.org/10.1007/978-1-4939-3578-9_5
- Efficient and Scalable Generation of Human Ventral Midbrain Astrocytes from Human-Induced Pluripotent Stem CellsJournal of Visualized Experiments 176https://doi.org/10.3791/62095
- Human stem cell-derived ventral midbrain astrocytes exhibit a region-specific secretory profileBrain Commun 5https://doi.org/10.1093/braincomms/fcad114
- The LRRK2 G2019S mutation alters astrocyte-to-neuron communication via extracellular vesicles and induces neuron atrophy in a human iPSC-derived model of Parkinson’s diseaseElife 10https://doi.org/10.7554/eLife.73062
- The Transcription Factor NFIA Controls the Onset of Gliogenesis in the Developing Spinal CordNeuron 52:953–968https://doi.org/10.1016/j.neuron.2006.11.019
- Patient-Specific iPSC-Derived Astrocytes Contribute to Non-Cell-Autonomous Neurodegeneration in Parkinson’s DiseaseStem cell reports 12:213–229https://doi.org/10.1016/j.stemcr.2018.12.011
- STAR: ultrafast universal RNA-seq alignerBioinformatics 29:15–21https://doi.org/10.1093/bioinformatics/bts635
- Spatiotemporal expression patterns of Pax6 in the brain of embryonic, newborn, and adult miceBrain Structure and Function 218:353–372https://doi.org/10.1007/s00429-012-0397-2
- Molecular basis of astrocyte diversity and morphology across the CNS in health and diseaseScience 378https://doi.org/10.1126/science.adc9020
- Single-cell atlas of early human brain development highlights heterogeneity of human neuroepithelial cells and early radial gliaNature Neuroscience 24:584–594https://doi.org/10.1038/s41593-020-00794-1
- Capture of neuroepithelial-like stem cells from pluripotent stem cells provides a versatile system for in vitro production of human neuronsPLOS ONE 7https://doi.org/10.1371/journal.pone.0029597
- Single-cell transcriptome analysis reveals cell lineage specification in temporal-spatial patterns in human cortical developmentScience Advances 6https://doi.org/10.1126/sciadv.aaz2978
- MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing dataGenome Biology 16https://doi.org/10.1186/s13059-015-0844-5
- Deregulation of dorsoventral patterning by FGF confers trilineage differentiation capacity on CNS stem cells in vitroNeuron 40:485–499https://doi.org/10.1016/s0896-6273(03)00637-8
- Pro-maturational Effects of Human iPSC-Derived Cortical Astrocytes upon iPSC-Derived Cortical NeuronsStem cell reports 15:38–51https://doi.org/10.1016/j.stemcr.2020.05.003
- Generation of human pluripotent stem cell reporter lines for the isolation of and reporting on astrocytes generated from ventral midbrain and ventral spinal cord neural progenitorsStem Cell Research 15:203–220https://doi.org/10.1016/j.scr.2015.05.014
- Gene and cluster-specific expression of the Iroquois family members during mouse developmentMechanisms of Development 107:169–174https://doi.org/10.1016/S0925-4773(01)00451-8
- Cell-specific and region-specific transcriptomics in the multiple sclerosis model: Focus on astrocytesProceedings of the National Academy of Sciences 115:E302–E309https://doi.org/10.1073/pnas.1716032115
- Temporally controlled modulation of FGF/ERK signaling directs midbrain dopaminergic neural progenitor fate in mouse and human pluripotent stem cellsDevelopment 138:4363–4374https://doi.org/10.1242/dev.066746
- GABAergic immunoreactivity is predominant in neurons derived from expanded human neural precursor cells in vitroExp Neurol 182:113–123https://doi.org/10.1016/s0014-4886(03)00055-4
- Single-cell genomic profiling of human dopamine neurons identifies a population that selectively degenerates in Parkinson’s diseaseNature Neuroscience 25:588–595https://doi.org/10.1038/s41593-022-01061-1
- A rosette-type, self-renewing human ES cell-derived neural stem cell with potential for in vitro instruction and synaptic integrationProc Natl Acad Sci U S A 106:3225–3230https://doi.org/10.1073/pnas.0808387106
- Subregional differences in astrocytes underlie selective neurodegeneration or protection in Parkinson’s disease models in cultureGlia 67:1542–1557https://doi.org/10.1002/glia.23627
- Specification of transplantable astroglial subtypes from human pluripotent stem cellsNature Biotechnology 29https://doi.org/10.1038/nbt.1877
- Molecular Diversity of Midbrain Development in Mouse, Human, and Stem CellsCell 167:566–580https://doi.org/10.1016/j.cell.2016.09.027
- Spatiotemporal transcriptome atlas reveals the regional specification of the developing human brainCell 186:5892–5909https://doi.org/10.1016/j.cell.2023.11.016
- The Subread aligner: fast, accurate and scalable read mapping by seed-and-voteNucleic Acids Res 41https://doi.org/10.1093/nar/gkt214
- APOE4 Causes Widespread Molecular and Cellular Alterations Associated with Alzheimer’s Disease Phenotypes in Human iPSC-Derived Brain Cell TypesNeuron 98:1141–1154https://doi.org/10.1016/j.neuron.2018.05.008
- CD44 expression identifies astrocyte-restricted precursor cellsDevelopmental Biology 276:31–46https://doi.org/10.1016/j.ydbio.2004.08.018
- Regionally Distinct Astrocytes Display Unique Transcription Factor Profiles in the Adult BrainFrontiers in Neuroscience 14https://doi.org/10.3389/fnins.2020.00061
- Region-Specific Response of Astrocytes to Prion InfectionFrontiers in Neuroscience 13https://doi.org/10.3389/fnins.2019.01048
- Role of Pax3/7 in the tectum regionalizationDevelopment 128:4069–4077https://doi.org/10.1242/dev.128.20.4069
- Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in RBioinformatics 33:1179–1186https://doi.org/10.1093/bioinformatics/btw777
- Astrocytes and disease: a neurodevelopmental perspectiveGenes & Development 26:891–907https://doi.org/10.1101/gad.188326.112
- Molecular and Functional Properties of Regional Astrocytes in the Adult BrainJ Neurosci 37:8706–8717https://doi.org/10.1523/JNEUROSCI.3956-16.2017
- Generation of high-purity human ventral midbrain dopaminergic progenitors for in vitro maturation and intracerebral transplantationNature Protocols 12:1962–1979https://doi.org/10.1038/nprot.2017.078
- Uniquely hominid features of adult human astrocytesJ Neurosci 29:3276–3287https://doi.org/10.1523/jneurosci.4707-08.2009
- Generation of the Human Pluripotent Stem-Cell-Derived Astrocyte Model with Forebrain IdentityBrain sciences 11https://doi.org/10.3390/brainsci11020209
- Astrocytes in neurodegenerative diseaseCold Spring Harbor perspectives in biology 7https://doi.org/10.1101/cshperspect.a020628
- A Single-Cell Transcriptomic Atlas of Human Neocortical Development during Mid-gestationNeuron 103:785–801https://doi.org/10.1016/j.neuron.2019.06.011
- Human stem cell-derived spinal cord astrocytes with defined mature or reactive phenotypesCell Rep 4:1035–1048https://doi.org/10.1016/j.celrep.2013.06.021
- Fiji: an open-source platform for biological-image analysisNature Methods 9:676–682https://doi.org/10.1038/nmeth.2019
- A new measure for functional similarity of gene products based on Gene OntologyBMC Bioinformatics 7https://doi.org/10.1186/1471-2105-7-302
- Foundations and implications of astrocyte heterogeneity during brain development and diseaseTrends in neurosciences 45:692–703https://doi.org/10.1016/j.tins.2022.06.009
- Astrocyte pathology and the absence of non-cell autonomy in an induced pluripotent stem cell model of TDP-43 proteinopathyProceedings of the National Academy of Sciences of the United States of America 110:4697–4702https://doi.org/10.1073/pnas.1300398110
- Human Astrocyte Maturation Captured in 3D Cerebral Cortical Spheroids Derived from Pluripotent Stem CellsNeuron 95:779–790https://doi.org/10.1016/j.neuron.2017.07.035
- Metabolic alterations in Parkinson’s disease astrocytesScientific Reports 10https://doi.org/10.1038/s41598-020-71329-8
- CellProfiler 4: improvements in speed, utility and usabilityBMC Bioinformatics 22https://doi.org/10.1186/s12859-021-04344-9
- The Sox9 transcription factor determines glial fate choice in the developing spinal cordGenes & Development 17:1677–1689https://doi.org/10.1101/gad.259003
- Growth/differentiation factor-15/macrophage inhibitory cytokine-1 is a novel trophic factor for midbrain dopaminergic neurons in vivoJ Neurosci 20:8597–8603https://doi.org/10.1523/jneurosci.20-23-08597.2000
- Comprehensive Integration of Single-Cell DataCell 177:1888–1902https://doi.org/10.1016/j.cell.2019.05.031
- Long-term tripotent differentiation capacity of human neural stem (NS) cells in adherent cultureMol Cell Neurosci 38:245–258https://doi.org/10.1016/j.mcn.2008.02.014
- REVIGO Summarizes and Visualizes Long Lists of Gene Ontology TermsPLOS ONE 6https://doi.org/10.1371/journal.pone.0021800
- Cortical Layer 1 and Layer 2/3 Astrocytes Exhibit Distinct Calcium Dynamics In VivoPLOS ONE 3https://doi.org/10.1371/journal.pone.0002525
- An Efficient Platform for Astrocyte Differentiation from Human Induced Pluripotent Stem CellsStem cell reports 9:600–614https://doi.org/10.1016/j.stemcr.2017.06.018
- A Language and Environment for Statistical Computing
- Physiology of astrogliaPhysiological Reviews 98:239–389https://doi.org/10.1152/physrev.00042.2016
- Ventral midbrain astrocytes display unique physiological features and sensitivity to dopamine D2 receptor signalingNeuropsychopharmacology 44:344–355https://doi.org/10.1038/s41386-018-0151-4
- Generation of Anterior Hindbrain-Specific, Glial-Restricted Progenitor-Like Cells from Human Pluripotent Stem CellsStem Cells and Development 28:633–648https://doi.org/10.1089/scd.2019.0033
- A single-cell RNA-seq survey of the developmental landscape of the human prefrontal cortexNature 555:524–528https://doi.org/10.1038/nature25980
- Neurosphere Based Differentiation of Human iPSC Improves Astrocyte DifferentiationStem Cells Int 2016https://doi.org/10.1155/2016/4937689
- Identification of spatially associated subpopulations by combining scRNAseq and sequential fluorescence in situ hybridization dataNat Biotechnol 36:1183–1190https://doi.org/10.1038/nbt.4260
Article and author information
Version history
- Preprint posted:
- Sent for peer review:
- Reviewed Preprint version 1:
Copyright
© 2024, Li et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
- views
- 102
- downloads
- 0
- citations
- 0
Views, downloads and citations are aggregated across all versions of this paper published by eLife.