SOX9-induced Generation of Functional Astrocytes Supporting Neuronal Maturation in an All-human System

Abstract

Astrocytes, the main supportive cell type of the brain, show functional impairments upon ageing and in a broad spectrum of neurological disorders. Limited access to human astroglia for pre-clinical studies has been a major bottleneck delaying our understanding of their role in brain health and disease. We demonstrate here that functionally mature human astrocytes can be generated by SOX9 overexpression for 6 days in pluripotent stem cell (PSC)-derived neural progenitor cells. Inducible (i)SOX9-astrocytes display functional properties comparable to primary human astrocytes comprising glutamate uptake, induced calcium responses and cytokine/growth factor secretion. Importantly, electrophysiological properties of iNGN2-neurons co-cultured with iSOX9-astrocytes are indistinguishable from gold-standard murine primary cultures. The high yield, fast timing and the possibility to cryopreserve iSOX9-astrocytes without losing functional properties makes them suitable for scaled-up production for high-throughput analyses. Our findings represent a step forward to an all-human iPSC-derived neural model for drug development in neuroscience and towards the reduction of animal use in biomedical research.

Keywords: All-human co-culture system; Astrocytes; Differentiation protocol; Genome engineering; Pluripotent stem cells.

Fig. 1

Six days of SOX9 overexpression in iSOX9-derived NPCs upregulates the expression of astrocyte-specific proteins. a Overview optimization of iSOX9-astrocyte differentiation. First, the number of days to overexpress SOX9 were tested by comparing 6 and 10 days of doxycycline exposure. Next, the effect of removing doxycycline was assessed. b Flow cytometry to assess the percentage of S100B and EAAT1 expressing cells after exposing iSOX9-derived NPCs for 6 and 10 days to doxycycline (N = 4–5 independent differentiations; *p < 0.05). c RT-qPCR of SOX9, S100B, GFAP, ALDH1L1 and OCT4 transcripts after exposing iSOX9-derived NPCs for 6 and 10 days to doxycycline (N = 3–4 independent differentiations; *p < 0.05).  d RT-qPCR of SOX9, S100B, GFAP, ALDH1L1 and OCT4 transcripts at different time points throughout the optimized differentiation protocol compared to fHA. (N = 3 independent differentiations; ^Δp < 0.05 versus fHA, ^####p < 0.0001 versus 6D doxy, °p < 0.05 versus 5D no doxy, ⁺p < 0.05 versus 30D no doxy). e Flow cytometry to assess the percentage of S100B and EAAT1 expressing cells at different time points throughout the optimized differentiation protocol compared with fHA (N = 3–5 independent differentiations;***p < 0.001 versus 6D doxy). f Representative immunofluorescence images of fHA and iSOX9-astrocytes at 5 and 30 days after stopping doxycycline treatment for S100B (green), GFAP (red), EAAT1 (green) and ALDH1L1 (red) (scale bar: 50 μm). Quantification of the percentage of positive cells and integrated density was performed using Columbus Image analysis software (PerkinElmer) (N = 3 independent differentiations; *p < 0.05, ***p < 0.001). g Cell roundness quantification of fHA and iSOX9-astrocytes after stopping doxycycline treatment for 5 and 30 days (N = 3 independent differentiations; ***p < 0.001). h Quantification of cell numbers throughout iSOX9-astrocyte differentiation (N = 3 independent differentiations). All data represented as mean ± SEM

Fig. 2

Functional characterization of iSOX9-astrocytes based on glutamate transport, patch clamp measurements and cytokine secretion. a Glutamate concentration was measured via mass spectrometry from the medium conditioned for 24, 48 and 72 h by iSOX9-astrocytes (N = 8 independent differentiations with two technical replicates for each differentiation) or fHA (N = 4). In addition, medium with no cells was taken into account as negative control (N = 4) (***p < 0.001 versus fHA and iSOX9). b Resting membrane potential and cell capacitance (reflecting cell size) measurements of fHA (N = 15 cells of 3 independent differentiations) and iSOX9-astrocytes (N = 15 cells of 3 independent differentiations) (**p < 0.01). c Na⁺ and K⁺ current measurements via patch clamp at voltages between − 120 mV and + 160 mV of fHA (N = 14) and iSOX9-astrocytes (N = 17 cells of 3 independent differentiations). d After exposing iSOX9-astrocytes (N = 3 independent differentiations with two technical replicates for each differentiation) and fHA (N = 3) for 5 days to IL-6, TNF-α, IL-1β or the combination of TNF-α and IL-1β, the concentration of IL-1α, IFN-α, IFN-β, HGF, IL-12, IL-10, LIF, GM-CSF, IL-6, TNF-α, IL-1β, PDGFaa and IL-23 was measured in the medium using a bead-based immunoassay. Each time, the measured concentration was divided by the concentration of untreated cells (*p < 0.05;**p < 0.01;****p < 0.0001). All data represented as mean ± SEM

Fig. 3

iSOX9-astrocytes show comparable calcium responses as fHA. Calcium responses (plotted as F340/F380 ratio) measured in Fura-2-loaded iSOX9-astrocytes and fHA in the presence of extracellular Ca²⁺-chelating agent BAPTA (3 mM), ensuring only Ca²⁺ release from internal Ca²⁺ stores is measured. a After stimulating fHA and iSOX9-astrocytes with ATP, they show a similar calcium response measured by Fura2 fluorescence emission (N = 6 independent differentiations). b After stimulating fHA and iSOX9-astrocytes with Ach, they show a similar calcium response measured by Fura2 fluorescence emission (N = 6 independent differentiations). c iSOX9-astrocytes and fHA show a similar calcium response after stimulation with GPN, FCCP and ionomycin measured by Fura2 fluorescence emission. A significantly higher calcium response is observed in fHA compared to iSOX9-astrocytes after thapsigargin stimulation (**p < 0.01) (N = 6 independent differentiations). All data represented as mean ± SEM

Fig. 4

RNASeq analysis of iSOX9-astrocytes reveals maturation after stopping doxycycline treatment and resemblance to isolated postnatal astrocytes. a Clustering of the iSOX9-astrocytes 5 (early) and 30 days (late) after stopping doxycycline treatment using a healthy donor-derived iPSC line and iPSCs derived from an ALS patient (each time N = 3 independent differentiations). Commercially available primary human fetal astrocytes (fHA) (N = 3) and iPSC-derived astrocytes (“iCell”) (N = 3) were also included. b Ranked GSEA of the differentially expressed genes between early and late iSOX9-astrocytes. Biological processes with a positive normalized enrichment score (NES) (blue bars) are enriched in the early iSOX9-astrocytes, while the processes with a negative NES (orange bar) are enriched in the late iSOX9-astrocytes. This analysis was performed for both the healthy donor and patient-derived iPSC-derived cells. c PCA plot using only the genes that showed at least a 16-fold increased or decreased expression between the postnatal and fetal samples of Zhang et al., of the previous samples mentioned in panel (a) combined with RNASeq data from isolated postnatal and fetal human astrocytes from Zhang et al. and the iPSC-derived astrocytes from Li et al. and Tchieu et al. d Expression values of the genes that showed at least a 16-fold increased or decreased expression between the postnatal and fetal samples of Zhang et al. Those genes can be divided into 11 clusters and are considered important for development/age

Fig. 5

iSOX9-astrocytes support iPSC-derived neuron maturation. DIV4 iNGN2-PSC-derived neurons were co-cultured 1:1 with DIV40 iSOX9-astrocytes for 37 days and electrical activity was measured using MEA plates (16 electrodes per well of a 48-well plate). As control, iSOX9-astrocytes alone and rat primary cortical neurons were also evaluated. Depicted are the % of active electrodes (a), number of bursts (b) and synchronicity index (c). Data is represented as mean ± SEM of N = 3–4 independent experiments