Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 15 (3), 477-86, S1

Human Cerebral Cortex Development From Pluripotent Stem Cells to Functional Excitatory Synapses


Human Cerebral Cortex Development From Pluripotent Stem Cells to Functional Excitatory Synapses

Yichen Shi et al. Nat Neurosci.


Efforts to study the development and function of the human cerebral cortex in health and disease have been limited by the availability of model systems. Extrapolating from our understanding of rodent cortical development, we have developed a robust, multistep process for human cortical development from pluripotent stem cells: directed differentiation of human embryonic stem (ES) and induced pluripotent stem (iPS) cells to cortical stem and progenitor cells, followed by an extended period of cortical neurogenesis, neuronal terminal differentiation to acquire mature electrophysiological properties, and functional excitatory synaptic network formation. We found that induction of cortical neuroepithelial stem cells from human ES cells and human iPS cells was dependent on retinoid signaling. Furthermore, human ES cell and iPS cell differentiation to cerebral cortex recapitulated in vivo development to generate all classes of cortical projection neurons in a fixed temporal order. This system enables functional studies of human cerebral cortex development and the generation of individual-specific cortical networks ex vivo for disease modeling and therapeutic purposes.


Figure 1
Figure 1. Directed differentiation of human ES and iPS cells to cortical stem and progenitor cells
A. Over the 15 day neural induction period, Oct4-expressing hES cells differentiate at high efficiency to Pax6-expressing neural stem cells. Asterisks indicate the absence of detectable Pax6-expressing cells at day 0 and of Oct4-expressing cells at day 15. Error bar, s.e.m., n=3 samples for each marker at each timepoint. B. Quantitative RT-PCR for the cortical stem cell-expressed transcription factor Foxg1 demonstrates that the induction of cortical stem cells begins after 5 days and peaks after 20 days, whereas Tbr2-expressing intermediate progenitor cells and newly-born neurons begin to appear almost a week later. Error bars, s.e.m. Semi-quantitative RT-PCR for the cortical stem cell-expressed Emx1 and the ventral and/or caudally expressed transcription factors Dlx1, Nkx2.1, HoxB4 and Isl1 further shows that neural rosettes generated by this method do not express mRNAs for the ventrally and caudally expressed genes. This is in contrast with rosettes ventralised by treatment with the hedgehog agonist purmorphamine (hiPSCs + Pur.). C. Confirmation of the cortical identity of hESC-derived stem and progenitor cells (Ki67-positive) by expression of proteins characteristic of cortical stem cells: Pax6, the intermediate filament protein Vimentin and Otx1/2. Scale bars, 50 μm. D. Quantification of the efficiency of cortical induction, as assayed by the percentage of Pax6-expressing cells (percentage of nuclei, detected with DAPI), in the presence or absence of retinoids in two hESC and four hiPSC cell lines. Values are the average of three cultures for each cell line. Error bars, s.e.m. E-P Phase contrast and Pax6 immunofluorescence images of hESC and iPSC-derived cultures 12-14 days after the initiation of neural differentiation by dual SMAD inhibition in the absence or presence of exogenous retinoids. Neural rosettes (arrows) were infrequently observed when retinoids were not added to the cultures. Small clusters of Pax6-expressing cells were also observed in these cultures. Efficient induction of neuroepithelial rosettes was observed in all lines, with the majority of cells expressing Pax6. Scale bars, 50 μm.
Figure 2
Figure 2. PSC-derived cortical stem/progenitor cells form a polarised neuroepithelium in vitro analogous to the cortical ventricular zone
A-C. hES (A) and hiPSC (B,C)-derived cortical stem and progenitor cells form polarized neuroepithelial rosettes of proliferating cells (Ki67) in which many mitoses (phosphohistone H3) take place near a central lumen (white arrows) formed from the apical surfaces of the neuroepithelial cells (CD133/Prominin1, red in all panels). D-L. In addition to CD133/Prominin1, cortical rosettes localize aPKC (D-F), Transferrin receptor (TfR, G-I) and ASPM (J-L) to the apical, luminal pole of the neuroepithelial cells. Scale bars A-C, 100 μm; D, 25 μm. M-O. Centrosomes (detected by CEP135 immunostaining) are located apically in hESC (M) and hiPSC (N, 0) derived cortical rosettes, as they are in the neuroepithelium in vivo. Acetylated tubulin (white arrows) extends throughout the cortical stem/progenitor cells. Scale bars, 25 μm. P-R. Proteins localized apically at adherens junctions in cortical stem and progenitor cells in vivo, ZO-1 and N-cadherin , are found enriched at the luminal, apical surface of cortical stem/progenitor cells derived from human PSCs. Scale bar, 100 μm. S. Phase contrast still images, taken at 30-minute intervals of an apically dividing cortical stem/progenitor cell in a hESC-derived cortical rosette. Yellow arrow indicates the nucleus of a neuroepithelial cell (green dot) which translocates apically, dividing into two daughter cells, one of which (yellow) remains at the centre of the rosette, while the other (blue) migrates radially away to the periphery of the rosette. Scale bar, 100 μm. T. Phase contrast images of a basal (abventricular) mitosis within the rosette shown in (S). In this case, the cell indicated by the arrow/green dot undergoes M-phase at the periphery of the rosette. Scale bar, 100 μm.
Figure 3
Figure 3. Cortical rosettes differentiated from PSCs generate basal progenitor and outer radial glial cells
A-C. The majority of cells within hESC (A) and hiPSC (B, C)-derived cortical rosettes are Vimentin-expressing neuroepithelial cells with apical processes (arrows) oriented to the centre of each rosette (asterisks). Scale bar, 50 μm. D-F. A subpopulation of rosette cells towards the periphery of each rosette express the basal progenitor cell/subventricular zone cell transcription factor Tbr2. A subset of Tbr2+ cells are proliferating cells, as they co-express Ki67 (white arrowheads), whereas others do not (yellow arrows). Scale bar, 50 μm. G-I. Many of the Tbr2+ cells are newly-born neurons, as they also express Doublecortin (yellow arrows). Scale bar, 50 μm. J-L. Whereas the majority of mitoses (phospho-histone H3+ cells) are located apically, at the centre of each rosette (yellow arrows), mitoses are also found displaced from the centre or towards the periphery of rosettes (white arrows). These abventricular mitoses indicate the presence of a secondary, basal progenitor cell population within the rosettes. Scale bar, 50 μm. M, N. Quantification of the proportions of Tbr2+ cells found in cortical rosettes – between 15 and 20% of cells within rosettes derived from different hESC and hiPSC lines express Tbr2 (M). Of the Tbr2+ population, approximately 40% are Ki67+ cycling progenitor cells (N). Error bars, SD. O. Quantification of the relative proportion of abventricular, basal cortical stem/progenitor cell mitoses (phospho-histone H3+ cells) occurring in cortical rosettes generated from hESCs (H9) and two hiPSC lines. Phospho-histone H3+ mitotic cells were defined as abventricular if found more than 5 cell diameters from the luminal surface. Error bars, SD. P-R. Representative images of GFP-labelled individual radial glial (ventricular) cells (P) and oRG cells (Q, R). The radial glial cell (P) expresses Pax6 protein (blue nucleus) has an apical process (red arrowhead, P) that contains a centrosome at its apical extreme (yellow arrow) and a long basal process (white arrowhead, P). In contrast, oRG cells (Q, R), which also express Pax6 (blue nuclei), have a single basal process (white arrowheads) and their centrosomes are found in the cell body, near the nucleus (yellow arrows). Scale bars: P, 25 μm; Q, R, 10 μm. S-U. Two types of mitotic progenitor cells are found in abventricular rosette locations: Pax6-expressing cells with phospho-vimentin+ basal processes (white arrowheads, S, T) that do not express Tbr2 (white arrows, U), which correspond to oRG cells; and Pax6-negative cells lacking basal processes (yellow arrows, T) that express Tbr2 (yellow arrowhead, T) and correspond to basal progenitor cells. Scale bars: S, T, 20 μm; U, 25 μm.
Figure 4
Figure 4. PSC-derived cortical stem cells produce cortical glutamatergic projection neurons before astrocytes
A. Quantification of neuronal production from PSC-derived cortical stem/progenitor cells. Neurogenesis was stimulated by withdrawal of mitogens, following which increasing numbers of neurons were generated over several weeks. Values are percentage of cells (DAPI+ nuclei) that express the neuron-specific tubulin Tuj1 and are the average of three cultures at each timepoint. Error bars, s.e.m. B. Reelin mRNA is expressed by PSC-derived cortical neurons at day 50, as detected by RT-PCR. C. Glutamatergic neuron production from PSC-derived cortical stem/progenitor cells. The overwhelming majority of neurons contain vGlut1-positive punctae in their neurites. Scale bar, 100 μm. D. Tbr1-expressing, corticothalamic projection neurons differentiate relatively early from PSC-derived cortical stem/progenitor cells. Representative cultures from day 50 for hES and hiPS lines are shown (Tbr1, red; Tuj1, green; DAPI, blue). Scale bar, 50 μm. E. Astrocytes (S100, red) are generated at a relatively late stage in this system (day 70, as shown here), as also occurs during in vivo development. Scale bar, 50 μm. F1, F2. GFAP expression follows S100 expression in developing astrocytes: white arrows denote GFAP+/S100+ cells and yellow arrows denote GFAP−/S100+ cells at day 90. Scale bar, 50 μm. G. Numbers of S100+ glia increase from approximately 1% to 7% of cells in culture between days 60 and 100, consistently across five different hESC and hiPSC lines. H. Diagram of the mouse brain slice assay system used here. Human PSC-derived cortical neurons were plated as a single cell suspension on coronal slices of embryonic mouse brain and co-cultured for 14 days. I, J. Confocal images of human cortical neurons (detected by human NCAM expression; I, hESCs; J, iPSCs) in mouse cortical slices after 14 days. The majority of neurons are radially arranged (pial surface is up in each panel), with some neurons tangentially oriented within the marginal zone. Scale bar, 50 μm. K, L. Quantification of the orientation of human ESC and iPSC-derived cortical neurons in mouse cortical slice culture, excluding neurons in the marginal zone. Neurons were classified as radially oriented if their major neurite was directed towards the pial surface within a 90° segment centered on the vertical (±45° of the perpendicular between the pial and ventricular surface, K). Almost 90% of the neurons from each line were radially oriented (L). n=3 regions counted, total of 500 cells for each cell line; error bars, SD. M, N. Expression of the layer 6 transcription factor, Tbr1, in human cortical neurons found in deep cortical layers of the mouse cortex (M) and also in upper cortical layers (N). Scale bar, 25 μm.
Figure 5
Figure 5. Production of human cortical excitatory neurons from PSCs in vitro recapitulates in vivo development
A. Diagram of classes of cortical projection neurons in the layers of the adult cortex, based on mouse data, with transcription factors expressed in each class of neuron as indicated. CPNs, callosal projection neurons. B. The order of cortical neurogenesis from human ES cells recapitulates normal development. Deep and upper layer neurons are generated in a temporal order from hESCs, with layer 5 and 6 neurons (Tbr1 and CTIP2) generated before layer 2/3 neurons (Brn2). N=3 experiments scored for each marker. Error bars, s.e.m. C-E. Differentiation of early born cortical neurons from hESCs. Corticothalamic projection neurons of layer 6 (Tbr1/CTIP2-expressing neurons, A-C) and corticospinal motor neurons of layer 5 (CTIP2-positive, Tbr1-negative neurons, white arrows in C) are both present. Scale bars, 50 μm. F-H. Upper layer, later born cortical neurons differentiate in this system several weeks after the early born, deep layer neurons. These neurons express Cux1, Satb2 and Brn2. Scale bars, 50 μm. I-K. Reproducible generation of deep and upper layer projection neurons by four different iPSC lines. L, M. Birthdating by BrdU administration demonstrates that upper layer, Satb2-expressing neurons are generated from cortical stem/progenitor cells at late stages (as late as day 90) in this system. BrdU addition for 48 hours at day 50, followed by harvesting 7 days later (L), found that the majority of Tbr1-expressing, deep layer neurons have already been generated at this stage, with only a small number of Tbr1+/BrdU+ neurons found (white arrows, L), whereas no Satb2-expressing neurons have differentiated. In contrast, BrdU administration at day 90, followed by cell analysis at day 100, found that large numbers of Satb2-expressing neurons are generated from mitotic stem/progenitor cells at this stage (yellow arrows, N). Scale bar, 50 μm. N. Relative proportions of different classes of cortical projection neurons generated from humans ES and iPS lines. Approximately equal proportions of deep and upper layer neurons are generated from all lines. See methods for technical details of cell counting.
Figure 6
Figure 6. PSC-derived cortical neurons differentiate to acquire mature electrophysiological properties
A, B. Voltage-gated sodium and potassium channels in PSC-derived cortical neurons. Current responses to families of step depolarizations from a holding potential of −80 mV to +40 are superimposed. In A, fast-activating and inactivating inward sodium currents are completely blocked by applying TTX. In B, 4-AP blocks a fast-activating transient fraction of outward K current. C. The electrophysiological properties of PSC-derived cortical neurons mature over time, as exemplified by the change in action potential firing in response to step current injections. Day 28, n=3; day 35, n=4; day 42 n=4; day 49, n=3. D, E. hESC and hiPSC-derived cortical neurons develop robust regular-spiking behaviour in response to step current injection.
Figure 7
Figure 7. Formation of functional excitatory synapses among PSC-derived cortical projection neurons
A-F. Super-resolution microscope images of dendrites (MAP2, green, A, B) showing localization of foci of the excitatory synapse-specific PSD95 protein (A-D). Physical synapses (arrows in all images) were identified by juxtaposition of pre- and post-synaptic protein complexes, either synaptophysin and PSD95 (C, D) or Munc13 and Homer (E, F). Scale bars as shown. G. Quantification of the proportion of PSD-95 foci that were found juxtaposed to synaptophysin foci in cortical neuronal cultures derived from four hiPSC and one hESC line. In all cases approximately 90% of PSD-95 foci were found colocalised with synaptophysin. Error bars, SD. See methods for technical details. H, I. Detection of mEPSCs in whole cell recordings of hESC (H) or hiPSC (I)-derived cortical neurons. The AMPA receptor antagonist CNQX blocked the appearance of mEPSCs. Also shown are average mEPSC from hESC (n=20 events; H) and hiPSC (n=20 events; I) derived cortical neurons. J. Evoked post-synaptic potentials (PSPs) in hESC and hiPSC-derived cortical neurons are excitatory (n=9 neurons) and blocked by the AMPA receptor antagonist CNQX (50 μM; n=4 neurons). Addition of the NMDA receptor antagonist APV (50 μM) had no detectable effect on evoked PSPs (n=2 neurons). Averaged recordings (Control, 7 trials; APV 13 trials; CNQX, 10 trials; Washout, 10 trials) from a representative neuron are shown.

Similar articles

See all similar articles

Cited by 273 PubMed Central articles

See all "Cited by" articles

Publication types

MeSH terms