Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 May 3;22(5):684-697.e9.
doi: 10.1016/j.stem.2018.04.009.

Efficient Generation of CA3 Neurons From Human Pluripotent Stem Cells Enables Modeling of Hippocampal Connectivity In Vitro

Free PMC article

Efficient Generation of CA3 Neurons From Human Pluripotent Stem Cells Enables Modeling of Hippocampal Connectivity In Vitro

Anindita Sarkar et al. Cell Stem Cell. .
Free PMC article


Despite widespread interest in using human induced pluripotent stem cells (hiPSCs) in neurological disease modeling, a suitable model system to study human neuronal connectivity is lacking. Here, we report a comprehensive and efficient differentiation paradigm for hiPSCs that generate multiple CA3 pyramidal neuron subtypes as detected by single-cell RNA sequencing (RNA-seq). This differentiation paradigm exhibits characteristics of neuronal network maturation, and rabies virus tracing revealed synaptic connections between stem cell-derived dentate gyrus (DG) and CA3 neurons in vitro recapitulating the neuronal connectivity within the hippocampus. Because hippocampal dysfunction has been implicated in schizophrenia, we applied DG and CA3 differentiation paradigms to schizophrenia-patient-derived hiPSCs. We detected reduced activity in DG-CA3 co-culture and deficits in spontaneous and evoked activity in CA3 neurons from schizophrenia-patient-derived hiPSCs. Our approach offers critical insights into the network activity aspects of schizophrenia and may serve as a promising tool for modeling diseases with hippocampal vulnerability. VIDEO ABSTRACT.

Keywords: CA3; DG; disease-in-a-dish; hippocampus; neuronal diversity; pyramidal neurons; rabies tracing; schizophrenia; single cell sequencing; synaptic connectivity.


Figure 1.
Figure 1.. Generation of hCA3s from Human ESCs
(A) Hierarchical clustering based on differential transcriptomes between hpNPCs (hp) and panneuronal NPCs (pan). (B and C) Immunostaining (B) and quantitation (C) of OTX1/2 expression in hpNPCs and pan-NPCs. Arrowheads represent OTX+ NPCs. *p = 0.022. (D) Schematic of E12 telencephalon showing the cortical hem (red), DG (yellow), and CA primordia (green and blue). (E) Wnt signaling pathway is upregulated in hpNPCs (padj ≤ 1.87E-06) (F) Schematic showing hCA3 differentiation paradigm. (G) Arrowheads show VGLUT1+ (red) and neuronal marker NeuN+ (green) 4–5 WIV hCA3. (H) Arrowheads show GRIK4+ MAP2+ neurons in hCA3 (4 WIV). Results are presented as mean ± SEM. n = 3, *p < 0.05. Scale bar, 50 μm. See also Figure S1.
Figure 2.
Figure 2.. Comprehensive CA3 Patterning in hCA3 Protocol
(A)Heatmap showing log2 fold change in 23 CA3 genes hCA3 (3–6 WIV) and hpNPCs (padj < 0.05). (B) Heatmap showing log2 fold change in expression of markers for CA3 (Elavl2, Grik4, Neurod6, Scgn, and Dkk3), CA2 (Cacng5 and Pcp4), CA1 (Pou3f1), DG (Prox1, Calb1, C1ql2, and Dock10), and NPC (Wnt 8b and Foxg1) markers in hCA3 (6 WIV) and hpNPC (padj < 0.01). (C and D) Immunostaining (D) and quantitation (C) of ELAVL2 (*p = 0.02), SCGN (*p = 0.01), PRKCD (*p = 0.02), CALB1 (****p < 0.0001), and GABA expression in hCA3 and hDG (4–5 WIV). Arrowheads show positive cells. See also Figure S1.
Figure 3.
Figure 3.. Molecular Census of hCA3 Neuronal Population
(A–D) tSNE plot showing 380 single cells (A) K-means clustering revealed 3 optimal numbers of clusters within the population (green, blue, and red). (B) Average expression of neuronal genes (Map2, Nefl, Tuj1, Rbfox3, and Dcx) and Sox2. Red cells indicate expression; black indicates low or unidentifiable expression. (C) Average expression of 23 CA3 genes is enriched in the green cluster. (D) tSNE plot showing examples of DEGs (upregulatd: Lhx9, L1cam, Elavl2, Tspan7, Vglut2 (Slc17a6), Scn3b, Elavl3 and downregulated:Apoe) in the green cluster. (E) Heatmap showing the enrichment analyses based on GO and KEGG pathways. (F) Comparison of LHX9 expression (arrowheads) in hpNPC, hCA3 (4 WIV). Scale bar, 50 μm. See Figure S2.
Figure 4.
Figure 4.. hCA3 Constitutes a Heterogeneous Population
(A) Expression of CA1 (Wfs1 and Pou3f1) and CA2 genes (Pcp4 and Rgs14) (B and C) Heatmap showing expression of individual CA3 genes (B), as well as neuronal genes and housekeeping genes (C). (D) Elavl2 and Scgn (green) expression in green cluster cells) (E) Sparse SCGN expression in the hCA3s (4 WIV). Scale bar, 10 μm.) (F) SCGN is expressed (arrowheads) in human CA3, but not in mouse CA3. Scale bar, 50 μm.) (G) Number of ELAVL2+:number of SCGN+ ratio in human brain and hCA3s. ns, p > 0.05.) See also Figure S3.
Figure 5.
Figure 5.. hCA3s Are Functionally Active
(A) Electrophysiology of hCA3s. (a) hCA3 morphology at 4 WIV. Dendrites are color-coded from yellow, white, magenta, and green to pale yellow to show consecutive branch orders. Scale bar, 10 μm. (b) Whole-cell patch clamping of eGFP+hCA3 neurons (marked with lentiviral-Grik4-GFP or Elavl2-GFP reporter at 6 WIV). (c) An example of evoked AP (depolarizing step of 500 ms in current clamp). (d) Na+ and K+ current in response to a series of depolarizing steps in voltage clamp. (e and f) Examples of spontaneous AP in current clamp (e) and EPSCs recorded in voltage clamp (−70 mV) (f). See also Figures S4E–S4T. (B) Comparison of transplanted hCA3s with mouse CA3 neurons. 7 mouse CA3s and 4 eGFP+hCA3 neurons were recorded from mouse hippocampal slices at 3–3.5 MPT. (a and d) 40× images (top, differential interference contrast [DIC]; bottom, epifluorescence) of patched mCA3 neurons(a) and hCA3 (d) in mouse brain slices. Scale bars, 100 μm. (b, c, e, and f) APs were evoked by somatic current injection steps (b and e) and ramps (c and f) in examples of a mCA3 (b and c) neuron and hCA3 (e and f). (g–j) Voltage steps (g and i) and ramps (h and j) elicit sodium spikes at depolarized potentials in an example of a mCA3 neuron (g and h) and hCA3 (i and j). (k and l) Spontaneous excitatory postsynaptic currents (sEPSCs) recorded from examples of mCA3 (k) and hCA3 neurons (l). (m) Resting membrane potentials (mV) (p = 0.1760), (n) capacitances (pF) (p = 0.2196), (o) input resistance (p = 0.0596), (p) sEPSC frequency (p = 0.3236), and (q) amplitude (p = 0.7628) of mCA3 and hCA3 neurons. (r) Current-voltage (IV) curves (generated from current injection steps) of mCA3 and hCA3 (repeated-measures two-way ANOVA, F(1,7) = 4.442, p = 0.0731, post hoc test multiple comparisons, Sidak’s multiple comparison test, not significant at all points). (s–w) Rheobase or current injection needed to elicit first AP from ramp (pA) (p = 0.2494), AP threshold (mV) (p = 0.1109), latency to first AP (ms) (p = 0.2438), AP peak amplitude (mV) (p = 0.3266), or AP half-width (ms) (p = 0.5140) is not significantly different between mCA3 and hCA3 neurons. (x) Current-voltage (IV) curves (generated from voltagesteps) of mCA3 and hCA3 (repeated-measures two-way ANOVA, F(1,7) = 0.4102, p = 0.5423, post hoc test multiple comparisons, Sidak’s multiple comparison test, not significant at all points. In (m)–(q) and (s)–(w), a two-tailed t test was performed. See also Figures S4E–S4T. (C) Increase in spontaneous activity of hCA3 network with time (a–e). Number of spikes (****p < 0.0001) (a), active electrodes (****p < 0.0001) (b), bursts (****p < 0.0001) (c), network bursts (***p = 0.0002) (d), and synchrony index (***p = 0.0008) (e) over a 10-min MEA recording of hCA3s at 0–3 WIV (gray) versus 4–8 WIV (green). Number of recordings per group ≥ 7; number of neuronal cultures = 96. (D) Spike rate (Hz) of hDG and hCA3s over a 10-min MEA recording. n = 3 neuronal cultures, recorded ≥2 times per week for 0–8 WIV and 1 time in the ninth WIV (**p = 0.0071). (E) Schematic showing postsynaptic (green+red+) and presynaptic (red) neurons (left). hCA3 culture shows more connectivity (red+/red+green+) compared to hDGs and mouse irradiated fibroblasts in vitro (right). ***p = 0.0003 and ****p < 0.0001. Results are presented as mean ± SEM (recorded on 12-well MEA plate). See also Figure S3 and S4.
Figure 6.
Figure 6.. hCA3s Connect to hDGs
(A) Reconstruction of DG-CA3 neuronal connection (left) in microfluidic devices (right). (B) hCA3s and hDGs were transfected with LV-Grik4-tdTomato (shown in green) and LVProx1-GFP (shown in red), respectively. (C) Rabies tracing in microfluidic devices containing hCA3 (left) and hDG (right). Postsynaptic hCA3s (yellow arrows) are labeled with LV-Grik4-G-GFP (green; C, left) and rabies virus (green+red+; C, right); rabies virus traces the presynaptic cells (white arrows) in the DG compartment (red). (D) Schematic showing postsynaptic (green+red+), presynaptic (red) and CALB+ presynaptic (red+blue+) neurons. (E) hCA3s are marked CA3-specific LV-Grik4-GGFP (green, yellow arrow) in DG-CA3 co-culture in combination with rabies tracing (red). Immunostaining showing hCA3 (green+red+), presynaptic neurons (red), and CALB1+ (blue, white arrows) presynaptic neurons (red+blue+) as traced by rabies virus (red). Scale bar represents 50 μm in (B) and (C) and 10 μm in (E). See also Figure S5.
Figure 7.
Figure 7.. Reduced Spontaneous Activities in Hippocampal Neurons from a Cohort with SZ
(A) ELAVL2 (left, green, arrowheads), SCGN (right, green, arrowheads), and MAP2 (red) expression in hCA3s (4–5 WIV) from healthy and SZ patients. Scale bar, 10 μm. (B) Numbers of ELAVL2+MAP2+ neurons in healthy and SZ groups (n = 4 individuals per group; ns, p = 0.39). (C) Reduced average spike (**p = 0.008) and network bursts (*p = 0.046) over 10 min of recordings in SZ-hCA3-hDG co-culture at 6 WIV. Bursts and synchrony index were not different (ns, p > 0.05). n = 12 neuronal cultures per individual). (D) Reduced average number of spikes (*p = 0.04) and synchrony index (*p = 0.04) over a 10-min MEA recording of SZ-hCA3s (6 WIV). Number of bursts and network bursts were not different (ns, p > 0.05). n = 6 neuronal cultures per individual. (E) Example of evoked APs from healthy (top) and SZ-hCA3 (bottom). At least 7 neurons per individual and a total of 73 neurons, including 40 neurons from the healthy cohort and 33 neurons from the SZ cohort, were analyzed. Total number (*p = 0.032) and maximum number of evoked Aps (**p = 0.008) were significantly reduced in hCA3 SZ at 4–5 WIV. Input conductance (nS) was increased in SZ-hCA3 (**p = 0.004). Each dot shows an individual. Data are presented as mean ± SEM. n ≥ 3 iPSCs in healthy and SZ group. At least 3 (A, B, and E) or 6 (C and D) neuronal cultures per individual were used. ns, p > 0.05. Recordings were made on a 96-well MEA plate. See also Figures S6 and S7.

Comment in

Similar articles

See all similar articles

Cited by 22 articles

See all "Cited by" articles

Publication types

MeSH terms