HippoCA3mpal Stem Cell Models Expose Dysfunctional Circuits in Schizophrenia

Abstract

In this issue of Cell Stem Cell, Sarkar et al. (2018) describe an efficient method for the generation of human hippocampal pyramidal neurons from stem cells. They developed a compelling in vitro model that recapitulates synaptic connectivity within the hippocampus and showed that cells derived from patients with schizophrenia exhibit abnormal electrical activity.

Schizophrenia is a common, polygenic, multifactorial psychiatric disorder for which partially effective treatments are available but none are disease modifying. In order to develop more effective therapies, new model systems have to be developed to allow interrogation of the disorder at a molecular and cellular level. A properly functioning hippocampus is essential for ‘‘normal’’ learning and memory, and smaller hippocampal volumes have previously been reported in patients diagnosed with schizophrenia (Nelson et al., 1998). Focusing on the hippocampus to model neuronal networks in schizophrenia confers certain advantages, as it is a particularly well characterized area at a cellular, molecular, and circuit level. Hippocampal plasticity and its’ vulnerability to environmental cues means that it is ideally suited to aid in the study of neurodevelopmental disorders such as schizophrenia where an interplay of genes and environment is believed to be critical in the underlying pathophysiology of the disorder, ultimately resulting in dysfunctional ‘‘wiring’’ within the hippocampus and altered functional connectivity to other brain regions such as the prefrontal cortex.

The hippocampal formation consists of three major subfields: cornu ammonis 1–3 (CA1–3), as well as the dentate gyrus (DG), the subiculum, and the entorhinal cortex (EC) (Figure 1). Several pieces of evidence support hippocampal vulnerability in schizophrenia: verbal declarative memory encoding (a hippocampal function) is predominantly affected in the disorder, and post-mortem (Kolomeets et al., 2007) and imaging studies have shown that CA3 synapse density and DG glutamate transmission is reduced, suggesting that mossy fiber (MF) pathways, connecting presynaptic DG axons and postsynaptic CA3 neurons, are particularly susceptible. However, a more recent study of post-mortem brains from schizophrenia patients has shown increased expression of GluN2B-containing NMDA receptors in CA3, suggesting either an increase in excitatory signaling or an elevation in ‘‘silent’’ synapses in CA3 (Li et al., 2015). Until now, an in vitro human cellular model that permits the study of neuronal network properties between synaptic pairs in the hippocampus has been lacking due to the relative inaccessibility of the human brain.

Figure 1. Schematic Overview of the Experimental Design Adopted in the Sarkar et al. Study.

The 2D cyto-architecture of the hippocampus was recapitulated in vitro by the generation and characterization of a panel of human iPSC and hESC-derived hippocampal cells (CA3s and DGs). Labeling the CA3 neurons (using lentiviral GRIK4-eGFP or Elval2-reporter constructs), which permitted immunocytochemistry of the DG-CA3 co-cultures as well as Rabies virus trans-neuronal tracing, demonstrated connectivity. Collectively, the studies demonstrated important structural and functional differences in synaptic connectivity between schizophrenia patient-derived cells and unaffected controls.

Over the past several years, there has been a plethora of papers using human stem cells to model neuropsychiatric disorders in a dish (Brennand et al., 2011, 2015; Wen et al., 2014; Yoon et al., 2014; Yu et al., 2014). These studies have investigated cell migration, polarity, and synaptic maturation, yet none have specifically facilitated the study of mature neuronal network properties. The Gage group has blazed a trail in this regard from the earliest seminal paper by Brennand et al. (2011) that was the first study to describe the use of patient-derived stem cells to model schizophrenia. The latest paper by Sarkar et al. (2018) (in this issue of Cell Stem Cell) is yet another landmark study from this group. It is unique in that it describes the production of multiple subtypes of functionally active human CA3 pyramidal neurons from the directed differentiation of patterned hippocampal progenitor cells (hpNPCs) derived from human embryonic stem cells (ESCs) or induced pluripotent stem cells (iPSCs) and uses these to highlight key differences between case and control-derived cell lines from patients with schizophrenia.

To generate hpNPCs, embryoid bodies were treated with DKK, SB431542, cyclopamine, and N2/B27 supplements for 20 days, which results in the combined inhibition of the Wnt, TGFb, Shh, and BMP pathways. Sarkar et al. (2018) showed that these hpNPCs are enriched with medial pallium markers that give rise to the hippocampus. Sarkar and colleagues obtained mature hCA3 neurons by plating these hpNPCs in the presence of ascorbic acid (AA), cAMP, BDNF, laminin, and Wnt3a, which was then removed after 3 weeks growth to allow differentiation to proceed (Yu et al., 2014). Morphogens, such as Wnts, have been shown to play an important role in orchestrating cell fate decisions in vitro in a concentration-dependent fashion (Liu and Zhang, 2011). In a similar fashion, hDG neurons were prepared from progenitors. The differentiated CA3 neurons are capable of forming synaptic connections in vitro with DG neurons, which recapitulates the development of normal hippocampal neuronal connectivity. Having generated this platform of pyramidal neurons, Sarkar et al. (2018) went on to characterize these at the molecular level using single-cell transcriptomic techniques. They showed enrichment of more than 20 CA3 genes in the hCA3 cells, reflecting heterogeneity of these cultures in terms of the presence of subpopulations of cells. On this journey, Sarkar et al. (2018) discovered a novel hCA3 subtype that is not found in mice, highlighting again the need for human-derived cell lines to model human disorders and warning of potential limitations in studying brain disorders that target the hippocampus using mouse models.

This patient-iPSC-derived DG-CA3 neuronal co-culture system was then utilized to specifically address connectivity in schizophrenia by means of both multielectrode array (MEA) recordings and whole-cell patch-clamp electrophysiology, both of which showed a reduction in spontaneous and evoked neuronal activity in CA3 neurons from schizophrenia cases, indicating a patient-specific decrease in intrinsic neuronal activity. The beauty here is that the lineage-specific hippocampal circuits that Sarkar et al. (2018) have modeled can potentially be scaled up for the study of a larger number of patient lines in the future and also for drug development and screening. Furthermore, in vitro experiments were cleverly combined with transplantation studies in mice to show that the transplanted human hCA3s survive, differentiate, mature, and integrate into functional hippocampal circuits in vivo after receiving functional excitatory inputs.