Schizophrenia Research Forum - A Catalyst for Creative Thinking

SfN 2011—Reprogramming Human Cells to Model Brain Diseases

Elise Malavasi, a graduate student at the University of Edinburgh, was kind enough to volunteer as a meeting correspondent for this year's Society for Neuroscience meeting, 12-16 November 2011 in Washington, DC.

7 December 2011. A symposium on Monday morning, 14 November 2011, chaired by Guo-li Ming from Johns Hopkins University, Baltimore, Maryland, and Marius Wernig from Stanford University, California, provided an exciting overview of how somatic cell reprogramming is being successfully applied to the study of brain disorders ranging from Alzheimer’s to schizophrenia.

Guo-li Ming introduced the session with a historic perspective on the seminal work of Yamanaka and colleagues, who first reported the groundbreaking discovery that both mouse and human fibroblasts can be reverted to pluripotent stem cells by exposure to a cocktail of four selected factors (Takahashi and Yamanaka, 2006; Takahashi et al., 2007). Once differentiated into neurons or glial cells, these reprogrammed stem cells—originally designated iPSCs (induced pluripotent stem cells)—provide a new model system to identify cellular phenotypes, connectivity, and circuitry defects associated with brain diseases. Ming emphasized how the identification of disease-specific phenotypes in iPSC-derived neurons and glia will allow the development of drug screening platforms for the search for compounds that can improve such phenotypes.

Oliver Brüstle from the University of Bonn, Germany, opened the symposium with a remark on how the availability of standardized procedures to produce neurons from human stem cells is a fundamental prerequisite for comparability among studies. Brüstle suggested that this could be achieved through the employment of homogeneous and stable self-renewing intermediates such as the long-term human embryonic stem cell-derived neuroepithelial stem cells (lt-hESNSCs, or lt-NES) described in a recent publication from his group (Koch et al., 2009). The lt-NES cells described by Brüstle, which are isolated from hESC-derived neural rosettes, are highly phenotypically homogeneous and retain the ability to form neural rosettes, denoting their undifferentiated status. Most importantly, lt-NES cells can be efficiently differentiated into neurons—preferentially GABAergic, even after prolonged passaging.

These cells can be used as an in-vitro model system to study diseases such as Alzheimer’s (AD). In fact, lt-NES-derived neurons express amyloid precursor protein (APP) as well as other AD-associated proteins, and possess the intact proteolytic machinery for the generation of amyloid-β (Aβ) from APP. The production of Aβ in these cells can be blocked by the γ-secretase inhibitor DAPT, confirming lt-NES-derived neurons as a valid model to screen for other compounds able to inhibit Aβ production. This model can be bettered by actively introducing AD-associated mutations via lentiviral delivery. In line with this idea, Brüstle and his team have recently generated lt-NES cells carrying artificially introduced presenilin mutations, and used them to study the pathogenesis of AD and test new compounds. The advent of the technology to generate iPSCs provided a further opportunity to refine lt-NES cells as a disease model. Indeed, Brüstle and colleagues succeeded in generating lt-NES cells from patient-derived iPSCs, and he pointed out how these cells represent a valuable tool to “reverse time” and follow the sequence of events that eventually leads to disease.

Brüstle is currently employing this model to study spinocerebellar ataxia type 3 (SCA3), also known as Machado-Joseph disease (MJD). This condition arises from the expansion of a trinucleotide repeat in the ATXN3 gene, which results in the accumulation of protease-resistant polyglutamine fragments that aggregate into toxic neural inclusions in the cell nuclei, eventually leading to neuronal loss. The protein cleavage mechanism leading to the generation of the polyglutamine aggregates is still unclear, which led Brüstle and colleagues to exploit patient iPSC-derived lt-NES to gain insight into the molecular pathogenesis of SCA3. At first, no major phenotypic difference was evident between SCA3 and control neurons, but stimulating neuronal activity by applying glutamate to the cultures revealed that patient-derived neurons, but not controls, responded with markedly increased proteolytic cleavage of ataxin and accumulation of insoluble polyglutamine aggregates (Koch et al., 2011). Interestingly, as presented in a poster at the meeting authored by D. Poppe from Brüstle’s lab, correcting the genetic mutation in patient-derived cells by adenoviral-mediated targeting completely prevented the glutamate-induced aggregate formation. Screening for compounds able to inhibit polyglutamine aggregation in patient cells led to the identification of a single hit, namely a calpain inhibitor. By analyzing other iPSC-derived cell types, Brüstle's lab confirmed that aggregate formation is strictly neuron-specific and, in particular, restricted to electrically mature neurons. Further experiments confirmed that membrane polarization and secondary activity of voltage-operated Ca+2 channels are necessary to trigger the activation of a yet unknown cleaver and induce polyglutamine aggregation. Of note, patient-derived neurons did not degenerate despite the accumulation of aggregates, suggesting that this may represent the early stage of pathogenesis, and other hits may be needed for the disease to manifest itself. This is consistent with the late onset of the symptoms, which normally appear in individuals between 30 and 50 years of age.

Next on the podium, Alysson Muotri from the University of California, San Diego, told us how he and his team are applying patient-derived iPSCs to the study of autism. When analyzing neutralized iPSCs obtained from individuals with different MeCP2 mutations (which are known to be causative in Rett syndrome), Muotri and coworkers noticed that their somas were significantly smaller compared to controls. In addition, Rett (RTT) neurons had almost no spines, and formed much fewer glutamatergic synapses. Most importantly, all the above abnormalities could be rescued by expression of exogenous MeCP2, and could be induced in control neurons by shRNA-mediated knockdown of MeCP2. Other defects evidenced in RTT neurons included Ca+2 influx deficiencies and reduced frequency and amplitude of spontaneous post-synaptic currents. Administration of IGF1, described to rescue some of the RTT-like neurological abnormalities in a mouse model, corrected the neuronal phenotype in iPSC-derived RTT neurons, providing the proof of principle that this model can be employed in drug screening platforms (Marchetto et al., 2010).

As Muotri pointed out, unlike Rett syndrome, many forms of autism are currently considered idiopathic, as they cannot be linked to any known mutation. Muotri and his group have recently reported the successful generation of iPSCs from dental pulp stem cells, providing a new accessible and non-invasive method to obtain patient-derived cells for the study of pediatric conditions. To extend his studies to idiopathic autism, Muotri and colleagues adopted a clever “tooth fairy” strategy to collect tissue from children with sporadic autism. This could then be subjected to genetic analysis to search for novel mutations, and also be used to generate iPSC-derived neurons in which to study the molecular consequences of such mutations. By adopting this approach, Muotri’s group found a novel autism-associated inversion disrupting two genes, one of which (TRPC6) had never been previously implicated in autism. The expression of TRPC6, which encodes for a Ca+2 channel, was found to be lower in autistic subjects carrying the inversion and, consistently, gene expression profiling of mutant cells evidenced deregulation of CREB target genes. Biochemical analyses also revealed lower levels of CREB phosphorylation in these patients. TRPC6 levels were found to regulate several aspects of neuronal morphology, including cortical spine density, number of glutamatergic synapses, and neurite length, all of which were impaired in iPSC-derived neurons from patients carrying the inversion, and, most interestingly, idiopathic patients. Intriguingly, many of the morphological abnormalities observed in TRPC6 mutant neurons closely resembled those seen in RTT neurons, indicating the possibility that these could arise from the disruption of a common molecular pathway. Muotri concluded by suggesting that the deregulation of TRPC6 levels may represent a common molecular feature of different forms of autism, a hypothesis strongly corroborated by his finding that MeCP2 regulates the expression of TRPC6.

The next speaker of the session was Guo-li Ming, who started by summarizing her group’s recently reported work that led to the generation of iPSCs from members of an American family with schizophrenia and a frame-shift mutation located at the exon-intron 12 region of DISC1 (Chiang et al., 2011). Of note, reprogramming of skin fibroblasts from mutation carriers and non-related controls was achieved without the use of virus-based methods, leaving the host genome intact. Ming and coworkers initially carried out a morphological characterization of DISC1 mutant and control iPSC-derived neurons maintained in vitro for up to four weeks, and observed some key differences between the two. In particular, DISC1 mutant neurons exhibited larger cell bodies, increased dendritic length up to two weeks of age, and increased synaptic density until three weeks of age. These morphological changes were intriguingly similar to the ones seen in mouse hippocampal dentate granule cells after shRNA-mediated knockdown of DISC1 (see SRF related news story). Ming and colleagues extended the analysis of these cells to a more physiological setting by co-transplanting iPSC-derived neural progenitors from patients and controls to the dentate gyrus of adult SCID mice. DISC1 mutant and control cells were labeled with different fluorophores before transplantation, so that they could be easily traced and distinguished in the recipient tissue. Seven days post-injection, Ming and coworkers detected active proliferation of the grafted cells, as well as some signs of differentiation triggered by endogenous signals. At this stage, both mutant and control cells expressed Prox1, a dentate granule cell-specific marker, denoting efficient integration in the host tissue. Twenty-eight days post-injection, both DISC1 mutant and control neurons were still present in the host tissue, where they had developed normally and formed spines. Interestingly, DISC1 mutant, but not control, neurons exhibited ectopic dendrites, a feature previously observed by Ming’s group in their DISC1 knockdown mouse model (see SRF related news story).

Lorenz Studer from Memorial Sloan–Kettering Cancer Center, New York, then described work building on their recent report that dual inhibition of SMAD signalling (SMADi) induces highly efficient neural differentiation of hESCs and iPSCs (Chambers et al., 2009). Studer and colleagues set out to identify molecules that can further increase the speed and efficiency of the differentiation process. They found that combining dual SMADi with the GSK-3β inhibitor CHIR99021, a FGF/VEGF inhibitor, and a γ-secretase inhibitor induced differentiation of 75 percent of the cells into nociceptive neurons expressing different types of nociceptive markers. Studer remarked that these cells represent a valuable source for modeling peripheral pain, and a novel platform for drug discovery.

In the second part of his talk, Studer presented his group’s recent work on familial dysautonomia (FD), a fatal peripheral neuropathy caused by a rare point mutation in the gene IKBKAP, involved in transcriptional elongation (Lee et al., 2009). The pathogenesis of this disease and the determinants of its specificity for the peripheral nervous system are still unclear, but the IKBKAP mutation is known to result in missplicing of the primary transcript, resulting in reduced protein levels and loss of autonomic and sensory neurons. To gain a better understanding of the molecular pathogenesis of FD, Studer and colleagues derived iPSCs from patients and controls, and directed their differentiation towards different cell types, representing all three embryonic cell layers. Consistent with the tissue specificity of the disease, the level of normal IKBKAP transcript was lowest in iPSC-derived neural crest precursors (the precursors of the peripheral nervous system) from FD patients. Further validation of FD iPSC-derived neural crest precursors as a disease model came from the observation that kinetin, a drug known to reduce the levels of mutant IKBKAP transcript in patient-derived lymphoblastoids, had the same beneficial effect on FD neural crest cells. Next, adopting high levels of mutant IKBKAP transcript as the FD marker, Studer set up a large-scale screening to identify new compounds that could rescue this phenotype. Several hits demonstrated the ability not only to rescue the normal transcript levels and induce protein expression, but also to favor the differentiation of neural crest precursors into autonomic neurons. When investigating the mechanism of action of one of these compounds, an antagonist of the α2 adrenergic receptor, Studer and coworkers found that it activates CREB, and that its action can be blocked by the PKA inhibitor H89, indicating that it functions at least in part by activating the cAMP/PKA signaling pathway. Studer concluded by suggesting that, since the established drug kinetin is known to act via a different mechanism, combining kinetin with compounds that stimulate CREB activation may significantly improve the therapeutic effect.

The final speaker of the session was Marius Wernig from Stanford University, California, whose recent work demonstrated the possibility of directly converting fibroblasts into neurons. In a recent publication, Wernig and his group reported that forced expression of a cocktail of three factors (Brn2, Ascl1, and Myt1l) is necessary and sufficient to rapidly and efficiently convert mouse fibroblasts into functioning neurons, which they named induced neural (iN) cells (see SRF related news story). Wernig’s team went on to show that iN cells can also be generated from human fibroblasts using the same combination of factors plus the transcription factor NeuroD1 (see SRF related news story). Next, they set out to identify the minimal combination of factors that can reprogram human fibroblasts to neural progenitor cells, an intermediate and self-renewing cell type that can be easily maintained and expanded in vitro. To do this, they started by expressing a combination of 11 candidate factors in a Sox-2 EGFP fibroblast line. Two of these factors, Sox-2 and FoxG1, were identified as sufficient to generate cells that possessed phenotypic and functional features of neural progenitors, but that could not be consistently induced to differentiate both into neurons and glial cells. By extending their research to other factors, Wernig and colleagues discovered that the inclusion of a third factor, Brn2, generated stable neural progenitors that could be efficiently differentiated not only into neurons, but also into functional myelinating oligodendrocytes, providing a novel valuable model to study a large spectrum of diseases of the central nervous system.—Elise Malavasi.

Comments on Related News

Related News: Neural Progenitor Cells Model Aspects of Schizophrenia

Comment by:  Nao GamoAkira Sawa (SRF Advisor)
Submitted 7 May 2014
Posted 7 May 2014

This study introduces a novel use of neural progenitor cells (NPCs) derived from human induced pluripotent stem cells (hiPSCs) to address mechanisms that may possibly underlie a predisposition to schizophrenia. Brennand et al. (2014) generated hiPSC-derived NPCs from patients with schizophrenia and control subjects. These NPCs, as well as six-week-old neurons differentiated from them, showed gene expression profiles similar to those of the fetal forebrain. Thus, these cells were used to address early disease etiology, in particular, focusing on mechanisms related to disruptions in prefrontal cortical development. Interestingly, the researchers found overlap in gene signatures between the six-week-old neurons and NPCs from patients, raising the possibility that disease predisposition may already be established at the NPC stage.

Particularly striking is the reduced migration of schizophrenia NPCs relative to control NPCs as they differentiated into neurons. This reduced migration may be due to schizophrenia NPCs remaining in a proliferative state before differentiating, as suggested by previous work from our group (Ishizuka et al., 2011). The authors also proposed that this reduced migration might lead to reduced synaptic connectivity, which they previously reported in schizophrenia hiPSC-derived neurons (Brennand et al., 2011).

The study also found differential expression of various genes and proteins, including those involved in neuronal differentiation and migration, glutamate receptor signaling, and cellular adhesion. The schizophrenia NPCs showed smaller mitochondria with altered cellular distribution relative to control NPCs, as well as oxidative stress, although the effects of oxidative stress might be limited to a subset of schizophrenia NPCs. This is a telling observation, in light of recent human and animal studies that suggest a role for oxidative stress in schizophrenia (Emiliani et al., 2014).

While it is as yet unknown whether these observations truly reflect disease predisposition, this work is innovative in taking advantage of the tools at hand. It is understood in the field that hiPSC-derived neurons can take months to fully functionally mature, and it would be difficult to simulate experience-dependent shaping of neuronal networks in a dish. However, instead of tolerating such shortcomings, the authors have used them to their advantage by addressing mechanisms that may occur at the fetal stage. Furthermore, NPCs are proliferative and suitable for high-throughput assays. This point is particularly useful when studying a disease with such heterogeneous etiology.

The true value of this experimental system will be revealed when it can predict actual brain mechanisms and clinical characteristics of individuals as well as groups of patients. The authors acknowledge that the sample size is currently small, and that the effect sizes of their observations are insufficient to predict diagnosis. It would be interesting to observe neuronal phenotypes in cells from patients with similar clinical characteristics. In addition, cells from patients with similar genetic backgrounds should be tested to control for possible biases in genetic architecture, for example, in cells from family members, or known mutations that can be created in control cell lines. We are optimistic that hiPSCs will prove a useful tool to study biological mechanisms of schizophrenia (Gamo et al., 2014, in press).


Brennand KJ, Simone A, Jou J, Gelboin-Burkhart C, Tran N, Sangar S, Li Y, Mu Y, Chen G, Yu D, McCarthy S, Sebat J, Gage FH. Modelling schizophrenia using human induced pluripotent stem cells. Nature . 2011 May 12 ; 473(7346):221-5. Abstract

Emiliani FE, Sedlak TW, Sawa A. Oxidative stress and schizophrenia: recent breakthroughs from an old story. Curr Opin Psychiatry . 2014 May ; 27(3):185-90. Abstract

Gamo and Sawa (in press). Human Stem Cells and Surrogate Tissues for Basic and Translational Study of Mental Disorders. Biol. Psychiatry.

Ishizuka K, Kamiya A, Oh EC, Kanki H, Seshadri S, Robinson JF, Murdoch H, Dunlop AJ, Kubo K, Furukori K, Huang B, Zeledon M, Hayashi-Takagi A, Okano H, Nakajima K, Houslay MD, Katsanis N, Sawa A. DISC1-dependent switch from progenitor proliferation to migration in the developing cortex. Nature . 2011 May 5 ; 473(7345):92-6. Abstract

View all comments by Nao Gamo
View all comments by Akira Sawa

Related News: Neural Progenitor Cells Model Aspects of Schizophrenia

Comment by:  Bryan MowrySamuel Nayler
Submitted 29 May 2014
Posted 29 May 2014

In a recent follow-up to their 2011 paper, Brennand et al. report considerable progress toward generation of a defined neuronal population generated from patient-derived induced pluripotent stem (iPS) cells. The advent of the iPS cell has been somewhat Promethean in that pluripotent stem cells are now a commonly utilized laboratory tool for disease modeling. While marked progress has occurred on a number of fronts, it is still not known to what degree stem cell-derived neurons truly resemble mature neurons that exist in the brain of a living human. Moreover, it is an open question what these cells can tell us about the onset of a clinically heterogeneous, polygenic disease such as schizophrenia.

Using gene expression analysis, Brennand et al. compare their samples to a developmental spectrum of samples from the Allen Brain Atlas to show that their iPSC-derived neurons most closely resemble early fetal forebrain neurons. This may provide precisely the model system that will allow researchers to validate the neurodevelopmental theory of schizophrenia, provided early molecular mechanisms can be identified that predispose to schizophrenia in later life. Brennand and colleagues go on to show functional phenotypic differences in schizophrenia patient-derived neurons relating to elevated oxidative stress and extra-mitochondrial oxygen consumption, as well as reduced migrational ability. It remains to be seen how relatable these phenomena are to events which occur in vivo and whether they may be informative in identifying and characterizing the underlying molecular and cellular mechanisms in schizophrenia.

View all comments by Bryan Mowry
View all comments by Samuel Nayler