Final Report Summary - DDPDGENES (Identification of genes important for human midbrain dopamine neuron development and Parkinson’s disease) Executive Summary:Developmental genes are well known for their ability to regulate multiple aspects of embryogenesis. However, some of these genes have been recently found to serve critical functions in the maintenance of the adult nervous system. For instance, midbrain dopaminergic (DA) neurons die in adult transgenic mice heterozygous for transcription factors necessary for midbrain DA neuron development, such as Nurr1, Pitx3, FoxA2, or En1. Interestingly the loss of DA neurons is responsible for the main symptoms of Parkinson‟s disease. Surprisingly, however, genes associated with genetic forms of Parkinson‟s disease (PD), such as LRKK2 and PINK1, do not cause cell death. We (the consortium) therefore hypothesized that adult onset PD may involve a dysregulation of the expression of developmental genes involved in the specification of DA neuron subtypes and their interaction with PD-associated genes. The goal of this project was to explore this hypothesis. More specifically we aimed to identify molecularly-defined subtypes of midbrain dopamine neurons and determine which genes and pathways are functionally relevant and differentially regulated in PD. In this way the project directly contributed to the objective of the Call, namely to elucidate the relationship between development and ageing.Our main findings were: (1) an unexpected heterogeneity of dopaminergic neuron subtypes, found in mouse and confirmed in human, with five adult subtypes emanating from two embryonic types; (2) human and mouse ventral midbrain development are highly conserved, including homologous cell types and similar molecular properties, albeit with key differences in crucial transcription factros; (3) preliminary findings show that multiple subtypes can also be defined by electrophysiological properties, but these may not align with molecularly defined subtypes; (4) stem cell-derived dopaminergic neurons, intended for transplantation or disease modelling, are highly heterogeneous and appear to recapitulate normal development in distorted form; (5) selective LSD1 and HDAC inhibitors are potent modulators of dopaminergic neuron differentiation in vitro. Project Context and Objectives:In DDPDGenes we have focused both on understanding the gene regulatory networks that control the specification of midbrain DA neuron subtypes and determine their importance for the selective degeneration of substantia nigra DA neurons in PD. We have examined the development of DA neuron subtypes, first in rodents, then in human cells obtained from fetal ventral medial (VM) tissue, and finally in DA neurons derived from human and rodent neural and embryonic stem cells and human PD-specific induced pluripotent stem (iPS) cells. Findings in these cells were compared to results for postmortem tissue from PD and control patients.Initial studies were conducted at the tissue level using immunohistochemistry and in situ hybridization (ISH), to define cell compartments. These analyses were followed by studies at the single cell level. We used gene expression profiling, 3D reconstructions of neuron morphologies, histological techniques and electrophysiological recordings, which will make possible to characterize different neuron subtypes, to map their expression in the developing rodent brain, to compare findings in developing, adult and PD DA neurons and to identify genes that are differentially expressed in different subtypes and/or in PD. The specific objectives were:1.- To characterize mouse midbrain DA neurons in the developing and the adult rodent brain by multiplex single-cell RNA-seq, 3D reconstructions of single neuron morphology, and multichannel electrophysiology in single cells. (MS1, MS2, MS3)2.- To characterize human midbrain DA neurons by multiplex single-cell RNA-seq, 3D reconstructions of single neuron morphology, and multichannel electrophysiology in single cells. (MS14, MS15)3.- Characterize specific DA neuron populations produced in vitro, from human and rodent neural stem cells (NSCs), embryonic stem (ES) cells and induced pluripotent stem (iPS) cells. (MS6, MS7, MS8, MS11, MS12)4.- Compare midbrain DA neurons derived from human fetal midbrain tissue with those derived from hNSC, hES and iPS cells. These results will allow to determine the quality of the DA differentiation protocols and to instruct improvements in current protocols. (by month 48, D4.4)5.- Identify changes in gene expression and interactions between developmental and disease- associated genes, between control and PD-iPS-derived DA neurons. (by month 48, D4.4)6.- Identify the similarities and differences, if any, between subtypes of human midbrain DA neurons in parkinsonian and control brains, compared to control and PD-iPS cells. (by month 48, D4.4)7.- Verify the function of genes expressed in DA neuron subtypes by gain and loss of function experiments in vitro and by expression in postmortem PD specimens. (MS4, MS8, MS9)8.-Test novel drugs aimed at the treatment of PD in human DA neurons subtypes, specifically the highly selective and potent LSD1 and HDAC inhibitors (MS5, MS10).Project Results:1. Heterogeneity of dopaminergic neuron subtypesand 2. Human and mouse ventral midbrain development is highly conservedDuring the four-year period of the DDPDGENES project, we have developed protocols and generated data with the aim to describe the development of dopaminergic (DA) cells into fully mature DA cell subtypes. To this end we invested considerable effort into generating single cell transcriptome (SCT) profiles across embryonic and post natal development, as this would enable us to unravel DA cell subtypes and the molecular processes underlying their development. The mouse was used as an animal model representing mammalian DA system, while we supplemented this with data from human embryos to highlight similarities and differences between human and mouse midbrain development and maturation. Mouse SCT profiles were generated both by random collection of cells and by FACS sorting fluorescent DA cells from the developing midbrain. In situ hybridization (ISH) and histochemistry (HC) was used identify the anatomical distribution of the identified cell subtypes during development. We found in both mouse and human that dopaminergic neurons developed in a series of steps from neural progenitors, to neuroblasts and then diverged into two distinct populations at late embryonic stages (e.g. E18.5 in the mouse). We believe that the first identified cell of the dopaminergic lineage proper was a neuronal progenitor (mNProg) in the ventricular zone, which expressed Lmx1a, a gene that defines the midbrain floor plate (m7) and is required for dopaminergic neuronal development in this region, Two types of intermediate zone neuroblasts in the intermediate zone were found to contribute to the dopaminergic: a more generic medial neuroblast (mNbM), characterized by the expression of not only Nr4a2 (also known as Nurr1), but also transcription factors such as NeuroD1, NeuroD2, Klf12 and Nhlh1. Notably, mNbM and mNProg both expressed Hes6 and Neurog2, the latter required for dopaminergic neurogenesis. The second neuroblast was a specific dopaminergic neuroblast (mNbDA) that expressed Pbx1 as well as Pitx3, a transcription factor required for the development of the substantia nigra. Finally, we found three distinct types of embryonic dopaminergic neurons (Figure 5A): (i) mDA0, which in addition to the factors above expressed tyrosine hydroxylase (Th), the rate-liming enzyme in the synthesis of dopamine; (ii) mDA1 neurons, which additionally expressed the dopamine transporter, Slc6a3; and (iii) mDA2 neurons, distinguished by the specific addition of Aldh1a1 and the transcriptional co-regulator Lmo3, a LIM domain only protein regulated by dopamine which interacts with basic-loop-helix proteins to regulate neurogenesis. Based on these findings, we believe that mDA1 and mDA2 represent two distinct subtypes of dopaminergic neurons in the newborn mouse and late-stage human embryo.To further elucidate the impact of postnatal development and final maturation of dopaminergic neurons, we performed a separate analysis of young adult mouse midbrain dopaminergic neurons by single-cell RNA-seq. Clustering revealed five distinct types of molecularly defined dopaminergic neurons, which mapped to partly different anatomically defined regions. All types shared expression of TH as well as a newly identified pan-dopaminergic marker, AJAP1, validated by immunohistochemistry. One cell type mapped to the substantia nigra pars compacta (mDA-SNC), two to the ventral tegmental area (mDA-VTA1 and mDA-VTA2), and two to different domains of the VTA and cells in the SNC (mDA-VTA3 and mDA-VTA4). These five cell types resembled those previously found by others by single-cell Q-PCR analysis of a limited number of genes on FACS- sorted Slc6a3+ cells. However, the whole-transcriptome analysis adopted in our study provides an unbiased and systematic analysis of all Th+ cells in the adult midbrain and helped identify a large number of cell type-specific genes, providing a comprehensive molecular definition of adult dopaminergic neuron subtypes.Thus in summary, we achieved our goal of discovering subtypes of dopaminergic neurons in the developing and adult mouse and human nervous system. We found that human development closely follows that found in the mouse, with a few key differences. For example, the transcription factor Pitx3, which is required for DA neuron development in the mouse, was not detected in human (or only at very low levels), suggesting its role has been superseded by another factor in humans.3. Multiple subtypes can also be defined by electrophysiological properties, but these may not align with molecularly defined subtypesThe aim was to classify DA cells of the Substantia Nigra pars compacta (SNC) according to their electrical and anatomical properties across post natal development, and to associate these functional subtypes with their genetically defined counterparts. In this way, it would be possible to explain functional differences in terms of differential gene expression, e.g. by associating properties of expressed ion channels with electrophysiological characteristics. It would also enable us to establish to what extend SCT profiles can be used to predict functional characteristics. Such a description would be of tremendous value for research into PD, as it would allow researchers to examine which of the identified subtypes that are most vulnerable to degeneration and if there is anything in their genetic or functional markup that can help explain their particular susceptibility to degeneration. It would also help inform cell replacement therapies, providing a unique reference for DA cell subtype characteristics. This was a very ambitious endeavor, requiring the development of novel protocols not included in the original proposal to accommodate unforeseen difficulties encountered during the project. This was without doubt one of the more challenging parts of the proposal. We established a genetic strain of mice at EPFL that express a fluorescent marker in DA neurons (DAT-Tomato), which allowed us to specifically target DA cells in the SNpc for whole-cell patching, morphological reconstruction and SCT profiling. The functional analysis was focused on post natal development, as these properties largely mature in the DA cells after birth. Our analysis showed that prior to birth these cells cannot be induced to fire action potentials and have very indistinct morphological features. In fact, many of the electrophysiological properties manifest themselves in the first few post natal days and continue to mature in the following weeks. This coincides with maturation of the DA system, as the DA cells in the VTA and SNC grow in size and the axons reach their target in the striatum. It also coincides with the period of diversification of DA neurons from two embryonic subtypes to five adult.Morphologically, we observe what appears to be a few distinct subtypes, that mature during post natal development. In the first post natal days the morphologies are typically smaller with less elaborate dendritic arborizations. At older ages the cell body is quite large (30-50um) with a dendritic tree covered with boutons that doesn’t extend much beyond the SNC. Cells are typically bitufted or have a few dendrites merging in different directions from the cell body. The axons project out of the SNC toward their target in the striatum and have not been reconstructed in their entirety. Axons are therefore not used as a morphological feature of DA cells in this study, only dendrites and cells bodies are used to distinguish DA cell morphologies. As for the electrophysiology, it remains a challenge to find unique features that can be used to classify DA cell morphologies into distinct subtypes. Currently, we have a pipeline (curtsey of the Blue Brain Project) that automatically extracts almost 300 morphology features from the raw digital reconstruction. Additional manually identified features were added to the morphology feature table, which remained difficult to extract automatically from the digital reconstructions (e.g. regarding cell body shape). Preliminary analysis using these features suggests that there are seven distinct morphologies. A decision tree was reconstructed based on the seven parameters, and interestingly six out of seven groups can be identified with 100% accuracy based on just one single parameter. Only group 6 and 7 remains difficult to group based on the decision tree. More work is required to refine and validate the morphology subtypes until we can devise a strategy that allows us to classify DA cells based on visual inspections of key features. 4. Stem cell-derived dopaminergic neurons, intended for transplantation or disease modelling, are highly heterogeneous.Single cells from hLT-NES, hES and cells reprogrammed into mDA neurons have been analyzed and compared to midbrain cells. These results will help guide the production of mDA neurons from stem/reprogrammed cells and will allow researchers to enrich for the desired subtypes such as nigral DA neurons that may be more resistant to stress (to be used for transplantation), or that may capture some of the features of PD (to be developed in WP3).The cell content and quality of commercially available human induced PS (iPS)-derived dopaminergic neurons generated using GSK3β inhibitors was examined by immunocytochemistry and single cell RNA-Seq. Clustering of RNA-seq data revealed an unexpected heterogeneity within these cultures, with at least 13 distinct iPS derived cell types, which resembled the 15 human neural cell types present in the human fetal midbrain tissue. Importantly none of these cells showed a positive proliferation index. The cell types identified included three radial glia-like cell types and three neuroblast-like cell types. Notably, we found three iPS cell-derived dopaminergic cell types (iDAa, iDAb and iDAc) with features of human fetal dopaminergic neurons, hDA0, hDA2 and hDA1, respectively. Surprisingly, four additional neuronal types, iRNa-b and iOTNa-b, which resembeled human red nucleus (hRN) and oculomotor and trochlear neurons (hOMTN) were also found.The quality of iPSC-derived dopaminergic neurons was assessed by a newly developed tool to score the similarity of each in vitro cell (iPS-derived) to canonical cell prototypes (human midbrain) based on the transcriptomes of major cell types in vivo. Surprisingly, 25% of the iPS-derived cells could not be classified based on endogenous week 7 midbrain cell types, which showed distinct and unambiguous identities. Most of the remaining 75% iPS-derived cells clustered along the desired phenotypes (DA and Nb), but only few high-scoring dopaminergic neurons were found. Many iPSCs scored lower quality than the corresponding cells in the brain and showed a spectrum of ill-defined intermediate forms, which indicates that there is room and possibly need for further improvement in DA differentiation protocols before such cells are used in cell replacement therapy.Finally, considering the safety of stem cell preparations, the finding that all of the cells exhibited low cell cycle indexes, is reassuring. However, it remains to be determined whether the unclassified cells in the iPSC preparations may give rise to abnormal or even harmful cell types. Future single-cell analysis of iPS-derived midbrain cells transplanted in animal models should reveal the final destiny of such cells. More broadly, our work demonstrates a powerful strategy to quantitatively and qualitatively assess the composition of cell preparations intended for human transplantation in any disease. We propose that single-cell analysis, and quantitative comparison with in vivo cell prototypes, should be a preferred strategy for assessing the safety and performance of cell preparations for clinical applications.5. Selective LSD1 and HDAC inhibitors are potent modulators of dopaminergic neuron differentiation in vitro. (the precise nature of these findings is confidential; please refer to the project reports for details)Potential Impact:1. Main impact of the projectAn increased understanding of mouse and human ventral midbrain development into adulthood, reported in a manuscript currently in revision at Cell; as well as a number of ancillary publications and manuscripts. In particular, including a comprehensive analysis by single-cell RNA-seq, electrophysiology and morphology.A comprehensive resource of high-quality single-cell RNA-seq expression data from developing and adult mouse and human ventral midbrain; deposited at the Gene Expression Omnibus (GEO) repository.A computational tool to assess the composition and quality of stem cell-derived dopaminergic neurons intended for transplantation or disease modelling. This tool aligns single-cell RNA-seq data from cultured cells with the reference atlas of normal mouse or human development, and defines cell types and their quality in relation to prototypic cells in the the reference.Perhaps the greatest impact will be on future clinical trials involving transplantation of stem cell-derived dopaminergic neurons. Such trials will almost surely incorporate quality assurance using single-cell RNA-seq and evaluation against our reference developmental atlas. Such qualito assessment will improve outcomes and ease the interpretation of unsuccessful transplants.2. Dissemination activities2.1. Web presenceDuring the first semester of the project a Facebook page called “The Dopaminergic Neuron” was created. At the petition of the commission representatives, after the evaluation of the first reporting period we also created a more traditional web site, containing the project and partner description, the announcements of the DDPDGENES meetings at http://www.ddpdgenes.eu/.The DDPDGENES project is also described on the web sites of individual partners.2.2. Patient outreach and communication to the publicWe organized at least two Open Day events annually in collaboration with local PD Patient’s organizations, and science was mixed with the personal experiences of Parkinson’s patients. These Open Days were well received by the invited participants, and were a source of motivation to the collaborating scientists in the DDPDGENES project. Around ten public talks were given every year of the project, of which a substantial number was directed at patients or the public at large. 2.3. Dissemination to the scientific communityApart from numerous talks at sceintific meetings, so far, six scientific papers have been published or submitted for publication.3. Exploitation of resultsThe major part of the results are in their nature basic research findings and thus not immediately exploitable in the narrow sense. One partner (Oryzon) has commercial interests in the findings specifically using HDAC and LSD1 inhibitors, which will be protected and exploited possibly for the use as drugs. These findings are detailed in the confidential part of this report (e.g. deliverables and project report of the final period).List of Websites:http://www.ddpdgenes.eu
