Neurons are the primary information-processing cells in higher eukaryotes, performing essential functions in stimulus reception, signal transmission, and adaptive responses. In humans, neuronal dysfunction can cause a variety of clinical disorders, including developmental conditions such as autism, psychiatric illnesses such as schizophrenia, and degenerative diseases such as Alzheimer’s dementia. Neurons are unique among cell types in their large and functionally polarized structure, post-mitotic state, and electrochemical activity. The human nervous system is unique in both its complexity and susceptibility to disease, and animal models often fail to recapitulate key characteristics of human neuronal disorders (D’Alton et al., 2014; Kitazawa, Medeiros, & Laferla, 2012). In order to understand normal human neuronal physiology as well as how cells and networks malfunction in disease, better experimental models of neuronal cell biology are needed.
Historically, the two main in vitro model systems for studying cellular and molecular neuroscience have been rodent primary neurons and human immortalized cell lines. Primary rodent neurons have specialized machinery unique to neurons, but, as they originate from another species, these cells may not recapitulate relevant aspects of human genetics or disease pathophysiology (Kloskowska et al., 2010; Nuber et al., 2013). Practically, primary neurons are time-consuming to isolate, can vary in quality from preparation to preparation, are difficult to scale for some applications, and are difficult to genetically engineer once isolated. Human immortalized lines such as HeLa, HEK293T, and U2OS, along with neuroblastoma lines such as SH-SY5Y, circumvent many of these challenges; they are easily cultured, relatively homogenous, scalable, and readily manipulated genetically.
However, they have widespread and unstable genotypic abnormalities and lack a truly neuronal phenotype, so they are poorly suited to study neuron-specific biology such as axonal or synaptic phenomena.
The first derivation of human embryonic stem cell lines in 1998 was soon followed by techniques to manipulate developmental pathways in order to promote differentiation into cell types of interest (Bain & Gottlieb, 1998; Kawai et al., 1998; Shimazaki, Arsenijevic, Ryan, Rosenfeld, &Weiss, 1999; Thomson et al., 1998). The scalability and genetic tractability of stem cells finally permitted large populations of normal human cells to be grown in vitro, and for the first time human neuronal cell culture became a viable model system. The landmark development of induced pluripotent stem cell (iPSC) technology further enabled the reprogramming of patient-derived cells to establish stem cell lines capable of differentiating and recapitulating cellular disease phenotypes in culture (Takahashi & Yamanaka, 2007). The field quickly embraced these transformative tools (Chambers et al., 2009; Wernig et al., 2008), and, together, their success revolutionized the study of human neurobiology; no longer were scientists limited to choosing between cell-type and species specificity, and the use of stem cells greatly facilitated genetic engineering. Still, initial methods for differentiating iPSCs to neurons or neuralprecursors were far from simple; all involved complex media formulations and lengthy protocols (Hu & Zhang, 2009; Karumbayaram et al., 2009; Zeng et al., 2010). Recent approaches have simplified the process by employing primarily small molecules, and the use of overexpressed or inducibly-expressed transcription factors have simplified it further.
Most small molecule-based differentiation approaches rely on a combination of pathway inhibitors (e.g., noggin, SB431542) to drive ES or iPS cells toward neuroectodermal development. This process results in neural progenitor cells (NPCs), which must be coaxed to differentiate further with different small molecules and growth factors to finally produce the desired neuronal subpopulation. There are three major drawbacks with these strategies: small molecules are not highly efficient, individual cells transduce signals at different rates, and different iPSC clones (even from the same patient) can respond differently to the same small molecules. Together, this leads to a mixed population of neural progenitors and various neural and glial cell types, producing batch-to-batch and line-to-line variability. Particularly for long-term cultures, proliferative cells can quickly outcompete post-mitotic neurons of interest, and the presence of multiple cell types complicates downstream analysis, especially for high-throughput microscopy screens or “-omics” applications. Finally, small molecule-based methods are often laborious, expensive, and slow. From a survey of articles describing small molecule-mediated iPSC differentiation to neurons in the past year (Cao et al., 2017; Kikuchi et al., 2017; Qi et al., 2017), timelines extended from 13 to 70 days, and involved between four and six medium-formulation changes over those periods. This technical and time burden decreases laboratory output, increases the likelihood of contamination, and creates barriers to entry for new users.
Transcription factor overexpression is a new approach to neuronal differentiation from iPSCs that circumvents many challenges associated with small molecule pathway inhibitors. Initially demonstrated by Zhang et al. (2013), overexpression of the master neuronal transcriptional regulator neurogenin-2 (NGN2) results in rapid, one-step differentiation of iPSCs to functionally mature glutamatergic cortical neurons. Similar results were independently obtained by the study of Busskamp et al. (2014), in which NGN2 was virally delivered and inducibly expressed in iPSCs, resulting in synaptically mature cells in 14 days. Both studies recorded upwards of 90% differentiation efficiency and purity, as measured by immunostaining of characteristic cortical neuron markers. With fewer medium changes and relatively rapid differentiation time, NGN2 overexpression offers an appealing alternative to small molecule differentiation strategies.
The set of protocols described here follows an improvement to the NGN2 method by Wang et al. (2017). In that study, the neurogenin-2 transgene was stably integrated into a safe-harbor locus in iPSCs under a doxycycline-inducible promoter. Clonal isolation of this stably integrated line enables near 100% efficiency and purity of differentiation to glutamatergic cortical neurons within the previously observed 14-day timeline, and simplifies differentiation to a two-step protocol. These cells, termed i3Neurons (integrated, inducible, and isogenic), offer a substantial improvement in efficacy and ease-of-use over other existing iPSC-to-neuron differentiation strategies.
Here we provide a detailed set of protocols for the generation and use of i3Neurons, and a related technique to generate lower motor neurons (i3LMNs), which also includes overexpression of the transcription factors Islet-1 (ISL1) and LIM Homeobox 3 (LHX3) along with NGN2. The described techniques mostly require only basic laboratory instrumentation and reagents, and can be completed without any specific training beyond mammalian cell culture proficiency, making them appealing to a wide range of laboratories. Basic Protocols 1 to 4 provide an update on the state of the art of human iPSC culture and transgenic-line generation, with Support Protocol 1 describing a genotyping strategy for confirming stably integrated lines. Basic Protocols 5 to 8 discuss the differentiation and culture of i3Neurons and i3LMNs, with Support Protocols 2 to 5 providing specific instructions on immunocytochemistry, transfection, transduction, and live imaging of differentiated neurons. Support Protocols 6 to 7 provide instructions on assessing induction efficiency, as well as optional culture supplementation with astrocytes.
The use of transcription factors (TFs) to specify iPSC differentiation follows naturally from the use of Yamanaka factors (Oct4/Sox2/Klf4/c-Myc) to generate iPSCs from differentiated cells. Whereas small molecules elicit inconsistent effects based on cell-to-cell variability in uptake and metabolism, stably integrated transcription factors enable uniformly efficient differentiation. These strategies are gaining momentum in the iPSC field, with recent reports detailing TF-based strategies for iPSC-derived motor neurons, oligodendrocytes, and even pancreatic beta cells (Goto et al., 2017; Major, Powers, & Tabar, 2017; Zhu, Liu, Zhou, & Ikeda, 2017). Ultimately, TF-based differentiation offers higher efficiency, higher purity, and a shorter timeline for producing the desired cell model when compared to small molecule differentiation methods. For laboratories with expertise in iPSC culture, TF-based methods have the potential to accelerate productivity, while for laboratories new to iPSC technologies, TF-based methods offer a low barrier to entry. The i3Neuron and i3LMN strategies described in this article express these TFs in an integrated and inducible fashion, offering the user maximal facility and control. For instance, multi-transgenic iPSC lines can be constructed piecemeal over time, with no concerns over loss of hNGN2 or hNIL expression. These lines can then be differentiated on a small or large scale, matured and assayed experimentally, or cryopreserved in a convenient pre-differentiated stock.
The main limitation with our strategy is the unavailability of the Tet-On promoter system for precise control of other transgenes, such as toxic genes. The 3-day neuronal differentiation period with doxycycline will also induce any other genes controlled by the Tet-On promoter, and overexpression of toxic genes (e.g., TDP-43 overexpression for ALS modeling) at this critical timemight compromise the quality of the resulting neurons.Methods that we have used to counter this challenge include implementing a shorter differentiation period (terminal differentiation is induced within 24 hr of continuous doxycycline exposure; Busskamp et al., 2014), inserting Kozak sequences with short upstream open reading frames (uORFs) to attenuate expression from the Tet-On promoter, and employing orthogonal inducible systems.
Critical Parameters and Troubleshooting
DNA quality for transfections
Efficient transfections are critical for facilitating downstream enrichment and clonal isolation when constructing a new iPSC line. Low efficiency transfections will require several weeks of enrichment, whereas an exceptionally high efficiency transfection could result in clonal isolation after the initial serial dilution or FACS enrichment of transfected cells.We have observed that the best indicators of DNA quality (apart from proper 260/280 and 260/230 spectrophotometric ratios) are concentration and preparative scale. Concentrating DNA, either through a new prep or ethanol precipitation, almost always improves efficiency. DNA for transfection should ideally be at a concentration higher than 1 μg/μl, and should almost certainly be concentrated if below 300 ng/μl. Miniprep DNA tends to produce less efficient transfections than either midi- or maxiprep DNA at the same concentration, likely due to higher bacterial endotoxin levels in smaller preps. Since endotoxins cannot be removed from existing preps, DNA to be used for transfection should ideally be prepared in a maxiprep.
Coating and freshness of medium for neurons
Plate coating for neurons (PLO for hNGN2 neurons and PLO/PEI with laminin for hNIL neurons) is the single most important parameter determining long and short-term neuronal health. Coating solutions should always be freshly prepared, and any stocks should be used within 1 week. Coated and washed dishes may also be stored covered at 4°C for only 1 week. Though we have not observed a striking cutoff point for coating viability, we have observed a steady decline in neuronal health when older coating solutions have been used. Likewise, neuronal media (IM, CM, or MM) should be prepared fresh with supplements in small scale batches (i.e., 50 to 100 ml), and supplemented media should be used within 1 week.
FACS conditions for iPSC sorting
FACS conditions must be carefully selected to ensure iPSC viability. Microfluidic flow sorting machines (e.g., Sony SH800S, NanoCellect WOLF Cell Sorter) employ low pressures for sorting, and are therefore ideal to preserve iPSC viability. If only conventional sorting machines are available, the nozzle should be adjusted to 100 μm (versus the 30-μm standard nozzle) in order to lower the sorting pressure. For bulk sorting, iPSCs should be resuspended for sorting in E8 + ROCK inhibitor, and they should be sorted into Matrigel-coated dishes with E8 + ROCK inhibitor. Improved survival and outgrowth from single cells may be accomplished by resuspending in StemFlex medium (Gibco, cat. no. A3349401) with ROCK inhibitor and plating on wells that have been coated with rhLaminin-521 (Gibco, cat. no. A29249). If routine passaging was done in E8 and E8 + ROCK inhibitor, these cells should be pre-equilibrated to StemFlex for at least 2 passages.
Neurons can be stained in situ, but special care must be taken to prevent dissociation of individual cells or sheets of neurons from the dish. It is therefore critical to reduce the total number of washes (e.g., using 2× fixative solution added directly to the neuronal medium) and carry out washes slowly (see Support Protocol 2).
Lentiviral vectors for i3Neuron and i3LMN transduction
We have found that iPSCs and iPSCderived neurons (i3Neurons and i3LMNs) silence certain viral elements in transgenic constructs. In particular, CMV promoters and IRES elements consistently fail to express.We suggest replacing CMV promoters with CAG, PGK, or EF-1α promoters, and replacing IRES elements with T2A linkers for polycistronic gene expression.
Enhancing stable cell line generation
Establishing stable lines of iPSCs using the safe-harbor integration strategy requires induction of double-stranded breaks at defined genomic loci (e.g., AAVS1 or CLYBL). One of the major roadblocks in the clonal enrichment and isolation process is the inefficiency of stable transgene integration (<1%). In addition to the low frequency of homologous recombination events between the donor plasmid and the genomic integration locus, a high frequency of p53-mediated death has been observed in cells that successfully undergo TALEN- or CRISPR/Cas9-mediated doublestranded breaks (Ihry et al., 2017). To counter the latter problem, we co-transfect a nonintegrating dominant negative p53 construct (Addgene #41856) along with donor DNA and TALENs when establishing a new line. We have seen major improvements in donor construct integration through this strategy, as measured by increased numbers of fluorescent cells both 24 hr after transfection and 1 week after FACS.
The construction of i3Neuron and i3LMN iPSC lines enables versatility in studying various aspects of neuronal cell biology through additional gene knockouts, knockdowns, or overexpression systems. The integrated, inducible, and isogenic transgene system allows the production of a more pure and mature population of neurons in less time than neurons differentiated by growth factors and small molecules alone. These cells are furthermore easily amenable tomultiple experimental applications, including microscopy, biochemistry, -omics, and electrophysiology (with addition of glia).
The estimated time to generate a stable hNGN2 or hNIL-expressing iPSC line is approximately 1 month (transfection, manual or FACS fluorophore enrichment, clonal isolation, genotyping, karyotyping). The doxycycline differentiation period is 3 days, and culture to mature neurons takes an additional 11 days (mature d14 neurons). These neurons can also be cultured longer, as we have observed good viability and morphology through 30 days after initial doxycycline induction.
This protocol is based on work that was partially carried out at the Gladstone Institutes (San Francisco) and implemented at the NIH. The work was financially supported by an NIH K08 award (M.E.W.), as well as by the NINDS intramural research program. We thank Kurt Fischbeck (NINDS/NIH) and Daniel Felkner (WiCell) for their helpful comments and suggestions.
Bain, G., & Gottlieb, D. I. (1998). Neural cells derived by in vitro differentiation of P19 and
embryonic stem cells. Perspectives on Developmental Neurobiology, 5, 175–178.
Beers, J., Gulbranson, D. R., George, N., Siniscalchi, L. I., Jones, J., Thomson, J. A., & Chen, G. (2012). Passaging and colony expansion of human pluripotent stem cells by enzyme-free dissociation in chemically defined culture conditions. Nature Protocols 7, 2029–2040. doi: 10.1038/nprot.2012.130.
Busskamp, V., Lewis, N. E., Guye, P., Ng, A. H., Shipman, S. L., Byrne, S. M., . . . Church, G. M. (2014). Rapid neurogenesis through transcriptional activation in human stem cells. Molecular Systems Biology, 10, 760–760. doi: 10.15252/msb.20145508.
Cao, S.-Y., Hu, Y., Chen, C., Yuan, F., Xu, M., Li, Q., . . . Liu, Y. (2017). Enhanced derivation of human pluripotent stem cell-derived cortical glutamatergic neurons by a small molecule. Scientific Reports, 7, 3282. doi: 10.1038/s41598-017-03519-w.
Carlson-Stevermer, J., Abdeen, A., Kohlenberg, L., Goedland, M., Molugu, K., Lou, M., & Saha, K. (2017). Assembly of CRISPR ribonucleoproteins with biotinylated oligonucleotides via an RNA aptamer for precise gene editing.
Nature Communications, 8, 1711. doi: 10.1038/s41467-017-01875-9.
Cerbini, T., Funahashi, R., Luo, Y., Liu, C., Park, K., Rao, M., . . . Zou, J. (2015). Transcription activator-like effector nuclease (TALEN)-mediated CLYBL targeting enables enhanced transgene expression and one-step generation of
dual reporter human induced pluripotent stem cell (iPSC) and neural stem cell (NSC) lines.
PloS One, 10, e0116032. doi: 10.1371/journal.pone.0116032.
Chambers, S. M., Fasano, C. A., Papapetrou, E. P., Tomishima, M., Sadelain, M., & Studer, L. (2009). Highly efficient neural conversion of human ES and iPS cells by dual inhibition of SMAD signaling. Nature Biotechnology, 27, 275–280. doi: 10.1038/nbt.1529.
Chen, G., Gulbranson, D. R., Hou, Z., Bolin, J. M., Ruotti, V., Probasco, M. D., . . . Thomson, J. A. (2011). Chemically defined conditions for human iPSC derivation and culture. Nature Methods, 8, 424–429. doi: 10.1038/nmeth.1593.
D’Alton, S., Altshuler, M., Cannon, A., Dickson, D. W., Petrucelli, L., & Lewis, J. (2014). Divergent
phenotypes in mutant TDP-43 transgenic mice highlight potential confounds in TDP-43 transgenic modeling. PloS One, 9, e86513. doi: 10.1371/journal.pone.0086513.
Gossen, M., & Bujard, H. (1992). Tight control of gene expression in mammalian cells by tetracycline-responsive promoters. Proceedings of the National Academy of Sciences of the U.S.A., 89, 5547–5551. doi: 10.1073/pnas.89.12.5547.
Goto, K., Imamura, K., Komatsu, K., Mitani, K., Aiba, K., Nakatsuji, N., . . . Inoue, H. (2017). Simple derivation of spinal motor neurons from ESCs/iPSCs using sendai virus vectors. Molecular Therapy—Methods&Clinical Development,
4, 115–125. doi: 10.1016/j.omtm.2016.12.007.
Hu, B.-Y., & Zhang, S.-C. (2009). Differentiation of spinal motor neurons from pluripotent human stem cells. Nature Protocols, 4, 1295–1304. doi: 10.1038/nprot.2009.127.
Ihry, R. J., Worringer, K. A., Salick, M. R., Frias, E., Ho, D., Theriault, K., Kommineni, S., . . . Kaykas, A. (2017). P53 toxicity is a hurdle to CRISPR/CAS9 screening and engineering in human pluripotent stem cells. bioRxiv, 168443.
doi: https://doi.org/10.1101/168443.
Karumbayaram, S., Novitch, B. G., Patterson, M., Umbach, J. A., Richter, L., Lindgren, A., . . . Lowry,W. E. (2009). Directed differentiation ofhuman-induced pluripotent stem cells generates active motor neurons. Stem Cells, 27, 806–811. doi: 10.1002/stem.31.
Kawai, H., Sango, K.,Mullin, K. A., & Proia, R. L. (1998). Embryonic stem cells with a disrupted GD3 synthase gene undergo neuronal differentiation in the absence of b-series gangliosides. Journal of Biological Chemistry, 273, 19634–19638. doi: 10.1074/jbc.273.31.19634.
Kent, L. (2009). Culture and maintenance of human embryonic stem cells. Journal of Visualized Experiments, pii, 1427. doi: 10.3791/1427.
Kikuchi, T., Morizane, A., Doi, D., Magotani, H., Onoe, H., Hayashi, T., . . . Takahashi, J. (2017). Human iPS cell-derived dopaminergic neurons function in a primate Parkinson’s disease model. Nature, 548, 592–596. doi: 10.1038/nature23664.
Kitazawa, M., Medeiros, R., & Laferla, F. M. (2012). Transgenic mouse models of Alzheimer disease: Developing a better model as a tool for therapeutic interventions. Current Pharmaceutical Design, 18, 1131–1147. doi:
10.2174/138161212799315786.
Kloskowska, E., Pham, T. M., Nilsson, T., Zhu, S., Oberg, J., Codita, A., . . . Benedikz, E. (2010). Cognitive impairment in the Tg6590 transgenic rat model of Alzheimer’s disease. Journal of Cellular and Molecular Medicine, 14, 1816–
1823. doi: 10.1111/j.1582-4934.2009.00809.x.
Kramer, M. F., & Coen, D. M. (2001). Enzymatic amplification of DNA by PCR: Standard procedures and optimization. Current Protocols in Cell Biology, 10, A.3C.1–A.3C.14. https://doi.org/10.1002/0471143030.cba03fs10.
Ludwig, T. E., Bergendahl, V., Levenstein, M. E., Yu, J., Probasco, M. D., & Thomson, J. A. (2006). Feeder-independent culture of human embryonic stem cells. Nature Methods, 3, 637–646. doi: 10.1038/nmeth902.
Major, T., Powers, A., & Tabar, V. (2017). Derivation of telencephalic oligodendrocyte progenitors from human pluripotent stem cells. Current Protocols in Stem Cell Biology, 39, 1H.10.1– 1H.10.23. doi: 10.1002/cpsc.17.
Matheson, N. J., Peden, A. A., & Lehner, P. J. (2014). Antibody-free magnetic cell sorting of genetically modified primary human CD4+ T cells by one-step streptavidin affinity purification. PloS One, 9, e111437. doi: 10.1371/journal.pone.0111437.
Mazzoni, E. O., Mahony, S., Closser, M., Morrison, C. A., Nedelec, S., Williams, D. J., . . . Wichterle, H. (2013). Synergistic binding of transcription factors to cell-specific enhancers programs motor neuron identity. Nature
Publishing Group, 16, 1219–1227. doi: 10.1038/nn.3467.
Merkle, F. T., Ghosh, S., Kamitaki, N., Mitchell, J., Avior, Y., Mello, C., . . . Eggan, K. (2017). Human pluripotent stem cells recurrently acquire and expand dominant negative P53 mutations. Nature, 545, 229–233. doi: 10.1038/nature22312.
Nuber, S., Harmuth, F., Kohl, Z., Adame, A., Trejo, M., Sch¨onig, K., . . . Riess, O. (2013). A progressive dopaminergic phenotype associated with neurotoxic conversion of α-synuclein in BAC-transgenic rats. Brain, 136, 412–432.
doi: 10.1093/brain/aws358.
Phelan, K., & May, K. M. (2015). Basic techniques in mammalian cell tissue culture. Current Protocols in Cell Biology, 00, 1.1.1–1.1.22. https://doi.org/10.1002/0471143030.cb0101s66.
Qi, Y., Zhang, X. J., Renier, N., Wu, Z., Atkin, T., Sun, Z., . . . Studer, L. (2017). Combined smallmolecule inhibition accelerates the derivation of functional cortical neurons from human pluripotent stem cells. Nature Biotechnology,
35, 154–163. doi: 10.1038/nbt.3777.
Shi, Y., Lin, S., Staats, K. A., Li, Y., Chang, W., Hung, S., . . . Ichida, J. K. (2017). Haploinsufficiency Leads to Neurodegeneration in C9ORF72ALS/FTD human induced motor neurons. Nature Medicine, 24(3), 313-325. doi:
Shimazaki, T., Arsenijevic, Y., Ryan, A. K., Rosenfeld, M. G., & Weiss, S. (1999). A role for the POU-III transcription factor Brn-4 in the regulation of striatal neuron precursor differentiation. EMBO Journal, 18, 444–456. doi: 10.1093/emboj/18.2.444.
Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K., . . . Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell, 131(5), 861–872. doi:
https://doi.org/10.1016/j.cell.2007.11.019.
Thomson, J. A., Itskovitz-Eldor, J., Shapiro, S. S., Waknitz, M. A., Swiergiel, J. J., Marshall, V. S., & Jones, J. M. (1998). Embryonic stem cell lines derived from human blastocysts. Science, 282(5391), 1145–1147. doi: https://doi.org/10.1126/science.282.5391.1145.
Voytas, D. (2001). Agarose gel electrophoresis. Current Protocols in Molecular Biology, 51, 2.5A.1–2.5A.9. doi: 10.1002/0471142727.mb0205as51.
Wang, C.,Ward, M. E., Chen, R., Liu, K., Tracy, T. E., Chen, X., . . . Gan, L. (2017). Scalable production of iPSC-derived human neurons to identify tau-lowering compounds by high-content screening. Stem Cell Reports, 9, 1221–1233.
doi: 10.1016/j.stemcr.2017.08.019.
Wernig, M., Zhao, J. P., Pruszak, J., Hedlund, E., Fu, D., Soldner, F., . . . Jaenisch, R. (2008). Neurons derived from reprogrammed fibroblasts functionally integrate into the fetal brain and improve symptoms of rats with Parkinson’s disease. Proceedings of the National Academy of Sciences of the United States of America, 105,
5856–5861. doi: 10.1073/pnas.0801677105.
Zeng, H., Guo, M., Martins-Taylor, K., Wang, X., Zhang, Z., Park, J. W., . . . Xu, R. H. (2010). Specification of region-specific neurons including forebrain glutamatergic neurons from human induced pluripotent stem cells. PloS One,
5, e11853. doi: 10.1371/journal.pone.0011853.
Zhang, Y., Pak, C., Han, Y., Ahlenius, H., Zhang, Z., Chanda, S., . . . S¨udhof, T. C. (2013). Rapid single-step induction of functional neurons from human pluripotent stem cells. Neuron, 78, 785–798. doi: 10.1016/j.neuron.2013.05.029.
Zhu, Y., Liu, Q., Zhou, Z., & Ikeda, Y. (2017). PDX1, Neurogenin-3, andMAFA: Critical transcription regulators for beta cell development and regeneration. Stem Cell Research & Therapy, 8, 240. doi: 10.1186/s13287-017-0694-z.