CALIFORNIA STATE UNIVERSITY, NORTHRIDGE
Growing Brains to Study Development: Optimization of Cerebral Organoid Culture from
Embryonic Stem Cells
A thesis submitted in partial fulfillment of the requirements
For the degree of Master of Science in Biology
By
Jessie Erin Buth
December 2015 The thesis of Jessie Erin Buth is approved:
______
Dr. Aida Metzenberg Date
______
Dr. Randy Cohen Date
______
Dr. Cindy Malone, Chair Date
California State University, Northridge
ii Table of Contents
Signature Page ii
List of Figures iii
Abstract vi
Chapter 1: Introduction 1
Chapter 2: Methods 15
Chapter 3: Results 23
Chapter 4: Discussion 50
References 55
Appendix: Supplemental Methods 59
List of Figures
Figure 1: Human organoid protocol (Cell line H9) and morphology at various time points. 23
Figure 2: Review of cortex development. 24
Figure 3: V-bottomed 96-well plates improve aggregate formation and generates a thick neuroepithelial layer on day 18 compared to
U-bottomed 96-well plates. 25
Figure 4: Plating at 9,000 cells per well on day 0 generates thickest continuous neuroepithelial layer compared to other cell numbers. 26
Figure 5: Addition of the BMP inhibitor (LDN193189) does not improve the efficiency of producing FOXG1+ cortical progenitors and inhibits formation of continuous N-cadherin+ apical membrane. 27
Figure 6: Fold change in gene expression of cortical markers increases as stem cell markers decrease in human cortical organoids over time by qPCR. 28
Figure 7: Human embryonic stem cells efficiently form cortical progenitors with 81.7% of total live cells per organoid positive for cortical marker FOXG1 at day 18. 29
Figure 8: Initial apical-basal polarity of human cortical organoids. 31
Figure 9: Organoids form cortical neurons with signs of laminar organization. 32
Figure 10: Basal-radial glial-like cells form in the subventricular zone of human cortical organoids and increase in number over time. 34
iii Figure 11: Basal radial glia and intermediate progenitor cells dividing in the subventricular zone increase in number over time. 35
Figure 12: Summary of current human organoid culture. 36
Figure 13: Optimized mouse organoid protocol (Cell line MM13). 37
Figure 14: Plating mouse embryonic stem cells at 5,000 cells per well on day 0 generates most consistent aggregates compared to other cell numbers and transfer to high oxygen produces more rosettes. 38
Figure 15: 3µM IWR1e (Wnt inhibitor) induced highest fold change in
Foxg1 expression at day 7 relative to undifferentiated mESCs by qPCR. 39
Figure 16: qPCR data shows mouse cortical organoids upregulate cortical genes by day 5 (Line MM13 + 3µM IWR1e). 40
Figure 17: Mouse embryonic stem cells efficiently form cortical progenitors with 92.9% of total live cells per organoid positive for cortical marker
Foxg1 at day 5. 41
Figure 18: 1:1 mixture of GMEM/KSR based media and N2 based media followed by N2 media supplemented with CDLC on day 7 “D” produces largest rosette structure positive for cortical markers. 42
Figure 19: 1:1 mixture of GMEM/KSR based media and N2 based media followed by N2 media supplemented with CDLC on day 7 “D” produces largest rosette structure with intermediate progenitors and cortical markers. 44
Figure 20: Summary of current mouse organoid culture. 45
iv Figure 21: Foxp1 expression overlaps with apical and basal radial glial cells and Foxp4 expression overlaps with intermediate progenitors in human cortical organoids over time. 46
Figure 22: Foxp1/Foxp4 expression partially overlaps and Foxp4 expression overlaps with apical radial glia in mouse cortical organoids over time. 47
v Abstract
Growing Brains to Study Development: Optimization of Cerebral Organoid Culture from
Embryonic Stem Cells
By
Jessie Erin Buth
Master of Science in Biology
The outermost portion of the brain, the cortex, is a six-layered structure of neurons that is responsible for higher cognitive functions such as thinking and speech.
This region is the most evolutionarily new part of the brain and is disproportionately large in humans than predicted by body size. What has caused the cortex to expand across evolutionary time? Little is known about this process due to the inaccessibility to human tissue. Much of what is currently known stems from static images of human fetal tissue or extrapolation from model organisms. Mice, and other model organisms, fail to recapitulate hallmark features of human brain development. There is a need for a better system to study brain development and its disorders. One alternative is to use stem cells.
Stem cells have the ability to form any cell type in the body, and in the lab can be
vi directed to form any cell type, or tissue, of interest. Recently, Kadoshima et al. (2013) and Lancaster et al. (2013) published protocols to create organ-like structures, called organoids, that model some aspects of brain development in vitro using stem cells. Stem cells are exposed to factors normally present during development and spontaneously form cortical tissue. This system has not been well characterized and it remains to be determined how reproducible these studies are with other cell lines. This study determined cortical organoids derived from H9 human embryonic stem cells and MM13 mouse embryonic stem cells can model some aspects of in vivo cortex development with some modifications of previously published protocols. Cortical organoids derived from human cells efficiently and reproducibly formed cortical tissue that exhibited some established features of cortex development. Human cortical organoids produced a thick layer of FOXG1+ cortical progenitors, initially showed the correct apical-basal (inside- out) polarity, and formed cortical neurons with signs of laminar organization. Mouse cortical organoids also efficiently formed FOXG1+ progenitors and produced cortical neurons, but with less laminar organization than in the human organoids. This study is the first to thoroughly characterize how well cortical organoids model in vivo development and in which areas the methods could be improved.
vii CHAPTER 1: INTRODUCTION
Humans have one of the largest brains of all animals relative to body mass
(Hofman 2014). The outermost region, the cortex, is a six-layered structure of neurons that is involved with higher cognitive functions such as thinking and fine motor control.
The cortex is the most evolutionarily new part of the brain and is disproportionately large in humans compared to other primates. How the human cortex expanded across evolutionary time remains difficult to study due to the inaccessibility to tissue. In vitro models of cortex development are a promising way to study human-specific features of cortical development that could not be recapitulated in model organisms. This chapter will review what is known about cortex development, and the following chapters will report on optimized protocols to model cortex development in 3D organ-like structures derived from human and mouse embryonic stem cells and discuss two genes hypothesized to be involved with cortical expansion over evolutionary time.
Early in development of the brain, regional cues alter gene expression in a concentration-dependent manner in order to persuade cells to adopt different fates
(Shimogori et al. 2004). These cues are called morphogens; the major classes of morphogens that set up early patterning in the brain include wingless-wnt (Wnt), bone- morphogenic proteins (BMP), fibroblast growth factor (FGF), and sonic hedgehog (Shh).
These proteins vary in concentration across the rostral-caudal and dorsal-ventral axes
(Shimogori et al. 2004). The cortex arises from the telencephalon in an area with a low concentration of these signals and begins as a population of multipotent neural stem cells
(Shimogori et al. 2004).
1 The cortex extends from a layer of neural stem cells that line the lateral ventricles.
First, neural stem cells expand in cell number by proliferative divisions, and then later differentiate into the six-layered structure of neurons seen in adults (Rakic 1995; Jessell and Sanes 2000). In 1995, Rakic first proposed the radial-unit hypothesis, which has become the basis for our understanding of cortex expansion. The radial unit hypothesis postulates that changes in regulatory genes that control how long the proliferative and differentiative phases are control the surface area and thickness of the cortex, with a larger surface area having more convolutions and folds as in the human cortex (Rakic
1995). The radial unit is defined as a group of neurons that arise from one founder cell, to form a vertical column of neurons (Rakic 1995). The number of founder cells determines the surface area of the cortex, and the number of neurons within a column determines the thickness (Rakic 1995). Rakic provides a simplified example to represent how timing could influence cortex size, by comparing the timing of the proliferative and differentiative phases of human and macaque. It is argued that during the proliferative phase, there is an exponential increase in cells, because each round of cell division doubles the pool of founder cells; thus, even a few days increase in the proliferative phase could allow the human cortex to have 23 or 24 more founder cells than the macaque
(Rakic 1995). During the second phase, neurons are formed in a linear fashion because each founder cell can add one neuron to the column for each round of cell division (Rakic
1995). In the macaque, the second phase is 100 days, while in humans it is 120 (Rakic
1995). During this phase, a 20-day increase would only increase the thickness of a column by ten cells, a 10% increase in thickness (Rakic 1995). This hypothesis explains how regulatory genes controlling the timing of different phases can result in cortical
2 expansion. Recent findings have shown that the development of the cortex is a much more complicated phenomena than Rakic originally described. The founder cells actually produce multiple types of progenitor cells that also give rise to neurons and other cortical cells, creating a cone of division rather than a columnar unit (Betizeau et al. 2013).
Across species, certain progenitor cells are more abundant than others (Hansen et al.
2010). In humans, there are three types of multipotent progenitors (apical radial glia, intermediate progenitors, and basal radial glia) that give rise to neurons in the cortex
(Jessell and Sanes 2000; Fietz and Huttner 2011). It has also been suggested that there are four distinct morphological types (morpho types) of basal radial glia cells, all of which can self-renew, and form neurons (Betizeau et al. 2013). As development proceeds, apical radial glia form neurons and other progenitors cells, which in turn form more neurons
(Jessell and Sanes 2000; Fietz and Huttner 2011). As the neurons mature, they migrate out of the ventricular zone (where apical radial glia divide) to the cortical plate using apical radial glial cell fibers attached to the basal surface as a scaffold (Jessell and Sanes
2000; Fietz and Huttner 2011). When the neurons reach the cortical plate, a six-layered structure is formed in an inside-out matter, where early-born neurons reside in deeper layers, and early-born neurons migrate further to more superficial upper layers (Jessell and Sanes 2000). Layer I is the outermost ‘basal’ side, and layer VI is the innermost
‘apical’ side (Jessell and Sanes 2000) (Figure 2). Before six distinct layers are established, the cortex is organized into different zones containing progenitors and migrating neurons; the ventricular zone, subventricular zone, cortical plate, and marginal zone.
3 In the ventricular zone, neural stem cells reside in a neuroepithelial layer, forming a niche, where there is a balance between symmetric (self-renewing) cell divisions to expand the epithelial-progenitor layer, and asymmetric (differentiative) cell divisions to form neurons and other progenitor cells (Jessell and Sanes 2000). Once the niche is expanded, there is a switch from neural stem cells to apical radial glial cells. Apical radial glia cells are bipolar in morphology, with a process extending to the basal surface and a cell body maintained near the apical membrane in the ventricular zone (Jessell and Sanes
2000). Apical processes are connected to the ventricular lumen by adherens junctions mediated by N-cadherin, a calcium-dependent adhesion molecule (Fietz and Huttner
2011). Neural stem cells and radial glia cells can be characterized by expression of paired-box gene 6 (PAX6) and sex-determining region Y-box 2 (SOX2) (see Table 1 for summary of molecular markers). Unlike neural stem cells, radial glial cells are also positive for astrocytic markers such as glial fibrillary acidic protein (GFAP) and the astrocytic glutamate transporter (GLAST), which is why they were first identified as glial cells, providing a scaffold for neurons to migrate along to their prospective layers
(Malatesta et al. 2008). Lineage-tracing experiments revealed these cells not only provide a scaffold for neuronal migration, but are also progenitor cells that self-renew, give rise to other neural progenitor subtypes, neurons, and only later, to astrocytes across the central nervous system (Noctor et al. 2001; Malatesta et al. 2008; Fietz an Huttner 2011; Florio and Huttner 2014). When dividing, apical radial glial cells undergo interkinetic nuclear migration. Here, the cell body moves away from the apical membrane during G1, toward the apical membrane during S/G2, and then divide at the apical surface creating a pseudostratified appearance (Fietz and Huttner 2011).
4 Two more progenitor subtypes contribute to the neuronal population in the cortex. These cells are located just outside the ventricular zone (VZ) in the subventricular zone (SVZ).
Table 1: Cell Specific Markers for Neural Progenitor Subtypes. Cell Type Abbreviation Location Molecular Markers Apical Radial aRG Ventricular Zone PAX6+ Glia (VZ) SOX2+ Phosopho-Vimentin+ (when in M phase) Intermediate IP Inner TBR2+ Progenitor Subventricular Zone (iSVZ) Basal Radial bRG Outer PAX6+ Glia Subventricular SOX2+ Zone (oSVZ) Phosopho-Vimentin+ (when in M-phase) HOPX+ TNC+ ITGB5+
The SVZ contains intermediate progenitors (IPs) in the inner-subventricular zone, and basal radial glial cells (bRGs) in the outer subventricular zone (Betizeau et al.
2013). Intermediate progenitor cells are nonpolar. These cells delaminate from the adherens junction belt and lose their apical and basal processes (Fietz and Huttner 2011).
Intermediate progenitors are labeled by T-box brain 2 (TBR2), proliferate for multiple rounds of divisions and then form neurons (Lui et al. 2011; Fietz and Huttner 2011;
Florio and Huttner 2014). Basal radial glial cells (bRGs), also known as outer radial glia, are monopolar radial glial cells that maintain one process attached to the basal membrane
(Hansen et al. 2010). These cells delaminate from the adherens junction belt and lose their apical process, although it has been shown that some morpho-types do partially retain this process (Fietz and Huttner 2011; Betizeau et al. 2013). Basal radial glial cells migrate to the outer subventricular zone by shortening their basal process and move the cell body out of the ventricular zone to the outer subventricular zone, a process called mitotic somal translocation (Pollen et al. 2015). Basal radial glial cells are characterized
5 by expression of apical radial glial cell markers (PAX6, SOX2, and actively mitosing cells with Phosphorylated Vimentin) as well as recently discovered basal radial glial specific markers such as HOPX, TNC, and ITGB5 (Pollen et al. 2015). Basal radial glial cells have the ability to self-renew, and form other morpho-types of bRGs, IPs, and neurons (Hansen et al. 2010; Betizeau et al. 2013). Abundance of bRGs is specific to animals with a gyrencephalic (folded) cortex, such as humans and ferrets, while few are seen in animals with lissencephalic (smooth) cortices like mice (Hansen et al. 2010; Fietz and Huttner 2011). Differing amounts of a third type of progenitor cell reveal a mode for differential neuronal output across species, specifically expansion in humans (Florio and
Huttner 2014). More superficial to the subventricular zone is the cortical plate; this is where neurons migrate to and form layers II through V1 in an inside-out manner as described above. The most superficial layer (layer I) is the marginal zone, which contains
Cajal-Retzius cells.
Most Cajal-Retzius cells migrate to layer I from nearby brain regions such as the cortical hem, but some are formed directly from apical radial glial cells before the formation of the cortical plate (Fietz and Huttner 2011). Cajal-Retzius cells release the extracellular matrix glycoprotein reelin which aids in radial neuronal migration (Bar et al.
2000). Evidence indicates reelin may have played a critical role in mammalian cortex development. The three-layered cortex of lizards and turtles develops in an outside-in manner, opposite to that seen in mammals, suggesting that the inside-out pattern was acquired during mammalian evolution (Bar et al. 2000; Nadarajah & Parnavelas 2002).
Interestingly, reelin production is minimal in lizards and turtles, while it is greatly enhanced in mammals (Bar et al. 2000; Nadarajah & Parnavelas 2002). Reelin appears to
6 be crucial for the formation of correct inside-out pattern in mammals. If reelin is knocked out in mice, cortical lamination is highly disrupted and separation of distinct layers is not seen (Dekimoto et al. 2010; Meyer 2010).
Model organisms such as mice and primates have provided much of the information known about cortex development, but fail to recapitulate aspects of human development. Ascertainment of the mechanisms underlying human development has remained elusive for ethical reasons. What is known about developmental processes stems from studies on model organisms and static images of human fetal tissue. Recently,
3D cell culture techniques have emerged that can model some aspects of whole organ development in vitro using stem cells (Lancaster and Knoblich 2014a). 3D cell culture techniques create organ-like structures, called organoids, that can easily model the complexity of 3D structural components of whole organs compared to previous flat culture techniques (Lancaster and Knoblich 2014a). By mimicking in vivo development cues, stem cells can now be directed to form any cell type of interest in vitro. Pluripotent stem cells or organ progenitor cells can be directed to form multiple cell types that self- organize to form complex structures similar to whole organs (Lancaster and Knoblich
2014a). To date, organoids have been used to model intestine, kidney, retina, and brain development (Lancaster and Knoblich 2014a). In the past few years, multiple groups have generated brain organoids from pluripotent stem cells (Mariani et al. 2012;
Lancaster et al. 2013: Kadoshima et al. 2013; Lancaster and Knoblich 2014b; Paşca et al.
2015). These groups have focused on modeling the most evolutionarily new region of the brain, the cortex. Currently there are multiple names given to the process of making cortical tissue from pluripotent stem cells: serum-free floating cultures of embryoid body-
7 like aggregates with quick reaggregation (SFEBq) (Kadoshima et al. 2013), human cortical spheroids (hSCs) (Paşca et al. 2015), and cerebral organoids (Lancaster et al.
2013). Herein the term cerebral organoid will be used.
One of the first studies on cortical organoids was from Watanabe and colleagues in 2005. Watanabe et al. (2005) showed that mouse embryonic stem cells could adopt a neural lineage in aggregates of cells in suspension culture, a method they called serum- free floating culture of embryoid body-like aggregates (SFEB). The SFEB culture could form telencephalic tissue, and be selectively altered to form different dorsal or ventral fates by addition of extrinsic factors, such as wingless-int (Wnt) or sonic hedgehog (Shh) proteins. The ability to manipulate cell fate to model different regions of the brain made this result very promising, but initially this protocol did not hold for human embryonic stem cells because these cells normally died upon single-cell dissociation. Later,
Watanabe et al. (2007) found that Rho-associated kinase inhibitor (ROCKi) could protect human embryonic stem cells from dissociation-induced apoptosis; thus, the SFEB protocol could now be applied to human cells as well. This protocol was later improved to quickly form aggregates of the same size (SFEBq), which more efficiently produced cortical progenitors and reduced variability between aggregates (Eiraku et al. 2008). In
2008, two groups showed that cortical organoids recapitulated hallmarks of early corticogenesis and that cortical neurons generated from organoids were functional, transplantable, and capable of forming connections in vivo (Eiraku et al. 2008; Gaspard et al. 2008). It was also shown that organoids derived from mouse cells did not form any human-specific features seen in human cerebral organoids, suggesting an intrinsic mechanism within the cells (Eiraku et al. 2008). In 2012, Mariani et al. showed that
8 induced pluripotent cells could also model telencephalic development using a combination of 3D and adherent cultures. This study established the potential for disease modeling using the organoid system; cells from patients with known disease mutations could be transformed to induced pluripotent stem cells and used to compare normal and abnormal brain development in vitro. This group later showed induced pluripotent cells could be used to model autism spectrum disorder and found that organoids derived from autistic patients showed an accelerated cell cycle and an overproduction of GABAergic inhibitory neurons (Mariani et al. 2015). Lancaster et al. (2013) also used induced pluripotent stem cells from a microcephaly patient and, using the 3D organoid technique, found organoids with a mutation in CDK5RAP2 prematurely differentiated to neurons and depleted the progenitor pool, leading to smaller sized organoids than controls.
Each of these method use different approaches to dissociate cells and cultures in different media. In each case, pluripotent stem cells are subjected to minimal media without morphogens or with morphogen inhibitors to adopt a telencephalic fate.
Additional factors can then be added to promote growth and neuronal maturation in a differentiation media. The cells can then self-organize to form a layered structure. To achieve this, embryonic stem cells are subjected to minimal media with inhibitors of morphogens of the TGFβ, Wnt, or BMP pathways (Mariani et al. 2012; Kadoshima et al.
2013; Paşca et al. 2015) or without inhibitors (Lancaster and Knoblich 2014b). Specific inhibitors can then be added to control for regional identity of the culture by mimicking morphogen gradients seen in vivo. For example, cortex development is affected by combinations of low TGFβ, Wnt, and BMP. Inhibitors are not needed, as neural is the default state in minimal media, and cortex can spontaneously form in culture without
9 added morphogens (Muñoz-Sanjuán and Brivanlou 2002). Addition of TGFβ and Wnt inhibitors described in Kadoshima et al. (2013) induce the culture to become more homogeneously cortical cells, which can reduce the variability of the culture.
Alternatively, absence of inhibitors allows other brain structures such as choroid plexus to develop, and is argued to be an advantage since choroid plexus cells have been shown to secrete factors that influence further development of cortex-like structures in other parts of the organoid (Lancaster et al. 2013; Lancaster et al. 2014). On the day differentiation begins, day zero, cells are lifted with enzyme, usually dispase or accutase, and either dissociated to single cells with Trypsin (Lancaster et al. 2013; Kadoshima et al.
2013) or allowed to spontaneously form embryoid body-like spherical structures (Paşca et al. 2015). Dissociating to single cells allows one to control the size of the aggregates, which allows for optimization of cell number for individual cell lines. Spontaneous formation of embryoid body-like aggregates in Paşca et al. (2015) is argued to be simple, scalable, and reproducible.
While all of these methods result in mono-telencephalic tissue and mono developmental timing similar to in vivo corticogenesis, there has yet to be developed a standardized, unbiased method to quantify exactly which area is being modeled and its timing. Recently a bioinformatics program called CoNTExT has emerged that can be used to comprehensively assess the status of an in vitro culture (Stein et al. 2014).
CoNTExT is a new tool that has the ability to assign a developmental time and location to in vitro cell culture assays based on transcriptomic analysis (Stein et al. 2014). It can assess how well culture models adhere to in vivo development, the maturity of cells after differentiation, and what anatomical location is being modeled (Stein et al 2014). Paşca et
10 al. (2015) analysis uses CoNTExT to assess the developmental maturity and regional identity of cortical organoids. Transcriptome analysis of the Paşca et al. (2015) cortical organoid culture revealed overlap with late mid-fetal periods, 19-24 gestation weeks, in contrast to flat culture techniques, which overlapped with earlier stages of development.
CoNTExT analysis is currently in progress for optimized protocols reported later in this chapter. When considering the limitations of each technique, it is important to realize that the regional identity of many culture techniques has been incompletely assessed. Use of the CoNTExT program may resolve this issue. Variability between these results could also be attributed to differences between cell lines tested. Thus far, few cell lines have been used for creating cortical organoids; it remains unclear if all cell lines can recapitulate previous in vitro results. It is also notable that addition of neurotrophic and other growth factors may promote the culture to favor one cell type versus another. Each protocol has its advantages and limitations. It remains unclear what parameters will make the best model of cortical development and disease. Once these parameters are established, the cortical organoid system can provide a platform to ask different developmental questions that have been previously difficult to study in humans.
The cerebral organoid system has the potential to provide a model system of brain development that can be altered to study different aspects of development, disease, and evolution. This study focuses on cortical development, and investigates the influence of two genes proposed to be involved in cortical expansion, FOXP1 and FOXP4.
Manipulation of these genes in future studies could lead to a better understanding of the function of these genes in relation to cortical expansion. FOXP1 and FOXP4 are members of the forkhead box (FOX) family of genes.
11 The FOX family of genes consists 40 members in mammals, divided into 19 subfamilies, FOXA to FOXS, that share a highly conserved ~100 base pair DNA-binding domain called the forkhead or winged-helix domain (Hannenhalli and Kaestner 2009).
This family serves a wide array of functions from developmental processes to insulin signaling (Hannenhalli and Kaestner 2009). The FOXP subfamily consists of four members; FOXP1-P4 (Hannenhalli and Kaestner 2009). FOXP1/2/4 are thought to be involved with central nervous system development (Bacon and Rappold 2012).
FOXP1/2/4 form heterodimers and homodimers through their zinc finger and leucine zipper domains. As dimers they bind DNA and regulate transcription through transcriptional repression (Li et al. 2004; Bacon and Rappold 2012).
FOXP1 consists of 34 exons that are alternatively spliced to many different isoforms, one of which is specific to embryonic stem cell maintenance while others are more predominant in differentiated cell types (Gabut et al. 2011). FOXP4, a close homologue to FOXP1, has 18 exons that are also alternatively spliced, and is expressed in the heart, brain, lung, liver, kidney, and testis (Teufel et al. 2003). Foxp1/2/4 share a partially overlapping, but distinct, expression pattern in the cerebral cortex (Takahashi et al. 2008). Mutations in FOXP1 in humans have been implicated with global development delay, autism, speech delay, intellectual disability, and deficits in motor coordination
(Palumbo et al. 2013; Le Fevre et al. 2013). FOXP1/4 and FOXP1/2/4 have been shown to regulate epithelial cell fate in lung development and endochondral ossification (bone formation) through a suppressor complex (Li et al. 2012; Zhao et al. 2015). In lung development, FOXP1/4 act to repress goblet cell fate, knockdown of FOXP1/4 results in premature differentiation and loss of epithelial regeneration (Li et al. 2012). During bone
12 formation, overexpression of FOXP1/2/4 led to loss of bone formation, while loss of function resulted in premature differentiation (Zhao et al. 2015). Rousso et al. (2012) has shown Foxp1/2/4 play a role in neural delamination and migration in the developing spinal cord. N-cadherin maintains adherens junctions between neural progenitor cells, keeping progenitor cells docked in the neuroepithelium, and is down regulated when cells differentiate and migrate. Foxp1/2/4 repress N-cadherin, which leads to breakdown of the adherens junctions between cells in the ventricular zone and allows neural migration
(Rousso 2012). Knockdown of Foxp2/4 by shRNA in the spinal cord prevents cells from detaching from the neuroepithelium, but does not halt neuronal differentiation (Rousso
2012). Misexpression of Foxp2/4 prompts disassembly of adherens junctions leading to production of motor neurons and other neuronal cells (Rousso 2012). Together, these results suggest FOXP1/2/4 act as a suppressor complex of various genes in the development of multiple tissues, by regulating cell fate.
FOXPs may act by a similar mechanism on cortical progenitors in the brain. In human fetal tissue, FOXP1 is expressed in apical radial glia in the ventricular zone and basal radial glia in the outer subventricular zone, while FOXP4 is expressed in intermediate progenitors in the inner subventricular zone (Pearson and Novitch, unpublished data). Unpublished data from the Novitch Lab suggest that Foxp1 mutant mice show a reduced number of basal radial glia and premature production of neurons as wells as intermediate progenitors. Foxp4 mutant mice show increased apical radial glia and decreased generation intermediate progenitors (Pearson and Novitch, unpublished data). These studies suggest that FOXP1 and FOXP4 may play roles in production and maintenance of cortical progenitor subtypes, with FOXP1 promoting apical radial glia
13 maintenance and formation of basal radial glia and FOXP4 promoting intermediate progenitors production and neuron formation (Pearson and Novitch, unpublished data;
Rousso et al. 2012). Organoid culture can be used to assess the role of FOXP1 and
FOXP4 in the production and maintenance of cortical progenitor subtypes. Cell lines carrying deletions in the functional regions of these genes could help elucidate their functions. To assess the validity of the organoid culture to model in vivo development, it was investigated whether expression of these genes reproduced expression patterns found in fetal human and mouse tissue (Pearson and Novitch, unpublished data).
The following chapters will examine in vitro models of cortex development, provide an optimized protocol to model cortex development using human and mouse embryonic stem cells (addressed in Q1, Q2, and H1), and suggest potential applications for this system (addressed in Q3 and H1).
Research Questions
Q1. Can other stem cell lines reproduce previous cortical organoid differentiation results?
Q2. How well do cortical organoids model in vivo development? In which areas is improvement needed?
Q3. Can organoids reproduce expression patterns of genes hypothesized to be involved in cortical expansion (FOXP1 and FOXP4) seen in human and mouse fetal tissue?
Hypotheses
H1. Cortical organoids derived from human and mouse embryonic stem cells can recapitulate established features of cortex development.
14 CHAPTER 2: METHODS
Human Cortical Organoid Culture
This portion of the chapter focuses on optimization of human organoid culture methods. Modified human organoid culture protocols from Kadoshima et al. (2013), and
Lancaster and Knoblich (2014b) and the experiments leading to this protocol are outlined below.
Human embryonic stem cell (hESC) maintenance:
Human embryonic stem cell line H9 were obtained from the UCLA stem cell core. H9 was cultured in DMEM/F12 supplemented with 20% KSR (Life Technologies
Lot Number 1670543), 0.1mM NEAA (Life Technologies), 2mM Glutamax (Life
Technologies), 0.1mM 2-Mercaptoethanol (Life Technologies), and 10ng per mL bFGF
(Life Technologies). H9 hESCs were grown to 70-80% confluence and then split 1:3 or
1:4 on 6-well plates coated with gelatin and a feeder layer of irradiated mouse embryonic fibroblasts. Cells were passaged manually with a STEMPRO EZPassage tool (Life
Technologies).
Summary of optimized hESC differentiation protocol:
hESC differentiation protocols followed were modified from Kadoshima et al.
(2013) and Lancaster et al. (2014). Prior to dissociation, cells were pretreated for 1 hour with 20uM Y-27632 (Stemgent). Dissociation of hESCs, defined as day 0, was done with
500µL 5U/mL dispase (Stem Cell Technologies) per well on a 6-well plate to lift the stem cell colonies. Next, 0.05% Trypsin (Life Technologies) with 0.05 mg/mL DNase I
(Worthington) and 20uM Y-27632 (Stemgent) was used to dissociate the colonies to single cells. Trypsin was partially inactivated with Glycine Max (Sigma), and cell
15 colonies were dissociated to single cell suspension with a P1000. Single cells were then strained on a 100µM cell strainer (Fisher Scientific) to remove clumps. Then the cells were centrifuged at 200g for 5 minutes. Cells were removed and plated at
9000cells/100µL per well on low-attachment V-bottom 96-well plate (Sumitomo
Bakelite) and then centrifuged again at 304g for 2 minutes for quick reaggregation. Cells were plated in cortical differentiation media containing GMEM (Life Technologies) supplemented with 20% KSR (Life Technologies), 0.1mM MEM-NEAA (Life
Technologies), 1mM Sodium Pyruvate (Life Technologies), 0.1mM 2-mercaptoethanol
(Life Technologies), 50units/mL Penicillin and 50 µg/mL Streptomycin (Life
Technologies), 20µM ROCKi (Y-27632) (Stemgent), 3µM IWR1e (IWR1endo)
(Calbiochem), and 5µM TGFβi (SB431542) (Stemgent) and kept at 37°C/20% O2 in a
HERAcell 150i CO2 Incubator (Thermo Scientific).
The resulting organoids were fed every 3 days. On day 3 100µL of GMEM/KSR based media was added per well. On day 6 125µL of media was removed, and 80µL fresh media without ROCKi (Y-27632) (Stemgent) was added. On days 9,12, and 15 75uL of media was removed, and 80uL of fresh GMEM/KSR based media was added per well.
On day 18, aggregates were moved to 10cm petri dishes containing 30-40 organoids and kept at 37°C/40% O2. Organoid media was changed to cortical maturation media containing DMEM/F12 (Life Technologies) supplemented with 1:100 N2 (Life
Technologies), 1:100 Chemically Defined Lipid Concentrate (Life Technologies), 1:500
Primocin (Life Technologies), and 50units/mL Penicillin and 50 µg/mL Streptomycin
(Life Technologies) and 0.2% Methylcellulose (Sigma). Media was changed every 3 days, all media was removed and 10mL fresh DMEM/N2 based media was added. At day
16 35, 1% growth factor reduced matrigel (Corning), 5 µg/mL Heparin (Sigma), and 1:100
B27 without vitamin A were added to the media. At day 56, aggregates were moved to
Lumox dishes (Sarstedt) to allow better oxygen absorption. Organoids were fed every 3 days, all media was removed and 5mL fresh DMEM/N2/B27 based media was added for each lumox dish. At days 35, 55, 75, 84, and 90 organoids are cut in half with a dissection scissors to allow better oxygen absorption.
Optimization Experiments:
A series of optimization experiments were done to achieve the procedure described above. First, round-bottom versus v-bottomed 96 well plates were investigated to see which formed better aggregates. H9 embryonic stem cells were dissociated to single cells as previously described above and plated at 9,000 cells per well in low- attachment U-bottomed 96-well plates and V-bottomed 96-well plates. Cell death, aggregation, and growth were assessed after 18 days in culture by morphology and immunohistochemistry. Next, the optimal cell numbers (3000/6000/9000/12000 cells per well on day 0) were investigated. H9 embryonic stem cells were plated on V-bottomed
96-well plates at varying concentrations (3,000, 6,000, 9,000, and 12,000 cells per well).
Organoids were cultured for 18 days and then growth and aggregation were evaluated by morphology and immunohistochemistry. Next, it was tested whether addition of BMP inhibitor (LDN193189) (Stemgent) would enhance formation of cortical tissue by increased Foxg1 expression; media was supplemented with 100nM LDN193189 for days
0-6, days 0-9, days 0-12, days 0-15, and days 0-18. Organoids were collected after 18 days in culture and cortical identity was characterized by immunohistochemistry, looking
17 for expression of cortical markers FOXG1, LHX2, and apical membrane marker N- cadherin. Organoids were then cultured up to 100 days and characterized at various time points to assess how well in vivo development was recapitulated, and in which areas improvement would be needed. Characterization of human cortical organoids at various time-points was done using fixation, cryosectioning, immunohistochemistry, cell counting with NIH image processing program ImageJ, and qPCR techniques as described in Appendix 1.
Outline of Human Organoid Culture:
Day 0 Day 6 Day 18 Day 35 Day 56-100 Cortical Differentiation Media Cortical Maturation Media/40% Cortical Maturation in V-bottom 96-well plate O2 in petri dish Media /40% O2 / Lumox Dishes + 20µM ROCKi + 1% Matrigel (growth factor (Y-27632) reduced)
+ 3µM IWR1e (IWR1endo) + 5 µg/mL Heparin + 5µM TGFβi (SB431542) +B27 without vitamin A
Mouse Cortical Organoid Culture
This portion of the chapter focuses on optimization of mouse organoid culture methods. Modified mouse and organoid culture protocols from Eiraku et al. (2008), Nasu et al. (2012), Kadoshima et al. (2013), and the experiments leading to this protocol are outlined below.
Mouse Embryonic Stem Cell (mESC) Maintenance:
Optimization of mouse organoid culture was done using the feeder-dependent mouse embryonic stem cell (mESC) line MM13. MM13 mESCs were maintained on a feeder layer of mouse embryonic fibroblasts (MEFs) irradiated in Novitch lab or gelatin-
18 coated dishes before differentiation. MM13 mESCs were cultured in DMEM high glucose (Life Technologies) supplemented with 15% ES-FBS (Life Technologies Lot
Number 1119159), 2mM Glutamax (Life Technologies), 50units/mL Penicillin and 50
µg/mL Streptomycin (Life Technologies), 0.1mM MEM-NEAA (Life Technologies), 1x
Nucleosides (Millipore), 0.1mM 2-Mercaptoethanol (Life Technologies), and 1000U/mL
LIF (Millipore). MM13 mESCs were passaged every 2-3 days with TrypLE Express
(Life Technologies) at 1:6-1:10.
Mouse Embryonic Fibroblasts (MEFs):
Mouse embryonic fibroblasts were grown and irradiated in Novitch Lab. MEFs were cultured in DMEM high glucose (Life Technologies) supplemented with 10% FBS
(Hyclone AYE161472), 2mM glutamax (Life Technologies), and 50units/mL Penicillin
& 50 µg/mL Streptomycin (Life Technologies).
Summary of mESC Differentiation:
For mouse organoid culture, modified protocols from Eiraku et al. (2008) and
Nasu et al. (2012) were used. On day 0, MM13 mES cells were dissociated to single cells in TrypLE Express (Life Technologies) and plated 5000 cells per well on U-bottomed low-attachment 96-well plates (Corning). 1mL TrypLE (Life Technologies) was added per well for 4 minutes. Colonies were broken down to a single cell suspension with a
P1000 and then centrifuged at 200g for 5 minutes. Cells were then plated at 5,000 cells per well in low attachment U-bottomed 96-well plates, and the plate was centrifuged at
304g for 2 minutes to allow quick reaggregation. Cells were plated in GMEM (Life
Technologies) supplemented with 10% KSR (Life Technologies), 2mM Glutamax (Life
Technologies), 1mM Pyruvate (Life Technologies), 0.1mM MEM-NEAA (Life
19 Technologies), 0.1mM 2-Mercaptoethanol (Life Technologies), and 3µM IWR1e
(Calbiochem). On day 1, 100µL of GMEM/KSR media was added per well. On day 3,
125µL is removed and 80µL fresh GMEM/KSR based media was added. On day 5, organoids were moved to 6-well low attachment plates (Corning) (8-12 organoids per well) in DMEM/F12 (Life Technologies) supplemented with 1:100 N2 (Life
Technologies), 2mM Glutamax (Life Technologies), 1x Chemically defined lipid concentrate (Life Technologies), 0.1mM MEM-NEAA (Life Technologies), 1mM
Sodium pyruvate (Life Technologies) and 0.2% Methylcellulose (Sigma). Organoids were fed every 2 days. All media was removed and 3mL of DMEM/N2 based media was added per well. Organoids were kept at 37°C/20% O2 in a HERAcell 150i CO2 Incubator
(Thermo Scientific) for day 0-5 and then transferred to 37°C/40% O2 day 5-12.
Optimization Experiments:
A series of optimization experiments were done to achieve the procedure above.
First, the optimal cell numbers (3000-6000 cells per well on day 0) were investigated.
MM13 embryonic stem cells were dissociated to single cells with TrypLE (Life
Technologies) and then plated in U-bottomed 96-well plates at varying concentrations
(2,000, 3,000, 4,000, 5,000 and 6,000 cells per well). After 12 days in culture, aggregation and growth were assessed by morphology and immunohistochemistry. Then it was investigated whether 40% O2 versus the standard 20% O2 would enhance organoid growth as described for human cortical organoids in Kadoshima et al. (2013). MM13 embryonic stem cells were plated on U-bottomed 96-well plates. On day 5, organoids were switched to N2 based media and either kept at 20% O2 or 40% O2. On day 12, organoids were collected to fix and evaluate by immunohistochemistry. Next, the
20 concentrations of Wnt inhibitor (IWR1e) (Calbiochem) and TGFβ inhibitor (SB431542)
(Stemgent) to induce Foxg1 expression were determined; 0-10µM IWR1e alone, 0-10µM
SB431542 alone, and varying combinations were tested. MM13 embryonic stem cells were plated at 5,000 cells per well on U-bottomed 96 well plates in media with varying concentrations of IWR1e and SB432542. Concentrations tested included 0µM, 0.5µM,
1µM, 2µM, 3µM, 5µM, and 10µM IWRe1 alone, 1µM, 3µM, 5µM, and 10µM SB431542, and in combination 1µM IWR1e with 1/3/5/10µM SB432542, 3µM IWRe1 with
1/3/5/10µM SB432541, 5µM IWR1e with 1/3/5/10µM SB432541, and 10µM IWR1e with 1/3/5/10µM SB431542. Inhibitors were added for the first 5 days in culture. On day
5, organoids were switched to DMEM-F12/N2 based media. On day 7, organoids were collected for quantitative polymerase chain reaction (qPCR) analysis for the genes Foxg1,
Lhx2, and Pax6. Next, the optimal timing to switch to N2 media was determined. MM13 cortical organoids were cultured to day 5 as previously described, and then switched to
N2 based media under four conditions. (1) The control, regular N2 based media as described in Eiraku et al. (2008), (2) N2 based media supplemented with chemically defined lipid concentrate, (3) kept in DMEM/KSR based media until day 7, or (4) a gradual switch to N2 based media where day 5-7 is a 1:1 mixture of DMEM/KSR media and N2 media followed by N2 media alone after day 7. The size of rosettes and formation of cortical neurons were evaluated by fixation of immunohistochemistry at various time points. Characterization of mouse cortical organoids at various time-points was done by fixation, cryosectioning, immunohistochemistry, cell counting with NIH image processing program ImageJ, and quantitative polymerase chain reaction (qPCR) analysis as described in Appendix 1.
21 Mouse organoid differentiation protocol:
Day 0 Day 5 Day 12 GMEM/10% KSR media (20% DMEM-F12/N2 media (40% O2) O2) U96w low attachment plate 6-well low attachment plate
Assessment of FOXP1 and FOXP4 expression in cortical organoids
Optimized cerebral organoids were derived from human and mouse embryonic stem cells and processed as previously described above and in Appendix 1.
Immunohistochemistry (Appendix 1) was done at various time points to characterize
FOXP1 and FOXP4 expression pattern and compared to previous unpublished data from
Novitch Lab on fetal human and mouse tissue. Antibodies for immunohistochemistry are listed in Appendix 1.
22 CHAPTER 3: RESULTS
Establishment of reliable cortical organoid culture from human embryonic stem cells.
Organoid culture from stem cells is a promising new approach to model development in vitro. It is predicted that organoids derived from stem cells can model established features of in vivo cortical development. An issue in the field has been to create a reproducible protocol that has the potential to work with multiple cell lines, while maintaining established features of in vivo development. This study found cortical organoids derived from a human embryonic stem cell line commonly used for neural differentiations (line H9) could model some aspects of cortical development. An optimized protocol, shown in Figure 1, was established that is reproducible, consistent, and efficient at producing cortex progenitors positive for the forebrain marker FOXG1.
Figure 1. Human organoid protocol (Cell Line H9) and morphology at various time points. On day zero cells are dissociated to single cells and plated 9,000 cells per well in low-attachment V-bottomed 96- well plates.
23
To assess how well the optimized protocol for organoid differentiation discussed in this report models in vivo cortex development, three characteristic features of cortical progenitors were examined: (1) production of a thick neuroepithelial layer of FOXG1+ progenitor cells, (2) correct apical-basal (inside-out) polarity, and (3) formation of upper and lower layer neurons in distinct locations, as shown in Figure 2.
Figure 2. Review of cortex development. (A) Inside-out development of cortical layers. (B) Three classes of distinct progenitor cell subtypes and their locations.
H9 human embryonic stem cells form cortex progenitors under optimized conditions.
In this study, line H9, a line used for neural differentiation, was studied in the attempt to create cortical organoids. It was first determined whether the initial plating
24 conditions described in Kadoshima et al. (2013), would be optimal for this cell line as well. Kadoshima et al. (2013) reports use of a particular plate shape (V-bottomed 96- well) to force cells together into tighter aggregates to allow better cell-cell interactions that promote growth. Cell number was also controlled (9000 cells per well) in order to produce consistent cell aggregates of the same size with a thick neuroepithelial layer of cells. Inhibition of morphogen pathways has been shown to increase the efficiency at producing cortical cells, and so addition of inhibitors of the Wnt and TGFβ pathways were added to produce exclusively dorsal forebrain progenitors. In this study, plate shape was tested to see if the V-shape was necessary in order to produce consistent cortical progenitors. Organoids plated in V-bottomed 96-well plates produced larger aggregates after eighteen days in culture compared to those plated on U-bottomed 96-well plates, as shown in Figure 3.
Figure 3. V-bottomed 96-well plates improve aggregate formation and generates a thicker neuroepithelial layer on day 18 compared to U-bottomed 96-well plates. Cortical marker (FOXG1) at day 18 shows larger aggregates in V-bottomed plate with thicker continuous neuroepithelial layer. Scale bar=200µM.
25 In the V-bottomed plates there was tighter aggregation and less cell death at day one and by day eighteen a thicker neuroepithelial layer of cells positive for the cortical marker
FOXG1 formed.
Once the initial plating condition was established, what was then tested was which cell number to add per well on day zero. It was found that 9,000 cells per well produced the largest aggregates with the thickest neuroepithelial layer positive for
FOXG1, as shown in Figure 4.
Figure 4. Plating cells at 9,000 cells per well on day 0 generates thickest continuous neuroepithelial layer compared to other cell numbers. Cortical marker (FOXG1) at day 18 shows 9,000 cells per well on V-bottomed 96-well plates formed the largest aggregates with the thickest neuroepithelial layer. Scale bar= 200µM.
Because bone-morphogenic protein (BMP) inhibitors have been shown to enhance the conversion to cortical cell fate (Chambers et al. 2009), they were added to the culture in an attempt to improve cortical organoid synthesis.
26 The BMP inhibitor (LDN193189) was added to the culture for varying amounts of time ranging from the first six days to the entire eighteen days, as shown in Figure 5.
Figure 5. Addition of the BMP inhibitor (LDN193189) does not improve the efficiency at producing FOXG1+ cortical progenitors and inhibits formation of continuous N-cadherin+ apical membrane. Addition of LDN193189 disrupted continuous neuroepithelial layer. Cortical marker (FOXG1), nuclear stain (Hoescht), and apical membrane marker (N-cadherin). Scale bar=200µM.
Surprisingly, LDN193189 did not improve the efficiency of production of cortical progenitors, and disrupted the continuous apical membrane (marked with N-cadherin) present in the control group. In all conditions where LDN193189 was added the formation of a continuous, donut-shaped pool of cortical progenitors with N-cadherin lining the outside was inhibited.
27 These data suggest the same conditions for the first eighteen days in culture established for the FOXG1::Venus reporter line presented in Kadoshima et al. (2013) are also optimal for H9.
Figure 6. Fold change in gene expression of cortical markers increases at stem cell markers decrease in human cortical organoids over time by qPCR. Fold change compared to reference gene GAPDH and day 0 (undifferentiated) human embryonic stem cells. Day 0 (d0) denotes start of differentiation. Stem cell marker OCT4 decreases over time. Cortical markers LHX2, FOXG1, SOX1, and PAX6 increase over time.
28
Figure 7. Human embryonic stem cells efficiently form cortical progenitors with 81.7% of total live cells per organoid positive for cortical marker FOXG1 at day 18. Cortical markers (FOXG1 and LHX2) and nuclear stain (Hoescht). Percent efficiency (number of FOXG1+ cells out of total live cells per organoid) on day 18. (n=16, 5 independent experiments). Scale bar= 100µM.
29 Organoid differentiation efficiently recapitulates in vivo cortical progenitor formation.
Cortical organoids were characterized at various time-points to establish how well this protocol matches in vivo cortical development and identify how the methods could be improved. First, the regional identity of the organoids was determined. As shown in
Figure 6, quantitative Polymerase Chain Reaction (qPCR) analysis showed that after seven days in culture canonical cortical progenitors genes such as FOXG1, LHX2, PAX6, and SOX1 were upregulated, while the expression of the pluripotent stem cell marker
OCT4 decreased. These data are consistent with the establishment of a dorsal cortical identity within the first seven days in culture that is maintained over time. The induction was robust, after eighteen days in culture on average 82% of the live cells in each organoid were positive for the cortical marker FOXG1, as shown in Figure 7. At that time, organoids coalesced to form a donut-like structure with cortical progenitors in a continuous thick layer lining the periphery and dead cells in the middle. By day 35, smaller donut-like “rosette” structures formed along the periphery with dead cells remaining in the middle. At later time-points, the rosette structure expanded in size while maintaining the FOXG1/LHX2+ cortical progenitor identity.
Human embryonic stem-cell derived organoids exhibit laminar organization similar to in vivo development.
Next, three established features of cortical development were examined (1) formation of apical (inside) basal (outside) polarity, (2) formation of cortical neurons in an inside-out manner i.e. neurons positive for upper layer neuronal markers located more superficial to neurons positive for lower layer neurons, and lastly (3) formation of three
30 distinct classes of progenitor cells in a temporal manner (apical radial glia, intermediate progenitors, and basal radial glia). After 35 days in culture, organoids showed correct apical-basal polarity, as shown in Figure 8.
Figure 8. Initial apical-basal polarity of human cortical organoids. Cajal-Retzius neurons in marginal zone (Reelin+), basal marker (laminin), and apical membrane (N-caherin+) at day 35 and day 72. Reelin+ cells in marginal zone become disorganized over time in culture. Scale bar= 100µM.
Apical marker N-cadherin lined the inside of the rosette structure, while laminin lined the basal (outer) edge. Cajal-Retzius cells (marked with Reelin) that reside in the marginal zone (later becomes layer I), were initially confined to the marginal zone, but become disorganized over time in culture. At the same time (day 35) cortical neurons formed and began to organize into distinct layers, as shown in Figure 9. By day fifty-five a distinct separation of subventricular zone and cortical plate was present, shown by the separation of intermediate progenitors (TBR2+) and lower layer neurons (TBR1+). At later time points, the number of intermediate progenitors and lower layer neurons increased and then plateaued after day 90, when upper layer neurons began to form.
31
Figure 9. Organoids form cortical neurons with signs of laminar formation. Progression of layer formation over time. Apical radial glia (SOX2+), intermediate progenitors (TBR2+), lower layer neurons (TBR1+), and upper layer neurons (SATB2+). Quantification shows total number of cells and proportion for each marker at various time points. d35 SOX2/TBR2/TBR1 counts n=6. d90+ SOX2/TBR2/TBR1 counts n=5. d72/90+ double positive and SATB2 n=3. Scale bar= 100µM.
32 Upper layer neurons (SATB2+) began to form after 72 days in culture and increased over time to become similar in number to the lower layer neurons by day ninety. Initially at day 72 there was an abundance of cells double positive for both upper and lower layer markers (TBR1+ and SATB2+), and then by day ninety the neurons began to mature and express SATB2+ only. Unlike neuronal layers in vivo, there is not yet a distinct separation of upper and lower layer neurons by day ninety. Overall, there was a marked separation of subventricular zone and cortical plate over time and generation of both deep and superficial classes of neurons though without obvious separation into distinct layers.
Cortical organoids formed all three classes of progenitors, including basal radial glial cells. Apical radial glia cells (SOX2+) cells formed early in the culture and slightly increased in number over time, as shown in Figure 9. The proportion of these cells decreased over time, as neurons and other progenitors expanded in the rosette structure.
Intermediate progenitors (TBR2+) formed around day 35 and increased in number over time. These cells maintained a constant proportion relative to apical radial glia and lower layer neurons over time. Basal radial glial cells (PAX6+/SOX2+) began to form in the subventricular zone after day 55 and increased in number over time, as shown in Figure
10. To confirm that the number of basal radial glial cells was increasing over time, another proliferation marker was used (Ki67+). Ki67 also marks actively dividing cells, it was used to see which types of progenitors are dividing and where they are located at different time points.
33
Figure 10. Basal-radial glial-like cells form in subventricular zone of human cortical organoids and increase in number over time. Basal radial glia (PAX6+/SOX2+) progenitor cells form in subventricular zone and increase in number over time in culture. Arrowheads denote basal radial glia. Radial glial cells in M-phase marked by phosphorylated vimentin (pVim). Scale bar= 200µM. Quantification of average number of pVim+ cells outside ventricular zone per rosette. n=rosette. d35 n=72. d55 n=83. d72 n=32. d84 n=58. d90+ n=17.
34
Figure 11. Basal radial glia and intermediate progenitor cells dividing in the subventricular zone increase in number over time. White dashed line denotes boundary between ventricular zone (VZ) and subventricular zone (SVZ). Markers for apical radial glia (PAX6+), intermediate progenitors (TBR2+), and dividing cells (Ki67+). Scale bar= 100µM. n=rosette. d55 n=8. d90+ n=6.
35 The percentage of cells positive for Ki67 in the ventricular zone versus subventricular zone remained constant at two time-points, day 55 and 94, at about 10% of total cells, as shown in Figure 11. Of the cells positive for Ki67+, apical radial glia and few intermediate progenitors were found proliferating in the ventricular zone at day 55 and
94. In the subventricular zone, there was an increase in the number of all three classes of progenitors also positive for Ki67. Basal radial glia maintained the same proportion compared to other Ki67+ progenitors, but increased in number. Together, these data show that all three classes of progenitors are present in organoids, with intermediate progenitors and basal radial glia cells prominently increasing over time.
Figure 12. Summary of current human organoid culture.
Summary of human cortical organoids
As shown in Figure 12, human cortical organoids derived from the cell line H9 formed cortical tissue under specific conditions (V plate shape, 9,000 cell number etc.) and model some aspects of in vivo cortex development. Organoids initially showed the correct apical-basal polarity, but became disorganized over time. Upper and lower layer
36 neurons formed, but do not show distinct layer separation. Three classes of progenitors formed in culture and intermediate progenitors and basal radial glia increased in number over time. In conclusion, human cortical organoids formed the correct cell types present in vivo with some signs of laminar formation.
Establishment of reliable cortical organoid culture from mouse embryonic stem cells.
While the goal is to use organoids to further understand human development, the organoid system has yet to be validated. There is an extensive knowledge of genetic mechanisms of mouse development from gene knock-out mouse studies. Using mouse embryonic stem cells to create cortical organoids, one can knock-out particular genes with a known phenotype and assess how well organoids recapitulate this result. With that goal in mind, it was predicted that organoids derived from mouse embryonic stem cells could recapitulate established features of in vivo development. It was found that under a modified protocol organoids derived from mouse embryonic stem cells (line MM13) could reproducibly, consistently, and efficiently producing cortex progenitors, as shown in Figure 13.
Figure 13. Optimized mouse organoid protocol (cell line MM13). On day zero cells are dissociated to single cells and plated 5,000 cells per well in low-attachment U-bottomed 96-well plates.
37 To assess how well the differentiation models in vivo cortex development, three characteristic features of mouse cortical development were addressed: (1) production of a thick neuroepithelial layer of Foxg1+ progenitor cells, (2) formation of cortical neurons, and (3) production of two distinct classes of progenitors abundant in fetal mouse cortex
(apical radial glia and intermediate progenitors).
Figure 14. Plating mouse embryonic stem cells at 5,000 cells per well on day 0 generates most consistent aggregates compared to other cell numbers and transfer to high oxygen produces more rosettes.
MM13 mouse embryonic stem cells form cortex progenitors under optimized conditions.
It was first determined whether line MM13 mouse embryonic stem cells could form cortical progenitors with the conditions presented in Eiraku et al. (2008). This was not the case, so a series of optimization experiments were performed. As shown in Figure
14, 5,000 cells per well on U-bottomed 96 well plates formed the most consistent, round aggregates.
38 Organoids transferred to high oxygen conditions showed increased growth and overall health, as shown in Figure 14; high oxygen conditions were used as described in
Kadoshima et al. (2013) for human cortical organoids.
Figure 15. 3µM IWR1e (Wnt inhibitor) induced highest fold change in Foxg1 expression at day 7 relative to undifferentiated mESCs by qPCR. Fold change compared to reference gene GAPDH and day 0 (undifferentiated) MM13 mouse embryonic stem cells. d7 control is DMSO control without inhibitors. Varying concentration of Wnt inhibitor (IWR1e) and TGFβ (SB431542) added from d0-5. Fold change in cortical markers Foxg1, Lhx2, and Pax6.
39
Figure 16. qPCR data show mouse cortical organoids upregulated cortical genes by day 5 (Line MM13 + 3µM IWR1e). Fold change compared to reference gene GAPDH and day 0 (undifferentiated) MM13 mouse embryonic stem cells. Day 0 (d0) denotes start of differentiation. Cortical markers Lhx2 and Pax6 increase over time. Cortical marker Foxg1 increases after day 2.
Plating at 5,000 cells produced consistent cell aggregates, but were not positive for the cortical marker Foxg1. Next, the correct inhibitor concentration to coerce the cells to adopt a Foxg1 positive dorsal forebrain fate was investigated. Varying concentrations of
Wnt inhibitor (IWR1e) alone, TGFβ inhibitor (SB431542) alone, and in combination were added to the media for the first five days in culture. After seven days, gene expression levels were assessed by quantitative polymerase chain reaction (qPCR) analysis for cortical markers Foxg1, Lhx2, and Pax6. qPCR analysis showed 3µM of Wnt
40 inhibitor (IWR1e) induced the highest fold change of Foxg1 while maintaining high expression of Lhx2 and Pax6 compared to control, as shown in Figure 15. Figure 16 shows the expression of cortical markers over time. Lhx2 and Pax6 were upregulated over time in culture and highly expressed by day five.
Figure 17. Mouse embryonic stem cells efficiently form cortical progenitors with 92.9% of total live cells per organoid positive for cortical marker Foxg1 at day 5. Cortical markers (Foxg1 and Lhx2), apical membrane (N-cadherin and apical protein kinase 3 ‘aPKC’), and apical radial glial marker (Pax6). Scale bar= 200µM. Average percent efficiency (efficiency defined as percentage of cells positive for cortical markers) at producing Foxg1+ cortical progenitors at day 5 (n=3).
41 Foxg1 expression initially decreased, but was upregulated by day five. It was then confirmed with antibody staining that organoids were positive for Foxg1 and Lhx2, as shown in Figure 17. By day five the cells coalesced to form round aggregates with a
‘donut-shaped’ pool of Foxg1+ cortical progenitors and a periphery positive for apical marker (N-cadherin), as shown in Figure 17. This conversion to cortical tissue was highly efficient, with 93% of live cells positive for Foxg1 at day 5.
Figure 18. 1:1 mixture of GMEM/KSR based media and N2 based media followed by N2 media supplemented with CDLC on day 7 “D” produces largest rosette structure positive for cortical markers. Cortical markers (Foxg1 and Lhx2) and apical membrane marker (N-cadherin). Compared switch to N2 based media on day 5 (control) “A” to switch on day 5 to N2 media supplemented with chemically defined lipid concentrate (CDLC) “B”, switch on day 7 to N2 media supplemented with CDLC “C”, or switch day 5 to a 1:1 mixture of GMEM/KSR based media (SFEBq) and N2 based media followed by N2 media supplemented with CDLC on day 7 “D”. Addition of CDLC improved organoid growth. 1:1 mixture of SFEBq:N2 media “D” yielded larger rosette structures. Scale bar= 50µM.
42 Organoid differentiation recapitulates some aspects of in vivo cortical progenitor formation.
Once inhibitor concentration and cell number were established, two established features of cortical development were examined (1) formation cortical neurons and (2) formation of two distinct classes of progenitor cells (apical radial glia and intermediate progenitors). The initial protocol produced organoids with unhealthy looking cells with small/abnormal nuclei and small rosette structures. Additional factors were added to the media to try to improve the overall health of the cells and to achieve larger, more complex rosette structures. In order to achieve larger rosette structures, the transition from GMEM/KSR based media to N2 based media was optimized. Four conditions were tested: (1) the protocol outlined in Eiraku et al. 2008, where the switch to N2 based media is at day five, as show in Figure 18A, (2) whether addition of chemically defined lipid concentrate (CDLC) would improve the health of organoids, as shown in Figure 18B, (3) switch to N2 based media at day seven, as shown in Figure 18C, and (4) a gradual switch to N2 based media where day five to seven a 1:1 mixture of GMEM/KSR based media:N2 based media and then after day seven N2 based media is added, as shown in
Figure 18D. It was found that addition of chemically defined lipid concentrate improved the health of the organoids, as shown in Figure 18. The gradual shift to N2 media shown in Figure 18D produced the best rosette formation, whereas in conditions A and B small rosettes formed (indicated by N-cadherin staining) and in condition C the N-cadherin layer had not rolled over and no rosettes formed. In condition D, large rosettes were beginning to form, as shown by the partial rolling over of N-cadherin staining. It was then assessed how well the four conditions produced cortical progenitors and neurons, as
43 shown in Figure 19. Again condition D formed the largest rosettes, with many apical radial glia (Pax6+), intermediate progenitors (Tbr2+), and cortical neurons (Ctip2+).
Figure 19. 1:1 mixture of GMEM/KSR based media and N2 based media followed by N2 media supplemented with CDLC on day 7 “D” produced largest rosette structure with intermediate progenitors and cortical neurons. Apical radial glia marker (Pax6+), intermediate progenitor (Tbr2+), and lower layer neuronal marker (Ctip2+). 1:1 mixture SFEBq:N2 media yielded larger Pac6+ rosette structures with more intermediate progenitors. Scale bar= 50µM.
44
Figure 20. Summary of current mouse organoid culture.
Summary of mouse cortical organoids
Under optimized conditions, MM13 mouse embryonic stem cells formed cortical tissue that modeled some aspects of in vivo cortex development, as shown in Figure 20.
Under the conditions specified above, cortical organoids formed two classes of cortical progenitors and cortical neurons in large rosette structures. In conclusion, mouse cortical organoids formed the correct cell types present in vivo, but did not yet display a defined laminar organization characteristic of mouse cortex.
Expression of transcription factors hypothesized to be involved in brain development match in vivo expression.
Previous data suggest that transcription factors FOXP1 and FOXP4 may play distinct roles in cortical progenitor formation and/or maintenance. Use of the organoid system could provide insight into what role these genes play in human cortical development. This study investigated whether the expression of two transcription factors hypothesized to be involved with human cortical expansion (FOXP1 and FOXP4) matched expression patterns from human and mouse fetal tissue.
45
Figure 21. FOXP1 expression overlaps with apical and basal radial glial cells and FOXP4 expression overlaps with intermediate progenitors in human cortical organoids over time. Marker for apical radial glia/basal radial glial cells (SOX2+) and intermediate progenitors (TBR2+). FOXP1 expression overlaps with radial glial cells (SOX2+) and in the cortical plate. FOXP4 expression overlaps with intermediate progenitors (TBR2+) and in the cortical plate. FOXP1 and FOXP4 expression does not appear to overlap. Scale bar= 100µM. (n=3).
46
Figure 22. Foxp1/Foxp4 expression partially overlaps and Foxp4 expression overlaps with apical radial glia in mouse cortical organoids over time. Marker for apical radial glia (Sox2). Foxp1 expression overlaps with Foxp4 at early timepoints, but appears to be downregulated by day 7. Foxp4 expression overlaps with Sox2. Scale bar= 200µM. Scale bar= 100µM for higher magnification images on right.
47 FOXP1 expression in human cortical organoids overlaps with apical and basal radial glial cells and sparsely in the cortical plate
Human cortical organoids expressed the transcription factor FOXP1 over time in culture, as shown in Figure 21. Expression was restricted to 99% of apical radial glia cells in the ventricular zone and 85% of basal radial glial cells in the subventricular zone at day 94. While radial glia cells highly expressed FOXP1, only 67% of all the cells positive for FOXP1 were also positive for apical and basal radial glia cell marker SOX2.
The other 33% of cells positive for FOXP1 were sparsely present in the cortical plate.
FOXP4 expression in human cortical organoids overlaps with intermediate progenitors in the subventricular zone and in the cortical plate
Human cortical organoids expressed the transcription factor FOXP4 over time in culture, as shown in Figure 21. FOXP4 was highly expressed in intermediate progenitor cells (Tbr2+) in the subventricular zone at day 94, 91% of cells positive for Tbr2 were also positive for FOXP4. Only 23% of the cells positive for FOXP4 was also positive for intermediate progenitor markers. FOXP4 was also highly expressed in the cortical plate,
77% of cells positive for FOXP4 were present in the cortical plate and not positive for intermediate progenitor markers.
Summary of FOXP1 and FOXP4 expression in human cortical organoids
FOXP1 was highly expressed in apical and basal radial glial cells and sparsely in the cortical plate. FOXP4 was highly expressed in intermediate progenitor cells and in the cortical plate. Expression of FOXP1 and FOXP4 did not appear to overlap.
48 Foxp1 and Foxp4 are expressed in mouse cortical organoids with some overlap at various time points
Mouse cortical organoids expressed Foxp1 and Foxp4 for the first three days in culture with partially overlapping expression, as shown in Figure 22. Initially, Foxp1 and
Foxp4 expression overlapped with apical radial glia marker (Sox2+). At later time points
(day 5-7) Foxp1 appeared to be down-regulated to low levels, while Foxp4 maintained expression overlapping with apical radial cells.
Summary of Foxp1 and Foxp4 expression in mouse cortical organoids
Mouse cortical organoids expressed Foxp1 and Foxp4 at early time points and overlapped with apical radial glial cell markers. Foxp1 was down-regulated at later time points, while Foxp4 maintained expression overlapping with apical radial glial cells.
49 CHAPTER 4: DISCUSSION
This study investigated whether stem cells can form organoids that model aspects of in vivo cortex development. Under optimized conditions, both cell lines tested, H9 human embryonic stem cells and MM13 mouse embryonic stem cells, could form cortical tissue that recapitulated some aspects of cortical development. This suggests that with optimization of cell number, inhibitor concentration, etc. theoretically any stem cell line could be used to create cortical organoids. This is promising find for the field. If organoids were to be used for disease modeling, more than one genetic background (cell line) would need to be tested to make claims about the disease mechanism for a diverse population of individuals with a disease. The same logic applies for gene function studies. Thus, a larger set of cell lines will need to be tested to make claims about the human population as a whole.
It was first tested whether the V-shaped plates described in Kadoshima et al.
(2013) were necessary to form consistent cortical organoids with human embryonic stem cells. While both the U and V-shaped plates produced cortical progenitors, the V-shaped plate appeared to promote better growth and formation of a thicker neuroepithelial layer.
This is likely due to enhanced cell-cell contacts forced in the V versus U shape. Similarly while all cell numbers tested produced FOXG1+ cortical cells, plating at 9,000 cells per well produced aggregates with a thicker neuroepithelial cell layer than other cell numbers. Next, the addition of BMP inhibitor LDN193189 was tested. In all conditions where LDN193189 was added, FOXG1+ cortical progenitors formed, but the N- cadherin+ apical membrane was disrupted.
50 Previous studies using BMP inhibitors (Chambers et al. 2009) used adherent cultures; cell signaling may be different under these conditions compared to the 3D cell contacts in organoids.
Human cortical organoids were then characterized up to day one hundred to see how well current organoids match in vivo development and in which areas the methods could be improved. Organoids efficiently formed cortical progenitors, and maintained that fate after Wnt and TGFβ inhibitors were removed from the media at day eighteen.
Human cortical organoids were initially able to reproduce the apical-basal polarity seen in vivo. Over time the basal membrane broke down and cells within the marginal zone
(Cajal-Retzius cells) began to migrate inward. In vivo, the meninges line the outer membrane of the cortex and produce a variety of small molecules that aid in neuronal migration and maturation such as neuronal growth factors, laminin, and retinoic acid. The lack of these cues in the culture could explain why the basal membrane breaks down over time. Additional experiments to determine the requirement of different factors secreted by the meninges alone or in combination could improve the overall rosette structure within organoids to better match in vivo development. Human cortical organoids were able to produce all the cell types seen in vivo, with separation of the subventricular zone and cortical plate, but not of upper and lower neuronal layers within the cortical plate.
Again, this could be due to the lack of cell migratory signals normally secreted by the meninges or the fact that Cajal-Retzius cells migrated inward. Cajal-Retzius cells produce
Reelin, a neuronal migratory signal, by being scattered throughout the cortical plate proper migration could have been disrupted. Three types of cortical progenitors formed,
51 including basal radial glia cells, which abundance of is specific to primates. This suggests human cells have an intrinsic program to produce these cells.
MM13 mouse embryonic stem cells were also able to recapitulate some aspects of cortex development under optimized conditions. Plating cells at 5,000 cells per well produced the most consistent cell aggregates. Organoids transferred to high oxygen conditions showed improved overall health and growth. Kadoshima et al. (2013) reported high oxygen conditions are essential for the continued growth of human cortical organoids, the same appears to apply to mouse cortical organoids. Since cell aggregates were initially not positive for cortical markers, an extensive inhibitor concentration study was done. It was found that 3µM of Wnt inhibitor (IWR1e) induced the highest fold change in Foxg1 expression by quantitative polymerase chain reaction analysis. Eiraku et al. (2008) and Nasu et al. (2010) have reported different concentrations of the same inhibitors as optimal for other cell lines. Foxg1 expression was only upregulated in a small range of concentrations, suggesting concentration is cell line specific. Cortical markers were upregulated over time, and confirmed by antibody staining. After finding the optimal inhibitor concentration, the conversion to cortical cells was highly efficient;
92.9% of total live cells per aggregate were positive for Foxg1. Since mouse organoid health deteriorated after day five, the switch to N2 media was optimized. It was found that a gradual switch to N2 media produced the largest aggregates positive for cortical markers with abundant apical radial glia, intermediate progenitors, and neurons.
Separation of the different layers was limited, and further optimization is needed to allow growth past ten days. Additional factors may need to be added to the media to signal neuronal migration and maturation such a laminin, retinoic acid, or B27. Nasu et al.
52 (2012) reports addition of laminin to the culture media promotes formation of more continuous rosettes with thicker laminar structure. Currently, the media is devoid of any cues or growth factors so this could explain the lack of layer separation and continued growth. Mouse cortical organoids need further optimization to be used to validate the organoid system. Once this is done, genetic manipulations of stem cell lines could be done to match knock-out mice experiments done in vivo. If organoids can reproduce known phenotypic features seen in knock-out models, this proof-of-principle experiment could then be applied to human organoids. This would greatly increase the explanatory power behind human-specific mechanistic features provided by the organoid system.
Lastly, this study investigated two transcription factors hypothesized to be involved in cortical expansion across evolution. It has been shown that members of the
Forkhead-box subfamily P act as a repressor complex, influencing differentiation and cell fate in other tissues in the body (Li et al. 2012; Rousso 2012; Zhao et al. 2015). It is hypothesized that FOXP1 and FOXP4 act by a similar mechanism in the brain. To assess the function of these genes in cortical growth, genetic manipulations could be done to overexpress or delete the functional domain of these genes in stem cells, and then culture organoids and analyze differences in cortical growth etc. This study reports preliminary results that confirmed FOXP1 and FOXP4 were expressed in human and mouse cortical organoids. In human cortical organoids, this pattern matched previous unpublished data from the Novitch lab, where FOXP1 was expressed in radial glial cells and FOXP4 was expressed in intermediate progenitors. In mouse cortical organoids, these data were inconclusive. FOXP1 and FOXP4 was expressed, but the time points in culture are estimated to be earlier than what was assessed in mouse fetal tissue.
53 If mouse organoids can be optimized further, by addition of different factors to promote continued growth and laminar organization past ten days, the expression pattern may match what has been previously seen in mouse fetal tissue.
One limitation of the current culture is that the health and structure of organoids deteriorate after about one hundred days with human cells or ten days with mouse cells.
This could be due to limited oxygen penetration to inner tissues once the rosettes reach a certain thickness or because a lack of vascularization. The brain has a complex system of blood vessels that provide essential nutrients and oxygen. Without this, organoids are limited in growth potential. Future experiments could look into co-culture methods with vascular progenitors or how to get oxygen/nutrients to the inner portion of organoids.
Another limitation, like other previous studies, is an incomplete assessment of regional identity and developmental time-course. Currently CoNTExt transcriptome analysis is under way, and once completed will greatly strengthen the results presented here, but until then dorsal-forebrain identity and developmental timing remain estimates.
Overall, this study has confirmed cortical organoids derived from human and mouse embryonic stem cells can model some aspects of in vivo cortex development and suggested areas in which improvement is needed. These data suggest cortical organoids can be a reliable model for some aspects of cortical development. Human-specific features of development that could never before be studied can now begin to be addressed in the organoid culture system. Insight into normal and abnormal development from organoids has the potential to advance the field of developmental neuroscience and lead to discoveries of new ways to treatment and/or prevent neurodevelopmental disorders.
54 References
Bacon, C. and G. Rappold. 2012. The distinct and overlapping phenotypic spectra of FOXP1 and FOXP2 in cognitive disorders. Human Genetics. 131(11):1687–1698.
Bar, I., C., de Rouvroit, and A. Goffinet. 2000. The evolution of cortical development. A hypothesis based on the role of the Reelin signaling pathway. Trends in Neurosciences. 23(12):633–638.
Betizeau, M., V. Cortay, D. Patti, S. Pfister, E. Gautier, A. Bellemin-Ménard, M. Afanassieff, C. Huissoud, R. Douglas, H. Kennedy, and C. Dehay. 2013. Precursor diversity and complexity of lineage relationships in the outer subventricular zone of the primate. Neuron. 80:442–457.
Chambers, S., C. Fasano, E. Papapetrou, M. Tomishima, M. Sadelain, and L. Studer. 2009. Highly efficient neural conversion of human ES and iPS cells by dual inhibition of SMAD signaling. Nature Biotechnology. 27:275–281.
Dekimoto, H., T. Terashima, and Y. Katsuyama. 2010. Dispersion of the neurons expressing layer specific markers in the reeler brain. Development, Growth, and Differentiation. 52:181–193.
Eiraku, M., K. Watanabe, M. Matsuo-Takasaki, M. Kawada, S. Yonemura, M. Matsumura, T. Wataya, A. Nishiyama, K. Muguruma, and Y. Sasai. 2008. Self- organized formation of polarized cortical tissues from ESCs and its active manipulation by extrinsic signals. Cell Stem Cell. 3(5):519–532.
Fietz, S. and W. Huttner. 2011. Cortical progenitor expansion, self-renewal and neurogenesis – a polarized perspective. Current Opinion in Neurobiology. 21:23– 35.
Florio, M. and W. Huttner. 2014. Neural progenitors, neurogenesis, and the evolution of the neocortex. Development. 141:2182–2194.
Gabut, M., P. Samavarchi-Tehrani, X. Wang, V. Slobodeniuc, D. O’Hanlon, H. Sung, M. Alvarez, S. Talukder, Q. Pan, E. Mazzoni, S. Nedelec, H. Wichterle, K. Woltjen, T. Hughes, P. Zandstra, A. Nagy, J. Wrana, and B. Blencowe. 2011. An alternative splicing switch regulates embryonic stem cell pluripotency and reprogramming. Cell. 147:132–146.
Gaspard, N., T. Bouschet, R. Hourez, J. Dimidschstein, G. Naeije, J. van den Ameele, I. Expuny-Camacho, A. Herpoel, L. Passante, S. Schiffmann, A. Gaillard, and P. Vanderhaeghen. 2008. An intrinsic mechanism of corticogenesis from embryonic stem cells. Nature. 455:351–358.
55 Hannenhalli, S. and K. Kaestner. 2009. The evolution of Fox genes and their role in development and disease. Nature Reviews Genetics. 10:233–240.
Hansen, D., J. Lui, P. Parker, and A. Kriegstein. 2010. Neurogenic radial glia in the outer subventricular zone of human neocortex. Nature. 464:554–561.
Hofman, M. 2014. Evolution of the human brain: when bigger is better. Frontiers in Neuroanatomy. 8:15.
Jessell, T. and J. Sanes. 2000. Chapter 53: The Generation and Survival of Nerve Cells. In: Principles of Neural Science Fourth Edition. Kandel, E., J. Schwartz, and T. Jessell. (eds.) Mc-Graw-Hill, N.Y..
Kadoshima, T., H. Sakaguchi, T. Nakano, M. Soen, S. Ando, M. Eiraku, and Y. Sasai. 2013. Self-organization of axial polarity, inside-out layer pattern, and species- specific progenitor dynamics in human ES cell-derived neocortex. Proceedings of the National Academy of Sciences of the United States of America. 110(50):20284–20289.
Lancaster, M., M. Renner, C. Martin, D. Wenzel, L. Bicknell, M. Hurles, T. Homfray, J. Penninger, A. Jackson, and J. Knoblich. 2013. Cerebral organoids model human brain development and microcephaly. Nature. 501:373–379.
Lancaster, M. and J. Knoblich 2014a. Organogenesis in a dish: modeling development and disease using organoid technologies. Science. 345(6194):1–9.
Lancaster, M. and J. Knoblich. 2014b. Generation of cerebral organoids from human pluripotent stem cells. Nature Protocols. 9(10):2329–2340.
Le Fevre, A., S. Taylor, N. Malek, D. Horn, C. Carr, O. Abdul-Rahman, S. O’Donnell, T. Burgess, M. Shaw, J. Gecz, N. Bain, K. Fagan, and M. Hunter. 2013. FOXP1 mutations cause intellectual disability and a recognizable phenotype. American Journal Medical Genetics. 161A:3166–3175.
Li, S., J. Weidenfeld, and E. Morrisey. 2004. Transcriptional and DNA binding activity of the Foxp1/2/4 family is modulated by heterotypic and homotypic protein interactions. Molecular & Cellular Biology. 24:809–822.
Li, S., Y. Wang, Y. Zhang, M. Lu, F. DeMayo, J. Dekker, P. Tucker, and E. Morrisey. 2012. Foxp1/4 control epithelial cell fate during lung development and regeneration through regulation of anterior gradient 2. Development. 139:2500– 2509.
Lui, J., D., Hansen, and A. Kriegstein. 2011. Development and evolution of the human neocortex. Cell 146:18–36.
56 Malatesta, P., I. Appolloni, and F. Calzolari. 2008. Radial glia and neural stem cells. Cell and Tissue Research. 331:165–178.
Mariani, J., M. Simonini, D. Palejev, L. Tomasini, G. Coppola, A. Szekely, T. Horvath, and F. Vaccarino. 2012. Modeling human cortical development in vitro using induced pluripotent stem cells. Proceedings of the National Academy of Sciences of the United States of America. 109:12770–12775.
Mariani, J., G. Coppola, P. Zhang, A. Abyzov, L. Provini, L. Tomasini, M. Amenduni, A. Szekely, D. Palejev, M. Wilson, M. Gerstein, E. Grigorenko, K. Chawarska, K. Pelphery, J. Howe, and F. Vaccarino. 2015. FOXG1-Dependent dysregulation of GABA/glutamate neuron differentiation in autism spectrum disorders. Cell. 162:375–390.
Meyer, G. 2010. Building a human cortex: the evolutionary differentiation of Cajal- Retzius cells and the cortical hem. Journal of Anatomy. 217:334–343.
Muñoz-Sanjuán, I., and A. Brivanlou. 2002. Neural induction, the default model and embryonic stem cells. Nature Reviews Neuroscience. 3:271–280.
Nadarajah, B., and J. Parnavelas. 2002. Modes of neuronal migration in the developing cerebral cortex. Nature Reviews Neuroscience. 3:423–432.
Noctor, S., A. Flint, T. Weissman, R. Dammerman, and A. Kriegstein. 2001. Neurons derived from radial glia cells establish radial units in neocortex. Nature. 409:714– 720.
Palumbo, O., L. D’Agruma, A. Minenna, P. Palumbo, R. Stallone, T. Palladino, L. Zelante, and M. Carella. 2013. 3p14.1 de novo microdeletion involving the FOXP1 gene in an adult patient with autism, severe speech delay and deficit of motor coordination. Gene. 516:107–113.
Paşca, A., S. Sloan, L. Clarke, Y. Tian, C. Makinson, N. Huber, C. Kim, J. Park, N. O’Rourke, K. Nguyen, S. Smith, J. Huguenard, D. Geschwind, B. Barres, and S. Paşca. 2015. Functional cortical neurons and astrocytes from human pluripotent stem cells in 3D culture. Nature Methods. 12:671–678.
Pollen, A., T. Nowakowski, J. Chen, H. Retallack, C. Sandoval-Espinosa, C. Nicholas, J. Shuga, S. Liu, M. Oldham, A. Diaz, D. Lim, A. Leyrat, J. West, and A. Kriegstein. 2015. Molecular identity of human outer radial glia during cortical development. Cell. 163:55–67.
Rakic, P. 1995. A small step for the cell, a giant leap for mankind: a hypothesis of neocortical expansion during evolution. Trends in Neuroscience. 18:383–388.
57 Rousso, D., C. Pearson, Z. Gaber, A. Miquelajauregui, S. Li, C. Portera-Cailliau, E. Morrisey, and B. Novitch. 2012. Foxp-mediated suppression of N-cadherin regulates neuroepithelial character and progenitor maintenance in the CNS. Neuron. 74:314–330. Shimogori, T., V. Banuchi, H. Ng, J. Strauss, and E. Grove. 2004. Embryonic signaling centers expressing BMP, WNT, and FGF proteins interact to pattern the cerebral cortex. Development. 131(22):5639–5647.
Stein, J., L. de la Torre-Ubieta, Y. Tian, N. Pariksak, I. Hernández, M. Marchetto, D. Baker, D. Lu, C. Hinman, J. Lowe, E. Wexler, A. Muotri, F. Gage, K. Kosik, and D. Geschwind. 2014. A quantitative framework to evaluate modeling of cortical development by neural stem cells. Neuron. 83:69–86.
Takahashi, K., F. Liu, K. Hirokawa, and H. Takahashi. 2008. Expression of Foxp4 in the developing and adult rat forebrain. Journal Neuroscience Research. 86(14):3106– 3116.
Teufel, A., E. Wong, M. Mukhopadkyay, N. Malik, and H. Westphal. 2003. FoxP4, a novel forkhead transcription factor. Biochimica, et Biophysica Acta – Gene Structure and Expression. 1627(2-3):147–152.
Watanabe, K., D. Kamiya, A. Nishiyama, T. Katayama, S. Nozaki, H. Kawasaki, Y. Watanabe, K. Mizuseki, and Y. Sasai. 2005. Directed differentiation of telencephalic precursors from embryonic stem cells. Nature Neuroscience. 8:288– 296.
Watanabe, K., M. Ueno, D. Kamiya, A. Nishiyama, M. Matsumura, T. Wataya, J. Takahashi, S. Nishikawa, S. Nishikawa, K. Muguruma, and Y. Sasai. 2007. A ROCK inhibitor permits survival of dissociated human embryonic stem cells. Nature Biotechnology. 25:681–686.
Zhao, H., W. Zhou, Z. Yao, Y. Wan, J. Cao, L. Zhang, J. Zhao, H. Li, R. Zhou, B. Li, G. Wei, Z. Zhang, C. French, J. Dekker, Y. Yang, S. Fisher, H. Tucker, and X. Guo. 2015. Foxp1/2/4 regulate endochondral ossification as a suppresser complex. Developmental Biology. 398:242–254.
58 APPENDIX: SUPPLEMENTAL METHODS
Quantitative Polymerase Chain Reaction Analysis
Total RNA extraction was done using the Qiagen RNeasy Mini Kit. cDNA was synthesized by reverse transcription using the Superscript III First-Strand Synthesis
System (Invitrogen). qPCR reactions were done using cDNA combined with LightCycler
480 SYBR Green I Master Mix (Roche) and the following primers. Primer sequences used for mouse cells were Foxg1 (Forward: 5’-ACA AGA AGA ACG GCA AGT ACG-
3’, Reverse: 5’-CAT AGA TGC CAT TGA GCG TCA-3’), Lhx2 (Forward: 5’-CTG TTC
CAC AGT CTG TCG GG-3’, Reverse: 5’-CAG CAG GTA GTA GCG GTC AG-3’),
Pax6 (Forward: 5’-GCA CAT GCA AAC ACA CAT GA-3’, Reverse: 5’-ACT TGG
ACG GGA ACT GAC AC-3’, and GAPDH (Forward: 5’-GGC CTT CCG TGT TCC
TAC-3’, Reverse: 5’-TGT CAT CAT ACT TGG CAG GTT-3’). Primer sequences for human cells were OCT4 (Forward: 5’-GGA GAA GCT GGA GCA AAA C-3’, Reverse:
5’-ACC TTC CCA AAT AGA ACC CC-3’), LHX2 (Forward: 5’-TCG GGA CTT GGT
TTA TCA CCT-3’, Reverse: 5’-GTT GAA GTG TGC GGG GTA CT-3’), FOXG1
(Forward: 5’-CCA GAC CAG TTA CTT TTTCCC-3’, Reverse: 5’-TGA AAT AAT
CAG ACA GTC CCC C-3’), SOX1 (Forward: 5’-TCC TGG AGT ATG GAC TGT CCG
3’, Reverse: 5’-GAA TGC AGG CTG AAT TCG G-3’), PAX6 (Forward: 5’-TGT CCA
ACG GAT GTG TGA GTA-3’, Reverse: 5’-CAG TCT CGT AAT ACC TGC CCA-3’), and GAPDH (Forward: 5’-CAG TCT CGT AAT ACC TGC CCA-3’, Reverse: 5’-TGT
AGT TGA GGT CAA TGA AGG G-3’). Samples were run on the Roche LightCycler
480 real-time PCR system in duplicates or triplicates and relative fold change in
59 expression of each target gene was normalized to GAPDH and experimental day 0
(undifferentiated mouse or human embryonic stem cells).
Fixation/Cryosectioning
Organoids were fixed at various time points in 4% paraformaldehyde (Ted Pella) for 20 minutes on ice. After fixed, organoids were washed 3 times in PBS (Life
Technologies) and then allowed to sink in 30% sucrose for 1-2 hours on ice. Once sunk in 30% sucrose, organoids were transferred to a freezing block in OCT (VWR). Blocks were cryosectioned on a LEICA CM3050 S at 12µm.
Immunostaining/Imaging
Sections were immunostained in AB Block (1% HIHS, 0.1% TritonX, 0.01% Na
Azide in PBS) with primary antibodies overnight at 4°C and secondary antibodies for one hour at room temperature. Primary antibodies used were goat anti-Sox2 (Santa Cruz
Biotechnology sc-17320), 1:500; mouse anti-Pax6 (Developmental Studies Hybridoma
Bank), 1:100; rabbit anti-Pax6, (MBL pd022), 1:1,000; rabbit anti-aPKC (PKCζ), (Santa
Cruz Biotechnology SC-216), 1:100; mouse anti-N-cadherin (BD Biosciences 610920),
1:1000; rabbit anti-Laminin (Abcam ab30320), 1:1000; rabbit anti-Caspase3 (Cell
Signaling 96615), 1:250; rabbit anti-Foxg1 (Abcam ab18259), 1:1000; goat anti-Lhx2
(Santa Cruz Biotechnology sc-19344), 1:1000; chicken anti-Tbr2 (Millipore ab15894),
1:1000; rabbit anti-Tbr1 (Abcam ab31940), 1:2000; Rat anti-Ctip2 (Abcam ab18465),
1:1000; rabbit anti-Calretinin (Chemicon ab5054), 1:2000; mouse anti-Satb2 (Abcam ab51502), 1:100; mouse anti-Reelin (MBL D223-3), 1:300; mouse anti-Phosphorylated
Vimentin (MBL D076-3), 1:500; rabbit anti-Ki67 (ThermoScientific RM-9106-01),
1:2000; guinea pig anti-Foxp1 (Rousso et al., 2008);
60 and rabbit anti-Foxp4 (Millipore ab274), 1:1000. Alexa488-, Dylight649-, FITC-, Cy3-, and
Cy5- conjugated secondary antibodies from Jackson ImmunoResearch Laboratories, Inc.
(West Grove, PA) were used. Hoescht (Thermo Scientific) was used for nuclear staining.
Fluorescence images were taken on a Zeiss LSM 780 confocal microscope. Images were processed on Zen 2012 software and Adobe Photoshop. Quantification was done using
NIH image processing program, ImageJ.
61