We describe a simplified 3D differentiation protocol for hPSCs, using defined medium and reduced growth factors, capable of generating cell aggregates with early neuroepithelial structures and positive for cerebellar-associated markers, as well as an optional 2D modification for differentiating cells as a monolayer to generate functional neurons.
Reducing the complexity and cost of differentiation protocols is important for researchers. This interest fits with concerns about possible unintended effects that extrinsic patterning factors might introduce into human pluripotent stem cell (hPSC) models of brain development or pathophysiology, such as masking disease phenotype. Here, we present two cerebellar differentiation protocols for hPSCs, designed with simpler startup method, fewer patterning factors, and less material requirements than previous protocols. Recently, we developed culture procedures, which generate free-floating 3-dimensional (3D) products consistent with other brain "organoid" protocols, including morphologies relevant to modeling brain development such as sub/ventricular zone- and rhombic lip-like structures. The second uses an adherent, 2D monolayer procedure to complete differentiation, which is shown capable of generating functional cerebellar neurons, as products are positive for cerebellar-associated markers, and exhibit neuron-like calcium influxes. Together, these protocols offer scientists a choice of options suited to different research purposes, as well as a basic model for testing other types of streamlined neural differentiations.
In vitro protocols for differentiating hPSCs toward cerebellar lineages initially operated on the principle of mimicking in vivo cerebellar development1,2,3,4. As such, they required a succession of factors introduced at specific times to drive pro-cerebellar patterning and maturation. Chief among these were WNT, bone morphogenetic proteins (BMPs), and fibroblast growth factors (FGFs) with known roles in mid-hindbrain development and formation of the isthmic organizer5,6,7. Of course, each additional step and factor means an increase in labor-intensive manipulations and greater expense for the researcher, and so developing simpler protocols capable of achieving equal results is of interest. This practical issue dovetails nicely with the hypothetical question, whether cells require such tight, external control over their development in vitro.
For cerebellar differentiation, a protocol published in 2015 addressed the necessity of using an extensive number of growth factors by using only FGF2, FGF19, and stromal cell-derived factor 1 (SDF1) for patterning purposes8. This study also differed from prior cerebellar protocols, by using a free-floating 3D culture system. In addition to producing cells positive for cerebellar markers, the brain "organoids" generated by their technique were shown to exhibit relevant morphology, unavailable in traditional 2D monolayer cultures, such as rhombic lip-like structures. Although less complex and costly with respect to growth factors, other features such as formation of uniform embryoid bodies (EBs) and culture in 96-well plates (96WPs), made it procedurally complex during initial steps. Another 3D protocol published the same year, reported successful differentiation to neural lineages using common and inexpensive cell culture techniques9. Although this group was investigating cortical rather than cerebellar differentiation, application of their concept for cerebellar differentiation could not be discounted.
We recently reported a 3D cerebellar differentiation protocol using a reduced number of patterning factors (namely, FGF2, 4, and 8), as well as a simplified setup by keeping cells in 6-well plates (6WPs) throughout to minimize medium requirements10. To assist production of granule cells, smoothened agonist (SAG) was used during the final maturation step. SAG is a less expensive chemical alternative to sonic hedgehog (SHH), which had been used in earlier cerebellar protocols, due to its role in promoting the growth of granule cell precursors (GCPs) in vivo1,2,11,12,13. Differentiation products were consistent with those from other 3D protocols, including the presence of cerebellar-associated markers in morphologically relevant structures8,9. Such results reinforce the earlier message that detailed mimicry of the in vivo environment may not be necessary for complex 3D in vitro differentiation protocols.
In addition to the 3D protocol, this report describes a 2D protocol, designed with the same quick setup, basic materials, and reduced number of growth factors. It is capable of producing cells from human embryonic stem cells (hESCs) or induced pluripotent stem cells (hiPSCs), positive for markers associated with early neural, cerebellar, and granule cell identities. In addition, calcium imaging indicates the presence of functional human neurons. The ability to choose between protocols, adds a level of flexibility for researchers, for those interested in either: (1) generating specific cell types, (2) modeling human brain development and associated structures, (3) analysis optimized in monolayer settings (e.g., patch-clamp recordings), or (4) cell-cell interactions in mixed neural cultures. Their simple and low-cost nature makes them accessible for researchers who are new to the hPSC field, or need base hPSC procedures from which to explore further differentiation options.
1. Preparations
NOTE: For all steps see Table of Materials for specific items.
2. Protocol 1: Feeder-free hPSC Culture
NOTE: hESCs were obtained from a non-commercial organization (line H01, see Table of Materials). Three hiPSC control lines (hvs51, 60, and 88) were generated by reprogramming fibroblasts from three healthy human patients (fibroblasts were derived from anonymous, non-identifiable donors and therefore exempt from IRB approval)10,17.
3. Protocol 2: 3D "Organoid" Differentiation
4. Protocol 3: Alternative 2D Differentiation Culture
Visual Overview of Reduced Growth Factor 2D and 3D Cerebellar Differentiation Protocols
Figure 1 shows the overall timeline for the 2D and 3D cerebellar differentiation protocols, identifying extrinsic factors and time of plating. The typical progress for hPSCs undergoing 3D cerebellar differentiation is depicted in Figure 2: with hESC line H01 starting as colonies in feeder-free culture at day 0 (upper left of figure); undergoing EB formation by day 2 (upper middle); growing into larger cell aggregates with apparent lumen following neural induction with RA and FGF8 at day 14 (upper right); forming aggregates of varying size and shape at day 28 (lower left); developing in complexity with different structures of a single aggregate indicated at day 28 (lower middle); and with continued morphological changes to the same structures at day 35 (lower right). The typical progress for hPSCs undergoing 2D cerebellar differentiation is depicted in Figure 3: with hESC line H01 as colonies in feeder-free culture at day 0 (upper left of figure, with circle indicating area of differentiated cells among hESC colonies); undergoing EB formation by day 2 (upper middle); growing into larger cell aggregates with apparent lumen following neural induction with RA and FGF8 at day 13 (upper right); proliferating as adherent cells after plating at day 14 (lower left); and then as a monolayer of cells with more complex/mature morphology at day 35 under low (lower middle) and high magnification (lower right).
3D Products Exhibit Markers and Structures of Early Neuroepithelium
Figure 2 (lower left image) and Figure 4 show the heterogeneity of 3D aggregate morphology seen throughout culturing, due to varying growth and/or maturation rates, as well as the stochastic merging or breaking apart of aggregates. Despite heterogeneity, each differentiation produces aggregates exhibiting early neural and neuronal markers, including cerebellar granule marker ZIC1, as indicated by immunocytochemistry (ICC) staining in Figure 5. More importantly, Figure 5 and Figure 6 suggest that a simple 3D culture, with reduced growth factors, is capable of generating aggregates with complex structures related to brain development such as the early neuroepithelium and rhombic lip.
2D Products Exhibit Cerebellar Markers and Functional Neuronal Activity
While 2D cultures cannot reproduce complex 3D structures, they are capable of generating cells exhibiting early neural and neuronal markers, including cerebellar granule cell marker ZIC1, as indicated by ICC staining in Figure 7. Gene expression analysis via RT-PCR, as seen in Figure 8, supports ICC staining results, though the presence of early granule cell marker ATOH1 is variable between experiments and lines. Calcium imaging is more easily handled in 2D culture. As seen in Figure 9, Supplemental Video 1, and Supplemental Video 2, electrically stimulated cells show calcium influxes that are typical of neuronal firing patterns, suggesting generation of functional neurons.
Figure 1: Timeline of differentiation protocol (starting at day 0 of differentiation). Solid line boxes indicate when specific factors are added to the culture medium, and dotted line boxes indicate plate-coating for optional 2D modification. For FGF2, the down arrow refers to lower concentrations (4 ng/mL), and the up arrow refers to higher concentrations (20 ng/mL). This figure has been modified from Holmes and Heine10. Please click here to view a larger version of this figure.
Figure 2: Representative brightfield images of 3D protocol. hESC colonies at d0, EBs at d2, aggregates following induction at d14, aggregates of different size and morphology (numbered 1-5) at d28, a single aggregate with unique, identifiable features (indicated by letters a–c) at d28, and visible changes in the same features at d35. Scale bar = 100 µm. This figure has been modified from Holmes and Heine10. Please click here to view a larger version of this figure.
Figure 3: Representative brightfield images of 2D protocol. hESC colonies a d0, EBs at d2, aggregates after induction at d13, after plating aggregates at d14, after maturation at d35 as seen at 5x magnification, and 20x magnification. The white circle in the upper left panel shows the area of differentiated cells among hESC colonies. Scale bar = 50 µm. Please click here to view a larger version of this figure.
Figure 4: Brightfield images showing varying size and complexity in 3D culture. Aggregates at (A) day 8 (hESCs), and (B) d35 (hiPSCs). The latter image is composed of three separate images to show the entire aggregate. Both aggregates may have been affected by merging of smaller aggregates or loss (breaking off) of structures. Scale bar = 200 µm. This figure is republished from Holmes and Heine10. Please click here to view a larger version of this figure.
Figure 5: ICC images of 3D products show relevant markers and structures. At culture d35, 3D products exhibit: PAX6 (green) and TBR2 (red) by lumen of neural rosette-like formation (first row); DCX (green) and NeuN (red) spreading from outer edge ventricular zone (VZ)-like structure (second row); KIRREL2, a marker associated with cerebellar neuroepithelium (third row, left); and ZIC1 a marker associated with cerebellar granule cells (third row, right). The experiment was conducted multiple times using four different hPSC lines: hESC line H01 (n = 5), and iPSC lines hvs88 (n = 4), hvs60 (n = 3), and hvs51 (n = 1). Arrows point to Rhombic lip (RL)-like structure. Scale bar = 100 µm. This figure is republished from Holmes and Heine10. Please click here to view a larger version of this figure.
Figure 6: Large scale ventricular zone-like structure in 3D product. At culture d35, an hESC-derived aggregate is positive for PAX6 (green) and TBR2 (red) commonly associated with early neurons found within ventricular zones (VZs) and subventricular zones (SVZs), in vivo. (Top) An asterisk (*) marks the apical side of a VZ-like region running along the edge of the aggregate, with brackets indicating depth/division of VZ/SVZs. (Middle) Merged signals show scattered sections of PAX6+/TBR2- cells increasing in size toward the upper right end of the VZ. (Bottom) Higher magnification image of the section indicated by a rectangle in the middle panel. Scale bar = 100 µm. This figure has been modified from Holmes and Heine10. Please click here to view a larger version of this figure.
Figure 7: ICC images of 2D products show relevant markers. At culture d35, 2D products exhibit, from hESCs (top row) and hiPSCs (bottom row), cells positive for: granule cell marker ZIC1 and migratory cerebellar neuron marker TAG1 (left column); and neuronal marker neurofilament (NF) and TAG1 (right column). Scale bar = 50 µm. Please click here to view a larger version of this figure.
Figure 8: RT-PCR of 2D products. mRNA expression analysis of hESC line H01 (top row) and hiPSC line hvs60 (bottom row) at the end of the 2D protocol shows products with gel electrophoresis for: granule cell marker ZIC1, granule cell marker ATOH1, migratory cerebellar neuron marker TAG1, Purkinje cell marker Calbindin (CALB), voltage-dependent calcium channel CACNA1A (C-1A), CACNA1E (C-1E), gamma-aminobutyric acid (GABA) B receptor 1 (G-Br1), and housekeeping gene EIF4G2).
Figure 9: Calcium imaging/analysis of 2D products. At culture d35, hPSC differentiation products were recorded for 2 min under the microscope, after incubation with fluor5 dye. At 30 s, cells were electrically stimulated for 10 s at 10 Hz. (A) Still images show hESCs at 0 s (left) and after the start of electrical stimulation at 30 s (right). Arrows indicate region of interest (ROI) for analysis of calcium influx. (B) Graphical analysis showing change in relative fluorescence versus time for ROI (change in fluorescence = (F-F0)/F0, where F0 = (∑F1-n)/n), with neuron-like spikes occurring before, during, and after stimulation. Black bar indicates length of electrical stimulation. Scale bar (white) = 50 µm. Full recording for hESC (as seen here) and an hiPSC line (not shown) are available in Supplementary Video 1 and Supplementary Video 2, respectively. Recordings are in AVI format, at 4x speed. Please click here to view a larger version of this figure.
Supplemental Video 1: Calcium imaging video of 2D product from hESC line H01. At culture d35, differentiation products from hESC line H01 were recorded for 2 min under the microscope, after incubation with fluor5 dye. At 30 s, cells were electrically stimulated for 10 s at 10 Hz. Recordings were made at 2 frames/s, and processed into AVI video at ~7 frames/s, producing a video lasting ~30 s, at ~4x speed. Please click here to download this file.
Supplemental Video 2: Calcium imaging video of 2D product from hiPSC line hvs51. At culture d35, differentiation products from hiPSC line hvs51 were recorded for 2 min under the microscope, after incubation with fluor5 dye. At 30 s, cells were electrically stimulated for 10 s at 10 Hz. Recordings were made at 2 frames/s, and processed into AVI video at ~7 frames/s, producing a video lasting ~30 s, at ~4x speed. Please click here to download this file.
Complexity and costs are relevant factors for stem cell researchers when choosing or developing differentiation protocols. This is especially true as it is an open question of how much external control is required to generate desired cell types, or—to pose it differently—how competent hPSCs are at producing their own developmental environment, if left to themselves with sufficient nutrients. Introduction of extrinsic factors in vitro may very well produce desired cell products, but they could also interfere with the intrinsic developmental capacities cells would have exhibited in vivo. Such considerations are important, particularly if the goal is use of patient-derived iPSCs for disease modeling. Extensive use of patterning and/or growth factors could mask disease phenotypes. The protocols detailed in this report follow the trend of previous studies to reduce complexity, cost, and/or use of extrinsic patterning factors8,9.
Based on results reported by Muguruma et al., and our own recent study, it appears that it is possible to achieve differentiation towards cerebellar fates without concerted efforts to reproduce in vivo conditions, as earlier studies have done1,2,3,4,8,10. The intriguing part is that the two studies used different sets of growth factors, suggesting that neither set were necessary, though both used FGF2. We ran additional tests, where FGFs were selectively excluded from the protocol, and showed that cells were capable of generating the same products without extrinsic FGFs10. Differences between our studies were qualified by the fact that we used different hPSC lines and culture methods, induced neural differentiation with RA, and included components to support granule cell survival and maturation (BDNF, GDNF, SAG, and KCL)11–14. In addition, a less complex startup method, compared to Muguruma et al., was employed. Their protocol began by generating uniform EBs in 96WPs, which isolated them physically and chemically from each other. The protocol here had all PSCs relatively crowded together in 6WPs during EB formation, which allowed them to interact freely. How this may have differentially affected the physical and chemical environment of EBs and later organoids (including intrinsic production of signaling compounds) is unknown, and could be explored. Also, while we show expression of genes associated with—and so suggestive of—cerebellar origin, located within structures morphologically similar to those reported by Muguruma et al., we cannot exclude generation of neuronal-like structures that are of non-cerebellar identity. Future studies, using a large panel of antibodies like those reported by Muguruma et al. (i.e., ATOH1, CALB, etc.) would make such assignments, and comparison between products of both protocols, more conclusive.
Within the 3D protocol, it is important to start with and maintain a sufficient number of cells in culture to ensure sufficient numbers of end products for analysis. Given significant die-off early in the protocol, we recommend starting with more than 500 EBs/well during the first 3 days in culture (Figure 1). This should not be difficult to achieve given colony sizes for hPSCs in feeder-free culture, but might not be as easy for those still using feeder-dependent methods. Given the large number of cells, it is important to watch for color change in medium (indicating pH changes), and accumulation of dead cells. Both must be corrected to prevent collapse of the culture. There may also be clumping of cells and aggregates into massive structures. Although it may still result in aggregates that can be analyzed, product quantity will be greatly reduced, so breaking them up into smaller aggregates with gentle trituration can be useful. However, avoid disturbing normal aggregates, which themselves can grow to large sizes (Figure 4). If aggregates become too sparse, it is recommended to combine wells so that aggregates are not completely isolated. Product variability (in number, size, and morphology) is a well-known issue in 3D cell culture, including for those protocols starting with isolated, uniform EB formation steps, suggesting that a less complex startup procedure (such as the protocol described here) could be more practical8,15. While this heterogeneity is something researchers need to keep in mind, particularly during analysis, the reported protocol generated products consistent with those found in other 3D protocols8,9,15. Based on size and morphology, they fall within the range of neural rosette to cerebral organoid, as described in a recent review by Kelava and Lancaster15, with the most fitting the classification of spheroid. Particularly notable, are the presence of 3D structures suggestive of neural rosettes with lumen, (sub)ventricular zones, and rhombic-lip like features (Figure 5 and Figure 6) as identified by other groups8,15,16,17. Since every experiment produced at least one aggregate with putative VZ/SVZs and cerebellar-associated markers (ZIC1, KIRREL2), those are useful criteria for determining the success of a 3D differentiation using our protocol, with RL-like features providing additional support. Extending the length of culture past 35 days was not tested, but could be pursued to determine the maximum extent of growth, complexity, and maturity allowed by this technique.
The 2D protocol uses the same non-adherent EB formation and neural induction process as the 3D protocol and so the comments above also apply. Once plated, a different set of considerations should be considered. The EBs should adhere quickly for cells to proliferate outward onto the plate. If there are problems with adherence, addition of RI (if not already used), reduced volume of medium, or experimental changes in PLO/LAM concentration may be applied. It is important to keep cells from growing too dense or sparse (preferably grown between 20-80% confluency) in the wells; daily monitoring and timely passaging is important, to avoid over-confluency or floating cells. Unlike the 3D protocol, there should not be significant die-off during culture, though there may be areas of poor growth, or a slowing of proliferation rates. Passaging affects the maturation state of cells (for example, removing cell processes and developed networks between cells) and should be kept in mind when approaching points where cells will be collected or analyzed in some way. For example, for calcium imaging it is very important to passage cells between 2-6 days prior to analysis. Passaging too close to analysis might mean cells have not had time to connect and/or mature, and too far may result in cells overcrowding, making imaging difficult. Although variability between experiments may exist, results are consistent with those reported in initial 2D cerebellar protocols1,2. ICC staining and gene expression analysis corroborate the presence of cells positive for granule cell marker ZIC1, while also identifying markers associated with other neural and cerebellar identities (Figure 7 and Figure 8). Calcium imaging, which involves electrical stimulation of cells incubated with fluor5 dye, indicated functional neuronal activity (Figure 9, Supplemental Figure 1, and Supplemental Figure 2), though it is not confirmed if these were granule cells. It is arguable that by giving cells more time to mature by extending the length of culture past 35 days, the amount of functional neuronal activity should increase. This potential could be explored in the future.
In addition to the lines of research suggested above, it would be of interest to determine differences in product identities (quantity and quality) between the 2D and 3D protocols. The importance of extrinsic FGFs was not tested in the 2D protocol, and it would be useful to know if lack of 3D structure after plating, and so the associated signaling pathways, would make 2D cultures more or less dependent on those early patterning compounds. More stripped-down protocols (e.g., no RA, BDNF, SAG) are equally plausible lines for further investigation. Finally, future studies might benefit from new research tools to better characterize (and assess generation efficiency of) human-specific cerebellar neuronal subtypes.
With the given caveats in mind, both reported protocols may be used for cerebellar differentiations, with products suited to different purposes. They may serve as practical starting points for researchers conducting pilot studies, testing viability of cell lines for such differentiations, or as a basic model for other types of targeted neural differentiation.
The authors have nothing to disclose.
We are grateful to Gerbren Jacobs and Jurjen Broeke for their expert technical assistance, to Prisca Leferink for contributing to the generation and characterization of two control iPSC lines, and to Lisa Gasparotto for demonstrating our procedures.
DMEM/F12+Glutamax | Gibco | 31331-028 | glutamine fortified DMEM/F12 |
Neurobasal medium | Gibco | 21103-049 | neural basic medium |
N2 supplement | Gibco | 17502-048 | |
B27 supplement | Gibco | 17504001 | |
Insulin | Imgen | PT468-B | |
L-glutamine | Gibco | 25030-024 | |
Non-essential Amino Acids (NEAA) | Gibco | 11140-035 | |
beta-mercaptoethanol | Gibco | 21985-023 | |
Poly-L-Ornithine | Sigma | P3655 | |
Poly (2-hydroxyethyl methacrylate) | Sigma | P3932 | aka Poly-Hema |
Laminin | Sigma | L2020 | |
E8 medium and supplement | Gibco | A1517001 | hPSC medium and supplement |
Penicillin/Streptomycin | Gibco | 15140-122 | |
Sodium Chloride | Sigma | S-5886 | |
y-27632 (ROCK inhibitor) | SelleckChem | S1049-10mg | |
DMSO | Sigma | D-2650 | |
Geltrex | Gibco | A1413302 | hPSC-appropriate adherent coating (PAAC) |
0,5M EDTA | Gibco | 15575-020 | |
0.2 um filter | VWR | 28145-77 | |
1.5 mL Eppendorf tube | VWR | 525-0130 | |
DMEM/F12 | Gibco | 21331-020 | |
Ethanol | VWR | 83804360 | |
Parafilm | Sigma | PM996 | wrap for culture plates |
cryotubes | ThermoFisher | 368632 | |
TrypLE | Gibco | 12563-029 | trypsin-based dissociation agent |
Defined Trypsin Inhibitor (DTI) | Gibco | R-007-100 | |
FGF-2 | Peprotech | 100-18B | |
FGF-4 | R&D Systems | 100-31 | |
FGF-8B | Peprotech | 100-25 | |
Retinoic Acid | Sigma | R2625 | |
Brain Derived Neurotrophic Factor | Peprotech | 450-02 | |
Glial Derived Neurotrophic Factor | Peprotech | 450-10 | |
Potassium Chloride | Sigma | P5405 | |
Neurotrophic Factor 3 | Peprotech | 450-03 | |
Smoothened Agonist (SAG) | Cayman | 11914 | CAS 912545-86-9 |
Axiovert 40C microscope | Zeiss | Brightfield imaging microscope | |
Axiocam | Zeiss | Brightfield imaging – image aquisition | |
Eppendorf Centrifuge 5810 | Eppendorf | 521-0996 | centrifuge for cell culture |
PBS (gebufferde natrium oplossing) | Braun Medical | 3623140 | |
5 ml Serological pipets | VWR | 612-4950 | |
10 ml Serological pipets | VWR | 612-4951 | |
6-wells culture plates | VWR | 734-2323 | |
12-wells culture plates | VWR | 734-2324 | |
hESCs | WiCELL | line H01 |