3.1. Neuronal Differentiation Using a 5-Stage Embryoid Body Approach
The majority of published neuronal differentiation methods describe selected human embryonic stem cell (hESC) lines such as H9 or I6 and these protocols were optimized around these cell lines. Despite general reproducibility across multiple hESC lines, in patient-specific human iPSCs consistent reproducibility has not been demonstrated, posing a challenge for disease modeling and drug screening. [26
]. One recent publication points towards specific markers such as miR-371-3 and FoxA2 that could predict a priori the differentiation potential of iPSCs or ESCs into the neuronal lineage, which can be relevant for downstream applications [30
Our goal was to develop a reliable protocol reproducible across various patient-specific iPSC lines. We tested a 5-stage protocol for neuronal dopaminergic differentiation that was originally introduced by Lee and Studer in mouse embryonic stem cells [18
] and subsequently further developed [28
]. This protocol involves EB formation for four days, neural rosette formation, isolation of neural rosettes, and expansion and PSA-NCAM enrichment using magnetic bead sorting of neuroprogenitors. A final maturation stage utilizes FGF8 and sonic hedgehog (Shh) for the first ten days followed by BDNF, GDNF, B27, and dcAMP in Neurobasal media for another 20–50 days (). We reason that this 5-stage protocol generating EBs has several advantages in generating neural precursors that can be easily expanded without loss of differentiation potential [31
]. Thus, this protocol is suitable for studying disease-mechanisms at the neuroprogenitor stage and maintaining potential for derivation of other CNS cell types [28
In a control iPSC line, EBs incubated for 4 days in EB media showed a similar result of neural rosette formation as described by Swistowski et al., 2009 [28
] (data not shown). When we attempted to derive NPCs from additional iPSCs derived from controls and patients affected with PD we observed very little neural rosette formation. Furthermore, these rosettes were not expandable as NPCs.
Small molecules have been reported to improve directing ESC/iPSCs into neural lineage [32
]. We tested a combination of small molecules: Dor and SB, both of which have been described for SMAD inhibition. The synergistic mode of action of inhibitors of SMAD signaling, Noggin and SB431542, has been reported to rapidly induce neural conversion of hESCs [15
]. Noggin, a bone morphogenetic protein (BMP) antagonist, and the small molecule Dor have similar activities which selectively inhibit the BMP type I receptors: ALK2, ALK3, and ALK6 and block SMAD1/5/8 phosphorylation [35
]. SB has been shown to be a selective inhibitor of activin receptor-like kinase receptors ALK4, ALK5, and ALK7 [36
For successful generation of NPCs, it is crucial to start with pristine, undifferentiated iPSC cultures. IPSC colonies should be densely packed show low nucleus to cytoplasma ratios and have discrete borders and no differentiation along the peripheries and/or in the centers of the colonies. In this protocol, we found that 2
mm diameter sized colonies yield the best results for neural rosette formation ().
Figure 2 (a) Representative image of the quality of iPSC colonies used to produce EBs: distinct border with little to no differentiation. The recommended size for EB formation should be double the size as the depicted colony, approximately 2mm in diameter. (more ...)
3.2. Combination of Dorsomorphin and SB431542 Improved Neural Induction
IPSC colonies were enzymatically treated with collagenase. After detachment, half of the colonies in a dish were exposed to 5μ
M SB. The other half was left untreated. The colonies were then cultured for 4 days in EB media with or without Dor/SB. EBs cultured in EB media alone showed loose, less compact, and irregular shapes () while the majority of EBs treated with Dor/SB demonstrated compact, solid and round shaped aggregates and had an average size of 350μ
m in diameter (). On day 4, media was changed to NIM media containing N2 media which was freshly made of different individual components. None of the commercially available N2 supplements showed consistent results (data not shown). EBs were then plated onto Geltrex-coated culture dishes on day 6. During days 6–10, neural rosettes were detected by their characteristic morphology of radially arranged cells (Figures (A)–(D)). In the early stages of NIM incubation (approximately days 8–10), neural rosettes showed darker centers of “flower-shaped” structures with indiscrete boundary lines (Figures (A) and (B)). In the latter incubation with NIM, “flower-shaped” morphologies were more distinct and edges more clearly defined, shown in Figures (C) and (D). Dissected rosettes (Figures (E) and (F)) that are replated and manually isolated a second time generate NPC populations of higher purity.
We evaluated neural differentiation of EBs on day 10 () via immunocytochemistry. Neural markers Pax6 and Sox1 were used as well as the pluripotent cell marker Oct4. Pax6 and Sox1 showed positive staining in attached EB (Figures (G) and (H)), however, Oct4 showed no immunoreactivity (data not shown). At the same time point, the percentage of neural rosettes formed with and without addition of Dor/SB was quantified by manually counting the colonies containing neural rosettes divided by total colonies attached on the culture dish (Figures and (I)). Without Dor/SB, we observed low rosette formation between 0% and 31.9%, and we were not able to derive expandable NPCs. The combination of Dor/SB, on the other hand, increased the neural rosette formation substantially to 48% to 97.5% of EBs in both control and PD-specific cell lines. Overall, we did not notice a difference in the efficiency of rosette formation between PD lines and control lines.
At day 6, we performed gene expression analysis of multiple markers in attached EBs. In all six lines we studied neuroectodermal markers Sox1 and Nestin, mesodermal marker Brachyury, endodermal marker GATA4 and pluripotent marker Oct4. Surprisingly, there was a striking difference of >150-fold in the gene expressions of neuroectodermal markers Sox1 and Nestin in Dor-/SB-treated EBs compared to EBs without small molecules (). This suggests that the two small molecules very efficiently modulate the SMAD signaling pathway leading to this enormous increase in neuroectodermal markers. This increase was consistent in all six iPSC lines tested, and differences in neuronal differentiation were not observed between patient and control lines. Endo and mesodermal markers GATA4 and Brachyury as well as pluripotency marker Oct 4 were all lower compared to the untreated, normalized NPC lines.
Figure 4 Quantitative gene expression analysis of PD and control lines with and without Dor/SB. Expression levels of neuroectoderm (Sox1 and Nestin), mesoderm (Brachyury), endoderm (GATA4), and pluripotent markers (Oct4) were assessed by quantitative PCR. The (more ...)
Neural rosettes were manually cut and replated as pieces to produce a population of NPCs of higher purity. Rosettes were manually isolated once again, collected, enzymatically treated with Accutase, and plated and expanded in NPC media. Manual passaging and expansion of NPCs still yielded approximately 10% undifferentiated Oct4-positive cells in NPC cultures, which upon further expansion showed iPSC morphology (data not shown). Therefore, we used magnetic bead sorting with a neural cell adhesion molecule antibody against polysialic acid neural cell adhesion molecule (PSA-NCAM or CD56) (Figures (A) and (B)). We observed an approximately 20% cell loss after magnetic bead sorting. We characterized NPCs after sorting by immunocytochemistry with defined markers Nestin and Sox1. We detected >90% Nestin and Sox1 immunoreactive NPCs in all iPSC cell lines taken through this protocol (Figures (C) and (D)). NPCs were readily expandable at a passaging ratio of 1
2 to 1
3 with Accutase. Cultures grew well when media was prepared freshly every 2 to 3 days and B27 added freshly to NPC media, before media changes. We expanded NPC cultures for >15 passages after derivation and did not observe any changes in morphology or expression of Nestin and Sox1.
Figure 5 Characterization of NPCs. (A) and (B) NPC morphology was observed under phase-contrast microscopy with 5x (A and 10x (B) magnification. (C and D) NPCs expressed Nestin (green) (C) and Sox1 (red). (D) Nuclei were counterstained with DAPI (blue). Scale (more ...)
With this new approach for neural induction using small molecules, we have dramatically increased reproducibility and efficiency of neural rosette stage/NPC generation. This is invaluable when using patient-derived iPSCs for disease modeling, which may have an intrinsic disadvantage in culture when carrying potential disease-related deficiencies. Since NPCs can be easily expanded, this could become a suitable cell type for high throughput screening where a very large number of starting material is needed.
3.3. Substitution of Small Molecules Purmorphamine or Smoothened Agonist for Sonic Hedgehog Had Similar Effects on Neuronal Maturation
We investigated the substitution of sonic hedgehog (Shh) for small molecules purmorphamine (Pur) or smoothened agonist (SAG) during dopaminergic maturation. These chemicals that are considerably less expensive, have minimal lot-to-lot variabilities, and have longer shelf-life compared to recombinant proteins.
For final dopaminergic maturation, we used a 2-step approach. For the first ten days, we cultured NPCs in FGF8 and tested two small molecules SAG and Pur as substitutes for Shh in control and patient cell lines. At day 1 in DA1 media, the plating density of the NPCs should be approximately 60% to 70% ((A)). During this 10-day protocol, cells were split at 100% confluency using Accutase and replated at a cell density of approximately 80%. When cells were plated at a lower cell density (<50%), we observed remarkable cell death and low rates of cell attachment. After ten days, we switched to Neurobasal media supplemented with BDNF, GDNF, dcAMP, and B27 every second day, but added B27 daily preventing cell death. Cells were split until they began growing out processes ((B)). After day 30 of dopaminergic maturation, cells were fixed, immunostained with TUJ1 and TH, and counterstained with DAPI (Figures (C)–(F)).
To measure the efficiency of the neuronal differentiation, we evaluated the percentage of TH and TUJ1 expressing neurons relative to total cells using two approaches: stereology with systematic random sampling and flow cytometry. Flow cytometry was employed to minimize bias. The challenges of accurate counting of these cultures are the dense “patches” of neurons and the majority of TH immunoreactive neurons localized in these “patches” [37
In the scatter plots for flow cytometry, (Figures (G)–(J)), undifferentiated NPCs did not show immunoreactivity for TUJ1 and TH ((H)) and had a similar pattern in the scatter plot to unstained NPCs ((G)). Differentiated neurons were immunoreactive for TUJ1 and TH ((I)) and were compared to the total number of unstained differentiated neurons ((J)).
Both approaches, stereology and flow cytometry, showed no significant differences among the three different components Shh, SAG, or Pur used in dopaminergic differentiation in terms of the ratio of TUJ1/total, TH/TuJ1, and TH/total cells ((K)). Data from flow cytometry was slightly lower than those from the stereological approach. We suspect that we lost neurons during the handling process such as dissociation and passaging through a cell strainer to filter clumps from cell suspension before flow cytometry was performed.
Some studies have shown that with an extension of culturing time by up to 60 days, more neurons convert to TH-positive as well as become electrophysiologically mature [37
]. Other studies showed a higher percentage of TH-positive neurons, however, different quantification approaches may have introduced bias toward a higher percentage of neuronal yields.