In order to address the question whether SDIA acts on neural progenitors before or after the expression of Sox1, we exploited the Sox1-GFP reporter ES cells in combination with neural differentiation via co-culture with PA6 stromal cells. Sox1-GFP expressing neural progenitor cells were generated and purified by FACS sorting (Fig 1A–B, E). Most experiments were performed at day 7 of differentiation when the numbers of GFP expressing cells peaks (59 ± 6%). FACS sorted GFP^pos and GFP^neg cell populations, together with reconstituted mixed cultures as control cells, were collected and re-plated in parallel. The cells were plated either back on a layer of PA6 stromal cells or on PDL/laminin coated plastic in N2/B27, in the absence of exogenous growth factors (See flow scheme Fig. 1F, G). Following an additional 7 days of differentiation, cultures were examined for the presence of TH expressing neurons by immunocytochemical staining (Fig. 2A–C).

When the GFP^pos and GFP^neg populations were mixed again after sorting and plated back onto PA6 cells, they gave rise to TH positive neurons at a frequency of 27 ± 7%. This frequency is indistinguishable from that of unsorted parallel control cultures (28 ± 4%, Fig. 3A), verifying that the sorting procedure itself had no adverse effect on subsequent dopaminergic differentiation. Therefore, we used unsorted cultures as controls in subsequent experiments. The GFP^neg population, which consists mostly of cells of non-neural lineage and a minority of undifferentiated ES cells, rarely give rise to TH expressing neurons in the subsequent 7 day differentiation culture regardless of the re-plating condition after sorting (data not shown), consistent with previously reported kinetics of dopaminergic neuron differentiation 1015. Interestingly, we obtained a similar proportion of TH expressing neurons from FACS purified Sox1-GFP expressing neural progenitors as compared to unsorted culture after re-plating on PA6 cells or on PDL/laminin (Fig. 2A–C, Fig 3A), suggesting that SDIA acts early during ES cell differentiation and that neural progenitors in this differentiation paradigm are already patterned to generate TH expressing neurons before or at the time of Sox1 expression.

Since the production of neural progenitors during ES cell in vitro differentiation is not synchronized, some of the FACS purified cells might have been produced at an earlier time point which have already committed to neuronal differentiation and lost the competence to respond to SDIA despite the maintenance of GFP expression. We have therefore examined the early born neural progenitors by performing FACS sorting at day 3–4 after plating. This is the earliest time point when Sox1-GFP, nestin, and Sox1 expressing cells can be readily detected. We found that day 3–4 Sox1-GFP^pos neural progenitors produced a similar proportion of TH⁺ neurons as compared to day 7 neural progenitors. (Fig 2D–E, 3C). Thus this data further supports the notion that the competence to generate TH expressing neurons by neural progenitors is closely linked to the onset of Sox1 expression.

To investigate whether the observed early dopaminergic fate specification in Sox1 expressing neural progenitor cells reflects a general mechanism or is specific for the PA6 differentiation method, we examined dopaminergic neuron production in another differentiation paradigm: monolayer differentiation. In contrast to previous reports where few dopaminergic neurons are generated in the absence of Shh and FGF8 918, we found that a substantial number of TH expressing cells are consistently and robustly produced without feeder cells or the addition of extrinsic factors during standard monolayer differentiation in N2/B27 on gelatin-coated plastic (Fig 4 and 5). Under these conditions, very few, if any, TH expressing cells were present at 10 days of differentiation although many β-III-tubulin expressing neurons can be detected at this time (Fig. 4A). After another 4 days of differentiation (a total of 14 days), more neurons (Fig 4B) and many TH expressing cells can be detected (Fig. 4C), corresponding to 16 ± 3% of the number of β-III-tubulin expressing neurons. The TH expressing cells were frequently observed in clusters and all had the morphology of neurons (Fig. 4C). 91 ± 7% of these TH expressing neurons also expressed Nurr1.

To address whether the dopaminergic fate specification also occurred already in the Sox1-GFP expressing neural progenitors produced in monolayer differentiation, GFP expressing cells were sorted using the same criteria as described for PA6 cultures and re-plated on PDL/Laminin coated plastics in N2B27 medium. Cell sorting was performed after 7 days when 81 ± 9% of the cells expressed GFP (Fig. 4D–F). Unsorted cultures processed in parallel gave rise to TH expressing neurons at a frequency of 16 ± 3% (Fig. 4I, 5A). Again, re-mixed FACS purified GFP^pos and GFP^neg populations produced a similar number of TH expressing neurons to the unsorted controls (not shown). The Sox1-GFP expressing population gave rise to TH expressing neurons (Fig. 4G–I) at a frequency indistinguishable from that of unsorted controls (14 ± 2% versus 16 ± 3%) after being re-plated on PDL/laminin (Fig. 5A), suggesting that the ability to differentiate into TH expressing neurons by monolayer-derived progenitors is also specified at the level of Sox1-GFP expression.

Midbrain property is key for functional integration of transplanted dopaminergic neurons 19. Previous studies suggest that ES cell derived dopaminergic neurons are heterogeneous with regard to regional identities and only a proportion of ES cell-derived TH⁺ neurons are characteristic of midbrain dopaminergic neurons 36151820. Therefore we examined the acquisition of midbrain dopaminergic progenitor fate and its relation to Sox1-GFP expression during neuronal differentiation by co-culture with PA6 stromal cells or on PDL/laminin.

We first validated the purity of FACS sorted Sox1-GFP neural progenitors for contamination of undifferentiated ES cells using antibodies against Oct4 and Sox2. Very few (less than 5%) of the Sox1-GFP^pos cells, derived either from PA6 co-culture or from monolayer differentiation, co-expressed Oct4 (Additional file 1). As expected and similar to neural progenitors residing in the proliferative zone of the developing midbrain, more than 90% of the Sox1-GFP expressing cells also expressed Sox2 (Additional file 1).

We found that a proportion of the PA6 co-culture derived neural progenitors express midbrain markers including FoxA2 (Fig. 6A), En1 (Fig. 6B), and Lmx1a (Fig. 6C). In keeping with developmental studies, postmitotic midbrain DA precursor markers such as Nurr1 and Pitx3 were not detected in the Sox1-GFP^pos population. However, after an additional 7 days of neuronal differentiation, either on PDL/Laminin (Fig. 6E,H,K) or maintained in contact with PA6 (Fig. 6D,G,J), these Sox1-GFP^pos progenitors gave rise to cells that expressed Nurr1, Pitx3, and En1 (Fig. 6D,E,G,H,J,K). Pitx3 expressing cells always co-expressed TH (inset in Fig 6E). Nurr1 and En1 were often, but not always, co-expressed with TH (Inset in Fig 6H and 6K). In PA6-derived cultures, the proportion of TH expressing cells that also expressed Nurr 1, En1 or Pitx3 was similar regardless of subsequent differentiation condition after FACS sorting: 82% (PA6-differentiated, n = 102) and 86% (PDL/laminin-differentiated, n = 111) of the TH expressing cells also expressed Nurr1, 20% (PA6-differentiated, n = 98) and 24% (PDL/laminin-differentiated, n = 103) co-expressed En1, and 16% (PA6-differerntiated, n = 100) and 14% (PDL/Laminin-differentiated, n = 99) co-expressed Pitx3.

Similar to the PA6 derived progenitors, Sox1-GFP^pos progenitors obtained by monolayer differentiation co-expressed Sox2 but not Oct4 (Supplemental Fig. 1). The majority of monolayer-derived TH expressing neurons, either produced from FACS purified Sox1-GFP progenitors re-plated on PA6 cells or directly from unsorted control cultures, expressed post mitotic dopaminergic neuron precursor marker Nurr1 (Fig. 6F, M–O). However, very few, if any, displayed a midbrain character at the level of Sox1-progenitor stage as determined by the lack of expression of FoxA2, En1, and Lmx1a. Furthermore, markers normally expressed by midbrain dopaminergic neurons such as En1 and Pitx3 or by diencephalic or olfactory bulb dopamine neurons such as Pax6, Nkx2.1, or Gad67 could not be detected in monolayer derived TH expressing neurons under any differentiation conditions tested (Fig. 6I,L and data not shown). Thus, our data demonstrate that neural progenitors produced from PA6 co-cultures can differentiate into dopaminergic neurons with midbrain characteristics, whilst monolayer differentiation paradigm does not support the acquisition of a midbrain dopaminergic neuron fate.

We noticed that the GFP^pos cells plated on PDL/laminin or matrigel coated plastic in the absence of PA6 cells gave rise to TH expressing neurons at the same frequency as those re-plated on PA6 (Fig. 3B). Furthermore, embryonic mouse fibroblast cells (MEFs), which also exhibit some SDIA activity 10, did not have an effect on the production of TH⁺ neurons (Fig. 3B). Therefore, our data suggest that SDIA does not promote terminal dopaminergic neuronal differentiation. To further address this question, we compared TH⁺ neuron production of FACS purified monolayer derived Sox1-GFP progenitors re-plated either in co-culture with PA6 cells or alone on PDL/laminin. We found that a similar numbers of TH⁺ neurons were generated under both culture conditions independent of the presence of PA6 cells at later stages of differentiation (Fig. 4G,H). 14 ± 2% of the neurons expressed TH when Sox1-GFP progenitors were re-plated on PDL/Laminin (Fig. 5A), whilst a similar level (15 ± 3%) was obtained when progenitors were re-plated in co-culture with PA6 cells (Fig. 5A). Thus, while the overall frequency of dopaminergic neuronal production by monolayer-derived neural progenitors is significantly (p ≤ 0.05) lower than that of PA6-formed progenitors, the exposure to SDIA during later stages of neuronal differentiation has no effect on the production of TH⁺ neurons. Similar results were obtained when monolayer derived Sox1-GFP expressing progenitors were sorted on day 3 of ES cell differentiation and then plated on PA6 cells for the remainder of the differentiation period. Taken together, our data showed that, SDIA does not promote dopaminergic neuron differentiation or maturation after Sox1-GFP expression.

To address whether monolayer-formed Sox1-GFP^pos neural progenitors are able to respond to SHH and FGF8, molecules well known to promote dopaminergic neuron production in both mouse and human ES cells, we added SHH and FGF8, together with FGF2 to cultures after FACS sorting of monolayer-formed progenitors at concentrations previously described to promote dopaminergic specification during ES cell differentiation 921. While continuous differentiation cultures under the same treatment produced significantly more TH⁺ neurons (Fig. 5B, and as reported previously 261115, no increase in TH expressing cells was observed from FACS purified neural progenitors (Fig. 5B) and no midbrain dopaminergic marker expression was induced within the TH expressing population. Therefore, our results suggest that the competence for dopamine neuron differentiation is acquired very early on during mouse ES cell in vitro development and is in-separable from the establishment of a pan-neuroectodermal fate.

Sox1-GFP ES cells were maintained in Glasgow Modified Eagles Medium (GMEM) supplemented with 2-mercaptoethanol, nonessential amino acids, sodium bicarbonate, 10% fetal calf serum and leukemia inhibitory factor (LIF) in gelatinized tissue culture flasks. Differentiation on PA6 stromal cells was carried out as previously described 10. Briefly, ES cells were cultured on a layer of PA6 stromal cells for 7 days in GMEM supplemented with 10% knock-out serum replacement (Gibco) at a density of 60 cells/cm². At day 7 medium was replaced with N2B27 (StemCellSciences) for the remainder of the differentiation period.

Monolayer differentiation was carried out as previously described 89. Briefly, ES cells were plated on gelatinized tissue culture plastics in N2B27 at a density of 10 000 cells/cm². Both types of differentiation cultures were processed for FACS either at day 3–4 or 7 followed by for immunostaining at day 14.

Cells were trypsinized and resuspended in 1% BSA in phosphate buffered saline (PBS) and filtered using a cell strainer (BD Falcon) to ensure single cell suspension. All cell sorting experiments were performed on a MoFlo cell sorter (DAKO). Live cells were gated based on forward scatter and side scatter and/or by 7AAD dye exclusion. Where applicable, the majority of the PA6 feeder cells were excluded based on their differential appearance from that of ES cell derivatives on PE (autoflouresence)/FITC plot. Gates for the two cell populations were set using PA6 cells and differentiated parental E14TG2a ES cells (Figure 1C, D). In some experiments, PA6 cells were excluded based on their inability to exclude 7-AAD. The GFP^pos neural progenitor population and the GFP^neg differentiated ES cell fraction were collected and cell viability at the end of the FACS sorting procedure was determined using trypan blue dye exclusion method. FACS sorted cells were either plated for neuronal differentiation or directly processed for cytospin.

FACS sorted cells were re-plated either on Poly-D-lysine(PDL)/Laminin coated plastics or on PA6 stromal cells, at a density of 100 000 cells/cm² in both cases. Where indicated SHH (400 ng/ml, R&D), FGF8 (100 ng/ml, R&D), and FGF2 (10 ng/ml. R&D) was added to the N2B27 medium. At the end of the culture period, cells were fixed in ice cold 4% paraformaldehyde (PFA) for 15 min at room temperature followed by 3 rinses in PBS.

FACS sorted cells were diluted to a concentration of 1 × 10⁵ cells/ml. 100 μl of cell suspensions of each sample were spun at 1000 rpm for 4 minutes. Cells were immediately fixed in 4% PFA for 15 minutes at room temperature followed by 3 rinses in PBS.

Fixed cells were blocked with 5% normal serum and 0.2% triton X-100 for 1 hour followed by incubation overnight with primary antibodies: rabbit anti-TH (1:1000, Pel Freeze), mouse anti-TH (1:1000, chemicon or 1:500, PelFreeze), mouse anti-β-III-tubulin (1:500, Babco), rabbit anti-Pitx3 (1:500, gift of Dr. M Smidt), mouse anti-En1 (1:250, DSHB), rabbit anti-Nurr1 (1:500, SantaCruz), goat anti-FoxA2 (1:200, santaCruz), rabbit anti-Lmx1a (1:2000, M German), goat anti-Oct4 (1:200, santaCruz), rabbit anti-Sox2 (1:200, chemicon). Cells were washed 3 times in PBS followed by incubation for 1–2 hours with flourescence-labeled secondary antibodies (1:200, Jackson lab) and DAPI (1:1000).

The number of positively stained cells was quantified by counting 18 randomly selected fields per well corresponding to more than 1000 neurons in total. Similarly, the number of TH expressing cells that also co-expressed another marker was quantified by counting the total number of TH expressing cells and the proportion of these cells that also express another marker (Nurr1, Pitx3, En1). 3 wells per experiment were counted and the analysis was repeated 3–4 times for each condition. Analysis of variance (ANOVA) was used to compare means between groups. Differences between groups were determined using the Bonferroni Dunn test, with P-values less than 0.05 considered statistically significant.