From c-Photina ® Mouse Embryonic Stem Cells to High-Throughput Screening of Differentiated Neural Cells via an Intermediate Step Enriched in Neural Precursor Cells

Abstract

The use of engineered mouse embryonic stem (mES) cells in high-throughput screening (HTS) can offer new opportunities for studying complex targets in their native environment, increasing the probability of discovering more meaningful hits. The authors have generated and developed a mouse embryonic stem cell line called c-Photina^® mES stably expressing a Ca²⁺-activated photoprotein as a reporter gene. This reporter cell line retains the ability to differentiate into any cell lineage and can be used for miniaturized screening processes in 384-well microplates. The c-Photina^® mES cell line is particularly well suited for the study of the pharmacological modulation of target genes that induce Ca²⁺ mobilization. The authors differentiated this mES reporter cell line into neuronal cells and screened the LOPAC¹²⁸⁰™ library monitoring the agonistic or antagonistic activities of compounds. They also demonstrate the possibility to generate and freeze bulk preparations of cells at an intermediate stage of differentiation and enriched in neural precursor cells, which retain the ability to form fully functional neural networks once thawed. The proposed cell model is of high value for HTS purposes because it offers a more physiological environment to the targets of interest and the possibility of using frozen batches of neural precursor cells.

Keywords

flash luminescence HTS mouse embryonic stem cells neurons photoprotein

Introduction

Ca²⁺ ions are critical for many cellular functions, including activation of signaling pathways, changes in cell contraction and motility, or secretion of hormones and other factors. The intracellular concentration of Ca²⁺ is highly regulated and compartmentalized, and many of the most interesting target classes for the life sciences industry, such as G-protein-coupled receptors (GPCRs), ion channels, and transporters, use Ca²⁺ mobilization to transduce their signal upon activation by specific ligands.¹ Measurement of intracellular Ca²⁺ concentrations requires sensitive and noninvasive analytical methods. One sensitive and noninvasive way to measure Ca²⁺ mobilization involves photoproteins, which release photons upon Ca²⁺ binding in the presence of the cofactor coelenterazine.² Photoproteins are widely used in cell-based assays for high-throughput screening (HTS) as reporter genes to monitor Ca²⁺ movements associated with signals.^2-6 Relative to Ca²⁺-sensitive fluorescent dyes, photoproteins possess several important advantages: high sensitivity, virtually undetectable background, and correspondingly high signal-to-noise ratio as well as small assay volumes.^2,7,8

Cell-based assays often use transformed cell lines, which express both a photoprotein and a target receptor. The parental cells used for overexpressing the gene of interest are usually selected because of the limited expression of interfering receptors, their ease of culture, and rapid growth. However, exogenous expression of targets and the transformed environment can create artifacts of gene dosage, toxicity, or stoichiometry of the receptor target itself, particularly when it requires assembly of multiple subunits. Primary cells are a good alternative to transformed cell lines because they express the gene of interest endogenously in a more physiological environment, but they are often difficult to culture in sufficient number and sometimes very tricky to transfect.

Embryonic stem cells could represent a valid alternative. By retaining the self-renewal property of the undifferentiated state, they can be cultured and expanded in vitro for long periods and are easily transfected.⁹ Moreover, embryonic stem cells can differentiate into virtually any cell type resembling primary cells.^10,11 Accordingly, they offer a natural environment for the receptor targets, and they can form complex targets (like multi-subunit ion channels), with the correct stoichiometry and natively regulated.¹²

In our approach to the problem, we generate clones of mouse embryonic stem (mES) cells expressing a photoprotein as a Ca²⁺ reporter system under the control of a ubiquitous promoter.¹³ We adapted a neuronal differentiation protocol to suit HTS conditions to set up robust cell-based assays.^13,14

We describe the preparation of bulk frozen cells at an intermediate stage of neural differentiation and enriched in neural precursor cells (NPCs). We demonstrate that neural cells derived from thawed NPCs maintain the same characteristics, in terms of neuronal subtype composition and functional responses, as the neural cells directly derived from undifferentiated mES cells, opening a new opportunity for HTS applications.

Materials And Methods

c-Photina^® mES cell culture

c-Photina^® TBV2 (129S2/SvPas) mES cells^13,15 were cultured in the undifferentiated state on a monolayer of mitomycin C–treated mouse embryonic fibroblasts in the presence of leukemia-inhibiting factor (LIF) (Chemicon, Temecula, CA).

Neuronal differentiation protocol

The mES reporter cells were seeded at 1500 to 3000 cells/cm² on gelatin-coated dishes in KnockOut™–Dulbecco’s modified Eagle’s medium (DMEM) with 15% KnockOut™ Serum Replacement (KSR; Invitrogen, Carlsbad, CA) for 13 to 21 days with medium changes every 2 to 3 days, as described in Fico et al.¹⁴

On day 7, the cells were detached using 0.05% trypsin/EDTA solution and counted. This step was defined as the replating step. At this stage, cells could be used either for immediate replating at 5000 cells/well in gelatin-coated 384-well plates to continue the neuronal differentiation protocol or frozen in 90% fetal bovine serum (FBS) + 10% dimethyl sulfoxide (DMSO) to obtain a homogeneous population of NPCs.

Frozen NPCs were thawed and immediately plated at 5000 cells/well in gelatin-coated 384-well plates. Two days after thawing and plating, the medium was changed using multichannel automatic pipettes, and this procedure was repeated every 2 to 3 days until the end of the differentiation protocol.

The functional test was performed on differentiated cells using the FLIPR^TETRA® reader in the luminescent mode. Cells were incubated for 4 h with 40 µL/well of Tyrode’s buffer (130 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 5 mM NaHCO₃, and 20 mM HEPES, pH 7.4, 2 mM Ca²⁺) and 10 µM coelenterazine in the dark at 37 °C in a humidified atmosphere with 5% CO₂. Agonists were injected at a 5× concentration.

RT-PCR analysis

Total RNA was isolated by TRIzol^® (Gibco/BRL, Gaithersburg, MD). Reverse transcription–PCR (RT-PCR) was performed with the Invitrogen SuperScript^® II RT-PCR kit (Invitrogen), as recommended by the manufacturer. The primer lists and the PCR conditions are described in Fico et al.¹⁴

Immunofluorescence analysis

The cells were fixed with 4% paraformaldehyde (PFA, MERCK, Whitehouse Station, NJ) solution and simultaneously blocked and permeabilized with 10% normal goat serum (Chemicon)/0.2% Triton X-100 in 1× phosphate-buffered saline (PBS) directly in 384-well plates.

All antibodies were incubated in 10% normal goat serum 0.1% Triton X-100 in 1× PBS.

The primary antibodies used were as follows: mouse anti–beta III tubulin (Chemicon), rabbit anti–glial fibrillary acidic protein (GFAP) (Dako, Glostrup, Denmark), mouse antinestin (Rat-401) (Chemicon), and mouse anti-neuN (Chemicon).

The secondary antibodies used were the antimouse Alexa 488 and/or antirabbit Alexa 546. Hoechst 33342 dye (Invitrogen) (2 µg/mL final concentration) was incubated for 5 min at room temperature.

Images were acquired by using GE Healthcare’s IN Cell Analyzer 1000 System (GE Healthcare, Piscataway, NJ). The IN Cell Workstation v3.5 was used for automated analysis of cellular images. After the optimization of segmentation parameters, the number of Hoechst-positive cells was calculated as well as the number of GFAP and/or neuN or beta III tubulin-stained cells.

Cells were classified using the neuN intensity in the nuclei: cells with a neuN intensity greater than or equal to the threshold were classified as “neuN+” cells, whereas cells with a neuN intensity less than the threshold were classified as “neuN–” cells. The same procedure was followed to distinguish the cells based on beta III tubulin staining in the cytoplasm. With double staining, only the neuronal marker-negative cells were further filtered using GFAP intensity in the cytoplasm. All experiments were performed in duplicate, and 10 random areas were selected and analyzed for each well.

LOPAC¹²⁸⁰™ screening

All the compounds of the LOPAC¹²⁸⁰™ library were reformatted in 384-well plates and diluted to 50 µM in Tyrode’s buffer plus 2.5% DMSO (5× concentrated). All control wells were treated with DMSO at same concentration as assay wells.

Neural-differentiated mES reporter cells were seeded in triplicate at a concentration of 5000 cells/well 7 days before the test (conducted at day 14 of differentiation) in gelatin-coated 384-well plates. Prior to the test, all the samples were incubated in Tyrode’s buffer containing 10 µM coelenterazine for 4 h.

The LOPAC¹²⁸⁰™ library was injected onto cell plates at 10 µM final concentration. After 5 min, the quisqualate compound was injected at 10 µM final concentration. The light release, expressed as RLU (relative luminescence units), was measured for a total of 350 s. The test was performed on the FLIPR^TETRA® instrument in the luminescent mode.

Data analysis

For the bulk processing of data, an implemented version of Spotfire^® Decision Site 9.1 was used. The analysis was performed using both an agonist and an antagonist screen approach, considering the maximum RLU values.

First, the Z′ values were calculated for both approaches, computing the means and the standard deviations of the control samples after removal of all the outlier points (defined as those with a modified |z-score| >3.5). The Z′ was computed as follows: Z′ = 1 − 3(s _max − s _min)/| x̄ _max − x̄ _min| where s _max and s _min are standard deviations of positive and negative controls, and x̄ _max and x̄ _min are means of positive and negative controls, respectively. Control samples were defined in agonist screen mode as “max” signals and “min” signals, which correspond respectively to 100 µM glutamate (n = 8) and Tyrode’s + 0.5% DMSO solution (n = 8). While in antagonist screen mode, “max” signals and “min” signals were, respectively, the following: Tyrode’s buffer (in first injection) plus 10 µM quisqualate (in second injection) (n = 8) and 100 µM JNJ16259685 (in first injection) plus 10 µM quisqualate (in second injection) (n = 8).

Then, the “percent activity mean” was calculated for the agonist screen approach (after triplicate outliers removal, defined as those with |z-score| >3.5), relative to the mean max signal (100 µM glutamate) for each plate corresponding to 100% activity, as well as the mean min signal (Tyrode’s DMSO 0.5% solution) for each plate corresponding to 0% activity, with the following formula: 100 * [(response value in first injection − x̄ _min)/ x̄ _max − x̄ _min].

For the antagonist screen approach (after triplicate outliers removal, defined as those with |z-score| >3.5), the “percent inhibition mean” was computed relative to the mean compound responses for each plate corresponding to 0% inhibition (x̄ _{response values in second injection}) and the mean min signal response (100 µM JNJ16259685 and 10 µM quisqualate) for each plate corresponding to 100% inhibition, with the following formula:

100 * [1 - (response value in second injection - {\bar{x}}_{\min}) / {\bar{x}}_{response values in second injection} - µ_{\min}] .

Active compounds in both approaches were defined as those samples that were 3 standard deviations above the mean percent activity or inhibition assuming that the values were normally distributed.

FM4-64 uptake protocol

Neurons differentiated from c-Photina^® NPCs (day 14) were stimulated for 1 min with Krebs’-Ringer’s-HEPES (KRH) (in mM: 130 NaCl, 5 KCl, 1.2 MgSO₄, 1.2 KH₂PO₄, 2 CaCl₂, 6 glucose, and 25 HEPES/Na, pH 7.4) supplemented with 50 mM KCl, in the presence of 10 µM FM4-64 and 1 µM tetrodotoxin (TTX). After FM4-64 loading, neurons were washed 4 times for a total of 8 min with warmed KRH at 37 °C supplemented with 1 µM TTX and 10 µM CNQX (6-cyano-7-nitroquinoxaline-2,3-dione).

Video microscopy and image analysis

Specimens were observed with a Zeiss (Oberkochen, Germany) Axiovert 135 TV inverted microscope equipped with epifluorescence optics.

Multielectrode array (MEA)–based electrical recordings

Electrical recordings were performed in the neural culturing medium at 37 °C using the Multi-Channel System apparatus with the MEA60 200 Pt GND chip (Ayanda Biosystems SA, Lausanne, Switzerland). Data recorded at 25 kHz/ch from the 60 channels were then filtered from 200 Hz to 3 kHz, and spikes were sorted using a threshold algorithm. The threshold was defined as a multiple of the standard deviation of the biological noise computed during the first 100 ms of the recording (−7 × SD noise). Data were further analyzed with the Multi-Channel System software (MC_rack).

The cells were incubated with 1 µM TTX for 5 min, and then the toxin was washed out to study the recovery. The electrical activity was monitored for 180 s during and post TTX washout.

NBQX (2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione) AMPA receptor inhibition was performed at the final concentration of 100 µM. The electrical activity was monitored for 250 s before and after NBQX addition.

Results

Neuronal differentiation

Different approaches have already been described to induce neural differentiation of mES cells through the serum-free KSR-supplemented medium.^16,17

We selected the monolayer-based method described in Fico et al.¹⁴ and optimized it for HTS purposes. In the original protocol,¹⁴ cells were not dissociated, and differentiation took place directly in the plates chosen for initial cell seeding. We first adapted the neural differentiation protocol to the 384-well plate format: neural cells were forming over time, but we observed marked well-to-well differences in the growth rate with a consequential variation in cellular composition and cell numbers at the end of the differentiation period. To overcome these limitations, we decided to introduce a replating step at day 7 of differentiation ( Fig. 1A ). Cells are harvested, pooled, counted, and plated in a 384-well plate format to reach the final differentiation step ( Fig. 1A ). This modified protocol allows the preparation of bulk cells in ease to manage plate formats (e.g., 100-mm Petri dishes) and also reduces the well-to-well variability in terms of growth rate and cellular composition that we experienced with the initial protocol.

Fig. 1.

(A) Neural differentiation protocol scheme. The protocol for neuronal differentiation was adapted to miniaturized formats, includes a replating step at day 7, and lasts a minimum of 13 days. At day 7, the neural precursor-enriched cells (NPCs) can be either plated in 384-well plates and used for functional testing or frozen in bulk and used later. (B) Expression profile analysis. Experiment performed by RT-PCR during neural differentiation (from day 0 to day 13), with marker genes for undifferentiated mES cells (nanog), neural precursors (nestin), neurons (NF-L, neurofilament-light protein), radial glia (BLBP, brain lipid-binding protein), and astrocytes (GFAP, glial fibrillary acidic protein); different neuronal subtypes such as GABAergic (GAD65, glutamic acid decarboxylase 65), dopaminergic (TH, tyrosine hydroxylase), serotoninergic (TPH1, tryptophan hydroxylase 1), and motoneuronal (HB9, homeobox gene 9) neurons; cells derived from different germ layers such as the endodermal precursors (sox17, SRY (sex determining region Y)-box 17), endo/mesodermal precursors (GATA-4, GATA-binding factor 4), and mesodermal cells (αMHC, alpha myosin heavy chain); as a control, the ubiquitously expressed hypoxanthine phosphoriboxyltransferase (HPRT) marker gene was used. (C, D) IN Cell Analyzer immunofluorescence-based quantitative analysis. IN Cell algorithm selected and circled the positive stained cells (left figures) with the specific antibodies for (C) beta III tubulin (green) and GFAP (red) and (D) neuN (green) and counterstained with the Hoechst 33342 dye (blue). The (C) beta III tubulin and GFAP-positive cells or the (D) neuN neuronal-positive cells were counted and the relative percentages are reported in the graphs on the right. Scale bars, 20 µm.

Furthermore, the pool of harvested cells can be frozen and thawed upon need at any time, acting similarly to a reagent for HTS ( Fig. 1A ).

c-Photina^® mES cells were plated as described in Fico et al.¹⁴ for 1 week and then dissociated and replated in gelatin-coated 384-well plates at 5000 cells/well. In this way, we observed the presence of cells showing a neural-like morphology. To confirm that those cells were indeed neural cells, we analyzed the expression of markers for undifferentiated and differentiated mES cells at various time points by RT-PCR ( Fig. 1B ). The results confirmed a decrease in the expression of nanog, a marker of undifferentiated mouse ES cells, during differentiation, with a contemporary activation of the expression of markers for neuronal and/or glial cells such as nestin, NF-L (neurofilament-light protein), BLBP (brain lipid-binding protein), and GFAP (glial fibrillary acidic protein). As shown in Figure 1B , the nestin level is higher between day 6 and 8 than on the other days, suggesting an enrichment in neural precursors cells in those days.

The neuronal subtype composition of the differentiated neural cells and the presence of contaminant cells from different germ layers were checked by RT-PCR analysis. The results showed the expression of GABAergic (GAD65, glutamic acid decarboxylase 65), dopaminergic (TH, tyrosine hydroxylase), and serotoninergic markers (TPH1, tryptophan hydroxylase 1) and a small amount of the motoneuronal marker (HB9, homeobox gene 9), especially after 8 days of differentiation (see Fig. 1B ). The presence of cells originating from different germ layers was restricted to day 13 and found to be of an endo-mesodermic cell origin (GATA-4, GATA-binding factor 4; see Fig. 1B ). These results are consistent with the data obtained in the original protocol.¹⁴

To validate the method and the reproducibility of the results, we set up a quantification protocol using an immunofluorescence approach ( Figs. 1C,D ).

Differentiated cells were immunostained directly in 384 MTP format, with specific markers for neurons (beta III tubulin, Fig. 1C ; neuN, Fig. 1D ), for astrocytes (GFAP), and with the Hoechst dye for nuclei. The images were acquired with an automated acquisition platform, the IN Cell Analyzer 1000 instrument, and then analyzed with the IN Cell Investigator Software to extract the percentage of neuronal marker-positive and GFAP-positive cells from the images. The IN Cell algorithm identified and counted all the Hoechst-stained nuclei present in 10 random areas in the well and defined as “positively stained cells” all the cells in which there was a signal overlay between the Hoechst and the antibody staining. The neuronal network in some cases was so intricate that it interfered with the exact identification of beta III tubulin–positive cells. For this reason, the usage of the neuN nuclear marker was preferred ( Figs. 1C,D ). With this quantification approach, we demonstrated that in our population of differentiated cells, around 45% to 60% of the cells originate from the neural lineage ( Figs. 1C,D ). Nestin-positive cells that were still observed on day 13 of differentiation were not included in this calculation (see also Cainarca et al.¹³).

Screening of a library of pharmacologically active compounds

We have already shown that c-Photina^® mES cells differentiated into neural cells were suitable for HTS purposes.¹³ As proof of principle, we screened the neuronal differentiated and undifferentiated mES -reporter cells with an unbiased Library Of Pharmacologically Active Compounds (LOPAC¹²⁸⁰™).¹³ During this screen, we observed that both cell types were stimulated by several compounds capable of stimulating endogenous receptors/channels and induced Ca²⁺ flux. In particular, we detected a substantial presence of glutamate receptors.¹³ To investigate which class of glutamate receptor was the most represented, we designed a compound master plate containing specific agonists for the different subclasses of glutamate receptors. Using this approach, we observed the predominant presence of group I metabotropic glutamate receptors ( Fig. 2A ).

Fig. 2.

LOPAC¹²⁸⁰™ screening of neural cells derived from c-Photina^® mES cells at day 14 of differentiation. (A) Dose-response curves of neural cells derived from c-Photina^® mES cells at day 13 of differentiation stimulated with different agonists for different glutamate receptors: glutamate (generic agonist), quisqualate and DHPG (group I metabotropic glutamate receptor-specific), SYM 2081, and kainate (kainate receptor-specific agonists). Curves were generated with GraphPad Prism 4 software (nonlinear regression analysis). Data are shown using mean ± SEM (n = 4). (B) Quisqualate dose-response curves and EC₅₀ calculation on neural cells derived from c-Photina^® mES cells at day 13 of differentiation (generated with GraphPad Prism 4 software [nonlinear regression analysis]). Data are shown using mean ± SEM (n = 4). (C) Example of control compound injections in agonist and antagonist screening mode. (D) Schematic representation of the results of the LOPAC¹²⁸⁰™ screening in antagonist mode. The lower part is the Spotfire^® graphical analysis of all the results. The selected compounds (indicated with black circles) were those with a percent inhibition mean higher than 81% (defined as hits in the agonist screen mode). Those with a percent inhibition mean lower than 81% are indicated with gray circles. All compounds with a percent activity mean in agonist mode higher than 7% (defined as hits in the agonist screen mode) are indicated with star symbols. (E) Z′ values for the antagonist screen approach of the 4 LOPAC plates tested in triplicate (total of 12 plates). (F) Graphical distribution of the percent inhibition mean of the compounds. The black bars are those hit compounds with percent inhibition mean values higher than 81%.

The Ca²⁺-mediated high luminescence response obtained after stimulation with quisqualate suggested the possibility of using this compound in an antagonist screen approach to look for mGluR inhibitors ( Figs. 2A,B ). The effects of quisqualate were determined in a dose-response experiment ( Fig. 2B ).

Notably, quisqualate activates both the mGluR1 and mGluR5 metabotropic receptors, both present in neural cells. From literature data, it was shown that the EC₅₀ values of quisqualate for rat mGluR5 are 20 times lower than for rat mGluR1 (in µM range) when these receptors are transfected in Chinese hamster ovary (CHO) cells.¹⁸ It is also known that there is a different expression pattern of these receptor subtypes in brain regions, with a predominant presence of mGlur5 receptors, for example, in rat cortex and hippocampus areas.¹⁹

For the group I mGluR antagonist screen, we decided to use again the LOPAC¹²⁸⁰™ library because it is a validation library containing compounds active on all the major pharmacological target classes, including group I mGluRs. The screen was performed in triplicate on c-Photina^® mES cells, differentiated to neural cells, at day 14 of differentiation. We injected the cells first with the LOPAC¹²⁸⁰™ library and after 5 min with quisqualate to look for group I mGluRs inhibitors. Data were analyzed for each addition, to study both the agonist and antagonist screen approaches.

Results are expressed as “percent activity” in agonist screen mode, with respect to 100 µM glutamate (100% activity) and Tyrode’s solution (0% activity) (see Materials and Methods). To determine the hit cutoff threshold for the agonist approach, the mean and the standard deviation of the percent activity of all samples tested were estimated to be −0.580 and 2.58, respectively. Hits were defined as those samples with a percent activity mean of the triplicates equal to 7% or greater (3 standard deviations above the percent activity mean; Suppl. Table S1, upper part).

The Ca²⁺-mediated signals obtained confirmed the presence of glutamate receptors together with purinergic and adrenergic receptors (Suppl. Table S1, upper part), which are receptor families highly expressed by neural cells. These results were consistent and similar to those shown in the previous screen performed in our laboratories in which the cells were tested in quadruplicate at day 13 of differentiation.¹³

For the antagonist screen approach, the results are expressed as percent inhibition, with respect to the mean response per plate in the second injection of the tested samples stimulated with 10 µM quisqualate (0% inhibition) and the mean response per plate in the second injection of 100 µM JNJ16259685, a group I mGluR antagonist, plus 10 µM quisqualate (100% inhibition; Fig. 2C and Suppl. Table S1, lower part).

To determine the hit cutoff threshold for the quisqualate antagonist screen, the mean and the standard deviation of the percent inhibition for all samples tested were estimated to be −0.53 and 27.41, respectively. The hits were defined as those samples with a percent inhibition mean of the triplicates equal to 81% or greater (3 standard deviations above the percent inhibition mean; Fig. 2D and Suppl. Table S1, lower part).

The screening quality parameter Z′ was evaluated for every plate both in agonist and antagonist mode. An overview of the HTS performance in antagonist screen mode for Z′ is presented in Figure 2E . The mean Z′ for the overall group I mGluR antagonist screening campaign is 0.54 (with a standard deviation of 0.14). The graphical distribution of the compound’s percent inhibition mean is shown in Figure 2F .

Of the 8 compounds scoring as putative quisqualate antagonists (in italics in Suppl. Table S1, lower part), 3 were also present among the putative activators list (Suppl. Table S1, upper part). These were quisqualate itself and DHPG (3,5-dihydroxyphenylglycine), both agonists of group I mGluR receptors, and loxiprofen. The consistent reduction of the luminescent response after the second addition of quisqualate could be due to a desensitization of the group I mGluR receptors (only for the first 2 compounds) and/or to a photoprotein consumption induced by the first injection of the agonists. The role of loxiprofen should be further investigated and confirmed.

In the LOPAC¹²⁸⁰™ library, 5 antagonists for group I metabotropic glutamate receptors are present: 3 specific for the mGluR5 and 2 for the mGluR1 subtypes. We found 3/3 antagonists for mGluR5 and none for mGluR1 (in bold in Suppl. Table S1, lower part, and Fig. 2D ). This could be due to the different expression level of mGluR5 receptors present in the population of neural-differentiated mES cells.

The data obtained in both agonist and antagonist mode provide additional evidence for the robust and appropriate differentiation of the c-Photina^® mES cells into neural cells and their suitability for HTS purposes.

c-Photina^® neural precursor cells

To further characterize our differentiation protocol, we investigated if the cells frozen during the intermediate step of replating, named NPCs, could be used as a source of cells already neural committed and ready to use. These cells can be thawed directly in gelatin-coated 384-well plates, and the differentiation can be completed in this format.

First, we verified that the freeze-thaw procedure did not interfere with correct neural development. We compared the neural populations obtained from freshly differentiated mES reporter cells and from thawed c-Photina^® NPCs using RT-PCR, immunofluorescence, and functional analysis. We measured Ca²⁺-mediated luminescence in a miniaturized screening format (384-well plate) after stimulation with compounds targeting receptors highly expressed in neural cells.

We demonstrated that c-Photina^® NPCs retain the same characteristics as c-Photina^® mES cells in terms of neural differentiation efficiency, following the presence of markers both at RNA (see RT-PCR expression panel in Fig. 3A ) and protein level ( Fig. 3B ). The relative percentage of neuN-positive (neurons) and GFAP-positive (glial cells) cells ( Fig. 3C ) is the same in the 2 cell populations and is around 60% neural cells. During these comparative experiments, we noticed that neural cells derived from c-Photina^® NPCs seem to differentiate with a small initial time lag in comparison to the neural cells derived from freshly c-Photina^® mES cells. We believe this slight delay could be due to the freeze-thaw procedure that could initially slow down the differentiation process. Further experiments to clarify this point are ongoing.

Fig. 3.

Comparison between neural cells differentiated directly from mES cells and from frozen neural precursor cells (NPCs) at different days of differentiation. (A) RT-PCR analysis during neural differentiation (from days 10-13 for mES cells and from days 8-17 for NPCs) with marker genes for undifferentiated mES cells (nanog), neural precursors (nestin), neurons (NF-L, neurofilament-light protein), glia (BLBP, brain lipid-binding protein; GFAP, glial fibrillary acidic protein), and GABAergic (GAD65, glutamic acid decarboxylase 65), dopaminergic (TH, tyrosine hydroxylase), serotoninergic (TPH1, tryptophan hydroxylase 1), and motoneuronal (HB9, homeobox gene 9) neurons; for cells of different germ layers such as the endodermal precursor sox17, endo/mesodermal precursor GATA-4 (GATA-binding factor 4), and mesodermal αMHC (alpha myosin heavy chain) cell markers. Amplified hypoxanthine phosphoriboxyltransferase (HPRT) is shown as a control. (B) Nestin (green)/Hoechst (blue), beta III tubulin (green)/GFAP (red)/Hoechst (blue), and neuN (green)/GFAP (red)/Hoechst (blue) immunostaining of differentiated neural cells at day 13 of differentiation starting from mouse ES cells (upper panels) and NPCs (lower panel) reveals the presence of the specific markers. Scale bars, 20 µm. (C) IN Cell Analyzer immunofluorescence-based quantitative analysis performed on neural cells derived from mouse ES cells and NPCs at days 13 and 14 of differentiation. The percentage of neuronal cells is calculated upon the neuN-stained cells, whereas the percentage of glial cells is calculated on the GFAP-stained cells as described above.

We also performed comparative functional tests with compounds behaving as agonists for GPCRs and channels enriched in neural populations, such as glutamate receptors ( Figs. 4A,D ), purinergic receptors ( Figs. 4B,E ), TRPVR1 channel (transient receptor potential vanilloid receptor–1), histamine receptors, TRPM8 channel, acetylcholine, and serotonin receptors ( Figs. 4C,F ). The results obtained on neural cells derived from mES reporter cells and from c-Photina^® NPCs showed kinetics consistent with the activation of the endogenous receptors with comparable maximum luminescence intensity values ( Fig. 4G ).

Fig. 4.

FLIPR^TETRA® functional analysis of neural cells differentiated from mES cells and neural precursor cells (NPCs) at day 13 of differentiation. (A-C) Ca²⁺-dependent kinetics obtained after stimulation of neural cells derived from mouse ES cells with a compound master plate containing different agonists for (A) glutamate receptor subtypes (glutamate, quisqualate, DHPG, SYM 2081, kainate) and with a different compound master plate for (B) purinergic receptors (adenosine triphosphate [ATP], cytidine triphosphate [CTP], uridine triphosphate [UTP], uridine diphosphate [UDP]) and (C) TRPV1 channel (capsaicin), histamine receptor (histamine), TRPM8 (icillin), nicotinic receptors (acetylcholine), and 5HT receptors (serotonin) at a final concentration of 100 µM. (D-F) Ca²⁺-dependent kinetics obtained after stimulation of neural cells derived from NPCs with a compound master plate containing different agonists for (D) glutamate receptor subtypes (glutamate, quisqualate, DHPG, SYM 2081, kainite) and with a compound different master plate for (E) purinergic receptors (ATP, CTP, UTP, UDP) and (F) TRPVR1 channel (capsaicin), histamine receptor (histamine), TRPM8 (icillin), nicotinic receptor (acetylcholine), and 5HT receptors (serotonin). (G) Maximum luminescence intensity RLU (relative luminescence units) value measured at FLIPR^TETRA® instrument after stimulation of neural cells derived from mouse ES cells and NPCs with the agonists as labeled contained in the master plate used in B, C, E, and F. Each datum point represents the mean of 6 measures ± SEM.

We also performed functional studies with the neural cells derived from c-Photina^® NPCs at different days of differentiation using different agonists ( Figs. 5A-D ). The kinetics of the bioluminescent responses after stimulation of neural cells derived from NPCs at days 13, 17, and 21 with quisqualate, glutamate, and adenosine triphosphate (ATP) are reported in Figures 5A-C , as examples. We observed characteristic photoprotein-based kinetics in all the days tested, with maximum signals increasing on day 17 ( Fig. 5D ). The signal reduction observed around day 21 ( Fig. 5D ) could be likely due to an excessive cell confluence that induces cell suffering and consequent signal reduction. Besides, we could not exclude a reduction in target expression or variation in its subunit composition during differentiation. The possibility of using the cells for functional tests over different days suggests this cell model as suitable for application to screening purposes ( Figs. 5A-C ).

Fig. 5.

FLIPR^TETRA® functional tests performed on neural cells derived from neural precursor cells (NPCs) at different days of differentiation. (A-C) Response kinetics obtained after stimulation with (A) quisqualate, (B) glutamate, and (C) adenosine triphosphate (ATP) agonists for the stimulation of glutamate receptors and purinergic receptors, respectively, at days 13, 17, and 21 of differentiation. (D) Maximum luminescence intensity RLU value measured at FLIPR^TETRA® instrument after stimulation of neural cells derived from NPCs with the agonists as labeled. Each datum point represents the mean of 6 measures ± SEM.

Neural cells derived from NPCs are able to form functional neuronal networks. We demonstrated that these cells are able to form active synapses by studying synaptic vesicle (SV) recycling using the fluorescent styryl dye FM4-64 ( Fig. 6 ). The dye was taken up by neurons at day 14 of differentiation during the high K⁺ depolarization, through a protocol that induces multiple rounds of exo-endocytosis, leading to labeling of the entire pool of recycling SVs.²⁰

Fig. 6.

FM4-64 uptake test. (A) Phase contrast image of neurons derived from neural precursor cells (NPCs; at day 14 of differentiation). Bar, 10 µM. (B) Fluorescent image of the same neurons derived from NPCs loaded with FM4-64 by 1-min exposure to KRH-55 mM KCl and 1 µM tetrodotoxin (TTX; depolarizing solution). Neurons were subsequently washed 4 times with KRH buffer with 1 µM TTX and 10 µM CNQX (6-cyano-7-nitroquinoxaline-2,3-dione). Scale bar, 10 µm. Insert box scale bar, 4 µm.

We also succeeded in recording spontaneous action potentials using the MEA technology in neural cells derived from NPCs. The culture shows spontaneous activity from day 13 in a manner similar to primary cultures.²¹ We seeded the NPCs on gelatin-coated MEA chips and enabled growth and differentiation for 1 to 2 weeks (Suppl. Fig. S1A). We recorded the spontaneous electrical activities of the cells at different days of differentiation starting from days 13 to 21 (data not shown). In Supplementary Figures S1B-D are shown the electrical activities measured and expressed in µV/s. In particular, in Supplementary Figure S1B are shown typical spikes trains, and in Supplementary Figure S1C are reported the overlaid kinetics of all the action potentials recorded over the entire duration of the experiment (3 min). To confirm the nature of the action potentials, we first recorded the spontaneous activity in the c-Photina^® neurons differentiated from NPCs (day 17) and the electrical activity of the same cells after treatment with 1 µM TTX (a voltage-gated Na⁺ channel inhibitor) for 5 min. The same cells were then washed to remove TTX, and the electrical activity was measured again. As shown in Supplementary Figures S1D-F, the electrical activity measured before treatment was almost completely abolished by treatment with TTX and restored after the washout. The inhibition of the spontaneous electrical activity in c-Photina^® neurons derived from NPCs by TTX treatment strongly suggests the presence of action potentials mainly driven by the presence of functional voltage-gated Na⁺ channels.

To further characterize the neural network of the NPC culture, we decided to investigate also the glutamatergic transmission. The experiment was performed at day 20 of differentiation using 100 µM NBQX (an AMPA glutamate receptor inhibitor; Suppl. Figs. S1F-G). The result shows a predominant AMPA receptor-mediated electrophysiological activity in the neuronal network formed by c-Photina^® neural-differentiated NPCs at day 20. The NBQX treatment abolished the majority of the spontaneous activity but did not completely block action potentials because a residual electrical activity was still detectable after NBQX treatment.

Discussion

The neural cells differentiated from c-Photina^® mES cells offer many possibilities for the generation of cell-based assays to measure Ca²⁺ mobilization through the activation of endogenous receptors. A set of validation studies described here demonstrates that the protocol we set up is suitable for miniaturized formats and gives robust results in terms of marker expression and percentage of neural cells. We demonstrated that the system is sensitive enough to be used in screening aimed to identify agonist as well as antagonist compounds for endogenously expressed receptor modulation. As proof of principle, we screened neural cells derived from mES cells in agonist and antagonist mode (with quisqualate as the activator compound) with the LOPAC¹²⁸⁰™ library.

Furthermore, we developed a method that allows the preparation of a frozen bulk of cells (neural precursor cells) already committed to neural lineage and ready to use in miniaturized format after only 1 week postthawing. We demonstrated that the neural cells obtained from the frozen c-Photina^® NPCs are able to form mature neural structures, which express all the markers of neural cells directly derived from c-Photina^® mES cells and are able to respond to the same stimuli with similar pharmacology.

Ca²⁺ signals govern important processes in neurons, including neurotransmitter release, membrane excitability, neurite outgrowth, and neurodegeneration.^22,23 The cytosolic Ca²⁺ concentration ([Ca²⁺]_c) increases upon activation of plasma membrane Ca²⁺ channels or Ca²⁺ release from intracellular stores.²³ To detect Ca²⁺ concentration changes, we introduced a Ca²⁺-activated photoprotein reporter system into a complex endogenous neuronal network, which allows the accurate monitoring of Ca²⁺ fluxes. The presence of a Ca²⁺ sensor in a complex physiological environment could be very interesting from an HTS perspective.

We set up a robust differentiation protocol that allows the formation of a complex neural network containing different neuronal cell populations, with special attention to potential screening applications. For example, the idea of generating bulk cells at a precommitted neural stage allowed us to use cells as reagents that can be frozen and are ready to use when needed. This approach can consistently reduce the cell culturing time from plating to use, decreasing the problems of variability, in terms of cell numbers and cell type, and of reproducibility.

We also monitored the presence of different indicators of neural differentiation not only by RT-PCR and immunofluorescence analysis but also by functional tests. We observed that for some receptors, it is possible to perform functional tests in a time window of 1 week. This result is very important because it suggests the possibility of using the neural-differentiated cells over different days even though we suggest carefully investigating this parameter for each selected target.

The choice of the best differentiation time window for performing the functional test is not trivial, and several factors must be considered. In fact, the expression level of the selected target (receptor/channel) might not be stable in the chosen temporal window or could vary in terms of subunit compositions during differentiation, and this could significantly alter the pharmacological behavior.

We previously reported¹³ that on day 13 of differentiation, the populations of neural-differentiated c-Photina^® cells behave more like prenatal and early postnatal neurons rather than adult neurons. This kind of neuronal behavior is typical of the early CNS development stage (reviewed in Li and Xu²⁴). Furthermore, the electrophysiological analysis of neurons derived from c-Photina^® NPCs reveals that the spontaneous action potentials shown at day 13 were sporadic and of smaller amplitude compared with the one shown by neurons at days 17 and 20 of differentiation. At these latter time points, we observed spontaneous network bursts with spikes more frequent and more clustered, with higher amplitude and more organized in space compared to day 13, behavior that resembles adult neuronal networks. In addition, the incomplete block of excitatory activity by NBQX at day 13 provides evidence of a maturation trend of neuronal network activity from day 13 to day 21.

Even if the cell maturation continues over time, we elected days 13 and 14 of differentiation as optimal for our tests based on several considerations. First, the quisqualate signal was steady in that temporal window. Second, the choice of those days allows a reduction of time from cell plating to screening with a consequential reduction in cell number variability mainly due to proliferation of other cells such as glia and cells from other germ layers over time. These results demonstrated that the model is adaptable to screening purposes.

The high quality of the obtained neural networks was demonstrated not only by the expression of neural markers but also by the ability to respond to endogenous stimuli to form active synapses and to generate action potentials. The electrophysiological data also demonstrate that the neural network formed contains a significant proportion of excitatory glutamate-induced electrophysiological activity. The more homogeneous results and the reduction of differentiation time make this frozen batch of cells enriched in neural precursors a powerful tool for HTS purposes.

Footnotes

Acknowledgements

The authors thank Cinzia Nucci, Angela Molteni, and Alessandro Taddei (Axxam S.p.A.); Miriam Ascagni (ALEMBIC, San Raffaele Scientific Institute, DIBIT); Emilia Biffi and Giancarlo Ferrigno (Politecnico di Milano, NearLab, Italy); Cinzia Ferri and Lawrence Wrabetz (San Raffaele Scientific Institute, DIBIT, Division of Genetics and Cell Biology, Milan, Italy); and Stefania Filosa and Genesia Manganelli (Stem Cell Fate Lab, Institute of Genetics and Biophysics “Adriano Buzzati Traverso,” CNR, Naples, Italy).

References

Rink

: Receptor mediated calcium entry. FEBS Lett 1990;268:381-385.

Eglen

Reisine

: Photoproteins: important new tools in drug discovery. Assay Drug Dev Technol 2008;6:659-671.

Kendall

Badminton

: Aequorea victoria bioluminescence moves into an exciting new era. Trends Biotechnol 1998;16:216-224.

Campbell

Hallet

Daw

Ryall

MET

Hart

Herring

: Application of the photoprotein obelin to the measurement of free Ca⁺⁺ in cells. In deLuca

McElroy

(eds): Bioluminescence and Chemiluminescence: Basic Chemistry and Analytical Applications. New York: Academic Press, 1981:601-607.

Johnson

Shimomura

: Introduction to the bioluminescence of medusae, with special reference to the photoprotein aequorin. Methods Enzymol 1978;57:271-291.

Chiesa

Rapizzi

Tosello

Pinton

de Virgilio

Fogarty

: Recombinant aequorin and green fluorescent protein as valuable tools in the study of cell signalling. Biochem J 2001;55:1-12.

Mattheakis

: Seeing the light: calcium imaging in cells for drug discovery. Drug Discov Today 2000;(suppl 1):18-19.

Bovolenta

Foti

Lohmer

Corazza

: Development of a Ca(2+)-activated photoprotein, Photina, and its application to high-throughput screening. J Biomol Screen 2007;12:694-704.

Burdon

Smith

Savatier

: Signalling, cell cycle and pluripotency in embryonic stem cells. Trends Cell Biol 2002;12:432-438.

10.

Odorico

Kaufman

Thomson

: Multilineage differentiation from human embryonic stem cell lines. Stem Cells 2001;19:193-204.

11.

Wobus

Boheler

: Embryonic stem cells: prospects for developmental biology and cell therapy. Physiol Rev 2005;85:635-678.

12.

McNeish

: Embryonic stem cells in drug discovery. Nat Rev Drug Discov 2004;3:70-80.

13.

Cainarca

Fenu

Ferri

Nucci

Arioli

Menegon

: A photoprotein in mouse embryonic stem cells measures Ca²⁺ mobilization in cells and in animals. PLoS One 2010;5:e8882.

14.

Fico

Manganelli

Simeone

Guido

Minchiotti

Filosa

: High-throughput screening–compatible single-step protocol to differentiate embryonic stem cells in neurons. Stem Cells Dev 2008;17:573-584.

15.

Bolino

Bolis

Previtali

Dina

Bussini

Dati

: Disruption of Mtmr2 produces CMT4B1-like neuropathy with myelin outfolding and impaired spermatogenesis. J Cell Biol 2004;167:711-721.

16.

Parisi

Lonardo

Fico

Filosa

Minchiotti

: A versatile method for differentiation of multiple neuronal subtypes from mouse embryonic stem cells. J Stem Cells 2007;1:259-269.

17.

Watanabe

Kamiya

Nishiyama

Katayama

Nozaki

Kawasaki

: Directed differentiation of telencephalic precursors from embryonic stem cells. Nat Neurosci 2005;8:288-296.

18.

Bräuner-Osborne

Krogsgaard-Larsen

: Pharmacology of (S)-homoquisqualic acid and (S)-2-amino-5-phosphonopentanoic acid [(S)-AP5] at cloned metabotropic glutamate receptors. Br J Pharmacol 1998;123:269-274.

19.

Shigemoto

Nomura

Ohishi

Sugihara

Nakanishi

Mizuno

: Immunohistochemical localization of a metabotropic glutamate receptor, mGluR5, in the rat brain. Neurosci Lett 1993;163:53-57.

20.

Menegon

Bonanomi

Albertinazzi

Lotti

Ferrari

Kao

H-T

: Protein kinase A–mediated synapsin I phosphorylation is a central modulator of Ca²⁺-dependent synaptic activity. J Neurosci 2006;26:11670-11681.

21.

Chiappalone

Bove

Vato

Tedesco

Martinoia

: Dissociated cortical networks show spontaneously correlated activity patterns during in vitro development. Brain Res 2006;1093:41-53.

22.

Clapham

: Calcium signaling. Cell 1995;80:259-268.

23.

Ghosh

Greenberg

: Calcium signaling in neurons: molecular mechanisms and cellular consequences. Science 1995;268:239-247.

24.

: The role and the mechanism of gamma-aminobutyric acid during central nervous system development. Neurosci Bull 2008;24:195-200.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB

0.03 MB