J. Neurochem. (2011) 119, 957–971.
We characterized phenotype and function of a fetal human mesencephalic cell line (LUHMES, Lund human mesencephalic) as neuronal model system. Neurodevelopmental profiling of the proliferation stage (d0, day 0) of these conditionally-immortalized cells revealed neuronal features, expressed simultaneously with some early neuroblast and stem cell markers. An optimized 2-step differentiation procedure, triggered by shut-down of the myc transgene, resulted in uniformly post-mitotic neurons within 5 days (d5). This was associated with down-regulation of some precursor markers and further up-regulation of neuronal genes. Neurite network formation involved the outgrowth of 1–2, often > 500 μm long projections. They showed dynamic growth cone behavior, as evidenced by time-lapse imaging of stably GFP-over-expressing cells. Voltage-dependent sodium channels and spontaneous electrical activity of LUHMES continuously increased from d0 to d11, while levels of synaptic markers reached their maximum on d5. The developmental expression patterns of most genes and of the dopamine uptake- and release-machinery appeared to be intrinsically predetermined, as the differentiation proceeded similarly when external factors such as dibutyryl-cAMP and glial cell derived neurotrophic factor were omitted. Only tyrosine hydroxylase required the continuous presence of cAMP. In conclusion, LUHMES are a robust neuronal model with adaptable phenotype and high value for neurodevelopmental studies, disease modeling and neuropharmacology.
dopamine receptor D2
glial cell derived neurotrophic factor
green fluorescent protein
Lund human mesencephalic
microtubule-associated protein 2
receptor tyrosine kinase
Homogeneous cultures of human post-mitotic neurons are of interest in multiple research areas ranging from developmental neurobiology to toxicology. The demand on new model systems with regard to homogeneity and steady availability has increased. Furthermore, feasibility of molecular biological manipulations and the applicability of such cells for large screens are desirable, as such features have proven useful in studies unraveling mechanisms of genetic neurodegenerative diseases (Greer et al. 2010; Ittner et al. 2010).
Transformed cell lines provide the advantage of an easy supply, a relatively homogeneous culture, and the generation of genetically-modified subclones. For instance PC12, generated from a rat adrenal medullary pheochromocytoma, have greatly contributed to research on mechanisms of neurodegenerative diseases (Greene and Tischler 1976; Rabizadeh et al. 1993; Xia et al. 1995) and neurotoxicology (Das et al. 2004; Breier et al. 2010). Their strict neurotrophin-dependence (Greene and Tischler 1976) has been beneficial for some research questions, but also puts limits on the generalized use of the cells. Human cell lines derived from embryonic teratocarcinomas (e.g. NT2, hNT) (Pleasure et al. 1992) can be directed towards a post-mitotic neuronal phenotype, but the need for a very time-consuming differentiation protocol has limited their wide-spread use. Instead, human neuroblastoma cell lines, such as SH-SY5Y have been commonly applied in systematic toxicological evaluation programs (Forsby et al. 2009), as well as in studies of basic neurobiology (Biedler et al. 1978). Furthermore, they were used to examine the mechanisms of neurodegeneration (Tofaris et al. 2001) and for high-throughput screenings (Loh et al. 2008), although they are hard to differentiate to a genuine post-mitotic state. The field of stem cell research may become the most important source for various human cell types in the future. Neurons of different specificity may be derived from human embryonic stem cells (Reubinoff et al. 2001), from adult neural stem cells (Johansson et al. 1999) or from human induced pluripotent stem cells (Lee et al. 2010). However at present, handling of such cell cultures is still time-consuming, associated with high costs, and neither homogeneity nor synchronization of cells are always given.
For the rational design of human neuronal models that address the present gaps, a scientific paradox has to be overcome: proliferation is needed to create large numbers, but neurons are by definition in a stable post-mitotic state. A successful solution is the transformation of committed neural precursor cells with myc oncogenes to ensure immortalization and continuous proliferation. Inactivation of the oncogene by exposure to neurotrophic factors (Donato et al. 2007) or tetracycline-controlled gene expression then allows neuronal differentiation. The latter approach is based on the retroviral LINX-v-myc vector with regulated v-myc expression (Hoshimaru et al. 1996). Addition of low concentrations of tetracycline abolishes v-myc expression, which allows cells to exit the cell cycle and to differentiate. This construct was used to generate MESC2.10 cells to be used for neuronal transplantation in Parkinson’s disease (Lotharius et al. 2002). The source material was derived from 8-week-old human ventral mesencephalic tissue. Karyotyping of the cell line showed a normal set of chromosomes and female phenotype (Paul et al. 2007). However, these cells were reported to be unstable and heterogeneous with regards to tyrosine hydroxylase (TH) expression and they were not suitable for replacement of dopaminergic neurons upon transplantation (Paul et al. 2007; Fountaine et al. 2008). In 2005, the subclone LUHMES (Lund human mesencephalic) was created (Lotharius et al. 2005) and used to study dopamine related cell death mechanisms (Lotharius et al. 2005; Schildknecht et al. 2009). With respect to the parkinsonian toxin MPP+, LUHMES behaved similarly to primary cells (Schildknecht et al. 2009), while MESC2.10 were 1000-fold less sensitive (Fountaine et al. 2008). The general neuronal characteristics and the differentiation status of LUHMES at different culture conditions still await a comprehensive characterization. We addressed here the expression of neuronal markers, neurite outgrowth and electrophysiological properties of the cells, and present an optimized differentiation protocol leading to cultures with greatly improved homogeneity. The study also addressed the question whether functional and signaling studies may be performed with these cells in a simplified culture medium without added differentiation factors. These experiments revealed a robust, endogenous program driving the regulation of most neuronal genes independent of added cAMP/glial cell derived neurotrophic factor (GDNF). Tyrosine hydroxylase was the most prominent exception, and required external signals. Therefore, the kinetics and conditions for up- and down-regulation of this enzyme were further characterized.
Materials and methods
For LUHMES cell culture, Nunclon™ (Nunc, Roskilde, Denmark) plastic cell culture flasks and multi-well plates – pre-coated with 50 μg/mL poly-l-ornithine and 1 μg/mL fibronectin (Sigma-Aldrich, St. Louis, MO, USA) in H2O for 3 h – were used. After removal of the coating solution, culture flasks were washed once with H2O and air-dried before cell seeding. Proliferation medium consisted of Advanced Dulbecco’s modified Eagle’s medium/F12, 1× N-2 supplement (Invitrogen, Karlsruhe, Germany), 2 mM l-glutamine (Gibco, Rockville, MD, USA) and 40 ng/mL recombinant basic fibroblast growth factor (R&D Systems, Minneapolis, MN, USA). For standard differentiation, +/+ medium, consisting of Advanced Dulbecco’s modified Eagle’s medium/F12, 1× N-2 supplement, 2 mM l-glutamine, 1 mM dibutyryl cAMP (Sigma-Aldrich), 1 μg/mL tetracycline (Sigma-Aldrich) and 2 ng/mL recombinant human GDNF (R&D Systems) was used, whereas alternative differentiation approaches were performed in −/− medium lacking both cAMP and GDNF, or +/− medium (cAMP, but no GDNF) or −/+ medium (no cAMP, but GDNF).
LUHMES maintenance and differentiation
LUHMES cells were grown at 37°C in a humidified 95% air, 5% CO2 atmosphere. Proliferating cells were enzymatically dissociated with trypsin (138 mM NaCl, 5.4 mM KCl, 6.9 mM NaHCO3, 5.6 mM d-Glucose, 0.54 mM EDTA, 0.5 g/L trypsin from bovine pancreas type-II-S; Sigma-Aldrich) and passaged 1 : 10 when they reached 80% confluency. For differentiation, 8 × 106 LUHMES were seeded into a T175 flask in proliferation medium and differentiation was started after 24 h, that is, on day 0 (d0), by changing to +/+, −/−, +/− or −/+ differentiation medium. After 2 days of cultivation in culture flasks, cells were trypsinized and seeded into poly-l-ornithine/fibronectin pre-coated multi-well plates at a cell density of 1.5 × 105 cells/cm2 if not otherwise indicated. In case of cultivation for more than three additional days, fresh differentiation medium was added on day 5 (d5) of overall differentiation. For experiments involving d0–d2 LUHMES and for the comparison of the 1-step with the 2-step protocol, cells were seeded into multi-well plates at a density of 4.0 × 104 cells/cm2 and were differentiated directly in the plates as described above.
Generation of GFP-over-expressing LUHMES cells
Lentivirus production was carried out as already described (Vergo et al. 2007). Briefly, HEK293FT cells, grown in LUHMES proliferation medium, were transiently transfected by lipofection with a 3-plasmid vector system consisting of pHsCEW (transfer vector containing enhanced green fluorescent protein (EGFP) (Leander Johansen et al. 2005), pMD2.G (envelope vector, http://www.addgene.com) and pBR8.91 (packaging vector). The supernatant was filtered and stored at −80°C. Undifferentiated LUHMES cells were transduced by incubation for 12 h with virus-containing supernatant. A GFP-positive cell pool was obtained by FACS sorting (BD FACS ARIA II, Heidelberg, Germany). A mixture of approximately 2% GFP-positive cells and 98% normal LUHMES was used for differentiations and termed ‘mixed LUHMES/GFP-LUHMES’ cultures.
LUHMES or mixed LUHMES/GFP-LUHMES cultures were grown and differentiated on pre-coated 15 mm glass cover slips (Menzel, Braunschweig, Germany) in 12-well plastic cell culture plates at 1.3 × 105 cells/cm2. Cells were fixed with phosphate-buffered saline (PBS)/4% paraformaldehyde for 15 min at 21°C, washed, permeabilized with PBS/0.2% Triton X-100 and pre-incubated with PBS/1% bovine serum albumin (Calbiochem, San Diego, CA, USA) for 1 h at 21°C. Then, primary antibodies (see Figure S1) were added overnight at 4°C. After three washing steps with PBS/0.1% Tween, anti-mouse Alexa-594 and anti-rabbit Alexa-488 (1 : 1000, Molecular Probes, Eugene, OR, USA) were applied as secondary antibodies for 1 h at 21°C. Actin was labeled with phalloidin Alexa-568 (1 : 1000, Molecular Probes). Hoechst-33342 (1 μg/mL, Molecular Probes) was added for 10 min prior to the final washing step and cover slips were then mounted on glass slides with Fluorsave reagent (Calbiochem). Samples were imaged with an Olympus IX81 inverted epifluorescence microscope (Hamburg, Germany), using a 40× air objective or a 100× oil objective. Image processing was carried out with the Olympus CellP software.
DNA of LUHMES cells undergoing mitosis was labeled using the Click-iT EdU (5-ethynyl-2′-deoxyuridine) Imaging Kit (Invitrogen) according to the manufacturer’s instructions. Briefly, cells were differentiated on pre-coated 15 mm glass cover slips. EdU reagent (nucleoside analog; final concentration of 10 μM) was added at different time points to the differentiation medium. After 24 h, the cells were fixed (4% paraformaldehyde) and permeabilized with PBS/0.5% Triton X-100 for 15 min. After two washing steps with PBS/3% bovine serum albumin, cells were incubated with Click-iT reaction cocktail (modified fluorescein fluorophore) for 30 min in the dark. After additional washing steps and DNA staining with H-33342, cover slips were mounted on glass slides and imaged with the Olympus IX81 microscope. Nuclei were manually counted using the Olympus CellP software.
Scanning electron microscopy
LUHMES cells were grown and differentiated on pre-coated 10 mm glass cover slips in 24-well plates at a density of 1.3 × 105 cells/cm2, fixed for 30 min with cold 2% glutaraldehyde/3% formaldehyde in 0.1 M cacodylate buffer with 0.09 M sucrose, 0.01 M CaCl2 and 0.01 M MgCl2, pH 7.4 (Sigma-Aldrich). Samples were washed with cacodylate buffer, dehydrated in a graded series of ethanol, critical point dried from liquid CO2 and sputter coated with 5 nm gold/palladium. Cells were examined using a Philips SEM 505 (Eindhoven, The Netherlands) at 30 kV accelerating voltage. Images were digitally recorded and processed with the ImageJ 1.43s (National Institutes of Health, USA, http://rsb.info.nih.gov/ij/) or Adobe Photoshop CS2 software.
Neurite outgrowth was monitored in GFP-over-expressing LUHMES cells in mixed cultures. Pre-differentiated d2 cells were seeded on 35 mm pre-coated glass bottom dishes (WillCo Wells B.V., Amsterdam, The Netherlands) and incubated for 24 h at 37°C in a humidified 95% air/5% CO2 atmosphere before they were transferred to a Nikon Biostation inverted epifluorescence microscope (Düsseldorf, Germany). Time-lapse imaging of several fields in the plate was performed at 37°C with a 40× air objective for 48 h in intervals of 20 min. Videos were exported using Nikon Imaging Software and image processing was done with ImageJ 1.43s, applying the Simple Neurite Tracer algorithm for the measurement of neurite outgrowth speed.
Electrodes with a resistance of 2–5 MΩ were pulled of borosilicate glass (Clark, G150F, Warner Instruments, Hamden, CT, USA) on a Sutter Instruments (Novato, CA, USA) P-97 horizontal micropipette puller. All experiments were carried out on a temperature-controlled microscope stage (37°C), using a custom built Teflon recording chamber (800 μL volume). Pre-differentiated cells were seeded on 10 mm pre-coated glass cover slips, and whole-cell voltage and current clamp recordings were performed on d3–d12. Whole-cell currents were recorded using an L/M-EPC-7 amplifier (List Medical Electronic, Darmstadt, Germany), digitised at sampling frequencies between 10 kHz and 50 kHz using a DigiData 1320A AD/DA converter (Axon Instruments Inc., Foster City, CA, USA). For the recording of total ionic currents as well as spontaneous activity the patch pipettes were filled with (in mM) 90 K+-gluconate, 40 KCl, 1 MgCl2, 10 NaCl, 10 EGTA, 4 Mg-ATP, 10 HEPES/KOH (pH 7.4 at 37°C), whereas the bath solution contained (in mM): 155 NaCl, 1 CaCl2, 3 KCl, 10 d-(+)-glucose, 10 HEPES/NaOH (pH 7.4 at 37°C). To further dissect the sodium currents, the patch pipettes were filled with (in mM) 110 CsF, 10 NaCl, 20 tetraethylammonium chloride (TEA-Cl), 10 EGTA, 4 Mg-ATP, 10 HEPES/CsOH (pH 7.4 at 37°C), whereas the bath solution contained (in mM): 135 mM NaCl, 1 CaCl2, 2 MgSO4, 10 glucose, 5 TEA-Cl, 10 HEPES/NaOH (pH 7.4 at 37°C). The sodium currents were blocked by addition of 0.5 μM tetrodotoxin to the bath solution. Liquid junction potentials were measured and corrected, using the method described by Erwin Neher (1992). Current data were normalized for the cell size by using the cell capacitance as surrogate measure of the cell surface. The recorded current was divided by the individual cell capacitance, giving the current density. Stimulation, acquisition and data analysis were carried out using pCLAMP 10.2 (Axon Instruments Inc.) and ORIGIN 8.0 (OriginLab Corp., Northampton, MA, USA). Fast and slow capacitive transients were cancelled online by means of analogue circuitry. Residual capacitive and leakage currents were removed online by the P/4 method. Series resistance compensation was set to at least 50%. For analysis, traces were filtered offline at 5 kHz. Cells for measurements were chosen with respect to their morphological phenotype (small, round, phase-bright cell bodies with projections of at least five times cell body diameter).
RT quantitative PCR
For reverse transcriptase (RT) quantitative real-time PCR (qPCR) analysis, RNA was extracted with the RNeasy mini Kit (Qiagen, Hilden, Germany). Pathway-focused gene expression profiling was performed using the ‘Human Neurogenesis and Neural Stem Cell’ RT2 ProfilerTM PCR Array (SABiosciences, Frederick, MD, USA). All target genes are listed on the website (http://www.sabiosciences.com). For other transcript analyses of LUHMES on various differentiation days, primers (Eurofins MWG Operon, Ebersberg, Germany) were designed using AiO (All in One) bioinformatics software (Karreman 2002) and can be found in Figure S1. All RT-qPCRs were run in a Bio-Rad Light Cycler (Bio-Rad, München, Germany) and analyzed with Bio-Rad iCycler software (Zimmer et al. 2011a). The threshold cycles (Ct) were determined for each gene and gene expression levels were calculated as relative expression compared to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (2−(ΔCt)) or as fold change relative to d0 (2−(ΔΔCt)). ΔCt and ΔΔCt were calculated using following formulas: ΔCt = Ct(day X, gene Y) − Ct(day X, GAPDH). ΔΔCt = ΔCt(day X, gene Y) −ΔCt(day 0, gene Y).
Western blot analysis
Cells were scraped from the plates and lysed in ristocetin-induced platelet agglutination-buffer (50 mM Tris-base, 1 mM EDTA, 150 mM NaCl, 0.25% sodium deoxycholate, 1% Nonidet P 40 substitute, 1% phenylmethylsulfonyl fluoride, 1 mM Na3VO4, pH 7.4; all from Sigma-Aldrich) for 15 min on ice. After removal of cell debris via centrifugation, protein concentrations were determined by using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL, USA) and 25 μg of each sample were boiled for 5 min at 95°C, separated on 12% sodium dodecyl sulfate gels and transferred onto nitrocellulose membranes (Amersham, Buckinghamshire, UK). Equal loading and transfer were controlled by Ponceau S staining. The membranes were incubated with primary antibodies at 4°C over night in 5% milk in Tris-buffered saline/0.1% Tween. Horseradish peroxidase-conjugated secondary antibodies were added for 1 h at 21°C and visualized in a FUSION-SL 4.2 MP chemiluminescence system (Peqlab, Erlangen, Germany) using enhanced chemiluminescence western blotting substrate (Thermo Scientific). Primary antibodies: mouse anti-GAPDH (1 : 10 000, Invitrogen), mouse anti-TH (1 : 2000, Chemicon, Temecula, CA, USA), goat anti-Ret (1 : 1000, R&D Systems). Secondary antibodies: goat anti-mouse-horseradish peroxidase (1 : 10 000, Jackson Immuno Research Europe, Suffolk, UK), rabbit anti-goat-horseradish peroxidase (1 : 5000, Sigma-Aldrich).
MPP+-treatment and resazurin reduction assay
Cells were either differentiated in +/+, +/−, −/+ or −/− medium and seeded after 2 days of pre-differentiation in 96-well plates at a density of 140 000 cells/cm2 in the respective differentiation medium. On d6, cells were stimulated with MPP+ (Sigma-Aldrich). Cell viability was assessed after 72 h with the resazurin assay (Schildknecht et al. 2009) using 5 μg/mL resazurin sodium salt (Sigma-Aldrich).
[3H]-MPP+ and [3H]-dopamine uptake assay
Uptake measurements were performed as described earlier for murine dopamine neurons or LUHMES (Schildknecht et al. 2009; Zimmer et al. 2011b). In brief, cells were exposed to 15 nM (4625 Bq/well) [3H]-MPP+ (3.1635 TBq/mmol stock solution; Perkin Elmer, Boston, MA, USA) and 5 μM [1H]-MPP+ or 15 nM [3H]-dopamine (4625 Bq/well) and 5 μM [1H]-dopamine in 1 mM ascorbic acid. Dopamine transporter activity was blocked by 30 min pre-treatment with 0.1 μM GBR12909. Supernatants were collected at different time points, and cells were gently washed prior to lysis with PBS/0.1% Triton X-100. Radioactivity in cell lysates and corresponding supernatants was determined using a Beckman LS-6500 scintillation counter (Brea, CA, USA). Release studies were performed under non-equilibrium conditions with continuous removal of supernatant, and measurement of activity in all fractions, as well as determination of activity in cells at the onset of the experiment and after different times.
Detection of endogenous dopamine
Cells (about 1.7 × 106) were collected on ice in PBS containing 1 mM ascorbic acid to prevent autoxidation of dopamine. Following disruption of the cells by 10 sonication pulses, samples were analyzed by a dopamine ELISA (IBL International GmbH, Hamburg, Germany) according to the manufacturer’s protocol.
Data in figures are shown as means ± SEM of at least three independent differentiations. For statistical analysis, Student’s t-test, one-way anova with Newman–Keuls post-test, or two-way anova, followed by a post hoc Bonferroni’s test were applied as appropriate. All statistics were calculated using GraphPad Prism software (San Diego, CA, USA) and p < 0.05 was considered as being significant.
Conversion of undifferentiated LUHMES into post-mitotic neuronal cells
For biochemical and morphological analysis, it is important to obtain homogeneously distributed and developed cells. Synchronization of the differentiation process arises as additional demand in experiments examining the change of cells over time. In order to address these requirements, we established a 2-step procedure as new standard LUHMES differentiation protocol (Fig. 1a). This involved a 48 h pre-differentiation phase in flasks. The timing was chosen in a way to nearly halt proliferation of the cells, which allowed a better control of cell densities after the differentiation (Figure S2). The re-plating step on day 2 (d2) allowed the redistribution of cells to multi-well culture dishes, so that each well contained cells of exactly the same differentiation state and density. Experiments with 96-well dishes were not possible with the 1-step differentiation, as cells overgrew or died in this culture format (Figure S2). In the second phase, LUHMES cells were terminally differentiated for up to 10 days (d2–d12).
Scanning electron microscopy (SEM) showed that proliferating LUHMES grew in regularly distributed colonies (Fig. 1b). Neurites were absent at this stage, but small structures resembling the whip-like appendages called primary cilia were frequently observed. Similar structures have for instance been found on stem cells (Stearns 2009). When switched to differentiation medium, LUHMES cells underwent a rapid morphological change and formed an elaborate neurite network on d5 (Fig. 1b). Undifferentiated cells already expressed the neuronal cytoskeletal protein βIII-tubulin (Fig. 1c). The mRNA of this gene was up-regulated during differentiation, and staining of the respective protein showed its distribution along the neurites and in the somata of differentiated cells. In parallel, the mRNA of Fox-3/NeuN (marker of post-mitotic neurons) was up-regulated, whereas CDK1 (cell cycle regulator) was strongly down-regulated. On d5, 100% of cells displayed βIII-tubulin positive neuritic extensions and NeuN was maximally up-regulated (Fig. 1c and d). This indicates that the culture reached a stable neuronal-like state with high synchronicity. This was further characterized with single cell resolution. On d0, 85–95% of LUHMES were positive for the cell cycle marker Ki-67 and 7% were found to be mitotic as indicated by H3S10P (phosphorylated serine-10 in histone H3), whereas NeuN staining was not detectable (Figure S2). On d5, nearly all cells had left the cell cycle, as they were Ki-67 and EdU negative and uniformly expressed NeuN (Figs. 1d and S2). Immunodetection of Ki-67 and NeuN in d10 cells gave similar results as on d5, even when tetracycline was omitted from the culture medium from day 5 onwards (not shown). Thus, the new protocol results in the synchronized and irreversible conversion of LUHMES neuronal precursors into a homogenously post-mitotic neuronal population within 5 days.
Electrophysiological properties of post-mitotic LUHMES
The LUHMES cultures obtained by our differentiation protocol were characterized for basic electrophysiological properties, with a specific focus on homogeneity and the question whether full maturity was reached on d5. Using a whole-cell patch clamp approach, we investigated the total ionic currents in LUHMES cells under voltage clamp conditions. All cells recorded at d5–d9 showed fast-activating inward currents, followed by strong outward currents for testing potentials more positive than −20 mV (Fig. 2a). Under conditions which lead to a block of potassium (K+) channels, all outward currents were abolished (Fig. 2b), whereas application of the selective voltage-gated sodium (Na+) channel blocker tetrodotoxin eliminated almost all inward currents (Fig. 2c). These findings demonstrate the presence of functional voltage-gated Na+ and K+ channels in differentiated LUHMES cells. We took a more quantitative approach to examine the time course of the expression of Na+ channels. Cells of different maturity (d3–d11) were used for measurements of the peak amplitude of every voltage step. These data were normalized for the cell size and expressed as current density, using capacitance as a surrogate measure of the cell surface (Fig. 2e). The values increased continuously from d3 to d11 (Fig. 2f), indicating that the full electrophysiological differentiation may be delayed compared to the changes assessed by immunostaining and qPCR. This would be in agreement with reports from other cell systems. To follow up on these findings, we investigated the spontaneous electrical activity of d3–d12 LUHMES by current clamp recordings. Up to d9, approximately 40% of the cells were spontaneously active. After 10–12 days of differentiation, the culture behaved homogeneously with respect to this endpoint, as all measured cells (n = 23) generated spontaneous action potentials. The behaviour of the majority of cells was characterized by more or less regular firing, while some cells fired trains of bursts with breaks of several seconds in between (Fig. 2g). In summary, neuronal electrical features seemed to increase from d5 to d12 while the post-mitotic state was already reached.
Differential changes in phenotypic markers of neuronal maturation
The synchronized culture allowed a broad characterization of LUHMES cells at different maturation stages on the basis of gene transcript profiling. Initially, several precursor cell markers were examined on d0 and d5/6 using qPCR and immunocytochemistry. Undifferentiated LUHMES expressed SOX2 mRNA and immunostaining confirmed the expression of the cognate protein, but showed that it was exclusively localized in the cytosol on d0 and completely lost on d5 (Table 1 and not shown). A similar kinetic behavior was observed for PAX3, a marker of migrating neuroblasts. Day 0 cells also expressed nestin, BRN3A and ASCL1 as expected, but the expression of these precursor markers was only slightly down-regulated upon differentiation. The fact that the cells retain some phenotypic features usually associated with immature cells was also confirmed by staining of polysialylated neural cell adhesion molecule (PSA-NCAM), which was positive on d0 for all cells and still positive for some cells on d5 (Figure S3). Immunostaining indicated that nestin still formed cytoskeletal structures in fully post-mitotic cells on d10, although the intensity was weaker than on d5 and d0 (not shown). As expected, the mesodermal gene SOX17, the rosette marker Forse-1 and the two bone morphogenetic protein genes BMP4 and BMP15 were not expressed (Table 1).
In order to characterize the neuronal differentiation process, 84 genes with key roles in neuronal development were analyzed by qPCR. One large group of genes was expressed more than 4-fold higher in d6 cells compared to d0 precursors. These were linked to general neuronal function or were neurite- and synapse-related. Particularly pronounced up-regulations were observed for the neuronal differentiation inducer neuregulin 1, the synaptogenic protein neuronal pentraxin 1 and the pre-synaptic dopamine receptor D2 (DRD2) (Figure S4a–c).
A second group of 25 genes was already highly (> 1% of the GAPDH level) expressed on d0 and did not change until d6. For instance MEF2C, an effector of neurogenesis, reached 28% of GAPDH levels and midkine (neurite growth-promoting factor 2, MDK) or tyrosine activation protein (YWHAH) up to about 7% (Figure S4d). On protein level this was found for Nurr1 staining, which did not change (Figure S3), and also for most of the synaptic proteins or neuregulin 1 (Fig. 3 and not shown). These data indicate a partial expression of neuronal features already in d0 LUHMES, and suggest advanced maturation on d6 regarding neurite and synapse formation.
In order to correlate qPCR data with protein expression and localization as assessed by immunocytochemistry, we focused on synaptic markers. The mRNA of most markers was already detectable in undifferentiated LUHMES (Figure S4e) and then up-regulated about 10-fold (Figure S4f). Generally, up-regulation occurred within 2 days and reached saturation on d6. However, post-synaptic markers (GRIN1, DLG4, neuroligin 1) and synaptic vesicle 2a were up-regulated later than the other genes (Figure S4f). Immunocytochemistry confirmed these findings on protein level (Fig. 3 and not shown). All markers were found in every cell examined, and the staining patterns were highly uniform throughout the whole cell culture. In d0 LUHMES, the synaptic proteins were present in low amounts as dot-like patterns in the cytosol, and the protein amounts (staining intensity) were strongly increased in d5 cells (Fig. 3). Costaining with βIII-tubulin revealed that the proteins were mainly localized close to microtubules within the extensions. Post-synaptic markers like the scaffold protein discs large homolog 4 (DLG4) were usually stained much weaker and seemed to accumulate in particular at crossings or swellings of neurites (Fig. 3). All immunostainings were also performed with d10 cells and showed fluorescence signals indistinguishable from d5 cells regarding localization and strength (not shown).
Neurodevelopmental aspects of neurite growth
Neurite outgrowth and growth cone regulation are key features of neurons not observed in any other cell type. LUHMES may provide a model system to study such processes in human cells. To provide a basis for this, we evaluated the kinetics of neurite formation and extension by SEM and fluorescence microscopy. Already on d2, LUHMES homogeneously extended structures with the characteristics of lamellipodia and filopodia, and these stained positive for the cytoskeletal protein βIII-tubulin in the proximal part and for F-actin in the distal parts (Fig. 4a). On d3, neurites of 50–150 μm length had formed, and these often ended with large membrane protrusions (Fig. 4b). These membrane areas stained intensively for F-actin, and were mostly devoid of βIII-tubulin staining (Fig. 4b). Such morphological features and the staining pattern are characteristic for functional growth cones in primary neuronal cultures.
In order to characterize the dynamics of the growth behaviour, individual living cells were imaged. To this end, proliferating LUHMES were lentivirally transduced to stably over-express green fluorescent protein (GFP). A 1 : 50 mixture of GFP-expressing and non-transduced LUHMES was used for time-lapse imaging experiments to allow visualization of neurites and somata of distinct cells during differentiation (Fig. 4c). Using these mixed LUHMES/GFP-LUHMES cultures on d3–d5, we found that cell morphology was very similar to what is known from histochemical stainings of dopamine neurons in the substantia nigra (Arsenault et al. 1988; Jaeger et al. 1989). 60% of the cells displayed one, 40% two clearly defined neurites at opposite poles (n = 100 cells), but only one of them constantly elongated over the recorded period of time. The outgrowth of this main extension was a rapid process of on average 20 μm/h (n = 10 cells). Growth was not constant, but involved both rapid expansion periods with peak velocities of up to 50 μm within 30 min and stagnation or ‘orientation’ phases, when the neurite repeatedly extended and retracted into different directions. Frequently, the growing neurite formed a second extension, followed by the collapse of the original growth cone and formation of a new one on the new ending (Fig. 4c). On d5, neurite elongation slowed down considerably so that many (65%) LUHMES cells did not display a growth cone at a given time point (Fig. 4d). Average neurite length at this stage was in the range of 500–1000 μm, and on d5–d10 all these structures were positive for the microtubule-associated protein tau (not shown). Microtubule-associated protein 2 (MAP2) staining appeared to be restricted to the cells’ single main extension on d5 and d10 (Figure S5a and not shown). In order to examine this in more detail, we stained d5 mixed LUHMES/GFP-LUHMES cultures, since in the sparsely-distributed GFP-over-expressing LUHMES, a clear distinction of neurites belonging to the same cell body was possible. We found that in all cells with two neurites, MAP2 was expressed in both of them. However, a significant MAP2 fluorescence signal was only detected in the proximal part of the neurites (Figure S5b). This feature explained why in the LUHMES culture, in which cells form neurites that are considerably longer than the distance between the somata, MAP2-positive and -negative neurite fragments were observed. These properties of LUHMES hint at a not yet fully established polarity as it has also been observed in young in vitro differentiating hippocampal neurons (Pennypacker et al. 1991), and they resemble stainings of primary mesencephalic cultures where MAP2 and TH localized to all neurites (Mytilineou et al. 2003).
Progress along the dopaminergic lineage during LUHMES maturation
Previous reports took a dopaminergic phenotype of differentiated LUHMES cells for granted and used them for mechanistic studies on neurodegeneration. The present study used 11 different markers to provide basic background information about the extent of differentiation of d0–d10 LUHMES along the dopaminergic lineage in the context of overall neuronal characterization. Investigation of the transcript levels over time indicated that both the initial expression and the extent of regulation during LUHMES maturation depended on the marker examined (Fig. 5a). One group –VMAT-2 and KCNJ6– was not detected in d0 LUHMES, and was clearly expressed on d6 (Fig. 5a). A second group was already strongly expressed on d0. Some of these genes were further up-regulated on d6 (DAT, GFRA1), while others remained at the same level (NURR1, EN1) (Fig. 5a). Thus, the regulations of different ‘dopaminergic development markers’ ranged from no change between d0 and d6 (e.g. DRD1) to a 1000-fold increase (DRD2). Analysis of the expression time course showed that saturation was reached at about d6, but that the kinetics were highly variable (Fig. 5b). Expression of TH, which is one of the most important markers of mature dopaminergic neurons, was not raised over baseline level until day 6, whereas the dopamine transporter (DAT) was maximally up-regulated on d2 and then rather decreased in expression (Figs 5b and S6). DAT activity was tested as example for functional implications. Import of the specific radiolabeled transporter substrate [3H]-MPP+ was measured in d0 and d6 LUHMES in the presence and absence of the specific DAT inhibitor GBR12909. While d0 LUHMES had an activity close to baseline, d6 cells exhibited significant uptake of [3H]-MPP+, which was completely prevented by GBR12909 (Fig. 5c). These data provide a complex picture of the dopaminergic differentiation state of LUHMES which cannot be described by a single marker. As seen with general neuronal markers, LUHMES cannot be assigned to a defined neurodevelopmental stage, as they co-express markers of mature neurons simultaneously with those normally associated with dopaminergic precursors (Figure S6).
Robust predetermination of the neuronal fate under altered differentiation conditions
In many fields of neurobiology, neuronally differentiated LUHMES without the continued presence of high concentrations of cAMP and GDNF would be favorable. Therefore, we developed an alternative differentiation protocol based on medium without these two factors (−/−) to compare the phenotype of the resultant cells to the one of those obtained by the standard protocol (+/+). The −/− cells looked morphologically and immunocytologically (βIII-tubulin, NeuN, discs large homolog 4 (DLG4), synaptosomal-associated protein 25 (SNAP25) similar to cells derived under +/+ conditions (Fig. 6a). Both types of cells grew neurites at the same speed, and the absence of cAMP/GDNF facilitated the examination of pathways controlling neurite outgrowth with various inhibitors (data not shown). To evaluate the effects of −/− medium on the overall neuronal differentiation, we also analyzed the expression kinetics and levels of > 15 general neuronal mRNAs. Except for small changes in the levels of synaptic vesicle 2a at late time points, no significant differences to +/+ cells were detected (Fig. 6b and not shown). These findings show that LUHMES can be differentiated into post-mitotic neurons without exposure to cAMP and GDNF.
In a final step, we asked how dopaminergic markers behaved under the altered conditions. The alternative differentiation did not significantly affect the expression of six (including DAT) out of nine markers examined by PCR. However, DRD2 and receptor tyrosine kinase (RET) mRNA levels were significantly reduced by 70–80% in −/− cells and the induction of TH and aromatic amino acid decarboxylase was prevented nearly completely (Figs 6c and S6). This differential effect on the markers was further examined on the level of protein expression and function. Western blots of RET protein and immunostainings of DRD2 confirmed our mRNA findings (Figure S3). Activity of DAT remained essentially the same under −/− and +/+ conditions (Fig. 7a). However, +/+ cells were more sensitive to the toxicity of MPP+ than −/− cells (Figure S7), consistent with earlier findings that inhibitors of TH activity reduce MPP+ toxicity (Schildknecht et al. 2009). Western blots of LUHMES cell lysates showed that differentiation with cAMP and GDNF led to an up-regulation of tyrosine hydroxylase beginning on d4 and rising to d8, whereas −/− cells expressed very little TH protein, if any (Fig. 7b). These data correlated with the dopamine content, which was high in +/+ and low in −/− cells (Fig. 7c). They indicate that in LUHMES the expression of different features, normally associated with dopaminergic cells, can be regulated differentially by culture conditions. This was investigated further by experiments, designed to switch on TH expression in −/− LUHMES by a delayed shift to +/+ medium. It was also examined whether TH expression was turned off by a switch of +/+ cells to −/− medium on d6 of differentiation. The results show that TH expression depends on external signals and in particular on dibutyryl-cAMP (Fig. 7d). Since the mRNA expression of neurotransmitter-related genes (e.g. COMT, PNMT, GAD1, DBH), characteristic for other cell types, did not differ between −/− and +/+ cells (Figure S6), this suggests that LUHMES do not shift overall neurotransmitter phenotype when differentiated in −/− medium, but that only TH expression is reduced. We therefore tested whether the dopamine uptake and release machinery was functional in −/− cells. The −/− cells took up external radioactive dopamine to the same extent as +/+ cells (not shown). When these cells were exposed to methamphetamine or depolarized by K+, they released similar amounts of dopamine (Fig. 7e). The same was observed for cells loaded with MPP+. After exposure to methamphetamine, they released similar amounts of the radiolabel, and most of this release was blocked by the DAT inhibitors mazindol or GBR12909 (Fig. 7f). In summary, genes related to the neuronal phenotype, including also most dopaminergic markers, and the neurotransmitter machinery, seem to be predetermined to be regulated in a defined pattern in LUHMES cells upon transition to a post-mitotic state, while the most prominent effect of growth factors is the regulation of TH.
The goal of the present work was to establish the LUHMES cell line as general human neuronal cell model. We have demonstrated the high homogeneity and ease of handling of this cell line, which allows the use of biochemical analytical methods, in addition to single cell approaches. The immunocytochemical detection of multiple neuronal markers showed that d5/d6 cells resemble primary neuronal cultures in many respects. The differentiation that proceeds rapidly and in a relatively synchronized way makes this system suitable for developmental studies. This notion is further supported by our findings of typical growth cone and neurite elongation behaviour, and that such processes may also be studied in media free of serum, cAMP and growth factors.
The neuronal status of LUHMES and its robustness have been little defined in earlier studies examining mainly their degeneration (Lotharius et al. 2005; Selenica et al. 2007). Before establishment of the differentiation conditions described here, handling of the cells often lead to asynchronous differentiation and heterogeneity of the cultures (Figure S2). For instance, the toxicity of MPP+ was hard to control, as it is highly cell density-dependent (Figure S7). With the old 1-step differentiation procedure, the cells became post-mitotic in a less stringent way. With the new protocol presented here, the cell numbers obtained were reproducible within a narrow range and easy to control (Figure S2). The elimination of such variation has allowed the present basic characterization of the cells’ developmental status in the proliferating and differentiating state. The experiments with varying media supplements showed a surprisingly robust fate determination of the cells towards neuronal maturation independent of external stimuli. Within the course of these studies it also turned out that the LUHMES line may be used to generate neurons not containing the dopamine synthesis machinery, and that various features of dopaminergic neurons may be regulated independently.
One particular advantage of the LUHMES model is the extremely high conversion rate to post-mitotic neurons (> 99%), and that this can be followed both by biochemical methods addressing the entire population and on a cell-by-cell basis. In this respect they behave similarly to, for example, rat PC12 cells (Greene and Tischler 1976), which are the most widely used model cell line in neuroscience. A human model with such functions is still required. For instance, the commonly applied teratocarcinoma-derived hNT cells form an extensive neurite network (Pleasure et al. 1992), but they cannot be followed on a single-cell basis during their lengthy differentiation, although the protocols have been considerably improved lately (Podrygajlo et al. 2010). The same applies to neurogenic stem cells. Although highly improved protocols have been published for neuronal differentiation (Chambers et al. 2009) of pluripotent cells, they involve ongoing proliferation and death of the start population. Another favourable property of LUHMES cells is the formation of the very long neurites. Interestingly, LUHMES may be complementary to PC12 for such neurite studies, as their neurite growth occurs spontaneously and independently of the addition of exogenous nerve growth factor, which is required in that rat model (Greene and Tischler 1976; Rabizadeh et al. 1993; Parran et al. 2003).
We found here that LUHMES acquire the basic electrical properties of neurons more slowly than many other differentiation marks (e.g. synaptic protein expression). The development of functional synapses may be an even later step, and requires further investigation. Also, the axo-dendritic polarization status has not been clarified to a detailed extent as for some other cells, such as hNT2 (Pleasure et al. 1992). The significance of the overlap of MAP2 with tau in the proximal part of long neurites needs further investigation in the future. Particularly for in vitro mesencephalic cultures, clear information regarding MAP2 and tau distribution is sparse in the literature, and therefore does not give good guidance to what should be expected from LUHMES. In other types of neurons, MAP2 localization is not exclusively restricted to dendrites (Binder et al. 1986). Both for hippocampal and dopaminergic neurons it has been shown that axons frequently arise from MAP2 containing processes (Dotti et al. 1988; Jaeger et al. 1989), and strong axonal MAP2 staining has been detected in spinal cord neurons (Papasozomenos et al. 1985).
Our cell characterization study suggests multiple applications of LUHMES besides the area of Parkinson’s disease explored earlier (Lotharius et al. 2005; Schildknecht et al. 2009, 2011). For instance, LUHMES appear to be an interesting new system for neurodevelopmental studies (e.g. neurite outgrowth) or for studies on the regeneration of experimentally damaged neurites. In particular, biochemical studies and the characterization of the culture by analysis of transcript profiles become possible because of the homogeneous differentiation. For many such future applications, transfectability of the cells would be an advantage. We have demonstrated here with GFP expression that lentiviruses represent an efficient tool for gene transfer into LUHMES, while classical methods generally showed very low efficiency.
The predetermined neuronal fate of LUHMES was clearly illustrated in this work, but it became also obvious that there is a certain range in which differentiation conditions can determine the exact neuronal phenotype. In this context it was interesting that dopaminergic markers showed a differential behaviour. Some (e.g. DAT) were regulated during the differentiation independent of external growth factors. The situation was different for RET, TH and aromatic amino acid decarboxylase, which required the continuous presence of cAMP/GDNF for their maximal expression in differentiated cells. When the role of the individual factors was tested, the data on TH protein amounts showed that dibutyryl-cAMP alone was sufficient for up-regulation. This is consistent with regulation of the TH promoter by cAMP and cAMP response element binding protein (Hyman et al. 1988; Piech-Dumas and Tank 1999; Lewis-Tuffin et al. 2004). GDNF alone hardly showed any effect, in accordance with findings that some of its neuronal activities require concurrent activation of cAMP-dependent pathways (Engele and Franke 1996). The individual factors also affected MPP+ sensitivity of LUHMES cells in the low concentration range. Differentiation in the presence of cAMP alone was as efficient as cAMP/GDNF for sensitization (Figure S7). GDNF alone also had some sensitizing effect. Possibly this was because of an additional modulation of cAMP levels under MPP+ stress, but this was not yet followed up further.
The great majority of gene regulations and protein expression was not affected at all by the absence of cAMP/GDNF during the differentiation (−/−). This suggests that −/− differentiated cells do not acquire an alternative neurotransmitter phenotype, but rather only lack dopamine. This assumption is supported by two findings. First, the dopamine uptake and release machinery is still intact in −/− cells, and TH can be easily induced even in already differentiated cells by growth factor addition. Second, we did not find evidence that genes required for other neurotransmitter phenotypes were specifically up-regulated under −/− conditions. The only gene found to be higher expressed under this condition was PITX3, which is a key transcription factor typically involved in dopaminergic maturation and maintenance. Its up-regulation in the absence of dopamine might be regarded as compensatory response, which has been suggested to be regulated via miR133b (Li et al. 2009).
We cannot exclude that further exogenous factors affect the differentiation of LUHMES. These might include standard medium components, attachment factors and mediators produced or modified by the cells themselves, such as retinoic acid. This might explain a certain microinhomogeneity of the cultures, for example, concerning the kinetics of the loss of polysialylated neural cell adhesion molecule (PSA-NCAM). However, most other markers were found to be homogenously distributed.
Concerning the specification of LUHMES along the dopaminergic maturation pathway, a complex picture emerged (Figure S6). The proliferating d0 cells not only expressed markers of neuronal precursors, but also many neuronal markers and several dopaminergic features, including for example NURR1 expression, which was already at its maximal level. Thus, d0 have already a definite neuronal commitment. In d6 cells, additional neuronal and dopaminergic features were up-regulated, including TH, DAT, D2 and KCNJ6, which would typically be found in differentiated A9 dopaminergic neurons. However, some of the features usually associated with immature cells, such as nestin expression, were not lost at this stage. The mixed maturity status of differentiated LUHMES may have two different reasons. First, the maturation of different features may follow different time courses, and some processes may require more than 10 days differentiation time. Second, the cells may only have a limited maturation capacity under the chosen culture conditions. Possibly, contact to other cell types or exposure to other factors (sonic hedgehog, retinoic acid, ascorbic acid, growth factors) may be required for full maturation. This question will require more investigation in the future. This also applies to the optimization of long-term cultures, which are of high interest for the examination of synaptic integration. Immediate applications for the new model not requiring significant modifications are studies of neurite outgrowth or regeneration and of its guidance. Moreover, with the LUHMES cells, a new powerful system is available to characterize the modes of action of developmental neurotoxicants. It might be particularly suited, when highly quantitative high-throughput assays are to be performed.
We are indebted to many colleagues for valuable contributions and insightful discussions. We acknowledge in particular Dr. Joachim Hentschel from the Electron Microscope Service Facility, Dr. Elisa May from the Bioimaging Center (BIC), and Dr. Sabine Kreißl and Dr. Giovanni Galizia for assistance with electrophysiological measurements. The work was facilitated by grants from the Doerenkamp-Zbinden foundation, the Land Baden-Württemberg and the DFG (KoRS-CB). D.S. was funded by an IRTG1331 fellowship. The authors declare no conflict of interest.