In Vitro Cell.Dev.Biol.—Animal DOI 10.1007/s11626-015-9873-3

Media formulation influences chemical effects on neuronal growth and morphology Joshua A. Harrill & Brian L. Robinette & Theresa M. Freudenrich & William R. Mundy

Received: 25 November 2014 / Accepted: 21 January 2015 / Editor: T. Okamoto # The Society for In Vitro Biology 2015

Abstract Screening for developmental neurotoxicity using in vitro, cell-based systems has been proposed as an efficient alternative to performing in vivo studies. One tool currently used for developmental neurotoxicity screening is automated high-content imaging of neuronal morphology. While highcontent imaging (HCI) has been demonstrated to be useful in detection of potential developmental neurotoxicants, comparison of results between laboratories or assays can be complicated due to methodological differences. In order to determine whether high-content imaging-based developmental neurotoxicity assays can be affected by differences in media formulation, a systematic comparison of serum-supplemented (Dulbecco’s modified Eagle’s media (DMEM) + 10% serum) and serum-free (Neurobasal A + B27) culture media on neuronal morphology was performed using primary rat cortical neurons. Concentration–response assays for neuritogenesis, axon and dendrite outgrowth, and synaptogenesis were performed in each media type using chemicals with previously demonstrated effects. Marked qualitative and quantitative differences in the characteristics of neurons cultured in the two media types were observed, with increased neuronal growth and less basal cell death in Neurobasal A + B27. Media formulation also affected assay sensitivity and selectivity.

J. A. Harrill : B. L. Robinette : T. M. Freudenrich : W. R. Mundy (*) Integrated Systems Toxicology Division, National Health and Environmental Effects Research Laboratory (NHEERL), United States Environmental Protection Agency, 109 TW Alexander Drive, Research Triangle Park, NC 27711, USA e-mail: [email protected] Present Address: J. A. Harrill Center for Toxicology and Environmental Health (CTEH), LLC, 5120 North Shore Drive, North Little Rock, AR 72118, USA

Increases in assay sensitivity were observed in Neurobasal A + B27 media as compared to serum-supplemented DMEM. In some instances, a greater difference between effective concentrations for cell death and neurodevelopmental-specific endpoints was also observed in Neurobasal A + B27 media as compared to serum-supplemented DMEM. These data show that media formulation must be considered when comparing data for similar endpoints between studies. Neuronal culture maintained in Neurobasal A + B27 media had several features advantageous for HCI applications including less basal cell death, less cell clustering and neurite fasciculation, and a tendency towards increased sensitivity and selectivity in chemical concentration–response studies. Keywords High-content image analysis . Developmental neurotoxicity . Axon outgrowth . Dendrite outgrowth . Synaptogenesis . Culture media

Introduction Environmental exposure to chemicals, combined with interindividual susceptibility factors, may be playing a role in the increased incidences of neurodevelopmental disorders observed in industrialized countries (Jacobson and Jacobson 1996; Grandjean and Landrigan 2006; Bellinger 2009). Given this observation and the inefficient nature of traditional in vivo developmental neurotoxicity (DNT) guideline studies (USEPA 1998; OECD 2007), the need exists for more efficient methods of DNT chemical hazard identification and characterization. A proposed solution for this problem is high- and medium-throughput chemical screening strategies using in vitro models, including neuronal cell lines and primary neural cultures, which recapitulate the critical processes of

HARRILL ET AL.

nervous system development (Coecke et al. 2007; Lein et al. 2007; Crofton et al. 2011). The utility of high-content imaging (HCI) as a tool to assess chemical effects on biological processes critical to nervous system development has been demonstrated for neural progenitor cell proliferation, apoptosis, neurite outgrowth, neurite differentiation, and synapse formation (Breier et al. 2008; Radio et al. 2010; Harrill et al. 2011a, b; Culbreth et al. 2012; Krug et al. 2013). The results of in vitro chemical screening using these approaches could be used to prioritize chemicals for targeted in vivo DNT studies (Crofton et al. 2011). Culture media can influence the growth and survival of neural cells in vitro. It is, however, unclear how different media alter the response of neural cells to toxicant exposure. We have used HCI to compare total neurite length and the total number of neurons per field (i.e., a measure of cell death) following chemical treatment (Harrill et al. 2010, 2011a, 2013). This comparison allows one to determine if an observed decrease in neurite length is due to cell death or a specific effect of the chemical on the neurodevelopmental process of neurite outgrowth. Parallel analysis of HCI endpoints has also been used to gain insight into the cellular mechanisms underlying the toxicity response. For example, parallel analysis has been used to discriminate between chemical effects on axon or dendrite outgrowth (Harrill et al. 2013). Also, parallel analysis has been used to determine whether decreases in total synapse number are due to decreases in dendrite length or decreases in synaptic density along the dendrite or both (Harrill et al. 2011b). It is unclear how changes in the type of culture media used would affect selectivity and characterization of cellular mechanisms in HCI-based DNT screening studies. A milestone in the fields of in vitro neuroscience and neurotoxicology was the introduction of serum-free defined media which could support the growth and survival of primary neurons in culture (Snyder and Kim 1979; Romijn et al. 1982, 1984). The impetus for development of serum-free media arose from observations of batch-to-batch variability in the growth-stimulating capacity of commercially available serum preparations, uncertainty regarding the chemical composition of serum products, and the need to utilize mitotic inhibitors in serum-containing cultures to reduce glial cell growth (Romijn et al. 1984; Price and Brewer 2001). Both defined and serum-containing neuronal media can support the growth of primary neurons for time periods (e.g., 1 to 4 wk) necessary for studying biological processes like synaptogenesis which occur relatively later in the time course of in vitro neurodevelopment. Thus, both types of media can be used in HCI-based DNT assays focused on synapse formation. However, both media types have some disadvantages. The amount of basal cell death has been shown to be higher in serum-containing media as compared to defined media tailored to support the growth of primary rodent neurons

(Brewer and Cotman 1989; Brewer 1995). Differences in calcium signaling and neuronal excitability have also been observed between cultures maintained in defined versus serumcontaining media, with the former type being less sensitive to excitatory stimuli (i.e., excitotoxicity) and containing less basal neuronal activity (Velasco et al. 2003; Stoppelkamp et al. 2010). It is unclear how these differences would influence the output of HCI-based DNT assays. The purpose of this study was to characterize the effects of two common media for the culture of primary neurons (i.e., DMEM + 10% serum vs. Neurobasal A + B27) on neuronal growth and morphology. Concentration–response experiments were performed using a set of chemicals we have shown previously to alter these endpoints. The results were compared quantitatively to determine if different types of media affect the sensitivity, selectivity, and characterization of cellular mechanisms in HCI-based screening assays which use parallel endpoint analysis.

Materials and Methods Materials. GIBCO® DMEM, Neurobasal A media, and B27 supplement were purchased from Life Technologies (Grand Island, NY). Bisindolylmaleimide I (Bis-I), mevastatin, InSolutionTM K252a, anti-microtubule associated protein 2 (MAP2) mouse monoclonal antibody (AB3418), antisynapsin rabbit polyclonal antibody (AB1543P), and bisbenzamide H 33528 (Hoechst stain) were purchased from EMD Millipore, Inc. (Billerica, MA). Sodium orthovanadate (Na3VO4), potassium chloride (KCl), and poly-D-lysine were purchased from Sigma-Aldrich, Inc. (St. Louis, MO). U0126 was purchased from Promega, Inc. (Madison, WI). Anti-βIIItubulin rabbit polyclonal antibody (PRB-435P) was purchased from Covance, Inc. (Princeton, NJ). HyCloneTM horse serum was purchased from Thermo Scientific (Waltham, MA). Cell culture. Pregnant Long-Evans rat dams were purchased from Charles River Inc. (Wilmington, MA) and housed as described (Harrill et al. 2011b). The facility was approved by the American Association for Accreditation of Laboratory Animal Care (AAALAC), and all animal care and tissue harvesting procedures were approved in advance by the US EPA, NHEERL Animal Care and Use Committee. Primary mixed cortical cultures were prepared from postnatal day 0 Long-Evans rat pups as described (Harrill et al. 2011b). Briefly, neocortices were removed, dissociated with trypsin, and resuspended in Dulbecco’s modified Eagle’s media (DMEM) containing (in mM) GlutaMAXTM-I (2) D-glucose (25), sodium pyruvate (1), HEPES (10) plus penicillin (100 U/mL), streptomycin (0.1 mg/mL), and 10% horse serum, pH=7.4. Cultures were plated in Corning® 96-well polystyrene cell culture plates coated with poly-D-lysine. Cells

MEDIA FORMULATION AND NEURONAL MORPHOLOGY

were allowed a 2-h attachment period. At 2 h, medium was removed by gentle aspiration with a pipette and replaced with either DMEM + 10% serum as described above or with Neurobasal A + B27 (NBA + B27) serum-free, defined media. For wells switched to NBA, an intermediate rinse with the defined media was performed before dispensing the final volume of media into each well. Cultures were maintained in a humidified incubator at 37°C with a 95%/5% CO2 atmosphere. For synaptogenesis experiments, both DMEM and NBA cultures were supplemented with cytosine arabinoside (5 μM) at 5 d in vitro (DIV) to inhibit glial growth. In time course studies, medium was changed every 3 d. In concentration–response studies, medium was changed concordant with dosing of cells as described below. For experiments measuring neuritogenesis, both DMEM and NBA cultures were seeded at 5×103 cells/well (1.56×104 cells/cm2). For the purpose of this work, neuritogenesis corresponds to the first 24 h of neurite outgrowth following seeding and attachment of cultures. For experiments measuring axon and dendrite growth, both DMEM and NBA cultures were seeded at 1×104 cells/ well (3.12×104 cells/cm2). For experiments measuring synaptogenesis, DMEM cultures were seeded at 4×104 cells/well (1.25×105 cells/cm2), and NBA cultures were seeded at 1× 104 cells/well (3.12×104 cells/cm2). Seeding DMEM cultures at higher densities was necessary in synaptogenesis experiments to account for the higher basal rate of neuronal loss in DMEM as compared to NBA cultures and to ensure that similar numbers of neurons were present in each culture type at time points past 6 DIV. Chemical treatment. The chemicals used in this study have been shown to have specific effects on neurodevelopmental endpoints studied using HCI (Harrill et al. 2011a, b, 2013) and were selected to investigate the effects of media type on HCI endpoints. Stock solutions (1,000×) of the highest concentrations of Bis-1 (10 mM), U0126 (30 mM), and mevastatin (30 mM) were prepared in pure DMSO. K252a was received as an InSolutionTM product solubilized in DMSO at a concentration of 1 mM. A series of 1,000× stock solutions for the remainder of the concentrations tested for each of these chemicals were prepared by dilution in DMSO. Dosing solutions (10×) for each test concentration were then prepared by diluting 1,000× stocks in either DMEM or NBA media. A 10× dosing solution for the highest concentrations of Na3VO4 (1, 000 μM for neuritogenesis and axon/dendrite outgrowth; 100 μM for synaptogenesis) and KCl (300 mM) were prepared by dissolving chemical directly into DMEM or NBA containing 1% DMSO. The remainder of the 10× Na3VO4 and KCl dosing solutions were prepared by serial dilution in cortical media containing 1% DMSO. Cultures were exposed to chemicals by diluting a volume of 10× dosing solution 1:10 in the respective treatment wells to achieve the final nominal media concentrations cited throughout the text. Final DMSO

concentration for all treatments was 0.1%. A concentration of 0.1% DMSO was also used as a vehicle control. In neuritogenesis time course studies, cells were analyzed at either 6, 12, 18, or 24 h. In experiments examining chemical effects on neuritogenesis, chemicals were added at 4 h and cells were analyzed at 24 h. In axon/dendrite outgrowth time course studies cells were analyzed at either 3, 5, or 7 DIV. In experiments examining chemical effects on axon/dendrite outgrowth, chemicals were added at 4 h and cells were analyzed at 120 h (i.e., 5 DIV). In synaptogenesis time course studies, cells were analyzed at either 6, 9, 12, 15, or 19 DIV. In experiments examining chemical effects on synaptogenesis, chemicals were added at 9 DIV, and cells were analyzed on 15 DIV. Immunocytochemistry. Immunocytochemistry was performed as described with modifications (Harrill et al. 2013). Cells were fixed by addition of 100 μL of warm phosphatebuffered saline (PBS) fixative solution containing 8% paraformaldehyde, 8% sucrose, and 3 μg/mL Hoechst 33258 directly to each well without media aspiration. The volume of fixative solution was equivalent to the volume of media contained in each well. This method preserved the fine architecture of neurites throughout the culture. Following a 20-min fixation period, cells were washed with immunocytochemical staining buffer (ISB) (Harrill et al. 2011b). Primary antibody solutions were prepared in ISB as follows: rabbit anti-βIII tubulin (1:2, 000) for neuritogenesis assays, rabbit anti-βIII tubulin (1:2, 000) + mouse anti-MAP2 (1:800) for axon/dendrite outgrowth assays, and rabbit synapsin (1:500) + mouse anti-MAP2 (1:800) for synaptogenesis assays. Cells were incubated for 1 h at room temperature with primary antibody solutions. The cells were then rinsed with ISB and incubated with appropriate Alexa Fluor®-conjugated antibodies diluted 1:500 in ISB for 1 h at room temperature, protected from light. Cells were then rinsed with ISB, rinsed in PBS, sealed with optical adhesive film, and stored at 4°C prior to imaging. The final PBS rinse was retained in each well as a storage buffer. Microscopy and high-content image analysis. Qualitative images were obtained using a Leica DMI6000 inverted fluorescence microscope with either a ×10 or ×20 objective. The Cellomics® ArrayScan VTI was used for automated highcontent image acquisition and analysis. Components of this imaging system were as previously described (Harrill et al. 2010). A total of between 9 and 12 unique fields-of-view were acquired and analyzed within each well. The Cellomics® Neuronal Profiler BioApplication was used to develop algorithms for measuring neuritogenesis, axon/dendrite outgrowth, and synaptogenesis. For neuritogenesis assays, the BStaging Method^ algorithm as described in Harrill et al. (2013) was used. Quantitative endpoints for neuritogenesis include the average number of neurons per field, total neurite

HARRILL ET AL.

length per neuron, and the percentage of stage 1, stage 2, and stage 3 neurons as described for primary cortical neurons (de Lima et al. 1997). For axon/dendrite outgrowth assays, the BSubtraction Method^ algorithm was used (Harrill et al. 2013). Quantitative endpoints for this assay include the average number of neurons per field, total neuron length per neuron, total axon length per neuron, and total dendrite length per neuron. For synaptogenesis assays, the algorithm as described in Harrill et al. (2011b) was used. Quantitative endpoints for synaptogenesis include the average number of neurons per field, total dendrite length per neuron, the number of synapses per cell body, the number of synapses per dendrite length, and the number of total synapses per neuron. Synapsin-positive puncta in close proximity to either a MAP2+ cell body or MAP2+ dendrite are used to quantify synapses. In each assay, the number of neurons per field can decrease in response to chemical treatment and is considered a measure of cell death. For optimization of each algorithm, untreated DMEM and NBA cultures were immunolabeled at various time points within the assay window, and algorithms were visually assessed for accuracy of neurite tracing and spot detection in the case of synaptogenesis assays. Identical tracing parameters were used in each assay for both DMEM and NBA cultures. Data analysis and statistics. Statistics were performed using GraphPad Prism® v5 (La Jolla, CA). For concentration– Figure 1. Qualitative comparison of neuritogenesis in DMEM and NBA + B27 media. Primary cortical cultures were plated at a density of 1.56×104 cells/cm2 and cultured in DMEM+10% serum or NBA + B27 media. Cells were then fixed and immunolabeled with an antibody against βIII-tubulin (red) at 6 (A, B) or 24 (C, D) h in vitro. Arrowheads = examples of stage 1 neurons. Arrows = examples of stage 2 neurons. Double arrows = examples of stage 3 (i.e., axonbearing) neurons. Scale bar = 50 μm.

response analyses (except neuronal staging in the neuritogenesis assay), data for each endpoint were normalized to vehicle control values and analyzed as percent changes from control. For neuronal staging data, the percent of neurons in each stage was normalized to the total number of neurons analyzed in each condition in order to derive a normalized population distribution. Details of statistical tests, the number of wells measured, and the number of replicate cultures are described in figure captions and tables for each experiment. Lowest observable effect levels (LOELs) were defined as the lowest concentration of chemical in which a statistically significant difference from control was observed. Effective concentrations (EC30) were defined as the concentration of chemical producing a 30% decrease from control values. EC30 values were determined using empirical point-to-point functions. The point at which the response dropped below 30% of control and remained below that level was used in the analysis.

Results Time course of neuritogenesis. Representative images of cortical neurons grown in DMEM or NBA + B27 during the initial period of neuritogenesis and neurite outgrowth are

MEDIA FORMULATION AND NEURONAL MORPHOLOGY

shown in Fig. 1. Neurons were immunolabeled with βIII-tubulin (red) to visualize neurites and cell bodies. At 6 h in vitro, a large number of neurons had developed one or more immature neurites in both DMEM and NBA + B27 cultures (i.e., stage 2 neurons, arrows). Still other neurons had not developed any neurites (i.e., stage 1 neurons, arrowheads). By 24 h in vitro, both DMEM and NBA + B27 cultures contained axon-bearing neurons (i.e., stage 3 neurons, double arrows) as well as a number of stage 1 and stage 2 neurons. Axons appeared much longer than secondary neurites extending from the same cells. Our previous work utilized an HCI algorithm (i.e., the staging method) to quantify total neurite length per neuron and the percentage of stage 1, stage 2, and stage 3 neurons in cortical cultures during early phases of in vitro development (Harrill et al. 2013). This algorithm was used to quantify and compare the growth characteristics of neuronal cultures seeded at the same density and grown in either DMEM or NBA + B27 media (Fig. 2). There was no difference in the average number of neurons per field between culture types. Also, there were no observable decreases in the number of neurons per field at time points up to and including 24 h in vitro (Fig. 2A). Total

neurite length per neuron increased in both culture conditions during this time period (Fig. 2B). A significant interaction of time × media was observed (F3.88 =23.26, p

Media formulation influences chemical effects on neuronal growth and morphology.

Screening for developmental neurotoxicity using in vitro, cell-based systems has been proposed as an efficient alternative to performing in vivo studi...
5MB Sizes 2 Downloads 7 Views