As with any cell-based test system, determining the type of cell(s) to be used is critical. For example, a commonly used neuroblastoma cell line (Neuro-2a) was less sensitive to some neurotoxicants than cultured primary neurons (LePage et al., 2005). The authors suggest caution in interpreting neurotoxicity data obtained from tests using this transformed cell line. Tamm, et al., (2006) reported that neural stem cells "are more sensitive than differentiated neurones or glia" to methlymercury exposure. Both neural progenitor cell lines and primary cultures have been utilized in published reports. Various laboratories have reported that primary cells are not always more sensitive than cell lines (Costa 2007; Harrill et al., 2011), and that differences may be due to different culture conditions, such as the presence of serum. Any cell system, whether a primary or transformed cell line, or neuroprogenitor cells, needs to be characterized and preferably compared to primary cells and/or the in vivo response for the specific toxicity endpoint being assessed.
In addition to defining the appropriate cellular models, predictive in vitro toxicity assays require that neurotoxicity specific assay endpoints be determined and used in appropriate models, and in a combination (test battery/test strategy) that provides an overall response predictive of the in vivo human outcome. Commonly used cellular endpoints that are not neural specific include cytotoxicity, proliferation, migration, differentiation, and apoptosis (Culbreth et al., 2012; Moors et al., 2009). Some endpoints that have been described as relevant to cell-based neurotoxicity assays include electrical activity, neurotransmitter release, and neurite outgrowth (Bal-Price et al., 2008a; De Groot et al., 2013; Radio & Mundy, 2008).
The roles of the blood-brain barrier (BBB), metabolism, and toxicokinetics need to be included in assessing the potential neurotoxicity of a substance. Recent workshops have discussed how the results from in vitro assays might be coupled with (Q)SAR studies, computational modeling, and other techniques to form an integrated testing strategy for the prediction of neurotoxicity (Bal-Price et al., 2008a; 2008b).
A major hurdle in replacing animal tests is identification of one or more sets of in vitro models and assay endpoints that will provide an overall result predictive of in vivo neurotoxicity. Many test batteries have been proposed for assessing neurotoxicity (Gartlon et al., 2006; Harry & Tiffany-Castiglioni, 2005; Suñol et al., 2008), but this continues to be a challenge.
An assessment by ECVAM researchers of an in vitro testing strategy composed of undifferentiated and differentiated PC12 cells and primary cerebellar granule cells for the test’s ability to distinguish between cytotoxic and neurotoxic substances (Gartlon et al., 2006). The different cell systems responded differently to the multiple endpoints evaluated. However, the conclusions were that "further work is required to determine suitable combinations of cell systems and endpoints capable of distinguishing neurotoxicants from cytotoxicants." The evaluation of other test batteries of various combinations of cells and non-neural specific endpoints to distinguish neurotoxicants from non-neurotoxicants has provided mixed results (Breier et al., 2008; Costa et al., 2007). The inclusion of neural-specific endpoints such as neurite outgrowth might increase the predictive power of such test batteries (Radio et al., 2009; Radio & Mundy, 2008).
Another approach involved investigating the effect of metabolism on drug toxicity to cultured liver and neural cells. Four drugs were tested in metabolically competent mouse hepatocytes and human hepatoblastoma (HepG2) cells and in neuroblastoma (SH-SY5Y) and astrocytoma (U-373 MG) cells (Mannerström et al., 2006). The researchers reported "better estimations of neurotoxicity can be made by the combined use of metabolically competent hepatocytes and glial cells (e.g. U-373 MG) together with neuronal cells (e.g. SH-SY5Y)." The same experimental approach to evaluating post-metabolism cytotoxicity to other organ systems, including human neural cells, has been available for several years with the use of the integrated discrete multiple organ cell culture (IdMOC) system (Li et al., 2004).
An ongoing problem in interpreting the results from in vitro experiments is the extrapolation of the results to the in vivo situation. Two recent experiments indicate the importance of determining intracellular concentrations of test substances in in vitro experiments. Mundy, et al. (2004), examined the intracellular concentration of the lipophilic polybrominated diphenyl ethers (PBDEs) using cultured neuronal and glial rat cells. Results showed a magnification of the applied doses – 100-fold for a 1 µM exposure of 60 minutes. Many experimental factors affected the intracellular concentrations of PBDE, including serum in the media, total volume of the media, and duration of exposure. Investigators concluded that media concentrations significantly underestimated cellular concentrations and that tissue concentrations are the relevant parameter and should be determined for in vitro experiments.
Animal tests sometimes expose the test subjects to very high concentrations of chemicals; however, the concentration of neurotoxicants found in brain tissues is typically parts per million (ppm). Cells are exposed to concentrations that range from nontoxic to cytotoxic (causing cell death), usually in the micromolar (µM) range. Meacham, et al. (2005), addressed the question of whether the in vitro doses provided comparable intracellular concentrations to those found in vivo for certain lipophilic neurotoxicants. Intercellular accumulation of two compounds in three in vitro neuronal tissue models was tissue, time, and concentration-dependent and in the ppm range, leading to the conclusion that "tissue levels rather than exposure concentrations are a more appropriate metric for comparison of in vitro to in vivo effects."
A test of unique interest to CNS toxicology is BBB permeability. If a substance (or its metabolites) cannot penetrate the capillary bed separating the brain from the rest of the circulation (the BBB), then it cannot be toxic to the brain. Tähti, et al. (2003) and the report from an ECVAM workshop (Prieto et al., 2004), reviewed the state of in vitro BBB models. Various species of microvessel endothelial cells and endothelial cell lines were cultured as monolayers on membranes to model the BBB. Co-culture models, which contain cells of more than one type, have also been studied (Figure 1). Astrocytes were found essential for brain microvessel endothelial cells to form a good barrier. Recent attempts to co-culture human microvascular endothelial cells and astrocytes as BBB models have been successful (Cucullo et al., 2007; Siddharthan et al., 2007).
Figure 1. Schematic representation of a co-culture blood-brain barrer model. (Source: Prieto et al., 2004, Altern. Lab Anim.)
Neural progenitor (stem) cells are showing many favorable properties as cellular models for neurotoxicity testing. Breier, et al. (2009) describe the recent availability of rodent and human neuroprogenitor cells for neurotoxicity cellular models, and discuss their potential advantages over primary neuronal and transformed cell lines.
Neural stem cell lines have been developed from human umbilical cord blood. These stem cells could be induced to differentiate into neuronal, astrocytic, and oligodendroglial lineages (Buzanska et al., 2005). Various cellular toxicity endpoints were evaluated in 2D and 3D neurosphere cultures. Alterations in growth factors and the extracellular matrix were found to modify the proliferation and migration of attached neurosphere precursor cells. To further refine the neurosphere model, microarray wells coated with different extracellular components were evaluated for their effect on neurosphere cell adherence and differentiation (Buzanska et al., 2009).
Pluripotent human embryonic stem cells have also been differentiated into normal human neurons (Stummann et al., 2009; Zeng et al., 2006). Researchers claim these cells show potential as in vitro models for the study of developmental biology and embryotoxicity. Both human embryonic stem cells (Harrill et al., 2010) and human induced pluripotent stem cells (Xu et al., 2013) have been differentiated into neurons and used to assess neurite outgrowth.
New assay endpoints for identifying neurotoxicants and developmental neurotoxicants are being explored by researchers. For example, Mundy, et al., (2008) looked at changes in cell morphology and protein biomarkers expressed during neurite outgrowth and synaptogenesis using cerebellar granule cells as a model for neural development. Neurotypic proteins expressed during normal cell growth were characterized, and the proteins perturbed by chemical inhibition of cell growth (biomarkers) were identified. The researchers found that “neurotypic proteins can be used as biomarkers of neuronal development in vitro, and in some cases, may detect changes that are not apparent using morphologic measures.” In another novel approach, reaggregating brain cell cultures were coupled with mass spectrometry-based metabolomics to obtain cellular metabolic profiles and measure neurotoxicant-induced perturbations (van Vliet et al., 2008). Metabolites were identified that are potential biomarkers for neurotoxicity. The same approach to test target organ specific compounds supported the utility of this in vitro metabolomics approach in detecting neurotoxic substances.
High-throughput and high content cellular assays have been developed for screening large numbers of compounds in multiple types of cells and for multiple endpoints. High-throughput cellular assays using neuroprogenitor cells were developed as a screening assay for developmental neurotoxicant effects on proliferation (Breier et al., 2008; 2009; Culbreth et al., 2012). The researchers concluded their data were useful in identifying neurotoxicants, but that further information on stem cell differentiation and a larger data set to define the sensitivity and predictivity of assays are needed (Breier et al., 2009). Similarly, high-throughput assays were developed for neurite outgrowth and cytotoxicity using the PC12 cell line as well as rodent primary neuronal cultures. Neurite outgrowth was visualized by immunolabeling of cytoskeletal proteins with various markers (e.g. β-tubulin), and quantified using automated microscopy and image analysis (Figure 2). The PC12 cell line assay was able to detect many, but not all, developmental neurotoxicants (Radio et al., 2008). Further work using high-throughput/high content cellular assays compared changes in neurite outgrowth in response to chemicals in the PC12 cell line with primary cerebellar granular cells, and has explored the use of rodent primary neuronal cultures for examining toxicant effects on axon outgrowth, dendrite outgrowth, and synaptogenesis (Radio et al., 2009; Harrill et al., 2011; Harrill et al., 2013).
Figure 2. Neurite outgrowth in PC12 cells analyzed using automated microscopy and image analysis: A) cell body and neurites visualized using immunofluorescent staining for β-tubulin; B) neurite outgrowth quantified using image analysis software (Courtesy of William R. Mundy, USEPA).
Compartmentalized microfluidic devices and/or microelectrode arrays have been developed to study the mechanism of neurotoxicant effects on neural cells where individual neurons in a network can be exposed or stimulated (Ravula et al., 2007; Suzuki & Yasuda, 2007; Yang et al., 2009). These devices can be used to monitor injury-induced electrical activity and to differentiate injury to the neuronal cell body versus the axon, or to associated glial cells. A number of researchers have used multi-electrode arrays to record electrophysiological cell responses in brain cell cultures, and observed that neurotoxic chemicals evoked electrophysiological perturbations at sub-cytotoxic concentrations (Van Vliet et al., 2007; Valdivia et al., 2014). More recently, the growing awareness that cells in a 3D context maintain physiological signaling much better than cells growing in a 2D format led to the establishment of 3D models that recapitulate more physiological neurodevelopmental functions and can be used for testing (Hoelting et al., 2013). In addition, “brain-on-a-chip” models are being developed with the ability to mimic the in vivo brain microenvironment in an in vitro system (Park, 2014).
The application of in vitro methods for neurotoxicity for regulatory testing purposes has been discussed in workshop/symposium reports (Bal-Price et al., 2008a; 2008b; 2012; Coecke et al., 2007; Lein et al., 2007). These groups reviewed existing alternative methods for neurotoxicity testing and discussed approaches for integrating them into regulatory frameworks. At the present time, in vitro methods “are considered complementary to animal tests because they provide an understanding of the molecular/cellular mechanisms involved in neurotoxicity” (Bal-Price et al., 2008b). All agreed that new test methods must be standardized and formally validated to have an impact on reducing animal use in regulatory testing.
DNT testing has been cited as an area in particular need of alternative test methods, both to better protect human health and to reduce the large numbers of animals currently required for current regulatory tests. A coordinated effort in the form of the TestSmart DNT program was launched to address this need by bringing together international stakeholders from diverse areas to review the science and to propose approaches for incorporating DNT alternatives into regulatory policies. A report from the first DNT Workshop identified future priorities as follows (Lein et al., 2007): “initiating a systematic evaluation of alternative models and technologies, developing a framework for the creation of an open database to catalog DNT data, and devising a strategy for harmonizing the validation process across international jurisdictional borders.”
For some new perspectives on non-animal neurotoxicity test methods, read the following invited commentaries:
Sherry L. Ward, PhD, MBA
AltTox Contributing Editor
William Mundy, PhD
Environmental Protection Agency
Joshua Harrill, PhD
Center for Toxicology and Environmental Health, L.L.C.
AltTox Editorial Board reviewer(s):
William Mundy, PhD
Environmental Protection Agency
Joshua Harrill, PhD
Center for Toxicology and Environmental Health, L.L.C.