Neuronal cell models have been tested for their use in predicting in vivo effects of different neurotoxic substances [1, 2, 4, 5]. Attempts have been made to develop and to utilize these in vitro neuronal models to study the mechanisms of toxicity due to chemical and biological compounds at cellular and molecular levels. Moreover, these models have also been tested for their use in rapid screening of potential neurotoxicants out of which positive compounds would be selected for in vivo evaluation. Prior studies using in vitro cellular models were intended to generate preliminary mechanistic and toxicity information while reducing animal use and associated high cost of in vivo testing. The following are the three different types of in vitro cellular models primarily used in biomedical research; (1) primary cell cultures, (2) clonal cell lines, and (3) neural stem cells. The main advantage of using primary cell cultures is that they retain the morphological, neurochemical, and electrophysiological properties of neurons in situ. However, the disadvantages of primary cell cultures include (a) a limited life span, (b) increased genetic variability between model systems and cultures, (c) mixture of different neuronal populations in each preparation, as well as (d) high resource requirements . Neural stem cells have the ability for self renewal and generating multiple cell types representing different parts of the nervous system; moreover, they can be derived from humans. However, as with primary cultures, the stem cells require animal use resulting in high resource requirements and regulatory issues. After neural stem cells have reached the end of their short variable life span, a new line has to be generated and then characterized before use [2, 32]; this is neither time nor cost effective.
A clonal cell line is defined as a population of cells that originated from a single source and can be maintained in culture for an extended period of time. A clonal cell culture has a number of advantages that make them useful as in vitro models: easy to obtain; relatively easy to grow; divide rapidly; and can be continuously subcultured for a relatively high number of passages to provide a large number of cells in a short period of time . The clonal M17 neuroblastoma cell line used in this study has the characteristics described above as well as the ability to become differentiated into a neuroblastic (N) cell when cultured in the presence of RA for several days [11, 14]. These properties make the M17 cell line a good in vitro cell model for mechanistic and neurotoxicity testing. However, the functional changes in M17 cells due to RA differentiation have not been thoroughly characterized. A very relevant question is why do we need a differentiated neuronal model for neurobiology studies. The answer is that most of the neuronal functions such as membrane excitability, ion channels, neurotransmitter release, endocyctotic and exocyctotic events etc. are characteristics of mature neurons and cannot be studied in an immature neuronal model.
Treatment with RA results in the progressive development of neuronal morphologies and formation of neuronal networks seen in a more mature neuronal culture over time. A commonly measured characteristic of differentiation is extension of neurites that are akin to the axons and dendrites of fully differentiated neurons . These neuronal properties in RA differentiated M17 cells were evidenced by both light microscopic observations (Figure 1) and immunofluorescence staining (Figure 2B - D). Although differentiated M17 cells demonstrated evidence of maturing neuronal organization and properties, the functional verification of synaptic activity remains to be done.
Besides morphological characteristics, we also observed differences in expression of several neuron specific proteins between undifferentiated and differentiated M17 cells (Figure 4). The presence or levels of certain proteins can vary between immature and mature neurons. One such protein is neuron specific enolase (NSE) which is responsible for generating phosphopyruvate hydratase that participates in glycolosis/gluconeogenesis; the levels of NSE increase as the neurons mature. While undifferentiated M17 cells do express NSE, its level increases due to differentiation. The formation of neurite-like processes as a part of synaptic organization and activity can be further characterized with the differential expression of the neurofilament proteins, NF-M, and –H that help form the neurofibrils within axons . Developing neurons generally do not express either of these neurofilament proteins until they become post-mitotic, which is fairly late in development. Another neurofilament subunit, vimentin, decreases as neurons mature . We were able to detect vimentin (Figure 4C) and the neurofilament proteins -M as well as -H (data not shown). We observed a decreased level of vimentin, whereas neurofilaments H and M increased due to differentiation. This might be an indication that under the conditions used, M17 cells could be in an early stage of maturation. This hypothesis is supported by the wide-spread expression of the immature neuronal marker β3-tubulin and the accumulation of synapsin-1/2 at the tip of the growth cone (Figure 3). The presence of synapsin within the growth cone is consistent with studies suggesting an axonogenic role during neurite extension and branching, which is a early aspect of neuronal maturation [35, 36]. The levels and localization of these developmentally staged proteins are anticipated to further change during prolonged culture in the presence of RA.
Since M17 cells are multipotential with regard to neuronal enzyme expression, we looked at the effects of RA differentiation on the expression of the main isoforms of acetylcholine (ACh) receptors (M1 mAChR, nAChR – α7) and choline acetyltransferase (ChAT) to determine the type of neurons that RA differentiated M17 cells could be. In Figure 5, nAChR-α7 was the only one of the mentioned proteins that was able to be detected. This indicates that the RA differentiated M17 cells are not cholinergic but would most likely be involved in post- and pre-synaptic excitation in the brain and not post-ganglion nerves in the CNS or exocrine glands [37, 38].
The differential expression of other neuronal proteins than those previously described, expression of voltage-gated Ca2+ channels and ionotropic receptors, which ultimately lead to an increase in neurotransmitter release, can be used to confirm neuronal characteristics and neuroexocytosis. The presence of SNAP-25 and synapsin are indicative of the potential to form functioning pre-synaptic compartments that mediate synaptic vesicle fusion with the pre-synaptic membrane and neurotransmitter release under depolarizing conditions. Although immunoblot demonstrated that overall synapsin expression in M17 cells does not significantly change after differentiation with RA (Figure 4B), synapsin-1/2 becomes distributed along processes, with a punctuate appearance (Figure 2 and supporting Additional file 1: Figure S1) and within the growth cone during neuritogenesis (Figure 3). SNAP-25 is a major component of the SNARE complex that is required for the fusion of vesicle to the cell membrane for the exocytosis of neurotransmitters. A two fold increase in the level of SNAP-25 (Figure 4A) was observed. This increase in SNAP-25 level may correlate with the significant (p<0.01) increase in KCl stimulated [3H] glycine release seen in differentiated M17 cells (Figure 6). The increase in the stimulated release as shown in Figure 6 doesn’t look impressive; however, it is quite marked because we are comparing the fraction of the total pool of [3H] glycine that is released in differentiated cells vs. undifferentiated cells. As mentioned earlier, we studied [3H] glycine release because this assay has been utilized successfully in assessment of neurotoxicity in cell culture models [15–18]. Others have studied glutamate release and glutamate induced excitotoxicity in M17 cells  and as such these cells could be suitable to study glutamate neurotoxicity. Since M17 cells have been reported to have a poor GABAergic property  these cells might not be a suitable model for GABA studies. In this report we demonstrated that a representative neurotransmitter function is enhanced in differentiated M17 cells compared to immature cells.
For functional neuroexocytosis, neurons need both the ability to form the SNARE complex and to have functional voltage-gated Ca2+ channels. The ability of Ca2+ and other ions to move across the cell membrane is necessary for excitation and signal transmission between neurons. Therefore, we studied the uptake of Ca2+ in both undifferentiated and differentiated M17 cells. There was no increase in the uptake of radiolabeled 45Ca2+ using varying concentrations of KCl in undifferentiated M17 cells (Figure 7A); a strong increase in the uptake of radiolabeled 45Ca2+ was observed with RA differentiation of M17 cells with a maximum opening of voltage-gated Ca2+ channels at 25 mM KCl. The presence of both N and P/Q type Ca2+ channels was indicated by the 50 – 60% reduction in Ca2+ uptake when conotoxin GVIA (N-type blocker) (Figure 7D) or agatoxin IVA (P/Q blocker) (Figure 7E) were applied to the culture. Using NNC 55–0396, only a small amount of T-type Ca2+ channels were detected (Figure 7C); however, the assay may not be sensitive enough to pick up the small change in intracellular Ca2+ concentration due to the small unitary conductance of T-type Ca2+ channels. Since neuroexocytosis is Ca2+ dependent, the lack of functional voltage-gated Ca2+ channels in undifferentiated M17 cells is detrimental for its use as a cell model for neurotoxicity research. The treatment of M17 with RA for a minimum of 72 hrs may be essential for functional neuronal cultures. It has been postulated that neuronal functions are cell maturation dependent .
In toxicity studies, it is important to look at both morphological and functional changes to determine toxicological mechanisms. It has been shown here as well as in other studies that maturation of neuronal cultures is very important when studying the effects of toxicants . It is known that intracellular Ca2+ is highly regulated and involved in normal cell functions and in toxicological mechanisms. The lack of voltage-gated Ca2+ channel expression in undifferentiated M17 cells could limit their use as a neurotoxicity model. This is supported by our observation that differentiation of M17 cells with RA was required to see the changes in [Ca2+]i following exposure to CG. The [Ca2+]i decrease due to CG is a toxicant response in neuronal cells that can lead to apoptosis and death of neurons [39, 40]. Acquisition in voltage-gated Ca2+ channels in differentiated neurons may be a prerequisite for studying neurotoxicity due to chemicals other than CG.