**1. Introduction**

The bio-accumulation of marine biotoxins produced by phytoplankton in filter-feeding invertebrates and finfish not only poses significant health threats to consumers but also has detrimental effects on the economies of nations highly dependent on seafood consumption for their subsistence [1–3]. Among these potent marine biotoxins are neurotoxins acting on the voltage gated sodium channels (VGSCs) of excitable cells, namely VGSC activators such as brevetoxins (PbTxs) and ciguatoxins (CTXs), and VGSC inhibitors such as saxitoxins (STXs) and tetrodotoxins (TTXs) [1,2]. Detection and quantification of these groups of marine toxins remain highly challenging due to the wide range of congeners present in trace amounts in contaminated biological matrices [4]. In addition, an official

reference method is still lacking for most of these toxins due to the poor availability of certified reference standards [5]. Although the mouse bioassay has been recommended by the European Food Safety Authority (EFSA) for years until 2015, the European Union now strongly advocates the use of analytical techniques based on liquid chromatography coupled with mass spectrometry in tandem (LC-MS/MS) and high performance liquid chromatography with fluorescence detection (HPLC-FLD) for the detection of lipophilic and hydrophilic toxins, respectively [6–10]. However, the need for alternative high throughput methods for toxin screening and quantification purposes remains [8,11]. Among the readily available and most widely used in vitro methods, the functional assay known as the cell-based assay (CBA) that uses a neuroblastoma (N2a) cell line appears as the most promising one [2,12].

Based on the mode of action of VGSC toxins, the CBA-N2a was initially developed to detect two VGSC inhibitors, STXs and TTXs [13]. Those marine toxins have no cytotoxic effect on N2a cells and their detection requires the addition of the sodium/potassium (Na+/K<sup>+</sup>) ATPase pump blocker ouabain (O), together with the sodium-channel activator veratridine (V), which induces permanent activation of the VGSCs [14]. Under this O/V treatment (OV+ conditions), sodium influx resulted in the cellular swelling and subsequent death of N2a cells [13]. Following the addition of STX or TTX, an increase of cell viability is observed as the two inhibitors counteract the effect of ouabain and veratridine. At that time, viability measure was achieved by a visual estimation of the morphological changes of N2a cells and the enumeration of viable cells under an inverted microscope. This assay performed in a 96-well microplate format has been applied to the detection of Paralytic Shellfish Poisoning (PSP) toxin standards [13] and to assess the presence of PSP toxins in biological extracts [15,16]. Standardization was also achieved using a certified STX standard for the detection of PSP toxins in shellfish and dinoflagellate cell extracts using automated endpoint determination with final readout based on crystal violet staining [17]. Further, this CBA-N2a was applied to detect VGSC inhibitors in marine bacterial supernatants using neutral red for final estimation of toxicity [18] and was subsequently extended to the detection of PbTxs and CTXs [19,20]. Contrary to VGSC inhibitors, these two VGSC activators induce a decrease in N2a cell viability in OV+ conditions [19,20], whereas no cytotoxic effect is observed in the absence of O/V treatment [21]. These same authors also introduced a more workable measure of cell viability using the 3-(4,5-dimethyl-2-thiazol)-2,5-diphenyl-2H-tetrazolium bromide (methylthiazolyldiphenyl-tetrazolium bromide) colorimetric assay (known as MTT assay), as modified from previous protocols aiming at assessing either cellular growth and survival [22] or chemosensitivity in established cell lines [23]. The MTT assay is widely used to measure cell viability and proliferation [24–27], and is based on the capacity of mitochondrial dehydrogenase enzymes present in living cells to reduce tetrazolium salts into an insoluble purple formazan product [28] localized in lipids droplets [29]. The readout step was first established using dimethyl-sulfoxide (DMSO) to dissolve formazan products [23], the amount of the formazan produced is directly linked to the number of viable cells remaining at the end of the assay. This estimation is achieved by means of an automated readout. Two patents describing an improved and simplified CBA-N2a were eventually filed, leading to grea<sup>t</sup> expectations among the scientific community that this bioassay could soon be routinely used to detect both activators and inhibitors of sodium channels [30,31]. However, neither commercial kit nor detailed protocol is available to date.

Briefly, the CBA-N2a originally set by Manger et al. [19,20] consists of three major steps; (i) cell seeding of 100,000 cells/200 μL/well in a 96-well microplate left to grown for ≈ 24 h, (ii) O/V treatment followed by cell layer exposure to toxin standards and/or biological samples during 24 to 48 h, depending on the targeted toxin activity (final reaction volume of 230 μL/well) and (iii) N2a cell viability readout at 570 nm using the MTT colorimetric assay. Practically, the detection of VGSC activators and inhibitors by CBA-N2a requires 3 and 4 days, respectively.

Since 1993, the CBA-N2a has been widely used in a number of studies for the detection of VGSC toxins and related toxin families. However, a comprehensive review of the literature shows that the protocol initially defined by Manger et al. [19,20] has undergone numerous

changes: for instance, (i) up to 13 different cell seeding densities have been tested, ranging from 10,000 to 250,000 cells/well [32–88] suggesting that cell confluence likely varied substantially between studies, especially when estimated by eye measurement; (ii) the culture medium established for cell layer implementation, which formerly used 10% of fetal bovine serum (FBS), was often reduced to 5% in many studies [32,33,36–39,42,46,48,54,59,69,70,74,77,78,81–83,85,86,89]; (iii) the confluence level reached by N2a cells after 24 h growth varied from non-confluent cells [66] to >90% confluence [48,59,63–65,72,73,77,78,83] or even was not specified in some studies [33–38,40–43,47,49–58,61,66,67,69,70,74–76,79–82,85,87–90].

Other steps of the CBA-N2a have also been the subject of substantial modifications. For instance, although many studies used a 500/50 μM (1:10 ratio) for O/V treatment [32–49,76,89,91,92] as initially defined by Manger et al. [19,20,30,31], numerous changes in O/V concentrations [54–57,59,61,63–65,71–75,77–83,85,87,93–95], O/V ratios [50–52,69,86] and reaction volumes [32,33,35,36,49,51,53,59,61,63,64,69,73,75,77–80,84,86–90] were further proposed. Interestingly, in many of these studies, the effects on cell viability resulting from all these modifications were rarely addressed but, when reported, showed incongruent results. For instance, O/V treatment at 500/50 μM was shown to induce highly variable effects between 20–92% [34,45,68,96]. Conversely, similar effects (10–56%) were reportedly obtained at lower O/V concentrations [50,86]. The most significant modification was the use of two different O/V treatments, i.e., 100/10 and 300/30 μM, in order to induce 20% and 80% of cell mortality, respectively, for a more reliable assessment of the effects of VGSC toxin activators and inhibitors on N2a cells [83]. Several changes were also made to the final reaction volume (230 μL) at the last step of the CBA-N2a, i.e., cell exposure to toxin standards or biological samples, with volumes varying from 100 to 210 μL [32–36,45,49,51,53,59,61,63,64,69,73,75,77–80,84,86–90,95]. Moreover, measuring cell viability by means of the MTT colorimetric assay is classically performed at 570 nm, however, different wavelengths were also tested that ranged from 490 to 595 nm [33,35,36,38–42,45,50,56,57,61,64,67–69,71,75,76,79,80,84,85,87,88,90,93,97]. Finally, water-soluble tetrazolium reagents were sometimes preferred to the former water insoluble formazan product, with the use of WST-8 [49,60] or commercial kits such as CellTiter 96® Aqueous One solution [33,38,69,86], Cell Counting Kit-8 [47] or XTT Cell Proliferation Assay kit [98].

This wide range of "in-house" methods described in the literature highlights the present lack of a consensus, standardized protocol for CBA-N2a, making any attempt to compare CBA data between assays and/or laboratories difficult or impossible. It also greatly hampers all current efforts to establish the CBA-N2a tool as a potential alternative reference method to LC-MS/MS. In this context, the present work aims at a comprehensive revisit of the CBA-N2a by evaluating the effects of six key parameters that are critical in obtaining reliable toxicity results, i.e., cell seeding densities, cell layer viability after 26 h growth, MTT incubation time, O/V treatment and solvent and matrix effects. To evaluate its robustness, the newly improved protocol thus established was further applied to the detection of toxins active on VGSC families (activators vs. inhibitors), and the estimation of CTX-like toxicity in fish samples of known ciguatoxic status.
