AO was provided by Water and Eco-Bio (Gunsan, Korea). The solvents and reagents used to obtain methanol extracts from the samples (methanol, acetonitrile [ACN], n-hexane, ammonium formate, and formic acid) were purchased from Sigma-Aldrich (St. Louis, MO, USA). For the analysis of GYM from AO extract, SPE (Oasis HLB 3cc [60 mg] Extraction Cartridges, WAT094226) and ODS (PREP C18 55–105 μm 125 Å) were purchased from Waters (Milford, MA, USA). All solvents used for MS/MS analysis were of liquid chromatography mass spectrometry grade and purchased from Alfa Aesar (Ward Hill, MA, USA). All chemicals and reagents used were of analytical grade.

A schematic diagram of the extraction method is presented in Supplementary Fig. S1. Briefly, the obtained AO (200 g) was added to 2 L of methanol and stirred thoroughly, followed by extraction at 20–23°C for 15 min. This was repeated three times. After each extraction, the mixture was centrifuged at 10,000 rpm speed using a centrifuge to separate the supernatant, which was then collected and stored. The remaining residue was extracted by adding 2 L of methanol. After completing the three extraction cycles, the extract was concentrated to pH 8.5 using NH₄OH, and solvent partitioning was performed using dichloromethane (CH₂Cl₂) in a 1 : 3 ratio of extract to solvent. The layers were thoroughly separated, and the CH₂Cl₂ layer was collected and concentrated. The concentrated CH₂Cl₂ layer was dissolved in 80% MeOH (+20% H₂O containing 0.1% formic acid), and solvent partitioning was performed using n-hexane. The aqueous layer was collected, concentrated, and used as the final sample (AO fraction, AOF).

To mitigate the matrix effect, that can distort analytical signals due to impurities or interfering substances in the sample, thereby improving the accuracy and reproducibility of the analysis, we used ODS and SPE (Supplementary Fig. S2). Briefly, ODS was packed into an open-column chromatography setup with a diameter of 50 mm and height of 300 mm, filled to a height of 200 mm. The column was conditioned with 100% methanol and equilibrated with 40% ACN containing 2 mM ammonium formate. The AOF was prepared at a concentration of 10 mg mL⁻¹ in 100% methanol, and 5 mL of this solution was loaded onto the column. A mobile phase consisting of 40% ACN with 2 mM ammonium formate was used to elute the sample. The eluted fractions were concentrated using a rotary evaporator and analyzed using UPLC-Q-TOF MS/MS.

Purified AOF was analyzed using UPLC-Q-TOF MS/MS at the Jeonbuk National University Center for University-Wide Research Facilities (Jeonju, Korea). A Xevo G2-XS Q-TOF (Waters, Milford, MA, USA) was used as the analytical instrument. The column was ACQUITY UPLC BEH C8 column (2.1 × 100 mm, 1.7 μm), and mobile phase consisted of 10 mM ABC DW-10 mM ABC 90% ACN with a gradient method (flow rate 0.2 mL min⁻¹; 0 min 100 : 0 v/v; 0–5 min 100 : 0 v/v; 5–30 min 75 : 25 v/v; 30–35 min 50 : 50 v/v; 35–40 min 0 : 100 v/v; 40–45 min 0 : 100 v/v; 45–45.1 min 100 : 0 v/v; 45.1–50 min 100 : 0 v/v). Gymnodimine mass analysis was performed in positive mode using a Q-TOF MS/MS mass spectrometer. The electrospray ionization-MS/MS conditions were set to a scan range of m/z 100 to 1,200 and a scan rate of 2 spectra/s, and helium was used at a gas temperature of 450°C and 30 L min⁻¹ as the collision gas. The collision energy was set to 20–45 eV. MS and MS/MS chromatograms and spectra were obtained using MassLynx v4.1 SCN888 (Waters Corporation, Wilmslow, UK). Electron transfer dissociation (ETD) was performed according to previously established protocols (Habeck et al. 2024). 1,4-dicyanobenzene was employed as the radical anion source for ETD fragmentation. All experiments were performed using a glow discharge current of 80 μA and a make-up gas flow rate of 50 mL min⁻¹. A capillary voltage ranging from 1.8 to 2.0 kV was applied for the sample analysis with a fixed cone voltage of 50 V. The Trap gas flow was optimized to 15 mL min⁻¹, and the transfer gas flow was set to 0.9 mL min⁻¹. The Trap wave height was adjusted between 0.3 and 0.4 V. The assignment of ETD fragment ions was carried out manually to achieve efficient fragmentation.

Microalgae exhibit diverse genetic and physiological characteristics at the strain level, that is the final stage of taxonomic classification. These variations influence environmental adaptability, growth rate, nutritional content, and metabolite production. In microalgal research, strain-level characterization is crucial, as it affects the biosynthesis of valuable compounds, such as omega-3 fatty acids, pigments, and antioxidants, remarkably affecting industrial applications. A. ostenfeldii (AO) is a marine dinoflagellate recognized for its production of marine biotoxins, including STX, paralytic shellfish poisoning (PSP), and cyclic imine (CI) toxins, such as spirolides (SPX) and gymnodimine (GYM). Numerous AO strains have been identified, with research by Helge Martens et al. (2017) reporting 68 distinct strains in the Netherlands. These strains exhibit variable GYM production, leading to the formation of different derivatives or analogs (Martens et al. 2017). However, limited research has been conducted on the growth rate and toxicity of K-1354 strain. This strain produces marine biotoxins, such as GTX 2/3 and STX (13 dmC) (Kremp et al. 2014). This study aimed to identify the optimal growth conditions for AO based on salinity levels and establish analytical conditions for obtaining new standards. The experiment was conducted under varying salinity conditions while maintaining a constant light intensity, light cycle, and incubation temperature, as outlined in the materials and methods section. Salinity is essential for marine microalgae and influences their growth. Four salinities (5, 10, 20, and 35 psu) were tested. No significant changes in cell counts were observed across all salinity conditions until day 3 (Fig. 1A). Growth increased by 10 psu from day 4, reaching a peak concentration of 18,000 cells mL⁻¹ by day 9. The 5, 20, and 35 psu groups maintained similar cell densities until day 4; however, growth decreased slightly under 35 psu conditions from day 5 onward. In contrast, continuous growth was observed at 5 and 20 psu, with a fourfold difference in cell count compared to that at 10 psu. According to Martens et al. (2016), growth is generally unproblematic at salinities of 6–34 psu. Growth ceased below a salinity of 5 psu, and rapid cell death was observed when transferred from a salinity of 5 to 3 psu. Growth rates ranged from 0.13 to 0.2 d⁻¹, with maximum cell counts reported between approximately 9 × 10³ and 12 × 10³ cells mL⁻¹. Another study reported a maximum cell count of approximately 22,266 cells mL⁻¹ at 20.9°C and salinity of 17 psu, with GYM A derivatives (12-methyl GYM) reaching up to 73.3 pg per cell. A salinity of 10 psu appears to be optimal for rapid and safe AO growth (Salgado et al. 2015). Salinity significantly affects membrane permeability and osmoregulation; high salinity can cause ion imbalance and osmotic pressure changes across the cell membrane, thereby increasing permeability and reducing physical stability (Shetty et al. 2019, Shahid et al. 2020). High salinity promotes the generation of reactive oxygen species within cells, damaging the photosynthetic machinery, reducing photosynthetic efficiency, and negatively affecting cell growth (Bazzani et al. 2021). AO strain No. K-1354 exhibited a growth rate of approximately 0.21 d⁻¹ under conditions of a light-dark cycle of 14 h : 10 h, light intensity of 80 μmol photon m⁻² s⁻¹, temperature of 20°C, and salinity of 10 psu, with a maximum cell count of approximately 40,000 cells mL⁻¹ (Fig. 1B). In medium-scale (20 L) cultures, the relationship between cell quantity and biomass was R² = 0.6822 (Fig. 1C). The positive correlation between cell quantity and biomass indicated that intracellular material content increased continuously with growth. Research focusing on increasing biomass during microalgal cultivation is ongoing to produce intracellular fatty acids and oils for industrial applications (Maltsev et al. 2021). An increase in biomass relative to cell quantity may also increase GYM production.

The extraction and purification of AO samples involved multiple complex and systematic steps that are carefully designed to isolate and concentrate the target compounds from natural substances effectively. First, sonication extraction using 100% MeOH was performed according to the method detailed in Supplementary Fig. S1. This method effectively destroys plant tissues and extracts intracellular substances into solvents (Liu et al. 2022, Lomartire and Gonçalves, 2022). Through this process, 6 L of the AO MeOH extract was obtained, which was sufficient for subsequent analyses. The extracted material was concentrated to dryness using a rotary evaporator, yielding 8.13 g of the AO MeOH extract. The extract was then dissolved in an ammonium hydroxide solution (NH₄OH, pH 8.5) for liquid-liquid partitioning. Partitioning was performed using dichloromethane (CH₂Cl₂) to obtain an organic layer. This extraction was repeated at least three times to ensure exhaustive partitioning before proceeding to the next phase. The resulting organic layer was concentrated to dryness and reconstituted in 80% methanol containing 0.1% formic acid. The second partitioning step was performed using n-hexane. The aqueous layer from the final partitioning was isolated and concentrated to dryness, yielding 1.2 g of sample. The final yield was 0.6% of the initial microalgal biomass (200 g). The fractionated AOMeOH extract was further purified and underwent ODS column chromatography. ODS, a representative stationary phase in reverse-phase chromatography, enables effective separation by strongly interacting with non-polar compounds (Yabré et al. 2018). This process followed the method detailed in Supplementary Fig. S2. The AOF was prepared at a concentration of 10 mg mL⁻¹ and loaded onto an ODS column with a total volume of 5 mL, the optimized concentration and volume considering the column capacity. Subsequently, stepwise elution was performed using 0, 20, 40, and 100% ACN. This stepwise elution method allowed for the effective separation of compounds with different polarities. Each fraction was analyzed using liquid chromatography-quadrupole time-of-flight (LC-Q-TOF) MS/MS. This highly sensitive analytical instrument provides accurate mass measurements and structural information, making it useful for identifying target compounds in complex mixtures (Ramos et al. 2023). The results showed that the fraction eluted with 40% ACN contained the highest intensity of the target compound. Based on these results, all subsequent extraction samples were eluted with 40% ACN to obtain the final sample (AO 40). The ODS purification process was repeated until all fractionated AO samples were exhausted, allowing a sufficient amount of the 40% ACN fraction to be obtained. Finally, the separated and purified 40% ACN fraction was further purified using SPE, an effective method for further concentrating the target compounds and removing impurities (Mahdavijalal et al. 2024). Analysis of the results revealed multiple peaks in the total ion chromatogram. Upon examining the MS spectrum of the peaks between 4 and 6 min, which showed the highest intensity (Fig. 2A), m/z 523.33 was detected, prompting further investigation of the XIC for this molecular mass. As shown in Fig. 2B, peaks were observed at 4.10, 4.61, and 5.14 min. A detailed analysis of the MS spectra of these three major peaks revealed that all three exhibited the same precursor ion value of m/z 524.34 (Fig. 2C–E). These results suggested that the analyzed substances may be isomers with the same molecular formula but different structures commonly found in natural products (Huo et al. 2022). Future research will require additional analyses for the accurate structural identification and characterization of these compounds. For example, structural analysis can be performed through NMR spectroscopy, 3D structure confirmation using X-ray crystallography, and biological activity tests (Barba-Ostria et al. 2022). Through such a comprehensive analysis, the complete characterization of these interesting compounds found in the AO sample will be possible, potentially leading to the discovery of new natural substances and important research results.

Accurate analysis and identification of marine toxins are crucial for public health and for the management of marine ecosystems. In particular, toxins produced by dinoflagellates are gaining worldwide attention because of their toxicity and ecological effect (Deng et al. 2023). This study focused on the analysis and characterization of GYM series toxins produced by AO using LC-MS/MS, an advanced analytical technology. Such research could provide essential baseline data for assessing marine ecosystem health and ensuring the safety of seafood. A notable discovery in the research process was the observation of three peaks with different retention times but identical precursor ions (Fig. 2). This phenomenon strongly suggests the presence of isomers with structurally similar but slightly different properties. The existence of isomers provides significant evidence of the diversity and complexity of marine toxins, and their accurate identification and characterization are key challenges in toxin research (Gerssen et al. 2010). For precise analysis, library matching using Waters Software was performed prior to the MS/MS analysis. This is an effective method for the rapid identification of unknown compounds by comparing them with known compounds. The matching results confirmed that these peaks corresponded to the precursor ions of GYMs B, C, and D. This initial identification plays a crucial role in determining the direction of subsequent analyses. Gymnodimines are primarily produced by marine dinoflagellates and are found in various regions across the globe (McCarron et al. 2014, Van de Waal et al. 2015). They have been detected in large quantities, particularly after prolonged algal blooms along the coasts of New Zealand and Tunisia (Seki et al. 1995). This suggests that GYM production is closely related to specific environmental conditions, highlighting the need for further research on the effects of climate and marine environmental changes on the production and distribution of these toxins. However, the toxicity of GYMs is extremely high. With an LD₅₀ of 96 μg kg⁻¹ when injected and 755 μg kg⁻¹ when administered orally, even small amounts can pose serious health risks. This high toxicity is due to the action of GYMs as neurotoxins that block acetylcholine receptors (Munday et al. 2004). Therefore, the accurate detection and quantification of these toxins are crucial for ensuring seafood safety and protecting public health. AO used in this study has also been confirmed to produce GYMs (Harju et al. 2016). This demonstrates that this species is an important source of GYM and suggests the need for further research on its ecological characteristics and toxin production mechanisms. Gymnodimines B, C, and D had the same natural mass value (m/z 523.32) and similar structural features (Fig. 3). MS/MS analysis detected similar fragment ions, such as m/z 506.35 (with H₂O removed) and m/z 162.12 and 136.12, derived from the CI portion (Naila et al. 2012, Martens et al. 2017). These characteristic fragment ion patterns reflect the structural features of GYM series toxins and serve as important indicators for their identification.

Additional analyses were performed on the respective peaks (Fig. 2) to acquire structural data to determine whether these compounds corresponded to GYMs B, C, or D. Previous studies have mainly distinguished these compounds using NMR results of isolated single compounds (Miles et al. 2003, Zurhelle et al. 2018). Although NMR is a powerful tool for elucidating compound structures at the atomic level, it is limited in its ability to analyze trace amounts of compounds present in complex mixtures (Judge and Ebbels 2022). Therefore, an integrated approach utilizing LC-MS/MS and NMR is necessary for the comprehensive characterization of these marine toxins. ETD is a method commonly used in proteomics that analyzes molecules by fragmenting them through a mechanism where electrons are transferred to multiply positively charged peptides, forming radical cations and subsequently cleaving N-Cα bonds (Molina et al. 2008). Unlike Collision-induced dissociation (CID), which primarily cleaves the most labile bonds, ETD induces the fragmentation of precursor ions through a gentler reaction pathway by transferring electrons from radical anions to the analyte, enabling more uniform fragmentation across the molecular structure (Han and Costello 2011). Given the inherent limitations of conventional fragmentation techniques in widely utilized structural analysis instrumentation, we investigated the potential of integrating ETD with CID to augment analytical capabilities and improve structural elucidation methodologies. This approach aims to leverage the complementary strengths of both fragmentation techniques, potentially providing more comprehensive structural information and overcoming the constraints associated with individual methods. The detected peaks were prepared by coupling the same protocol and column used in the LC-Q-TOF MS/MS analysis with UPLC. Because the remaining peaks were identified in very small quantities, making further analysis challenging, the peak detected at 4.61 min was analyzed using ETD. As shown in Fig. 4, a peak at m/z 524.3461 was observed at 4.61 min in the MS/MS spectrum. Upon examining the spectrum, the CI of GYM was observed at m/z 136.1130 and 162.1272 (Fig. 4B). Furthermore, when the intensity of m/z 200–500 in the MS/MS spectrum was lowered, we confirmed the detection of [M+H-H₂O] at m/z 506.33, m/z 496.33, without carbon monoxide (CO), and at m/z 402.26, 346.24, and 316.23. These fragment ions are characteristic of GYM D (Harju et al. 2016). When purchasing GYM standards, most are limited to GYM A (MUSECHEM, BENCHCHEM, etc.) and GYM B (LGC). In contrast, other isomers are not commercially available because of difficulties in their isolation and purification. Therefore, this study demonstrates the potential for isolating and purifying GYM D from culturable AO samples. Based on these results, we anticipate the ability to secure standard materials for future monitoring of marine toxin.