Fractions from D. tertiolecta extracts were analyzed by HPLC-UV-HRESIMS on an Agilent 1200 RR coupled to a Bruker maXis QToF spectrometer with electrospray ionization, as reported by Martín et al. (2014). Data were analyzed following the guidelines developed by Fundación MEDINA (Perez-Victoria et al., 2016), and compared with the data available on the Fundación MEDINA internal database and the Dictionary of Natural Products database. ¹H NMR and HSQC spectra were recorded on a Bruker Avance III spectrometer (500 MHz for ¹H NMR) equipped with a 1.7 mm TCI MicroCryoProbe, using the signal of the residual solvent as internal reference (δH 7.24 and δC 77.23 ppm for CDCl₃).

Dunaliella tertiolecta was bought from NCMA Bigelow (code CCMP 1320) in September 2010 and has been maintained in the culture collection of the Stazione Zoologica Anton Dohrn since then. The medium used to maintain and grow this green alga was Guillard’s f/2 medium (Guillard, 1975) with silica depletion. Briefly, D. tertiolecta was inoculated in a volume of 30 mL and further scaled up to 2 L at an initial concentration of 5,000 cells/mL to produce enough biomass for chemical and biological analyses. Culture growth was monitored daily from samples fixed with one drop of Lugol (final concentration of about 2%) and counted in a Bürker counting chamber under an Axioskop 2 microscope (20×) (Carl Zeiss GmbH, Jena, Germany) (Orefice et al., 2015). Once the stationary phase was reached (day 15), aliquots of 50 mL were harvested by centrifugation at 3,000 rpm and 4°C, for 10 min. The wet biomass harvested was immediately used for chemical extraction. A biological triplicate was performed in order to generate solid results.

The wet biological material (80% water content) was extracted by soaking in methanol (proportion 1:5, w/v) and 30 min of maceration. The organic mixture was then vortexed for 1 min, sonicated with three bursts of 30 s in an icy water bath to lysate the cells, and centrifuged at 3,000 rpm and 4^°C to precipitate the solid material in suspension that was discarded. The organic phase was then transferred to a rounded Pyrex flask and evaporated under reduced pressure. The extraction method was performed according to the protocol described by Cutignano et al. (2015), a specific method developed to avoid or at least ameliorate many of the difficulties associated with screening of marine natural extracts while extracting the most common classes of natural products present in such samples.

Fractionation of each extract (an aliquot with a maximum weight of 20 mg) was performed by solid phase extraction (SPE) using CHROMABOND^® HR-X cartridges (6 mL/500 mg) as reported in Cutignano et al. (2015). In this study, the elution process was achieved by gravity. Briefly, the cartridge was washed with 3 mL of methanol and equilibrated with 6 mL of distilled water. The extract was suspended in 1 mL of distilled water and sonicated for a few seconds in a bath of icy water before loading it into the cartridge. After the extract suspension was absorbed by the resin, the following elution steps were performed to produce five different fractions: Washing step 100% H₂O (2 mL, discarded); 100% H₂O (6 mL, fraction A); CH₃OH/H₂O (50:50, 9 mL, fraction B); CH₃CN/H₂O (70:30, 9 mL, fraction C); 100% CH₃CN (9 mL, fraction D); CH₂Cl₂/CH₃OH (90:10, 9 mL, fraction E). Each fraction was then evaporated under reduced pressure, weighted, and preserved at −20^°C.

Human cells were bought at ATCC.² Human melanoma cells (A2058; ATCC^® CRL-11147™) were cultured in DMEM and human lung adenocarcinoma cells (HCC827; ATCC^® CRL-2868™) were cultured in RPMI. Human lung adenocarcinoma cells (Calu-3; ATCC^® HTB-55™), human hepatocellular liver carcinoma cells (HepG2; ATCC^® HB-8065™) and normal lung fibroblast cells (MRC-5; ATCC^® CRL-171™) were cultured in EMEM medium. The media were supplemented with 10% fetal bovine serum, 50 U/mL penicillin, and 50 μg/mL streptomycin (Sigma Aldrich). To estimate the in vitro antiproliferative effects of D. tertiolecta, A2058, HepG2, HCC827, Calu-3, and MRC-5 cells line were seeded in 96-well microtiter plates at a density of 1 × 10⁴ cells/well and incubated at 37°C to allow for cell adhesion in the plates. After 24 h, the medium was replaced with fresh medium containing increasing concentrations of the fractions (10, 50, and 100 μg/mL) dissolved in dimethyl sulfoxide (DMSO) and further incubated for 72 h. The maximum concentration of DMSO used was 1% (v/v). Each concentration was tested at least in triplicate. Doxorubicin, used as positive control, was tested at 5 mM, and a dose response curve with 8-point serial dilution (1:2 dilutions) was performed. Cell viability was 0% at each concentration tested. After 72 h, cell viability was assessed using the MTT test 3-(4,5-dimethyl-2-thizolyl)-2,5-diphenyl-2H-tetrazolium bromide; A2231,0001, Applichem Panreac Tischkalender, Darmstadt, GmbH) (Baudelet et al., 2013; Russo et al., 2016). Briefly, the medium was replaced with medium containing MTT at 0.5 mg/mL and the plates were incubated for 3 h at 37°C. After incubation, cells were treated with isopropyl alcohol (used as MTT solvent) for 30 min at room temperature. Absorbance was measured at OD = 570 nm using a microplate reader (Multiskan™ FC Microplate Photometer, Thermo Fisher Scientific, Waltham, MA, United States) (Sansone et al., 2014). The same mentioned protocol was used to treat the A2058 cell lines with glycerol 1-(9Z,12Z,15Z-octadecatrienoate) 2-(4Z,7Z,10Z,13Z-hexadecatetraenoate)-3-O-β-D-galactopyranoside (1) purified from the raw extracts of D. tertiolecta. Cell survival was expressed as a percentage of viable cells in the presence of the tested samples, with respect to untreated control cultures with only DMSO. Extracts/fractions with cell viabilities above 60% were not considered active.

Ten grams of lyophilized D. tertiolecta biomass were extracted by sonication and further maceration in methanol as reported in the chemical extraction method. The organic phase was filtered and evaporated at 35°C under reduced pressure, yielding 2.1 g of raw extract. The SPE-method for fractionation of marine extracts reported by Cutignano et al. (2015) was scaled-up for 2.1 g of raw extract and performed to yield five different fractions of decreasing polarity (A–E).

Fractions B to D, containing the compound of interest, were pooled and processed by reversed phase preparative HPLC (Agilent Zorbax SB-C8, 21.2 × 250 mm, 7 μm; 20 mL/min, UV detection at 210 nm) with a linear gradient of CH₃CN in water from 5 to 100% over 30 min and a wash step at 100% over 10 min where the fraction containing 1 (7.6 mg) eluted at 30.5 min. This fraction was further purified by reversed phase semipreparative HPLC (XBridge C18, 10 × 150 mm, 5 μm, 3.8 mL/min, UV detection at 210 nm) with a linear gradient of CH₃CN in water from 80 to 92% over 37 min to yield 2.4 mg of compound 1 (r_T 31 min, >90% purity calculated in terms of area of the peak in the UV 210nm chromatogram of the aliquot analyzed).

The total extract of D. tertiolecta obtained during the stationary growth phase was tested against human cell lines using the MTT assay. In particular, A2058, HepG2, HCC-827, Calu-3, and MRC-5 cells were incubated in the presence or absence of three different concentrations (10, 50, and 100 μg/mL) of D. tertiolecta total extract. After 72 h of incubation at 37°C, cell survival was measured with the MTT assay. The screening was performed using three biological replicates. As shown in Figure 1, total extracts of D. tertiolecta showed antiproliferative activity only on A2058 cells with 44 and 35% of cell viability at 50 and 100 μg/mL concentrations, respectively. On the other hand, no cytotoxic effects were observed for the HepG2, HCC827, Calu-3 cancer cell lines, and for MRC-5 normal cell lines.

D. tertiolecta raw extract was then fractionated to obtain five fractions (Fractions A to E). As reported in the solid phase extraction (SPE) method to fractionate marine organic extracts (Cutignano et al., 2015), the method yielded an amino acid- and saccharide- rich fraction (Fraction A), a nucleoside-rich fraction (Fraction B), a glycol- and phospholipid-rich fraction (Fraction C), a free fatty acid- and sterol-rich fraction (Fraction D), and a triglyceride-rich fraction (Fraction E). Fraction A was not considered for testing since it was mainly composed by amino acids, saccharides (primary metabolites) and remaining salts. For the antiproliferative assay and bioactivity evaluation, the same cell lines (i.e., cancer cell lines A2058, HepG2, HCC827, Calu-3, and MRC-5) were used. Bioactivity testing of the other four fractions identified fractions D and E as the most active against A2058, HepG2, HCC827, and Calu-3 (Figure 2). In particular, cell viability of A2058 cells was reduced to 16% when incubated with Fraction D at 100 μg/mL; 38% in HepG2 cells at 100 μg/mL; 46% in HCC827 cells at 50 μg/mL and 9% at 100 μg/mL; and 21% in Calu-3 cells at 100 μg/mL. Fraction E lowered cell viability to 36% in A2058 cells and 45% in HCC827 cells at 100 μg/mL. Fractions B and C did not show any significant antiproliferative activity on any of the cell lines tested. Regarding toxicity, the fractions did not display any cytotoxic effect against normal cells MRC-5. Activities above 60% cell viability were not considered.

HPLC-UV-HRESIMS based dereplication was used to identify the components in bioactive fractions. LC-UV traces of the most active fractions (D and E) and HRESIMS spectra of the most abundant components are reported in the figures below (Figures 3, 4). The compounds most likely present in fractions D and E were identified according to their mass spectra, UV maxima, natural source (taxonomy) and biological activity data.

Peak 1 (P1), observed in both fraction D and E, was suggested to have a molecular formula of C₄₃H₆₈O₁₀ based on the existence of an [M + NH₄]⁺ ion at m/z 762.5208 (calcd. for C₄₃H₇₂NO₁₀⁺ 762.5151). The best match in the Dictionary of Natural Products database (DNP) according to the HRMS data, biological source, retention times and UV (λ) maxima, was glycerol-1-(9Z,12Z,15Z-octadecatrienoate)-2-(4Z,7Z,10Z,13Z-hexadecatetraenoate)-3-O-β-D-galactopyranoside (1) (Figure 5). This compound has already been found and isolated from extracts of another two green algae: Dunaliella acidophila (Della Greca et al., 1989) and Tetraselmis chui (Banskota et al., 2013), and it was reported to possess nitric oxide inhibitory activity. However, this is the first time it is tested for its antiproliferative activity on cancer cells. This compound is one of the monogalactosyldiacylglycerols (MGDG, one type of glycoglycerolipid) present in high amounts in the thylakoid membranes of photosynthetic organisms (Boudière et al., 2014) including plants, algae and cyanobacteria. The genes involved in the biosynthesis of MGDGs are present in several microalgal species, as reported in Riccio et al. (2020a). Maeda et al. (2010) studied the effect of a pool of MGDGs from spinach on six different cancer cell lines: lung cancer A549, acute lymphoblastoid leukemia BALL1, colon carcinoma HCT116, cervix cancer HeLa, promyelocytic leukemia HL-60 and stomach cancer NUGC3. The authors extracted this glycoglycerolipid from spinach and demonstrated that this molecule affected the activity of DNA polymerases α, γ, δ and ε (DNA metabolic enzymes). They showed that spinach MGDGs may be able to penetrate cancer cells and reach the nucleus, where the inhibition of the DNA polymerase activities may lead to selective cancer cell growth suppression. MTT assay was used to demonstrate that spinach MGDGs reduced the proliferation of all six cancer cell lines, and promyelocytic leukemia (HL60) cells were the most affected with an IC₅₀ value of 39.2 μg/mL. Such results fall into the same order of magnitude as the results obtained for Fraction D in the current work (for instance, 46% cell viability at 50 μg/mL in HCC827 cell line) even though they didn’t test the individual compounds. An MGDG could hence explain the biological activity observed within Fraction D.

Peak 2 (P2), observed only in fraction E, was suggested to have a molecular formula of C₄₀H₅₄O according to the existence of an [M + H]⁺ peak at m/z 551.4248 in its HRMS spectrum (calcd. for C₄₀H₅₅O⁺ 551.4247). The best coincidence in the DNP corresponded to crocoxanthin, according to the HRMS data, biological source, retention times and UV (λ) maxima. Crocoxanthin is a bioactive carotenoid of microalgal origin that was first isolated from the mixotroph marine microalgae Hemiselmis virescens (Chapman, 1966). In particular, de Oliveira-Júnior et al. (2020) demonstrated that crocoxanthin purified from Rhodomonas salina showed antiproliferative activity against A2058 cells, inhibiting cell growth by 50% at 50 μM (IC₅₀). In that study, Crocoxanthin promoted growth inhibition, decreased cell migration, and induced apoptosis and sub-G1 cells accumulation after 72 h of treatment. Since the second component of peak 2 is most likely to be a known pigment with antiproliferative activity on cancer cells, it was not further confirmed nor studied in this work.

Compound 1 was purified from a methanolic extract of the green microalga using solid phase extraction to yield three different fractions containing the compound, which were then processed by low pressure chromatography followed by preparative and semipreparative reversed phase HPLC, as reported in materials and methods. Mass spectrometry and NMR analysis confirmed the structure of the molecule as glycerol 1-(9Z,12Z,15Z-octadecatrienoate)-2-(4Z,7Z,10Z,13Z-hexadecatetraenoate)-3-O-β-D-galactopyranoside (Della Greca et al., 1989; Banskota et al., 2013). ¹H NMR, HSQC 2D-NMR, UV-HRMS spectra, MS/MS spectra and fragmentation pathway are reported in Supplementary Figures 1–5. Two tables containing the NMR and MS/MS signals were also included in Supplementary Tables 1, 2. NMR spectra were consistent with bibliography, being identical to the original data by Della Greca et al. (1989), and almost identical to the data published by Banskota et al. (2013) regarding (2S)-1-O-(6Z,9Z,12Z,15Z-octadecatrinoyl)-2-O-(4Z,7Z,10Z,13Z-hexadecatetranoyl)-3-O-D-galactopyranosylglycerol, except for the presence of the extra double bond (6Z). In addition, MS/MS data allowed to build a fragmentation pathway that is consistent with the proposed structure. Metfrag³ was used to assign mass peaks to the corresponding ions.

The antiproliferative activity of compound 1 was tested against A2058 melanoma cell line, to confirm that growth inhibition observed for the fraction was displayed by this compound. Compound 1 induced a 50% reduction of cell viability when tested at 50 μg/mL, while at 100 μg/mL cell viability was 10%, respect to the vehicle (Figure 6).