For each extract, the concentration inhibiting 50% of cell growth (Growth Inhibition 50%; GI₅₀) was determined (Table 1).

DCM and EtOH DT extracts inhibited MCF-7 growth with equivalent potency and at low concentrations (GI₅₀ ≈ 60 μg·mL⁻¹). The DT DCM extract also inhibited LNCaP growth, with a GI₅₀ close to the value determined on MCF-7 (GI₅₀ = 60.9 μg·mL⁻¹). No extract inhibited MDA-MB-231 growth. The DT DCM extract, active both on MCF-7 and LNCaP cells, was selected to purify antiproliferative molecules by fractionation.

Figure 1 presents the DT DCM extract chromatogram obtained at 435 nm, with the definition of the fractions and sub-fractions tested on MCF-7.

Identification of the 11 major peaks present on the chromatogram was not performed at this step as we hypothesized that some fractions may not be active on cancer cells. The antiproliferative activity of each fraction was studied on the MCF-7 cell line as it was the most sensitive to the starting DCM extract, and grew faster than LNCaP. Table 2 presents the antiproliferative activity of the four DT DCM fractions and the four F1 sub-fractions on MCF-7.

Fraction 1 (F₁) was identified as the only active fraction in the DT DCM extract, with a GI₅₀ = 14.3 μg·mL⁻¹. Decrease of the GI₅₀ value compared to the DCM extract confirmed that this fraction was concentrated in active molecules (Table 2). The GI₅₀ of F₂, F₃ and F₄ were superior to 100 μg·mL⁻¹ (Table 2), indicating that they did not contain potent antiproliferative molecules. F_1.2, F_1.3 and F_1.4 strongly inhibited MCF-7 growth, with GI₅₀ values of 20.5, 18.9 and 11.7 μg·mL⁻¹, respectively (Table 2). The GI₅₀ values of these three sub-fractions were in the range of that of the F₁ fraction, and confirmed that the three sub-fractions contained active molecules (Table 2). The GI₅₀ of F_1.1 was greater than 40 μg·mL⁻¹. Figure 2 presents the GI₅₀ (μg·mL⁻¹) measured on MCF-7 with the starting DT DCM extract, the F₁ fraction and the F_1.4 subfraction.

The GI₅₀ decreased with purification steps, indicating that the antiproliferative activity measured in the initial crude extract was not due to a synergistic action between several molecules in the mixture.

The antiproliferative activity of the most active sub-fraction, F_1.4, was assessed on MCF-7 continuously exposed for 72 h to increasing concentrations in the cell culture medium. F_1.4 inhibited MCF-7 growth at a concentration as low as 0.1 μg·mL⁻¹ and in a dose-dependent manner from 0.1 to 40 μg·mL⁻¹ (Figure 3).

A concentration of 40 μg·mL⁻¹ was necessary to observe a cytostatic activity on MCF-7 (Figure 3). MCF-7 cells were also exposed for 72 h to various concentrations of F_1.4 in the cell culture medium, before changing the medium to a fresh control cell culture medium (Figure 4).

At all tested concentrations, the growth rate increased when the culture medium containing F_1.4 was replaced with fresh culture medium, demonstrating that F_1.4 exerted a strong antiproliferative effect, without however killing all cells, in the range of the pharmacological concentrations studied.

Analytical RP-HPLC at 435 nm of F_1.4 demonstrated that 95% of this fraction corresponded to a single molecule M eluting at t = 17.326 min (Figure 5).

Iterative semi-preparative RP-HPLC allowed the collection of 0.910 mg of F_1.4 from 1500 mg freeze-dried DT cells, indicating that M represented 0.0576% (w:w) of the freeze-dried microalgae content, considering the extraction yield to be 100%. The absorption spectrum of M was characteristic of a carotenoid pigment and presented maximal absorption peaks at 417.2, 441.5 and 471.9 nm, with a band III:II ratio of 96% (Figure 5). These values were compared to data from reference spectra [16] which suggested that M was most probably violaxanthin, as its maximal absorption wavelengths and band III:II ratio were very close to the values measured with standard violaxanthin in ethanol. HRMS analysis unambiguously confirmed that M corresponded to violaxanthin (Table 3; Figure 6C).

Figure 7 shows epifluorescence microphotographs of MCF-7 cells labeled with the DNA marker BOBO-1 (green) and Annexin-V-Alexa 568 (red) after 48 h incubation in the presence of various concentrations of violaxanthin purified from DT.

Only a small proportion of control cells (Figure 7A) were identified as apoptotic (red arrows) or early necrotic (bright green spots). Violaxanthin 8 and 20 μg·mL⁻¹ (Figures 7B and 7C, respectively) evoked MCF-7 apoptosis as indicated by the important increase in the number of cells binding annexin V without being labeled by BOBO-1. Only a few cells were identified as late necrotic after the violaxanthin treatment (green arrows).

Violaxanthin doses evoking phosphatidylserine translocation in MCF-7 cells did not induce DNA fragmentation, even after a 72 h treatment (Figure 8B).

A higher dose of violaxanthin was tested, in the range of rational pharmacological doses (40 μg·mL⁻¹) (66.4 μM), but no DNA fragmentation was observed even after 72 h incubation (Figure 8).

Dunaliella tertiolecta (DT) belongs to the Chlorophyceae class (green microalgae). The genus Dunaliella exhibit a small ovoid unfrustulated cell (<20 μm), covered of cellulose, xylans, mannans and/or glycoproteins, with two equal flagella inserted at the apexes of the cell. Chlorophyll a, chlorophyll b, β,β-carotene, neoxanthin, lutein and violaxanthin are the main pigments described in DT [17]. DT was selected as it does not contain fucoxanthin, which has already been extensively studied for its antiproliferative activity [9–15].

Microalgae were grown at IFREMER PBA, Nantes, France. DT strain CCMP364 (CCMP, USA) was cultivated in 10 L flasks under continuous illumination at an average light intensity of 180 μmol·m⁻²s⁻¹. Growth was performed at 20 °C, in pH unregulated batch culture, in Walne (Conway) medium diluted in 0.22 μm sterile-filtered natural seawater. The DT cell suspension was harvested at the end of the exponential growth phase, and cells were separated from culture medium by soft centrifugation (4000 g, 20 min, 10 °C). Cells were frozen at −20 °C, sent to laboratory LIENSs, La Rochelle and freeze-dried at −55 °C and P < 1 hPa, on a freeze-dryer equipped with a HetoLyoPro 3000 condenser and Heto cooling trap (Therma Electron Corporation, France).

In order to extract most microalgal organic molecules, on a wide range of polarity, successive extractions were performed in dichloromethane (DCM), ethanol (EtOH) and ultrapure water. A 1 g sample of freeze-dried microalgae powder was first extracted for 2 h in 100 mL DCM (1% w/v), at room temperature, under continuous shaking and in the dark. The mixture was filtered through a PVDF 0.22 μm membrane, and the DCM extract was evaporated to dryness in dark vials (45 °C, vacuum). The insoluble residue was collected, dried and successively extracted in ethanol and ultrapure water, in the same conditions, except that the water extract was collected by centrifugation (8000 g, 10 min, 4 °C) and filtered through a nitrocellulose membrane. The EtOH extract was evaporated to dryness in dark vials (45 °C, vacuum) and the aqueous extract was freeze-dried.

The RP-HPLC system was composed of a binary pump (Waters, W600), an autosampler (Waters, W717), a thermostated (20 °C) column compartment, and a photodiode-array detector (Waters, W486). Analytical RP-HPLC of the DCM extract was performed on a 20 μL sample injected in a Phenomenex Luna C18 (2) analytical column (250 × 4.6 mm, 10 μm), the mobile phase consisting of a ternary gradient of solvent A (Methanol/water (80/20)); solvent B (Acetonitrile/water (90/10)) and solvent C (ethyl acetate). The gradient flow program was adapted from Jeffrey and collaborators [18], as follows: 0 min, 100% A; 3 min, 100% B; 35 min, 30% B and 70% C; 38 min, 100% C; 41 min, 100% C; 43 min, 100% B; 45 min, 100% A. The flow rate was 1.0 mL·min⁻¹ and elution was monitored at 435 nm. Fractionation of the active extract was performed in semi-preparative conditions, using a semi-preparative Phenomenex Luna C18 (2) column (250 × 10 mm, 10 μm) with a flow rate fixed at 5 mL·min⁻¹.

MCF-7 and MDA-MB-231 human mammary carcinoma cell lines, as well as A549 human lung adenocarcinoma cell line (LGC Standards, France) were grown as monolayers, at 37 °C in a 5% CO₂–95% air humidified atmosphere, in DMEM (Gibco, France) supplemented with 10% heat-inactivated (56 °C, 30 min) FCS (Dutscher, France) to which penicillin 100 U·mL⁻¹ and streptomycin 100 μg·mL⁻¹ were added. LNCaP human prostatic carcinoma was grown as monolayer in RPMI (devoid of phenol red) supplemented with 10% heat-inactivated (56 °C, 30 min) FCS, penicillin 100 U·mL⁻¹ and streptomycin 100 μg·mL⁻¹.

The algal extracts and related fractions were evaporated to dryness and stock solutions were prepared in DMSO before being diluted in the cell culture medium. The final DMSO concentration was lower than 1% and tested as a negative control. Cell viability was studied using the MTT assay.

Accurate molecular weight of the bioactive molecules contained in DT extracts was determined by HRMS at the “Centre Régional de Mesures Physiques de l’Ouest”, University of Rennes 1, France. The mass spectrometer was a Bruker MicrO-Tof-Q 2, equipped with an ESI source, and samples were dissolved in CH₂Cl₂:CH₃OH (90:10).

All solvents used in this study were HPLC grade. Standard pigments were obtained from Chromadex, France. Ultra-pure water was obtained using a Milli-Q system (Millipore, France).