Differences and Similarities in Lipid Composition, Nutritional Value, and Bioactive Potential of Four Edible Chlorella vulgaris Strains

The microalga Chlorella vulgaris is a popular food ingredient widely used in the industry, with an increasing market size and value. Currently, several edible strains of C. vulgaris with different organoleptic characteristics are commercialized to meet consumer needs. This study aimed to compare the fatty acid (FA) and lipid profile of four commercialized strains of C. vulgaris (C-Auto, C-Hetero, C-Honey, and C-White) using gas- and liquid-chromatography coupled to mass-spectrometry approaches, and to evaluate their antioxidant and anti-inflammatory properties. Results showed that C-Auto had a higher lipid content compared to the other strains and higher levels of omega-3 polyunsaturated FAs (PUFAs). However, the C-Hetero, C-Honey, and C-White strains had higher levels of omega-6 PUFAs. The lipidome signature was also different between strains, as C-Auto had a higher content of polar lipids esterified to omega-3 PUFAs, while C-White had a higher content of phospholipids with omega-6 PUFAs. C-Hetero and C-Honey showed a higher content of triacylglycerols. All extracts showed antioxidant and anti-inflammatory activity, highlighting C-Auto with greater potential. Overall, the four strains of C. vulgaris can be selectively chosen as a source of added-value lipids to be used as ingredients in food and nutraceutical applications for different market needs and nutritional requirements.


Introduction
Microalgae are emerging organic and green ingredients with a wide diversity of nutrients and bioactive compounds, such as vitamins, minerals, proteins, and lipids. The consumption of microalgae-based products is on the rise, largely due to their beneficial health properties. They are excellent allies for a healthy and sustainable diet and lifestyle. Among commercialized microalgae, Chlorella vulgaris has one of the highest market values, expected to reach USD 412.3 million by 2028, and an actual yield of 5000 tons of dry matter per year [1]. Chlorella vulgaris is widely used in the food and feed industries, but also in the cosmetic and pharmaceutical fields, among others, as reviewed elsewhere [2].
Currently, the production of microalgae biomass at large scale is considered one of the most promising strategies to meet the needs of the next generations for food, feed, and pharmaceutical products [3]. Several cultivation methods have been established

Microalgae Strain and Culture Media
Chlorella vulgaris strains (C-Auto, C-Hetero, C-Honey, and C-White) were provided by Allmicroalgae. The C-Auto and C-Hetero C. vulgaris strains were produced as reported in previous works [3,7], whereas the chlorophyll-deficient mutants (C-Honey and C-White) were obtained by chemically induced random mutagenesis from C-Hetero, as described [6]. Briefly, the heterotrophic culture was grown in Guillard's F2 culture medium adapted to local water using nitrate and glucose with a C:N ratio of 6.7:1. This culture was obtained sequentially from a 50-250 mL Erlenmeyer flask to a 5 L bench-top fermenter (New Brunswick BioFlo ® CelliGen ® 115; Eppendorf AG, Hamburg, Germany), and later to 200 L and 5000 L industrial fermenters. All fermenters were operated in fed-batch mode under controlled temperature, pH, and dissolved oxygen. Chlorella vulgaris cells were cultured and treated with different concentrations of ethyl methanesulfonate (EMS, Merck, New York, NY, USA) for 1 h in the dark. A colony of C. vulgaris mutants colored yellow (C-Honey) was selected by visual observation of the plates in dim light, and then isolated and sub-cultured several times on count agar plates (VWR, Carnaxide, Portugal). Then, C-Honey was grown to an exponential phase and subjected to a second round of random mutagenesis with 300 mM EMS. This time, a colony with a white colour (C-White) was selected and sub-cultured on count agar plates with 10 µM norflurazon (a metabolic inhibitor of carotenoids biosynthesis) and incubated at 30 • C in the dark for 1 week. C-White colonies were sub-cultured multiple times, with and without norflurazon to confirm mutant phenotypic stability. Mutant strains were considered stable after 50 generations expressing the same phenotype. To further ensure mutation stability, stock cultures of the mutants were maintained with the same inhibitor (10 µM norflurazon) to maintain the selective pressure used for the initial mutagenesis process and, thus, prevent reversion of the mutation.

Lipid Extraction
Total lipids were extracted according to the modified Folch's method, as previously described [7,14]. The extraction was carried out individually for each C. vulgaris strain (C-Auto, C-Hetero, C-Honey, and C-White), from 25 mg of spray-dried biomass using a solvent mixture of CH 2 Cl 2 :CH 3 OH (2:1, v/v). All experimental procedures were carried out individually for five replicates (n = 5) obtained from a single large batch of biomass. The suspension was centrifuged at 2000 rpm for 10 min (Selecta JP Mixtasel, Abrera, Barcelona, Spain) and the supernatants were collected and dried under a stream of nitrogen. This step was repeated three times. Then, the dried extracts were re-extracted with 2 mL of CH 2 Cl 2 , 1 mL of CH 3 OH, and 0.75 mL of Milli-Q water, centrifuged (2000 rpm for 10 min), and then the lipid-containing layers were collected and dried. The aqueous phase was re-extracted with 2 mL of CH 2 Cl 2 two more times. Finally, the extracts obtained were weighed and the extraction yield was calculated as a percentage of dry weight (DW), with the following Equation (1):

Fatty Acid Analysis by GC-MS
The fatty acid composition of C. vulgaris lipid extracts was analyzed by gas chromatography coupled to mass spectrometry (GC-MS). Fatty acid methyl esters (FAMEs) solution was prepared using alkaline trans-esterification of 60 µg of lipid extract and the addition of methyl nonadecanoate (internal standard) at 0.96 µg mL −1 , as performed routinely in our laboratory [15,16]. The GC-MS was performed using an Agilent Technologies 8860 GC System (Santa Clara, CA, USA) interfaced with an Agilent 5977B Mass Selective Detector (Agilent, Santa Clara, CA, USA). The electron impact ionization was 70 eV, and full scan mode was used in 1 s cycles with a scanning range of m/z 50-550. A DB-FFAP capillary column (30 m long × 0.32 mm internal diameter, 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA) was used. In the chromatographic analysis, the following conditions were used: injection volume of 2 µL (splitless), a constant flow rate of 1.4 mL min −1 of helium gas, inlet temperature 230 • C, and detector temperature 220 • C. The oven temperature was programmed at 58 • C for 2 min, 25 • C min −1 to 160 • C, 2 • C min −1 to 210 • C, 30 • C min −1 to 225 • C (held for 10 min). The data acquisition software was the GCMS5977B/Enhanced MassHunter. Agilent MassHunter Qualitative Analysis 10.0 software, NIST library, and the literature were used to identify FAs. The relative abundance (RA) of fatty acids was obtained by integrating the area under the peak and normalizing the data with the internal standard.
The mass spectrometer operated simultaneously in positive (ESI 3.0 kV) and negative (ESI −2.7 kV) modes. The capillary temperature was 320 • C and the sheath gas flow was 35 U. Acquisition of data was performed in full scan mode with a high resolution of 70,000 and automatic gain control (AGC) target of 3 × 10 6 , in an m/z range of 300-1600, 2 micro scans, and a maximum injection time (IT) of 100 ms. Tandem mass spectra (MS/MS) were obtained with a resolution of 17,500, AGC target of 1 × 10 5 , 1 micro scan, and a maximum IT of 100 ms. The cycles consisted of a full-scan mass spectrum and 10 data-dependent MS/MS scans, which were repeated continuously throughout the experiments with a Lipid species were identified using mass spectrometry-data independent analysis (MS-DIAL) v4.70 software and manual data analysis [19,20], and integrated in the MZmine v2.53 software [21]. The areas of the peaks of each lipid species were normalized by calculating the ratio against the area of the selected internal lipid standard (with the closest retention time). The relative abundance of each lipid species was estimated by dividing the normalized peak areas of each lipid species by the sum of the total normalized peak areas.
The antioxidant activity expressed in Trolox equivalents (TE) was calculated with Equation (7), in which IC20 (or IC50) values are the concentration of lipid extract per sample and of Trolox, which promotes a 20% (or 50%) reduction in the radicals DPPH • or ABTS •+ .

Anti-Inflammatory Activity
The anti-inflammatory activity of the lipid extracts from C. vulgaris strains was evaluated with the cyclooxygenase-2 (COX-2) inhibition assay. This assay was carried out using the commercial COX-2 inhibitory screening assay kit, a Cayman test kit-701080 (Cayman Chemical Company, Ann Arbor, MI, USA), and was performed according to the manufacturer instructions. Briefly, lipid extracts were dissolved in 100% DMSO to a final concentration of 20, 60, and 125 µg mL −1 . The amount of prostaglandin F2α produced was quantified by spectrophotometry (415 nm, Multiskan GO 1.00.38, Thermo Scientific, Hudson, NH, USA) and processed with the software SkanIT version 3.2 (Thermo Scientific). The results were expressed as a percentage of inhibited COX-2.

Statistical Analysis
Statistical analysis was performed using R version 4.0.2 [22], in Rstudio version 1.3.1093 [23]. Lipidomic data were log-transformed. All data was tested for normality using the Shapiro-Wilk test and homoscedasticity using the Levene test. When these assumptions were satisfied, ANOVA followed by post-hoc Tukey tests were used. Non-parametric data were examined using the Mann-Whitney U test, followed by post-hoc Dunn test. The pvalues were corrected for multiple testing using the BH Benjamin, Hochberg, and Yekutieli method (q-values). All univariate analyses were performed using the R package 'rstatix' [24]. The principal component analysis (PCA) and the ellipses were created using the R libraries FactoMineR [25] and factoextra [26]. Heatmaps and hierarchical cluster analysis (HCA) were created using the R package pheatmap using "Euclidean" as clustering distance and "ward.D" as the clustering method [27]. The heatmaps were constructed based on the 50 lipid species with the lowest q-values (q < 0.05).

Lipid Content and Fatty Acid Profile of C. vulgaris Strains
The lipid contents expressed in dry weight (DW) of C-Auto, C-Hetero, C-Honey, and C-White, estimated by gravimetry, were 7.38 ± 0.3%, 5.18 ± 0.2%, 5.0 ± 1.2%, and 4.74 ± 0.5%, respectively ( Figure 1). Statistical analysis revealed that lipid content from C-Hetero, C-Honey, and C-White was significantly different from C-Auto. No significant differences were observed between the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White).
The results were expressed as a percentage of inhibited COX-2.

Statistical Analysis
Statistical analysis was performed using R version 4.0.2 [22], in Rstudio version 1.3.1093 [23]. Lipidomic data were log-transformed. All data was tested for normality using the Shapiro-Wilk test and homoscedasticity using the Levene test. When these assumptions were satisfied, ANOVA followed by post-hoc Tukey tests were used. Non-parametric data were examined using the Mann-Whitney U test, followed by post-hoc Dunn test. The p-values were corrected for multiple testing using the BH Benjamin, Hochberg, and Yekutieli method (q-values). All univariate analyses were performed using the R package 'rstatix' [24]. The principal component analysis (PCA) and the ellipses were created using the R libraries FactoMineR [25] and factoextra [26]. Heatmaps and hierarchical cluster analysis (HCA) were created using the R package pheatmap using "Euclidean" as clustering distance and "ward.D" as the clustering method [27]. The heatmaps were constructed based on the 50 lipid species with the lowest q-values (q < 0.05).
To estimate the lipid nutritional and health benefits of C. vulgaris strains, the lipid quality indexes including the n−6/n−3 ratio, atherogenic index (AI), thrombogenic index (TI), the hypocholesterolemic/hypercholesterolemic (h/H) index, and the peroxidation index (PI) were calculated ( Table 2). C-White recorded the highest n−6/n−3 ratio (13.9 ± 0.6), with much lower ratios for the other three strains, with C-Hetero (3.9 ± 0.0), C-Honey (2.1 ± 0.0), and C-Auto (0.6 ± 0.0). The AI and TI are predictors of cardiovascular disease risk, which measure the probability of reducing atherogenic plaques and blood clot formation, respectively [28,29]. Lower values of AI and TI are associated with a higher protective effect against cardiovascular disease. The AI values (0.4 ± 0.0, 0.4 ± 0.0, and 0.5 ± 0.0) and TI (0.5 ± 0.0, 0.5 ± 0.0, and 1.1 ± 0.2) of C-Hetero, C-Honey, and C-White, respectively, were higher than for C-Auto (0.2 ± 0.0 for AI and TI). The h/H values were significantly lower in C-Hetero (2.8 ± 0.0), C-Honey (2.5 ± 0.0), and C-White (2.2 ± 0.1), when compared to C-Auto (4.7 ± 0.2). Higher values of h/H ratio are regarded as being beneficial for human health since they are correlated with a higher PUFA content [30]. In addition, the stability of FAs to peroxidation was assessed by calculating the PoxI [18]. C-Auto obtained the highest PoxI value, while the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White) showed lower values. The higher the PoxI value, the more vulnerable the FAs to peroxidative damage. Table 1. GC-MS analysis of the fatty acid composition (in %) of total lipid extract from the four C. vulgaris strains (C-Auto, C-Hetero, C-Honey, and C-White). Data are presented as mean ± SD (n = 5). The most abundant FAs are highlighted in bold. Values with the same letter ( a , b , or c ) are statistically different (q < 0.05).
Abbreviations: ∑ PUFA, the sum of polyunsaturated fatty acids; ∑ MUFA, the sum of monounsaturated fatty acids; ∑ SFA, the sum of saturated fatty acids. Table 2. Lipidic quality indexes of the total lipid extracts of four Chlorella vulgaris strains, C-Auto, C-Hetero, C-Honey, and C-White. Data are represented as means ± SD, n = 5. Values with the same letter ( a , b , or c ) are statistically different (q < 0.05).

Lipidomics Profile of Chlorella vulgaris Strains
The lipidome characterization of C-Auto, C-Hetero, C-Honey, and C-White allowed the identification of 316 lipid species present in 4 C. vulgaris strains distributed across polar lipids (GL, PL, and sphingolipids) and neutral lipids (NL). The GL classes identified were the digalactosyldiacylglycerol (DGDG) monogalactosyldiacylglycerol (MGDG) and sulfoquinovosildiacylglycerol (SQDG). The identified PL classes include phosphatidylcholine (PC), lysophosphatidylcholine (LPC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and phosphatidylinositol (PI). Sphingolipids, including ceramides (Cer) and neutral lipids, diacylglycerols (DG) and TG, were also reported. The molecular compositions of the identified lipid species are described in Supplementary Table S1.
The relative abundances of the main lipid groups identified (PL, GL, and TG) were calculated and are shown in Figure 2. Results show that C-Auto exhibits a balanced total relative abundance in PL, GL, and TG, while the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White) show differences in the relative contents. The three heterotrophic strains have a lower content of GL compared to C-Auto. Moreover, C-Hetero and C-Honey have a higher relative abundance in TG, and C-White is characterized by a higher abundance in PL.

Lipidomics Profile of Chlorella vulgaris Strains
The lipidome characterization of C-Auto, C-Hetero, C-Honey, and C-White allowed the identification of 316 lipid species present in 4 C. vulgaris strains distributed across polar lipids (GL, PL, and sphingolipids) and neutral lipids (NL). The GL classes identified were the digalactosyldiacylglycerol (DGDG) monogalactosyldiacylglycerol (MGDG) and sulfoquinovosildiacylglycerol (SQDG). The identified PL classes include phosphatidylcholine (PC), lysophosphatidylcholine (LPC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and phosphatidylinositol (PI). Sphingolipids, including ceramides (Cer) and neutral lipids, diacylglycerols (DG) and TG, were also reported. The molecular compositions of the identified lipid species are described in Supplementary Table S1.
The relative abundances of the main lipid groups identified (PL, GL, and TG) were calculated and are shown in Figure 2. Results show that C-Auto exhibits a balanced total relative abundance in PL, GL, and TG, while the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White) show differences in the relative contents. The three heterotrophic strains have a lower content of GL compared to C-Auto. Moreover, C-Hetero and C-Honey have a higher relative abundance in TG, and C-White is characterized by a higher abundance in PL.  (Figure 3) showed discrimination between the four strains, namely between C-Auto and the remaining three strains grown under heterotrophic conditions (C-Hetero, C-Honey, and C-White) along PC1 (63.0%). The strain C-White split from C-Hetero and C-Honey along PC2 (28.2%).  (Figure 3) showed discrimination between the four strains, namely between C-Auto and the remaining three strains grown under heterotrophic conditions (C-Hetero, C-Honey, and C-White) along PC1 (63.0%). The strain C-White split from C-Hetero and C-Honey along PC2 (28.2%).  Hierarchical clustering and heatmap analysis comprising a set of the top 50 most discriminant species with the lowest q-values (q < 0.05) between all 4 strains of C. vulgaris is represented in Figure 4. The hierarchical clustering showed that the C-Auto split from the other three strains on a first level, followed by a split between C-White and the remaining strains, in a similar trend as seen in PCA analysis. The 50 most discriminating lipids comprised lipids from all the identified classes. A total of 42 lipids (9 MGDG, 3 DGDG, 3 SQDG, 4 PC, 1 LPC, 5 PE, 9 PG, 2 TG, 3 DG, and 3 Cer) were up-regulated in C-Auto. The strain C-White had 1 MGDG, 1 DGDG, and 2 PI up-regulated compared to the remaining Hierarchical clustering and heatmap analysis comprising a set of the top 50 most discriminant species with the lowest q-values (q < 0.05) between all 4 strains of C. vulgaris is represented in Figure 4. The hierarchical clustering showed that the C-Auto split from the other three strains on a first level, followed by a split between C-White and the remaining strains, in a similar trend as seen in PCA analysis. The 50 most discriminating lipids comprised lipids from all the identified classes. A total of 42 lipids (9 MGDG, 3 DGDG, 3 SQDG, 4 PC, 1 LPC, 5 PE, 9 PG, 2 TG, 3 DG, and 3 Cer) were up-regulated in C-Auto. The strain C-White had 1 MGDG, 1 DGDG, and 2 PI up-regulated compared to the remaining conditions, with the lowest content of these lipid species found in C-Auto. The remaining 46 species from the heatmap were downregulated or presented lower abundances in C-White. In C-Hetero, four SQDG were up-regulated when compared to the other strains, while the remaining discriminating lipid species were identified with lower abundance. Hierarchical clustering and heatmap analysis comprising a set of the top 50 most discriminant species with the lowest q-values (q < 0.05) between all 4 strains of C. vulgaris is represented in Figure 4. The hierarchical clustering showed that the C-Auto split from the other three strains on a first level, followed by a split between C-White and the remaining strains, in a similar trend as seen in PCA analysis. The 50 most discriminating lipids comprised lipids from all the identified classes. A total of 42 lipids (9 MGDG, 3 DGDG, 3 SQDG, 4 PC, 1 LPC, 5 PE, 9 PG, 2 TG, 3 DG, and 3 Cer) were up-regulated in C-Auto. The strain C-White had 1 MGDG, 1 DGDG, and 2 PI up-regulated compared to the remaining conditions, with the lowest content of these lipid species found in C-Auto. The remaining 46 species from the heatmap were downregulated or presented lower abundances in C-White. In C-Hetero, four SQDG were up-regulated when compared to the other strains, while the remaining discriminating lipid species were identified with lower abundance. The level of discrimination between C-Auto and the other strains of C. vulgaris hinders comprehension of how the identified lipid species vary within C-Hetero, C-Honey, and C-White. To overcome this, hierarchical clustering and heatmap analysis comprising a set of the top 50 most discriminant species with the lowest q-values (q < 0.05) between C-Hetero, C-Honey, and C-White was performed ( Figure 5). Results seem to indicate that C-Honey shows similarities with C-Hetero, while C-White exhibits a distinct lipid profile. A set of lipids comprising 13 lipid species (1 PC, 1 PG, 1 Cer, 2 DGDG, 3 DG, and 5 TG) had higher abundances in C-Hetero and C-Honey, but lower abundance in C-White. On the other hand, C-White was highlighted from C-Hetero and C-Honey, with a set of 12 lipids more abundant comprising GL (5 MGDG) and PL (1 PI, 2 PE, and 4 PC) esterified to LA. Overall, C-Honey showed a higher abundance of TG species with a low unsaturation degree, esterified to C18:1 and LA, while C-White showed the lowest abundance in TG species. In the case of C-Hetero, it discriminates from C-Honey, and C-White with 17 species of lipids, including mostly GL (1 DGDG, 5 MGDG, 7 SQDG) and 4 LPC.
The level of discrimination between C-Auto and the other strains of C. vulgaris hinders comprehension of how the identified lipid species vary within C-Hetero, C-Honey, and C-White. To overcome this, hierarchical clustering and heatmap analysis comprising a set of the top 50 most discriminant species with the lowest q-values (q < 0.05) between C-Hetero, C-Honey, and C-White was performed ( Figure 5). Results seem to indicate that C-Honey shows similarities with C-Hetero, while C-White exhibits a distinct lipid profile. A set of lipids comprising 13 lipid species (1 PC, 1 PG, 1 Cer, 2 DGDG, 3 DG, and 5 TG) had higher abundances in C-Hetero and C-Honey, but lower abundance in C-White. On the other hand, C-White was highlighted from C-Hetero and C-Honey, with a set of 12 lipids more abundant comprising GL (5 MGDG) and PL (1 PI, 2 PE, and 4 PC) esterified to LA. Overall, C-Honey showed a higher abundance of TG species with a low unsaturation degree, esterified to C18:1 and LA, while C-White showed the lowest abundance in TG species. In the case of C-Hetero, it discriminates from C-Honey, and C-White with 17 species of lipids, including mostly GL (1 DGDG, 5 MGDG, 7 SQDG) and 4 LPC.

Antioxidant Activity of Lipid Extracts from C. vulgaris Strains
The lipid extracts of the different C. vulgaris strains showed antioxidant activity, being able to distinctly inhibit the DPPH • and ABTS •+ radicals in a dose-dependent manner ( Figure 6). The results obtained in the DPPH • assay revealed that lipid extracts from C-Auto and C-Honey showed the highest antioxidant activity. C-Auto and C-Hetero extracts reached the inhibition of 20% of DPPH • radicals corresponding to an IC20 of 108.2 ± 9.8 µg mL −1 with a TE of 93.1 ± 8.7 µmol Trolox g −1 for C-Auto and an IC20 of 228.9± 21.5 µg mL −1 and TE of 44.0 ± 4.2 µmol Trolox g −1 for C-Honey ( Figure 6A). The lipid extracts from C-Hetero and C-White only promoted approximately 10% radical inhibition in all the concentrations evaluated. Statistically significant differences were observed at the 62.5 µg mL −1 of lipid extract between C-Auto and the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White), as well as at the concentration 250 µg mL −1 between C-Auto and C-Hetero, C-Auto and C-White and C-Honey and C-White.
The results obtained in the DPPH • assay revealed that lipid extracts from C-Auto and C-Honey showed the highest antioxidant activity. C-Auto and C-Hetero extracts reached the inhibition of 20% of DPPH • radicals corresponding to an IC20 of 108.2 ± 9.8 µg mL −1 with a TE of 93.1 ± 8.7 µmol Trolox g −1 for C-Auto and an IC20 of 228.9± 21.5 µg mL −1 and TE of 44.0 ± 4.2 µmol Trolox g −1 for C-Honey ( Figure 6A). The lipid extracts from C-Hetero and C-White only promoted approximately 10% radical inhibition in all the concentrations evaluated. Statistically significant differences were observed at the 62.5 µg mL −1 of lipid extract between C-Auto and the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White), as well as at the concentration 250 µg mL −1 between C-Auto and C-Hetero, C-Auto and C-White and C-Honey and C-White.
In the ABTS •+ assay, the lipid extracts from C-Auto showed the best antioxidant activity ( Figure 6B). C-Auto promoted 50% of inhibition of the ABTS •+ radical, reaching an IC50 value of 230.7 ± 1.8 µg mL −1 and a TE activity of 854 ± 0.7 Trolox µmol g −1 , while C-Honey and C-Hetero reached 20% inhibition of this radical, corresponding to an IC20 of 190.7 ± 13.4 µg mL −1 and TE of 42.1 ± 3.1 µmol Trolox g −1 for C-Hetero and an IC20 of 147.4 ± 10.6 µg mL −1 and a TE of 53.8 ± 4.0 µmol Trolox g −1 for C-Honey. C-White did not promote more than 10% radical inhibition in all concentrations evaluated. Statistically significant differences were observed in all concentrations of lipid extract tested (12.5, 62.5, 125, and 250 µg mL −1 ). Significant differences were observed between C-Auto and the three heterotrophically grown strains (C-Hetero, C-Honey, and C-White) at 62.5, 125, and 250 µg mL −1 . Moreover, between the three heterotrophically grown strains, significant differences were seen at 125 µg mL −1 between C-Honey and C-White, and at 250 µg mL −1 between C-Hetero and C-White and C-Honey and C-White. These results suggest that C. vulgaris strains exhibit different antioxidant potential. Overall, lipid extracts from C-Auto had the best antioxidant activity in both scavenging assays (ABTS •+ and DPPH • ), followed by C-Honey, C-Hetero, and, at last, C-White.

Anti-Inflammatory Activity of Lipid Extracts of C. vulgaris Strains
The lipid extracts from all four C. vulgaris strains were able to inhibit COX-2 activity at a concentration of 125 µg mL −1 of lipid extract (Figure 7). However, at lower concentrations of 20 µg mL −1 of lipid extract, none of the C. vulgaris strains showed COX-2 inhibitory activity, and at 60 µg mL − , only C-Auto and C-Hetero lipid extracts demonstrated the ability to inhibit COX-2.
At the concentration of 125 µg mL −1 of lipid extract, C-Auto showed the highest inhibitory potential of COX-2 activity (79.1 ± 7.7%), followed by C-Hetero (76.8 ± 5.9%), C-White (34.9 ± 1.5%), and finally C-Honey (13.6 ± 10.5%). Statistical analysis revealed significant differences in the inhibitory COX-2 potential between (a) C-Auto and C-Honey and (b) C-Auto and C-White, as well as between (c) C-Hetero and C-Honey and (d) C-Hetero and C-White lipid extracts. These results suggest that lipid extracts from C. vulgaris strains, with different lipid composition, exhibited different anti-inflammatory potential.
vulgaris strains exhibit different antioxidant potential. Overall, lipid extracts from C-Auto had the best antioxidant activity in both scavenging assays (ABTS •+ and DPPH • ), followed by C-Honey, C-Hetero, and, at last, C-White.

Anti-Inflammatory Activity of Lipid Extracts of C. vulgaris Strains
The lipid extracts from all four C. vulgaris strains were able to inhibit COX-2 activity at a concentration of 125 µg mL −1 of lipid extract (Figure 7). However, at lower concentrations of 20 µg mL −1 of lipid extract, none of the C. vulgaris strains showed COX-2 inhibitory activity, and at 60 µg mL − , only C-Auto and C-Hetero lipid extracts demonstrated the ability to inhibit COX-2. At the concentration of 125 µg mL −1 of lipid extract, C-Auto showed the highest inhibitory potential of COX-2 activity (79.1 ± 7.7%), followed by C-Hetero (76.8 ± 5.9%), C-White (34.9 ± 1.5%), and finally C-Honey (13.6 ± 10.5%). Statistical analysis revealed significant differences in the inhibitory COX-2 potential between (a) C-Auto and C-Honey and (b) C-Auto and C-White, as well as between (c) C-Hetero and C-Honey and (d) C-Hetero and C-White lipid extracts. These results suggest that lipid extracts from C. vulgaris strains, with different lipid composition, exhibited different anti-inflammatory potential.

Discussion
Chlorella vulgaris is one of the most commercialized microalga for food applications, with a high market value, and which has been widely used as food and as food ingredient [1,2]. Moreover, this microalga is rich in valuable nutrients, including lipids, such as PUFAs, which are described as having beneficial health effects. Most C. vulgaris biomass is produced under autotrophic conditions but other strains of C. vulgaris are also cultivated and commercialized, including strains obtained under heterotrophic conditions (herein called C-Hetero, C-Honey, and C-White), which reduces cultivation costs [3]. The strains C-Honey and C-White are also characterized by low pigment content and improved organoleptic properties, such as less flavor, smell, and colour [6]. Previous studies have shown that C-Auto and C-Hetero have different lipid profiles [7], while the lipidome of C-Honey and C-White strains were not identified before and were studied in this work for the first time. In addition, the bioactive potential of lipid extracts obtained from all C. vulgaris strains were evaluated and compared.
Between the four strains, C-Auto exhibited the highest lipid content (7.38 ± 0.3% in DW), while the remaining three strains showed similar content in lipids (approximately 5% in DW), but lower than C-Auto. These results are in agreement with previous reports on the lipid content of autotrophic C. vulgaris (approximately 9% in DW) [2,29] and the lower content of strains grown in heterotrophic conditions [7,9]. The dissimilar lipid

Discussion
Chlorella vulgaris is one of the most commercialized microalga for food applications, with a high market value, and which has been widely used as food and as food ingredient [1,2]. Moreover, this microalga is rich in valuable nutrients, including lipids, such as PUFAs, which are described as having beneficial health effects. Most C. vulgaris biomass is produced under autotrophic conditions but other strains of C. vulgaris are also cultivated and commercialized, including strains obtained under heterotrophic conditions (herein called C-Hetero, C-Honey, and C-White), which reduces cultivation costs [3]. The strains C-Honey and C-White are also characterized by low pigment content and improved organoleptic properties, such as less flavor, smell, and colour [6]. Previous studies have shown that C-Auto and C-Hetero have different lipid profiles [7], while the lipidome of C-Honey and C-White strains were not identified before and were studied in this work for the first time. In addition, the bioactive potential of lipid extracts obtained from all C. vulgaris strains were evaluated and compared.
Between the four strains, C-Auto exhibited the highest lipid content (7.38 ± 0.3% in DW), while the remaining three strains showed similar content in lipids (approximately 5% in DW), but lower than C-Auto. These results are in agreement with previous reports on the lipid content of autotrophic C. vulgaris (approximately 9% in DW) [2,29] and the lower content of strains grown in heterotrophic conditions [7,9]. The dissimilar lipid accumulation between the four C. vulgaris strains may be due to cultivation conditions, such as light intensity, temperature, or CO 2 supply [31,32]. The lower lipid content observed in the heterotrophically grown strains (C-Hetero, C-Honey, and C-White) can also be associated with the lower amount of chloroplasts in these strains, inferred by the lower glycolipid (lipids typical of chloroplasts) content and pigments, particularly chlorophylls, which are in lower abundance in these strains [3,6].
The FA profile analysis revealed that C-Auto showed a higher content of omega-3 PUFAs, with ALA representing the most abundant FA, while the lowest omega-3 content was reported for C-White. In contrast, C-Hetero and C-White exhibited the highest content in omega-6 PUFAs, in similar levels, whereas C-Auto showed the lowest omega-6 PUFAs amount. These results are in accordance with previous reports on the comparison of C. vulgaris grown auto-and heterotrophically [7,30]. The four C. vulgaris strains can be used as source of essential PUFAs, such as ALA and LA, which can only be obtained through diet and are important precursors of longer omega-3 and omega-6 PUFAs, respectively. The studied C. vulgaris strains have higher abundances in ALA and LA when compared to other microalgae, and other abundant sources of LA (e.g., canola oil) [33]. Omega-3 PUFAs are often associated with health benefits in immunity regulation, while a moderate intake of omega-6 PUFAs is associated with lower risk of cardiovascular disease, highlighting the importance of having dietary sources of both omega-3 and omega-6 PUFAs in different abundances [10,34]. Despite the four C. vulgaris strains having a distinct FA profile, they exhibit an interesting nutritional composition in terms of FA composition. Therefore, they can represent a source of omega-3 and omega-6 PUFAs, providing health benefits for people with different nutrient deficiencies in these PUFAs. Another abundant FA in C. vulgaris strains produced heterotrophically (C-Hetero, C-Honey, and C-White) was the OA. This FA is quite abundant in olive oil and is described with a protective effect against inflammation and cancer [35,36].
According to the calculated index values, the four C. vulgaris strains had good lipidic nutritional indicators, including AI and TI, which are cardiovascular disease risk predictors, and h/H ratio ( Table 2). Lower values of AI and TI usually indicate a higher protective effect against cardiovascular disease [30], and were observed in the four C. vulgaris strains. On the other hand, an elevated h/H is associated with a high content in terms of PUFAs, and a beneficial effect by lowering cholesterol levels [34,37]. The four C. vulgaris strains showed high values of h/H, and C-Auto excelled with the best ratio. These results support the nutraceutical value of FAs from the studied C. vulgaris strains. Indeed, the health potential of FAs from C. vulgaris were already addressed in the literature. For instance, in mice fed with a high-fat diet, the administration of FAs from C. vulgaris promoted a beneficial effect on hyperglycemia, hyperlipidemia, hepatic steatosis, and adipocyte hypertrophy [38]. Another study evaluated the effect of auto-and heterotrophic C. vulgaris in zebrafish and showed that both strains have anti-obesity, anti-steatosis, and anti-inflammatory activity [39]. The susceptibility of the lipid extracts of the four C. vulgaris strains was also assessed. Lower values of PoxI were observed for the heterotrophically grown strains (C-Hetero, C-Honey, and C-White), of which C-White achieved the best index value, suggesting a higher stability of the FAs present in the lipid extract [18,40].
The lipid profile of the four strains of C. vulgaris was determined by lipidomic analysis using high-resolution RP-LC-MS/MS, which allowed the identification of 316 lipid species, including polar and neutral lipids. The polar lipid species were similar in all four strains, and in accordance with data previously reported in the literature for C-Auto and C-Hetero analyzed by HILIC-LC-MS/MS [7,8]. However, differences in the relative amounts of the lipid species were found between the strains (Figure 4), suggesting a specific lipid signature for each strain but with similarities between the strains grown heterotrophically. For instance, C-Auto stands out, with a higher amount of lipids esterified to PUFAs, especially GL (MGDG and DGDG), while C-Hetero, C-Honey, and C-White were characterized by abundant lipids with a lower unsaturation degree. Comparing the three strains grown under heterotrophic conditions (C-Hetero, C-Honey, and C-White), changes in lipid composition were observed between them. Lipid extracts from C-Hetero and C-Honey strains exhibited a high abundance of TG species, including a wide diversity of TG molecules mainly esterified to OA and essential FAs, namely ALA and LA. The TGs are main components of edible oils, and the physiochemical and nutritional properties of these are determined by the TG molecules present [41]. Edible oils are rich in omega-3 and -6 PUFAs, and are currently obtained from fish (e.g., fish oil), which represents a limited supply, reinforcing the need for sustainable sources of edible oils [42]. Moreover, the production volume as well as the market value of fish oil is expected to increase, demanding fast and sustainable alternatives [43,44]. Therefore, microalgae including C. vulgaris, can be a promising source of edible oils and an alternative to fish oil for applications in food, feed, and in dietary supplements.
Polar lipids, such as GL, are highlighted as efficient delivers of PUFAs, with better bioavailability compared with TGs [12,45]. Polar lipids are the main components of lecithins, which are used as food ingredients and in the formulation of nutraceuticals [46]. In addition, a few polar lipids from microalgae have been reported as modulators of inflammation, such as some PC, PG, MGDG, DGDG, and SQDG lipids [29], which were all identified in the four C. vulgaris strains. Moreover, species, such as MGDG (32:0), MGDG (32:2), MGDG (34:1), and SQDG (32:0), identified in all C. vulgaris strains and with a higher relative abundance in C-White, were previously reported with anti-viral, anti-obesity, and anti-tumor activities [47][48][49][50] The bioactive potential of C. vulgaris strains was assessed through DPPH • and ABTS •+ radicals scavenging assays. All lipid extracts from C. vulgaris strains exhibited antioxidant activity, with C-Auto showing the best antioxidant potential and C-White the lowest ( Figure 6). Lipid extracts from C-Hetero, C-Honey, and C-White have a similar content in the total amount of PUFAs, with different proportions in omega-3 and omega-6 PU-FAs, and dissimilar antioxidant activities. The lower antioxidant activity of C-Hetero, C-Honey, and C-White extracts could be correlated with the n−6/n−3 (Table 2), and the lower abundance of omega-3 PUFAs when compared with C-Auto. However, differences among the antioxidant effect between the strains grown heterotrophically can be due to the different omega-3 and omega-6 PUFA contents, or the contribution of the synergistic effects of other compounds with antioxidant properties, such as pigments, cannot be ruled out [6]. Altogether, results suggest that C. vulgaris strains can be considered a promising source of antioxidant lipids for the formulation of added-value products for the food and nutraceutical industries. The anti-inflammatory potential of lipid extracts was also assessed by the in chemico COX-2 inhibitory assay. Lipid extracts from all four strains of C. vulgaris inhibited the activity of COX-2, with C-Auto and C-Hetero recording the highest anti-inflammatory activity. Lipid extracts from C-Auto and C-Hetero were already described with anti-inflammatory potential [7,39]. When compared to other microalgae (Gloeothece sp., Tetraselmis sp. and Phaeodactylum tricornutum) tested at similar and higher concentrations (500 µg mL −1 )) of the lipid extracts [51,52], C-Auto and C-Hetero (125 µg mL −1 )) revealed higher anti-inflammatory potential, while C-White showed similar activity. Differences in the anti-inflammatory potential observed between heterotrophically grown strains (C-Hetero, C-Honey, and C-White) can be associated with the lipid species competing as a substrate for COX-2. As previously mentioned, algal lipids have been reported as modulators of inflammation, and several PL and GL species may play an important role in reducing unregulated inflammatory processes [53].
The four C. vulgaris strains evaluated in this study seem to represent a source of bioactive lipids with functional properties, ones that can be applied in the formulation of promising new ingredients in the development of foods and dietary supplements. Moreover, they are an interesting alternative for consumers with different nutritional requirements or with plant-based diets, such as vegetarians and vegans.

Conclusions
This work described the differences in FA and lipid profiles, as well as the antioxidant and anti-inflammatory properties of four strains of C. vulgaris (C-Auto, C-Hetero, C-Honey, and C-White). These strains are already approved for human consumption and marketed as food ingredients and dietary supplements. The same FAs, lipid classes, and lipid species were found in the lipidome of all C. vulgaris strains. Despite the similarities, statistical analysis showed a clear difference in the relative abundance of certain FAs, lipid classes, and species, revealing a strain-specific lipid signature. The autotrophically grown C. vulgaris was highlighted with a high abundance in omega-3 PUFAs and GL esterified to PUFAs, while the heterotrophically grown strains showed a higher abundance of low unsaturated species. Moreover, C-Hetero and C-Honey were abundant in TG species, esterified to OA and LA, while C-White had a higher content in polar lipids. The four strains exhibited antioxidant properties, with the highest value verified in C-Auto. Furthermore, among the four strains, C-Auto and C-Hetero exhibited the best anti-inflammatory potential. Altogether, these results contribute to the valorization of all four strains of C. vulgaris as food ingredients and supplements with nutritional and functional value for different markets.