Introduction

Inadequate protection against UV radiation is related to sunburn, hyperpigmentation, the inhibition of the skin’s immune system, and the development of melanoma and nonmelanoma skin cancers, among other health risks1,2. Further than being an esthetic issue, rising levels of UV radiation are a global public health concern that demands the development of new skin photoprotection products3,4.

UV radiation corresponds to 10% of the total radiation reaching the Earth’s surface and is of major interest for skin photoprotection because of its high energy, which is capable of penetrating both the epidermis and dermis. This can result in adverse effects such as sunburns, mutations, and the appearance of skin cancer cells5. Within the UV spectrum, UVC radiation (λ = 100–290 nm)3 is entirely absorbed by the ozone layer, whereas UVB radiation (λ = 290–320 nm)3 accounts for 4–5% of the UV radiation that reaches the Earth’s surface. The majority, 95%, is UVA radiation (λ = 320–400 nm)6. The remaining 90% of the radiation that reaches the Earth’s surface comprises 40% infrared radiation (λ = 760 nm–1 mm) and 50% visible light (λ = 400–760 nm). Additionally, artificial sources such as LED lights, flashlights, and halogen bulbs can emit these types of radiation7. They can interact with the epidermis, dermis, and subcutaneous tissue, contributing to various skin-related issues, including photoaging, inflammation, overheating, hyperpigmentation, collagen degradation, and oxidative damage5,8.

There are two mechanisms by which radiation interacts with the skin. The primary mechanism, or absorption mechanism, involves chromophores in skin tissues, such as melanin, urocanic acid, and flavins, which absorb radiation and induce biological responses, including skin darkening and, unfortunately, the development of certain melanomas3,9. The secondary mechanism, or photosensitization reactions, involves the transfer of high energy from light to skin biomolecules, inducing the production of reactive oxygen (ROS) and nitrogen (NOS) species. This results in structural damage and further adverse effects3,6.

Accordingly, there are two photoprotection mechanisms. The primary mechanism comprises organic or inorganic sun filters (sunscreens) that absorb or reflect radiation, and the secondary mechanism includes molecules that neutralize ROS and NOS when they are produced6,10.

Although many sunscreens exist, some traditional synthetic organic UV–Vis filters can have adverse effects on both human health and the environment11. These effects on the skin include photosensation, phototoxicity, dermatitis, allergies, and alopecia12,13. In aquatic ecosystems, the negative impacts include water pollution, bioaccumulation, coral bleaching11,14, and hormonal effects on aquatic organisms15,16. These issues have drawn the attention of health and environmental authorities worldwide, prompting efforts to limit the use of some of these organic filters in cosmetic products and to explore safer products for skin photoprotection15,17.

In this context, algae are an important source of active compounds that are already being employed in the pharmaceutical, food, and cosmetic industries because of their unique chemical structures and health-promoting properties18,19,20,21. Compounds isolated from algae have multiple biological activities, such as photoprotective, skin enzyme inhibitory, and anti-inflammatory effects. These compounds include pigments18,22, phenolic compounds23,24, sulfated polysaccharides25, and mycosporine-like amino acids26,27, which are valuable for cosmetic applications17,21,28,29,30. Some of these natural products offer photoprotective activity through both primary and secondary mechanisms with fewer secondary effects, and they can be added to traditional sunscreen formulations to reinforce a product’s skin-protecting effect11,19,20. On the other hand, algal biomass is considered a renewable source, as it can be either sustainably obtained from the environment, harnessing massive algae blooms31, or cultivated in large-scale systems where targeted active metabolites can be induced32,33.

Considering that tropical countries such as Colombia experience high UV indices annually34, marine algae in these ecosystems are expected to develop different photoprotection mechanisms, particularly through secondary metabolites. These compounds help them adapt to variable environmental and sunlight conditions, and can be seized for skin photoprotection18,19,35. Despite the region’s unique algae diversity, with at least 619 reported algal species36, there are currently no systematic studies on natural products derived from marine algae on Colombia’s Caribbean coast37.

Even though there are various methods for evaluating photoprotective activity, handling large numbers of samples requires an efficient approach in which information can be efficiently analyzed. This should facilitate the rational selection of samples with the most promising activities by integrating data from complementary methodologies. To the best of our knowledge, literature lacks such methodologies specifically applied to algae. Most existing approaches compare samples by separately evaluating activity and composition parameters. However, there are a few reports in which information on variations in the polyphenol content, photoprotective activity, and antioxidant activity variations in different seasons has been analyzed simultaneously in Lamiaceae species38. This approach allows the identification of the seasonal influence on polyphenol content and, consequently, on photoprotective activity.

This work aims to implement a new screening methodology for the selection of photoprotective brown macroalgae extracts. The approach integrates results from in vitro photoprotection activity assays (sun protection factor (SPF), UVA ratio (UVAr), and critical wavelength (λc)), and the antioxidant activity of brown algae extracts. These data are integrated with HPLC-DAD metabolic profile evaluations using Heatmaps and Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS) algorithm. This strategy facilitates the selection of the most promising photoprotective extracts and the identification of compounds responsible for the observed activity. This is one of the first reported methodologies attempting to explore algae diversity as a source of photoprotective compounds using methodologies that are commonly used in natural product research laboratories and are easily affordable for researchers worldwide.

Results and discussion

Although the Rhodophyta phylum has been the most studied in terms of photoprotective activity, the most abundant algae in the Caribbean Sea are those belonging to the Heterokontophyta phylum, whose abundance has notably increased in response to coral reef degradation. Considering this high biomass availability and the adaptation of these organisms to high sunlight exposure year-round, they possess photoprotection mechanisms, in terms of metabolite production, that can be explored.

To do so, a methodology for the screening of promising photoprotective brown macroalgae extracts was implemented, enabling the evaluation and tracking of this activity in each sample set. For this purpose, 17 algae samples belonging to the Ochrophyta phylum were collected in Providencia Island in September 2021, in which the Dictyota, Canistrocarpus, Lobophora, Padina, Sargassum, Stypopodium, and Turbinaria genera are represented (Supplementary Table S.1). The selection included samples from different genera, as well as samples from the same genera, with the goal of identifying differences in metabolic production and photoprotection activity within the implemented screening methodology.

From the 17 samples and their biological triplicates (51 in total), 204 extracts were obtained: FO, FB, FM, and FW. The extraction yield ranged from 2 to 9% for the FO extracts, from 0.01 to 24% for the FB extracts, from 3 to 15% for the FM extracts, and from 0.5 to 3% for the FW extracts.

General remarks on the UV–vis absorption spectra of crude extracts

The UV–Vis spectra of the crude extracts in Fig. 1 show the differences between the extracts. The FO extracts (Fig. 1a and e) presented the highest absorbance at the examined wavelengths, followed by the FB (Fig. 1b and f), FW (Fig. 1d and h), and FM extracts (Fig. 1c and g).

Fig. 1
figure 1

UV–Vis absorption spectra (λ = 290–500 nm) of the extracts obtained from 17 selected brown seaweed samples collected in Providencia. (a–d) spectra of extracts from Dictyota ( DI01, DI02, DI03, DI04, DI05, DM, DP01, and DP02) and Canistrocarpus ( CCe and CCr) genera: (a) FO extracts, (b) FB extracts, (c) FM extracts, and (d) FW extracts. (e–h) spectra of extracts from Stypopodium ( SS01, SS02, and SS03), Saragassum ( SP), Lobophora ( LV), Padina ( PG), and Turbinaria ( TT) genera: (e) FO extracts, (f) FB extracts, (g) FM extracts, and (h) FW extracts.

Full size image

The samples from the Dictyota and Canistrocarpus genera (Fig. 1a–d) presented absorbance maxima within the 350–480 nm region which corresponds to typical UVA absorbance bands of pigments such as chlorophylls and carotenoids, which are the main photoprotection mechanisms in photosynthetic organisms. Although there were differences between the samples, the Canistrocarpus cervicornis (CCe)-FO extract had the highest absorbance and the broadest absorption band (λmax = 420 nm and 440 nm) (Fig. 1a). A similar trend was observed in samples belonging to the Stypopodium (SS), Turbinaria (TT), Padina (PG), Sargassum (SP), and Lobophora (LV) genera (Fig. 1e–h), in which there were two absorbance maxima regions: between 290 and 295 nm and between 415 and 445 nm. Among those samples, SS03-FO showed the highest absorbances in the studied ranges (λmax = 295 nm and 415 nm, respectively) (Fig. 1e).

The FB extracts (Fig. 1b and f) presented distinctive characteristics, particularly in samples from other genera than the Dictyota or Canistrocarpus genera. Among those extracts, the absorbance maxima were distributed across the 290–500 nm range, with maxima in the PG (λmax = 295, 312, 326, and 437 nm), LV (λmax = 450 nm) and SS (λmax = 297 nm) extracts, whereas the Dictyota (DI, DM, and DP) and Canistrocarpus (CCe and CCr) extracts presented lower absorbance maxima at approximately 430–450 nm, with no significant differences among them. Comparatively, there were minimal differences between the FM and FW extracts (Fig. 1c, d, g, and h); however, they consistently yielded the lowest absorbances among all the studied extracts.

In vitro photoprotection and antioxidant activity assays

SPF, UVAr, and λc are parameters used to estimate the photoprotection capacity of the extract against UVA and UVB radiation. Integration of these parameters into the early-stage search for photoprotective compounds is proposed as a tool to guide the selection of promising extracts. The results from the in vitro photoprotection and antioxidant activity assays (Supplementary Table S.2) were simultaneously analyzed through a heatmap, grouping samples on the basis of their similarities (Figure 2).

Fig. 2
figure 2

Heatmap representation of the calculated in vitro sun protection factor (SPF at 50 ppm), UVA ratio (UVAr*), critical wavelength (λc), and antioxidant activity (AOX) assays. FO/FB/FM/FW stands for extract type. The numbers on the left differentiate the seven identified clusters, and the numbers in red indicate the clusters with the most promising photoprotective activity. * UVAr values correspond to the mean of three replicates at different concentrations, as no significant differences were observed between extracts at the evaluated concentrations.

Full size image

The results showed that extracts spontaneously grouped into seven clusters based on their polarity. Clusters 5 to 7 grouped mainly FO extracts which displayed the most promising photoprotective activity. Specifically, cluster 5 included the extract with the highest recorded SPF value (SS03-FO = 2915), whereas cluster 7 had the highest UVAr value (CCe-FO = 3623). All the extracts within clusters 5 and 6 demonstrated antioxidant activity. Clusters 2 and 3 included FB and FW extracts with intermediate values for the studied photoprotection parameters, whereas cluster 1 included FM and FB extracts with no antioxidant activity and the lowest SPF (DI01-FM = 0.146), UVAr (DI01-FM = 0.167), and λc (DI03-FM = 336) values. The λc parameter did not have a clear maximum among the evaluated extracts, but the higher values were identified in clusters 6 and 7, whereas the minimum values were in cluster 1.

These results reflect not only the differences between extracts from different species but also the differences between extracts from the same species obtained with several solvents. While the first one is an estimation of the chemical diversity of the evaluated samples, the second one shows how the extraction procedure allowed extracts with different compositions to be obtained, a helpful trait in further purification steps. In this way, extracts located in clusters 5, 6, and 7 were prioritized with the most promising photoprotection activity, yielding the selection of 17-FO and one FB extract.

The 18 preselected extracts and their triplicates (54 samples in total) were further analyzed using HPLC-DAD. Each chromatogram was baseline-corrected using the Asymmetric Least Squares (AsLs) algorithm, fragmented into 28 retention-time variable fragments, and processed via MCR-ALS. With this methodology, chromatograms in the dataset were simultaneously corrected, aligned, and compared, obtaining pure spectra (UV–Vis spectra) and elution profile information for each detected component in the chromatographic dataset. This also affords retention time-shifts, coelution, and asymmetric peak phenomena, among other characteristics of HPLC-DAD chromatographic datasets39.

This analysis resulted in the detection of 268 components for the 28 segments in the 54 samples in the dataset. Details on the estimated number of detected components and figures of merit for each fragment are presented in Supplementary Table S.3. The R2 (explained variance) and %LOF (% lack of fit) values were above 98% and less than 12% of the fragments, respectively. These values were interpreted as acceptable for most of the fragments, considering the complexity and differences between samples and their replicates. The MCR-BANDS algorithm showed small differences between the maximum (fnmax) and minimum (fnmin) values of the relative contribution function (fn) for each component, indicating that rotation ambiguities were removed within the applied constraints39,40.

After MCR-ALS analysis, the results were retrieved in a relative concentration matrix (RCM) comprising 18 samples (54 chromatograms) and 268 components, each characterized by a UV-Vis spectrum between 200 and 800 nm. We applied the rule proposed by Vinaixa and coworkers41 to distinguish between analytical and biological variations in the detected features, identified as components (N) with specific retention times (tr) and UV-Vis spectra. As proposed by the authors, features were retained when the Coefficient of Variation Threshold (CVT) exceeded 20%, indicating that the variation was predominantly attributed to biological differences among samples. When this rule was applied, no components were excluded from the dataset, and subsequent multivariate analysis was conducted on the 54 X 268 RCM matrix.

Principal component analysis (PCA), Figure 3, revealed that the first two components accounted for 60.4% of the variance, yielding an R2 value of 0.853 and a Q2 value of 0.502. The score plot presented in Figure 3a shows the clustering of the samples, with no significant differences among them. However, within this representation, the three replicates of C. cervicornis (CCe-FO) did not cluster together, either because of changes in their chemical composition or the amount of the detected components. In contrast, C. crispatus (CCr), a sample belonging to the same genus, had no such differentiation, grouping within the largest cluster with Dictyota, Sargassum, and Lobophora samples. Along with the mentioned differences for C. cervicornis extracts, Hierarchical Cluster Analysis (HCA) in Figure 3b revealed that all S. zonale FO extracts (SS01-FO, SS02-FO, and SS03-FO) clustered together as a second differentiable group.

Fig. 3
figure 3

Principal Component Analysis (PCA) of the relative concentration matrix (RCM). (a) Score plot, (b) Hierarchical cluster analysis (HCA), (c) Loadings plot, and (d) UV–Vis spectra of selected loadings. C: Canistrocarpus cervicornis (CCe) cluster; S: Stypopodium zonale (SS) cluster; O: other samples cluster. The pure spectra matrix (ST) was used to characterize the absorbance regions and color the variables in the loadings plot: () (UV absorption; λ = 290–400 nm); () (visible absorption; λ = 400–800 nm); () (UV–Vis absorption; λ = 290–800). The black circumferences represent sample clusters and important variables for extract differentiation.

Full size image

The pure spectra matrix (ST) was used to characterize the absorbance regions corresponding to each (N) component. This characterization was incorporated into the loading plot, as depicted in Figure 3c. Variables exhibiting absorption in the UV region are represented in green, those absorbing in the visible region are denoted in blue, and variables absorbing in the UV and Vis regions are indicated in red. The results showed that variables that have a strong influence on the PCA components and the location of the observations in the Score plot (e.g., 51, 159, 254, 220, 18) absorb mainly in the visible and UV-Vis regions, with absorption maxima at 270 nm and between 400 and 600 nm, as shown in Figure 3d. These findings suggest that these pigments are photosynthetic pigments with distinct distributions among samples.

Partial Least Squares (PLS) analysis, Figure 4, integrates the chemical information obtained from HPLC-DAD metabolic profiling (RCM: 54 X 268) and the measured photoprotective activity parameters (SPF, UVAr, λc, and antioxidant activity) as outputs (Y-matrix: 54 X 4) in the analysis. The Score Plot (Figure 4a) shows three distinct groups: C. cervicornis (C), S. zonale (S), and other genera (D) extracts. Once again, the C. cervicornis (Ce-FO) extract, on the one hand, and the S. zonale (SS-FO) extract, on the other hand, were different from the other extracts, reflecting both higher photoprotection parameters and the most differentiable chemical profile. The VIP plot (Figure 4b) showed that 74 variables were important for this projection, each characterized by its retention time (tr) and UV–Vis spectrum. The main important variables for S. zonale extracts differentiation were 81, with two absorption maxima in the UV region (tr = 19.0–19.9 min; λmax = 210 and 310 nm), and 107, with absorption maxima in the UV region (tr = 21.9-22.5 min; λmax = 204, 230, 250 and 296 nm) (Figure 4d). In C. cervicornis extracts, variables that influence their differentiation include 254 (tr = 40.0–43.0 min; λmax = 480, 524, and 648 nm), 220 (tr = 33.5–36.0 min; λmax = 472, 528, and 648 nm), and 18 (tr = 5.0–7.9 min; λmax = 476, 524, and 648 nm). To the same extent, the loading plot (Figure 4c), confirmed that other important variables in extract differentiation were 127 (tr = 23.2–24.0 min; λmax = 432, 664 nm), 51 (tr = 15.5–17.3 min; λmax = 268, 452 nm), and 159 (tr = 24.6–25.3 min; λmax = 264, 340, 432, 624 and 660 nm).

Fig. 4
figure 4

Partial least square analysis (PLS) of the relative concentration matrix (RCM). The outputs (Y-matrix) corresponded to the measured in vitro photoprotection parameters (SPF at 50 ppm, UVAr*, λc, and antioxidant activity). (a) Score plot, (b) VIP plot for variables with VIP scores grater than 1 (important variables), (c) Loadings plot, and (d) UV–Vis spectra of selected variables (81, 107, and 127). C: Canistrocarpus cervicornis (CCe) cluster; S: Stypopodium zonale (SS) cluster; O: Other samples cluster. The pure spectra matrix (ST) was used to characterize the absorbance regions and color the variables: ( ) (UV absorption; λ = 290–400 nm); () (visible absorption; λ = 400 -800 nm); and () (UV–VIS absorption; λ = 290–800). The black circumferences represent sample clusters and important variables for extract differentiation.* UVAr values correspond to the mean of three replicates at different concentrations, as no significant differences were observed between extracts at the evaluated concentrations.

Full size image

Considering these differences, which grouped the most promising photoprotection activity and the most differentiated chemical profile, C. cervicornis (CCe-FO) and S. zonale (SS01, SS03, and SS03 FO) extracts were selected for further identification of the compounds accounting for the tracked photoprotection activity.

Chemical compositions of the organic extracts of C. cervicornis (CCe) and S. zonale (SS)

Supplementary Figure S.1 shows the results of the 1H-NMR analysis of the C. cervicornis and S. zonale extracts. The spectra revealed that the extracts primarily consisted of fatty acids with characteristic signals at δH 1.27 for methylene protons in fatty acids, δH 2.1 for protons in allylic methylene, δH 2.3 for α methylene in carbonyl carbons, and δH 2.8 for protons of the methylene allylic to two double bonds. Downfield, there are less intense signals between δH 6 and 7 and between δH 9 and 10.5 that are associated with pigments (chlorophylls42 and carotenoids43) in the extracts. In addition, signals between δH 3 and 6 were related to the presence of terpenes44.

Fractionation of the CCe-FO extract by column chromatography allowed the recovery of 10 fractions. The photoprotection activity assays for the fractions are presented in Supplementary Figure S.2. The results were used to track the activity, indicating the highest SPF values in fractions D09, D08, and D07 and the highest UVAr value observed for fraction D08. The selection criterion did not include λc, as no significant differences were observed between the crude extract and its fractions.

The 1H- and 13C-NMR spectra of fractions D07 and D08 indicated the presence of pheophytin a45,46 (Supplementary Figure S.3 and Supplementary Figure S.4) as the major component, which was attributed to the observed activity in these fractions. Fraction D09 was subjected to preparative HPLC due to its abundance and significant value in photoprotection activity assays (Supplementary Figure S.2). This process yielded 32 new fractions (D0901 to D0932) in which NMR analysis revealed the presence of isolinearol (D0914; 1.7 mg; tr = 21.6–23.0 min; Figure 5a, Supplementary Figure S.5 and Supplementary Figure S.6)47,48, and fucoxanthin (D0924; 3.0 mg; tr = 40.5–41.8 min; Figure 5b, Supplementary Figure S.7 and Supplementary Figure S.8)49 with absorbance maxima at 210 nm and 470 nm, respectively.

Fig. 5
figure 5

Chemical structures of (a) isolinearol, (b) fucoxanthin (c) 2,5,7-trihydroxy-2-pentadecylchroman-4-one(1), (d) stypodiol, (e) stypoldione identified in Canistrocarpus cervicornis (CCe) and Stypopodium zonale (SS02) FO extracts.

Full size image

Fucoxanthin was isolated as an orange solid, that was subjected to further photoprotection activity assays, as presented in Fig. 6. These results show that fucoxanthin is the compound that accounts for the tracked photoprotective activity of the CCe-FO extract and demonstrates the potential for photoprotection within the UV region and protection against visible light, which is not measured with classical in vitro photoprotection activity assays but is important because of the negative effects of this radiation on skin health5,8.

Isolinearol was identified by comparison of its 1H- and 13C-NMR spectra with those previously reported by Texeira and coworkers47. This secodolastane diterpenoid is considered a chemotaxonomic marker for the Dictyota/Canistrocarpus genera50 and specifically for Canistrocarpus cervicornis in the tropical Atlantic American region51.

On the other hand, the S. zonale (SS02-FO) extract was fractionated by column chromatography yielding 17 fractions (S01 to S17). The results for the photoprotection activity of these fractions are presented in Supplementary Figure S.2. The data suggested that the highest photoprotection activity was concentrated in fractions S06, S07, and S11, as shown by the SPF values, whereas the highest photoprotective activity against UVA was found in fraction S06. The λc values were not considered, as both the crude extract and the fractions presented comparable values. 1H- and 13C-NMR spectral analysis allowed the identification of pheophorbide a52 (Supplementary Figure S.9) and fucosterol53 in fraction S06.

Fraction S07 was separated by preparative HPLC, considering the higher values in photoprotection activity assays (Supplementary Figure S.2). The results yielded 6 new fractions (S0701 to S0706). NMR analysis revealed the presence of compound 1 in fraction S0705 (10.1 mg; tr= 17.8-20 min; Figure 5c, Supplementary Figure S.10, Supplementary Figure S.11, Supplementary Figure S.12, Supplementary Figure S.13, and Supplementary Figure S.14) and stypodiol in fraction S0703 (2.7 mg; tr = 10–10.3 min; Figure 5d, Supplementary Figure S.15 and Supplementary Figure S.16)54, with absorption maxima at 207, 227, 287, and 320 nm, and 205 nm, respectively. Stypoldione was identified in fractions S08 and S0955,56 (Figure 5e, Supplementary Figure S.17 and Supplementary Figure S.18), fucoxanthin was the main compound in fraction S10, and fatty acids constituted fractions S11 and S12.

Compound 1 was isolated as an off-white solid, with a molecular formula of C24H38O5 based on its (+) HRESIMS data calculated for the protonated molecular ion [M+H]+ at m/z 407.2789 (calculated for C24H39O5+ at m/z 407.2792, Δ 0.74 ppm). The 13C-NMR spectrum showed the presence of 20 signals, including one quaternary carbonylic carbon (δC 197.16), three sp2 oxygenated quaternary carbons (δC 168.17, 165.0, and 162.17), one hemiacetal carbon (δC 103.62), one sp2 quaternary carbon (δC 103.09), two sp2 methines (δC 96.88, 96.68), one sp3 methylene (δC 46.00), one sp3 methyl (δC 14.46) and sp3 aliphatic methylenes characteristic of an aliphatic chain (between δC 41.98 and δC 24.76). The 1H-NMR spectrum suggested the presence of an aliphatic methyl proton (δH 0.88), three protons of aliphatic carbons (δH 1.27, 1.47, and 1.85), two protons of cyclic methylenes (δH 2.60, 2.91), two aromatic protons (δH 5.82, 6.00) and two aromatic hydroxyl protons (δH 10.22, 11.94). (+) HRESIMS data (Supplementary Figure S.19) was used to identify the C-15 side chain length of compound 1. In addition, the HRESIMS/MS spectra of 407.2789 showed ions at 153.0185 (C7H5O4+, Δ 1.31 ppm) and 127.0395 (C6H7O3+, Δ 3.94 ppm), confirming the proposed structure. The first ion corresponds to fragmentation between C2 and O1, C3 and C4, and the oxidation of the alcohol in C5, whereas the second ion corresponds to the fragmentation between C10 and C4, and between C2 and O1 (Supplementary Figure S.19).

These spectroscopic data are identical to those reported for 2,5,7-trihydroxy-2-pentadecylchroman-4-one (spiralisone C), which was previously isolated from the brown alga Zonaria spiralis57, and the plant seeds of Syzygium cumini58. Chroman-4-ones are important heterobicyclic secondary metabolites with various reported biological activities including anticancer, antioxidant, antimicrobial, and antifungal activities [58,59]. This is the first report of a chromanone in the Stypopodium genus, although the corresponding chromones have been previously identified in the brown algae Zonaria tournefortii59, Stypopodium flabelliforme60, and Stypopodium zonale61 (Phaeophyceae). As discussed by Kamboj and Singh (2020), chromones and chromanones are structurally related, but the absence of the C2–C3 double bond in chromanones results in different biological activities.

The results of further photoprotection activity assays of pure compound 1, shown in Figure 6, revealed the highest SPF values and the lowest UVAr values from all the isolated compounds, as well as a λc at 335.9 nm. These results are comparable with those of BP-3, which was used as a positive control, indicating its potential as a UVB filter. In contrast, the UVAr and λc values for compound 1 are lower than those reported for pheophytin a, pheophorbide a, and fucoxanthin, which presented better UVA photoprotection and critical wavelengths that classify them as broad-spectrum filters. These differences highlight the potential of using combinations of the isolated compounds in sunscreen formulations to provide complete UV photoprotection.

Fig. 6
figure 6

Results of the in vitro photoprotection activity assays for benzophenone-3 (BP-3), pheophytin a, fucoxanthin, 2,5,7-trihydroxy-2-pentadecylchroman-4-one (compound 1), and pheophorbide a identified in C. cervicornis (CCe-FO) and S. zonale (SS02-FO) organic extracts. (a) SPF (Sun Protection Factor); (b) UVAr (UVA ratio); (c) λc (critical wavelength). The error bars represent the standard deviation (For SPF, n = 3, mean of three replicates at each concentration. For UVAr, n = 3, mean of three replicates at different concentrations as no significant differences were observed between extracts at the evaluated concentrations).

Full size image

UV-Vis spectra of the VIP loadings identified in PCA and PLS analysis in Figure 3 and 4, were compared with those of the identified compounds, and it was observed that fucoxanthin corresponded to variable 51. The Relative Concentration Matrix of the extracts (Supplementary Figure S.20) revealed that even though this variable was detected in the organic extracts of all the species, the C. cervicornis (CCe) and S. polyceratium (SP) extracts presented relatively high relative concentrations of this compound. This explains the extract differentiation and the higher UVAr values found in these FO extracts. Similarly, the UV-Vis spectra of variables 127 and 159 correspond to chlorophyll a derivatives such as pheophytin a and pheophorbide a which were identified in extracts with the highest UVAr values. These variables were detected in most of the samples, with high concentrations in DP02, PG, and DM, and CCe, LV, and SP, respectively. For the S. zonale extracts, variable 81 corresponds to compound 1, and variable 33 corresponds to stypoldione, with absorption maxima in the UVC and UVB regions. These variables have a limited distribution among samples, with S. zonale (SS01, SSS02, and SS03) having the highest concentrations, accounting for this extract differentiation and the highest SPF values found for these FO extracts.

In brief, a photoprotective activity screening methodology was implemented for the evaluation of algae extracts. This approach integrates in vitro photoprotection evaluation, antioxidant TLC activity assays, and HPLC-DAD metabolic profiling. Algal samples with promising photoprotective activity and distinct metabolic profiles were selected, and photoprotective activity was monitored until chemical characterization of the active compounds. This work represents one of the first reported methodologies aiming to explore algae diversity as a source of photoprotective compounds, using widely used and affordable methodologies in natural product research laboratories. This methodology could be further implemented for continuous screening, tracking, and comparison of photoprotective activity in different algal species.

Methods

General experimental procedures

HPLC-grade methanol (MeOH) (99.9%), HPLC-grade 2-propanol (IPA) (99.9%), hexane (95%), silica gel 60 (0.040–0.060 mm), and silica gel 60 F254 TLC plates were purchased from Merck Millipore, USA. Butanol (BuOH) (99.5%), dichloromethane (DCM) (99.8%), ethyl acetate (EtOAc) (99.8%), and HPLC grade acetonitrile (ACN) (99.9%) were purchased from Panreac AppliChem, USA. Ethanol (EtOH) (99.8%), 2-hydroxy-4-methoxybenzophenone (BP-3) (98%), 2,2-diphenyl-1-picrylhydrazine (DPPH), phloroglucinol, and acetic acid (CH3COOH) were purchased from Sigma Aldrich, USA. Oleic acid (99%) was from Alfa Aesar, USA. Ultrapure Type I Milli-Q water (membrane 2.2 μm) was used.

UV–Vis spectra of the crude extracts were recorded on a BertholdTech TriStar2S microplate reader (Berthold Technologies, Germany) in 96-well flat-bottom polystyrene plates (NEST® Biotechnology, China). UV–Vis spectra of the fractions were recorded on a Thermo Scientific Multiskan GO Microplate Spectrophotometer (Thermo Fisher Scientific, USA). UV–Vis spectra of the pure compounds or enriched fractions were recorded in a Varioskan LUX Multimode microplate reader (Thermo Fisher Scientific, USA) using 96-well UV flat-bottom clear polystyrene plates (Corning®, USA). HPLC–DAD profiling was performed with a Thermo Fisher Scientific Dionex UltiMate 3000 UHPLC equipped with DAD and Sedere 85L ELSD (Sedere, France) detectors, using an X-Terra Waters RP-18 (5 μm, 4.6 × 250 mm) column (Waters Corporation, Massachusetts, USA). Preparative HPLC separations were performed with a Merck Hitachi HPLC–DAD equipment (Pump L-6000A, DAD Detector L-4500, UV–Vis Detector L-4250) using a Kromasil 100–5-C18 (10 × 250 mm) column (Nouryon, Sweden). NMR analysis was performed in a 400 MHz Bruker Advance Neo device, using deuterated chloroform (CDCl3) (99.5%) and deuterated methanol (MeOD) (99.8%) from Merck Millipore, USA.

Collection and preparation of algae material

Seventeen brown algae samples were collected at Providencia Island, Colombia, in September 2021. The samples were kept at -20 °C and were identified by Dr. Natalia Rincón Díaz, Dr. Mónica Puyana, Dr. Vanessa Urrea-Victoria, and M.Sc. Felipe de la Roche. Detailed information on the samples studied can be found in Supplementary Table S.1. A voucher for each sample was deposited at the JIWUKORI Algae Herbarium at the Biology Department, Sciences Faculty, Universidad Nacional de Colombia.

For each sample, three biological replicates were selected. Prior to extraction, the samples were washed with artificial seawater to eliminate debris and epiphytes and subsequently freeze-dried for 24–48 h until dryness.

Sample extraction

Crude extracts were obtained following the procedure reported by Sepúlveda, 202362. Briefly, 1 g of dry material was extracted with 20 mL of MeOH/water (1:1), obtaining the polar crude extract (A) and the residual algal material. The (A) extract was dried under vacuum, resuspended in 15 mL of water, and partitioned with 15 mL of BuOH, yielding butanolic (FB) and aqueous (W) extracts. Each (W) extract was dried and extracted with 10 mL of cold MeOH, yielding methanol (FM) and aqueous (FW) extracts. The remaining algae material was further extracted with 20 mL of DCM/MeOH (1:1) solution, yielding the organic (FO) extract. Each extraction was conducted in triplicate with ultrasound assistance for 30 min. The supernatants were recovered by filtration, dried under vacuum, and stored in darkness at − 20 °C.

In vitro photoprotection activity assays

As an international standard, photoprotection-related assays must evaluate photoprotection in the UVA and UVB regions. In vitro assays are approaches based on absorbance measurements that can be used for screening in early-stage research prior to in vivo assays, which can be used for final product testing.

2 mg/mL ethanolic solutions of each algae extract were prepared, followed by ethanolic dilutions to obtain 200, 100, 50, and 10 ppm solutions. BP-3, at the same concentrations, was used as a positive control, and ethanol (EtOH) was included as a blank. For FW extracts, solutions and blanks were prepared in deionized water.

Absorbance was recorded for each concentration in a TriStar2S microplate reader (Berthold Technologies) within the 290–400 nm range at 1 nm intervals. These values were used to compute the Sun Protection Factor (SPF), UVA ratio (UVAr), and critical wavelength (λc) as follows:

Sun protection factor

The Sun Protection Factor (SPF) measures UVB photoprotection, representing the minimum quantity of radiation required to produce sunburns in sunscreen-protected vs. sunscreen-unprotected skin63. It was calculated for each solution using Mansur’s in vitro methodology64, with some modifications as proposed by Nunes and coworkers65. SPF was computed using Mansur’s Eq. (1), using absorbances between 290 to 320 nm in 5 nm increments:

$${\text{SPF}} = {\text{FC}} \times \mathop \sum \limits_{{{290}}}^{{{320}}} {\text{EE}}\left(\uplambda \right) \times {\text{I}}\left(\uplambda \right) \times {\text{Abs}}\left(\uplambda \right)$$
(1)

where FC stands for correction factor = 10. EE (λ) corresponds to the erythemal effect spectrum, and I (λ) stands for solar intensity spectrum, in which the EE X I product are constants (Supplementary Table S.4)66. Abs (λ) corresponds to the recorded absorbance at the evaluated wavelength.

UVA ratio

The UVA ratio (UVAr), measures the balance between total UVA and UVB absorbances, providing a quantitative evaluation of UVA-photoprotection and assessing the overall balance between UVA and UVB photoprotection. It was determined according to the method proposed by Rojas and coworkers’67, using the recorded absorbances between 290 and 400 nm in 1 nm intervals and Eq. (2):

$${\text{UVAr}} = \frac{{\mathop \smallint \nolimits_{{{320}}}^{{{400}}} {\text{Abs}}\left(\uplambda \right){\text{d}}\uplambda }}{{\mathop \smallint \nolimits_{{{290}}}^{{{320}}} {\text{Abs}}\left(\uplambda \right){\text{d}}\uplambda }}$$
(2)

where Abs (λ) corresponds to the recorded absorbance within the studied UVB (290–320 nm) and UVA ranges (320–400 nm).

Critical wavelength

The critical wavelength (λc) records the balance between UVA and UVB photoprotection and measures the overall photoprotection capacity within this spectral range. This corresponds to the wavelength at which 90% under the total absorbance curve is reached. Mathematically, it is determined using the Eq. (3):

$$\mathop \smallint \limits_{290}^{\lambda c} {\text{Abs}}\left(\uplambda \right){\text{d}}\uplambda = {0}{\text{.9}} \times \mathop \smallint \limits_{290}^{400} {\text{Abs}}\left(\uplambda \right){\text{d}}\uplambda$$
(3)

where Abs (λ) represents the recorded absorbance between 290 and 400 nm in 1 nm increments. Following Rojas and coworkers’ procedure67, the total area under the absorbance curve is designated 100%, and the critical wavelength is determined by interpolation as the wavelength at which 90% of the curve is attained.

Antioxidant bioautography DPPH assay

10 µL of 2 mg/mL extract solutions were spotted on TLC plates. The mobile phase for the (FO) and (FB) extracts consisted of hexane/ethyl acetate (7:3), whereas for the (FM) extracts a butanol/water/acetic acid (3:1:1) mixture was used. Phloroglucinol and oleic acid were used as positive controls, and amino acid standards served as negative controls. After developing the TLC plate and according to Plazas and coworkers’ procedure68, the plate was sprayed with a 0.2% DPPH solution and kept in darkness for 15–30 min. The active bands appeared as bleached spots against a purple background.

Metabolic HPLC–DAD profiling

A UHPLC Thermo Fisher Scientific Dionex UltiMate 3000 equipped with DAD (3000RS) and ELSD (Sedex LT-ELSD) detectors was used for HPLC-DAD metabolic profiling, using an X-Terra Waters RP-18 (5 μm, 4.6 × 250 mm) column (Waters Corporation, Massachusetts, USA). For chromatographic elution, water (A) and MeOH + 10%isopropanol (B) were used, and a gradient solvent system was used as follows: 0 min (70% B), 2 min (80% B), 17 min (100% B), 32 min (100% B), 40 min (80% B), 42 min (70% B) at a 0.6 mL/min flow rate. ELSD and DAD (λ = 200–800 nm) detectors were used to monitor chromatographic separation and obtain the UV–Vis spectrum for each peak.

Chromatogram analysis was performed using Tauler and coworkers’69 and Peng and coworkers’39 procedures for HPLC-DAD data deconvolution using multivariate curve resolution—alternating least squares (MCR-ALS) methodology in the MATLAB © interface69,70, using the MCR-ALS toolbox71. The detailed step-by-step procedure is described in the Supplementary information and Supplementary Figure S.21. Briefly, baseline correction for each selected chromatogram was performed using the Asymmetric Least Squares (AsLs) algorithm, resulting in a 54-chromatogram dataset. Chromatograms were fragmented into 28 retention time-variable segments (f1 to f28) by visual inspection, keeping the same fragment sizes in the entire dataset. Data for each (fn) segment in each chromatogram, were columnwise augmented to obtain Df matrices (D1 to D28) with elution times (tr) as rows and wavelengths (λ) as columns. The number of components (N) in each segment (fn) was determined using the SVD (Singular Value Decomposition) algorithm, varying the number of accepted components and evaluating the reliability of the results considering R2 (explained variance) and %LOF (% lack of fit) values.

The starting set of spectral profiles (STn) was predicted by the purest variable detection method, allowing 10% noise39. Non-negativity constraints were applied in rows and columns using the fnnls (fast non-negative least squares) algorithm. Unimodality constraints were applied in the row direction and correspondence constraints were used for all the experiments72. The results were normalized using the Euclidean norm39,72,73 and rotation ambiguities within the applied constraints were evaluated by applying the MCR-BANDS algorithm by comparing the maximum (fnmax) and minimum (fnmin) values of the relative contribution function (fn) for each component40. Each obtained Cn matrix was integrated to obtain the relative concentration matrix (RCM: Chromatogram X area) and further analyzed in SIMCA 17® by principal component analysis (PCA), hierarchical cluster analysis (HCA), and partial least square data analysis (PLS). Information on UV–Vis absorbance regions of each component from the ST matrices was used to color the variables in the PCA and PLS loadings plot. The evaluated photoprotection parameters and the results of the antioxidant activity assay were used as outputs (Y-matrix) for PLS analysis.

Statistical analysis

The systematic analysis of the data is summarized in the scheme shown in Supplementary Figure S.22. A heatmap was constructed to graphically analyze the results of the in vitro photoprotection activity and antioxidant assays for each extract. It was constructed in RStudio®74 from a 68 X 4 matrix, allowing data centering and scaling in the column direction (SPF (50 ppm), UVAr, λc, and AOX) and clustering in the row direction (samples).

The metabolic profile and its correlation with photoprotection activity assays were analyzed by multivariate data analysis tools in Simca 17®75. The models were validated using R2, Q2, CV-ANOVA tests (p < 0.05) and permutation test (999 permutations, R2 < 0.4, and Q2 < 0).

Chemical characterization

The most promising extracts, FO extracts from Canistrocarpus cervicornis (CCe) and Stypopodium zonale (SS02), were separately fractionated using silica gel flash chromatography. A discontinuous solvent gradient was employed using hexane, ethyl acetate, and methanol as the mobile phases. The solvent gradient included hexane (100%), hexane/ethyl acetate (90:10; 80:20; 70:30; 50:50; and 20:80), ethyl acetate (100%), ethyl acetate/methanol (80:20; 50:50), and methanol (100%). Fractions were collected, vacuum-dried, and stored at -20 °C in darkness. 36 fractions were collected from CCe and regrouped into 10 fractions (D01–D10) according to TLC analysis. 31 fractions were collected from the SS02 extract and subsequently regrouped into 17 fractions (S01 to S17) upon TLC analysis. In vitro photoprotection assays were evaluated for the obtained fractions as detailed above. Measurements were conducted in triplicate for each fraction.

Fractions D09 from (CCe) and S07 from (SS02) were further purified by preparative HPLC chromatography, using a Merck Hitachi HPLC system (L-6000A pump; L-4500 DAD detector) equipped with a Kromasil 100–5-C18 (10 × 250 mm) column. For D09, the mobile phase consisted of water (A) and acetonitrile (B) in gradient mode as follows: 0 min (10% B), 3 min (10% B), 10 min (70% B), 25 min (70% B), 30 min (100% B), 45 min (100% B), 50 min (10%B) and 60 min (10%B), at a flow rate of 2 mL/min. 32 sub-fractions (D0901-D0932) were collected and further analyzed by NMR.

For S07, separation was performed in isocratic mode with 100% acetonitrile at a flow rate of 2.5 mL/min. 6 subfractions (D0701–D0706) were collected and further analyzed by NMR. UPLC-DAD-MS/MS analysis of subfraction D0705 was performed on a Bruker Elute UHPLC system (DAD Elute, Bruker Impact II Q-TOF) equipped with a Kromasil 100–2.5-C8 (4.6X100 mm) column. The mobile phase consisted of water with 0.1% formic acid (A) and acetonitrile with 0.1% formic acid (B). Elution was performed in isocratic mode at 95% B. HRESIMS analysis was performed in positive ion mode (m/z 100–2000 Da, Capillary voltage: 4500 V, ion source temperature: 200 °C, dry gas flow: 9 L/min, nebulizer gas pressure: 2 bar76).

Extracts and fractions were analyzed by 1H-NMR on a Bruker Advance Neo 400 MHz instrument, with CDCl3 or MeOD as solvents. Compound identification was achieved through 1H and 13C-NMR analyses as well as two-dimensional COSY, HMBC, and HSQC experiments and comparisons with literature reports.

2,5,7-trihydroxy-2-pentadecylchroman-4-one (1) Off-white solid. \({\left[\alpha \right]}_{D}^{26.9}=\) 3.43 (c = 0.0005, C3H6O). UV (ACN) λmax = 207, 227, 287, 320 nm. 1H-NMR (MeOD, 400 MHz) δ11.94 (s, OH-5), 6.00 (s, H-6), 5.82 (s, H-8), 2.91 (d, J = 16.9 Hz, H-3), 2.60 (d, J = 16.9 Hz, H-3), 1.85 (m, H-1′), 1.47 (m, H-2′), 1.27 (m), 0.86 (t, J = 6.8 Hz, H-15′). 13C-NMR (MeOD, 100 MHz): 197.16 (C-4), 168.17 (C-7), 165.00 (C-5), 162.17 (C-10), 103.62 (C-2), 103.09 (C-9), 96.88 (C-8), 96.68 (C-6), 46.00 (C-3), 41.98 (C-1′), 33.10 (C-13′), 30.83 (C-12′), 30.81–30.50, 24.73 (C-2′), 23.76 (C-14′), 14.46 (C-15′). Positive ion HRESIMS m/z 407.2789 (calcd for C24H38O5 [M + H]+, 407.2792), m/z 389.2674 (C24H37O4+, Δ3.34 ppm), HRESIMS/MS of 407.2789: 153.0185 (C7H5O4+, Δ1.31 ppm), 127.0395 (C6H7O3+, Δ3.94 ppm).