Flower Extracts from Ornamental Plants as Sources of Sunscreen Ingredients: Determination by In Vitro Methods of Photoprotective Efficacy, Antigenotoxicity and Safety

Plants are sources of sunscreen ingredients that prevent cellular mutations involved in skin cancer and aging. This study investigated the sunscreen properties of the extracts from some ornamental plants growing in Colombia. The UV filter capability of the flower extracts obtained from Rosa centifolia L., Posoqueria latifolia (Rudge) Schult, and Ipomoea horsfalliae Hook. was examined. Photoprotection efficacies were evaluated using in vitro indices such as sun protection factor and critical wavelength. UVB antigenotoxicity estimates measured with the SOS Chromotest were also obtained. Extract cytotoxicity and genotoxicity were studied in human fibroblasts using the trypan blue exclusion and Comet assays, respectively. Major compounds of the promising flower extracts were identified by UHPLC–ESI+–Orbitrap–MS. The studied extracts showed high photoprotection efficacy and antigenotoxicity against UVB radiation, but only the P. latifolia extract showed broad-spectrum photoprotection at non-cytotoxic concentrations. The P. latifolia extract appeared to be safer for human fibroblast cells and the R. centifolia extract was shown to be moderately cytotoxic and genotoxic at the highest assayed concentrations. The I. horsfalliae extract was unequivocally cytotoxic and genotoxic. The major constituents of the promising extracts were as follows: chlorogenic acid, ecdysterone 20E, rhamnetin-rutinoside, cis-resveratrol-diglucoside, trans-resveratrol-diglucoside in P. latifolia; quercetin, quercetin-glucoside, quercetin-3-rhamnoside, kaempferol, kaempferol-3-glucoside, and kaempferol-rhamnoside in R. centifolia. The potential of the ornamental plants as sources of sunscreen ingredients was discussed.


Introduction
Photoprotection is a preventive strategy to defend human skin against cancer and photoaging [1]. The sunlight UV rays that reach the Earth's surface, such as ultraviolet A (320-400 nm) and ultraviolet B (280-320 nm), cause DNA damage (e.g., cyclobutane pyrimidine dimers), which initiates inflammatory processes and skin cancer [2,3]. The use of sunscreens is among the most popular strategies in photoprotection [4]. Sunscreens contain compounds that act like filters, either absorbing or refracting the UV radiation [5]. Relevant properties of sunscreens used commercially have been reviewed [6,7].

In Vitro Photoprotection Efficacy of the Flower Extracts
The UV absorbance spectrum of each flower extract is shown in Figure 1; high levels of absorbance across the UV spectrum were observed for all extracts. The extracts showed the highest absorbance peaks for λ between 225 and 320 nm. All the absorbance peaks depended on the extract concentrations. It has been indicated that active ingredients of commercial sunscreens can be toxic to humans and coral reefs [8][9][10][11], which has increased the interest in using phytochemicals in sunscreen formulations [12][13][14][15][16][17]. Phytochemicals are genoprotective and anticancer agents [18,19] that are considered to be non-toxic and pharmacologically safe for human consumption [13]. These phytochemicals can prevent cellular mutations involved in skin cancer and aging by regulating UV-induced mutability [20,21].
Among the wide plant diversity, ornamental species have been widely used as raw materials for the cosmetic industry due to both their fragrance and therapeutic properties [22]. Colombia is the second highest exporter of ornamental flowers worldwide [23]; however, until today, their use has been exclusively for decorative purposes. In the present study, we took advantage of the cosmetic and therapeutic properties of ornamental plants to find phytochemicals that can be used as sunscreen ingredients.
The work aims were as follows: (i) to evaluate the UV absorption capability of flower extracts obtained from some ornamental plants growing in Colombia by means of in vitro protection efficacies (SPFin vitro, and λc), (ii) to study the correlation between in vitro UVB protection efficacy (SPFin vitro) and complementary SOS Chromotest-based DNA protection using data from flower extracts, (iii) to evaluate in human fibroblast the cytotoxicity and genotoxicity of promising extracts, and (iv) to study the chemical composition of promising extracts as sources of sunscreens. We showed that the flower extracts isolated from ornamental plants, R. centifolia and P. latifolia, were rich in photoprotective compounds. P. latifolia extract, which is photostable and relatively safe, seems to be a good candidate for a sunscreen active ingredient.

In Vitro Photoprotection Efficacy of the Flower Extracts
The UV absorbance spectrum of each flower extract is shown in Figure 1; high levels of absorbance across the UV spectrum were observed for all extracts. The extracts showed the highest absorbance peaks for λ between 225 and 320 nm. All the absorbance peaks depended on the extract concentrations.  The SPF in vitro values calculated for each flower extract are presented in Table 1. According to the European Commission categories (see Section 4), the four flower extracts showed UVB photoprotection efficacy (SPF in vitro ≥ 6.0); the SPF in vitro values increased with concentration (R = 0.97, p < 0.001, Figure 2A). Except for the I. horsfalliae extract, other extracts also showed broad-spectrum (UVA-UVB) protection efficacy (λc ≥ 370 nm) at a high extract concentration (750 µg/mL). Table 1. List of the studied plant species. For each plant extract, the following indices are shown: (i) UVB protection efficacy (SPF in vitro ); (ii) critical wavelength (λc); (iii) genotoxicity inhibition percentage (%GI) obtained using the SOS Chromotest; and (iv) percentages of human fibroblast cell viability (%CV). At photoprotective and non-cytotoxic extract concentrations, the percentage of effectiveness (E ff ) values was also estimated for the minimum erythema dose (MDE) according to the Fitzpatrick skin scale.

Relations between SPFin vitro and %GI Estimates in Flower Extracts
None of the flower extracts had increased I values at concentrations assayed in the SOS Chromotest (data not shown); therefore, they were considered not genotoxic in the model E. coli PQ37 cells. All photoprotective extracts also showed antigenotoxic against UVB radiation ( Table 1). The SPFin vitro and %GI values in flower extracts were highly correlated (R = 0.82, p ˂ 0.001, Figure 2B). That is, the greater the UVB photoprotective efficiency, the lower the genetic damage.

Relations between SPF in vitro and %GI Estimates in Flower Extracts
None of the flower extracts had increased I values at concentrations assayed in the SOS Chromotest (data not shown); therefore, they were considered not genotoxic in the model E. coli PQ37 cells. All photoprotective extracts also showed antigenotoxic against UVB radiation ( Table 1). The SPF in vitro and %GI values in flower extracts were highly correlated (R = 0.82, p < 0.001, Figure 2B). That is, the greater the UVB photoprotective efficiency, the lower the genetic damage.

Extract Cytotoxicity in Human Fibroblast (MRC5) Cells
Cytotoxicity dose-response relations were studied for each flower extract at a concentration range between 62 and 750 µg/mL (Table 1). Next, lethal concentrations, 50% (LC 50 ) and 30% (LC 30 ), in human fibroblast (MRC5) cells were obtained by interpolation. Based on the LC 50 values, extract cytotoxicity was as follows: I. horsfalliae (398 µg/mL) > R. centifolia pink, commercial variety (492 µg/mL) > R. centifolia fuchsia, commercial variety (702 µg/mL). The extracts were safe for fibroblast cells at concentrations lower than LC 30 values as follows: I. horsfalliae (250 µg/mL); R. centifolia pink, commercial variety (363 µg/mL); and R. centifolia fuchsia, commercial variety (492 µg/mL). The P. latifolia extract was unique in that it was shown to be non-cytotoxic to fibroblast cells at the concentration range studied. All the extracts were relatively less cytotoxic than the commercial sunscreen (Eau Thermale Avène SPF 50+) and the sunscreen active ingredient (titanium dioxide) used for comparison.

Extract Genotoxicity in Human Fibroblast (MRC5) Cells
Genotoxicity dose-response relations were studied for each flower extract ( Table 2). According to genotoxicity criteria, the extracts produced some degree of DNA damage at the following concentrations: I. horsfalliae (187.5 µg/mL) > R. centifolia (375.0 µg/mL) > P. latifolia cated that the solvent (methanol) was not genotoxic in human fibroblasts for the concentration range tested. Therefore, except for P. latifolia, the extracts showed low-to-moderate genotoxicity across the concentration range studied. Such genotoxicity showed a clear dose-response relationship (extract concentration-DNA damage), and demonstrates the importance of establishing the safe extract concentrations for potential use as sunscreen ingredients. These data also suggest a cytotoxic mode of action depending on the genotoxicity for all extracts. Table 2. Genotoxicity of the flower extracts in MRC5 human fibroblasts cells. The genetic damage index (GDI) values and their corresponding standard errors, calculated from at least three independent experiments, are given. Pearson correlation coefficients (R) showing the relationship between GDI extract concentrations are also presented.

GDI (
was unique in that it was shown to be non-cytotoxic to fibroblast cells at the concentration range studied. All the extracts were relatively less cytotoxic than the commercial sunscreen (Eau Thermale Avène SPF 50+) and the sunscreen active ingredient (titanium dioxide) used for comparison.

Extract Genotoxicity in Human Fibroblast (MRC5) Cells
Genotoxicity dose-response relations were studied for each flower extract ( Table 2). According to genotoxicity criteria, the extracts produced some degree of DNA damage at the following concentrations: I. horsfalliae (187.5 µ g/mL) ˃ R. centifolia (375.0 µ g/mL) ˃ P. latifolia (750 µ g/mL). Evaluation of the equivalent solvent concentrations (dilutions) in extracts indicated that the solvent (methanol) was not genotoxic in human fibroblasts for the concentration range tested. Therefore, except for P. latifolia, the extracts showed low-tomoderate genotoxicity across the concentration range studied. Such genotoxicity showed a clear dose-response relationship (extract concentration-DNA damage), and demonstrates the importance of establishing the safe extract concentrations for potential use as sunscreen ingredients. These data also suggest a cytotoxic mode of action depending on the genotoxicity for all extracts. The standard mutagen 4-nitro-quinoline-1-oxide (0.89 µ g/mL) was used as positive control (PC). † , The DNA damage criteria were as follows: (i) GDI values between 0 and 1 (no DNA damage); (ii) GDI values between 1 and 2 (little DNA damage); (iii) GDI values between 2 and 3 (moderate DNA damage); and (iv) GDI values between 3 and 4 (severe DNA damage). In addition, a clear dose-response relationship (concentration-DNA damage) must exist. ‡ , SEC: GDI values for solvent (methanol) equivalent concentrations (between 39 and 618 mM) in the extracts.

In Vitro Photoprotection Efficacy and Photostability at Safe Extract Concentrations
At photoprotective and safe extract concentrations, namely, at non-cytotoxic (≤LC30) and non-genotoxic concentrations, only the P. latifolia extract showed high UVB and broad-spectrum (UVA-UVB) photoprotection efficacy values ( Table 1). The other extracts demonstrated a reduction in their SPFin vitro values, which indicates their low photoprotection efficacy (6.0 ≤ SPFin vitro ≤ 15.0; λc ˂ 370 nm) values. Conversely, extract photostability or effectiveness (Eff) when they were irradiated at Fitzpatrick's MDE was consistently high at photoprotective and safe extract concentrations. According to these results, the P. latifolia and R. centifolia extracts at non-cytotoxic concentrations could be good candidates for use as sunscreen active ingredients. Conversely, the I. horsfalliae extract at non-cytotoxic

In Vitro Photoprotection Efficacy and Photostability at Safe Extract Concentrations
At photoprotective and safe extract concentrations, namely, at non-cytotoxic (≤LC 30 ) and non-genotoxic concentrations, only the P. latifolia extract showed high UVB and broadspectrum (UVA-UVB) photoprotection efficacy values ( Table 1). The other extracts demonstrated a reduction in their SPF in vitro values, which indicates their low photoprotection efficacy (6.0 ≤ SPF in vitro ≤ 15.0; λc < 370 nm) values. Conversely, extract photostability or effectiveness (E ff ) when they were irradiated at Fitzpatrick's MDE was consistently high at photoprotective and safe extract concentrations. According to these results, the P. latifolia and R. centifolia extracts at non-cytotoxic concentrations could be good candidates for use as sunscreen active ingredients. Conversely, the I. horsfalliae extract at non-cytotoxic concentrations was shown to be genotoxic (Table 2) and poorly photoprotective (Table 1); therefore, this extract was excluded from further analyses.

Discussion
The present work evidenced that the flower extracts isolated from several ornamental plants cultivated in Colombia contained compounds that could be used as sunscreen ingredients. Among the eleven plant species studied in our project (data not shown), three plants (R. centifolia, P. latifolia, and I. horsfalliae) in particular showed good photoprotective properties. This finding provides new evidence on plant extract applicability as a source of solar filters (Table 4).

Discussion
The present work evidenced that the flower extracts isolated from several ornamental plants cultivated in Colombia contained compounds that could be used as sunscreen ingredients. Among the eleven plant species studied in our project (data not shown), three plants (R. centifolia, P. latifolia, and I. horsfalliae) in particular showed good photoprotective properties. This finding provides new evidence on plant extract applicability as a source of solar filters (Table 4).
We also showed that the photoprotection efficacy depended on extract concentration. The photoprotective extracts were also antigenotoxic against UVB. This indicates that they acted as filters that absorbed or refracted the UV rays and reduced genotoxicity. As in the previous studies [17,[31][32][33], our data support the use of DNA damage detection assay (in this case, the SOS Chromotest) as an effective complement that improves the efficacy of photoprotection measurement. As we indicated in the Introduction, some organic ingredients of commercial sunscreens can be toxic to humans and coral reefs [8][9][10], with inorganic filters (i.e., titanium dioxide or zinc oxide) being a safer alternative [11]. At photoprotective concentrations, the extracts were relatively less cytotoxic than a commercial sunscreen (Eau Thermale Avène SPF 50+), and its active ingredient (titanium dioxide) was used for comparison. In this sense, the plants studied here could be new and safer sources of ingredients for the development of commercial sunscreens.
The P. latifolia extract was the most promising for use as a sunscreen, because it showed high photoprotective efficacy, antigenotoxicity, photostability, and relatively low cytotoxicity and genotoxicity in human fibroblasts. P. latifolia, commonly known as jasmine tree-mountain lily (azuceno)-monkey apple, is a native plant from Central and South America [46]. This tree possesses flowers with a very intense scent, whose main fragrance compound is (Z)-3-hexenyl acetate [47]. We found that P. latifolia ethanolic extract was rich in cisand trans-resveratrol diglucosides, two stilbenoid isomers, which were previously identified in Vitis vinifera [48] and Glycine max [49]. Chlorogenic acid (35 ± 1 mg/g), ecdysterone (64 ± 8 mg/g), and rhamnetin-rutinoside (17 ± 1 mg/g) were present as well at relevant concentrations. Chlorogenic acid, resveratrol, and its derivatives have been reported as compounds with solar filter activity [50][51][52][53]. Resveratrol and its derivatives have also shown antigenotoxicity against UV radiation and skin cancer chemopreventive properties [54][55][56]. We hypothesized that some of these compounds, or their combinations, were responsible for the sunscreen/antigenotoxic properties of the extracts studied here. Among the plant species studied, the P. latifolia extract appeared to be the best candidate and the safest for use as a sunscreen active ingredient.
The flower extracts from R. centifolia had high UVB and broad-spectrum protection efficacies as well. However, these extracts showed between low and moderate cytotoxicity and genotoxicity in human fibroblasts measured at the highest concentrations compared with those Rosa species extracts previously studied [57]. This supported the importance of establishing safe extract concentrations for use of the extracts as sunscreen ingredients, as has been previously suggested [22]. Thus, these extracts should be used with caution until more details on their genotoxicity are known. The R. centifolia extracts were rich in kaempferol and quercetin and their derivatives (quercetin-glucoside, quercetin-3rhamnoside, kaempferol-3-glucoside, kaempferol-rhamnoside), as has also been found for other Rosa species [26,[58][59][60][61]. For R. centifolia species, the antimutagenic activity has been previously reported against ethyl methanesulfonate [62]. Kaempferol and quercetin showed solar filter activity [52,53,63]. These main compounds were possibly responsible for the sunscreen/antigenotoxic properties of the R. centifolia extracts studied in this work.
Conversely, the extract obtained from I. horsfalliae, a species with previously reported sunscreen properties [40], was shown to be cytotoxic and genotoxic to human fibroblast cells. Little is still known about the genotoxicity of the major compounds (chlorogenic acid, dicaffeoylquinic acid, and scopoletin) of this extract. Chlorogenic acid and scopoletin have shown clastogenic activity [64,65], which suggests that flower extract's genotoxic activity could be related to these compounds, although, a synergistic effect of these compounds could also be the cause of their genotoxicity. Confirmation of this hypothesis requires further studies.

Plant Material and Extracts
The flowers from Rosa centifolia (Rosaceae), pink and fuchsia commercial varieties, were supplied by Flexport-Colombia S.A.S. (Bogotá, Cundinamarca, Colombia). The flowers from Posoqueria latifolia (Rubiaceae) and Ipomoea horsfalliae (Convolvulaceae) species were collected from experimental plots at the Agroindustrial Pilot Complex of the National Center for Agroindustrialization of Aromatic and Medicinal Tropical Vegetal Species (CENIVAM). The P. latifolia e I. horsfalliae species were identified at the Colombian National Herbarium, where their vouchers (COL; voucher number in parentheses) were placed (Table 1).
For each specimen, undamaged and fully developed flowers were dried in an Advantage Plus Tray Lyophilizer (Virtis Co., Gardiner, ME, USA) and were used for solvent extraction as indicated by Sierra et al. [40]. In brief, dried flowers (1 g) were mixed with acidified ethanol solution (20 mL, 0.5% HCl, 1:1 v/v) and put for 5 min in an S15H ultrasound bath (Elmasonic, Singen, Germany). The mixture was filtered, and the residue was extracted twice more. Extracts were roto-evaporated and then were dried as indicated above. Extract stock solutions were prepared from the dried powder (30 mg), which were dissolved in methanol (1 mL), vortexed (3 min), exposed to ultrasound (10 min, 40 • C), and centrifuged (5000× g, 10 min). The supernatant (1 mL) was then filtered and was stored at −80 • C in a Thermo Scientific ® Series-86 DEG C ultra-low-temperature freezer (Thermo Scientific, Waltham, MA, USA). Before their use, the extract stock solutions were defrosted and refrigerated (5-8 • C) for 24 h.

UV Absorption Capability
Aqueous aliquots (1.5 mL) of each extract, prepared at different concentrations (between 62.5 and 750.0 µg/mL) using distilled water, were placed in a quartz cuvette (1 cm step length and 1.5 mm glass thickness), and their UV absorption spectra (λ = 200-400 nm) were recorded in triplicate using the Skanlt 3.2 function on a Multiskan GO UV spectrophotometer (Thermo Scientific, Waltham, MA, USA). Distilled water was used as a blank. A minimum of three independent experiments were carried out per extract dilution. The absorbance values were recorded for wavelength intervals of 10 nm. The average absorbance values and their corresponding standard errors were plotted using the program ggplot2 of the R platform [66].

In Vitro Photostability
Extract aliquots (1 mL) were distributed into Petri plates with a 5 cm diameter for their irradiation. The Petri plates were irradiated in darkness using a UVA/UVB irradiation chamber BS-02 (Dr. Grobel UV-Elektronik GmbH, Etlingen, Germany) equipped with a radiation controller, UV-MAT, from the same commercial house. This radiation controller continuously measured the irradiance, calculated the irradiation dose, and switched the lamps after reaching the target dose. Operating at 100% intensity, the UVB lamps in the irradiation chamber had an irradiance value of 4 mW/cm 2 . The UVB radiation doses applied were those corresponding to minimum erythema dose (MDE) in humans according to the Fitzpatrick skin scale [24]. These were as follows: type I (0.035 J/cm 2 = 350 J/m 2 ), type II (0.056 J/cm 2 = 560 J/m 2 ), type III (0.070 J/cm 2 = 700 J/m 2 ), and type IV (0.084 J/cm 2 = 840 J/m 2 ). The relative photostability of the extracts was expressed as the percentage of effectiveness (E ff ) of SPF in vitro after UV exposure and was calculated as follows: E ff = SPF in vitro after irradiation SPF in vitro before irradiation × 100 [14].

Antigenotoxicity against UVB Radiation Estimates Based on SOS Chromotest
Before extract antigenotoxic effects were assayed, their genotoxicities were investigated using the SOS Chromotest [72], as was described previously by Quintero et al. [73]. The antigenotoxicity assay was conducted using a co-incubation procedure as described by Fuentes et al. [74]. Briefly, the cells were simultaneously treated with extracts (between 62.5 and 750.0 µg/mL) and 10 J/m 2 of UVB radiation, which largely induced the SOS function Escherichia coli PQ37 cells [75]. After, they were cultured for 2 h at 37 • C and shaken at 300 rpm in a Thermomixer apparatus (Eppendorf, Sao Paulo, Brazil). Negative (distilled water) and positive (10 J/m 2 of UVB) controls were always included in each assay. A minimum of four independent experiments per treatment with two replicates were conducted. β-Galactosidase (βG) and alkaline phosphatase (AP) activities were assayed in 96-well plates (Brand-GMBH, Wertheim, Germany), as described by Fuentes et al. [74]. The antigenotoxicity (the ability of the plant extract to protect against UV-induced DNA damage) was measured as a significant reduction in the SOS induction factor (I) in E. coli PQ37 cells and was expressed as a percentage of the genotoxicity inhibition, as follows: %GI = 1 − (I ct − I nt )/(I UVB − I nt ) × 100, where I ct was the SOS induction factor in the co-incubation procedure; I nt was the SOS induction factor in non-treated cells, and I UVB was the SOS induction factor in UVB-treated cells. Negative values of %GI were considered as zero; therefore, this parameter ranged from 0% to 100%. The minimal concentration that produces a significant (p ≤ 0.05) genotoxicity inhibition (CGI) in PQ37 cells was used for comparison of the genoprotective potential of the tested samples.

Extract Cytotoxicity in Human Fibroblast (MRC5) Cells
Cytotoxicity of flower extracts in MRC-5 cells was evaluated using the trypan blue exclusion assay [76]. The lung embryo fibroblast (MRC5) cells [77] were grown on DMEM/F-12 medium (5 mL) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% of GIBCO Penicillin-Streptomycin, at 37 • C, and under CO 2 (5%) conditions in a Midi 40 incubator (Thermo Scientific, Marietta, OH, USA). Every three days, cells were grown in fresh medium to reach a confluence of 80%. Cell cultures were mixed with each extract at final concentrations between 62.5 and 750.0 µg/mL, and kept at 37 • C (24 h) under CO 2 (5%) atmosphere conditions. After 24 h, trypsin EDTA-treated cells were centrifuged (4000× g, 6 min), dissolved in PBS buffer (100 µL), and mixed (10 µL) with the same volume of trypan blue (0.4%) to assess cell viability. Live and dead cells were counted using a Neubauer chamber and Eclipse E200 optical microscope (Nikon Instruments Inc., NY, USA). At least three independent experiments were carried out for each treatment. The results are expressed as percentages of cell viability (%CV) per treatment, as follows: %CV = (Living cells)/(Total cells) × 100. Lethal concentrations 50% (LC 50 ) and 30% (LC 30 ) for each flower extract were calculated by interpolation using the graphic method [78]. LC 50 and LC 30 were considered as bordering cytotoxic and non-cytotoxic concentrations, respectively. That is, extracts were cytotoxic at values ≥ LC 50 and non-cytotoxic at values ≤ LC 30 .

Extract Genotoxicity in Human Fibroblast (MRC5) Cells
Genotoxicity of flower extracts in MRC-5 cells was evaluated using the Comet assay. For this purpose, a high-throughput Trevigen CometChip ® platform (Gaithersburg, MD, USA) was used as indicated by Sykora et al. [79], with some minor modifications. Firstly, a CometChip ® slide, cleaned previously with ethanol, was covered with an agarose solution prepared in PBS at 1.3% and tempered at 45 • C; then, the agarose was solidified for 24 h at 4 • C. Trypsin-treated cells (3 mL) were collected by centrifugation (2000× g, 6 min), were washed twice using NaCl solution (0.75%), centrifuged and suspended in fresh NaCl solution (3 mL), and then were quantified using a Neubauer counting chamber. A cell suspension (3 mL) prepared at 4.4 × 10 4 cell/mL was mixed with an equal volume of lowmelting-point agarose prepared in PBS at 1%, and the mix was poured on the CometChip ® slide. The agarose was solidified for 15 min at 4 • C. Finally, the CometChip ® chamber was ensembled by hermetically sealing to prevent mixing between the wells.
For cell treatments, 100 µL extract samples (prepared between 62.5 and 750.0 µg/mL) or standard mutagen used as a positive control (4-nitroquinoline 1-oxide prepared at 0.89 µg/mL) were loaded into wells of the CometChip ® chamber. A sample of DMEM medium (100 µL) was considered as the negative control. The chamber with treatments was incubated for 30 min at 37 • C under CO 2 (5%) atmosphere conditions. The solutions of each treatment were removed from the wells, and 30 µL of Bioultra Proteinase K (0.19 mg/mL) was loaded in each well for cell enzymatic lysis (1 h, 37 • C).
After that, the CometChip ® slide was removed from the CometChip ® chamber and submerged for 15 min at 4 • C in a Comet electrophoresis tank (Cleaver Scientific Ltd., Rugby, Warwickshire, UK), which contained alkaline buffer (0.3 N NaOH, 1 mM EDTA, pH 13). The electrophoresis was carried out for 30 min at 300 mA and 25 V. The CometChip ® slide was submerged for 15 min in a tray containing neutralizing solution (0.4 M TRIS, pH 7.5); then, it was washed with distilled water and dried at 37 • C in a Midi 40 incubator. Finally, cell nuclei contained in each microgel were stained with 7 µL of YOYO solution (1 mM prepared in 5% DMSO) and were immediately scored using a Zeiss Axio Observer 7 fluorescence microscope (GmbH, Oberkochen, Germany).
DNA damage was expressed in arbitrary units based on the classification of Comets into five categories (0-4), as was proposed by Collins et al. [80]. A genetic damage index (GDI) was calculated for each treatment, as follows: GDI = (N 0 × 0 + N 1 × 1 + N 2 × 2 + N 3 × 3 + N 4 × 4)/n, where Ni was the number of nuclei scored in each category and n was the number of scored cells per slide [81]. Two hundred cells per slide and two slides per treatment were analyzed. The results from at least three independent experiments were averaged to obtain the GDI values for each treatment.

UHPLC-ESI+-Orbitrap-MS Analysis
The flower extracts were analyzed by a UHPLC Dionex™ UltiMate™ 3000 (Thermo Fisher Scientific, Bremen, Germany) coupled to an Orbitrap™ mass detector (Exactive Plus, TFS, Bremen, Germany), using a heated-electrospray interface (HESI-II) operated in positive-ion acquisition mode (350 • C). The extract component separation was carried out on a Hypersil GOLD™ aQ column (TFS, Sunnyvale, CA, USA), of 100 mm × 2.1 mm id, ×1.9 µm particle size, at 30 • C. The mobile phase was as follows: A: water (0.2% formic acid) and B: acetonitrile (0.2% formic acid). Analysis started with 100% A and changed linearly up to 100% B in 8 min, remained for 4 min, and then returned to 100% A in 1 min; then, it remained in equilibrium for 3 min. The mobile phase flow was 0.3 mL/min, and the injection volume was 1 µL. Capillary voltage (3.5 kV, 320 • C) and higher-energy-collisional dissociation cell (HCD) were used in 10-40 eV range. Mass range in all experiments was set at m/z 80-1000. The data obtained were processed with the Thermo XCalibur™ Roadmap software, version 3.1.66.10. Compound identification was based on the extracted ion currents (EICs) of the protonated molecules, the exact masses of the ions, mass spectra fragmentation patterns, and by comparison of the experimental mass spectra with those of standards compounds.

Statistical Analysis
The SPF in vitro , λc, E ff , %GI, survival (%), and GDI values and their corresponding standard errors were calculated. In all cases, the data passed the Kolmogorov-Smirnov and F-maximum tests for normality and variance homogeneity, respectively; therefore, the parametric tests were used in subsequent data analyses. When a significant F-value was obtained in one-way analysis of variance (ANOVA), the groups were subsequently compared with Tukey's test. The Pearson correlation analysis was used to examine the relationship between extract and compounds concentrations, SPF in vitro , and %GI estimates. For all statistical analyses, a value of p < 0.05 indicated significance. The R program [66] was used for all analyses.

Conclusions
Our findings show that flower extracts from ornamental plants were rich in photoprotective compounds. Several studied flower extracts showed solar filter activity and photoprotection, which was attributed to their major compounds or to their combinations. P. latifolia extract, which is photostable and safe, appeared to be the best candidate for use as a sunscreen active ingredient. R. centifolia extracts showed between low and moderate cytotoxicity and genotoxicity in human fibroblasts at the highest concentrations assayed, while the I. horsfalliae extract was unequivocally cytotoxic and genotoxic. Therefore, R. centifolia and I. horsfalliae extracts should be used with caution until more details on their toxicity and genotoxicity are obtained. In addition, it is necessary to test these phytochemicals using in vivo mammalian assays for practical use of these extracts in photoprotection. Sunscreen based on phytochemicals will require cost-effective processes that combine plant tissue culture and enriched-fraction extraction techniques, to guarantee a stable supply of these raw materials.