Oxo-Carotenoids as Efficient Superoxide Radical Scavengers

Oxo-carotenoids containing conjugated carbonyl groups in their chains were designed to be more efficient superoxide radical scavengers than natural carotenoids, β-carotene and canthaxanthin. A practical chain-extension method for polyene dials (e.g., crocetin dial) was also proposed based on Horner–Wadsworth–Emmons olefination. Double aldol condensation between polyene dials and acetophenones with ring substituents produced oxo-carotenoids with substituted benzene rings. The antioxidant activity of oxo-carotenoids was measured using DPPH (radical) and ABTS (cationic radical) scavenging assays and compared with the analysis with the superoxide (anionic radical) probe. An effective conjugation length by carbon–carbon double bonds is important to provide superior antioxidant activity for oxo-carotenoids, regardless of the type of radical probe used in the assay. Increasing electron density is favorable to strong antioxidant activity in DPPH, and the phenol group is favored in ABTS, whereas electron deficient oxo-carotenoids are very potent in the superoxide radical assay. All oxo-carotenoids exhibited 105~151% better superoxide radical scavenging activity compared to beta-carotene (100%), whereas 38~155% in DPPH and 16~96% in ABTS radical scavenging activities were observed.


Introduction
Carotenoids are important secondary metabolites produced from microalgae to higher plants for energy production in photosynthesis and self-defense such as antibacterial and antioxidant activities [1]. The conjugated polyene chain plays vital roles in lightharvesting [2,3], energy-transference [4,5], and radical-quenching [6]. Provided as a dietary source of fruits and vegetables, human carotenoids benefit healthy living by reducing the risk of disease and aging by eliminating reactive oxygen species (ROS) that cause lipid peroxidation, DNA mutations, protein defects and other oxidative damages [7][8][9]. ROS includes hydroxy radical (HO•), hydroxide (HO − ), triplet oxygen ( 3 O 2 ), hydroperoxide (HOO − ), peroxide (O 2 − ), and superoxide radical (O 2 −· ), among which superoxide radical is a precursor to most other ROS [10][11][12]. It is the reactivity of the polyene chain towards ROS that exhibits the antioxidant activity of carotenoids [6].
Carotenoids readily undergo one-electron oxidation by photoionization or chemical oxidation to afford carotenoid cationic radical, which is known as a key intermediate in the radical scavenging process by electron transfer [13]. The maximum absorption wavelengths in benzene are 920 nm (astaxanthin), 940 nm (canthaxanthin), 1000 nm (zeaxanthin), 1020 nm (β-carotene), and 1050 nm (lycopene), which are red-shifted from those of the parent carotenoids [14]. The energy gap between HOMO and LUMO of the cationic radicals is the smallest for lycopene and the largest for astaxanthin. The reduction potentials of carotenoid cationic radicals are similar in the range of 1020 ± 40 mV, but the relative ease of electron-transfer to the carotenoids is in the order of astaxanthin > canthaxanthin > zeaxanthin > β-carotene > lycopene [14]. Oxo-carotenoids are not considered as good reducing (radical) agents compared with simple carotenoids. It is reported that carbonyl groups indeed increase the reduction potential of the carotenoids [15,16].
The efficacy of antioxidant activity of carotenoids is often measured by standard radical scavenging assays utilizing 2,2-diphenyl-1-picrylhydrazyl (DPPH) [17][18][19] or 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radicals [20]. DPPH radical readily absorbs H at the allylic positions including methyl groups of carotenoids to produce carotenoid neutral radical. ABTS cationic radical, on the other hand, takes a single electron from the polyene of carotenoids to yield the resonance-stabilized carotenoid cationic radical. Quantitative measurements of the reduction in UV absorption for the above standard radicals by carotenoids represent the antioxidant activities of the carotenoids. These assays work very well for simple carotenoids such as β-carotene and lycopene [21]. However, antioxidant activities of oxo-carotenoids such as canthaxanthin, halocynthiaxanthin, and capsanthin as shown in Figure 1 may be underestimated by these standard radical assays [22,23]. The electron-deficient carotenoids with conjugated carbonyl groups would destabilize the resulting carotenoid (cationic) radicals, and thus the loss of H or a single electron from the polyene would not be a main mechanism of action for antioxidant activity. On the other hand, anionic (or radical) ROS such as superoxide radical (O2 − ) or singlet oxygen would readily transfer a single electron or add to the electron deficient oxocarotenoids to produce the resonance-stabilized oxo-carotenoid (anionic) radicals [24][25][26]. It is thus necessary to use a (anionic) radical ROS scavenging assay for the fair judgement of the antioxidant activities of oxo-carotenoids. Pursuing the synthesis of strong carotenoid antioxidant, a series of novel ketonic carotenoids (e.g., 1a and 2a in Figure 1) were devised and synthesized by a series of chainextension and aldol condensation protocols. Antioxidant activities of the ketonic carotenoids using standard DPPH and ABTS radical assays were compared with those of superoxide radical scavenging assays (typically detecting formazan conversion from nitroblue tetrazolium) [27][28][29]. The biggest obstacle in the study of antioxidant activity of the ketonic carotenoids is the fact that their UV absorption wavelengths (496~551 nm) mostly coincide with those of probe molecules such as DPPH radical (517 nm) and formazan (562 nm). This problem can be avoided by separating the probe molecule from ketonic carotenes (or from any other interferences) by HPLC [30] or by choosing a different orthogonal probe system [31]. Herein, we report the details of our studies on the syntheses and anti-oxidant activities of novel oxo-carotenoids.

Materials and Methods
Reactions were performed in a well-dried flask under argon atmosphere unless noted otherwise. Solvents for extraction and chromatography were reagent grade and used as Pursuing the synthesis of strong carotenoid antioxidant, a series of novel ketonic carotenoids (e.g., 1a and 2a in Figure 1) were devised and synthesized by a series of chain-extension and aldol condensation protocols. Antioxidant activities of the ketonic carotenoids using standard DPPH and ABTS radical assays were compared with those of superoxide radical scavenging assays (typically detecting formazan conversion from nitroblue tetrazolium) [27][28][29]. The biggest obstacle in the study of antioxidant activity of the ketonic carotenoids is the fact that their UV absorption wavelengths (496~551 nm) mostly coincide with those of probe molecules such as DPPH radical (517 nm) and formazan (562 nm). This problem can be avoided by separating the probe molecule from ketonic carotenes (or from any other interferences) by HPLC [30] or by choosing a different orthogonal probe system [31]. Herein, we report the details of our studies on the syntheses and anti-oxidant activities of novel oxo-carotenoids.

Materials and Methods
Reactions were performed in a well-dried flask under argon atmosphere unless noted otherwise. Solvents for extraction and chromatography were reagent grade and used as received. Column chromatography was performed with silica gel 60 (70-230 mesh) using a Antioxidants 2022, 11, 1525 3 of 15 mixture of EtOAc/hexane as eluent. 1 H-and 13 C-NMR spectra were, respectively, recorded on a 400 MHz and 100 MHz NMR spectrometer in deuterated chloroform (CDCl 3 ) with tetramethylsilane (TMS) as an internal reference unless noted otherwise.
Scavenging Activity = {1 − (PA sample /PA blank )} ABTS (2,2'-azino-bis(3-ethylbenzohiazoline-6-sulfonic acid) cationic radical assay [20]: Each stock solution of 7.4 mM ABTS diazonium salt in H 2 O and 2.6 mM potassium persulfate in H 2 O was prepared. The working solution was prepared by mixing the two stock solutions in the ratio of 1:1 and performing a reaction at 25 • C for 12 h in the dark. The aliquot of 5 mL resulting solution was then diluted with 250 mL MeOH to obtain an absorbance of 0.7 at 734 nm. Fresh ABTS cationic radical solution (600 µL) was added to 2.5 M carotenoid solution in 50 µL of CH 2 Cl 2 /DMSO (1:2 v:v) and reacted at 25 • C for 2 h in the dark. The resulting mixture (0.3 mL) was placed in a microcuvette in a double beam UV-Vis spectrophotometer (Scinco NEOSYS-2000), in which the absorbance was measured at 734 nm. Particular care was taken to minimize the loss of free radical activity of the ABTS radical stock solution by storing at 4 • C until it was used. The absorbances (A) of ABTS cationic radical in the blank and in the carotenoid were calculated and the ABTS cationic radical scavenging activity was obtained using the following formulation.
Scavenging Activity = {1 − (A sample /A blank )} Superoxide radical Assay [27][28][29]: Superoxide radicals are generated in the PMS-NADH-O 2 (phenazine methosulfate-nicotinamide adenine dinucleotide-oxygen) system, which normally utilize the reduction of nitroblue tetrazolium (NBT) into formazan with 562 nm absorption for superoxide radical scavenging assay. Du to overlapping absorption of formazan with keto-carotenes; however, NBT could not be used in this experiment. Instead, t-butylhydroquinone (TBHQ) was utilized for the oxidation to the corresponding benzoquinone by superoxide radicals with 290 nm absorption [31]. The tubes bearing the reaction mixture {100 µL of 1.0 mM of TBHQ in EtOH/H 2 O (1:1) + 200 µL of 468 µM NADH in Phosphate-buffered saline solution (pH = 7.0) + 200 µL of 60 µM PMS + 100 µL of 2 mM ketonic carotene in CH 2 Cl 2 /DMSO (1:2)} were vortexed for 60 s, and then incubated at 25 • C for 1 h in the dark. The above resulting solution was filtered through 0.45 µm polyether sulfone membrane filter (HYUNDAI Micro Co., LTD., Seoul, Korea) and an aliquot (10 µL) of the sample was injected for HPLC analysis. The reversed-phase HPLC system (Waters Corporation) consisted of a binary pump, a system controller (Empowers 3 Software), an auto-injector, and a photodiode array detector with an Agilent Poroshell 120 EC-C18 column (4 µm, 4.6 × 150 mm, Carbon Load: 1%). Isocratic elution was performed with MeOH (0.7 mL/min) and with H 2 O (0.3 mL/min) at a total flow rate of 1 mL/min. The TBHQ peaks (~2.4 min retention time) were monitored at 290 nm in trifold analyses. The capability of O 2 •− scavenging activity was calculated using the following equation.
where A 1 and A x are the peak areas of the TBHQ probe in the absence and presence of superoxide anion radical scavenger (keto-carotenoid), respectively; and A 0 is the peak area of the TBHQ probe at initial concentration in the reaction mixture.

Syntheses of Oxo-Carotenoids
We designed ketonic carotene 2a as a main model substrate for oxo-carotenoids in superoxide radical scavenging assay, which would be prepared by aldol condensation between acetophenone and C 20 crocetin dial 2 (Scheme 1) [37]. Sliwka and we independently reported the chain-extension protocol for the synthesis of polyene dials 2 and 3 based on Wittig and Julia-Kocienski olefinations, respectively [32,36]. Even though high yields were reported for the syntheses of 2 and 3 by Sliwka, only milligram quantities were obtained after HPLC separation from the microwave-assisted Wittig reaction [36]. On the other hand, the chain-extension by double Julia-Kocienski olefination produced polyene dials 2 and 3 in sub-gram quantities, but mono-coupling was a major problem for the elongated polyene dial 3 [32]. The yield of C 30 polyene dial 3 was dropped significantly because of incomplete deprotection of acetal units. It was thus necessary to develop a practical synthetic method of polyene dials 2 and 3 in over-gram quantities for the preparation of various ketonic carotenoids by aldol condensation with acetophenones. with MeOH (0.7 mL/min) and with H2O (0.3 mL/min) at a total flow rate of 1 mL/min. The TBHQ peaks (~2.4 min retention time) were monitored at 290 nm in trifold analyses. The capability of O2 •− scavenging activity was calculated using the following equation.
where A1 and Ax are the peak areas of the TBHQ probe in the absence and presence of superoxide anion radical scavenger (keto-carotenoid), respectively; and A0 is the peak area of the TBHQ probe at initial concentration in the reaction mixture.

Syntheses of Oxo-Carotenoids
We designed ketonic carotene 2a as a main model substrate for oxo-carotenoids in superoxide radical scavenging assay, which would be prepared by aldol condensation between acetophenone and C20 crocetin dial 2 (Scheme 1) [37]. Sliwka and we independently reported the chain-extension protocol for the synthesis of polyene dials 2 and 3 based on Wittig and Julia-Kocienski olefinations, respectively [32,36]. Even though high yields were reported for the syntheses of 2 and 3 by Sliwka, only milligram quantities were obtained after HPLC separation from the microwave-assisted Wittig reaction [36]. On the other hand, the chain-extension by double Julia-Kocienski olefination produced polyene dials 2 and 3 in sub-gram quantities, but mono-coupling was a major problem for the elongated polyene dial 3 [32]. The yield of C30 polyene dial 3 was dropped significantly because of incomplete deprotection of acetal units. It was thus necessary to develop a practical synthetic method of polyene dials 2 and 3 in over-gram quantities for the preparation of various ketonic carotenoids by aldol condensation with acetophenones. We investigated the synthesis of polyene dials 2 and 3 based on Horner-Wadsworth-Emmons (HWE) olefination in the hope that more nucleophilic phosphonate (e.g., 5) carbanion would efficiently produce the chain-extended higher homologues and that the byproduct phosphenic acid would turn into phosphoric acid in aqueous acidic medium (1M HCl) to help with the acetal deprotection (Scheme 1). Acetal-protected allylic chloride 4, prepared from isoprene at an 86% overall yield using the published procedure [32], was converted into C5 phosphonate 5 at a 76% yield using the sequence of Finkelstein (NaI) and Arbuzov reactions (triethyl phosphite). On the other hand, DMSO-mediated We investigated the synthesis of polyene dials 2 and 3 based on Horner-Wadsworth-Emmons (HWE) olefination in the hope that more nucleophilic phosphonate (e.g., 5) carbanion would efficiently produce the chain-extended higher homologues and that the byproduct phosphenic acid would turn into phosphoric acid in aqueous acidic medium (1M HCl) to help with the acetal deprotection (Scheme 1). Acetal-protected allylic chloride 4, prepared from isoprene at an 86% overall yield using the published procedure [32], was converted into C 5 phosphonate 5 at a 76% yield using the sequence of Finkelstein (NaI) and Arbuzov reactions (triethyl phosphite). On the other hand, DMSO-mediated oxidation of acetal-protected allylic chloride 4 produced the corresponding C 5 aldehyde 6 in 44% yield [34]. The HWE reaction of C 5 phosphonate 5 with C 5 aldehyde 6 under NaH in THF at 25 • C, followed by the direct hydrolysis of the acetal protection in aqueous acidic (1M HCl) solution smoothly produced C 10 2,7-dimethyl-2,4,6-octatrienedial (1) in 83% yield.
We then tested a chain extension protocol utilizing C 5 phosphonate 5 for the practical synthesis of polyene dials 2 and 3. The HWE reaction of C 10 dial 1 (6.0 g, 1 equiv.) with C 5 phosphonate 5 (2.2 equiv.) under t-BuOK in THF/t-BuOH at ambient temperature, followed by 1M HCl addition, produced C 20 crocetin dial 2 in 92% yield (10.0 g) after purification by trituration from Et 2 O. Further extension to C 30 polyene dial 3 (590 mg, 1.38 mmol) was performed equally well with an 82% yield by conducting a HWE reaction of C 5 phosphonate 5 (1.50 g, 4.90 mmol) with C 20 crocetin dial 2 (500 mg, 1.69 mmol) even though a higher temperature of 120 • C was required for the double olefination. Once again, deprotection of acetal was effectively completed upon the addition of aqueous acidic solution presumably by the effect of the by-product, phosphoric acid.
With the series of polyene dials 1-3 in hand, the optimal conditions for aldol condensations with acetophenone, which seemed easy to achieve but were not, were carefully studied (Table 1). Initial screening of C 10 dial 1 with acetophenone under aqueous NaOH or NaH in THF conditions was disappointing (entries 1-2). No condensation product was obtained when using K 2 CO 3 base in MeOH at 25 • C (entry 3). The single condensation product 1a-al started to form in low yields at 70 • C in MeOH using DBU or NaOMe as a base (entries 4-5), which was optimized in THF at −78 • C using NaHMDS base (35% yield, entry 7). The double aldol condensation product 1a was first obtained in 20% yield when Triton B was used in MeOH at 70 • C, which was maximized with the NaOH base in MeOH/H 2 O (10:1) at 25 • C to obtain a 72% yield (entries 8-9). Homogeneity of reagents and base in solvent appeared to be most important in the double aldol condensation between C 10 dial 1 and acetophenone. a Acetophenone (3 equiv.) and C 10 dial 1 (1 equiv.) were reacted with each base (5 equiv.) under the specified condition in each solvent. b no detectible product was obtained (decomposed).
The conditions for double aldol condensation between C 20 crocetin dial 2 and acetophenone were then investigated starting with the above optimal condition using NaOH base in MeOH/H 2 O (10:1) at 25 • C ( Table 2). It was surprising that no condensation product was obtained under this condition (entry 1). The homogeneity issue again was decisive. The longer chain of 2 resulted in lower polarity, and removal of water from the reaction medium provided a single condensation product of 2a-al at a 48% yield when LiOMe was used as a base at 25 • C (entry 2). The highest yield (57%) of 2a-al was obtained from the NaOH base in MeOH at 70 • C, while low temperature (−78 • C) reactions with the metal amide bases in THF produced much lower yields (9~13%) of 2a-al (entries [3][4][5]. The reaction using Triton B as a base in MeOH produced no condensation products at 25 • C and no detectible product at 70 • C by decomposition (entries 6 and 7). The double aldol condensation product 2a was finally obtained in 40% yield under a lower polar medium of THF at 25 • C with Triton B as a base (entry 8) [38]. Double aldol condensation of acetophenone with C 30 polyene dial 3 was carried out under the above optimum condition of Triton B in THF at 25 • C to give 3a in 25% yield. Acetophenones with different ring substituents were selected to diversify the oxocarotenoid end groups. Various novel oxo-carotenoids 2a-2j and 3a-3b were prepared by conducting double aldol condensation between polyene dials 2 or 3 and acetophenones with the ring substituents to evaluate the effects of hydroxy and methoxy groups in terminal benzene rings on the antioxidant activities of the ketonic carotenoids ( Figure 2) [39]. MOM substitution (2b and 3b) was chosen for solubility reasons and electronegative chlorine substitution (2j) for comparison with other electron-rich groups. The specific reaction conditions for double aldol condensation and their yields are summarized in Table 3 (Supplementary Materials provide the experimental procedure and analytical data for 2b-2j and 3b), together with the maximum UV absorption wavelengths (λ max ) for the conjugated polyene system. The Density Functional Theory (DFT) calculation {rb3lyp/6-31 g(d,p) as an optimization function set} was performed to obtain the most stable geometries (Supplementary Materials), as well as the energy gaps (∆E) between HOMO and LUMO levels and the dihedral angles between the benzene ring and the polyene chain ( Table 3).
The wavelength (λ max ) of maximum UV absorption is related to the energy gap between HOMO and LUMO levels and represents the effective conjugation length of the π-system in oxo-carotenoids. The number (N) of conjugated C=C bonds for 3a and 3b was 13, whereas it was N = 9 for 2a-2j, and N = 5 for 1a. The UV absorption (λ max ) for 3a is 550 nm, for 2a it is 496 nm, and for 1a it is 441 nm, which decreases systematically as N decreases. Although DFT calculations predicted similar energy gaps between HOMO and LUMO levels within the 2a-2j series (N = 9), a significant red-shift of λ max was evident for ketonic carotenoids 2b-2j with auxochrome substituents on the benzene ring. Relatively precise values (510~520 nm) were observed except for 2c (549 nm) and 2i (497 nm), which can be explained by the effectiveness of conjugation. The dihedral angle between the benzene ring and the polyene chain for 2c was predicted to be 0 • for maximum conjugation due to intramolecular hydrogen bonding between the ortho-OH and carbonyl groups of the chain, whereas the 53.5 • dihedral angle for 2i indicates poor conjugation because of the sterically crowded ortho-dimethoxy substituents. a A = NaOH was used as a base in MeOH/H2O (10:1) at 25 °C; B = Triton B was used as a base in THF at 25 °C; C = NaOH was used as a base in MeOH at 70 °C; D = t-BuOK was used as a base in toluene at 110 °C. b Energy gap between HOMO and LUMO levels. c Dihedral angle between benzene ring and polyene chain. d 2e was not purified by SiO2 column but recrystallized from cold MeOH. The double aldol condensation of C20 crocetin dial 2 and acetophenones substituted with -OMOM, -OH, -OMe, and -Cl was mostly performed under NaOH base in MeOH at 70 °C to obtain oxo-carotenoids 2b, 2c, 2d, 2f, 2g, and 2j. Moderate to good yields (44~89%) were obtained except for 2j with para-chlorine substituent (16% yield). Oxo-carotenes 2h (69% yield) and 2i (39% yield) containing a para-methoxy group were successfully prepared by the condition using Triton B in THF at 25 °C. Preparation of oxo-carotenes 2e with para-hydroxy and 3b with para-methoxymethoxy substituent required harsh condition using t-BuOK base in toluene at 110 °C, but low yields (18% and 9%, respectively) were obtained.
The wavelength (λmax) of maximum UV absorption is related to the energy gap between HOMO and LUMO levels and represents the effective conjugation length of the πsystem in oxo-carotenoids. The number (N) of conjugated C=C bonds for 3a and 3b was 13, whereas it was N = 9 for 2a-2j, and N = 5 for 1a. The UV absorption (λmax) for 3a is 550 nm, for 2a it is 496 nm, and for 1a it is 441 nm, which decreases systematically as N decreases. Although DFT calculations predicted similar energy gaps between HOMO and LUMO levels within the 2a-2j series (N = 9), a significant red-shift of λmax was evident for ketonic carotenoids 2b-2j with auxochrome substituents on the benzene ring. Relatively precise values (510~520 nm) were observed except for 2c (549 nm) and 2i (497 nm), which can be explained by the effectiveness of conjugation. The dihedral angle between the benzene ring and the polyene chain for 2c was predicted to be 0° for maximum conjugation due to intramolecular hydrogen bonding between the ortho-OH and carbonyl groups of the chain, whereas the 53.5° dihedral angle for 2i indicates poor conjugation because of the sterically crowded ortho-dimethoxy substituents.  Table 3. The reaction condition and the yield for oxo-carotenoids in Figure 2, their UV maximum absorption wavelength, energy gap between HOMO and LUMO levels, and dihedral angle between benzene ring and polyene chain by DFT {rb3lyp/6-31 g(d,p)} calculation. a A = NaOH was used as a base in MeOH/H 2 O (10:1) at 25 • C; B = Triton B was used as a base in THF at 25 • C; C = NaOH was used as a base in MeOH at 70 • C; D = t-BuOK was used as a base in toluene at 110 • C. b Energy gap between HOMO and LUMO levels. c Dihedral angle between benzene ring and polyene chain. d 2e was not purified by SiO 2 column but recrystallized from cold MeOH.

Antioxidant Activities of Oxo-Carotenoids
Antioxidant activities for the prepared novel oxo-carotenoids together with apo- carotenedials (1-3) and apo-carotenals (1a-al, 2a-al) were then tested by applying the standard DPPH (radical) and ABTS (cationic radical) scavenging assays with β-carotene and canthaxanthin as references for simple and oxo-carotenoid, respectively. Unlike ABTS cationic radical which absorbs UV at 734 nm, the maximum absorption wavelength (517 nm) of DPPH radical overlaps with that of the above oxo-carotenoids (496~550 nm). DPPH radical is stable and the HPLC method for separation and measurement of the pure DPPH radical without interferences from oxo-carotenoids and other sources is reliable [30]. Therefore, the DPPH radical scavenging analysis of oxo-carotenoids and apocarotenoids was performed using the HPLC method.
Superoxide anionic radical (O 2 −· ) can be generated nonenzymatically by the reaction of phenazine methosulfate (PMS) and nicotinamide adenine dinucleotide (NADH) under aerobic conditions, which readily reduces nitroblue tetrazolium (NBT) to formazan [27][28][29]. The antioxidant activity of carotenoids is regarded as the competition with NBT for superoxide radical, which is often measured quantitatively by the reduction of the formazan absorption peak at 560 nm. Unfortunately, the coincidence of the UV absorption peaks with oxo-carotenoids prevented us from utilizing a formazan probe in the superoxide radical assay, and formazan was not even detected when using the HPLC method. Instead, t-butylhydroquinone (TBHQ) was chosen as an orthogonal probe in the oxidation to the corresponding benzoquinone by superoxide radicals [31]. The reduction in the TBHQ peak at 290 nm by oxo-carotenoids represents the antioxidant activity in superoxide radical assay. The HPLC method was utilized again to eliminate any potential interferences for the assay in the short wavelength UV region.
The antioxidant activity of the oxo-carotenoid was represented by the scavenging activity for each probe radical, which was measured as the reduced fraction in the UV absorption peak of each probe by oxo-carotenoids. The scavenging activity values of oxocarotenoids for DPPH radical (DR), ABTS cationic radical (AR), and superoxide anionic radical (SR) were measured three times and the mean and standard deviation are summarized in Figure 3 (see also Table S1 in Supplementary Materials) together with β-carotene and canthaxanthin as references. It is very interesting to compare the trend of radical scavenging activity of carotenoids in DR/AR (neutral and cationic radical) assays with that of the SR (anionic radical) assay in Figure 3.
The antioxidant activity of β-carotene in DR and AR assays is higher (20~30%) than that of canthaxanthin as was reported [22,23], whereas almost the same activities were observed in the SR assay. β-Carotene is a better antioxidant than oxo-carotenoids in DR and AR assays except for in the case of 2c (2'-OH) and 2i (2',4',6'-trimethoxy) in DR assay (vide infra), but it is the weakest antioxidant in SR scavenging except the short 1a (N = 5) series. The carbonyl groups in the rings are better in terms of scavenging DR and AR; on the other hand, those in the chain are effective in scavenging SR based on the comparison between 2a and canthaxanthin.
The antioxidant activities of the DR and AR assays were of a similar trend for the oxocarotenoids as outlined in the following: (1) apo-carotenedials 1-2 are much better in radical scavenging than the corresponding apo-carotenals (1a-al and 2a-al); (2) the substitution effect in terminal rings for radical quenching follows in the order of ortho > para > meta -OH and -OMe groups (2c-2h); (3) electron-rich terminal ring (2i) scavenges the radicals better than the electron-deficient one (2j). There also exist certain differences in that electron density provided by the substituents in the terminal rings is more important in DR assay (2i vs. 2j), and that phenolic end groups (2c-2e) are superior to the corresponding anisole end groups (2f-2h) in the AR assay. Effective conjugation by intramolecular hydrogen bonding between ortho-phenol and the carbonyl groups in 2c (0 • dihedral angle) is noteworthy for the 90.9% scavenging activity in the AR assay.
The antioxidant activity of oxo-carotenoids with the SR assay is much different to those for the DR and AR assays. In the SR assay, apocarotenals (1a-al and 2a-al) were found to be good antioxidants given the corresponding apo-carotenedials (1)(2). The Phenol end groups (2c-2e) are still better than anisole end groups (2f-2h), but there is no systematic positional effect of the substituent OH or OMe. Oxo-carotene 2j with an electron-deficient para-chlorobenzene ring is superior to the electron-rich oxo-carotene 2i, which is opposite to the results for the DR and AR assays. radical (SR) were measured three times and the mean and standard deviation are summarized in Figure 3 (see also Table S1 in Supplementary Materials) together with β-carotene and canthaxanthin as references. It is very interesting to compare the trend of radical scavenging activity of carotenoids in DR/AR (neutral and cationic radical) assays with that of the SR (anionic radical) assay in Figure 3. The antioxidant activity of β-carotene in DR and AR assays is higher of canthaxanthin as was reported [22,23], whereas almost the same activi in the SR assay. β-Carotene is a better antioxidant than oxo-carotenoids in except for in the case of 2c (2'-OH) and 2i (2',4',6'-trimethoxy) in DR ass it is the weakest antioxidant in SR scavenging except the short 1a (N = 5) s groups in the rings are better in terms of scavenging DR and AR; on the in the chain are effective in scavenging SR based on the comparison be thaxanthin.
The antioxidant activities of the DR and AR assays were of a sim oxo-carotenoids as outlined in the following: (1) apo-carotenedials 1-2 radical scavenging than the corresponding apo-carotenals (1a-al and 2a tution effect in terminal rings for radical quenching follows in the orde meta -OH and -OMe groups (2c-2h); (3) electron-rich terminal ring (2i) icals better than the electron-deficient one (2j). There also exist certain electron density provided by the substituents in the terminal rings is DR assay (2i vs. 2j), and that phenolic end groups (2c-2e) are superior All the experimental results on antioxidant activity of the oxo-carotenoids for each radical probe agree well with the mechanism in Scheme 2. Facile hydrogen radical abstraction from many allylic positions of β-carotene makes it a better antioxidant compared with most oxo-carotenoids in the DR assay, whereas high electron-density in the phenyl rings of oxo-carotene 2i (1,3,5-trimethoxy substitution) reversed the activity by stabilizing the resultant radical species (1). A single electron transfer from the polyene chain of oxocarotenoids to ABTS cationic radical produces unstable electron-deficient cationic radical activity of keto-carotenes, which positions β-carotene as a better antioxidant in the AR assay (2). On the other hand, the electron-deficient oxo-carotenoids would readily accept an electron [15] or form an adduct [16] from superoxide radical to produce carotenoid anionic radical, which means oxo-carotenoids are better antioxidants than β-carotene (3). Because the size of the frontier molecular orbital coefficient is correlated with the reaction site [40], DFT calculation was performed for 2b using the Multiwfn program [41] to predict the position of superoxide radical addition (Supplementary Materials). It is the central carbon (C-16) of the polyene chain that produces the highest coefficient of 7.33% in LUMO. It was demonstrated that superoxide anionic radical readily reacts with the conjugated carbonyl chain of oxo-carotenoids to produce the resonance-stabilized carotenoid anionic radical. Therefore, the antioxidant activity of oxo-carotenoids including xanthophylls should be fairly evaluated by anionic (radical) probe, such as the superoxide radical scavenging assay.
predict the position of superoxide radical addition (Supplementary Materials). It is the central carbon (C-16) of the polyene chain that produces the highest coefficient of 7.33% in LUMO. It was demonstrated that superoxide anionic radical readily reacts with the conjugated carbonyl chain of oxo-carotenoids to produce the resonance-stabilized carotenoid anionic radical. Therefore, the antioxidant activity of oxo-carotenoids including xanthophylls should be fairly evaluated by anionic (radical) probe, such as the superoxide radical scavenging assay.

Conclusions
We developed a synthetic method of ketonic carotenoids as a simple model for oxocarotenoids by performing aldol condensation of polyene dials and acetophenones containing different ring substituents. A practical chain-extension protocol for polyene dials was also established on a multigram scale based on the Horner-Wadsworth--Emmons reaction. The antioxidant activity of oxo-carotenoids was measured by performing a standard radical scavenging assay utilizing DPPH and ABTS radicals and we compared the results with those of the superoxide anionic radical assay. Although β-carotene was demonstrated to be a very potent antioxidant by DPPH and ABTS assays, the oxo-carotenoids exhibit superior antioxidant activity than the natural carotenoid in the superoxide radical scavenging assay. Each assay method suggests different antioxidant activity for the carotenoids, and the selection of the right probe for the ROS species of interest is important to obtain an impartial evaluation of carotenoids. It is the conjugated polyene chains that provide antioxidant activity to carotenoids. Increasing the electron density and increasing the effective conjugation length would be beneficial for superior antioxidant activity in DPPH and ABTS assays. The latter is also true for the superoxide radical scavenging assay but reducing the electron density in this case would favor superior antioxidant activity. Oxo-carotenoids are powerful antioxidants that were demonstrated to be valuable using the scavenging assay for superoxide radical, which is a precursor of most ROS.
Supplementary Materials: The following supporting information can be downloaded at www.mdpi.com/xxx/s1, Experimental Procedure for 2b-2j and 3b, Cartesian coordinates for the optimized geometry by DFT calculation, LUMO coefficient of 2b, Table S1: Antioxidant activity by DPPH, ABTS, and superoxide radical scavenging assays, 1 H-and 13 C-NMR spectra, Assignment of E-stereochemistry for 4 by N.O.E., and Peak assignment for 2b by COSY, HMBC, and DEPT.

Conclusions
We developed a synthetic method of ketonic carotenoids as a simple model for oxocarotenoids by performing aldol condensation of polyene dials and acetophenones containing different ring substituents. A practical chain-extension protocol for polyene dials was also established on a multigram scale based on the Horner-Wadsworth--Emmons reaction. The antioxidant activity of oxo-carotenoids was measured by performing a standard radical scavenging assay utilizing DPPH and ABTS radicals and we compared the results with those of the superoxide anionic radical assay. Although β-carotene was demonstrated to be a very potent antioxidant by DPPH and ABTS assays, the oxo-carotenoids exhibit superior antioxidant activity than the natural carotenoid in the superoxide radical scavenging assay. Each assay method suggests different antioxidant activity for the carotenoids, and the selection of the right probe for the ROS species of interest is important to obtain an impartial evaluation of carotenoids. It is the conjugated polyene chains that provide antioxidant activity to carotenoids. Increasing the electron density and increasing the effective conjugation length would be beneficial for superior antioxidant activity in DPPH and ABTS assays. The latter is also true for the superoxide radical scavenging assay but reducing the electron density in this case would favor superior antioxidant activity. Oxo-carotenoids are powerful antioxidants that were demonstrated to be valuable using the scavenging assay for superoxide radical, which is a precursor of most ROS.

Supplementary Materials:
The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/antiox11081525/s1, Experimental Procedure for 2b-2j and 3b, Cartesian coordinates for the optimized geometry by DFT calculation, LUMO coefficient of 2b, Table S1: Antioxidant activity by DPPH, ABTS, and superoxide radical scavenging assays, 1 H-and 13 C-NMR spectra, Assignment of E-stereochemistry for 4 by N.O.E., and Peak assignment for 2b by COSY, HMBC, and DEPT.