2024 Volume 10 Article ID: 2023-0045
Plants can produce anthocyanins to survive their given environments. A model plant Arabidopsis thaliana (Arabidopsis) accumulates the 11 anthocyanins (A1–A11) with various substitutions to their basic structure. However, it is difficult to detect all 11 anthocyanins in every analysis using conventional analytical techniques. In this study, we predicted ultraviolet–visible (UV-vis) absorption spectra of A1–A11 by an ab initio calculation based on density functional theory. We could identify substituents attached to the basic structure of anthocyanins from the predicted absorption peaks. We found that glucose substitution significantly reduced the solvation free energy due to five hydroxyl groups, whereas sinapoyl substitution increased the solvent effect due to the methyl group in a sinapoyl moiety. Our findings may explain the antioxidant capacity of each anthocyanin by comparing the predicted UV-vis absorption spectra with the measured UV-vis spectra. These outcomes can provide clues to uncover anthocyanin biosynthesis in different stress conditions/tissues.
Plants can produce complex compounds for their survival in any given environment because they cannot move by themselves. These compounds are the so-called plant-specialized metabolites that benefit the human health [1,2,3]. There are various plant-specialized metabolites. For example, phenolic compounds including anthocyanins have multiple biological effects, such as antioxidant activity, detoxification activity, antiproliferation, induction of apoptosis, antiangiogenic activity, and anti-inflammatory activity [1, 4, 5]. Anthocyanins are colored water-soluble pigments found in fruits, flowers, leaves, and roots [6]. The biosynthetic pathways for producing anthocyanins in plants have been extensively studied by applying natural product chemistry, biochemistry, transcriptomic, and metabolomic analyses [7,8,9,10]. A study of wild-type and PAP1/MYB75 in a pap1 dominant mutant (pap1D) of Arabidopsis thaliana (Arabidopsis) revealed that Arabidopsis can produce anthocyanins A1–A11 (Figure 1) [8, 10,11,12,13,14]. The contents of A1–A11 in the wild-type plants under nonstress conditions differ. A1, A2, and A4 could not be detected in the aerial parts, whereas A11 is the most abundant [11, 15, 16].
Anthocyanin modification pathway. The symbols in the table represents substituents; H, hydrogen; sina, sinapoylation; cou, coumaroylation, gly, glycosylation. The arrow colors correspond to coumaroylation (black), glucosylation (blue), malonylation (green), and sinapolylation (red).
Anthocyanins can absorb ultraviolet (UV)-B (300 nm) and visible light (520 nm). Antioxidant activities against reactive oxygen species generated by abiotic stress conditions (e.g., excess light, nutrient-limited conditions, salt, cold, and drought) could be correlated with UV absorption levels [17]. The UV spectra of anthocyanins can be obtained by referring to articles and comparing the UV spectra of reference compounds with the results of a liquid chromatography-photodiode array (LC-PDA) or liquid chromatography-mass spectrometry analyses. As these processes consume enormous time, a novel computational method for predicting each ultraviolet–visible (UV-vis) absorption spectrum with a short time and high accuracy is required. A1, A2, and A4 are intermediates in the biosynthetic pathway of anthocyanins. However, experimentally detecting all 11 anthocyanins simultaneously is difficult because some anthocyanin levels show low abundance in plants.
Predicting chemical properties from functional groups has been tried, but these have failed due to the complexity between functional groups [18]. Moreover, quantum chemical calculations have potential in this area because the structures of anthocyanins can be directly used as parameters to represent compounds’ physicochemical characteristics, including UV-vis spectra. Although ab initio quantum chemical calculations offer highly accurate predictions, such as absorption spectra of phenolic compounds and PTCDI (3,4,9,10-perylene-tetracarboxylic-diimide) [19, 20], the high computational cost limits the molecule size [18]. Meanwhile, semiempirical methods can calculate larger molecule sizes at lower costs. However, semiempirical methods require parameter adjustments to reproduce experimental values with qualitative accuracy.
In this study, we performed an ab initio calculation based on density functional theory (DFT) of UV-vis spectra to complement the experimental values of compounds with antioxidant activity, i.e., anthocyanins (A1–A11), as highly accurate calculated values of anthocyanins are required. This study could provide hints to elucidate why the abundance of the 11 anthocyanins differs under different stress conditions and tissues. By assigning the spectral changes caused by various substituents, the influence of substituents was systematically clarified. Anthocyanins undergo structural modifications accompanied by photophysical and chemical changes in response to solvent environments [21]. Therefore, the solvation free energy was also calculated to investigate the interaction between anthocyanin and the solvent.
A1. Cyanidin 3-O-[2”-O-(xylosyl) glucoside] 5-O-glucoside.
A2. Cyanidin 3-O-[2”-O-(xylosyl) glucoside] 5-O(6”’-O-malonyl) glucoside.
A3. Cyanidin 3-O-[2”-O-(xylosyl) 6”-O(p-coumaroyl) glucoside] 5-O-glucoside.
A4. Cyanidin 3-O-[2”-O-(2”’-O-(sinapoyl) xylosyl) glucoside] 5-O-glucoside.
A5. Cyanidin 3-O-[2”-O-(xylosyl)-6”-O-(p-coumaroyl) glucoside] 5-O-[6”’-O-(malonyl) glucoside].
A6. Cyanidin 3-O-[2”-O-(xylosyl)-6”-O-(4-O-(glucosyl)-p-coumaroyl) glucoside] 5-O-glucoside.
A7. Cyanidin 3-O-[2”-O-(2”’-O-(sinapoyl) xylosyl) 6”-O-(p-coumaroyl) glucoside] 5-O-glucoside.
A8. Cyanidin 3-O-[2”-O-(xylosyl) 6”-O(4-O-(glucosyl)-p-coumaroyl) glucoside] 5-O-[6”’-O-(malonyl) glucoside].
A9. Cyanidin 3-O-[2”-O-(2”’-O-(sinapoyl) xylosyl) 6”-O-(p-O-coumaroyl) glucoside] 5-O-[6”’-O-(malonyl) glucoside].
A10. Cyanidin 3-O-[2”-O-(2”’-O-(sinapoyl) xylosyl) 6”-O-(4-O-(glucosyl)-p-coumaroyl) glucoside] 5-O-glucoside.
A11. Cyanidin 3-O-[2”-O-(6”’-O-(sinapoyl) xylosyl) 6”-O-(4-O-(glucosyl)-p-coumaroyl) glucoside] 5-O-[6”’-O-(malonyl) glucoside].
The measured UV-vis spectrum values of A3, A4, A8, A9, A10, and A11 were collected from previous studies (Table S1) [10, 15, 17, 22, 23].
2.2 DFT calculationsDFT calculations with the level of B3LYP functional and 6-31G (d) for anthocyanins A1–A11 were performed using Gaussian 09 package [24]. After structural optimization with the initial planar structures, frequency calculations were performed to confirm the absence of imaginary frequencies and obtain the Gibbs free energies. We confirmed that all three rings of anthocyanin skeleton remained in the plane. UV-vis spectra were obtained from a single point of time-dependent DFT (TDDFT) calculations at the same level. The number of excited states for the calculation was in the range of 40–55. The calculated absorption energies were broadened by Gaussian functions with a half-width of 0.33 eV. The TDDFT method in conjunction with the B3LYP functional has been shown to successfully simulate the UV-vis spectra of some polyphenols [19, 25]. To obtain the solvation properties, we also performed structural optimization and frequency calculations using the self-consistent reaction field polarizable continuum model [26] with a dielectric constant of 78.3553. The solvation free energy (SFE) was obtained from the difference between the Gibbs free energy in the gas-phase and solvation.
The UV-vis spectra of the 11 anthocyanins predicted by DFT calculation are shown in Figure 2. Although the absorption strengths in solvents were slightly larger than those in the gas-phase, the shapes were not significantly different. The dielectric constants of other solvents, such as methanol (32.613) or DMSO (46.826), were smaller than that of water (78.3553); therefore, the physical properties related to UV-vis absorption of anthocyanin were unlikely to be affected by the solvent effect. For example, the spectrum of A3 (Figure 2) has peaks at 255 (shoulder), 312, and 483 nm in the gas-phase and 254, 362, and 486 nm in water. As these peaks were similar in shape to the 312-, 440-, and 524-nm peaks from experimental data [21], our calculations can sufficiently predict peaks to reproduce the real measurement spectra. We chose the spectra in the gas-phase for further analysis, as they were more consistent with the experimental peaks (Table S2).
UV-vis spectra of anthocyanins in the gas-phase (black) and solvent (blue). The x axis represents the wavelength (nm), whereas the y axis represents the calculated absorbance (ε).
Next, we focused on how the spectrum changed by substituting the following four molecules: malonyl, coumaroyl, sinapoyl, and glycosyl groups. When a malonyl group was substituted on A1 to A2, the UV-vis spectra of A1 and A2 showed no significant changes in peak shapes. The possible reason is that the malonyl group has a low absorption coefficient, less than 500 L mol⁻1 cm⁻1 at 225 nm, with the level of B3LYP/6-31G (d). This trend was also found in other malonylation, such as A3 to A5 and A6 to A8. Luo et al. [10, 15] analyzed the UV-vis spectra of A10 and A11 using LC-PDA and found that there was no significant spectral change between A10 and A11 by substituting the malonyl group in A10. This observation agrees with our computational results, i.e., malonylation has no significant effects on the absorption spectrum (Figure 2). Our calculation showed that it is possible to predict the absorption spectra of molecules such as A1 and A2 that are difficult to measure directly by experiment due to the very low yield of A1 and A2 in Arabidopsis [11, 15, 16].
3.2.2 CoumaroylationA3 was formed when a coumaroyl group was added to A1 (Figure 1). The major difference in the spectra of A1 and A3 is the appearance of a new peak at 310 nm in A3 (Figure 2). This new peak corresponds to the highest occupied molecular orbital (HOMO) to the lowest unoccupied molecular orbital (LUMO) + 1 transition (left side of (Figure 3). The HOMO and LUMO + 1 of A3 have π-bonding and π*-bonding characteristics, respectively; therefore, the absorption at 310 nm can be assigned to a π → π* transition. The HOMO and LUMO + 1 of A3 correspond to the HOMO and LUMO of coumaroyl, respectively. The coumaroyl group has an absorption peak at 290 nm, about 26,000 L mol⁻1 cm⁻1 of the HOMO to LUMO transition (Figure S1), suggesting that the new peak of A3 in the UV-vis spectrum was derived from the addition of a coumaroyl moiety. The new peak of A3 at 310 nm was slightly red-shifted from the peak of the coumaroyl group at 290 nm, attributable to the energy gap between LUMO + 1 and HOMO of A3 (4.085 eV) being lower than that between the LUMO and HOMO of coumaroylation (4.363 eV). This trend was also found in other coumaroylation, such as A2 to A5 and A4 to A7 (Figure 2).
MO energies and isodensity plots of A3 and coumaroylation. The red and blue surfaces represent the MO phases.
A4 was formed when a sinapoyl group was added to A1 (Figure 1). The major difference in the spectra of A1 and A4 is the appearance of a new peak at 350 nm in A4. This new peak corresponds to the HOMO − 2 to LUMO + 1 and HOMO to LUMO + 1 transition (Figure 4). The HOMO − 2 and HOMO of A4 have π-bonding characteristics and the LUMO + 1 has a π*-bonding characteristic; therefore, the absorption at 350 nm can be assigned to a π → π* transition. The HOMO − 2, HOMO, and LUMO + 1 of A4 correspond to the HOMO − 1, HOMO, and LUMO of the sinapoyl group, respectively. The sinapoyl group has an absorption peak at 320 nm, about 21,000 L mol⁻1 cm⁻1 (Figure S1). Therefore, the new peak of A4 can be predicted to be derived from a sinapoyl modification. The new peak of A4 at 350 nm was slightly red-shifted from the peak of a sinapoyl group at 320 nm because of the energy gap between LUMO + 1 and HOMO − 2 for A4 (4.199 eV) is lower than that between LUMO and HOMO − 1 (4.674 eV).
MO energies and isodensity plots of A4 and sinapoylation. The red and blue surfaces represent the MO phases.
A6 was formed when a glucosyl group was added to A3 (Figure 1). Although the absorption spectra of the two molecules were similar, the maximum absorption coefficient of A6 (39,000 L mol⁻1 cm⁻1) was slightly higher than that of A3 (32,000 L mol⁻1 cm⁻1) (Figure 2). The main peak of 320 nm in A3 and that of A6 corresponded to the transition between HOMO and LUMO + 1. Figure 5 shows the HOMO and LUMO + 1 of A3 and A6. Because all these orbitals originate from a coumaroyl moiety, the peak at 320 nm in A3 and A6 was considered to be derived from a coumaroyl modification. Comparing the MOs between A6 and A3, the MOs in A6 slightly extended to a glycosyl group at the position of the p-coumaroyl glucoside, particularly HOMO (the inset of Figure 5). The higher absorption coefficient in A6 may be due to the larger transition dipole moment. This trend was also found in other glucosylation, such as A5 to A8, A7 to A10, and A9 to A11.
MO energies and isodensity plots of A6 and A3. The inset shows the HOMO of A6 in wire modes. The HOMO of p-coumaroyl expanded to the glycosyl group. The red and blue surfaces represent the MO phases.
Solvent effects are essential for the function of compounds, including anthocyanidins [27]. Table 1 shows the SFE of each molecule. As a general trend, the molecules from A5 to A11 showed larger energy, except for A7 and A9.
| SFE | |
| A1 | −83.6 |
| A2 | −88.5 |
| A3 | −86.0 |
| A4 | −81.9 |
| A5 | −90.0 |
| A6 | −98.6 |
| A7 | −83.9 |
| A8 | −102.8 |
| A9 | −88.8 |
| A10 | −96.5 |
| A11 | −102.3 |
To examine the effect of substituents, the differences in each molecule were taken and summarized for each substituent (Table 2). Glucosylation significantly reduced the SFE in the 11 anthocyanins, followed by malonylation, coumaroylation, and sinapoylation (Table 2). The glucose substituent provided five hydroxyl groups, which showed significantly lower SFE than other substituents. Malonylation possessed two hydroxyl groups due to the reduction of the SFE, although the change was smaller than that of the glucosyl group. Compared with the coumaroyl and sinapoyl groups attached to the basic structure of anthocyanin, a methyl group at a sinapoyl moiety contributed to hydrophobic characteristics and hence increased the SFE of A1 to A4, A3 to A7, A5 to A9, A6 to A10, and A8 to A11 (Table 2).
| Malonylation | A1 -> A2 | −5.0 | Sinapoylation | A1 -> A4 | +1.7 | |
| A3 -> A5 | −3.9 | A3 -> A7 | +2.2 | |||
| A6 -> A8 | −4.2 | A5 -> A9 | +1.2 | |||
| A7 -> A9 | −4.9 | A6 -> A10 | +2.0 | |||
| A10 -> A11 | −5.8 | A8 -> A11 | +0.5 | |||
| Coumaroylation | A1 -> A3 | −2.5 | Glucosylation | A3 -> A6 | −12.6 | |
| A2 -> A5 | −1.4 | A5 -> A8 | −12.8 | |||
| A4 -> A7 | −2.0 | A7 -> A10 | −12.7 | |||
| A9 -> A11 | −13.5 |
We performed DFT calculations to investigate the effects of substituents on the photo absorption properties and solvent effects of anthocyanins. We found that malonylization had no significant effect on the absorption properties due to its low absorption coefficient. The substitution of coumaroyl and sinapoyl groups yielded the appearance of new peaks at 310 and 350 nm, respectively. MO analysis revealed that these new peaks were derived from the corresponding substituents. The substitution of glucose could intensify the peak at 320 nm, which was derived from a coumaroyl moiety. The MOs of the coumaroyl group were slightly expanded compared with those of glycosylation. This expansion of MOs increased the dipole transition moment and enhanced the absorption peak of the coumaroyl group. Several anthocyanins are of low abundance in plants and hence are difficult to measure experimentally, e.g., A1 and A2. In addition, the IUPAC nomenclature for complex compounds such as anthocyanins makes it difficult to search compound information including physical properties in SciFinder (https://scifinder-n.cas.org/) and Dictionary of Natural Products (https://dnp.chemnetbase.com/faces/chemical/ChemicalSearch.xhtml). This study shows that the calculated spectra agree with experimental ones, indicating that quantum chemical calculations can predict the physical properties of anthocyanins with high accuracy. We also investigated the effects of substituents on solubility. The substitution of glucose significantly reduced the SFE and made anthocyanins easier to dissolve in water, whereas sinapoylation could increase the SFE due to its methyl group.
In this study, we focused on the anthocyanins produced by Arabidopsis. The overaccumulation of A11 in plants remains unknown, although the biosynthesis of A11 requires higher costs than that of A1 in terms of metabolism. A11 has two times higher absorbance than A1 and cyanidin 3-O-glucoside at around 300 and 500 nm, respectively (Figure 2) [17]. Further, sinapoylation such as A11 can increase the antioxidant capacity of anthocyanins [28], which could be a reason plants accumulate more A11 than other anthocyanins.
Moreover, A2 attached with a sinapoyl group has not been detected in Arabidopsis. Our developed method can generate “artificial” anthocyanins with calculated physical properties. This study may broaden our understanding of anthocyanin biosynthesis. Semiempirical methods are required to calculate larger and more diverse molecules in plants. This study also provides a reference for adjusting the parameters of semiempirical methods for future research.
Table S1: Measured UV-vis spectra obtaited from the articles.
Table S2: Maximum absorption wavelength in the experiments and calculations.
S2: Absorption spectra of modified moleculesFigure S1: UV-Vis spectra of (a) malonyl, (b) coumaroyl, (c) sinapoyl, and (d) glucose.
We would like to thank Dr. Umpei Nagashima for fruitful discussions. We acknowledge Enago for English language review. This work was supported by the “Sustainable Food Security Research Project” in the form of an operational grant from the National University Corporation. This work was also supported partly by JSPS KAKENHI (Grant Number 19K05711 to M.K).
