Molecular Modeling Based on Time-Dependent Density Functional Theory (TD-DFT) Applied to the UV-Vis Spectra of Natural Compounds

Abstract

As diseases caused by solar radiation have gained great prominence, several methods to prevent them have been developed. Among the most common, the use of sunscreens is customary and accessible. The application of theoretical methods has helped to design new compounds with therapeutic and protective functions. Natural compounds with described photoprotective potential properties (3-O-methylquercetin, gallic acid, aloin, catechin, quercetin, and resveratrol) were selected to perform theoretical studies. Computational methods were applied to predict their absorption spectra, using DFT and TD-DFT methods with functional B3LYP/6−311+g(d,p) basis sets and methanol (IEFPCM) as a solvent. The main electronic transitions of the compounds were evaluated by observing whether the differences in HOMO and LUMO energies that absorb in the UV range are UVA (320–400 nm), UVB (290–320 nm), or UVC (100–290 nm). Experimental validation was carried out for EMC, quercetin, and resveratrol, demonstrating the consistency of the computational method. Results obtained suggest that resveratrol is a candidate for use in sunscreens. The study provided relevant information about the in silico predictive power of natural molecules with the potential for use as photoprotective adjuvants, which may result in fewer time and resource expenditures in the search for photoprotective compounds.

1. Introduction

A significant focus in public health has been placed on the risks caused by unprotected exposure to sun radiation, especially skin cancer, which has shown a significant increase in cases every year [1]. In light of this, studies to understand and develop better and safer photo-protective compounds for sunscreen formulations have become more desirable. Thus, it is necessary to expand the knowledge regarding compounds used in the prevention of the damage caused by ultraviolet (UV) radiation [2,3,4].

UV radiation is divided into three distinct categories according to their wavelength: UVA radiation (320–400 nm) is known to induce direct DNA damage, utilizing reactive oxygen species to create free radicals, which mostly lead to photoaging. UVB radiation (290–320 nm) has high energy, causes erythema and immunosuppression, and is associated with skin cancer. UVC radiation (100–290 nm) has especially high energy, making it extremely harmful to living beings; however, hardly any of it can get through the ozone layer present in the stratosphere. Therefore, UVB radiation is the one we are exposed to the most and the most harmful to human beings [5,6,7,8].

Among the main methods of preventing sun damage, the use of sunscreens is strongly recommended [9,10]. Sunscreens are composed of molecules that can absorb or reflect the sun’s radiation in the UVA and UVB ranges. Currently, they are divided into two classes: organic, which can absorb solar radiation mostly in the UVA and UVB range, and inorganic, which act by scattering and reflecting UV radiation [11,12].

Consequently, because of its growing importance in everyday life, new strategies are being sought to develop better sunscreen formulations with wider UV coverage, more favorable aesthetics, greater adherence, and minimum skin penetration [13]. Computational studies utilizing molecular modeling methods have shown promising results in predicting the behavior of different compounds when interacting with UV radiation. This can be applied to extracting relevant information about the mechanism of action, understanding electronic transitions, and planning novel and potent compounds with photoprotective properties [14,15].

Molecular modeling is a group of computational methods that simulate physicochemical systems, seeking to generate, manipulate, and analyze realistic representations of molecular structures obtained from physicochemical properties calculated by computational chemistry techniques. Thereby, these methods can extract valuable information, which allows for a better understanding of the spectral, structural, and electronic behavior, that can be used to optimally plan for new compounds with the desired physicochemical properties [16,17].

Density functional theory (DFT) is a quantum mechanical method that assumes that all the system properties are charge density functions. This enables the calculation of the exact description of the structure, energy, and molecular properties of the compound. Utilizing DFT for excited states, the time-dependent density functional theory (TD-DFT) can obtain the electronic spectra for absorption and study several processes that involve the excited state [17,18,19,20,21,22,23,24].

Therefore, the main goals of this work were to study the electronic structures as well as the excited states of natural molecules to understand the electronic mechanism related to UV radiation absorption employing TD-DFT methods.

2. Materials and Methods

2.1. Dataset

Six compounds from natural sources with possible UV filter properties described in the literature were selected. Their chemical structures and names are presented in

Table 1. Compounds with two-dimensional (2D) structures and names.

Compound	Structure
3-O-methylquercetin
gallic acid
aloin
catechin
quercetin
resveratrol

[25,26,27,28,29].

The compound ethylhexyl methoxycinnamate (EMC) (Figure 1) was used as a reference for the study since it is a consolidated UVB filter [30,31,32].

2.2. Computational Protocol

A UV spectra simulation was carried out for EMC and the natural compounds presented in Table 1. The protocol described in Figure 2 was applied individually for each compound.

The structural model construction was done with the program GaussView 4 [33]. After that, the compounds were submitted to conformational calculation in Spartan’14 [34], using the semi-empirical method PM6. The most stable conformation had its geometry optimized by the DFT method, with functional B3LYP and base 6−311+G(d,p) in Gaussian 09 [35,36,37,38]. The solvent effect used was evaluated by the methanol (ε = 32.61) implicit solvent method IEFPCM, which was chosen because it is the same solvent employed in the experimental assay. The vibrational analysis was carried out in Gaussian 09 at the same theory level used in the geometrical optimizations, which ensures the local minimums are confirmed by the absence of an imaginary mode in the vibrational analysis calculations. Excited states were calculated with the TD-DFT method using the same functionals and base optimization calculations as the optimized geometry fundamental state. GaussView was used to extract the data regarding the epsilon coordinates (molar absorption) at the respective wavelengths from 100 to 500 nm, obtaining theoretical UV-vis absorption graphs. To validate the most suitable methodology to simulate the UV-vis, the calculation was performed for EMC, since it is the most commonly used UVB filter in sunscreen formulations and was used to validate the last studies published by the research group [14,39].

2.3. Experimental Protocol

The absorption profiles were generated by spectrophotometry. Quercetin and resveratrol were used to compare the experimental and theoretical absorption profiles equivalence. To obtain the experimental data, a spectrophotometer was used in the range of 240–400 nm, and a methanol solution of 3.048 mg/L was used for quercetin and 3.06 mg/L for resveratrol.

3. Results and Discussion

After the construction of the compound’s structures in GaussView 4 [33], they were submitted to conformational analyses in Spartan’14 [34]. The conformational search by the semi-empirical PM6 method was carried out to select, among all possible 3D conformations, the most stable one, which has the lowest energy [40,41,42,43].

The most stable conformation was subjected to DFT calculations, seeking to understand its energy transitions and perform a vibrational analysis, different studies also used DFT for conformational and energy calculations [44,45].

For the DFT methods, functional B3LYP and base 6−311+G(d,p) were used, using the implicit solvent methanol (IEFPCM) in Gaussian 09; the functional B3LYP was used because it is a functional already consolidated in the prediction of UV-vis. Previous studies by the group, by Garcia 2015, also evaluated the effect of different functionals on theoretical UV-vis predictions, and the one that had the highest accuracy for organic molecules with photoprotective qualities was B3LYP [14,15,39,41,46,47].

In addition to maintaining the use of a validated methodology for the simulation of UV-vis spectra for molecules with photoprotective characteristics, diffusion (+) and polarization (d,p) functions were added. This generates more accurate and reliable data in studies of electronic distribution for conjugated systems [48].

The UV-vis predictions on EMC were performed to acquire the knowledge and skills necessary to execute the calculations and to validate the method using Gaussian 09. Accordingly, the results obtained corroborate with other results previously found by our group [14], which enables us to consider the method validated and appropriate for advancing the project. The UV-vis spectrum obtained shows molar absorptivity, the main electronic transitions (mainly singlet states, also known as absorption bands), and oscillator force (Figure 3).

Vibrational (frequency) analysis was carried out at the same theoretical level as the geometrical optimizations to confirm the absence of imaginary frequencies. It is essential that the conformation obtained be localized at the minimum energy point with a positive frequency, since negative frequencies show imaginary conformational conditions [24,49]. For the studied compounds, there were no negative frequency implications, showing that the conformational analysis performed using semi-empirical PM6 and DFT methods was adequate and that the conformations were both real and of sufficient quality to advance the study.

The next step was to calculate the electronic properties using the TD-DFT method with functional B3LYP and 6−311+G(d,p) basis sets. These calculations allow us to generate enough data to comprehend the main theoretical electronic states and UV-vis absorption profiles (molecular absorptivity data). Using GaussView 4, the information related to transition energy studies and images of the main electronic transition for each compound and their UV-vis spectra were extracted and are presented in Figure 4 [14].

The UV-vis prediction results obtained by applying the TD-DFT described method correspond satisfactorily to data obtained through experimental studies found in the literature, with a difference between the theoretical and experimental data ranging between 1 and 5.8% (

Table 2. Theoretical λmax obtained by the work methodology and the experimental λmax found in the literature and its % error.

Compound	λmax (nm) Theoretical	λmax Experimental (Literature)	% Error
3-O-metilquercetina	361.77	358 nm	1
gallic acid	284.74	290 nm	1.8
aloin	371.79	353 nm	5
catechin	258.74	274 nm	5.8
quercetin	383.01	375 nm	2
resveratrol	324.93	307 nm	5.5

).

The compound 3-O-methylquercetin (Figure 4A) had the lowest error rate in the study. Its theoretical λmax calculated at 361.77 nm, while experimental studies suggest that its λmax is 358 nm, presenting an error rate of 1% [50]. Gallic acid (Figure 4B) also had a low error rate; its theoretical λmax was calculated at 284.74 nm, and experimental studies indicate that its λmax is 290 nm, showing a difference of 1.8% [51,52]. Aloin (Figure 4C) had its theoretical λmax at 371.79 nm, and experimental studies suggest that its λmax is 353 nm, presenting a difference of 5% [53,54]. Catechin (Figure 4D) showed the highest error rate, with a theoretical λmax of 258.74 nm and an experimental λmax suggestive of 274 nm; the difference found was 5.8% [55]. Quercetin (Figure 4E) showed great proximity between the results; its theoretical λmax is 383.01 nm; experimental studies suggest that its λmax is 375 nm, and the error rate found is 2% [56]. Resveratrol (Figure 4F) has a theoretical λmax of 324.93 nm, experimental studies in the literature report a λmax of 307 nm with a 5.5% error rate [57].

A comparison between the previous theoretical studies and the results obtained in this work showed that the methodologies are converging. In these studies, gallic acid was assessed using the functions B3LYP and 6−311+G(d,p) basis sets; in water (IEFPCM) the absorption peak (λmax) is 275 nm, while our study showed 284 nm (a 3% difference between the results), probably due to the solvent used [47]. For quercetin, the difference between the studies performed by Conard et al. [46] using the B3LYP and 6−31G(d,p) basis sets and methanol (PCM) as solvent was negligible, with about 1% of the calculated absorption peak at 381 nm. Moreover, the theoretical studies performed on resveratrol, again using the B3LYP function with 6−31G* in vacuum, resulted in an absorption peak at 315 nm [58]. In our work, we obtained the λmax of 324 nm with a difference of 2.8% from previous work, and we attribute this difference to the distinct basis function as well as the solvent employed.

The main electronic transitions calculated by TD-DFT were also analyzed. They demonstrate each orbital’s contribution so that the main electronic transition and the area responsible for the molar absorptivity peak (λmax) can occur.

Using GaussView, it was also possible to calculate the orbital involved in the main electronic transition and obtain the energies from the orbital between HOMO -3 and LUMO +3. The results of the electronic transition, expressed in eV, are presented in

Table 3. Energy and contribution of molecular orbitals related to electronic transitions calculated using functional B3LYP.

Compound	λmax (nm)	Energy (eV)	Electronic Transition Contribution (%)
ethylhexyl methoxycinnamate (EMC)	322.24	3.84	H → L (+99%)
3-O-methylquercetin	361.77	3.42	H → L (+93%)
gallic acid	284.74	4.35	H-1→ L (+62%) H → L (+27%) H → L+1 (+9%)
aloin	371.79	3.33	H → L (+98%)
catechin	258.74	4.79	H → L (+74%) H-1 → L (+9%) H-3→ L+1 (+7%)
quercetin	383.01	3.24	H → L (+96%)
resveratrol	324.93	3.81	H → L (+97%)

and were used to generate the graphs in Figure 5. The representations and images were generated in GaussView, demonstrating the orbitals involved in the main electronic transition and the energy necessary for this transition to occur. For all compounds, this transition happens in the near area of the molecule’s λmax, in the UV-vis range.

Generally, most compounds presented a HOMO→ LUMO transition in λmax. However, gallic acid presents a contribution in HOMO −1 → LUMO and HOMO → LUMO +1, and catechin shows a contribution from HOMO −1 → LUMO and HOMO -3 → LUMO +1.

Regarding the compounds studied, this correlation can be observed in the UV-vis graphs generated. The results suggest that the closer the wavelength is to visible light (400 nm), the smaller the energy gap. For catechin (Figure 5E), the calculations show an energy gap of 4.79 eV and λmax of 258.74 nm, which is confirmed by its UV absorption profile, demonstrating that it acts in the UVC range (100–290 nm) (Figure 4D). Quercetin (Figure 4), on the other hand, has shown an energy gap of 3.24 eV and λmax of 383.01 nm, indicating that its highest UV absorption is in the UVA range (320–400 nm) (Figure 4E).

Thus, according to the studies realized in this work, compounds 3-O-methylquercetin (Figure 4A), aloin (Figure 4C), quercetin (Figure 4E), and resveratrol (Figure 4F) show an absorption range similar to EMC (Figure 5B,D,F,G), which implies that they deserve attention and are promising candidates for photoprotection.

The other compounds studied, such as gallic acid (Figure 4B) and catechin (Figure 4D), presented absorption profiles that were mostly in the UVC (100–290 nm) and UVB (290–320 nm) ranges and showed a wider energy gap than the other compounds studied. This indicates that these molecules might be interesting for chemical sunscreen compositions, along with other compounds that encompass other areas in the UV spectrum.

Thus, these correlated pieces of information can be promising if utilized to better predict the compound’s electronic behavior and photoprotective properties.

Experimental Validation

To validate the results, a study of the absorption profiles generated through the experimental protocol for the compounds quercetin and resveratrol was carried out, evaluating whether the absorption profiles would be equivalent to those obtained by the computational protocol (Figure 6).

The results obtained from both compounds validate the prediction results of the computational protocol, as both the absorption profiles and the λmax presented themselves in similar areas of the spectra. For quercetin, the experimental λmax was located at 371 nm and the theoretical λmax at 383.01 nm, and for resveratrol, the experimental λmax was located at 305 nm and the theoretical λmax at 324.93 nm. These comparative results obtained by experimental and computational techniques are needed to demonstrate that, even for molecules that differ from each other, the predictive results are within the expected range when analyzing the absorption curves.

4. Conclusions

The aim of this work was to use computational methods, mostly TD-DFT (B3LYP), to simulate different electronic conditions and seek to better understand how the compounds studied act as photoprotective candidates. Examining the UV-vis absorption profiles revealed that the study was quite satisfactory. The HOMO and LUMO electronic transitions for the compounds 3-O-methylquercetin, aloin, gallic acid, catechin, resveratrol, and quercetin showed promising results when evaluating the energy gap and the UV-vis absorption range. The calculated energy gap values correspond to a range of 4 eV, which confirms that the molecules studied have an absorption in the UV region. The experimental protocol for the absorption profiles of quercetin and resveratrol molecules validated the absorption profiles obtained through the computational protocol (TD-DFT).

In conclusion, our study demonstrated, through computational methods, that the UV spectra prediction method is applicable for compounds of natural origin with photoprotective properties, two of which were experimentally validated. Thus, our work opens the possibility of tracking the most promising compounds in silico, which can minimize failures in experimental validations and save laboratory resources. However, we highlight the need for additional experimental validations to refine the mathematical calculations and error rates.