Introduction

There is considerable interest in studying polyphenols, and thus, experimental and computational studies on the isolation/extraction, synthesis, characterization, and reactivity of such compounds are prevalent in the very recent literature [1,2,3,4,5,6,7,8,9,10]. Dihydrochalcones (DHCs) are a class of polyphenols that are present in extracts of more than forty weed plants and deciduous trees from families such as Leguminoseae, Laureaceae, and Lorantaceae [11]. These compounds attract considerable attention for being used in the human diet as functional foods due to their pharmacological activity, including antioxidant, antibacterial, anti-inflammatory, antitumor, and antiviral properties [12,13,14,15,16]. In terms of antioxidant activity, it is well known that the potential of a given substance can be probed through its capability to scavenge free radicals by (mainly) the mechanisms of hydrogen-atom transfer (HAT) and single electron transfer (SET) [17,18,19,20,21]. In HAT, where an H-atom is transferred to a free radical, i.e.,

$$ \text{ArOH} + \mathrm{R}^{.} \rightarrow \text{ArO}^{.} + \text{RH}, $$
(1)

there is a strong dependence on the O-H bond dissociation enthalpy (BDE), since there will be higher antioxidant activity when there is a weaker O-H bond [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37]. The BDE (i.e., energy to break the O-H bond) is computed as the difference in the heat of formation between the molecule (ArOH) and corresponding radical (ArO.)[38]. On the other hand, in SET, a single electron is transferred from the molecule (ArOH) to the free radical (R.), i.e.,

$$ \text{ArOH} + \mathrm{R}^{.} \rightarrow \text{ArOH}^{+} + \mathrm{R}^{-} \rightarrow \text{ArO}^{.} + \text{RH}. $$
(2)

Clearly, the smaller the ionization potential (IP) for ArOH, the lower the energetic cost to abstract an electron. By examining the BDEs and IPs of DHCs, additional information can be obtained about potential compounds with applications as phytotherapeutics [39]. Very recently, we performed a computational study on two flavonols that were isolated from Loranthaceae family plant extracts: kaempferol 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside and quercetin 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside [40], which are glycosylated versions of kaempferol and quercetin, respectively. One of the goals of the previous study was to probe how much the presence of the sugar group would affect the values of BDEs and IPs. In the present work, we extend this idea to examining phloretin and its glycoside phlorizin. The structures of both compounds can be seen in Fig. 1. Phloretin and phlorizin have been studied as candidate molecules for skin-based drug delivery [41], anti-inflammatories [42], and UV light-induced photodamage protectors [43]. There are various pharmacological studies on phloretin [44,45,46,47], including in vivo and in vitro investigations on the antioxidant activity [48, 49]. Two studies reporting BDEs for phloretin have previously been published: Kozlowski et al. [50] computed BDEs for a series of chalcones and dihydrochalcones in the gas-phase, methanol, and water using density functional theory (DFT) with the B3P86 functional and the 6-31 + G(d,p) basis set. Yan et al. [51] utilized the B3LYP functional [52,53,54,55] and the 6-311G(d,p) basis set [56,57,58,59] to compute the BDEs for phloretin in the gas-phase, benzene, and water; no IPs were determined. To the best of our knowledge, no investigation on BDEs or IPs for phlorizin were performed to date. Hence, the present work provides additional information on the potential antioxidant activity of these compounds including the role of the sugar. The choice of functional and basis set on the effect of the sugar has been examined to determine that a consistent story emerges independent of the computational method. The geometries of phloretin and its glycoside phlorizin were optimized to obtain structural parameters, and then based on these structures, the corresponding BDEs and IPs were determined in the gas-phase, n-hexane, ethanol, methanol, and water. A comparison of the present values to previous computational and experimental results for other flavonols suggest both compounds should exhibit antioxidant activity. In addition, the presence of the sugar moiety has only a modest effect on the BDEs and IPs. Therefore, experimental investigations on BDEs of both compounds are encouraged to verify the predictions made in the present work.

Fig. 1
figure 1

Representation of the chemical structures, along with atom numbering, for the compounds of interest in the present work

Full size image

Computational methods

DFT was the computational approach applied in the present work. In a very recent study, La Rocca et al. [60] benchmarked twenty-one (21) commonly used exchange-correlation functionals for the determination of the BDEs and IPs for two selected molecules that are well known to present antioxidant activity: quercetin and edaravone. The conclusion was that M05-2X [61], M06-2X [62], and LC-ω PBE [63] were the preferred functionals to compute the antioxidant behavior. Hence, in the present work, we decided to use one of Minnesota family functionals (M06-2X), LC-ω PBE and the widely used B3LYP (to directly compare our results with BDEs for phloretin avaliable in the literature). Geometry optimization (using default convergence criteria and without symmetry constraints) of the neutral molecules and their radicals were carried out using the B3LYP, M06-2X, and LC-ω PBE exchange-correlation functionals [52,53,54,55, 62, 63] with both the 6-311G(d,p) and 6-311 + G(d,p) basis sets [56,57,58,59]. The initial structure for the optimization of the neutral phloretin was built following the results achieved in the conformational analysis performed by Kozlowski et al. [50]. In terms of phlorizin, the structure used was based on the crystal structure presented by Aufmkolk et al. [64]. Additionally, computations on the energies of other conformers were performed to verify if the crystallographic data corresponded to the lowest energy conformer for the given computational method. Regarding the differences, there are several low-lying conformers that are very close in energy (0.001–0.4 kcal/mol). Therefore, these are clearly energetically accessible; a figure with the details of the structural parameters of the conformer utilized can be found in the Supplementary Information. For the open-shell species, the unrestricted formalism (i.e., UB3LYP, UM06-2X, U-LC-ω PBE) was used for all computations. To confirm the conformations as minima, to evaluate the zero-point energy (ZPE) corrections, and to determine enthalpies (at 298 K), vibrational frequencies were computed for all optimized structures. Solvent effects were included using the integral equation formalism polarizable continuuum model (IEF-PCM; which shall hereafter be simply referred to as PCM) [65,66,67]. The solvents considered included n-hexane (𝜖= 1.8819), ethanol (𝜖= 24.852), methanol (𝜖= 32.613), and water (𝜖= 78.3553). Single-point second order Møller-Plesset perturbation theory computations, MP2/6-31G(d,p)//B3LYP/6-311 + G(d,p), were also performed to provide comparison between the results obtained using a wave function based method with those obtained with DFT.

The BDEs were determined using the enthalpies of formation for the radical generated via H-atom abstraction H f (ArO.), for the H-atom H f (H.), and for the neutral molecule H f (ArOH), such that

$$ \text{BDE} = \mathrm{H}_{f}(\text{ArO}^{.}) + \mathrm{H}_{f}(\mathrm{H}^{.}) - \mathrm{H}_{f}(\text{ArOH}). $$
(3)

For each compound, the IPs were determined as the difference between the (electronic + ZPE) energies for the neutral (ArOH) and the cation (ArOH+). All computations in the work presented here were accomplished using the Gaussian 09 software suite [68].

Results and discussion

Structures

Table 1 presents selected optimized bond lengths, bond angles, and dihedral angles for phloretin and phlorizin determined in the gas-phase using the B3LYP, M06-2X, and LC-ω PBE functionals with the 6-311G(d,p) and 6-311 + G(d,p) basis sets. For both compounds, the bond lengths and bond angles agree very well with each other (within 0.003 Å and 1 degree) for a given functional when the basis set is changed (as expected); the dihedrals for phlorizin show a larger (up to 15 degrees) basis set dependence. Thus, for the B3LYP, M06-2X, and LC-ω PBE approaches, there is only a very weak dependence when describing the bond lengths and angles for the two compounds of interest. This is in agreement to what was reported in previous studies on other phenolic compounds [25, 40].

Table 1 Selected bond lengths (Å) and angles (degrees) from the optimized structures of phloretin and phlorizin in the gas-phase
Full size table

The present results (determined in all the approaches used) suggest there is a moderate interaction between H8 and O a in phloretin (with the distance between H8 and O a being approximately 1.6 Å). Hence, the \(r_{(\mathrm {H}_{8}-\mathrm {O}_{7})}\) distance is slightly larger than any other r(H−O) bond for this compound. This behavior was also observed in previous studies for other molecules containing a hydroxyl neighboring the carbonyl moiety [39, 40]. In phlorizin, the presence of the galactose causes an enlargement in the \(r_{(\mathrm {C}_{2^{\prime }}-\mathrm {O}_{7})}\) distance and a small change in the angle \(\theta _{(C_{1^{\prime }}-C_{2^{\prime }}-O_{7})}\). In addition, the torsion angles \(\theta _{(C_{\alpha }-C_{\beta }-C_{1}-C_{2})}\), \(\theta _{(C_{\alpha }-C_{\beta ^{\prime }}-C_{1^{\prime }}-C_{2^{\prime }})}\), and \(\theta _{(C_{\alpha }-C_{\beta }-C_{1^{\prime }}-C_{6^{\prime }})}\) experience major changes due to the sugar moiety, which indicates the loss of coplanarity between ring A and the keto group. The rotation around \(\theta _{(C_{\alpha }-C_{\beta ^{\prime }}-C_{1^{\prime }}-C_{2^{\prime }})}\) can have a significant impact on the antioxidant activity of phlorizin.

Determination of BDEs and IPs

Tables 2 and 3 show the BDEs computed at the B3LYP/6- 311 + G(d,p), M06-2X/6-311 + G(d,p), and LC-ω PBE/6-311 + G(d,p) levels of theory in the gas-phase, n-hexane, ethanol, methanol, and water for phloretin and phlorizin, respectively. Corresponding DFT results with the 6-311G(d,p) basis set are provided in the Supplementary Material. MP2/6-31G(d,p)//B3LYP/6-311 + G(d,p) results in the gas-phase and water are included for comparison; however, as has been noted previously [69]. MP2 overbinds when compared to DFT and experimental results. In general, polar solvents (such as ethanol, methanol, and water) are the most important ones for experimentally testing antioxidant activity. However, in a previous study by our group [40], we computed BDEs in n-hexane and, thus, this solvent was included in the present work to compare the behavior for phloretin and phlorizin. For phloretin, results from Yan et al. [51] determined at the B3LYP/6-311G(d,p) level of theory in the gas-phase and water and results from Kozlowski et al. [50] computed at the B3P86/6-31 + G(d,p) level of theory in the gas-phase, methanol, and water are also included in Table 2 for comparison purposes.

Table 2 Bond dissociation enthalpies (BDEs) (in kcal/mol) for different OH groups of phloretin
Full size table
Table 3 Bond dissociation enthalpies (BDEs) (in kcal/mol) for different OH groups of phlorizin
Full size table

A very small basis set dependence (0.01-1 kcal/mol) can be observed for the BDEs computed in the gas-phase. In addition, most of the BDEs obtained exhibit values where 6-311 + G(d,p)< 6-311G(d,p) using B3LYP, M06-2X and LC-ω PBE for both phloretin and phlorizin.

In both compounds, the computationally determined BDEs using DFT (in gas phase) for the majority of the OH groups have values lower than 91.98 kcal/mol, that is the BDE determined previously [70] for gallic acid at B3LYP/6-31G(d,p) in the gas phase. Additionally, the value of 90.93 kcal/mol for gallic acid (in water) is also higher than most of BDEs computed for phloretin and phlorizin in the same solvent environment. These observations suggest that both compounds studied in this work are promising in presenting antioxidant activity, given that gallic acid is one of the standards in experimental tests of antioxidant activity [71,72,73] and present a common structural moiety with phloretin and phlorizin.

For both phloretin and phlorizin (for all the functionals and basis sets and in all the solvents), the relative BDEs for the OH groups are 4-ArOH < 6′-ArOH < 4′-ArOH < 2′-ArOH. A slightly different trend (6′-ArOH < 4-ArOH) was observed in the results of Yan et al. [51] for phloretin; however, the difference between the two BDEs is less than 2 kcal/mol and Yan et al.’s values exhibit differences with the present BDE values at the same level of theory. This discrepancy is due to the use of a different conformer of phloretin by those authors as well as (perhaps) subtle differences in the computational methodology. Yan et al. [51] utilized the GAMESS [74] software package (hence different convergence thresholds and definition of B3LYP) and a different solvation model. Interestingly, the MP2/6-31G(d,p)//B3LYP/6-311 + G(d,p) results for the present conformer also show the BDE for 6′-ArOH to be lower than that for 4-ArOH but these results are strongly overbound compared to the DFT ones as has been observed previously [69]; hence, the primary discussions focus on the DFT-determined BDEs. On the other hand, BDEs computed by Kozlowski et al. [50] are comparable to the ones obtained in the present work. Our BDEs computed using the B3LYP functional are shifted by approximately 4 kcal/mol when compared to the ones determined using the B3P86 functional; in agreement to what is stated by Kozlowski et al. [50] with regard to relative energies determined with these two functionals. 2′-ArOH is well known for presenting the highest BDE due to the occurrence of the hydrogen bonding interaction between the H atom from hydroxyl and the O atom from the keto group already discussed in the previous section. The inclusion of the galactose causes the following effects on the values of BDEs: results obtained from the LC-ω PBE/6-311 + G(d,p) approach in the gas-phase indicate that the sugar moiety would cause a decrease of 1.22 kcal/mol in the 4-ArOH BDE which is in agreement with the behavior observed for kaempferol 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside and quercetin 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside in a recent study of our research group [40]. The decrease is also observed for computations taking into account the solvents n-hexane, ethanol, and methanol; however, in smaller fashions. The 4-ArOH BDEs determined at the B3LYP/6-311 + G(d,p) level of theory in methanol and n-hexane for phlorizin also present a decrease (of approximately 1.16 kcal/mol) when compared to the ones computed at the same level for phloretin. In the case of employing M06-2X/6-311 + G(d,p), the BDEs of the 4-ArOH are slightly higher for phlorizin than the ones for phloretin. The BDE values of 6′-ArOH and 4′-ArOH are changed (sometimes increasing a small amount and other times having a significant decrease) with the sugar inclusion. Interestingly, the presence of the galactose causes different effects on all the BDE values of 6′-ArOH and 4′-ArOH, respectively. For instance, at the M06-2X/6-311 + G(d,p) level of theory the BDEs of the 6′-ArOH barely increase while the BDEs of the 4′-ArOH decrease up to 5.75% in their values when the sugar moiety is present. This can be attributed to the rotation of the A ring (see torsion angles and \(\theta _{(C_{\alpha }-C_{\beta }-C_{1^{\prime }}-C_{6^{\prime }})}\)) that can lead to a sterically more (or less) crowded chemical site depending on the configuration established by the compound.

There are previous investigations reporting that the inclusion of a sugar moiety in a polyphenol increases the BDEs [36, 37]. In the previous study we performed on kaempferol 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside and quercetin 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside [40], which are glycosylated versions of kaempferol and quercetin, respectively, we observed a small decrease and/or increase in the BDEs (particularly for quercetin 3-O-α-L-arabinofuranosyl-(1→3)-α-L-rhamnoside) when comparing to the ones of the parent molecules. In the present study, as seen in Tables 2 and 3, the BDE values for phlorizin (4-ArOH) are comparable (sometimes slightly lower or slightly higher) than those computed for phloretin (6′-ArOH). Thus, the presence of the sugar substituent can lead to an increase (or decrease) in the antioxidant activity of phlorizin when in comparison to phloretin which is in agreement with our previous study [40].

The IPs for phloretin and phlorizin computed using B3LYP, M06-2X, and LC-ω PBE with the 6-311G(d,p) and 6-311 + G(d,p) basis sets in the gas-phase, n-hexane, ethanol, methanol, and water are presented in Tables 4 and 5, respectively. Again, MP2/6-31G(d,p)//B3LYP/6-311 + G(d,p) results are included. In general, it is possible to notice a considerable decrease in the IPs for both phloretin and phlorizin when in the presence of polar solvents (i.e., ethanol, methanol, and water) than for non-polar n-hexane. As reported in several previous studies [17,18,19], the surrounding environment has a strong impact on the IP values, and thus, the polar solvent facilitates the SET mechanism (related to the IP values) by providing charge separation. The IPs of phlorizin are always lower than those of phloretin at B3LYP with 6-311-G(d,p) and 6-311 + G(d,p) basis sets, which is in agreement to what was observed in our recent work for other sugar substituted compounds [40]. Using M06-2X and LC-ω PBE, the pattern is not the same; in some solvents phlorizin has higher IPs (in others, lower IPs) than phloretin. In water, B3LYP predicts phloretin to have a lower IP than phlorizin, while M06-2X and LC-ω PBE both predict the reverse trend. Thus, regarding the SET mechanism the presence of the sugar moiety should have a minor influence, but the nature of the influence (positive or negative) cannot be unambiguously assigned from the present computations.

Table 4 Ionization Potentials (IPs) (in kcal/mol) for phloretin as determined with the B3LYP, M06-2X, and LC-ω PBE functionals
Full size table
Table 5 Ionization Potentials (IPs) (in kcal/mol) for phlorizin as determined with the B3LYP, M06-2X, and LC-ω PBE functionals
Full size table

Summary

The parent molecule phloretin and its glycosylated version phlorizin, see Fig. 1, have been examined computationally utilizing DFT with the B3LYP, M06-2X, and LC-ω PBE functionals and with both the 6-311G(d,p) and 6-311 + G(d,p) basis sets. The focus of the investigation has been on the structural and energetic parameters including both BDEs and IPs, which provide information on the potential antioxidant activities. As the values of BDEs are considerably lower than the ones probed for IPs (in the gas phase or in any given solvent environment), the HAT mechanism is preferred over the SET mechanism. The BDEs determined suggest both phloretin and phlorizin hold promise for exhibiting antioxidant activity, especially phlorizin. Interestingly, the present BDE results suggest that (depending of the functional chosen) the sugar moiety can lead to an increase, decrease, or no change in the antioxidant activity. Hence, further experimental tests are encouraged in order to mitigate the substituent effect on phloretin and test the best choice of functional to probe BDEs for DHCs.

Supplementary Information

Cartesian coordinates for the optimized geometries determined using the B3LYP, M06-2X, LC-ω PBE functionals and the 6-311 + G(d,p) basis set in the gas-phase, n-hexane, ethanol, methanol, and water, can be found in the Supplementary Materials.