3.1. Derivatives and Properties
By inserting -OH, -NH
2, -SH and -COOH groups in sites R
1 to R
5, 185 new FA derivatives were built (
Table S1). Twenty of them with one functional group, 160 with any possible combination of two functional groups, and 5 with three functional groups. The latter were constructed from the most promising bi-functionalized species.
A selection score (S
S) was computed. It is meant to identify the FA derivatives with the most likely drug-like behavior and corresponds to that included in the CADMA-Chem protocol.[81-83, 306-308] The associated equations are provided in
Table S6. The higher the value of S
S the more likely the drug-like behavior. S
S takes into account eight ADME properties: water/octanol partition coefficient (logP), topological polar surface area (PSA), number of heavy atoms (
XAt), molecular weight (MW), number of H-bond acceptors (HB
A), number of H-bond donors (HB
D), rotatable bonds (RB), and molar refractivity (
MR); two toxicity descriptors: oral rat 50 percent lethal dose (LD
50) and Ames’ mutagenicity (M); and the synthetic accessibility (SA).
The S
S for all the designed FA derivatives is presented in
Figure 1. The parent molecule and the average S
S value for the reference set are included for comparison purposes. The individual values of all the FA derivatives are reported in
Table S2, together with those of the above-mentioned descriptors. Higher values of S
S suggest better drug-like behavior, lower toxicity, and easier synthesis. The first screening was based on this score, and twelve FA derivatives were selected. However, before moving them forward to the next stage of the investigation (
Scheme 2) a double-check analysis was performed using exclusion scores (S
E), which allows verifying if any of the selected molecules significantly deviates (in any of its properties) from the average value of the reference set.
Four exclusion scores were analyzed (S
E,ADME2, S
E,ADME8, S
E,ADMET and S
E,ADMETSA). Their equations are provided
Table S7. S
E,ADME8, S
E,ADMET and S
E,ADMETSA are extensions of the well-known S
E,ADME2, based on two descriptors (logP and MW).[309, 310] S
E,ADME8 uses the same kind of strategy as S
E,ADME2, but includes six additional terms (PSA, XAt, HB
A, HB
D, RB, and
MR). S
E,ADMET and S
E,ADMETSA also include toxicity (LD
50 and M) and synthetic accessibility (SA) descriptors.
S
E,ADME2 values were previously estimated to be between 1.2 and 1.5 for 1791, 152 and 35 oral drugs.[309, 310] For the 12 selected FA derivatives the average S
E,ADME2 value was found to be 1.06, with individual values ranging from 0.56 to 1.42 (
Table S4). The estimated average values for the other elimination scores were found to be: S
E,ADME8 = 4.86 (ranging from 2.59 to 7.34). S
E,ADMET = 8.13 ( ranging from 4.53 to 11.54), and S
E,ADMETSA = 9.59 (ranging from 5.50 to 12.52). It seems worthwhile mentioning that high values of the exclusion scores might result from either worse or better behavior than the average of the reference drugs. Thus, a detailed analysis is required to determine if any particular candidate should be removed from the selection or not.
According to the gathered results (
Figure 2), toxicity is responsible for the largest deviation. Regarding ADME, the six additional descriptors lead to largest deviation than logP and MW. Synthetic accessibility also has rather small influence on the deviations from the reference molecules. A more detailed examination, considering the individual contribution of all the investigated descriptors is presented in
Figure 3.
The more important deviations arise from LD50, followed by M, PSA and HBD. the FA derivatives with the largest LD50 deviations from the reference set are FA, FA-173, FA-175 and FA-26. However, they correspond to a lower toxicity to rats than the reference average (LD50 = 960.8), with values 4742.7, 4471.9, 4040.7, and 3635.2, respectively. Regarding Ames mutagenicity, a similar trend was found. The FA derivatives predicted as the least mutagenic are just those that deviate the most from the reference set average (M = 0.41). They are FA-88, FA-106, FA-115 and FA-142, all with M = 0.01. Thus, these deviations imply that the above-mentioned derivatives have a more desirable behavior than that of the reference set. Accordingly, they were not excluded from the chosen subset.
The largest PSA deviation were found for FA-41, FA-26, FA-88 and FA-173 (124.3, 113.0, 104.1 and 104.1 Å2, respectively). However, their PSA values are all below the Veber’s limit: 140 Å2. Thus, these derivatives were also kept in the chosen subset. The largest deviations for HBD correspond to FA-26 with HBD = 5 and FA-8, FA-41 and FA-138 with HBD = 4. Since they do not represent violations of the Lipinski’s rule, these candidates were not eliminated.
After carefully examining elimination scores for the 12 FA derivatives with the highest SS values, none of them were excluded from the selection. Thus, they move forward to the next stage, which is the investigation of the antioxidant capacity through electron and H-atom donation. This detailed analysis is important since it allows interpreting deviations for all the used descriptors, and prevents from excluding suitable candidates for no good reason.
3.2. pKa and Antioxidant Activity
As previously mentioned, acid-base equilibria are crucial for medical drugs intended to passively cross biological barriers. The
pKa values and molar fractions (
Mf), at physiological
pH, were estimated for the 12 FA derivatives chosen in the first stage of the investigation as those with the best drug-like behavior (
Table 2). Additionally, the corresponding deprotonation routes and distribution diagrams are provided in
Figures S2 and S3.
The calculated molar fractions (
Table 2) revealed that 7 of the 12 derivatives, selected based on the S
S value, would have negligible population (<10
-4) at physiological pH, i.e., at pH=7.4. Thus, they were excluded as viable candidates. Albeit the
Mf(0) for other 3 (FA-26, FA-118, and FA-175) are rather small, they are very similar to that of FA. Since there is abundant data on the biological activities of FA (
Table 1), it can be inferred that such fractions are enough. Consequently, 5 derivatives (FA-8, FA-26, FA-118, FA-138, and FA-175) were further investigated. Among the studied derivatives, FA-138 is the only one that is predicted to have similar fractions of neutral (q= 0) and anionic (q= -1) species. This feature might be relevant to its possible use as multifunctional antioxidant. The rather large neutral fraction (59.0%) is expected to promote passive crossing through biological membranes, while the anionic fraction (40.3%) is likely to be the key one for the free radical scavenging activity as it is the case for many phenolic compounds.
The ionization energies (IE), electron affinities (EA), and the lowest bond dissociation energies (BDE) for the acid-base species with non-negligible population (
Mf(q) ≥ 10
-4) of FA and its derivatives, at
pH=7.4, are reported in
Table 3. The complete set of BDEs, i.e., considering all viable H-donating sites, is provided as
Supplementary Materials (
Table S8). IE and BDE reactivity indexes are related to the viability of electron and H atom donation. Thus, they were used to compare the efficiency of the derivatives with that of reference antioxidants, as free radical scavengers via single electron transfer (SET) and formal hydrogen atom transfer (HAT) mechanisms, respectively.
IE and BDE values were used to build the electron and hydrogen donating ability map for antioxidants (eH-DAMA,
Figure 4). This graphical tool has been recently proposed to simultaneously accounting for the likeliness of molecules as H donors (formal HAT reaction route) and electron donors (SET reaction route).
92,93 The dominant acid-base species of the investigated FA derivatives, at physiological pH, were included in this map; as well as two antioxidant references (Trolox and α-tocopherol), the parent molecule, and the H
2O
2/O
2•−. This pair represents the potential oxidant target. The best radical scavengers are expected to be located at the bottom left, i.e., lower IE and lower BDE. The species in this region are likely to simultaneously act as electron and H-atom donors.
Based on the eH-DAMA (
Figure 4) it is predicted that the five FA derivatives included in it should be efficient for scavenging peroxyl radicals through both mechanisms, SET and
f-HAT. Their efficient for that purpose is expected to surpass that of α-tocopherol and ferulic acid. On the contrary, only the anionic form of FA-138 is predicted to be more efficient than Trolox for that purpose. FA-8 may be better electron donor than Trolox, but not as good for donating H-atoms. However, further investigation dealing with other aspects of antioxidant activity, kinetics in particular, are still needed to confirm of refute the foreseen trends.
3.3. Polygenic Activity
The possible neuroprotection activity of the ferulic acid derivatives was investigated using molecular docking studies. Only the compounds with a significant molar fraction of the neutral specie, at physiological pH, were studied. To evaluate the general neuroprotection activity, a polygenic score (S
P) was developed. S
P is a measure of the tested compounds capacity to bind to the enzymes compared with natural substrates (COMT: dopamine (dopa), MAO-B phenylethylamine (pea) and AChE: acetylcholine (ACh). It was defined according to our previously reports [81, 82] as:
The scoring values are presented in the
Table 4. When the values of S
P are examined, it can be predicted that the compounds exhibit neuroprotection activity since their scores are higher than the those of the corresponding natural substrates (S
P= 3.00), i.e., the investigated ferulic acid derivatives may present stronger affinities towards the enzymes. Among the studied compounds, the FA-26 analog is expected to have the best neuroprotection activity. Interestingly, according with the docking results, the parent molecule (ferulic acid) is also likely act as a neuroprotector.
The examination of individual DG
BW values reveals that the studied compounds could be better inhibitors for AChE and MAO-B than they are for the COMT enzyme. Negative values of COMT (blue fragment of the bars, in
Figure 5) indicates that this enzyme forms more stable complexes with dopamine than with the tested FA derivatives. Only FA-118 shows a slightly higher score than dopamine. Interestingly, this compound has a catechol moiety, which is recognized to exhibit effective COMT inhibition potential.[
311] FA-175 presents almost the same score as dopamine (log DG
WB/DG
B,
sub= - 0.001). On the other hand, for MAO-B and AChE (green and red fragments, respectively,
Figure 5), the neuroprotection behavior of FA derivatives was evidenced by their positive values. Between these two enzymes, the inhibitor potential of the studied derivatives is expected to be stronger for AChE, which suggests an improved activity against acetylcholine degradation. The complete set of binding energies can be founded in
Table S9,
Supplementary Materials.
The main interactions for the complexes with the highest S
P are shown in
Figure 6. They are FA-26 with AChE (left), FA-26 with MAO-B (middle), and FA-118 with COMT (right). For all of them FA-26 is in its anionic form, which is the most abundant species at physiological pH (X~0.97). To understand the interactions formed in the protein-ligand complexes, it is important to know the architecture of the enzymes and the function of the key residues. AChE has a highly specialized structure, which allows it to be one of the fastest known enzymes. The catalytic triad (H447, E334 and S203) is founded at the bottom of the enzyme and surrounded by 14 well conserved aromatic residue.[
312] Among them, W83 plays an essential role since it forms a substrate union site, while Y70, Y121, and W279 conform the anionic peripheric site.[
312] Additionally, AChE has a high dipole moment with the axis oriented towards the substrate entry site. It has been suggested that this moment may serve to pull down the cationic substrate of AChE. This dipole is controlled mainly by residues D71, E199, and E443.[
313]
FA-26 has several H-bond donors and acceptors, and an aromatic ring that contribute to generate intermolecular connections with the AChE key amino acids. In fact, complex FA-26:AChE is formed by several interactions, mainly hydrogen bonds and p-interactions. This derivative is bonded to the active site of AChE through four hydrogen bonds (D71, Y121, F292, and Y338), one p-stacking interaction (Y334) and one p-alkyl interaction (W83). The observed interactions suggest that FA-26, although not bonded to the catalytic triad, can inhibit the ACh degradation, blocking the entry and union sites.
MAO-B function involves two hydrophobic pockets, an entry pocket and an active site pocket with I199 acting as a gatekeeper between two cavities. The catalytic reaction site comprises a redox cofactor, flavin adenosin dinuleotide (FAD). The active site is completed by residues Y398 and Y435, orienting the substrate to the proper position.[
314]
Four H-bonds involving Q206, L171, and FAD; a p-stacking (F343); and non-conventional C-H bonds stabilize the complex formation. An important feature of the conformation adopted by FA-26 in the complex is the formation of an H-bond with N5 in the FAD moiety. This atom is required for the redox activity of the cofactor[
314] and, hence, for the catalytic function of the enzyme. This conformation could not be achieved without the orientation promoted by the L171 and Y398 residues, which suggests that FA-26 could inhibit some enzymes with the same mechanism of action as MAO (type A) or other flavoenzymes as lactate oxidase.[
315] According to these findings, FA-26 is predicted to act as a reversible or non-covalent MAO-B inhibitor as Safinamide o Moclobemide,[299, 316], which are recognized antidepressant drugs. This way of inhibition is preferable since it has been proven to be associated with less toxicity than others.[
317]
COMT is a selective enzyme that catalyzes the transfer of methyl groups to the 3-OH position of catecholamines. COMT is a Mg-dependent enzyme, with the metal bound to D141, D169, and N170 residues. This enzyme uses the Mg atom to bind the substrate and make it more easily ionizable.[
318] The methyl group is transferred by S-Adenosylmethionine cofactor. The binding substrate site is completed with several hydrophobic residues M40, L198, W143, and the gatekeepers W38 and P174.[
318].
FA-118 has a catechol moiety that binds the Mg atom, according to the docking simulations by two metal-donor unions. A hard acid-base interaction (Mg-O) stabilizes the formation of this adduct. In addition, H-bonds between the catechol fragment and the residues K144 and N170 also contribute to the binding energy. Finally, several hydrophobic interactions with key residues of the active site (M40 and P174) complete the stabilization of the FA-118:COMT complex. Such an arrangement explains the good score obtained in the simulations and suggests that FA-118 can be efficient as a COMT inhibitor.
The docking simulations indicate that while all the investigated FA derivatives can act as neuroprotectors of acetylcholine and phenylethylamine (with FA-26 being predicted as the best one for that purpose), only FA-118 would be able to protect dopamine against COMT-induced degradation. Accordingly, FA-118 is proposed as a promising candidate in the context of Alzheimer’s and/or anti-anxiety disorders, while FA-26 was identified as the best candidate (among the studied molecule) for Parkinson´s. All of them certainly deserve further investigations related to their potential as neuroprotectors.