Brewers’ spent hop revalorization for the production of high added-value cosmetics ingredients with elastase inhibition capacity

To maximize the efficacy of the BSH extracts, it is necessary to identify the hop compounds with the highest activity and also to optimize the parameters influencing the production of these extracts so that the highest elastase inhibition capacity can be achieved. Thus, a three-step methodology was carried out, starting with a computational analysis to test the in-silico inhibition capacity of the proposed hop humulones ((iso-)α-acids), lupulones (β-acids) and prenylflavonoids¹³. Next, BSH extracts were prepared following a procedure rooted on a Box–Behnken factorial design to optimize the extraction agent and conditions, and then testing their anti-elastase capacity¹⁴. Last, to characterize the composition of these extracts, a HPLC–MS/MS analysis was performed to guarantee the presence of those molecules that were previously identified as potent inhibitors by the computational methodology.

Molecular modeling: in-silico identification of potent inhibitors

A blind docking analysis was employed to detect the region of interaction of the elastase receptor against the proposed α and β-acids and prenylflavonoids. The elastase binding site is well-known and is stablished by the key residues Hys57, Asp102 and Ser195¹⁵, so this analysis allows to detect whether those α and β-acids and prenylflavonoids are able to interact with the binding site or whether they remain attached to non-active areas of the receptor.

Accordingly, this study is carried out with the α-acids adhumulone, cohumulone and n-humulone, iso-α-acids cis-iso-α-humulone, cis-iso-α-cohumulone, trans-isocohumulone, trans-isohumulone, iso-α-cohumulone, iso-α-adhumulone, cis-tetrahydroisocohumulone, cis-tetrahydroisohumulone, trans-isocohumulone and trans-tetrahydroisohumulone, trans-tetrahydroisocohumulone, the β-acids lupulone and colupulone and the prenylflavonoids 6-prenylnaringenin, 8-prenylnaringenin, 6-geranylnaringenin, desmethylxanthohumol, xanthohumol and the isomerized form isoxanthohumol. Blind docking performs a conformational analysis of each ligand on all α-carbons of the studied protein. In this way, all possible ligand-receptor binding sites are evaluated, and a score is assigned to quantify the suitability of the binding. In other words, all binding positions and their corresponding binding energy are analyzed, establishing a quantitative ranking of the most favourable positions for each α and β-acid and prenylflavonoid in the elastase receptor¹⁶.

The interaction of the aforementioned ligands into the elastase binding site is reported in Table 1. The suitability of these positions is identified by a cluster number (CL), being CL1 the most favourable and CLn the most unfavourable (n = 1 – total number of dockings). Favourable bond energies were detected for every ligand inside the inspected binding site, and also the binding site was reached within first three dockings (n CL ≤ 3) so it can be considered that these ligands have a high potential to bind elastase.

These molecules have significant structural differences, so there is no common pattern in the way they interact with the receptor and, therefore, they are not always able to interact with the same key residues. Among them, cis- and trans-tetrahydroisohumulone, cis-iso-α-cohumulone and trans-isohumulone have the most favourable key residue binding energies, followed by 8-prenylnaringenin and lupulone. This high affinity seems to be caused by the hydroxy groups, which are correctly positioned to function as acceptor groups and bind the key residues (Figs. 1, Fig. si. 1–22).

Blind docking of elastase and (A) Cis-tetrahydroisocohumulone, (B) Cis-tetrahydroisohumulone. (A) Shows an incorrect positioning of cis-tetrahydroisocohumulone in the binding site, not being able to interact with the key residues His57, Asp102 nor Ser195, while (B) shows the interaction of cis-tetrahydroisohumulone, structurally similar to cis-tetrahydroisocohumulone, bound to the key residues His57 and Ser195.

The presence of α and β-acids and prenylflavonoids is determined by HPLC–MS/MS in different samples of hop and BSH extracts to determine and quantify those molecules which are present in a more pronounced way (Table 2). Isoxanthohumol, xanthohumol, 8-prenylnaringenin, iso-α-adhumulone and/or n-humulone, iso-α-cohumulone, trans and/or cis-tetrahydroisocohumulone, tetrahydroisohumulone and/or tetrahyroisoadhumulone, lupulone and colupulone were detected as major α and β-acids and prenylflavonoids from hops, being iso-α-adhumulone and/or n-humulone, iso-α-cohumulone and lupulone in higher concentration (Table 2). Several publications state that the most common form, at slightly acidic pH (pH = 5.8–7), is the cis-form¹⁷. Therefore, it is believed that cis- isomers have higher concentration in those cases where it is not possible to differentiate between these two molecules by mass spectrometry analysis methodologies.

If the BD results (Table 1) and concentrations of these molecules in hops are considered (Table 2), the most important would be trans-isohumulone, followed by cis-iso-α-cohumulone and lupulone. Thus, a molecular dynamics analysis is carried out with these molecules which offer the best BD results. Also, these other molecules which are in much higher concentration in hop samples were analyzed, because although they may not be the most active, the large difference in concentration may compensate this expected lower bioactivity. Accordingly, trans-isohumulone, cis-iso-α-cohumulone, lupulone, cis- and trans-tetrahydroisohumulone, and 8-prenylnaringenin were selected to test their performance in a dynamic environment, taking into consideration the bound protein residues and their stability within the binding site.

MD results show that trans-isohumulone (Fig. s.i.25) has a low RMSD value for elastase and higher but constant for ligand. This molecule interacts with the key residues Hys57 and Ser195, being the first one an intermittent and unstable contact. Ser195 and the non-key residues Arg61 and Gly193 are the most durable contacts over time, with an intermittent interaction from 10 to 50 ns.

Cis-iso-α-cohumulone (Fig. 2) shows low and constant RMSD values both for protein and ligand, especially from 15 ns onwards. It interacts slightly intermittently, but during the whole simulation period, with the key residues Hys57 and Ser195. Besides, the non-key contacts Val99, Gln92, Ser214 and Val216, which are maintained during the whole simulation help to fix the ligand in a high stable position. Ser214 and Val216 play a major role, so they are permanent contacts after the first 25 ns, and this is a key aspect to promote the interaction with the aforementioned key residues.

Molecular dynamics analysis of cis-iso-α-cohumulone and the standard elastase inhibitor epigallocatechin gallate. Final docking position, protein–ligand contacts, and RMSD values are represented.

Lupulone (Fig. s.i.27) has high and unstable RMSD values for ligand, especially at the end of the simulation period. Regarding the evolution, it can be assumed that this ligand turns from one position to another during the course of the simulation. No interaction was detected with the key residues, and, as it was assumed due to the high RMSD instability, the protein–ligand contacts show that there is an important conformational change of the ligand, which turns from a first interaction with Asn25, Tyr117 and Gln119 to an interaction with Trp27, Tyr137, Gln157 and Tyr207.

Trans-tetrahydroisohumulone (Fig. s.i.28) shows low and constant RMSD values for both protein and ligand throughout the simulation, especially from 8 ns onwards. It only interacts very intermittently and during the first 20 ns of simulation with the key residue His57, although this interaction is practically negligible. Regarding non-key residues, Val216, stable throughout the simulation, and Arg217, slightly more intermittent, stand out. It also shows more intermittent interactions, but throughout the simulation, with Val99, Gln192 and Phe215.

Cis-tetrahydroisohumulone (Fig. s.i.29) has low and constant RMSD for protein throughout the entire simulation. The ligand has a slightly elevated RMSD from 15 ns onwards, at which point a conformational change that modifies its binding to the receptor occurs. This ligand interacts highly intermittently and only in the first 20 ns of the simulation with the key residues His57 and Ser195, although these interactions are practically negligible. All ligand interactions are rather intermittent, except with the non-key residues Ala99, Trp172 and Phe215, whose interaction remains constant especially from 20 ns onwards.

8-prenylnaringenin (Fig. s.i.30) has low RMSD for both protein and ligand, especially from 20 ns onwards. It interacts intermittently throughout the simulation with the key residue His57 and more constantly with the key residue Ser195. It also has slightly intermittent interactions during the whole simulation with the non-key residues Gly193, Ser214, Val216 and Arg217, giving the ligand a high stability.

Two standard references as quercetin and epigallocatechin gallate (EGCG) were tested to contrast their performance against these molecules. Quercetin (Fig. s.i.26) shows slightly high RMSD values around 6.5 Å, and interacts intermittently but during the whole MD simulation with His57. It also interacts with Ser195, but only in certain moments. As non-key residues, quercetin interacts permanently with Val216 and Arg217, and intermittently but throughout the entire simulation with Thr96, Val99, Thr175 and Phe215. This high number of non-key contacts helps to stabilize the ligand and makes quercetin a potent elastase inhibitor.

Epigallocatechin gallate (Fig. 2) shows a really low RMSD value around 3.5 Å and interacts with the three key residues His57, Asp102 and Ser195. The most relevant contact is His57, which is intermittent but maintained during the whole simulation. Asp102 is maintained during the last 5 ns and Ser195 is also intermittent but maintained throughout the entire simulation, especially during the last 30 ns. As non-key residues, epigallocatechin gallate interacts permanently with Gly193 and Ser214, and also slightly more intermittently with Thr96, Asp97, Val216 and Arg217, making this molecule a high stable ligand and thus a strong elastase inhibitor.

According to these MD results, cis-iso-α-cohumulone and 8-prenylnaringenin are the most stable ligands (Figs. 2, Fig. s.i.30). Compared against the standard inhibitors, these two molecules show a similar MD behaviour to quercetin and epigallocatechin gallate (Figs. 2, Fig. s.i.26), so they are expected to act as elastase inhibitors. Cis-iso-α-cohumulone, which was detected as the most potent inhibitor by MD analysis (Fig. 2), is also present in high concentration in hop samples (Table 2), so probably this molecule would inhibit elastase in a more effective way than 8-prenylnaringenin due to this higher concentration.

Optimization and characterization of BSH extracts

To contrast these theoretical results, BSH extracts were prepared using propanediol-water mixtures (Table s.i.2) following a Box–Behnken factorial design (Fig. s.i.23) combined with the response surface methodology (RSM) which is explained on “Preparation of hop extracts” section¹⁴. Also, an experimental design using water as extraction solvent was performed (Table s.i.2). Results show that temperature and propanediol percentage improve the phenolic content, antioxidant activity and elastase inhibition (Fig. 3, Table s.i.1). Water extraction procedures, despite of being able to achieve high concentrated phenolic extracts with good antioxidant capacities, do not show anti-elastase activity, regardless of time or temperature used during the extraction process (Table s.i.2). In this case, epigallocatechin gallate was taken as a control standard. Between the two standard inhibitors computationally tested, quercetin and epigallocatechin gallate, the latter was chosen because it has a molecular structure of similar complexity to the bioactive compounds present in hops.

Phenolic content, antioxidant activity and elastase inhibition percentage evolution related to propanediol content, extraction time and extraction temperature.

The desirability function of the RSM was used to optimize the extraction variables on the process of obtaining bioactive extracts, seeking to maximize multiple responses (phenolic content and/or antioxidant activity and/or anti-elastase activity) (Fig. s.i.24). The optimal operating factors to maximize antioxidant capacity and anti-elastase activity were achieved at 74 °C, 79.92% propanediol and 51.45 min extraction. BSH extract 18 obtained under these conditions shows 100% elastase inhibition capacity, antioxidant capacity equivalent to 81.9 mmol Trolox/L and a high phenolic content (Table 3). These predicted values are close those obtained for BSH extract 14 under conditions that also maximize the phenolic content (Table 3).

Epigallocatechin gallate was taken as a control standard for anti-elastase activity assay, being the effective or inhibitory concentration that reduces the enzyme activity to 50% of the control EGCG (EC₅₀ value) 2.3 mg/L (experimental data, n = 3). On the other hand, BSH 14 reduces de activity of elastase to 100% at 4.7 mg GAE/L total phenolic concentration (predicted data). Between the two molecules computationally tested, quercetin and epigallocatechin gallate, the latter was chosen because it has a molecular structure of similar complexity to the bioactive compounds present in hops.

Then, to obtain a quantitative molecular characterization of BSH extracts a HPLC–MS/MS analysis was carried out following the 4.4.2 procedure. These results (Table 2) show that those propanediol-water extracts where cis-iso-α-cohumulone and 8-prenylnaringenin molecules are in higher concentration (BSH 7, 7′ and 12) have higher anti-elastase activity, meanwhile water extracts where there is lower concentration of cis-iso-α-cohumulone and no presence of 8-prenylnaringenin do not have anti-elastase potential (BSH 2W, 5W and 8W). This proves the strong inhibition capacity of these two molecules, which was also previously reported during the computational analysis procedure.