Dihydroisocoumarins, Naphthalenes, and Further Polyketides from Aloe vera and A. plicatilis: Isolation, Identification and Their 5-LOX/COX-1 Inhibiting Potency

The present study aims at the isolation and identification of diverse phenolic polyketides from Aloe vera (L.) Burm.f. and Aloe plicatilis (L.) Miller and includes their 5-LOX/COX-1 inhibiting potency. After initial Sephadex-LH20 gel filtration and combined silica gel 60- and RP18-CC, three dihydroisocoumarins (nonaketides), four 5-methyl-8-C-glucosylchromones (heptaketides) from A. vera, and two hexaketide-naphthalenes from A. plicatilis have been isolated by means of HSCCC. The structures of all polyketides were elucidated by ESI-MS and 2D 1H/13C-NMR (HMQC, HMBC) techniques. The analytical/preparative separation of 3R-feralolide, 3′-O-β-d-glucopyranosyl- and the new 6-O-β-d-glucopyranosyl-3R-feralolide into their respective positional isomers are described here for the first time, including the assignment of the 3R-configuration in all feralolides by comparative CD spectroscopy. The chromones 7-O-methyl-aloesin and 7-O-methyl-aloeresin A were isolated for the first time from A. vera, together with the previously described aloesin (syn. aloeresin B) and aloeresin D. Furthermore, the new 5,6,7,8-tetrahydro-1-O-β-d-glucopyranosyl- 3,6R-dihydroxy-8R-methylnaphtalene was isolated from A. plicatilis, together with the known plicataloside. Subsequently, biological-pharmacological screening was performed to identify Aloe polyketides with anti-inflammatory potential in vitro. In addition to the above constituents, the anthranoids (octaketides) aloe emodin, aloin, 6′-(E)-p-coumaroyl-aloin A and B, and 6′-(E)-p-coumaroyl-7-hydroxy-8-O-methyl-aloin A and B were tested. In the COX-1 examination, only feralolide (10 µM) inhibited the formation of MDA by 24%, whereas the other polyketides did not display any inhibition at all. In the 5-LOX-test, all aloin-type anthranoids (10 µM) inhibited the formation of LTB4 by about 25–41%. Aloesin also displayed 10% inhibition at 10 µM in this in vitro setup, while the other chromones and naphthalenes did not display any activity. The present study, therefore, demonstrates the importance of low molecular phenolic polyketides for the known overall anti-inflammatory activity of Aloe vera preparations.


Phytochemical Investigations on A. vera
The three dihydroisocoumarins, isolated for the first time from Aloe vera (L.) Burm.f., were identified by the following 1 H/ 13 C-NMR and ESI-MS results.
3R-feralolide; (3R-(2 -acetyl-5 -hydroxyphenyl)methyl-3,4-dihydro-8-hydroxy-2(1H)benzopyran-1-on) (  13  3  These experimental data are in good accordance with those published for the first isolation of feralolide (1) from A. ferox [12] and could be further validated by additional NOE measurements. The structure of 3 -O-β-D-glucopyranosyl-feralolide (2) was identified by the above given NMR data, in comparison with the literature [13] and additional NOE measurements, whereby the attachment of the glucosyl moiety at C3 -OH could be verified for the first time in the present study. The structure of the new 6-O-β-D-glucopyranosylferalolide (3) was identified by the above-given NMR, including HMBC data and additional NOE measurements. In contrast to 3 -O-β-D-glucopyranosyl-feralolide (2), the attachment of the glucosyl moiety at C6-OH was determined by the corresponding HMBC and NOE data (see Figures 3 and 4). All feralolides (1, 2, and 3) show identical negative and positive signs in the comparison of their CD spectra, thus enabling a clear identification of their 3R configuration ( Figure 4). In terms of quantity, aglycone 1 is the major feralolide in the A. vera drugs, while glucosides 3 and 2 are present in a ratio of 2.5:1, as shown by HPLC separation ( Figure 5). These experimental data are in good accordance with those published for the first isolation of feralolide (1) from A. ferox [12] and could be further validated by additional NOE measurements. The structure of 3′-O-β-D-glucopyranosyl-feralolide (2) was identified by the above given NMR data, in comparison with the literature [13] and additional NOE measurements, whereby the attachment of the glucosyl moiety at C3'-OH could be verified for the first time in the present study. The structure of the new 6-O-β-D-glucopyranosyl-feralolide (3) was identified by the above-given NMR, including HMBC data and additional NOE measurements. In contrast to 3′-O-β-D-glucopyranosyl-feralolide (2), the attachment of the glucosyl moiety at C6-OH was determined by the corresponding HMBC and NOE data (see Figures 3 and 4). All feralolides (1, 2, and 3) show identical negative and positive signs in the comparison of their CD spectra, thus enabling a clear identification of their 3R configuration ( Figure 4). In terms of quantity, aglycone 1 is the major feralolide in the A. vera drugs, while glucosides 3 and 2 are present in a ratio of 2.5:1, as shown by HPLC separation ( Figure  5).     In detail, the chromatographic and spectroscopic data of (2) and (3) disp high degree of similarity. The only significant difference between these two subst the cross peak between the carbon atom C6 and the anomeric H1″ in the spectrum in the latter compound, which proves the bond of the glucose moiety to carbon. The protons at C5 and C7 exhibit cross peaks with the carbons C6 and C respectively. The protons at C4′ and C6′ display cross peaks with the carbons C3 and C5′, respectively ( Figure 3). Further cross peaks observed in the HMBC sp are those between the protons at C5 and C6′ and the carbons C4 and C9, respecti well as between the proton C3 and the carbons C4 & C9 ( Figure 3). The me groups of C4 and C9 display further cross peaks with the carbons C1′, C1, C4a (b C4 and C5), and C6. Finally, three further cross peaks appear between H3 a carbon atoms C1′, C1, and C4a (  In detail, the chromatographic and spectroscopic data of (2) and (3) displayed a high degree of similarity. The only significant difference between these two substances is the cross peak between the carbon atom C6 and the anomeric H1" in the HMBC spectrum in the latter compound, which proves the bond of the glucose moiety to the C6 carbon. The protons at C5 and C7 exhibit cross peaks with the carbons C6 and C6 & C8, respectively. The protons at C4 and C6 display cross peaks with the carbons C3 & C5 and C5 , respectively ( Figure 3). Further cross peaks observed in the HMBC spectrum are those between the protons at C5 and C6 and the carbons C4 and C9, respectively, as well as between the proton C3 and the carbons C4 & C9 ( Figure 3). The methylene groups of C4 and C9 display further cross peaks with the carbons C1 , C1, C4a (between C4 and C5), and C6. Finally, three further cross peaks appear between H3 and the carbon atoms C1 , C1, and C4a ( Figure 3). Judging both from the ESI-MS ( Figure 2) and NMR data, the aglycone of 6-O-β-D-glucopyranosyl-feralolide (3) could clearly be identified as feralolide. However, in contrast to 3 -O-β-D-glucopyranosyl-feralolide (2), the glucose moiety was demonstrated to be attached to the 6-position as an O-glycoside, as strongly hinted by the low-field shift of about 2 ppm observed for the 13 C-NMR values at C3 of 3 -O-β-D-glucopyranosyl-feralolide and C6 of 6-O-β-D-glucopyranosyl-feralolide, respectively. As far as the stereochemistry at C3 and C9 is concerned, the measured CD data are clearly identical to those of feralolide ( Figure 4), proving all three molecules to exhibit an R-configuration. Therefore, (3R)-6-O-β-D-glucopyranosylferalolide (3) was firstly isolated from Aloe spp. as a new natural product in the study at hand. As far as 3 -O-β-D-glucopyranosyl-feralolide (2) is concerned, its steric structure at the C3 carbon had not been reported when it was first discovered in A. hildebrandii by [13], and has now been established as 3R, recently independently confirmed from flowers of A. arborescens [16]. Surprisingly, feralolide (1) was recently found in A. vera gel in two Yemeni Aloe spp. by LC-MS [25], confirming our previous assumption from TLC experiments [14,15] that dihydroisocoumarins are also gel constituents.

Biological Screening of 15 Polyketides on Their 5-LOX/COX-1 Inhibiting Potency
In the applied 5-LOX screening system [11] the aloin/anthraquinone derivatives (at a concentration of 10 µ M) displayed inhibition activities between 25 and 41% of LTB4

Biological Screening of 15 Polyketides on Their 5-LOX/COX-1 Inhibiting Potency
In the applied 5-LOX screening system [11] the aloin/anthraquinone derivatives (at a concentration of 10 µM) displayed inhibition activities between 25 and 41% of LTB 4 production. Among the tested chromones, only aloesin (4) displayed any inhibitory activity that was also limited to the LOX system. All tested naphthalene derivatives were inactive in both applied test systems. Among the tested substances, only feralolide (1) (at a concentration of 10 µM) achieved a 24% reduction of MDA production in the COX-1 test. All other tested substances were inactive in this test system. Feralolide did not, however, show any effect on the production of LTB 4 . Even if the glycosylated feralolide derivatives were inactive in the applied test system, they may very well be active in the human body, as the function of glycosides as inactive prodrugs that are transformed into active components by the human metabolism after ingestion is a well-documented fact in pharmacokinetics [30]. The anthraquinones emodin and aloe emodin (10) have already been identified as lipoxygenase inhibitors [31]. In particular, the individual findings of the anti-inflammatory activity of the chromones are obvious, especially under consideration of the described COX-2 inhibition by aloesin (4) itself [22], in contrast to the COX-1 inhibition observed here (Figure 9). Possibly, a desired dual inhibition of COX-2 and 5-LOX may be a new strategy to provide safer non-steroidal anti-inflammatory drugs. Consequently, equivalent studies should be carried out for many previously reported anti-inflammatory chromones, including free and esterified aloesins [18][19][20][21]32].
The documented reduction of MDA production in the COX-1 test by feralolide constitutes the first attribution of biological activity to a dihydroisocoumarin of an Aloe species.
The presented proof of its anti-inflammatory activity is of high relevance in the context of the typical application of A. vera in wound healing and a variety of inflammatory diseases, all of which have been validated by successful clinical studies [42][43][44][45] and even meta-analysis [46]. Feralolide might be of especial interest in these dermatological indications, as similar dihydroisocoumarins have been identified as active constituents of Japanese amacha (meaning "sweet tea") (Hydrangea macrophylla Seringe var. thunbergii Makino) [36,37] and the Brazilian Xyris pterygoblephara Steud. [40], both of which are traditionally used against skin inflammations. Although the Aloe spp. dihydroisocoumarins were isolated from the laxative aloe resin drug in the present research project, there is no reason to assume that the presence of these compounds in the living plants should be limited to the resin-storing vessels as described above. They may therefore very well be of high interest in the context of the numerous dermatological applications of the Aloe spp. gel drugs. Furthermore, anticancer activities reported from Brazilian traditional medicine for dihydroisocoumarins in the literature [38,39] might also be of interest for the feralolide derivatives of Aloe spp., as the use of whole leaf macerates of Aloe arborescens Mill. as an anticancer drug is one of its most famous applications in this ethnopharmacological tradition [47]. However, how much dihydroisocoumarins might really contribute to this effect has to be clarified in further, independent experiments.

Isolation of Pure Dihydroisocoumarins (1, 2, and 3)
Aloe barbadensis (syn. A. vera) drug (Ph.Eur.) (Lot No. Tot.J/W27.02.84/038) was obtained from Müggenburg Pflanzliche Rohstoffe (Bad Bramstedt, Germany). For the analysis, 100 g of the powdered drug were extracted with 2 l of a mixture of EtOAc/H 2 O (9:1, v/v) under constant shaking for 17 h. After filtration through a frit for removing insoluble components, the organic and aqueous phases were separated. Afterward, the organic phase was extracted again, first with 100 mL of water, then with 100 mL of a saturated aqueous NaCl solution, and then once again with 100 mL of water. Finally, the organic phase was dried over Na 2 SO 4 and subsequently evaporated to dryness under reduced pressure, resulting in 22.8 g of dry extract. A part of the organic fraction (20 g) was adsorbed to 80 g of silica gel (40-63 µm) and subjected to normal phase silica gel CC

TLC Examination
All extract fractions in the above-described isolation protocol were controlled for the presence of desirable substances using silica gel 60 F 254 TCL plates (Merck, Darm-  During the catalysis of PGH 2 (prostaglandin H2) into TXA 2 (thromboxane A2) by the thromboxane synthase system, malondialdehyde (MDA) is produced at a ratio of 1:1 with 12-HHT (12(S)-hydroxyheptadeca-5Z, 8E, 10E-trienoic acid). MDA can be detected photometrically after aldol condensation with thiobutyric acid [48]. For the in vitro screening, 1.5 l of pig blood was collected in a plastic bucket that already contained 150 mL of sodium-EDTA solution (0.077 M), immediately after sacrificing the animal. The blood was diluted with 700 mL of isotonic PBS (phosphate-buffered saline) solution and centrifuged for 20 min at 200× g. The supernatant, which is rich in thrombocytes, was carefully collected using a syringe and centrifuged once again for 20 min at 200× g to remove all remaining erythrocytes. In the subsequent centrifugation step at 1000× g for 15 min, the thrombocytes sediment and can finally be re-suspended in isotonic PBS at a defined cell density of 10 9 cells/mL. All isolation steps can be performed in plastic vessels at room temperature. For the test, 700 µL of this thrombocyte suspension was placed in 1.5 mL Eppendorf vessels (Eppendorf AG, Hamburg, Germany), together with 10 µL of the respective inhibitor solution, and incubated for 10 min at 37 • C. The enzyme reaction was started by adding 50 µL of calcium ionophore buffer (final concentration: 5 µM) and terminated after 10 min at 37 • C by adding 400 µL of trichloroacetic acid. All samples were centrifuged at 4000× g for 15 min, after which the supernatant was used for determining the MDA concentration. A negative control was performed in four repetitions, in which only 10 µL of DMSO solvent were added to the reaction mixture instead of the respective isolated compounds. For the positive control (n = 2), 700 µL of thrombocyte suspension was mixed with 10 µL of solvent and incubated at 37 • C. After 10 min, 400 µL of trichloroacetic acid and subsequently 50 µL of calcium ionophore buffer were added to the mixture. The subsequent steps were identical to those described for the extracted Aloe constituents, as described above. For determining the MDA concentration, 0.5 mL of the abovementioned supernatant was mixed with 0.5 mL of thiobarbituric acid solution. This mixture was incubated in a water bath for 30 min at 70 • C. After another 30 min of incubation at room temperature, the amount of MDA was measured in a spectral fluorometer (λ-excitation: 533 nm, λ-emission: 550 nm).

In Vitro Screening for Anti-Inflammatory Activity by Inhibiting 5-LOX (LTB 4 Assay)
For the LTB 4 (Leukotriene B4) test, the products of the arachidonic acid production can be directly measured via HPLC after the enzyme reaction in the thrombocytes and the subsequent activation reaction have taken place. The produced amounts of LTB 4 and 5-HETE (5-Hydroxyeicosatetraenoic acid) were directly detected at 270 nm, revealing the inhibitory potential of the tested substance (inhibition by zileuton as positive control). For the screening, 1.0 L of cow blood was collected in a plastic bucket that already contained 100 mL of sodium-EDTA solution (0.077 M) as an anticoagulant, immediately after sacrificing the animal. The blood was centrifuged at 200× g for 20 min, after which the thrombocytes containing the supernatant were separated. In order to facilitate the lyses of the erythrocytes, the remaining pellet was re-suspended in 800 mL of water. After 30 s of incubation, 400 mL of hypertonic PBS solution was added, followed by another centrifugation step at 485× g for 10 min. The resulting cell pellet was re-suspended in 50 mL of isotonic PBS solution and transferred to 20 mL of Histo-Paque (Sigma Aldrich, Taufkirchen, Germany). After yet another centrifugation for 45 min at 675 g, lymphocytes and macrophages that did not sediment in this setup were removed, while the granulocyte pellet was re-suspended in 10 mL of isotonic PBS solution. After an additional washing step in isotonic PBS solution, the granulocytes were re-suspended in O 2 -saturated, isotonic PBS solution, which had been continuously stirred on a magnetic stirrer at maximum speed for 5 min, at an exact cell density of 6 × 10 7 cells per ml. Although this granulocyte isolation protocol was performed at room temperature, the light exposure of the cells had to be limited as much as possible. For the enzyme reaction, a glass centrifuge tube was filled with 2.5 µL of the Aloe constituent test solution, or 2.5 µL of DMSO for the kinetic and control measurements. Subsequently, the tubes were filled with 0.8 mL of the above-described granulocyte suspension and incubated for 5 min at 37 • C in the water bath under constant shaking. Thereafter, 0.2 mL of calcium chloride was added, followed by an additional 5 min of incubation at 37 • C. In the next step, leukotriene production was induced by the addition of 2.5 µL of calcium ionophore buffer to the reaction mixture. After an additional 5 min (this variable was varied, respectively, in the case of the kinetic measurements) of incubation at 37 • C, the enzyme reaction was terminated by adding 1 mL of a solution of acetonitrile/methanol (1:1) that contains NDGA (nordihydroguaiaretic acid) as an antioxidant and PGB 2 (prostaglandin B2) as an internal standard. All glass centrifuge tubes were put into an ice-water bath immediately after termination of the reaction, sealed, and incubated in the ice-water bath for 20 min. Thereafter, centrifugation was performed for 15 min at 4000 × g under constant cooling at 4 • C. The supernatants were filled into vials, which were flanged and stored at −20 • C for further preparation. Each incubation setup contained 4 control and 8 sample measurements. Each defrosted sample was diluted with 10 mL of water and applied to an RP octadecyl extraction column (Macherey-Nagel, Düren, Germany), which had previously been washed successively with 10 mL MeOH, 5 mL water, and 5 mL of a 0.1% EDTA solution. After washing the column two times with 5 mL of water, the adsorbed substances were eluted with 3 mL of MeOH. This eluate was diluted with 3 mL of water. Subsequently, the concentration of LTB 4 was directly measured by HPLC using a Nucleosil 7 µm C18 column (250 mm × 4.6 mm) (Macherey-Nagel, Düren, Germany) with tetrahydrofuran/MeOH/aqueous EDTH solution (0.1%)/acetic acid (25:30:45:0.1; v/v), pH 5.5 (conc. NH 4 aq ) as a mobile phase, at a flow rate of 0.9 mL/min. UV detection took place at 270 nm. For measuring the 5-HETE concentration, only the mobile phase was changed to MeOH/H 2 O/acetic acid (77:23:0.1; v/v), pH 5.5 (conc. NH 4 aq ) with a flow rate of 1.0 mL/min. UV detection took place at 232 nm. The inhibition of the 5-lipoxygenase systems was then calculated from the measured amounts of LTB 4 and 5-HETE. The relative concentration of LTB 4 /5-HETE in each sample equals the ratio of the peak areas of LTB 4 /5-HETE, relative to the internal standard PGB 2 . The inhibition of the 5-lipoxygenase system, therefore, equals the ratio between the LTB 4 /5-HETE value in the presence and in the absence of the respective Aloe constituent.