Original Papers

1657

Authors

Krasimir Rusanov 1, Eliane Garo 2, Mila Rusanova 1, Orlando Fertig 2, Matthias Hamburger 2, Ivan Atanassov 1, Veronika Butterweck 3

Affiliations

1 2 3

Key words " Rosa damascena l " Rosaceae l " rose oil distillation l " wastewater management l " adsorption resin l " polyphenols l

received revised accepted

July 11, 2014 Sept. 8, 2014 Sept. 12, 2014

Bibliography DOI http://dx.doi.org/ 10.1055/s-0034-1383145 Published online October 8, 2014 Planta Med 2014; 80: 1657–1664 © Georg Thieme Verlag KG Stuttgart · New York · ISSN 0032‑0943 Correspondence Prof. Dr. Veronika Butterweck Institute for Pharma Technology School of Life Sciences University of Applied Sciences Northwestern Switzerland Gründenstraße 40 CH-4132 Muttenz Switzerland Phone: + 41 6 14 67 46 89 Fax: + 41 6 14 67 47 01 [email protected]

AgroBioInstitute, Agriculture Academy, Sofia, Bulgaria Institute of Pharmaceutical Biology, Department of Pharmaceutical Sciences, University of Basel, Basel, Switzerland Institute for Pharma Technology, School of Life Sciences, University of Applied Sciences Northwestern Switzerland, Muttenz, Switzerland

Abstract !

The production of rose oil from rose flowers by water steam distillation leaves a water fraction of the distillate as main part of the waste. Therefore, the rose oil distillation wastewater represents a serious environmental problem due to the high content of polyphenols which are difficult to decompose and have to be considered as biopollutants when discarded into the drainage system and rivers. On the other hand, natural polyphenols are valuable compounds with useful properties as bioactive substances. Until now there is no established practice for processing of rose oil distillation wastewater and utilization of contained substances. Thus, it was the aim of this study to develop a strategy to separate this wastewater into a polyphenol depleted water fraction and a polyphenol enriched fraction which could be developed into innovative value-added products. In a first step, the phytochemical profile of rose oil distillation wastewater was determined. Its HPLC‑PDA‑MS analysis revealed the presence of flavan-3-ols, flavanones, flavonols and flavones. In a second step, the development of a stepwise

Introduction !

The oil bearing rose (Rosa damascena Mill. f. trigintipetala Dieck, Rosaceae) is a perennial plant native to Europe and Middle East countries [1, 2]. R. damascena is a characteristic and economically important plant for Bulgaria [1, 3], considering that the country produces about half of the worldʼs rose oil [3]. The estimated annual production of Bulgarian rose oil over the last 15 years ranged between 870 and 2000 kg, and the world annual consumption for rose oil accounted for 3000 to 4500 kg [3]. At present, rose plantations in Bulgaria cover an estimated total area of approximately 3500 ha [3]. Around 90% of the rose

concentration of rose oil distillation wastewater was performed. The concentration process includes a filtration process to eliminate suspended solids in the wastewater, followed by adsorption of the contained phenolic compounds onto adsorption resins (XAD and SP). Finally, desorption of the polyphenol fraction from the resin matrix was achieved using ethanol and/or aqueous ethanol. The result of the process was a wastewater low in soluble organic compounds and an enriched polyphenol fraction (RF20 SP-207). The profile of this fraction was similar to that of rose oil distillation wastewater and showed the presence of flavonols such as quercetin and kaempferol glycosides as major metabolites. These compounds were isolated from the enriched polyphenol fraction and their structures confirmed by NMR. In summary, a pilot medium scale system was developed using adsorption resins for the recovery of polyphenols from rose oil distillation wastewater suggesting an industrial scalability of the process. Supporting information available online at http://www.thieme-connect.de/products

flowers are currently processed to rose oil, and the rest is either used for production of concrete through solvent extraction or by the food industry for production of jams and liqueurs [3]. One of the challenges to the rose industry at present is the increase of the produced waste during the rose oil distillation process. More than 3000 kg of rose flowers yield 1 kg of rose oil and 1 kg of fresh raw material gives approximately 2 kg of residue on a wet weight basis [4]. One round of distillation uses 500 to 1000 kg of rose petals (e.g., 1000 kg petals are mixed with 4000 L water), and distillation takes at least 2 h [2]. During the distillation process the essential oil is extracted from the flowers. The chemical composi-

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Recovery of Polyphenols from Rose Oil Distillation Wastewater Using Adsorption Resins – A Pilot Study

Original Papers

Fig. 1 HPLC‑PDA-ESIMS chromatogram of rose oil distillation wastewater (A) and of resin fraction RF20 (B) after pilot medium scale up. Top A and B UV trace (200–700 nm). Bottom A and B ELSD trace. SunFire™ C18 column, A: 0.1 % aqueous formic acid, and B: MeCN. For step gradient, see text. Numbers 1–14 refer to the isolated compounds. Letters refer to compounds identified only by HPLC‑PDA‑MS (* refers to compounds identified in RODW only; # refers to compounds identified in RF20 only). Compounds 1 and 14 could be identified in the LC-ESIMS chromatogram only after their isolation, due to their weak ionization and peak overlap, respectively.

tion of rose oil is very complex, given that it contains more than 300 known compounds [5]. Upon completion of the distillation process, the hot residue from the distiller tank containing flower debris and distillation water is discharged in the drainage system or spread on the waste lagoon near the distilleries, thus representing a major environmental concern [4]. However, very limited data are available on the chemical characterization of the debris [4], and a detailed polyphenolic profile of the wastewater has not yet been established. During the water steam distillation process of rose flowers, the non-volatile phenolic compounds remain in the waste. Polyphenols are known to have a wide spectrum of biochemical and pharmacological effects, such as antioxidant, anticancer, anti-inflammatory, and lipid-lowering properties, as well as protective effects on the cardiovascular system (for a review see [6]). In particular, rose petals are known to contain compounds with potential antiproliferative activity, such as flavonoids, gallic and protocatechuic acids, and tannins [7, 8]. On the one hand, polyphenols have to be considered as biopollutants due to their antibacterial properties and therefore limited microbial degradability and, on the other hand, as potentially valuable compounds with pharmacological biological properties. For these reasons, it would be of interest to recover these polyphenols from the rose oil distillation wastewater (RODW) to obtain high-value products. Similar extraction procedures based on resin adsorption have been previously developed for recovery of valuable phenolic substances from olive mill wastewater [9, 10]. Thus, the aim of the present study was to develop an efficient, adequate and low-cost process for the recovery of polyphenols from RODW. The starting point for this wastewater management effort was a HPLC‑PDA‑MSbased qualitative characterization of the polyphenolic profile of RODW. Major constituents in RODW were first identified by

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

HPLC‑PDA-ESIMS analysis. Subsequently, small-scale resin adsorption/desorption experiments were then carried out for the selection of the best performing resin. Afterwards, a “mediumscale liquid waste sampling process” was designed based on these preliminary experiments. In a next step, the same HPLC‑PDA-ESIMS method was applied for the analysis of secondary metabolites obtained from a medium-scale resin fraction RF20.

Results and Discussion !

The production of rose oil from R. damascena flowers results in large quantities of RODW and solid debries as by-products of the distillation [4]. RODW contains high quantities of sugars and phenolic compounds, whereby the content varies depending on the overall organization of the rose oil distillation process used in the distilleries [3]. In the present study, the possibility of using adsorbent resins for the recovery of polyphenols in RODW was evaluated. This technology has been successfully applied for recovering phenolic substances from olive mill wastewater [9, 10]. The present experiments were performed in three steps: First, an HPLC‑PDA-ESIMS analysis of the original RODW was performed " Fig. 1 A). The UV spectra of peaks suggested the presence of (l flavan-3-ols, flavanones, flavonols, and flavone analogues [11, 12]. Compounds a–c eluting during the first 5 min of the run displayed only one major UV absorption maximum between 260– 280 nm (band II), which was typical for flavan-3-ols and flavanones [13]. UV spectra of most other flavonoids detected in RODW showed two major absorption bands and were assigned to flavonols (band I, 325–360 nm) and flavones (band I, 300– 350 nm). Flavonols were essentially quercetin and kaempferol

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1658

glycosides, and their substituents were identified based on their " Table 1). Peak l eluting at 13.6 min showed a pseuMS spectra (l do molecular ion [M + H]+ at m/z 641.3, and fragments ions at m/z 479.4 [M-162]+ and m/z 317.4 [M-162-162]+. The sequential loss of 2 × 162 u along with the fragment observed at m/z 317.4 allowed the identification of l as methylquercetin dihexoside. The presence of a galloyl moiety in compounds f, i, m, o, and q was in accordance with the pseudo molecular ion observed for these metabolites. The MS spectra of f showed a [M + H]+ ion at m/z 599.4, and a fragment ion at m/z 467.7 [M-132]+ suggesting that the galloyl moiety was not attached to the sugar moiety. These data did not correspond to any hitherto reported flavonoid. The MS spectra of peak r showed a pseudomolecular ion [M + H]+ at m/z 653.3, and a fragment ion at m/z 303.2. Hence, r was identified as quercetin acetyldisaccharide [4]. An acetyl residue was also present in compounds s and 5, and was confirmed for the latter molecule with the NMR spectral data recorded for the isolated compound (Table 5S, Supporting Information). Peak k could not be assigned to a specific flavonoid type based on the available UV and MS data. Subsequently, a small scale resin selection study was performed to identify a suitable resin, as well as RODW/resin ratio. Following the extraction procedure described below (Fig. 2S, Supporting Information), a total of 13 different XAD, IRA, and SP resins (1 volume each) were mixed with different volumes of RODW (1, 2, 4, 8, 12, and 16 times the volume of the resin, respectively). The aqueous supernatant (water fraction; WF) was removed, and resin fractions (RF1 to RF16) were then obtained by extraction of the resins with 1 volume 90 % EtOH. Total phenolics and total soluble sugar content were determined in each RF. Moreover, RFs were tested for their radical scavenging activity with a DPPH assay (Fig. 1S, Supporting Information). The amount of total phenolics present in samples recovered with 90 % EtOH from IRA resin was very low and suggested that these resins were not appropriate for the purpose. Most of the samples treated with XAD and SP resins showed a total phenolics content, soluble sugars content, and radical scavenging capacity that was proportional to the amount of RODW loaded onto the resin. Within the conditions tested, the best extraction capacity was observed with SP-825L and SP-207. For these two resins, results obtained with RF12 and RF16 were similar, suggesting a saturation of the resin when it was loaded with more than 12 volumes of RODW. XAD-2 and XAD-4 resins gave lower recovery and were not used for further development. In addition, the sedimentation speed of each resin was also taken into account for the selection of the best suited resin. In summary, the results of the small scale study suggested that 12 volumes RODW/SP-207 would be suitable for further scale-up extraction. A comprehensive qualitative analysis of the RF obtained with SP207 resin, including a fractionation and full NMR characterization of the pure compounds, was performed to complete the qualitative analysis of RODW. For this purpose, a medium-scale RF20 was prepared since it was more feasible to work with 20 than with 12 volumes RODW/SP-207. " Fig. 1 B) revealed the The HPLC‑PDA-ESIMS profile of RF20 (l presence of all major flavonoids that were previously detected in RODW. Only five minor compounds identified in RODW were not detected in RF20, and compounds h and j were seen only in RF20. HPLC‑PDA-ELSD analysis of RODW, RF20, and PW20 was performed to complete the data obtained with HPLC‑PDA-ESIMS " Fig. 2). Major amounts of very polar and non UV-absorbing (l compounds eluted during the first 5 min of the LC run of RODW

and PW20. These compounds were most likely sugars. The ELSD trace of RF20 corresponded well with the HPLC‑UV-ESIMS profile. The normalized HPLC-ELSD traces of RODW and RF20 were compared and confirmed that RF20 was indeed a polyphenol-enriched fraction when similar amounts of each sample was analyzed. Further fractionation of RF20 by a combination of column chromatography on Sephadex LH-20 and preparative RP-HPLC af" Fig. 3). Of these, 12 exhibited UV forded 14 pure compounds (l spectra characteristic of flavonoids and corresponded to peaks detected in the previous HPLC‑PDA-ESIMS analysis. Compounds 1 and 14 were detected in RODW and RF20 after their isolation. Structures were confirmed with the aid of HPLC-ESIMS, 1H‑NMR and 2D‑NMR (COSY, HSQC and HMBC) spectra and comparison with literature data [14–20]. Compound 1 (m/z 285.1 [M + H]+) was identified as a 2-phenylethyl-O-β-glucopyranoside. Compounds 2 (m/z 611.2 [M + H]+, 303.1 [(M + H)-162–146]+) and 3 (m/z 595.2 [M + H]+, 287.1 [(M + H)-162–146]+)) were identified as rutin (quercetin-3-O-β-rutinoside, 2) and kaempferol-3-O-βrutinoside (3). Compounds 4 (m/z 595.2 [M + H]+) and 5 (m/z 637.2 [M + H]+) were kaempferol diglycosides. An additional signal at δ 2.0 in the 1H‑NMR spectrum of 5 was consistent with the presence of an acetyl moiety, and the compounds were identified as kaempferol-3-O-β-glucopyranosyl (1→4)‑α‑rhamnopyranoside (4) and kaempferol-3-O-β-acetyl-glucopyranosyl (1→4)‑α‑rhamnopyranoside (5). Compounds 6 and 10 were isoquercitrin (quercetin-3‑O-β-glucopyranoside, 6) and hyperoside (quercetin-3-O-β-galactopyranoside, 10). The 1H‑NMR of compound 11 (m/z 449.1 [M + H]+, 303.1 [(M + H)-146]+) suggested the presence of a 6-deoxy-hexose residue and was identified as quercitrin (quercetin-3-O-α-rhamnopyranoside, 11). The remaining flavonoids 7–9 and 12–13 were kaempferol glycosides. The fragmentation pattern observed in the MS spectra of 7 (m/z 449.1 [M + H]+, 287.1 [(M + H)-162]+), 8 (m/z 449.1 [M + H]+, 287.1 [(M + H)162]+), and 9 (m/z 433.1 [M + H]+, 287.1 [(M + H)-146]+) suggested the presence of hexose residues. With the aid of 1H‑NMR spectra, compounds were identified as kaempferol 3-O-β-galactopyranoside (7), astragalin (kaempferol-3-O-β-glucopyranoside, 8), and kaempferol-3-O-α-rhamnopyranoside (9). Compounds 12 and 13 showed similar MS spectra (m/z 419.1 [M + H]+, 287.1 [(M + H)-132]+). In the 1H‑NMR spectra, 3JHH ‑coupling constants of the anomeric protons of 12 (δ 5.3, 7.6 Hz) and 13 (δ 5.27, 1.3 Hz) indicated the presence of a β-xylose and α-arabinose moiety, respectively, and the compounds were identified as kaempferol-3O-β-xylopyranoside (12) and kaempferol-3-O-α-arabinofuranoside (13). Compound 14 (m/z 303.5 [M + H]+) was identified as ellagic acid. The positions of the sugar residues and the interglycosidic linkages in all compounds were supported by HMBC correlations. 1H and 13C NMR data are provided as Supporting Information (Tables 1S–14S). Nine out of the 16 flavonols (compounds 2, 3, 6, 8, 9, 10, 6, 11, 15, and 16) have been previously identified with the aid of HPLC‑PDA and MS in extracts of fresh flowers of R. damascena or in flower debris after hydrodistillation [4, 11, 13, 21]. Five additional flavonols (4, 5, 7, 12, and 13) have not been reported for R. damascena flower extracts and were identified here for the first time in roses. Ellagic acid (compound 14) was found as a major compound in the SP-207 resin extract obtained from RODW. Ellagic acid has been identified in rose hips, and leaf extracts from several Rosa species [22, 23]. However, to the best of our knowledge, ellagic acid has not been detected so far in extracts of R. damascena flow-

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

1659

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

Original Papers

Original Papers

Table 1 Compounds detected in the HPLC‑PDA‑MS profiles of rose oil distillation wastewater and resin fraction RF20. Peaks 1–14 were isolated and their structures fully characterized by NMR (Tables 1S–14S, Supporting Information). Identity of aglycones 15 and 16 was confirmed by co-chromatography of reference compounds. Letters a–u refer to compounds identified by HPLC‑PDA‑MS only. Peak

Identity

[2 M – H]−

Trivial

Rt

HPLC‑DAD

[M + H]+

names

[min]

λmax [nm]

m/z

215, 270 215, 275 215, 270 220 sh, 310 210, 265, 350 sh 210, 270, 350 sh 205, 225 sh, 275, 350 sh

467.5 611.3 467.5 467.5 579.5

153.3 153.2 153.3 153.3

466.4 610.6 466.4 466.4 578.5

599.4

467.7 [M-132]+ 579.6, 467.5, 153.2

598.5

Other adducts

MW

m/z

a b* c d* e*

Flavan-3-ol hexoside Flavanone disaccharide Flavan-3-ol hexoside Flavonol Flavonol disaccharide

3.8 4.0 4.5 4.8 8.9

f*

Flavonol galloylpentoside Flavonol trisaccharide

9.3

Quercetin O-methyl trisaccharide Flavonol galloylglycoside Quercetin O-dimethyl trisaccharide n. c. Quercetin O-methyldihexoside

10.9

270, 350 sh

787.2

786.7

11.9

215, 270, 350 sh 255, 300 sh, 365 260, 350 sh 260, 300 sh, 350 sh

737.2

736.6

Phenyl-glucopyranoside Quercetin galloylhexoside Quercetin-3-O-rhamnosylglucoside Quercetin di-deoxyhexose pentoside Ellagic acid

13.8

250

285.5

14.8

220, 255, 290 sh, 355 220, 255, 350

617.4

616.5

611.2

610.5

220, 250, 300 sh, 355 250, 300 sh, 370 255, 265 sh, 300 sh, 355 255, 265 sh, 300 sh, 355 255, 265 sh, 302 sh, 355 265, 290 sh, 350 260, 295 sh, 350 265, 295 sh, 350 265, 295 sh, 350 255, 265 sh, 295 sh, 345 265, 290 sh, 350 255, 300 sh, 355 260, 295 sh, 320 sh, 345 260, 295 sh, 320 sh, 350 265, 295 sh, 320 sh, 345 215, 260 sh, 345 215, 260 sh, 350 220, 265, 315

727.3

g

h# i* j# k l

1 m 2 n 14 10 6 o 3 7 p 8 11 q 12 4 13 9 r s t

Quercetin-3-O-galactoside Quercetin-3-O-glucoside Quercetin galloylhexoside Kaempferol 3-O-rutinoside Kaempferol-3-O-galactoside Quercetin O-methyl disaccharide Kaempferol-3-O-glucoside Quercetin-3-O-rhamnoside Kaempferol galloylhexoside Kaempferol-3-O-xyloside Kaempferol-3-O-glucosylrhamnoside Kaempferol-3-O-arabinoside Kaempferol-3-O-rhamnoside Quercetin acetyldisaccharide Kaempferol acetyldisaccharide Flavone

Rusanov K et al. Recovery of Polyphenols …

9.6

12.1 12.4 13.6

Rutin

15.8 16.1 16.3

Hyperoside

16.8

Isoquercitrin

17.1 18.0 18.7 19.2 19.5

Astragalin

20.4

Quercitrin

20.8 21.8 22.8

Multiflorin B

23.4 24.3 25.3 26.4 27.8 29.0

Planta Med 2014; 80: 1657–1664

787.2

785.2 434.9 641.3

786.7

471.7

784.7

479.3 [M-162]+; 317.3 [M-162162]+ 307.4 [M + Na]+

640.5

284.3

303.4

726.6

303.5

302.2 303.3

465.2

303.3

464.4

617.3

303.4

616.5

595.2

594.5

449.1

449.3 [M-146]+ 287.3 287.2

611.2

303.3

610.5

449.1

287.2

449.1

303.3

448.4

601.3

600.5

419.1

315.2 [M-286]+ 287.2 287.5

595.2

287.3

594.5

419.1

418.4

653.4

859.2 [2 M + Na]+ 287.3 887.1 [2 M + Na]+ 287.2 303.3

637.3

287.3

433.1

611.3

927.5 [2 M‑H]−

464.4

465.2

448.4

895.5 [2 M‑H]−

448.4

418.4

863.3 [2 M‑H]−

432.4 652.6

610.5 cont.

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1660

Original Papers

1661

Table 1 Continued Peak 5

Identity

u 16

Kaempferol-3-O-acetylglucosylrhamnoside Flavone Quercetin

15

Kaempferol

Trivial

Rt

HPLC‑DAD

[M + H]+

names

[min]

λmax [nm]

m/z

Multiflorin A

30.3

215, 260 sh, 350 210, 265, 315 215, 255, 300 sh, 370 266, 298 sh, 320 sh, 365

637.2

30.6 30.8 31.9

Other adducts

[2 M – H]−

MW

m/z 287.2

636.5

595.4 303.3

594.5 302.2

287.2

286.2

Fig. 2 HPLC-ELSD chromatograms of rose oil distillation wastewater, resin fraction RF20, and water fraction WF20. Sample amount of each sample was injected and chromatograms were normalized and confirmed the extraction of RODW to obtain the flavonoid-enriched fraction RF20. SunFire™ C18 column, A 0.1 % aqueous formic acid, and B MeCN. Amount injected: 10 µg. For LC step gradient and extraction process, see text.

Fig. 3 Structures of compounds 1–16 identified in resin fraction RF20.

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

n. c. Flavonoid type could not be characterized with available data. * Compound identified in RODW only; # compound identified in RF20 only

Original Papers

ers. Over the last decades great attention has been paid to the health beneficial properties of ellagic acid [24, 25]. In particular, several studies demonstrated promising anticancer effects of the compound [26–29]. Finally, flavonoids were quantified in RF20. Given that the metabolites identified in RODW were essentially derivatives of quercetin and kaempferol, the aglycones were quantified after hydrolysis. The sample was subjected to acid hydrolysis (0.1% TFA, 2 h at 90 °C), followed by HPLC‑UV analysis which showed the presence of three major peaks, namely quercetin, kaempferol, and ellagic acid. The polyphenol content in RF20 was 4% for quercetin, 4.5 % for kaempferol, and 4.5 % for ellagic acid. In conclusion, our pilot study demonstrated that polyphenols in RODW can be efficiently obtained with the aid of activated adsorption resins. The study also suggested a possible industrial scalability of the process. Analysis of the composition of the RF20 showed that RODW is a rich source of phenolic compounds, such as ellagic acid, 2-phenylethyl-O-β-glucopyranoside, and several kaempferol and quercetin glycosides with potential health beneficial properties. Further studies on process optimization and investigation of the pharmacological properties of the concentrated resin fraction are currently underway.

Material and Methods !

General experimental procedures Technical grade solvents were used for extraction and CC. HPLC grade solvents were purchased from Scharlau and Lab-Scan. HPLC grade water was obtained using EASY-pure II (Barnstead) water purification system. Reagents for determination of total phenolic content, glucose content, and DPPH radical scavenging were purchased from Sigma. Resins used included nine Amberlite® resins, XAD-2, XAD-4, XAD-7HP, XAD-1180N, XAD-16N, IRA-67, IRA-404, IRA-410, and IRA-900 (all Dow Chemical), and four Sepabeads™ resins, SP-70, SP‑207, SP-825L, and SP-850 (all Mitsubishi Chemical). They were all purchased from Sigma. New batches of resin were activated before use with a 4-step washing cycle using, in sequence, MeOH, H2O, 1 M NaOH, and H2O. At each step, the mixture was shaken at 150 rpm (15 min for MeOH and H2O, and 12 h for 1 M NaOH), and the supernatant was subsequently removed after decantation. Sephadex® LH-20 was purchased from GE Healthcare. A P50 pump (GE Healthcare) and Superfrac fraction collector (Pharmacia Biotech) were used for CC on Sephadex LH-20. For TLC analysis, silica gel plates 60 F254 (Merck) were used with EtOAc/formic acid/glacial acetic acid/H2O (100/11/11/26) as mobile phase. Detection was at UV 254 and 366 nm before staining, and at 366 nm after staining with natural product reagent (1% diphenylboryloxyethylamine in MeOH; Sigma-Aldrich) followed by PEG solution (5% ethanolic polyethylene glycol-4000; Sigma-Aldrich). HPLC‑PDA-ELSD analyses were performed on an Alliance 2695 instrument (Waters) equipped with a 996 PDA detector, and an evaporative light scattering detector (ELSD) Series 2000 (Alltech). Separations were performed on a C18 SunFire™ column (3.5 µm, 3 × 150 mm i. d., Waters) equipped with a guard column (3 × 10 mm). The column was thermostatted at 50 °C. For the ELSD, N2 flow was 2.4 L/min, and evaporation temperature was 60 °C. HPLC‑PDA-ESIMS analyses were performed on an 1100 Series HPLC system (Agilent) consisting of a quaternary low-pressure mixing pump with degasser module, column oven, PDA detector,

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

and autosampler coupled to an Esquire 3000 plus mass spectrometer (Bruker). ESI‑MS spectra were recorded in positive and negative ion mode between m/z 200–1500. Same analytical HPLC column and method as for HPLC‑PDA-ELSD analyses were used. The column temperature was 25 °C. For all analytical HPLC analyses, the samples were dissolved in H2O or 30–100% DMSO, depending on the solubility of fractions. The concentration of fractions was 1 mg/mL, and 10 µL were injected. The mobile phase consisted of 0.1 % aqueous formic acid (A) and MeCN (B). The flow rate was 0.4 mL/min. The following long gradient profile was used for qualitative analysis: 10 % B isocratic for 4 min, gradient 4–5 min to 17% B, 5–26 min to 27 % B, 26–27 min to 50 % B, 27–35 min 50% B isocratic, 35–36 min to 100 % B, 36–41 min 100% B isocratic, 41–42 min to 10% B. UV spectra were recorded from 210 to 400 nm. The following short gradient was used for quantitative analysis: 10 % B to 80 % B in 11 min, holding for 1 min. Preparative HPLC was carried out on a HPLC system consisting of a SCL-10VP controller, LC-8A binary pumps, a UV‑Vis SPD-M10A VP detector and Class-VP 6.12 software (all Shimadzu). The mobile phase consisted of H2O (A) and ACN (B) for all fractions, except for fraction 12 where 0.1 % formic acid was added to solvents A and B. Separations were performed on a SunFire Prep C18 OBD (30 × 150 mm, 5 µm, Waters) column equipped with a guard column (20 × 10 mm i. d.). The flow was set to 20 mL/min. UV spectra were recorded from 200 to 400 nm; 1H NMR and 2D NMR (COSY, HSQC, HMBC) data were recorded in DMSO-d6 or in CD3OD on a Bruker Avance III™ 500 MHz NMR spectrometer equipped with a 1-mm TXI microprobe. Data were processed with Topspin 2.1 software (Bruker).

Rose oil distillation wastewater Wastewater obtained after distillation of full-blown R. damascena flowers was obtained from the distillery of the Institute of Roses, Essential and Medicinal Crops (IREMC) in Kazanlak, Bulgaria, in June 2013. Wastewater was filtered through a cheese-cloth to afford RODW which was stored at + 4 °C in 10 L plastic bottles. A voucher specimen of RODW_June2013 is stored at − 20 °C at the Institute for Pharma Technology, School of Life Sciences, FHNW.

Resin fractions RFs were prepared from RODW as follows (Fig. 2S, Supporting Information): (A) RODW (for amounts, see procedures below for different extraction batches) was mixed with activated resin (for volume and resin type, see procedures below) in a 1 L glass flask, and the suspension was shaken at 150 rpm for 30 min on an orbital shaker. After sedimentation of the resin, the supernatant (WF) was removed. The resin was then mixed with 600 mL H2O for 10 min at 150 rpm, and the supernatant was removed after decantation. (B) 90% EtOH were added to the resin, mixed for 10 min, and the supernatant (RF) was removed. All fractions obtained were stored at − 20 °C. (C) The resin was mixed with 600 mL H2O, and the suspension shaken for 10 min at 150 rpm. The supernatant was removed after decantation, and the resin was regenerated for use.

Resin fraction small scale extraction and resin selection Sample preparation: RFs were prepared as described above using 30 mL of different resin types (Nine Amberlite® resins, XAD-2, XAD-4, XAD-7HP, XAD-1180N, XAD-16N, IRA-67, IRA-404, IRA410, and IRA-900, and four Sepabeads™ resins, SP-70, SP-207, SP-825L, SP-850) and RODW at different RODW/resin volume ra-

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1662

Original Papers

Resin fraction small scale extraction with SP-207 and process optimization Sample preparation: RFs were prepared following the procedure described above. 40 mL activated SP-207 resin was used as starting material. RODW was mixed with the resin at following RODW/resin volume ratios: 1, 4, 8, 12, 16, 20, and 24. Extraction was with 60 mL 90% EtOH. Sample hydrolysis: 10 mg each RF was solubilized in 1 mL H2O with 0.1 mL TFA. The solution was heated at 90 °C for 2 h, and 0.1 mL DMSO was added to the solution after cooling at room temperature. This solution was used to prepare the 1 : 5 and 1 : 10 dilutions for HPLC analyses. Quantitative analysis by HPLC‑UV: Standard stock solutions (SS) of ellagic acid, quercetin and kaempferol were prepared in DMSO. At least five calibration samples were prepared by serial dilution with SS in DMSO covering the range from 0.01 and 0.1 mg/mL. Calibration curves were established based on the peak area for each three compounds at 255 nm for ellagic acid (Rt 7.3 min) and quercetin (Rt 9.5 min), and at 266 nm for kaempferol (Rt 10.5 min), respectively. RF samples obtained after hydrolysis were diluted tenfold and analyzed in duplicate.

Resin fraction pilot medium scale extraction and fractionation Sample preparation: RF20 was prepared in medium scale following the procedure described above using 600 mL RODW and 30 mL activated SP-207 resin selected in the previous study. 60 mL 90 % EtOH were used for the extraction (step B), and resin was regenerated as described above (step C). 3 × 31 extraction cycles were performed to obtain a sufficiently large amount of RF20 (11′160 mL) for qualitative analyses and compound isolation. RF20 was concentrated and divided into individual 10 mLvials (each vial containing the equivalent of 70 mL RF20 equivalent or 0.6% total RF). RF fractionation: A portion of RF20 (6.8 g) was dissolved in 60 mL 80 % MeOH and applied to a Sephadex LH-20 column (100 × 5.5 cm i. d.) eluted with 90% MeOH at a flow rate of 2 mL/min. Time-based fractions (10 min each) were collected and combined based on TLC pattern to 16 main fractions: F1 (1–74, 1.32 g), F2 (75–88, 0.22 g), F3 (89–104, 0.40 g), F4 (105–110, 0.17 g), F5 (111–122, 0.64 g), F6 (123–130, 0.31 g), F7 (131–136, 0.28 g), F8 (137–152, 0.42 g), F9 (153–174, 0.26 g), F10 (175–194, 0.23 g), F11 (195–226, 0.45 g), F12 (227–300, 0.56 g), F13 (0.19 g), F14 (0.47 g), F15 (0.07 g), and F16 (0.07 g). Preparative HPLC was performed under isocratic conditions. Fraction F1 was separated using 20% ACN. The sample was dissolved in 30% DMSO (50 mg/mL) and injected as 5 aliquots to afford compound 1 (26.6 mg, Rt 10.6 min). Fraction F3 was separated with 22 % ACN. The sample was dissolved in 30% DMSO (50 mg/mL) and injected as 7 aliquots to yield compounds 2 (13.4 mg, Rt 7.0 min) and 3 (30.1 mg, Rt 9.6 min). Fraction F4 was separated using 30 % ACN. The sample was dissolved in 30 % DMSO (50 mg/mL) and injected as 6 aliquots to provide compounds 4 (28 mg, Rt 6.2 min) and 5 (32.6 mg, Rt 10.6 min). Fraction F6 was separated with 25% ACN. The sample was dissolved in 30 % DMSO (50 mg/mL) and injected as 7 aliquots to provide compounds 6 (26.4 mg, Rt 7.3 min), 7 (19.5 mg, Rt 9.0 min), 8 (19.6 mg, Rt 10.0 min), and 9 (44.2 mg, Rt 16.2 min). Fraction F8 was separated using 25 % ACN. The sample was dissolved in 50 % DMSO (50 mg/mL) and was injected as 7 aliquots to afford compounds 10 (70.5 mg, Rt 7.3 min), 11 (26.8 mg, Rt 10.7 min), 12 (17.5 mg, Rt 13.1 min), and 13 (29.6 mg, Rt 14.9 min). Fraction F12 was separated by using 22% isocratic ACN + 0.1 % formic acid. The sample was dissolved in 100% DMSO (50 mg/mL) and was injected as 5 aliquots to provide compound 14 (15.7 mg, Rt 10.4 min).

Supporting information Comparison of adsorption and extraction capacity of different resins, a summary of RODW processing procedure, as well as 1H and 13C NMR data of compounds 1–13 are provided as Supporting Information

Acknowledgements !

This project was financially supported by the Swiss National Science Foundation (SNF), project number IZEBZ0_143 110/1 and the Ministry of Education and Science of Bulgaria – Project D02– 1148 within the frame of the Bulgarian-Swiss Research Program 2011–2016.

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

tios (1, 4, 8, 12, and 16). Each resin sample was washed with 90 mL H2O and extracted with 60 mL 90 % EtOH to afford RF1 to RF16 for each resin type. Determination of total phenolic content: Total phenolic content was determined using a modified Folin-Ciocalteu (FC) method [30] adapted for ELISA micro plate reader (LKB 6060–006). Briefly, 10 µL of sample were mixed with 790 µL distilled H2O and 50 µL Folin-Ciocalteuʼs phenol reagent (Sigma) in a 96 Nunc® DeepWell™ plate (Sigma). After vortexing, the plate was incubated for 5 min at room temperature, followed by the addition of 150 µL 20 % Na2CO3. The samples were again vortexed and incubated for 2 h at room temperature. Aliquots of 200 µL of each sample were transferred to an ELISA micro plate, and the absorption at 670 nm was measured. The phenolic content was calculated as gallic acid equivalents (GAE). DPPH scavenging: The free-radical-scavenging capacity of samples was measured following a modified DPPH method [31] adapted for ELISA micro plate reader. Briefly, in a 96 Nunc® DeepWell™ plate (Sigma), 190 µL of sample were mixed with 1.3 mL of 0.1 mM DPPH solution in ethanol and incubated at room temperature in the dark for 30 min. The absorption was measured at 540 nm. DPPH inhibition was calculated according to the following formula: % inhibition = [(Abs control – Abs sample)/Abs control] × 100. The data were expressed as L-ascorbic acid equivalents. Determination of total sugars: Total sugars content was determined using the anthrone method [32] adapted for the ELISA micro plate reader. Briefly, 125 µL sample were transferred into a 96 Nunc® DeepWell™ plate (Sigma) placed on ice. To each well, 250 µL of 75 % sulfuric acid was added, and the plate was vortexed. Then, 500 µL 0.2 % anthrone solution in 75% sulfuric acid were added, and the samples heated at 94 °C for 30 min. The samples were rapidly cooled on ice, and 200 µL of each sample were transferred into a 96 well ELISA micro plate. Absorption at 620 nm was measured, and the data were calculated as glucose equivalents.

1663

Original Papers

Conflict of Interest !

The authors declare no conflict of interest.

References 1 Rusanov K, Kovacheva N, Atanassov A, Atanassov I. Rosa damascena Mill., the oil-bearing Damask rose: genetic resources, diversity and perspectives for molecular breeding. Floriculture Ornamental Biotech 2009; 3: 14–20 2 Rusanov K, Kovacheva N, Rusanova M, Atanassov I. Traditional Rosa damascena flower harvesting practices evaluated through GC/MS metabolite profiling of flower volatiles. Food Chem 2011; 129: 1851– 1859 3 Kovacheva N, Rusanov K, Atanassov I. Industrial cultivation of oil bearing rose and rose oil production in Bulgaria during 21st century, directions and challenges. Biotechnol Biotech Eq 2010; 24: 1793–1798 4 Schieber A, Mihalev K, Berardini N, Mollov P, Carle R. Flavonol glycosides from distilled petals of Rosa damascena Mill. Z Naturforsch 2005; 60: 379–384 5 Rusanov KE, Kovacheva NM, Atanassov I. Comparative Gc/Ms analysis of rose flower and distilled oil volatiles of the oil bearing rose Rosa damascena. Biotechnol Biotechnol Eq 2011; 25: 2210–2216 6 Scalbert A, Johnson IT, Saltmarsh M. Polyphenols: antioxidants and beyond. Am J Clin Nutr 2005; 81 (Suppl. 1): S215–S217 7 Daglia M. Polyphenols as antimicrobial agents. Curr Opin Biotech 2012; 23: 174–181 8 Pan MH, Ho CT. Chemopreventive effects of natural dietary compounds on cancer development. Chem Soc Rev 2008; 37: 2558–2574 9 Agalias A, Magiatis P, Skaltsounis AL, Mikros E, Tsarbopoulos A, Gikas E, Spanos I, Manios T. A new process for the management of olive oil mill waste water and recovery of natural antioxidants. J Agric Food Chem 2007; 55: 2671–2676 10 Bertin L, Ferri F, Scoma A, Marchetti L, Fava F. Recovery of high added value natural polyphenols from actual olive mill wastewater through solid phase extraction. Chem Eng J 2011; 171: 1287–1293 11 Kumar N, Bhandari P, Singh B, Gupta AP, Kaul VK. Reversed phase-HPLC for rapid determination of polyphenols in flowers of rose species. J Sep Sci 2008; 31: 262–267 12 Mabry TJ, Markham KR, Thomas MB. The systematic identification of flavonoids. Berlin, New York: Springer-Publisher; 1970: 354 13 Velioglu YS, Mazza G. Characterization of flavonoids in petals of Rosa damascena by HPLC and spectral-analysis. J Agric Food Chem 1991; 39: 463–467 14 Markham KR, Geiger H, Jaggy H. Kaempferol-3-O-glucosyl(1–2)rhamnoside from Ginkgo biloba and a reappraisal of other gluco(1–2, 1–3 and 1–4)rhamnoside structures. Phytochem 1992; 31: 1009–1011 15 Markham KR, Ternai B. C‑13 NMR of flavonoids. 2. Flavonoids other then flavone and flavonol aglycones. Tetrahedron 1976; 32: 2607– 2612

Rusanov K et al. Recovery of Polyphenols …

Planta Med 2014; 80: 1657–1664

16 Markham KR, Ternai B, Stanley R, Geiger H, Mabry TJ. C‑13 NMR-Studies of flavonoids 3. Naturally occurring flavonoid glycosides and their acylated derivatives. Tetrahedron 1978; 34: 1389–1397 17 Nasr C, Lobstein-Guth A, Haag-Berrurier M, Anton R. Quercetin coumaroyl glucorhamnoside from Ginkgo biloba. Phytochem 1987; 26: 2869–2870 18 Pauli GF, Matthiesen U, Fronczek FR. Sulfates as novel steroid metabolites in higher plants. Phytochem 1999; 52: 1075–1084 19 Sikorska M, Matlawska I. Kaempferol, isorhamnetin and their glycosides in the flowers of Asclepias syriaca L. Acta Pol Pharm 2001; 58: 269–272 20 Yoshinari K, Shimazaki N, Sashida Y, Mimaki Y. Flavanone xyloside and lignans from Prunus jamasakura Bark. Phytochem 1990; 29: 1675– 1678 21 Kumar N, Bhandari P, Singh B, Bari SS. Antioxidant activity and ultraperformance LC-electrospray ionization-quadrupole time-of-flight mass spectrometry for phenolics-based fingerprinting of rose species: Rosa damascena, Rosa bourboniana and Rosa brunonii. Food Chem Toxicol 2009; 47: 361–367 22 Nowak R, Gawlik-Dziki U. Polyphenols of Rosa leaves extracts and their radical scavenging activity. Z Naturforsch C 2007; 62: 32–38 23 Nowak R, Olech M, Pecio L, Oleszek W, Los R, Malm A, Rzymowska J. Cytotoxic, antioxidant, antimicrobial properties and chemical composition of rose petals. J Sci Food Agr 2014; 94: 560–567 24 Sepulveda L, Ascacio A, Rodriguez-Herrera R, Aguilera-Carbo A, Aguilar CN. Ellagic acid: Biological properties and biotechnological development for production processes. Afr J Biotechnol 2011; 10: 4518–4523 25 Vattem DA, Shetty K. Biological functionality of ellagic acid: A review. J Food Biochem 2005; 29: 234–266 26 Mishra S, Vinayak M. Ellagic acid inhibits PKC signaling by improving antioxidant defense system in murine T cell lymphoma. Mol Biol Rep 2014; 41: 4187–4197 27 Mishra S, Vinayak M. Ellagic acid induces novel and atypical PKC isoforms and promotes caspase-3 dependent apoptosis by blocking energy metabolism. Nutr Cancer 2014; 66: 675–681 28 Umesalma S, Nagendraprabhu P, Sudhandiran G. Antiproliferative and apoptotic-inducing potential of ellagic acid against 1,2-dimethyl hydrazine-induced colon tumorigenesis in Wistar rats. Mol Cell Biochem 2014; 388: 157–172 29 Vanella L, Di Giacomo C, Acquaviva R, Barbagallo I, Li Volti G, Cardile V, Abraham NG, Sorrenti V. Effects of ellagic acid on angiogenic factors in prostate cancer cells. Cancers 2013; 5: 726–738 30 Singleton VL, Orthofer R, Lamuela-Raventos RM. Analysis of total phenols and other oxidation substrates and antioxidants by means of Folin-Ciocalteu reagent. Methods Enzymol 1999; 299: 152–178 31 Brandwilliams W, Cuvelier ME, Berset C. Use of a free-radical method to evaluate antioxidant activity. Food Sci Technol-Leb 1995; 28: 25–30 32 Scott TA, Melvin EH. Determination of dextran with anthrone. Anal Chem 1953; 25: 1656–1661

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.

1664

Copyright of Planta Medica is the property of Georg Thieme Verlag Stuttgart and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.