The Gastroprotective Effects of Eugenia Dysenterica (Myrtaceae) Leaf Extract: the Possible Role of Condensed Tannins

Home , Hydrolysable tannin, Plinia edulis, Prodelphinidin

722 Regular Article Biol. Pharm. Bull. 37(5) 722–730 (2014) Vol. 37, No. 5

The Gastroprotective Effects of Eugenia dysenterica (Myrtaceae) Leaf Extract: The Possible Role of Condensed Tannins Ligia Carolina da Silva Prado,a Denise Brentan Silva,c Grasielle Lopes de Oliveira-Silva,a Karen Renata Nakamura Hiraki,b Hudson Armando Nunes Canabrava,a and Luiz Borges Bispo-da-Silva*,a a Laboratory of Pharmacology, Institute of Biomedical Sciences, Federal University of Uberlândia; b Laboratory of Histology, Institute of Biomedical Sciences, Federal University of Uberlândia; ICBIM-UFU, Minas Gerais 38400– 902, Brazil: and c Núcleo de Pesquisas em Produtos Naturais e Sintéticos, Faculty of Pharmaceutical Science at Ribeirão Preto, University of São Paulo; NPPNS-USP, São Paulo 14040–903, Brazil. Received June 26, 2013; accepted February 6, 2014

We applied a taxonomic approach to select the Eugenia dysenterica (Myrtaceae) leaf extract, known in Brazil as “cagaita,” and evaluated its gastroprotective effect. The ability of the extract or carbenoxolone to protect the gastric mucosa from ethanol/HCl-induced lesions was evaluated in mice. The contributions of nitric oxide (NO), endogenous sulfhydryl (SH) groups and alterations in HCl production to the extract’s gastroprotective effect were investigated. We also determined the antioxidant activity of the extract and the possible contribution of tannins to the cytoprotective effect. The extract and carbenoxolone protected the gastric mucosa from ethanol/HCl-induced ulcers, and the former also decreased HCl production. The blockage of SH groups but not the inhibition of NO synthesis abolished the gastroprotective action of the extract. Tannins are present in the extract, which was analyzed by matrix assisted laser desorption/ionization (MALDI); the tannins identiﬁed by fragmentation pattern (MS/MS) were condensed type-B, coupled up to eleven ﬂavan-3-ol units and were predominantly procyanidin and prodelphinidin units. Partial removal of tannins from the extract abolished the cytoprotective actions of the extract. The extract exhibits free-radical-

scavenging activity in vitro, and the extract/FeCl3 sequence stained gastric surface epithelial cells dark-gray. Therefore, E. dysenterica leaf extract has gastroprotective effects that appear to be linked to the inhibition of HCl production, the antioxidant activity and the endogenous SH-containing compounds. These pleiotropic actions appear to be dependent on the condensed tannins contained in the extract, which bind to mucins in the gastric mucosa forming a protective coating against damaging agents. Our study highlights the biophar- maceutical potential of E. dysenterica. Key words Eugenia dysenterica; gastroprotection; condensed tannin; Myrtaceae; matrix assisted laser desorption/ionization (MALDI)

Numerous plant-derived extracts are gastroprotective and proach to plant collection relies on the premise that related represent a great source of bioactive compounds with the ther- taxa have inherited the genetic ability to produce similar apeutic potential to treat gastric and duodenal ulcers, diseases pharmacologically active secondary metabolites.14,15) There- that are considered to be worldwide heath issues.1) Thus, stud- fore, considering that many species of the Myrtaceae family ies of this subject are important because they provide data that including those from the Eugenia genus have been reported to assist governmental institutions in their policies concerning possess gastroprotective activities, we applied the taxonomic the use of plants in the phytotherapeutic industry, the sustain- approach to select E. dysenterica and study its cytoprotective able use of the biodiversity and consequently, the application actions on the stomach. In a pilot experiment published only of phytotherapeutics to health promotion programs. as part of an academic report, our co-author (Canabrava) and Many plants exhibiting gastroprotective actions that have his colleagues observed that E. dysenterica leaf extract indeed been studied belong to the Myrtaceae family and include protects the gastric mucosa of rats from damage caused by the Campomanesia lineatifolia,2) Campomanesia xanthocarpa,3) administration of indomethacin, a cyclooxygenase inhibitor.16) Eugenia jambolana,4,5) Myrtus communis,6) Plinia edulis,7) To study this subject further and to pharmacologically char- Syzygium aromaticum,8,9) and Syzygium cumini.10) Eugenia dy- acterize the E. dysenterica leaf extract, we investigated the senterica DC. (Myrtaceae), also known in Brazil as “cagaita” mechanisms involved in its gastroprotective effect on ethanol/ or “cagaiteira,” is a shrubby ornamental and melliferous tree HCl-induced ulcers in mice; moreover, a group of secondary that is widely distributed throughout the second largest Bra- compounds that are putatively responsible for this effect were zilian biome, the Cerrado or upland savannah.11) The fruit is analyzed and identiﬁed by matrix assisted laser desorption/ edible and has traditionally been used as a cathartic agent; ionization (MALDI)-MS and MS/MS. the leaves, in contrast, have been used to treat diarrheic diseases.11) Recently published data provide scientiﬁc support, at MATERIALS AND METHODS least in part, for these folk uses.12,13) In considering the selection of a plant for pharmacological Animals The experiments were performed using male studies many approaches can be used. The taxonomic ap- Swiss mice (35–45 g). The animals were housed in a room at 21°C with 12 h light/dark cycles and were provided free access The authors declare no conflict of interest. to tap water and standard chow. All the protocols used were

* To whom correspondence should be addressed. e-mail: [email protected]; © 2014 The Pharmaceutical Society of Japan [email protected] May 2014 723 reviewed and approved by the Ethics Committee on Animal mixed carefully. After 15 min, the samples were centrifuged at Use of the Federal University of Uberlândia (CEUA/UFU; 3000 rpm for 15 min. The supernatant was removed and used process n° 022/11, addendum 175/11). to treat mice (GD-ED group) that had been fasted for 24 h Plant Material Eugenia dysenterica DC. (Myrta- (1.0 mL/40 g, i.e., 1000 mg/kg) 30 min before the administra- ceae) leaves were collected at the Santa Rita farm located in tion of the ethanol/HCl solution. A control extract (ED group) Lassance city, state of Minas Gerais, Brazil (17°59′14.23″S; was prepared by adding 1.5 mL of the acetate buffer without 44°44′38.00″W) in September, 2011. The species was identi- gelatin to the same volume of the extract (80 mg/mL). To fied by Dr. Adriana Arantes, and a voucher specimen was de- verify the effectiveness of this method in removing polyphe- posited at the Herbarium of the Federal University of Uberlân- nolic compounds from the extract, 2.980 µL of FeCl3 (0.01 M dia (Uberlândia, MG, Brazil) with the number HUFU-45956. in 0.01 M HCl) was added to 20 µL of the supernatant from Preparation of the Leaf Extract The leaves were washed the gelatin-treated, control extract or distilled water (blank) (immersed for 10 min in 70% ethanol, 5 min in 0.2% hypochlo- to determine its absorbance at 510 nm 45 min later. The data rite and 10 min in running water; this procedure was applied from both groups were expressed as the absorbance units/mg to disinfect the surface of the leaves) and dried in an oven at of the extract. 40°C for 48 h. The powdered dried leaves were extracted with Determination of the Ability of the Extract to Bind to distilled water (20%, w/v) for 48 h at room temperature. The the Gastric Mucosa and Stain with Ferric Chloride This extract was lyophilized and then stored at −20°C and protect- experiment was conducted as previously described20) with ed from light until use (yield: 0.8%). The surface disinfection some modifications. Briefly, animals without any treatment procedure neither altered the biological effect of the extract on were sacrificed by cervical dislocation under anesthesia the gastric tissue, nor changed the structure of its condensed (sodium thiopental, 60 mg/kg, i.p.); the stomachs were then tannins when compared to extract obtained with leaves that removed, fixed in phosphate-buffered 4% formalin, embed- were washed only with water (data not shown). ded in paraffin and further sectioned in 5 µm slices. Depar- Qualitative Determination of the Tannins in the Extract affinized and hydrated sections were then treated with a 5% The presence of tannins was screened by reacting the extract solution of E. dysenterica leaf extract for 1 h, rapidly rinsed samples with gelatin or FeCl3 according to the method de- in water (3 times), treated with 2% FeCl3 for 15 min, washed scribed by Costa (2000), with some modifications as follows: in water, alcohol and xylene and mounted in gum damar. No (I) 1 drop of HCl (10%) was added to 1 mL of the extract counterstain was used, and the slices were observed under a solution (1 mg/mL), and gelatin solution (2.5% in 10% NaCl) light microscope. was added dropwise; the formation of a precipitate indicated a Effect of the Extract on Gastric Secretion The assay 21) positive reaction. (II) FeCl3 (1%) was added dropwise to 5 mL was performed according to the method of Shay et al., of the extract solution (1.0 mg/mL); the solution was monitored with some modifications as follows: the mice were fasted for for a brownish green or a blue-black color and/or precipitation. 36 h with free access to water; immediately after laparotomy Evaluation of the Gastroprotective Effect of the Extract and pylorus ligature under anesthesia (10 mg/kg xylazine The acidified ethanol-induced gastric lesion model17) was used and 100 mg/kg ketamine, i.p.), E. dysenterica leaf extract to evaluate the gastroprotective effect of the E. dysenterica (1000 mg/kg), cimetidine hydrochloride (100 mg/kg)9) or saline leaf extract. Mice were divided into 6 groups (n=6–10) and were administered intraduodenally, and the abdomen was fasted for 24 h prior to the oral administration (total volume: sutured. The animals were sacrificed 4 h later by cervical dis- 0.8 mL) of saline (0.9% NaCl), carbenoxolone (250 mg/kg)9) or location while they were under thiopental sodium anesthesia E. dysenterica leaf extract (100, 300, 550, 1000 mg/kg); 50 min (60 mg/kg, i.p.); the abdomen was then reopened, and another after the treatments, all the animals orally received 0.26 mL ligature was placed at the esophageal end near the diaphragm. of 0.3 M HCl/60% ethanol. The animals were sacrificed by The stomachs were removed, the gastric content was collected cervical dislocation under anesthesia (sodium thiopental, and weighed (in mg), and the gastric content volume was ad- 60 mg/kg, intraperitoneally (i.p.)) 1 h after the administration justed to 5 mL by the addition of distilled water. The solution of the HCl/ethanol solution. The stomachs were removed, was centrifuged at 3000 rpm for 10 min. The pH was deter- opened along the greater curvature, fixed between 2 glass mined using a pH meter, and the total acid in the supernatant plates and scanned. ImageJ software (http://rsb.info.nih.gov/ was determined by titrating to pH 7.0 using a 0.01 mol/L ij/) was used to analyze the stomach images, and the results NaOH solution and phenolphthalein as an indicator. The free were expressed as the ulcerative index (U.I.). To calculate the and total acidity values were expressed as pH values and + U.I., the lesions were scored according to the severity of the µeqH /g of the gastric content, respectively. gastric mucosal injury as follows: hemorrhagic lesions (3), Evaluation of the Participation of Nitric Oxide (NO) and high hyperemic area (2) and moderate or soft hyperemic area Endogenous SH-Groups in the Gastroprotective Action of (1). Thus, the U.I. was determined as previously described,18) the Extract These experiments were conducted as previous- with some modifications: U.I.=3×hemorrhagic lesion area ly described,22) with some modifications: mice were fasted for (mm2)+2×high hyperemic area (mm2)+1×moderate/soft hy- 24 h and divided into groups (n=9–10) that were treated with 2 G peremic area (mm ). NEM (N-ethylmaleimide, 10 mg/kg, i.p.), (N -nitro-L-arginine Determination of the Gastroprotective Effect of the methyl ester (L-NAME), 70 mg/kg, i.p.), or saline (0.5 mL, Extract with a Reduced Tannin Content Tannins were i.p.).9) Thirty minutes later, animals of all groups received oral partially removed from the extract as previously described,19) doses of the vehicle (saline), carbenoxolone (250 mg/kg) or E. with modifications as follows: 1.5 mL of a gelatin solution dysenterica leaf extract (1000 mg/kg). After 50 min, gastric (2.5%) in 0.2 mol/L acetate buffer, pH 5.0 with 0.17 mol/L mucosa lesions were induced. The animals were sacrificed NaCl was added to 1.5 mL of the extract (80 mg/mL) and 1 h after the administration of the HCl/ethanol solution, the 724 Vol. 37, No. 5 stomachs were removed, and the gastric mucosa lesions were measured as described above. Determination of the in Vitro Antioxidant Activity of the Extract The free-radical-scavenging activity of the E. dysenterica leaf extract was measured in vitro using the 1,1-diphenyl-2-picryl-hydrazyl (DPPH·; Sigma-Aldrich Chemi- cal Co., St. Louis, MO, U.S.A.) free radical method.23) Briefly, 0.5 mL of water or extract dissolved in water was added to 1.5 mL of an ethanolic DPPH· solution to reach the following · final concentrations in the reaction tube: 0.06 mM DPPH and 0.1, 0.2, 0.3, 1.0, 2.0, 3.0, 10.0 or 30.0 µg/mL extract. After a 30 min incubation at room temperature in a low luminosity environment, the absorbance of each sample was determined at 515 nm. The difference between the absorbance of the solu- Fig. 1. Evaluation of the Gastroprotective Effect of the Aqueous Euge- tions containing only DPPH· and that of solutions containing nia dysenterica Leaf Extract DPPH· plus extract was determined. Tests were performed The effects of E. dysenterica leaf extract (100–1000 mg/kg, p.o.) and carbenoxolone (CBX, 250 mg/kg, p.o.) on gastric lesions induced by the administration (p.o.) in duplicate using solutions without reagents as blanks and of acidified hydroalcoholic solution were analyzed in mice (n=6–10). The results ascorbic acid as a positive control. The sample concentration represent mean±S.E.M. of the ulcer index (U.I.) in mm2. * p<0.05 and ** p<0.01 versus control (SAL). providing 50% inhibition (IC50) was calculated by nonlinear regression analysis (hyperbolic equation) of the inhibition percentage plotted against the sample concentration using the was purchased from Cristália Produtos Químicos e Farmacêu- GraphPad Prism® software. ticos LTDA (Itapira, SP, Brazil). Liquid Chromatography-Tandem Mass Spectrometry Statistical Analysis The data were expressed as the The LC analyses were performed with an UPLC-DAD-ESI means± S.E.M. and were compared by ANOVA followed by TQ Acquity (Waters®, Milford, MA, U.S.A.), using a C18 Dunnett’s test or by Student’s t-test, as appropriate. The data Shim-pack XR-ODS (2.2 µm; 2.0 mm×50 mm, Shimadzu) were considered to be statistically significant when p<0.05. column. The mobile phase was acetonitrile (solvent B) and water containing 0.1% formic acid, and the flow rate was RESULTS 0.3 mL·min−1. The column temperature was 30°C and the injection volume was 5 µL. The elution profile was the fol- E. dysenterica leaf extract decreased the U.I. (in mm2) in a lowing: 5% B (0 to 0.9 min), 5 to 20% B (0.9 to 5.1 min), 20 dose-dependent manner and was as effective as the reference to 100% B (5.1 to 8.5 min), 100% B (8.5 to 9.2 min). The MS drug carbenoxolone (Fig. 1). The addition of gelatin and FeCl3 conditions were the following: cone energy of 25 kV, colli- to the E. dysenterica leaf extract resulted in the formation sion energy of 25 eV, capillary energy of 2.5 kV. Nitrogen was of a precipitate and the development of a blue-black color, used as the nebulizing and drying gas (650 L h−1, 350°C) and respectively. Moreover, the addition of gelatin to the extract argon was used as the collision gas. TIC chromatograms were decreased the extract’s absorbance at 510 nm after the addition recorded between m/z 50 and 1000 in both negative and posi- of FeCl3 (Fig. 2A) and greatly decreased its gastroprotective tive modes. effect; indeed, there was no significant difference between MALDI-MS and MALDI-MS/MS Analyses The MS the U.I. observed in saline (control) and GT-ED treated and MS/MS analyses were performed using the UltrafleX- groups (Fig. 2B). Moreover, the E. dysenterica extract/FeCl3 treme MALDI-TOF/TOF equipment (Bruker Daltonics, sequence stained the surface epithelial cells dark-gray (Fig.

Bremen, Germany). A mixture of peptides (standard II of 3); neither E. dysenterica extract nor FeCl3 alone stained the Bruker®) was used for the external and internal calibrations. epithelial surface (not shown). The extract reduced DPPH· to The ions were accelerated at 20 kV. The experimental condi- a yellow product in a concentration-dependent manner, with tions were the following: pulsed ion extraction of 120 ns, laser approximately half the potency of the standard antioxidant frequency of 1000 Hz, reflectron mode, positive ion mode; 800 agent, ascorbic acid (IC50: 3.97±0.05 µg/mL vs. 2.09±0.01 µg/ shots were averaged to record a mass spectrum. For MS/MS mL; extract vs. ascorbic acid). The intraduodenal administra- analyses, the selected ions were accelerated to 19 kV in the tion of E. dysenterica leaf extract or cimetidine decreased the LIFT cell for MS/MS analyses. 2,5-Dihydroxybenzoic acid total and free acidity of the gastric juice of mice subjected to (DHB) was used as the matrix (20 mg/mL with 30% acetoni- the pyloric-ligation procedure (Table 1). However, neither the trile and 70% H2O with 0.1% trifluoracetic acid). The aqueous extract nor cimetidine significantly altered the mass of the extract was solubilized with ACN : H2O (3 : 7) and DHB con- gastric content (Table 1). The U.I. observed after E. dysen- taining 0.1 M NaCl (1 : 1 : 0.1) was added. These mixtures (1 µL) terica leaf extract administration did not differ between the were spotted onto a ground stainless steel MALDI target. The control (saline-treated) animals and the NEM-treated mice compounds were identified by their MS data, fragmentation (Fig. 4A). L-NAME administration decreased the ability of pattern and accurate mass measures. carbenoxolone but not the extract to protect the gastric mucosa G Drugs Carbenoxolone, N-ethylmaleimide and N -nitro-L- from ethanol/HCl-induced lesions (Fig. 4B). arginine methyl were purchased from Sigma-Aldrich Chemi- The aqueous extract of E. dysenterica was analyzed by cal Co., cimetidine was purchased from Fluka Analytical LC-DAD-MS and LC-DAD-MS/MS. The compounds were (Buchs, Switzerland), xylazine and ketamina were purchased identified by analysis of the MS, MS/MS, and UV data, in from Syntec do Brasil LTDA (Cotia, SP, Brazil), thiopental comparison with previously published fragmentation patterns May 2014 725

and standard injections. The substances procyanidin B-1 (5), catechin (4) and a dimeric procyanidin gallate were identiﬁed (Table 2, Fig. 5). The aqueous extract of E. dysenterica was also analyzed by MALDI. Figure 6 shows the MALDI-TOF spectrum of the polymeric tannin mixture in the extract. MS/ MS data were analyzed to determine the presence of oligomeric units of flavan-3-ol or galloyl. The identiﬁed tannins were condensed type-B, coupled up to eleven flavan-3-ol units. Moreover, E. dysenterica condensed tannins were found to consist predominantly of procyanidin units and were also combined with prodelphinidin units in the series 2 and 3 (Table 3, Fig. 5).

DISCUSSION

Our data demonstrated that E. dysenterica leaf extract dose- dependently protects the mouse gastric mucosa from damage induced by administration of ethanol/HCl. This observation conﬁrms preliminary data from our laboratory (using cyclooxygenase inhibitor to induce gastric lesion in rat)16) and shows that the gastroprotective effect of E. dysenterica leaf extract is not species- or model-speciﬁc, which may reflect its ability to

Table 1. Effects the Eugenia dysenterica Leaf Extract (ED, 1000 mg/kg) or Cimetidine (100 mg/kg) Administered Intraduodenally on Biochemical Parameters of Gastric Contents

Parameters/Treatments Control Cimetidine ED Fig. 2. Determination of the Gastroprotective Effect of the Aqueous Gastric contents (g) 0.47±0.07 0.27±0.04 0.35±0.10 Eugenia dysenterica Leaf Extract with a Reduced Tannin Content pH 3.41±0.20 5.77±0.54** 5.36±0.54** Absorbance of Eugenia dysenterica leaf extract (ED) and gelatin-treated E. Free acidity (µeq H+/g) 6.42±5.69 0.86±0.51* 2.21±1.55* dysenterica leaf extract (GT-ED) at 510 nm after the addition of FeCl3 (A). Effects of saline (Sal, p.o.), ED (1000 mg/kg, p.o.) and GT-ED (1000 mg/kg, p.o.) on gastric Total acidity (µeq H+/g) 62.60±7.60 36.00±7.00* 34.60±7.60* lesions induced by the administration (p.o.) of acidiﬁed hydroalcoholic solution in mice (n=6–7) (B). The results represent mean±S.E.M. of the absorbance or of the Values represent the mean±S.E.M.; * p<0.05, ** p<0.01 versus control. n=5–8 ulcer index (U.I.) in mm2. * p<0.05 versus control. mice.

Fig. 3. Determination of the Ability of the Aqueous Eugenia dysenterica Leaf Extract to Bind to the Gastric Mucosa and Stain with Ferric Chloride

Mice gastric mucosa section (glandular portion) sequence treated with E. dysenterica and FeCl3. No counterstain. Note the dark-gray staining of the gastric mucosal surface after the addition of FeCl3 (arrows). 726 Vol. 37, No. 5 influence various defense mechanisms in the gastric mucosa. tective actions of plants from the Myrtaceae family.2,3) Our The inhibitory effect of the extract on gastric lesions was less preliminary phytochemical analyses suggested the presence potent than that of the standard treatment (carbenoxolone, of polyphenolic compounds in E. dysenterica leaf extract be-

250 mg/kg), however, the highest dose of the extract used was cause it became blue-black in the presence of FeCl3, a general as effective as carbenoxolone, which is evidence of the power- characteristic reaction for this class of chemicals.24) Moreover, ful cytoprotective action of E. dysenterica leaf extract. the extract effectively precipitated gelatin, a hallmark of tan- Polyphenolic compounds, including flavonoids and tannins, nic compounds,24) strongly suggesting the presence of tannins are among the secondary metabolites linked to the gastropro- among the polyphenolic compounds from the E. dysenterica leaf extract. Based on these observations, we hypothesized that the tannins in the E. dysenterica extract could account for the extract’s cytoprotective activity. To test this hypothesis, we treated animals with a partially puriﬁed extract from which the tannin content had been greatly decreased by gelatin precipitation (the effectiveness of our methods in removing polyphenolic compounds was spectrophotometrically conﬁrmed; in fact, the absorbance at 510 nm of gelatin-treated samples was decreased by almost 1/4 from that of the non-treated samples

after FeCl3 administration; Fig. 2A). After this precipitation, the extract lost its gastroprotective activity, suggesting that the tannins are indeed responsible for the E. dysenterica leaf extract-mediated cytoprotection of the mouse gastric mucosa. In addition to precipitate soluble proteins, tannins also bind to mucins in various tissues, including those on gastric surface epithelial cells.20) Thus, our observation that E. dysenterica leaf extract can bind epithelial mucins and that its presence can be detected as a dark-gray complex after ferric chloride treatment suggest that the tannin–mucin complex formed after E. dysenterica treatment could represent a protective coating that prevents gastric tissue damage caused by proteolytic enzymes, H+ and by ethanol, which solubilizes the mucus barrier. It is important to mention that this protective coating mechanism has also been proposed to explain the cytoprotective actions of others tannin-rich extracts.2,25) Therefore, considering that tannins appear to be extremely important for the gastroprotective effect of the extract, LC- DAD-MS/MS analyses were performed in an attempt to identify the tannins in the aqueous extract of E. dysenterica. It was possible to identify the compounds procyanidin B-1, catechin and procyanidin O-gallate; many additional tannins Fig. 4. Evaluation of the Participation of Nitric Oxide (NO) and Endo- were identified by MALDI analyses. The molecular formulae genous SH-Groups in the Gastroprotective Action of the Eugenia dysen- were confirmed by the accurate mass data (with an internal terica Leaf Extract calibrant), and the tannins were determined by the MS/MS Effects of E. dysenterica leaf extract (1000 mg/kg, p.o.) or carbenoxolone (CBX, data, which was used to define the series. 250 mg/kg, p.o.) on gastric lesions induced by the administration (p.o.) of acidified hydroalcoholic solution in mice (n=9–10) pre-treated or not with NEM (10 mg/kg, The tannins identified in E. dysenterica were condensed i.p.) (A) or L-NAME (70 mg/kg, i.p.) (B). The results represent the mean±S.E.M. with B-type linkages. The polymerization of flavan-3-ol units, of the ulcer index (U.I.) in mm2.* p<0.05 and ** p<0.01 versus respective control (pre-treated with saline, Sal). as procyanidin, prodelphinidin, prorobinetidin and others,

Table 2. Identiﬁcation of Chromatographic Peaks from the Aqueous Extract of E. dysenterica and UV, MS and MS/MS Data

t UV Negative (m/z) Positive (m/z) Peak R Compound (min) (nm) MS MS/MS MS MS/MS

1 3.01 Procyanidin B-1a) 278 577 [M−H]− 577 (20 eV)→451, 425, 407, 579 [M+H]+ 579 (20 eV)→427, 409, 301, 339, 299, 289, 245, 289, 275, 247, 205, 161, 125 163, 139, 127 2 3.41 Catechina) 278 289 [M−H]− 289 (20 eV)→248, 227, 217, 291 [M+H]+ 291 (20 eV)→165, 147, 139, 579 [2M−H]− 203, 188, 164, 151, 123, 119, 111 125, 123 3 4.73 Dimeric procyanidin 277 729 [M−H]− — 731 [M+H]+ 731 (20 eV)→471, 440, 427, gallate 410, 317, 301, 290, 247, 180, 163, 140, 127

tR: Retention time. a) Conﬁrmed by injection of standards. May 2014 727

Fig. 5. General Structure of the Tannins Presented in the Aqueous Eugenia dysenterica Leaf Extract Typical polymer structure of the condensed tannins repeating unit: procyanidin (PCY) and prodelphinidin (PDE). Typical linear condensed tannins type B with C4–C8 (1) and C4–C6 (2), linkages unit substituted by galloyl (3), catechin (4) and procyanidin B-1 (5). yields condensed tannins, and the flavan-3-ols units are linked fore, it was possible to determine the sequential monomer though C4–C6 or C4–C8 bonds between the units26) (Fig. 5). units and chemical constitution of individual polymers only by The MS data showed that the series were composed primar- MS/MS analyses. Hydrolysable tannins have been identified ily of procyanidin units (290 µ) (Table 3, Figs. 5, 6), and the in the Eugenia genus,27,28) but only condensed tannins were oligomers consisted of 11 units. The prodelphinidin units and detected in the extract analyzed in our study. Previous phyto- galloyl groups were also present in the identified tannins. For chemical investigation of E. dysenterica revealed compounds the identification of tannins, the MS/MS data were fundamen- such as a peptide,12) monoterpenes and sesquiterpenes,29) tal, starting with the confirmation number of flavan-3-ol units vitamins and carotenoids.30) The catechin derivatives were and the molecular weight of these units. The next step is the suggested to be present in E. dysenterica extract, but this re- confirmation of the identity of the units from the fragments port was based only on UV data.31) Our studies provide more produced by Retro Diels Alder (RDA) fragmentation in ring complete evidence of the chemical nature of E. dysenterica B or the substituents, such as galloyl group. For this reason, because the compounds were identified with certainty using the condensed tannin structures can be proposed with more UV, MS and MS/MS data. reliable manner. The MS/MS spectra confirmed, for example, It has been reported that catechins and procyanidins did not that the ions m/z 889, 1177, 1465, 1753, 2041, 2329, 2617, 2905 show antiulcer activity against HCl/etanol-induced ulcer,25) and 3193 compose this series, such as the MS/MS spectrum of moreover, these compounds have low or no capability to m/z 2329 that revealed the fragment ions m/z 2041, 1753, 1465, precipitate protein.32,33) These observations together with 1177 and 889, the same ions observed in the MS spectrum of the fact that gelatin-treated E. dysenterica leaf extracts lost the extract (see supplemental information). Therefore, the units their gastroprotective effect, strongly suggest that catechins added from ion m/z 889 have the molecular weight of 290 Da and procyanidin B-1 do not significantly contribute to this and suggesting the presence of procyanidin unit. The frag- effect. However, it has been suggested that the gastroprotec- mentation of ion m/z 889 yielded the fragment ions m/z 737, tive actions of tannic compounds are related to molecule size. 719, 601, 567, 449 and 313 (see supplemental information). Indeed, tannins yielded from procyanidin units as tetramers, The m/z 737 fragment ion represents the loss of 152 µ and pentamers and hexamers exhibit gastroprotective effects confirms the presence of the ring B substituents of the procy- whose magnitude is proportional to the size of the oligomeric anidin unit. The fragment ions of m/z 601 and 313 confirmed chain.25) Interestingly, the interaction of tannins with proteins the presence of the coupling of successive procyanidin units. also appears to be related to the molecule size; thus, procyani- The other important polymers in this fraction were formed by din hexamers and pentamers show higher protein-binding and one galloyl addition and successive procyanidin units up to 11 precipitation ability than tetramers, trimers show little precipi- units (series 4). The MS/MS analysis of the m/z 753 [M+Na] + tation ability, and as mentioned above, procyanidin dimers and produced the ion m/z 463, which confirms the presence of gal- monomers (as catechins) are unable to precipitate proteins.25) loyl in the starter unit. We used MS/MS to propose the tannin These reported data and the fact that many condensed tan- structures, and not only MS. The identification of tannins is nins were identified in the E. dysenterica leaf extract strongly more reliable from MS/MS analyses, but there are few studies suggest that condensed tannins are the main compounds reporting the use of MS/MS data to identify tannins. There- responsible for the powerful gastroprotective effect of the E. 728 Vol. 37, No. 5

Fig. 6. Mass Spectra of the Aqueous Eugenia dysenterica Leaf Extract Mass spectra recorded in positive ionization mode: series 1–2 (A) and 3–4 (B). dysenterica leaf extract. dysenterica leaf extract-mediated gastroprotection, as SH Ethanol/HCl induced-ulcers involve the formation of free groups possess electron-donating ability and are thus capable radicals34) that may induce lipid peroxidation and cell dam- of binding free radicals.36) NEM administration greatly re- age.35) Moreover, ethanol decreases the gastric levels of glu- duced the protective effect of the extract, thus highlighting tathione, a well-known antioxidant compound.36) Therefore, the participation of endogenous SH groups in the gastropro- based on our observation that E. dysenterica leaf extract tective effect. This observation suggests that the antioxidant possesses antioxidant activity, a property that is usually found activity of the extract alone is not sufﬁcient to fully protect in plants containing polyphenolic compounds,37) it is pos- the stomach from ethanol/HCl-induced damage. Therefore, sible that this effect is part of the mechanism by which the there may be a cooperative relationship between the activity extract protects the gastric mucosa from injury. To investigate of the extract and free-radical scavenging by the endogenous this possibility, we analyzed whether blocking endogenous SH-bearing molecules; alternatively or additionally, the extract SH groups with the SH alkylator NEM38) could decrease E. may augment the bioavailability of the endogenous SH groups. May 2014 729

Table 3. Identiﬁcation of Condensed Tannins by MALDI-TOF from Despite these doses appear even to be high for humans, it Aqueous Extract of E. dysenterica should be noted that the actually gastroprotective dose of some extract used in clinical studies41) represents only a frac- [M+Na]+ (error) MF Compound tion (ca. 28%) of that can be calculated by the body surface 42) Series 1 889.1905 (5.7 ppm) C45H38O18 3 PCY area normalization method. Therefore, the effective human 1177.2586 (0.3 ppm) C60H50O24 4 PCY equivalent doses of E. dysenterica leaf extract concerning its 1465.3279 (3.8 ppm) C75H62O30 5 PCY gastroprotective effect may be lower than that necessary to 1753.3888 (1.7 ppm) C90H74O36 6 PCY protect mice. Moreover, as pointed out by Souza-Formigoni 2041.4441 (2.5 ppm) C105H86O42 7 PCY and his colleagues, it is possible that lower doses of the ex- no 2329.5 C120H98O48 8 PCY no tract given chronically may also have therapeutic effects 2617.6 C135H110O54 9 PCY against gastric lesions.42) 2905.6no C H O 10 PCY 150 122 60 In summary, the present study demonstrates that E. dysen- 3193.7no C H O 11 PCY 165 134 66 terica leaf extract strongly protects the gastric mucosa from Series 2 905.1897 (0.9 ppm) C H O 2 PCY–1 PDE 45 38 19 ethanol/HCl-induced injury and highlights the biopharma- 1193.2517 (1.8 ppm) C H O 2 PCY–1 PDE–1 PCY 60 50 25 ceutical potential of this species. The gastroprotective effect 1481.3120 (3.5 ppm) C75H62O31 2 PCY–1 PDE–2 PCY 1769.3835 (1.6 ppm) C H O 2 PCY–1 PDE–3 PCY is linked to the inhibition of HCl production, to the extract’s 90 74 37 free-radical-scavenging activity (antioxidant property) and to 2057.4293 (7.2 ppm) C105H86O43 2 PCY–1 PDE–4 PCY no the presence of endogenous SH-containing compounds, but 2345.5 C120H98O49 2 PCY–1 PDE–5 PCY no does not involve NO-mediated cytoprotection. These pleio- 2633.6 C135H110O55 2 PCY–1 PDE–6 PCY no tropic actions appear to be primarily related to the condensed 2921.6 C150H122O61 2 PCY–1 PDE–7 PCY no 3209.7 C165H134O67 2 PCY–1 PDE–8 PCY tannins in E. dysenterica, which are predominantly composed

Series 3 1193.2517 (1.8 ppm) C60H50O25 3 PCY–1 PDE of procyanidin followed by prodelphinidin, in addition to

1481.3120 (3.5 ppm) C75H62O31 3 PCY–1 PDE–1 PCY the presence of the galloyl group in one polymeric series.

1769.3835 (1.6 ppm) C90H74O37 3 PCY–1 PDE–2 PCY Moreover, tannins present in the extract appear to bind to

2057.4293 (7.2 ppm) C105H86O43 3 PCY–1 PDE–3 PCY epithelial mucins in the gastric mucosa forming a protective no 2345.5 C120H98O49 3 PCY–1 PDE–4 PCY coating against damaging agents. no 2633.6 C135H110O55 3 PCY–1 PDE–5 PCY no 2921.6 C150H122O61 3 PCY–1 PDE–6 PCY Acknowledgments We thank Débora Cristina de Oliveira no 3209.7 C165H134O67 Nunes and Simone Ramos Deconte for technical assistance. Series 4 753.1418 (1.8 ppm) C37H30O16 PCYG–PCY This work was supported by a master fellowship by the Co- 1041.1998 (6.5 ppm) C52H42O22 PCYG–2 PCY ordenação de Aperfeiçoamento de Pessoal de Nível Superior 1329.2725 (1.9 ppm) C67H54O28 PCYG–3 PCY (CAPES) for LCSP. We are also indebted to Pró-Reitoria de 1617.3297 (2.2 ppm) C82H66O34 PCYG–4 PCY Pós-Graduação e Pesquisa da Universidade Federal de Uber- 1905.4042 (3.9 ppm) C97H78O40 PCYG–5 PCY lândia (PROPP-UFU) and to Fundação de Amparo à Pes- 2193.4679 (3.5 ppm) C112H90O46 PCYG–6 PCY quisa do Estado de São Paulo (FAPESP) for ﬁnancial support. 2481.5177 (2.3 ppm) C127H102O52 PCYG–7 PCY no Finally, we would like to thank Dr. Norberto Peporine and the 2769.6 C142H114O58 PCYG–8 PCY no Mass Spectrometry Facility of Physics and Chemistry from 3057.6 C157H126O64 PCYG–9 PCY no FCFRP-USP for the excellent customer service. 3345.7 C172H138O70 PCYG–10 PCY MF: molecular formula, PCY: procyanidin unit, PDE: prodelphinidin unit, PCYG: procyanidin O-gallate unit, no: not observed with internal calibrant (low intensity). REFERENCES

1) Borrelli F, Izzo AA. The plant kingdom as a source of anti-ulcer Several lines of evidence suggested that NO is an impor- remedies. Phytother. Res., 14, 581–591 (2000). tant endogenous gastroprotective mediator. The mechanisms 2) Madalosso RC, Oliveira GC, Martins MT, Vieira AED, Barbosa J, involved in NO-mediated gastroprotection include the main- Caliari MV, Castilho RO, Tagliati CA. Campomanesia lineatifolia RUIZ & PAV. as a gastroprotective agent. J. Ethnopharmacol., 139, tenance of mucosal integrity, the inhibition of leukocyte mi- 772–779 (2012). gration and platelet adherence, and the increase in mucosal 3) Markman BEO, Bacchi EM, Kato ETM. Antiulcerogenic effects of 39) blood flow. L-NAME, a nonselective inhibitor of the nitric Campomanesia xanthocarpa. J. Ethnopharmacol., 94, 55–57 (2004). oxide synthase enzymes that produce NO, did not signiﬁcantly 4) Chaturvedi A, Kumar MM, Bhawani G, Chaturvedi H, Kumar M, compromise the extract’s gastroprotective effect, suggesting Goel RK. Effect of ethanolic extract of Eugenia jambolana seeds on that this mediator is not involved in the cytoprotective actions gastric ulceration and secretion in rats. Indian J. Physiol. Pharma- of E. dysenterica. Finally, the extract inhibited the secretion of col., 51, 131–140 (2007). HCl similar to the standard treatment (cimetidine); this effect 5) El-Shenawy SM. Evaluation of some pharmacological activities of may also contribute to the decrease in mucosal injury induced ethanol extracts of seeds, pericarp and leaves of Eugenia jamolana by ethanol/HCl administration. in rats. Inﬂammopharmacology, 17, 85–92 (2009). 6) Sumbul S, Ahmad MA, Asif M, Saud I, Akhtar M. Evaluation of It has been suggested that the extrapolation of animal dose Myrtus communis LINN. berries (common myrtle) in experimental to human dose is correctly performed through normalization 40) ulcer models in rats. Hum. Exp. Toxicol., 29, 935–944 (2010). to body surface area. Accordingly to this approach, the 7) Ishikawa T, Donatini R dos S, Diaz IE, Yoshida M, Bacchi EM, human equivalent doses calculated for E. dysenterica extract Kato ET. Evaluation of gastroprotective activity of Plinia edulis are approximately 45 mg/kg and 81 mg/kg (take into account (VELL.) SOBRAL (Myrtaceae) leaves in rats. J. Ethnopharmacol., 118, 550 and 1000 mg/kg as active doses in mice, respectively). 730 Vol. 37, No. 5

527–529 (2008). 25) Saito M, Hosoyama H, Ariga T, Kataoka S, Yamaji N. Antiulcer 8) Agbaje EO. Gastrointestinal effects of Syzigium aromaticum (L.) activity of grape seed extract and procyanidins. J. Agric. Food MERR. & PERRY (Myrtaceae) in animal models. Nig. Q. J. Hosp. Chem., 46, 1460–1464 (1998). Med., 18, 137–141 (2008). 26) Schofield P, Mbugua DM, Pell AN. Analysis of condensed tannins: 9) Santin JR, Lemos M, Klein-Júnior LC, Machado ID, Costa P, de A review. Animal Feed Science and Technology, 91, 21–40 (2001). Oliveira AP, Tilia C, de Souza JP, de Sousa JP, Bastos JK, de An- 27) Omar R, Li L, Yuan T, Seeram NP. α-Glucosidase inhibitory hydro- drade SF. Gastroprotective activity of essential oil of the Syzygium lysable tannins from Eugenia jambolana seeds. J. Nat. Prod., 75, aromaticum and its major component eugenol in different animal 1505–1509 (2012). models. Naunyn Schmiedebergs Arch. Pharmacol., 383, 149–158 28) Lee MH, Nishimoto S, Yang LL, Yen KY, Hatano T, Yoshida T, (2011). Okuda T. Two macrocyclic hydrolysable tannin dimers from Euge- 10) Ramirez RO, Roa CC Jr. The gastroprotective effect of tannins nia uniflora. Phytochemistry, 44, 1343–1349 (1997). extracted from duhat (Syzygium cumini SKEELS) bark on HCl/etha- 29) Costa TR, Fernandes OF, Santos SC, Oliveira CM, Lião LM, Ferri nol induced gastric mucosal injury in Sprague-Dawley rats. Clin. PH, Paula JR, Ferreira HD, Sales BH, Silva M doR. Antifungal Hemorheol. Microcirc., 29, 253–261 (2003). activity of volatile constituents of Eugenia dysenterica leaf oil. J. 11) Almeida SP, Proença CEB, Sano SM, Ribeiro JF. Cerrado: espécies Ethnopharmacol., 72, 111–117 (2000). vegetais úteis, first ed., EMBRAPA-CPAC, Planaltina (1998). 30) Cardoso LM, Martino HSD, Moreira AVB, Ribeiro SMR, Pinheiro- 12) Lima TB, Silva ON, Oliveira JTA, Vasconcelos IM, Scalabrin FB, Sant’Ana HM. Cagaita (Eugenia dysenterica DC.) of the Cerrado Rocha TL, Grossi-De-Sá MF, Silva LP, Guadagnin RV, Quirino BF, of Minas Gerais, Brazil: Physical and chemical characterization, Castro CFS, Leonardecz E, Franco OL. Identification of E. dysen- carotenoids and vitamins. Food Res. Int., 44, 2151–2154 (2011). terica laxative peptide: a novel strategy in the treatment of chronic 31) Souza PM, Elias ST, Simeoni LA, de Paula JE, Gomes SM, Guerra constipation and irritable bowel syndrome. Peptides, 31, 1426–1433 EN, Fonseca YM, Silva EC, Silveira D, Magalhães PO. Plants from (2010). Brazilian Cerrado with potent tyrosinase inhibitory activity. PLoS 13) Lima TB, Silva ON, Silva LP, Rocha TL, Grossi-de-Sá MF, Franco ONE, 7, e48589 (2012). OL, Leonardecz E. In Vivo Effects of Cagaita (Eugenia dysenterica, 32) Hagerman AE, Butler LG. Protein precipitation method for the DC.) Leaf Extracts on Diarrhea Treatment. Evid. Based Comple- quantitative determination of tannins. J. Agric. Food Chem., 26, ment. Alternat. Med., 2011, 1–10 (2011). 809–812 (1978). 14) Cordell GA, Beecher WW, Douglas Kinghorn A, Pezzuto JM, Con- 33) Cala O, Pinaud N, Simon C, Fouquet E, Laguerre M, Dufourc EJ, stant HL, Chai H, Fang L, Seo E, Long L, Cui B, Slowing-Barillas Pianet I. NMR and molecular modeling of wine tannins binding to K. The dereplication of plant-derived natural products. Studies in saliva proteins: revisiting astringency from molecular and colloidal Natural Products Chemistry. (Atta-ur-Rahman ed.) Vol 19, Elsevier, prospects. FASEB J., 24, 4281–4290 (2010). the Netherlands, pp. 749–791 (1997). 34) Mutoh H, Hiraishi H, Ota S, Ivey KJ, Terano A, Sugimoto T. Role 15) Cordell GA. Biodiversity and drug discovery—a symbiotic relation- of oxygen radicals in ethanol-induced damage to cultured gastric ship. Phytochemistry, 55, 463–480 (2000). mucosal cells. Am. J. Physiol., 258, G603–G609 (1990). 16) Junqueira VMS, Silva MA, Canabrava LCMN, Rossi DA, Beletti 35) Nishida K, Ohta Y, Ishiguro I. Relation of inducible nitric oxide ME, Canabrava HAN. Avaliação antimicrobiana e antiulcerogênica synthase activity to lipid peroxidation and nonprotein sulfhydryl da Eugenia dysenterica. Horizonte Científico, 1, 1–10 (2007). oxidation in the development of stress-induced gastric mucosal 17) Mizui T, Doteuchi M. Effect of polyamines on acidified ethanol lesions in rats. Nitric Oxide, 2, 215–223 (1998). induced gastric lesions in rats. Jpn. J. Pharmacol., 33, 939–945 36) Kidd PM. Glutathione: Systemic protectant against oxidative and (1983). free radical damage. Altern. Med. Rev., 2, 155–176 (1997). 18) Szelenyi I, Thiemer K. Distention ulcer as a model for testing of 37) Repetto MG, Llesuy SF. Antioxidant properties of natural com- drugs for ulcerogenic side-effects. Arch. Toxicol., 41, 99–105 (1978). pounds used in popular medicine for gastric ulcers. Braz. J. Med. 19) Karamać M. Chelation of Cu(II), Zn(II), and Fe(II) by tannin con- Biol. Res., 35, 523–534 (2002). stituents of selected edible nuts. Int. J. Mol. Sci., 10, 5485–5497 38) Szabo S, Nagy L, Plebani M. Glutathione, protein sulfhydryls and (2009). cysteine proteases in gastric mucosal injury and protection. Clin. 20) Pizzolato P, Lillie RD. Mayer’s tannic acid-ferric chloride stain for Chim. Acta, 206, 95–105 (1992). mucins. J. Histochem. Cytochem., 21, 56–64 (1973). 39) Laine L, Takeuchi K, Tarnawski A. Gastric mucosal defense and 21) Shay H, Komarov SA, Fele SS, Meranze D, Gruenstein H, Siplet H. cytoprotection: bench to bedside. Gastroenterology, 135, 41–60 A simple method for the uniform production of gastric ulceration in (2008). the rat. Gastroenterology, 5, 43–61 (1945). 40) Reagan-Shaw S, Nihal M, Ahmad N. Dose translation from animal 22) Arrieta J, Benitez J, Flores E, Castillo C, Navarrete A. Purification to human studies revisited. FASEB J., 22, 659–661 (2007). of gastroprotective triterpenoids from the stem bark of Amphiptery- 41) Santos-Oliveira R, Coulaud-Cunha S, Colaço W. Revisão da May- gium adstringens; role of prostaglandins, sulfhydryls, nitric oxide tenus ilicifolia MART. ex REISSEK, Celastraceae. Contribuição ao and capsaicin sensitive neurons. Planta Med., 69, 905–909 (2003). estudo das propriedades farmacológicas. Rev. Braz. Fharmacogn., 23) Brand-Williams W, Cuvelier ME, Berset C. Use of a free-radical 19, 650–659 (2009). method to evaluate antioxidant activity. LWT—Food Science and 42) Souza-Formigoni ML, Oliveira MG, Monteiro MG, da Silveira- Technology, 28, 25–30 (1995). Filho NG, Braz S, Carlini EA. Antiulcerogenic effects of two May- 24) Costa AF. Farmacognosia: Farmacognosia Experimental, third ed., tenus species in laboratory animals. J. Ethnopharmacol., 34, 21–27 Fundação Calouste Gulbekian, Lisboa (2000). (1991).