Simple and rapid spectrophotometric method for quality determination of roselle (Hibiscus sabdariffa) TJPS 2016, 40 (4):

Transcription

1 Originl Article Thi Journl of Phrmceuticl Sciences Simple nd rpid spectrophotometric method for qulity determintion of roselle (Hibiscus sbdriff) Tsmporn Sukwttnsinit 1, Jnkn Burn-Osot 2, Uthi Sotnphun 1 1 Deprtment of Phrmcognosy, Fculty of Phrmcy, Silpkorn University, Nkhon-Pthom 73000, Thilnd, 2 Deprtment of Phrmceuticl Chemistry, Fculty of Phrmcy, Silpkorn University, Nkhon-Pthom 73000, Thilnd Corresponding Author: Uthi Sotnphun, Deprtment of Phrmcognosy, Fculty of Phrmcy, Silpkorn University, Nkhon-Pthom, 73000, Thilnd. Tel.: , Fx: E-mil: u.sotn@gmil. com ABSTRACT A simple nd rpid spectrophotometric method hs been developed to determine the qulity of roselle (Hibiscus sbdriff). The bsorbnce t 530 nm (coloring intensity) nd the rtio between the bsorbnces t 380 nd 530 nm (degrdtion index) of its 50% methnol extrct were suggested s the criterion prmeters to discriminte between the qulified nd unqulified smples. The contents of totl nthocynins nd totl phenolic compounds were lso estimted by the prtil lest squre nd multiple liner regression models estblished from the bsorbnces t 330, 380, nd 530 nm. Their relibility ws indicted by the low root-men-squre error of prediction = nd high correltion between the ctul nd prediction vlues of the tested smples (R = 0.99). Received: June 9, 2016 Accepted: Nov 15, 2016 Published: Dec 01, 2016 Keywords: Anthocynins, chemometrics, Hibiscus sbdriff, phenolic compounds, qulity, spectrophotometry INTRODUCTION Roselle (Hibiscus sbdriff L. fmily Mlvcee) is worldwide populr herb. Its phrmcologicl ctivities nd clinicl studies hve been investigted, nd mny extensive reviews hve been published [1-4]. The min ctivities were, for exmples, ntioxidtion, reducing blood pressure in pre- nd mild-hypertensive ptients, improving lipid profile, nd diuretic. Its min phytochemicl constituents were orgnic cids, nthocynins, nd other phenolic compounds [1,2]. The officil qulity specifiction of roselle is published in Europen Phrmcopoei; cid content is ssyed nd coloring intensity is determined [5]. Thilnd nd Chin were the world s lrgest suppliers [6]. However, there ws only one comprehensive study on the qulity of Thi roselle [7], nd none of the publiction hs been found for the Chinese roselle. Most of the qulity study ws on the Africn roselle [8-13]. The qulity of Sudnese roselle ws sid to be the best but its production quntity ws low nd ll ws exported to Germny. The other importnt producers were Mexico, Egypt, Senegl, Tnzni, Mli, nd Jmic but the production is mostly used domesticlly [6]. The other two qulity studies of roselle were tht from Mlysi [14] nd Turkey [15]. Beside the generl qulity control prmeters, the totl titrtble cid nd the contents of its ctive constituents, i.e., nthocynins nd phenolic compounds hve been reported [5,7,8,10,12,14-16]. The totl contents of nthocynins nd phenolic compounds were determined by spectrophotometric nd Folin-Cioclteu methods, respectively, wheres the contents of two min nthocynins (delphinidin 3-smbubioside nd cynidin 3-smbubioside) were nlyzed by high performnce liquid chromtogrphy [12]. One obstcle for the quntittion of totl nthocynin content in routine work is the high expenditure of stndrd nthocynins. The reddish color of roselle, which referred 194

2 to the color of nthocynins, ws the lterntive qulity prmeter [5,12,15,16]. Visul inspection is the esiest technique to clssify between the qulified nd the unqulified smples [7,11] nd it ws one of the commercil specifiction [6,13]. However, some borderline smples re not esy to be discriminted due to humn visul limittion. Color of smple could be more ccurtely determined by mesuring for its spectrophotometric bsorbnce t 520 nm [5] or CIELAB color [12]. Furthermore, the contents of totl nthocynins nd ech of its min nthocynins could be estimted from CIELAB color [17]. However, different cultivr of roselle possesses different color. It is ornge-red, drk purple, nd purplish-red for the smples from Sudn, Chin, nd Thilnd, respectively [6]. The qulity of roselle produced in Thilnd hs been investigted [7] but its color mesurement hs never been studied. Therefore, this study suggested simple nd rpid method for the qulity determintion of Thi roselle bsed on its color using spectrophotometry. Moreover, the models for prediction of the contents of totl nthocynin nd totl phenolic compounds were lso constructed from ultrviolet/visible (UV/Vis) spectrl dt using chemometric methods, i.e., prtil lest squre (PLS) nd multiple liner regression (MLR). MATERIALS AND METHODS Plnt Mterils 32 roselle smples throughout Thilnd were purchsed from locl mrket during 2003 to They were identified by Associte Professor Uthi Sotnphun. The voucher specimens (US03001-US03020 nd US04001-US04012) re deposited t the Herbrium of the Deprtment of Phrmcognosy, Silpkorn University, Thilnd. Their qulity grding ws ssigned bsed on their color s reddish group (n = 16) nd brownish group (n = 16). 60% of smples (10 reddish smples nd 9 brownish smples) were rndomly selected s trining set, nd the rest smples were used s test set. All smples were dried, ground to fine powder, pssed through sieve mesh mm, stored t 4 C nd wrmed to room temperture before use. UV/Vis Spectroscopic Dt Exctly g of the fine smple ws extrcted by soniction with 5 ml of 50% methnol for 15 min. The extrct (0.2 ml) ws diluted with 0.8 ml of 50% methnol nd mixed well. The solution (0.3 ml) ws further diluted nd mixed well with 4.9 ml of 50% methnol. UV/Vis spectr in the rnge of nm of the finl solution were collected by UV/Vis spectrophotometer (Agilent 8453 Model G 11C3A, USA). Norwy). Dt of ll UV/Vis spectr ( nm) were pretreted by subtrction with the bsorbnce t 700 nm before nlyses (Figure 1). RESULTS AND DISCUSSION This ws the continuous study on the qulity control of roselle of Thilnd. 32 smples nd the nlysis result of their totl nthocynin nd totl phenolic contents were obtined from our previous investigtion (Tble 1) [7]. Bsed on the color visulized by nked eye, ll smples (n = 32) were divided into reddish (qulified) nd brownish (unqulified) groups. Totl nthocynin contents of ll reddish smples were more thn 1.0% clculted s delphinidin 3-smbubioside, suggesting this vlue s criticl vlue for qulity grding of roselle smples. However, stndrd nthocynin ws need for the nlysis. It is expensive nd not esy to purify in lbortory. To Tble 1: Summrized totl nthocynins content (% clculted s delphinidin 3 smbubioside) nd totl phenolic content (% clculted s gllic cid) of ll smple set of reddish nd brownish roselle (men±sd [minimum mximum]) Smple set (n) Totl nthocynins Totl phenolics Reddish Totl (16) 1.80±0.56 ( ) Trining (10) 1.72±0.50 ( ) Test (6) 1.92±0.68 Brownish ( ) Totl (16) 0.50±0.20 ( ) Trining (9) 0.53±0.20 ( ) Test (7) 0.47±0.20 SD: Stndrd devition ( ) 3.01±0.40 ( ) 3.03±0.40 ( ) 3.00±0.45 ( ) 2.27±0.34 ( ) 2.26±0.34 ( ) 2.28±0.36 ( ) Determintion of Totl Anthocynin nd Totl Phenolic Contents All dt of the contents of totl nthocynins (% clculted s delphinidin 3-smbubioside) nd totl phenolics (% clculted s gllic cid) of ll smples were from our previous study (Tble 1) [7]. Dt Anlysis Principl component nlysis (PCA), PLS, nd MLR were crried out using the Unscrmble 9.8 (Cmo Process AS, Figure 1: Ultrviolet/visible spectr of 50% methnol extrcts of ll smples in the rnge of nm 195

3 develop fst, simple, nd relible method for qulity grding in routine work, 60% of smples (trining set, n = 19) were rndomly selected, nd the rest smples (test set, n =13) were used to chllenge the method. All smples were extrcted with 50% methnol to ensure tht nthocynins nd other phenolic compounds could be dissolved in the solvent. UV/Vis spectr of the extrcts of ll smples were mesured, nd the results re shown in Figure 1. The difference between the spectr of the reddish nd brownish smples in the trining set ws studied by PCA. As shown in Figure 2, the score nd loding plots of PC1 versus PC2 explined for 97% of the vribles. The reddish nd brownish groups were clerly distinguished t the opposite side of the digonl between the xes of PC1 nd PC2 (Figure 2). The reddish smples were minly locted in the qurter between the positive xis of PC1 nd the negtive xis of PC2 which corresponded to the higher bsorbnces t 330 nd 530 nm, nd the lower bsorbnce t 380 nm (Figure 2b) thn the brownish smples. A 380 ws the bsorbnce of browning degrdtion products of nthocynins. It ws possibly interfered by other constituents nd shifted from generlly t 420 nm [18]. However, A 380 of the reddish group (0.260 ±0.040) ws not significntly lower thn tht of the brownish group (0.301 ± 0.051). A 330 were the mximum bsorption wvelengths of hydroxycinnmtes nd nthocynins, respectively, with little shifts from the references [19]. Even though A 330 of the reddish group (1.388 ± 0.214) ws higher thn tht of the brownish group (1.129 ± 0.123), their rnges ( nd , respectively) were overlpped. A 530 or coloring intensity [5] ws the only wvelength tht the rnges between the two groups were seprtely. They were (0.372 ± 0.129) nd (0.159 ± 0.045), respectively. A 530 of ll reddish smples were more thn Therefore, this vlue might be ssigned s cut-off criterion for the grding distinguishing between the qulified (reddish) nd the unqulified (brownish) roselle. However, their different ws very smll nd esy to error. Degrdtion index ws the other suggested prmeter. It hs been generlly used to explin the qulity of the nturl products contining nthocynins. It ws clculted from the rtio between A 420 nd A 520 [18]. In this study, A 380 were used for insted. Degrdtion indexes of the reddish nd brownish smples were (0.761 ± 0.221) nd (2.173 ± 1.309), respectively. Their difference ws obviously enlrged. The lower degrdtion index indicted the better qulity of the smple. The vlue of not more thn 1.2 might be suggested s the cut-off criterion between the qulified nd the unqulified smples. Using of both coloring intensity nd degrdtion index for qulity grding ws vlidted by the smples in the test set. The results re shown in Figure 3. Coloring intensity nd degrdtion indexes of ll reddish nd brownish smples in the test set were in ccordnce with bove qulity criteri. For more detil on the contents of totl nthocynins nd totl phenolic compounds, PLS models bsed on UV/Vis spectrl dt in the rnge of nm were developed, nd the results re shown in Figure 4. The optiml number of PLS fctors ws determined by the lowest root-mensqure error of cross-vlidtion (RMSECV) obtined from the cross-vlidtion process. The models were tested by the tested smple set nd the error of prediction ws expressed s root-men-squre error of prediction (RMSEP). The vlidtion sttistics of the models of totl nthocynins nd totl phenolics re shown in Tbles 2 nd 3, respectively. Both models used only 2 PLS fctors nd hd stisfied vlidtion results (RMSEP = nd , respectively). However, PLS model constructed from vrible-rich spectroscopic dt set compre to the number of smples possibly cused n overfitting problem nd misleding prediction results of unknown smples [20]. Therefore, only dt relly involving the models should be included. As indicted in the previous section, A 330, A 380, were the importnt dt discriminting between the qulities of roselle smples. High vlues of regression coefficients of these wvelengths in PLS models (Figure 4) lso suggested them s the high influencing vribles which referred to the qulity of the smples. Regression coefficients of A 330 were the positive vlues, wheres tht of A 380 ws negtive. These dt confirmed the results of PCA tht the qulified (reddish) smples hd higher A 330, nd lower A 380 thn the unqulified (brownish) smples. Then, PLS nd MLR models using only these three vribles were constructed. A better prediction ws obtined for the models of totl nthocynins (RMSEP = nd for PLS nd MLR models, respectively) (Tble 2). Only A 380 were significnt in MLR model. However, when only these two wvelengths were included in the model, b Figure 2: Principl component nlysis of trining set, () score plot nd (b) loding plot, using ultrviolet/visible spectr in the rnge of nm; R = Reddish smples, B = Brownish smples 196

4 the vlidtion result ws not stisfied (RMSEP = ). Prediction results of the tested smples using these models re shown in Tble 4. The effective nd simple models for the prediction of totl nthocynin content re s follows: Figure 3: () Coloring intensity nd (b) degrdtion indexes of reddish nd brownish smples of trining nd test set b b Figure 4: Prtil lest squre prediction models of () totl nthocynin content (% clculted s delphinidin 3-smbubioside) nd (b) totl phenolic content (% clculted s gllic cid) using ultrviolet/visible spectr in the rnge of nm Tble 2: Vlidtion prmeters of the prediction models of totl nthocynin content Method Wvelength (nm) Number of PLS fctor RMSECV R (LOOCV) RMSEP R (test) b PLS PLS 330, 380, MLR 330, 380, MLR 380, Correltion between mesured nd predicted vlues during leve one out cross vlidtion, b Correltion between mesured nd predicted vlues of test set. RMSECV: Root men squre error of leve one out cross vlidtion, RMSEP: Root men squre error of prediction, PLS: Prtil lest squre, MLR: Multiple liner regression Tble 3: Vlidtion prmeters of the prediction models of totl phenolic content Method Wvelength (nm) Number of PLS fctor RMSECV R (LOOCV) RMSEP R (test) b PLS PLS 330, 380, PLS 330, MLR 330, 380, Correltion between mesured nd predicted vlues during leve one out cross vlidtion, b Correltion between mesured nd predicted vlues of test set. RMSECV: Root men squre error of leve one out cross vlidtion, RMSEP: Root men squre error of prediction, PLS: Prtil lest squre, MLR: Multiple liner regression 197

5 PLS: Totl nthocynins = A A A 530 MLR: Totl nthocynins = A A A 530 P < For totl phenolic content, vlidtion result of PLS model estblished from A 330, A 380, (RMSEP = ) ws not much different from tht of A , nd the result of MLR model using A 330, A 380, ws worsened (RMSEP = ). A 330 were significnt in PLS model nd when nother model estblished from these two vribles, the sme result ws obtined (RMSEP = ). Prediction results of the tested smples using these models re shown in Tble 5. The effective nd simplest model for the prediction of totl phenolic content is s follow: PLS: Totl phenolics = A A 530 P < CONCLUSION This study found tht the UV/Vis bsorbnce t 330, 380, nd 530 nm of 50% methnol extrct were the importnt prmeters referring to the qulity of roselle. The specifiction bsed on the bsorbnce t 530 nm (coloring intensity) nd the rtio between the bsorbnces t 380 nd 530 nm (degrdtion index) ws set up to specify the qulity of the smples. PLS nd MLR models estblished from the bsorbnces of these three wvelengths were lso suggested to estimte the contents of totl nthocynins nd totl phenolic compounds. All suggested methods were relible, simple, nd rpid; used less chemicl nd could be pplied in routine work. Tble 4: Totl nthocynin content (% clculted s delphinidin 3 smbubioside) of tested smples using the prediction models Smple Mesured vlue PLS MLR nm 330,380,530 nm 330,380,530 nm 380,530 nm PLS: Prtil lest squre, MLR: Multiple liner regression Tble 5: Totl phenolic content (% clculted s gllic cid) of tested smples using the prediction models Smple Mesured vlue PLS MLR nm 330, 380, 530 nm 330, 380 nm 330, 380, 530 nm PLS: Prtil lest squre, MLR: Multiple liner regression 198

6 ACKNOWLEDGMENTS The uthors would like to thnk Fculty of Phrmcy, Silpkorn University, for provision of lbortory fcilities. REFERENCES 1. D-Cost-Roch I, Bonnlender B, Sievers H, Pischel I, Heinrich M. Hibiscus sbdriff L. - A phytochemicl nd phrmcologicl review. Food Chem 2014;165: Ali BH, Al Wbel N, Blunden G. Phytochemicl, phrmcologicl nd toxicologicl spects of Hibiscus sbdriff L: A review. Phytother Res 2005;19: Ptel S. Hibiscus sbdriff: An idel yet under-exploited cndidte for nutrceuticl pplictions. Biomed Prev Nutr 2014;4: Gurdiol S, Mch N. Therpeutic potentil of Hibiscus sbdriff: A review of the scientific evidence. Endocrinol Nutr 2014;61: Europen Phrmcopoeil Commission. Roselle. In: Europen Phrmcopoei. 7 th ed. Strsbourg, Frnce: Europen Directorte for the Qulity of Medicine (EDQM); p Plotto A. Hibiscus: Post-Production Mngement for Improved Mrket Access. INPhO - Post-Hrvest Compendium. Rome, Itly: Food nd Agriculture Orgniztion of the United Ntions; Sukwttnsinit T, Burn-osot J, Sotnphun U. Spectrophotometric method for quntittive determintion of totl nthocynins nd qulity chrcteristics of roselle (Hibiscus sbdriff). Plnt Med 2007;73: Sulimn AM, Ali AO, Idriss SE, Abdulrhmn MA. A comprtive study on red nd white Krkde (Hibiscus sbdriff L.) Clyces, extrcts nd their products. Pk J Nutr 2011;10: Elhssn EHAR, Ahmmed EM, Sirg N. Stndrdiztion of Roselle (Hibiscus sbdriff L.) Clyx cultivted in Sudn. J Med Plnts Res 2014;8: Nnm NM, Onyeke NG. Chemicl composition of two vrieties of sorrel (Hibiscus sbdriff L.), Clyces nd the drinks mde from them. Plnt Foods Hum Nutr 2016;71: Brinckmnn J. Medicinl Plnts nd Extrcts - Mrket News Service (MNS) Qurterly Edition. Genev: Interntionl Trde Centre; p Julini HR, Welch CR, Wu Q, Diouf B, Mliny D, Simon JE. Chemistry nd qulity of Hibiscus (Hibiscus sbdriff) for developing the nturl-product industry in Senegl. J Food Sci 2009;74:S Agribusiness in Sustinble Ntionl Africn Plnt Products. Mrket Survey: Hibiscus sbdriff. Avilble from: herbs.org/fric/hibiscus.htm. [Lst ccessed on 2003 Dec 23]. 14. Wong PK, Yusof S, Ghzli HM, Mn YB. Physico-chemicl chrcteristics of Roselle (Hibiscus sbdriff L.). J Nutr Food Sci 2002;32: Pesenözdoğn F, Orhn N, Ergun F. Studies on the conformity of Hibiscus sbdriff L. Smples from Turkish mrket to Europen phrmcopei. FABAD J Phrm Sci 2011;36: Abou-Arb AA, Abu-Slem FM, Abou-Arb EA. Physico-chemicl properties of nturl pigments (nthocynin) extrcted from Roselle clyces (Hibiscus subdriff). J Am Sci 2011;7: Cmelo-Méndez GA, Jr-Plcios MJ, Escudero-Gilete ML, Gordillo B, Hernnz D, Predes-López O, et l. Comprtive study of phenolic profile, ntioxidnt cpcity, nd color-composition reltion of Roselle cultivrs with contrsting pigmenttion. Plnt Foods Hum Nutr 2016;71: Wrolstd RE. Color nd Pigment Anlyses in Fruit Products. Sttion Bulletin 624. Oregon: Agriculturl Communictions, Oregon Stte University; Tsi PJ, McIntosh J, Perce P, Cmden B, Jordn BR. Anthocynin nd ntioxidnt cpcity in Roselle (Hibiscus sbdriff L.) Extrct. Food Res Int 2002;35: Tnk K, Kub Y, Sski T, Hiwtshi F, Komtsu K. Quntittion of curcuminoids in curcum rhizome by ner-infrred spectroscopic nlysis. J Agric Food Chem 2008;56: