Pelargonium Sidoides and Pelargonium Reniforme — a Quality Control Perspective ⁎ J.E

Available online at www.sciencedirect.com

South African Journal of Botany 82 (2012) 83–91 www.elsevier.com/locate/sajb

Phytochemical distinction between Pelargonium sidoides and Pelargonium reniforme — A quality control perspective ⁎ J.E. Maree, A.M. Viljoen

Department of Pharmaceutical Sciences, Faculty of Science, Tshwane University of Technology, Private Bag X680, Pretoria 0001, South Africa

Available online 31 July 2012

Abstract

Pelargonium sidoides is indigenous to South Africa and highly valued by traditional healers as a remedy to treat coughs, upper respiratory tract irritations and gastrointestinal conditions. An ethanolic extract of P. sidoides is used in the proprietary herbal tincture known as Umckaloabo®. Despite the commercial development of P. sidoides very few studies have been conducted to document the full phytochemical range of variation for natural populations and to date only one paper has been published on the development of a fast accurate quality control method for the validation of raw material. In this study P. sidoides (n=96) and P. reniforme (n=57) fresh root samples were collected from different localities in southern Africa. Phytochemical ﬁngerprints of the two species were generated by analytical techniques including high performance thin layer chromatography (HPTLC), liquid chromatography coupled to mass spectrometry (LC–MS), Fourier transform near infrared (FT-NIR), Fourier transform mid infrared (FT-MIR) spectroscopy, proton nuclear magnetic (1H NMR) spectroscopy and NIR hyperspectral imaging. Fingerprinting results revealed distinct phytochemical differences and similarities between the two species. The HPTLC and LC–MS chromatograms conﬁrm reports that P. reniforme does not contain umckalin. All analytical methods investigated showed the potential to discriminate between P. sidoides and P. reniforme. © 2012 SAAB. Published by Elsevier B.V. All rights reserved.

Keywords: Metabolomic ﬁngerprinting; Pelargonium sidoides; Pelargonium reniforme; Quality control

1. Introduction often adulterated with the close taxonomic ally P. reniforme (Brendler and Van Wyk, 2008). P. sidoides and P. reniforme can Pelargonium sidoides is a highly valued medicinal plant only be distinguished by the shape of the leaves and the colour of locally and internationally. The roots of this species are exported its flowers. P. sidoides has crowded velvety, cordate, long stalked to a German phyto-pharmaceutical company, Dr. Willmar leaves with short glandular hairs and dark reddish purple to Schwabe Pharmaceuticals, for the preparation of a 30% ethanol almost black flowers, whereas P. reniforme has grey-green extract (herbal tincture) known as Umckaloabo®. It has been velvety kidney shaped leaves with pink to magenta flowers developed into a commercially successful phytomedicine over (Kolodziej, 2000; Van Wyk and Wink, 2004). the years which is for the treatment of various illnesses (Brendler P. sidoides is widely distributed in the Eastern Cape, Free and van Wyk, 2008; Lewu et al., 2007). The safety and efficacy State, North West and Mpumalanga Provinces of South Africa of the herbal medicine have been clinically researched and is as well as in Lesotho with a distribution range of approximately proven curative for a range of diseases such as sinusitis, acute 600 000 km2 (Newton et al., 2011; Van der Walt and Vorster, bronchitis and common colds (Brendler and van Wyk, 2008). 1988). Pelargonium reniforme canonlybefoundinthedryflats At present most of the plant material used in the preparation of and grasslands extending between Knysna and Umtata (Van der the phytomedicine is still being wild harvested by locals from Walt, 1977). Wild harvesting of P. sidoides is reported to be rural communities (Lewu et al., 2007). The raw plant material is limited to the Free State, Eastern Cape Provinces and Lesotho. A study carried out by De Castro et al. (2010) found that in the Free ⁎ Corresponding author. Tel.: +27 12 3826360; fax: +27 12 3826243. State and Lesotho areas, P. sidoides population densities varied E-mail address: [email protected] (A.M. Viljoen). from 100 000 plants per 100 ha to 652 400 plants per 100 ha.

They also projected that the P. sidoides population density near (Schleicher & Schuell, United Kingdom). This procedure was Cathcart and Bedford is about 17.8 million plants per 50 000 ha repeated three times and the combined extracts were dried in a and 4 million per 45 000 ha (Newton et al., 2011). vacuum oven (VISMARA, Labotec, Switzerland) at 45 °C. Despite the commercial interest in P. sidoides and P. reniforme few studies have been conducted to document the phytochemical 2.1. Vibrational spectroscopy variation for natural populations. In addition, researchers believe that it is not only one compound that is responsible for its 2.1.1. Fourier transform infrared spectroscopy pharmacological properties (Kolodziej, 2000; Kolodziej and Approximately 0.5 g of the power was placed in Chromacol Kiderlen, 2007; Seidel and Taylor, 2004). Therefore it is vials (Chromacol Ltd, United Kingdom). The vials containing important to develop quality control systems and validation the samples were introduced to the spectrometer (Büchi procedures which reflect a greater part of the entire metabolome. NIRFlex N500 Fourier transform NIR spectrophotometer, Quality control analysis is vital in the pharmaceutical industry Büchi, Labortechnik AG, Switzerland) and the reflectance to guarantee authenticityandqualityofproducts(Luypaert et al., spectra of the samples were collected from 1000 to 2500 nm 2003; Sánchez et al., 2000). Correct identification of raw material (10000–4000 cm−1), with NIRWARE 1.2.3000 ADVANCED is one of the first steps in the quality and safety assurance of herbal Edition software, yielding 1501 data points. Each spectrum was medicine products. Identification methods should be able to the average of 32 scans. The mid infrared absorbance spectra of discriminate between related species and or potential adulterants the dried powdered samples were collected from 500 to (Laasonen et al., 2002). A major challenge in quality assurance of 4000 cm−1 by positioning it on the horizontal diamond crystal herbal material is the vast variation of active constituents in plants of the Alpha-P FT-IR spectrometer (Bruker Optik GmbH, from the same species. This variation is due to a range of factors Germany) equipped with a diamond attenuated total reflectance including locality, climate, age, harvesting season, storage (ATR) unit. The spectra were obtained with 2407 data points conditions and soil type (Laasonen et al., 2002). As a result of from the accumulation of 32 scans. The sample platform was this variation, the selection of only a few compounds as criteria cleaned between every sample using a 30% ethanol solution. for quality control is inadequate (Woo et al., 1999). SIMCA-P (11.0.0.0) software package (Umetrics, Sweden) was Metabolomic fingerprinting can be defined as a rapid analysis used for all statistical analysis. Second derivative (cubic method for the non-biassed identification and quantification of polynomial order, gap of 5), multiplicative scatter correction all metabolites in the biological system and can allow sample (MSC) and standard normal variate (SNV) were evaluated as classification and discrimination between closely related species. pre-treatment methods to address the non-linearity problem of the Metabolomic quality control protocols can be constructed using data. analytical methods such as chromatography and spectroscopy. Examples of chromatographic methods are high performance 2.1.2. Near infrared hyperspectral imaging liquid chromatography (HPLC) and liquid chromatography–mass Nine representative powdered samples (P. sidoides n= 5 and spectrometry (LC–MS) and examples of spectroscopic techniques P. reniforme n= 4) were selected based on the FT-MIR are Fourier transform infrared spectroscopy (FT-IR) and nuclear spectroscopy PCA model results (Maree and Viljoen, 2011)and magnetic resonance spectroscopy (NMR). Chromatographic and analysed by means of hyperspectral imaging. Hyperspectral spectroscopic techniques can be used in combination to overcome images were acquired using a SisuChema short wave infrared limitations of the individual methods (Ellis et al., 2007; García- (SWIR) camera (Specim, Spectral Imaging Ltd, Finland) with an Pérez et al., 2008). imaging spectrograph coupled to a 2D array mercury–cadmium– telluride (HgCdTe) detector. Images were acquired at a spectral 2. Materials and methods range of 1000 nm to 2500 nm with a 10 nm resolution, 6.3 nm spectral sampling per pixel and a view of field of 100 mm A total of 153 fresh root samples of P. sidoides (n=96) and ×100 mm. White (100% standard) and black (closed shutter) P. reniforme (n = 57) were collected from different natural references were captured prior to each image. Images were populations in Lesotho and the Eastern Cape Province (South analysed with Evince version 2.5.0 hyperspectral image analysis Africa) and retention samples and voucher material is retained in software (UmBio AB, Sweden). Image calibration and correction the Department of Pharmaceutical Sciences, Tshwane University to absorbance was done automatically in the Evince software. of Technology (for full collection information refer to Maree Principal component analysis (PCA) model with 3 components (2009)). The samples were air-dried at room temperature and was calculated, the background (irrelevant pixels) was removed milled to a powder using a hammer mill (Pore size: 0.75 mm, and the model was recalculated. PCA score plots, score images and Dietz Motoren GmbH, Germany) and sieved (500 μm, Endecotts loading line plots were investigated to identify regions of interest. Ltd, United Kingdom) to obtain uniform particle size. Sixty-nine Partial least squares discriminant analysis (PLS-DA) model was samples (P. sidoides n=48; P. reniforme n=21) were selected to calculated (trainingset n = 80% of total pixels; testset n = 20% of represent the two species from different populations. One gram of total pixels), with species as y-variable (class 1: P. sidoides and each sample was placed into an Erlenmeyer flask with 50 ml class 2: P. reniforme), to discriminate between the pixel spectra of of MeOH:CH2Cl2 (1:1, v/v). The mixture was placed in a the two species. PLS-DA model is a classical partial least-squares water bath (Thermo Scientific, South Africa) at 40 °C for 1 h analysis (PLS) regression model where the Y-variables are and subsequently filtered with Whatman® filter paper No. 41 represented by distinct dummy variables describing the class J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91 85 membership of the statistical units (Menesatti et al., 2009). Two Table 1 – independent samples (P. sidoides n=1 and P. reniforme n=1) Gradient profile used during the LC MS analysis. were used as test matrix. The constructed PLS-DA model was Time (min) Solvent A (0.1% formic acid) Solvent B (Acetonitrile) used to classify each pixel of the test matrix. Colour prediction 0.00 100.0 0.0 maps were constructed where each pixel was classified and 7.00 75.0 25.0 coloured according to predicted class. 13.00 0.0 100.0 14.00 0.0 100.0 14.10 100.0 0.0 2.2. High performance thin layer chromatography 16.00 100.0 0.0

The extracts were reconstituted with methanol to yield a final concentration of 10 mg/ml. Two microliters of the sample was spotted in a 10 mm band with a CAMAG Automatic TLC Sampler 4 (ATS4, CAMAG, Switzerland) on 20 cm×10 cm described by Kim et al. (2010). In brief, 10 mg dried sample was pre-coated silica gel glass plates (HPTLC plates silica gel 60 F254, transferred into a 2 ml Eppendorf tube, to which 0.75 ml Merck, Switzerland). The HPTLC plates were developed in a CH3OH-d4 and 0.75 ml D2O(KH2PO4 buffer, pH 6.0), CAMAG automatic developing twin trough glass chamber (ADC containing 0.005% (w/v) TMSP-d4 were added. The mixture was 2, CAMAG, Switzerland) up to 85 mm using water:methanol: vortexed at room temperature (25 °C) for 1 min, ultrasonicated ethylacetate (10:14:76, v:v:v) as mobile phase. The chamber was (Sonorex Digital 10P, Bandelin GmbH & Co. KG, Germany) for saturatedwiththesamesolventasthemobilephaseatroom 20 min and then centrifuged at 13,000 rpm at room temperature temperature (20 °C ±2 °C) before development for 20 min, using for 20 min. The supernatant (0.75 ml) from each tube was a saturation pad. The plate was automatically dried by the ADC 2 transferredtoa5mmNMRtubeandusedforNMRanalyses. instrument for 5 min with cold air. The separated bands on the 1HNMRspectrawererecordedat25°Cona600MHzBruker HPTLC plate were viewed and recorded at 254 nm and 366 nm DMX-600 spectrometer (Bruker, Germany); operating at a proton with the CAMAG Reprostar 3. CAMAG winCats™ 1.4.1 planar NMR frequency of 600.13 MHz. CH3OD was used as internal chromatography manager software (CAMAG, Switzerland) was lock. 1H NMR data spectra were processed using MestReNova used in all the steps to control the instruments and to keep a software (version 7.1.2, Mestrelab Research, Spain). The resulting record of all analyses. spectra were manually phased, baseline corrected, calibrated at 0.00 ppm to TMSP. The regions of δ 3.28–3.34 were removed 2.3. Liquid chromatography–mass spectrometry analysis because of the residual signal of CH3OD.

All extracts were re-dissolved in a solution of 50% acetonitrile 3. Results and discussion (ACN) and 0.1% formic acid in water to yield a concentration of 50 mg/ml. Two microliters of each sample was injected into a 3.1. Vibrational spectroscopy Waters Alliance 2690 HPLC system (Waters, United Kingdom) with a Waters BEH C18 column, 1.7 μm (2.1 mm×50 mm) 3.1.1. Fourier transform infrared spectroscopy coupled to a 996 photodiode array (PDA) detector and a Waters The chemical composition of P. sidoides and P. reniforme are API Quattro Micro MS detector (Waters, United Kingdom). The very complex. It is not possible to establish a direct correlation PDA detector was set to scan a wavelength range of 100 nm to between the chemical composition and FT-NIR spectra of the 1400 nm. The MS detector was operated in electron impact mode species due to the complexity of overlapping overtones and with the capillary voltage at 3.5 kV, the cone voltage at 35 kV combination bands in the NIR region. It is difficult to ascribe (positive switching — ES+). The flow rate of the HPLC was specific bands of variation between the two species in the raw 0.3 ml/min, the nebuliser flow rate 50 l/h and the desolvation gas FT-NIR spectra, therefore the difference between the species flow rate 350 l/h. The source temperature was 100 °C and the spectra is not apparent. Hence mathematical data pre-processing is desolvation temperature 350 °C. The mobile phase used consisted needed. Figure 1 shows the average reflectance spectra of P. of 0.1% formic acid (solvent A) and acetonitrile (solvent B). The sidoides and P. reniforme. The baseline drift is mostly due to gradient profile used over the 16 min period is shown in Table 1. particle size differences of the samples since FT-NIR spectra also The Masslynx™ (Version 4.0) software package (Waters, United reflect the physical properties of the samples. Examples of Kingdom) and Microsoft® Excel (2003) was used to analyse and baseline shift can be seen in Fig. 1 at wavelengths 1316 nm– graphically display the results. Peaks of interest were identified in 1923 nm and 2000 nm–2500 nm. the ES chromatograms. The specific mass was extracted from Multiplicative scatter correction and second derivative were these peaks using the combined spectra function, which allows used to remove baseline drift and enhance the discrete spectral for background noise to be eliminated. features. The spectral characteristics associated with the different species subsequently became more apparent and is illustrated in 2.4. 1H NMR spectroscopy Fig. 2. Comparison of the spectra between the two species showed similar absorption peaks but with different peak intensities. This Twelve selected root samples of P. sidoides (n = 6) and suggests that similar chemical constituents are present in both P. reniforme (n=6) were extracted following the protocol species but at different concentrations (quantitative differences) 86 J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91

0.6 P. sidoides P. reniforme 0.5

0.4

0.3 Absorbance 0.2

0.1

0 1000 1020 1042 1064 1087 1111 1136 1163 1190 1220 1250 1282 1316 1351 1389 1429 1471 1515 1563 1613 1667 1724 1786 1852 1923 2000 2083 2174 2273 2381 2500 Wavelength (nm)

Fig. 1. Average raw (no pre-treatment) FT-NIR spectra of ground dried roots of P. sidoides (red) and P. reniforme (green).

(Liu et al., 2008). Examples of quantitative differences between the 3.1.2. Near infrared hyperspectral imaging species can be observed at the following wavelengths: 1730, 1891, After removing all irrelevant pixels, standard normal variate 1917 and 2145 nm (Fig. 2). The qualitative differences (presence (SNV) and univariate (UV) scaling was applied to all data and a 2 or absence of certain peaks in the two spectra) (Liu et al., 2008) PCA model with three PCs (R Xcum =0.9993) was constructed. were obtained at wavelengths 1751, 1956 and 1984 nm (Fig. 2). Three distinct clusters were visible in the scatter score plot (Fig. 5a) Characteristic FT-MIR spectra of P. sidoides and P. reniforme of PC 1 (R2X=0.9974) vs. PC 2 (R2X=0.00152). Inter-species are illustrated in Fig. 3. The spectra of the two species are very variation can be observed along the first PC (x‐axis) (Fig. 5a,b) similar. Chemical compositional variation (qualitative differences) while P. sidoides intra-species variation can be detected along the can only be observed in spectral pre-treated data (multiplica- 2nd PC (y‐axis). A three component PLS-DA model with an – 2 tive scatter correction and second derivative) between 1328 R Ycum of 0.8986 and RMSEP of 0.4790 (P. sidoides) and 0.4912 − − − 1341 cm 1,855–878 cm 1 and 626–655 cm 1 (Fig. 4). Quan- (P. reniforme) was constructed to predict the identity of unknown titative differences in the spectra are visible between 1610– samples. These model performance statistics indicates that the − − 1648 cm 1 and 803–770 cm 1. The comprehensive application PLS-DA model performed well when applied at pixel level and of vibrational spectroscopy and chemometrics to discriminate could be used to accurately classify unknown samples. The colour between the two species has recently been published (Maree and prediction map of independent test matrices (Fig. 6) showed that Viljoen, 2011). 129 pixels of P. sidoides sample (total number of pixels=1725)

5 P. sidoides P. reniforme 4

-1

-2

-3

-4 1600 1620 1641 1663 1685 1708 1732 1756 1781 1807 1834 1861 1889 1918 1948 1979 2011 2044 2078 2113 2149 2187 2226 2266 2308 2351 2396 2443 2492

Wavelenght (nm) SIMCA-P 11 - 2009/05/21 02:36:35 PM

Fig. 2. Average FT-NIR spectra of ground dried roots of P. sidoides (red) and P. reniforme (green) after treatment with second derivative and multiplicative scatter correction (1600–2500 nm) illustrating quantitative and qualitative differences between the two species. J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91 87

P. sidoides 0.20 P. reniforme

0.18

0.16

0.14

0.12

0.10

Absorbance 0.08

0.06

0.04

0.02 597. 739. 880. 3430 3855 3572 3288 3147 3005 2863 2438 2155 1872 1730 1589 1447 1305 1164 1022 3995 3713 2722 2580 2014 2297 Wavenumber (cm^-1) SIMCA-P 11 - 2010/07/27 01:46:23 PM

Fig. 3. Average raw (no pre-treatment) FT-MIR spectra of dried ground roots of P. sidoides (red) and P. reniforme (green).

were misclassified as P. reniforme and 210 pixels could not be 0.34, 0.42, 0.71 and 0.84 (Fig. 7). Umckalin (Dr. Willmar assigned to a specific class, while 5 pixels P. reniforme (total Schwabe Pharmaceuticals, Germany) and scopoletin (Sigma- number of pixels=2486) sample were misclassified as P. sidoides Aldrich®, South Africa) (marker compounds) had Rf values of and 352 pixels could not be classified (Table 2). 0.76 and 0.71, respectively. The results obtained from the HPTLC chromatogram confirm reports that P. reniforme does not 3.2. High performance thin layer chromatography analysis contain umckalin (Kolodziej, 2007; Latté et al., 2000).

A single HPTLC system can be used to differentiate between 3.3. Liquid chromatography–mass spectrometry analysis P. sidoides and P. reniforme. Viewing the chromatogram under 366 nm (UV) showed that P. sidoides separated into seven The LC–MS chromatogram examination of P. sidoides bands while P. reniforme separated into six bands. Reported and P. reniforme resulted in the detection of 18 diverse and phytochemical similarities, at different concentrations, do exist distinctive major peaks. LC–MS chromatograms of the samples between the two species and can be seen at Rf values 0.11, 0.26, representative of the two species are shown in Fig. 8.The

P. sidoides 0.0005 P. reniforme

0.0004

0.0003

0.0002

0.0001

0.0000

-0.0001

-0.0002

-0.0003

-0.0004

-0.0005

-0.0006 1800 1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 600 Wavenumber (cm^-1)

Fig. 4. Average FT-MIR spectra of ground dried roots of P. sidoides (red) and P. reniforme (green) after treatment with second derivative and multiplicative scatter correction (600–1800 cm−1) illustrating qualitative differences between the two species. 88 J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91

Fig. 5. (a) PCA score plot of PC1 (t[1]) vs. PC2 (t[2]) illustrating inter- and intra-species variation (P. sidoides red ellipse; P. reniforme green ellipse); (b) corresponding PC1 score image with P. sidoides samples bordered in red and P. reniforme bordered in green. chromatograms of the two species exhibited similar compounds used to discriminate between the two species. LC–MS analysis and can be seen at peak numbers 1, 4, 6, 8, 9, 14 and 16 (Table 3). also proved that umckalin is only present in P. sidoides. However, phytochemical differences between the two species do Umckalin with molecular mass of 222 g/mol, was detected at a exist and can be observed at peak numbers 2, 3, 5, 7, 10–13, 15, retention time of 8 min after mass extraction. 17 and 18. These results proved that LC–MS analysis could be 3.4. 1H NMR spectroscopy

The average 1HNMRspectraofP. sidoides (red) and P. reniforme (green) are shown in Fig. 9.The1HNMR spectra of both species are very complex in the regions between 4.0 and 3.5 ppm and no differences between the two species were visible. The spectra of the two species were

Table 2 Classification of pixels per image of the two independent test matrices. % pixels classified P. sidoides P. reniforme Unclassifiable Fig. 6. Colour prediction maps of two independent hyperspectral images a) P. sidoides 80.3% 7.48% 12.2% (a) P. sidoides and (b) P. reniforme where each pixel is coloured according to its b) P. reniforme 0.201% 85.6% 14.2% classified class (P. sidoides is green, P. reniforme is blue and unclassifiable is red). J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91 89

PS31S PPS32 PS334 PS38 PPS42 PS45 PSS48 PPS50 PS633 PSS64 11 12 PS66 PS69 PS71 PS722 PS774 PSS77 PS788 PS880 PS81PS82S 11 12 b

PRR1 PR5 PR77 PRR9 PR10 PR13 PRR14 PR16 PR22 PRR28 11 12 PR31 PR366 PR39 PR40 PR422 PR447 PR50 PR51 PR552 PRR53 11 12

Fig. 7. Chromatographic separation of a) P. sidoides and b) P. reniforme viewed under 366 nm, illustrating phytochemical variation between species. Standards: umckalin (track 11) and scopoletin (track 12). predominantly similar with the obvious exception in chemical 4. Conclusion shift regions of 8.4–8.2 ppm (aromatic region), 6.5–6.2 ppm (olefinic region) and 2.6–2.4 ppm (alkene region). Table 4 is a Development of phytochemical fingerprints of the two species summary of the discriminative regions between the species. by means of HPTLC, LC–MS, FT-IR spectroscopy, 1H NMR

100 a

15.30 3.30 249.0 305.1 7.62 301.0 10.71 % 3.01 415.2 2.63 358.1 328.0 12.24 9.54 564.4 13.68 305.1 617.4 4.48 4.16 289.1 12.05 305.1 6.91 9.95 311.2 1.20 5.96 551.2 329.2 341.1 307.1 8.96 289.1 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00 16.00

4.15 100 305.1 3.51 609.1 15.31 10.67 399.2 b 415.2

3.30 305.1 9.14 585.2 % 1.20 5.94 3.07 307.1 341.1 609.1 9.03 585.2 13.67 2.94 617.4 609.1 9.57 599.2 12.98 399.2 2.66 0.95 609.1 9.97 249.0 6.90 8.13 329.2 551.2 425.1 0 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00 16.00

Fig. 8. LC–MS chromatogram fingerprints in the negative ion mode of (a) P. sidoides and (b) P. reniforme (inset: umckalin). 90 J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91

Table 3 Table 4 Characteristic peaks of LC–MS fingerprints of P. sidoides and P. reniforme. Summary of discriminative chemical shifts of P. sidoides and P. reniforme. Peak no Retention time (min) P. sidoides P. reniforme Chemical shift (ppm) P. sidoides P. reniforme 1 1.20 (341) a √√8.40 (doublet) √ X 2 2.63 (328) √ x 8.25 (doublet) √ X 3 3.01 (358) √ x 6.45 (multiplet) √ X 4 3.30 (305) √√6.30 (multiplet) √ X 5 3.50 (305) x √ 2.55 (doublet) √ X 6 4.16 (305) √√2.50 (singlet) √ X 7 4.47 (289) √ x 8 5.95 (307) √√ 9 6.91 (551) √√accurate non-laborious uncomplicated technique with little √ 10 7.61 (301) x data processing needed. The challenge remains to establish if 11 9.14 (585) x √ 12 9.55 (305) √ x these methods have sufficient sensitivity to determine the level 13 9.57 (599) x √ of adulteration in a mixture containing both P. sidoides and 14 10.64 (415) √√P. reniforme. 15 12.24 (564) √ x 16 13.68 (617) √√ 17 15.30 (249) √ x Acknowledgements 18 15.31 (399) x √ a Base peak mass (m/z−1). The authors thank Tshwane University of Technology (TUT) and the National Research fund (NRF) for funding this project. The authors wish to thank Dr H. Kim, Dr Y. Choi and spectroscopy and NIR hyperspectral imaging revealed the Prof R. Verpoorte from the University of Leiden for their existence of distinct phytochemical differences between the two assistance and use of their NMR facility. species. Any of these methods can be used during quality control of raw materials to identify the specific species. NIR hyperspectral References imaging and FT-IR spectroscopy are more rapid, non-destructive and cost effective quality control methods but the difficulty of data Brendler, T., Van Wyk, B.-E., 2008. A historical, scientific and commercial interpretation and complex calibration procedures makes these less perspective on the medicinal use of Pelargonium sidoides (Geraniaceae). Journal attractive. On the other hand, HPTLC and 1HNMRspectroscopy of Ethnopharmacology 119, 420–433. are faster techniques as qualitative differences can be detected De Castro, A., Vlok, J., Mcllelan, W., 2010. Field survey of the distribution of 1 Pelargonium sidoides and size of selected sub-populations. Resource easier by means of visual inspection. Of the latter two, HNMR Assessment study conducted for the South African National Biodiversity spectroscopy data analysis is more complicated and time Institute. consuming. When determining the quality of raw material, Ellis, D.I., Dunn, W.B., Griffin, J.L., Allwood, J.W., Goodacre, R., 2007. LC–MS is a more sensitive and accurate method to quantify Metabolic fingerprinting as a diagnostic tool. Pharmacogenomics 8, 1243–1266. marker compounds. The results obtained from the HPTLC and García-Pérez, I., Vallejo, M., García, A., Legido-Quigley, C., Barbas, C., 2008. Metabolic fingerprinting with capillary electrophoresis. Journal of Chroma- LC–MS chromatogram confirm reports that only P. sidoides tography. A 1204, 130–139. contains umckalin. FT-MIR spectroscopy is suggested as best Kim, H.K., Choi, Y.H., Verpoorte, R., 2010. NMR-based metabolomic analysis option for the first step in quality control because it is an of plants. Nature Protocols 5, 536–549.

2200 P. sidoides P. reniforme 2000 1800 1600 1400 1200 1000 800 600 400 200 0

10 9 8 7 6 5 4 3 2 1 Chemical shift (ppm) SIMCA-P+ 12.0.1 - 2012-04-11 13:39:37 (UTC+2)

Fig. 9. Average 1H NMR spectra of P. sidoides (red) and P. reniforme (green) root extracts at 600 MHz. J.E. Maree, A.M. Viljoen / South African Journal of Botany 82 (2012) 83–91 91

Kolodziej, H., 2000. Traditionally used Pelargonium species: chemistry and sidoides and its close taxonomic ally P. reniforme. Vibrational Spectroscopy 55, biological activity of umckaloabo extracts and their constituents. Current Topics 146–152. in Phytochemistry 3, 77–93. Menesatti, P., Zanella, A., D'Andrea, S., Costa, C., Paglia, G., Pallottino, F., Kolodziej, H., 2007. Fascinating metabolic pools of Pelargonium sidoides and 2009. Supervised multivariate analysis of hyper-spectral NIR images to Pelargonium reniforme, traditional and phytomedicinal sources of the herbal evaluate the starch index of apples. Food and Bioprocess Technology 2, medicine Umckaloabo®. Phytomedicine 14 (Suppl. 1), 9–17. 308–314. Kolodziej, H., Kiderlen, A.F., 2007. In vitro evaluation of antibacterial and Newton, D., Raimondo, D., Motjotji, L., Lippai, C., 2011. National Environmental immunomodulatory activities of Pelargonium reniforme, Pelargonium sidoides Management: Biodiversity Act (10/2004): Draft Biodiversity Management and the related herbal drug preparation EPs® 7630. Phytomedicine 14 (Suppl. 1), Plan for Pelargonium sidoides. Retrieved from http://www.environment.gov. 18–26. za/sites/default/files/gazetted_notices/biodiversitymanagement_pelargonium_ Laasonen, M., Harmia-Pulkkinen, T., Simard, C.L., Michiels, E., Räsänen, M., sidoidesplan.pdf. Vuorela, H., 2002. Fast identification of Echinacea purpurea dried roots Sánchez, M.S., Bertran, E., Sarabia, L.A., Ortiz, M.C., Blanco, M., Coello, J., using near-infrared spectroscopy. Analytical Chemistry 74, 2493–2499. 2000. Quality control decisions with near infrared data. Chemometrics and Latté, K.P., Kayser, O., Tan, N., Kaloga, M., Kolodziej, H., 2000. Unusual Intelligent Laboratory Systems 53, 69–80. coumarin patterns of Pelargonium species forming the origin of the traditional Seidel, V., Taylor, P.W., 2004. In vitro activity of extracts and constituents of herbal medicine umckaloabo. Zeitschrift fur Naturforschung - Section C Pelagonium against rapidly growing mycobacteria. International Journal of Journal of Biosciences 55, 528–533. Antimicrobial Agents 23, 613–619. Lewu, F.B., Adebola, P.O., Afolayan, A.J., 2007. Commercial harvesting of Van der Walt, J.J.A., 1977. Pelargoniums of Southern Africa, vol. 1. Purnell & Pelargonium sidoides in the Eastern Cape, South Africa: striking a balance Sons, Cape Town, pp. 40–41 (ISBN: 0702112208). between resource conservation and rural livelihoods. Journal of Arid Van der Walt, J.J.A., Vorster, P.J., 1988. Pelargoniums of Southern Africa, vol. Environments 70, 380–388. 3. Juta, Kirstenbosch National Botanic Gardens, Cape Town, pp. 129–131 Liu, H.-X., Zhou, Q., Sun, S.-Q., Bao, H.-J., 2008. Discrimination of different (ISBN: 062011074). Chrysanthemums with Fourier transform infrared spectroscopy. Journal of Van Wyk, B.-E., Wink, M., 2004. Medicinal Plants of the World: An Illustrated Molecular Structure 883–884, 38–47. Scientific Guide to Important Medicinal Plants and Their Uses. Timber Luypaert, J., Zhang, M.H., Massart, D.L., 2003. Feasibility study for the use of Press. near infrared spectroscopy in the qualitative and quantitative analysis of Woo, Y., Kim, H., Cho, J., Chung, H., 1999. Discrimination of herbal medicines green tea, Camellia sinensis (L.). Analytica Chimica Acta 478, 303–312. according to geographical origin with near infrared reflectance spectroscopy Maree, J.E., 2009. Development of a quality control protocol for Pelargonium and pattern recognition techniques. Journal of Pharmaceutical and Biomedical sidoides DC using Fourier transform infrared spectroscopy. MTech Thesis. Analysis 21, 407–413. Pretoria, Tshwane University of Technology. Maree, J.E., Viljoen, A.M., 2011. Fourier transform near- and mid-infrared spectroscopy can distinguish between the commercially important Pelargonium

Edited by I Vermaak