Phytochemical Comparison of the Water and Ethanol Leaf Extracts of the Cree medicinal plant, Sarracenia purpurea L. (Sarraceniaceae).

PURPOSE
The Cree of Eeyou Istchee in Northern Quebec identified Sarracenia purpurea L. as an important plant for the treatment of Type 2 diabetes. Traditionally the plant is used as a decoction (boiling water extract) of the leaf, however, in order to study the extract in a laboratory setting, an 80% ethanol extract was used. In this study, the phytochemistry of both extracts of the leaves was compared and quantified.


METHODS
Two S. purpurea leaf extracts were prepared, one a traditional hot water extract and the other an 80% ethanol extract. Using UPLC-ESI-MS, the extracts were phytochemically compared for 2 triterpenes, betulinic acid and ursolic acid, using one gradient method and for 10 additional substances, including the actives quercetin-3-O-galactoside and morroniside, using a different method.


RESULTS
The concentrations of the nine phenolic substances present, as well as an active principle, the iridoid glycoside morroniside, were very similar between the two extracts, with generally slightly higher concentrations of phenolics in the ethanol extract as expected. However, two triterpenes, betulinic acid and ursolic acid, were 107 and 93 times more concentrated, respectively, in the ethanol extract compared to the water extract.


CONCLUSION
The main phytochemical markers and most importantly the antidiabetic active principles, quercetin-3-O-galactoside and morroniside, were present in similar amounts in the two extracts, which predicts similar bioactivity.This article is open to POST-PUBLICATION REVIEW. Registered readers (see "For Readers") may comment by clicking on ABSTRACT on the issue's contents page.


INTRODUCTION
Traditional medicine continues to be an important part of the culture for the Cree First Nation of Eeyou Istchee in Northern Quebec. They use plants to treat the symptoms and causes of many diseases, and traditionally prepare most of their plant extracts as decoctions (boiling water extracts). In 2003, a number of Cree First Nations of Eeyou Istchee as well as the Cree Health Board initiated a research project in collaboration with six university laboratories, named the CIHR Team in Aboriginal Antidiabetic Medicines (CIHR-TAAM) (http://www.taam-emaad.umontreal.ca/), to evaluate the safety and efficacy of culturally relevant medicines used to treat symptoms of Type 2 diabetes (1,2). This disease is particularly important as the prevalence of diabetes is greater than 20% in the Cree adult population (3) and compliance with conventional therapies is low.
It is well known that different solvents are able to extract different families of phytochemicals based on their polarity, hence varying the biological activity of the extracts (4)(5)(6). Alcohol extracts are preferred to water extracts in many pharmacological assays, as many lipophilic bioactive secondary metabolites, notably phenolics are more efficiently extracted than with water. In addition to the phytochemicals found in ethanolic extracts, hot water also tends to extract other biological components (7), such as hydrophilic polysaccharides and denatured proteins, which may interfere with some assays (8). However, the phytochemistry of the two types of plant extracts is _________________________________________ very seldom compared in medicinal plant research, even when traditionally prepared boiled water extracts are used for therapy.
One of the main plant species identified by the Cree for its antidiabetic properties was Sarracenia purpurea L. (Sarraceniaceae), the pitcher plant, also known as Ayigadash in Cree (1). It is a carnivorous plant found to inhabit bogs and other nitrogen-poor environments (9). A quantitative ethnobotanical survey by Leduc et al. (1) showed that S. purpurea was an important plant species used to treat the symptoms of diabetes by the Cree, and it was identified as one of the priority plant species for the TAAM to study. Therefore, a robust phytochemical comparison of the traditional water and the ethanol extracts was necessary to examine the targeted substances previously identified in S. purpurea (10) (Figure 1). The purpose of this work was to quantify the differences, particularly for those substances known to have antidiabetic activity.

Plant extracts
S. purpurea L. leaves were collected in August 2008 in a bog in the Eastern James Bay Region in Quebec, Canada according to healer's and Elder's instructions and under permit from the TAAM -Cree Nation Intellectual Property agreement. Its identity was confirmed by Dr. A. Cuerrier from the Montreal Botanical Garden and a voucher specimen #2003-05 was deposited at the Marie-Victorin herbarium at the Montreal Botanical Garden (1,2). The pitchers of the plant were cut open and cleaned of any potential insects or debris and dried in an electric food dehydrator (Nesco/American Harvest, Two Rivers, WI, USA) at 40˚C. The leaves were then ground to a powder using a Thomas Wiley Mill (1 mm filter) (Thomas Scientific, Swedesboro, NJ, USA). Subsequently, two different types of extracts were prepared.
The ethanol extract was prepared using 80% ethanol in a ratio of 10 mL per gram of dry ground plant material extracted at room temperature overnight under gentle agitation. The extract was then separated from the plant residue by vacuum filtration using a Whatman qualitative no.1 filter (11 µm pore size) (Whatman plc, Maidstone, Kent, UK). The residue was then extracted two more times using the same method described above. The filtered extracts were then combined and the ethanol evaporated using a rotary evaporator at 40°C and residual water removed through lyophilisation. The water extract was prepared by gently boiling the plant material in distilled water for one hour in a ratio of 10 mL per gram of dry plant material to approximate the preparation method used by the Healers. The extract was then separated from the plant residue as above and dried using a B-191 Büchi Mini Spray Dryer (Büchi, Switzerland) operating with an inlet temperature of 150ºC and an air flow rate of 600 L/hour. Both the ethanol and water extracts were ground with a mortar and pestle into homogenous powders and resuspended in methanol and filtered prior to analysis.

Ultra performance liquid chromatographyelectrospray ionization -mass spectrometry (UPLC-ESI-MS)
The analytical methodology was based on Saleem et al. (12) with slight modifications. Briefly, the identification and quantification of the 12 target substances was performed using a Shimadzu UPLC-MS system (Mandel Scientific Company Inc, Guelph, ON, Canada). The system consisted of a binary pump (LC30AD), a column thermostat (CTO20a), a high performance autosampler (SIL-30AC), a photodiode array detector (SPD-M20A) and a mass selective detector LC2020. Electrospray source was operated in positive and negative ionization modes ( Table 1). The nebulizing gas flow was set at 1.5 L/min and drying gas flow was at 10 L/min. The desolvation line temperature and heat block temperature were set at 300°C and 450°C respectively. Identification of separated substances was undertaken by co-chromatographic comparison of the retention time and mass data with commercially available purified standards. A Shimadzu Lab Solutions software (5.41.240) was used for the quantification of identified substances. Standard curves (5 to 100 ng) were built by injecting dilutions of the standard substances in order to obtain peak areas to automatically quantify the amount of each marker substance. Each sample was analyzed in triplicate and peak response averaged to account for instrumental variation; the r 2 values were all greater than 0.99 (data not shown).
Two separation gradients were employed. Method A for the triterpenes including betulinic acid and ursolic acid, and method B for the phenolics. Separations were performed on a Phenomenex Kinetex C18 column (2.6 μm particle size, 100 mm x 2.1 mm, Phenomenex Inc., Torrence, CA, USA) with corresponding guard column to achieve enhanced separation. The injection volumes were 1 µL, flow was set to 0.4 mL/min with a column temperature of 55°C. Method A, isocratic elution method with 62.5% acetonitrile (ACN) with 37.5% water for 5 min. Method B, gradient elution method with 0.1% formic acid (solvent A) and 70% ACN with 30% MeOH and 0.1% formic acid (solvent B). The gradient was initially set at 86% A for 5.5 min, followed by 65% A isocratic for 2 min. The column was then washed for 2 min with 100% of solvent B and then returned to initial condition.

Lipophilicity
The partition coefficient logP, a measure of lipophilicity, was evaluated using ChemBioDraw Ultra 13.0 software (Perkin Elmer, Waltham, MA, USA).

STATISTICAL ANALYSIS
Data were analysed using GraphPad Prism 6.0 software (GraphPad Software Inc., La Jolla, CA, USA) using a multiple t-test comparing the concentration of each marker substance between the extracts. A false discovery rate was set to 1%.

RESULTS
The water extraction produced a powder at a yield of 20.6% w/w dried plant material, while the ethanol extraction produced a resinous/glassy extract at a yield of 36.9% w/w. Two different UPLC-ESI-MS methods used to analyze the ethanol and water extracts of the leaves of S. purpurea provided good chromatographic resolution ( Figure  2) of the 12 targeted substances. In the ethanol extract, the peaks representing betulinic acid (5), ursolic acid (6), morroniside (2), (-)-epicatechin (3), taxifolin-3-O-glucoside (4) and quercetin-3-Ogalactoside (7) were present in high enough quantities to be visible on the chromatogram. Their identification in the extracts was achieved based on selective ion monitoring for particular mass over charge (m/z) and retention times (Table 1) compared to pure standards. In the water extract, the same substances were identifiable, however at a lower concentration.
As shown in Figure 2, the relative abundance of the phytochemicals varied between the water and ethanol extracts. This was quantified in Table 2, where the amount of (+)-catechin (1) and morroniside (2)   The triterpenes, betulinic acid (5) and ursolic acid (6), were also more abundant in the ethanol extract (p<0.001), and here the magnitude of the difference was much greater as they were about 107 and 93 times more abundant in the ethanol extract, respectively. Kaempferol-3-O-rutinoside (12) and rutin (8), minor substances in S. purpurea, were detected in trace amounts and below the limit of accurate quantitation. On the other hand, luteolin-7-O-glucoside (10) and myricetin (11) were not detected in the samples analyzed. Evaluation of lipophilicity as determined by logP values showed that the majority of the substances had values between -2.28 and 1.5 (Table  3). This included all the phenolics and the iridoid glycoside, morroniside (2). The two triterpenes, betulinic acid (5) and ursolic acid (6) had significantly higher logP values of 7.38 and 7.33, respectively.

DISCUSSION
This study is the first comparing the phytochemical content of S. purpurea extracts prepared in two different ways. The water extraction used in this study was consistent with the traditional preparation used by the Cree of Eeyou Istchee whereas the ethanol extract is preferentially used in most pharmacological studies examining the activity of medicinal plants. This comparison was important for the Cree Healers as they wanted to know that the phytochemicals present and active in these assays were also the same ones present in their traditional preparation. It was shown here that the 80% ethanol extract generally results in significantly higher quantities of phenolics present, as well as the triterpenes. The only exceptions were the two most polar substances present (+)-catechin (1) and morroniside (2).
This result was not surprising as higher quantities of phenolics have been consistently seen by other groups in alcohol extracts (7,13,14). In fact, this is one of the reasons why alcohol extracts are preferred in bioassays since relevant biological activities, such as antioxidant activity, are easier to observe in extracts with a greater phenolic concentration. Alcohols, being less polar solvents, are able to break down plant cell walls more efficiently and hence extract the necessary components (13). However, pure alcohol extracts are less efficient than alcohol/water mixtures (15). The 80% ethanol/water mixture normally used for extraction was optimized previously in the laboratory for phenolic extraction (16,17). Also, it is understandable that (+)-catechin (1) and morroniside (2) were present in significantly higher quantities in the water extract. They are the two most polar marker substances examined, as can be Table 2. Quantification of phytochemicals in ethanol and water extracts of Sarracenia purpurea using UPLC-ESI-MS. The concentration of each compound in each extract is presented in mg/g. Means of three injections and SEM (in brackets) are presented. Concentration ratio between the ethanol and water extract is also presented.  seen by their early retention time and logP values, and hence more readily extracted in the more polar solvent (13).
Quercetin-3-O-galactoside and morroniside, active substances of S. purpurea (11,18,19), were present in similar amounts between the two extracts. Betulinic acid and ursolic acid were present in much larger quantities in the ethanol extract but to date have not been identified as actives. This, however, may be due to their high logP values. According to Lipinski's Rule of Five describing drug like substances, substances with logP values higher than five are less likely to be absorbed (20). Quercetin-3-O-galactoside, as well as the other phenolics, are more drug like as their logP values are lower than five, while those for betulinic acid or ursolic acid are greater than seven. However, discrimination of bioactive substances based on their logP values alone is not ideal, as it does not apply to substances that are actively transported (21) or take into account a matrix effect, which in this instance would be the plant extract. Preliminary in vitro and in vivo work has shown that ursolic acid and betulinic acid exhibit some antidiabetic activity (22)(23)(24)(25)(26)(27). Thus, it would be interesting to further explore the potential antidiabetic mechanism of betulinic acid and ursolic acid in more detail and determine why they are not the active principles in our assays.
An ethanol extract is generally preferred in biological assays as water can extract other cellular components that may mask or interfere with potential active substances (7,8). This could translate to a decreased, altered, or no activity in an in vitro assay. An exception to this preference would be in immunological studies, where proinflammatory polysaccharides may be necessary (28). This has been particularly important in the case of Echinacea (29) or ginseng (30), where the water extract has been shown to be quite proinflammatory in comparison to the ethanol one. This highlights the complexities of examining botanical samples where the activity may be related to the antagonistic or synergistic effect of many substances.
Although previously identified as minor components of S. purpurea (31), luteolin-7-Oglucoside and myricetin were not present in these extracts at the detection limit. Previous phytochemical analysis of the leaves of this species showed that these two substances were present in trace quantities. Seasonal and geographic variability in phytochemical content may account for these substances not being present at this level in the plant material used in this study.
S. purpurea extracts and isolated major constituent substances showed activity in a wide variety of antidiabetic assays (Table 4). In conclusion, most of the phenolic substances with antidiabetic properties in S. purpurea and other Cree botanical plants (10) showed a somewhat similar profile in both water and alcohol extracts, although amounts tend to be higher in alcohol. Thus, these extracts are expected to have generally comparable pharmacology with some predictable differences. In other situations, where lipophilic substances such as betulinic acid or ursolic acid are the active principles, the bioactivity is likely to be dramatically different between alcohol and water extracts. These findings demonstrate that both aqueous and ethanolic extracts may need to be evaluated when examining traditionally prepared medicines which are based on some form of water extraction.