Mass Spectroscopic Fingerprinting Method for Differentiation Between Scutellaria lateriflora and the Germander (Teucrium canadense and T. chamaedrys) Species

Abstract

Scutellaria lateriflora, commonly known as skullcap, is used as an ingredient in numerous herbal products. However, it has been occasionally adulterated/contaminated with Teucrium canadense and T. chamaedrys, commonly known as germander, which contain hepatotoxic diterpenes. Due to the morphological similarities between the two genera, analytical methodologies to distinguish authentic S. lateriflora from the Teucrium species are needed to ensure public safety. In this study, a direct-injection electrospray ionization/MS method was used to generate spectral fingerprints of extracts from 21 skullcap and germander samples at a rate of 90 s/sample. MS fingerprints were analyzed by principal component analysis. The newly developed method offers a rapid and easy way to differentiate between skullcap and germander samples.

Product authentication, QA, and identification of adulterants are major issues facing the dietary supplement industry. This is particularly true when botanicals are very similar morphologically or when the product material is a powdered extract that lacks any morphological or genetic information. In such cases, chemical identification is an essential tool.

Scutellaria lateriflora (SL), commonly known as skullcap, is used by Western medical herbalists and frequently included as an ingredient in numerous herbal supplements available over the counter (1, 2). SL has historically been used to treat anxiety, sleeplessness, and various types of spasms (1, 2) and continues to be used for these purposes today (3, 4). Of the 55 herbs traditionally used to treat urinary system disorders (5), it is one of the top five with respect to radical-scavenging activity. Unfortunately, the only reported clinical trial as to its efficacy in relieving anxiety was based on a nonvalidated, subjective assessment scale (6). Previous studies have suggested that its modulation of γ-aminobutyric acid (GABA) and serotonin receptors may be partially responsible for SL's putative effects (7–9).

Historically, SL has been subject to adulteration with various species of the botanical germander, such as Teucrium canadense (TCA) and T. chamaedrys (TCH), which contain hepatotoxic diterpenes, specifically the neoclerodane diterpene teucrin A (10–12). The hepatotoxicity of teucrin A is due to its bioactivation by cytochrome P450 to create one or more hepatotoxic reactive metabolites (13, 14). The evidence suggests that oxidation of the furan ring moiety is necessary for hepatotoxicity. The diterpenes in Scutellaria contain a tetrahydrofurofuran moiety rather than the furan rings found in Teucrium (15–17).

Studies on the chemical composition of Scutellaria indicate that flavonoids, including flavones, flavanones, and chalcones, are the main phenolic components. The flavonoids are generally considered to be the active components of these plants (18–20). Unlike plants of the Scutellaria genus, plants of the Teucrium genus contain phenylethanoid glycosides as the main phenolic components, with some flavonoids also reported (19, 21–24).

Scutellaria and Teucrium genera are so similar morphologically that even trained botanists cannot easily distinguish between them. Analytical methods like TLC, HPLC-UV, and HPLC/MS have been used to distinguish between the two species. These traditional methods have their shortcomings. TLC lacks resolution and sensitivity, and requires carefully controlled conditions (20). Traditional HPLC-UV or HPLC/MS methods can be lengthy (12, 19, 20, 25); interpretation is time-consuming and requires trained professionals. For these reasons, a rapid and easy method capable of distinguishing between the two species was given a high priority by the Ingredient Ranking Subgroup of the AOAC Presidential Task Force on Dietary Supplements.

Spectral and chromatographic fingerprints that are analyzed using pattern recognition methods can be used for chemical discrimination between botanical materials and can place the analysis on a sound statistical base. The pattern of the spectra (with no prior separation) or the chromatograms have been used to distinguish genera and species (26–30). Use of the full spectra or chromatogram for principal component analysis (PCA) or analysis of variance (ANOVA) prevents undue reliance on one or two marker compounds, increases the robustness of the analysis, and provides for both visual and statistical evaluation (26, 28). In both cases, the chemical fingerprint is dependent on the extraction solvent. Spectral fingerprints are generally easier to analyze, since chromatographic fingerprints must be aligned with respect to retention times to provide uniform images.

In this paper, we present a method that discriminates between Scutellaria and Teucrium using mass spectral fingerprints acquired in less than 2 min using direct injection of a simple aqueous methanol extract. Data were processed using PCA or SIMCA (soft independent modeling of class analogy or statistical isolinear multicategory analysis); the latter is a multiple PCA fitting approach. This method demonstrates the efficacy of using mass spectral fingerprints for discriminating between a limited number of botanical genera and species.

Experimental

Reagents and Samples

1. Water.—Optima grade (Fisher Scientific, Pittsburgh, PA).

2. Acetonitrile.—Optima grade (Fisher Scientific).

3. Methanol.—Optima grade (Fisher Scientific).

4. Formic acid.—MS grade (Sigma-Aldrich, St. Louis, MO).

5. Verbascoside.—>95% pure (ChromaDex Inc., Irvine, CA).

6. HPLC mobile phase.—Mobile phase A was 0.1% formic acid in water, and mobile phase B was 0.1% formic acid in acetonitrile.

7. Samples.—Nine samples of the aerial parts of SL, five samples of TCA, and seven samples of TCH were obtained from the American Herbal Pharmacopoeia (AHP; Scotts Valley, CA). All samples except two were dried plant materials. All samples are botanically authenticated. Vouchers of authenticated materials and retention samples were deposited in the herbarium of the AHP.

Apparatus and Operating Conditions

1. HPLC system.—Agilent Technologies 1100 HPLC system (Palo Alto, CA) consisting of a quaternary pump with vacuum degasser, thermostated column compartment, autosampler, and diode array detector (DAD).

2. Mass spectrometer.—LCQ Classic ion-trap mass spectrometer (Thermo Fisher Scientific Inc., Waltham, MA).

3. Centrifuge.—IEC Clinical Centrifuge (Damon/IEC Division, Needham Heights, MA).

4. LC conditions.—The HPLC-UV/MS method used a mobile phase consisting of 0.1% formic acid in H₂O (A) and 0.1% formic acid in acetonitrile (B), with isocratic elution at 60 + 40 (v/v) for 1.5 min. The flow rate was 0.5 mL/min.

5. MS conditions.—Electrospray ionization (ESI) was performed in the negative-ion mode to obtain the MS fingerprints. The parameters of the mass spectrometer were optimized for verbascoside by autotune using the Xcalibur software through infusion of the verbascoside standard. The following conditions were used: sheath gas flow rate, 80 (arbitrary units); auxillary gas flow rate, 10 (arbitrary units); spray voltage, 4.50 kV; heated capillary temperature, 220°C; capillary voltage, −4.0 V; and tube lens offset, 25 V.

Sample Preparation

A 10 mg amount of dried ground sample was mixed with 5.0 mL methanol–water (60 + 40, v/v) in 15 mL centrifuge tubes and sonicated for 60 min at room temperature. The extracted samples were centrifuged at 5000 g for 15 min. The supernatant was filtered through a 17 mm (0.45 μm) PVDF syringe filter (VWR Scientific, Seattle, WA). The tinctures were diluted 1 to 100 (v/v) with the aqueous methanol extraction solvent described above and filtered prior to injection. The 11 SL samples were labeled SL01 to SL11, the five TCA samples were labeled TCA01 to TCA 05, and the seven TCH samples were labeled TCH01 to TCH 07.

To avoid errors arising from unexpected degradation of some phenolic compounds, the sample analysis was completed within 24 h after the extraction. The injection volume for all samples was 5 μL. Each sample was analyzed once per day on 5 consecutive days.

Data Acquisition

Spectral fingerprints were obtained for negative ions using direct injection ESI-MS. Spectra were summed over the 1.5 min interval required for the sample bolus to enter the MS. Five repeat analyses of the 21 different samples provided 105 spectra.

Data Processing

The mass spectra for each sample consisted of a one-dimensional matrix (counts versus mass for m/z 250–800). The spectra were exported to Excel (Microsoft, Inc., Belleview, WA) for preprocessing and then to SOLO (Eigenvector Research, Inc., Wenatchee, WA) for PCA and SIMCA. The preprocessing in Microsoft Excel involved combining the 105 spectra, sorting the data by sample names, and aligning the masses (each spectrum was a different length, since not all masses appeared in each spectrum). The resulting two-dimensional matrix (105 samples versus 551 masses) was then exported to SOLO for PCA and SIMCA. Preprocessing in SOLO, prior to PCA, consisted of normalization (the sum of squares of the counts for a spectrum was set equal to 1) and mean centering. Data were analyzed with and without autoscaling. Use of autoscaling produced tighter clusters in the PCA score plots, since the difference in absolute counts was minimized. Without autoscaling, clusters became more linear as the counts of specific ions had an increased effect.

Results and Discussion

Mass Spectra of Scutellaria and Teucrium

The method developed is intended to distinguish SL from TCA or TCH in minutes without chromatography. No analytical column is needed for the method, and a C18 RP guard column was only used as an online filter to protect the mass spectrometer. The guard column provided very limited separation between the inorganic and organic components of the extraction. The MS fingerprint for each sample was obtained as the sum of all the spectra of a sample for 1.5 min.

Typical MS fingerprints obtained for SL, TCA, and TCH are shown in Figure 1. The spectra for the three materials are very different from each other. This point is key to the fingerprinting method, since the morphological characteristics of these species are very similar. The most significant, or characteristic, m/z for SL is baicalin (5, 6, 7-trihydroxy-flavone-7-O-glucuronide) at 444.9 (M-H). It is found in SL, but not in TCA or TCH. The characteristic m/z for TCA is verbascoside at 623.2 (M-H), found in TCA, but not in SL or TCH. The characteristic m/z for TCH is teucrioside at 755.3 (M-H), found in TCH, but not in SL and TCA. These compounds were identified in an earlier study using HPLC/MS (25). By optimizing the MS parameters using verbascoside, the three peaks mentioned above dominated their respective spectra. There was, however, still considerable contribution from the other masses.

Figure 1. Typical MS fingerprints for SL, TCA, and TCH samples.

PCA of Mass Spectra of SL, TCA, and TCH

PCA involves a mathematical procedure that transforms a number of possibly correlated variables into a smaller number of uncorrelated variables called principal components. The first principal component accounts for as much of the variability in the data as possible, and each succeeding component accounts for as much of the remaining variability as possible. The advantage of using PCA to process the MS fingerprints is automation and simplification. It is easy to visually compare the mass spectra of a few samples, but the comparison of hundreds of samples quickly overwhelms our ability to see patterns. PCA provides visual patterns that everyone can easily understand and avoids subjective decisions.

Inspection of the PCA score in the plot shown in Figure 2 reveals two interesting observations. First, two samples (two groups of five, since the samples were analyzed five times each) are out of place. The PCA score of sample SL05 placed it within the TCA cluster, and the PCA score of TCA03 placed it within the TCH cluster. By studying their spectral fingerprints (Figure 3), it is obvious from the characteristic masses that these two samples were misidentified by the traditional morphological method. SL05 is a member of TCA, and TCA03 a member of TCH. They are easily identified by the MS fingerprints or PCA score plot. The PCA score plot with correct group assignment is shown in Figure 4. It is, in general, easy to distinguish the three groups using the PCA score plot. The exception is the merge point for TCA and TCH. The goal of this study was to distinguish between SL and the germanders. This is easily done using the PCA score plot in Figure 4, where the SL cluster is well separated from the TCA/TCH clusters.

Figure 2. PCA score plot for SL, TCA, and TCH samples.

Figure 3. MS fingerprints for SL05 and TCA03.

Figure 4. PCA score plot for SL, TCA, and TCH samples with corrected sample classification.

The second observation is that the three clusters (SL, TCA, and TCH) are elongated and seem to point to a central position. The data resemble three spokes on a wheel. An inspection of the MS fingerprints of the samples (data not shown) suggests that position on the spokes is dependent on the characteristic m/z intensity. As the characteristic m/z intensity increases, the sample moves further out on the spoke (toward the rim) and away from the central position (hub of the wheel; Figure 5). This conclusion is supported by the preprocessing method. The data in Figure 3 were preprocessed without autoscaling, which emphasizes the count intensity of the characteristic masses. Use of autoscaling reduces the length of the spokes, making each cluster more compact. Fingerprints for TCA01 and TCH07, the samples with the lowest characteristic signal intensities of TCA and TCH, respectively, are shown on Figure 6. Compared with Figure 1, it is easy to see that the signal intensities of the characteristic ions (m/z 623.2 and 755.3) are much lower. In both cases, the signals for the characteristic masses are not easily distinguished from the other mass signals. These data are consistent with our previous study that revealed that the concentration varied by x and y times, respectively. The concentrations reported earlier correlate well with the location of the samples on the spokes of the wheel; the higher the concentration, the farther away from the hub.

Figure 5. PCA score–characteristic m/z relationship.

Figure 6. MS fingerprints of the two samples closest to the center of the PCA score plot from the TCA and TCH groups.

These data show that, as the characteristic mass decreases, the germander species tend to look the same. However, the germander and SL spokes do not meet. At low concentrations of the characteristic masses, the germander and SL are still distinguishable from each other. The rest of the masses in the fingerprint allow discrimination between the genera. This emphasizes the necessity of comparing the total mass pattern rather than a single marker mass. It also points out that genus differences are greater than species differences, a fairly intuitive fact.

SIMCA Analysis of Mass Spectra of SL, TCA, and TCH

The SIMCA method effectively fits a principal components model around each class of samples in a data set. Figure 7 shows the samples/scores plot of the SIMCA test for all the samples, with the correct classification. The SL group (▼) was used to build the SIMCA model. The model then was used to test samples from other groups. None of the samples from the other two groups (TCA and TCH) fit into the SL model. Figure 8 shows the samples/scores plot of the SIMCA test for all the samples, with the original morphological classification. The SIMCA model differentiated the misidentified sample SL05 (a member of TCA) from other SL samples.

Figure 7. SIMCA model plot using SL as a model with corrected sample classification.

Figure 8. SIMCA model plot using SL as a model with the original sample classification.

Conclusions

This study demonstrated a simple and easy-to-use method that could distinguish SL from TCA and TCH in minutes. The fingerprinting method showed that the traditional morphological method to distinguish SL from Teucrium species is not infallible, even when samples are obtained with full documentation from a highly reliable source. The fingerprinting identification method described here nicely complements a more rigorous and lengthy chromatographic method previously developed in our laboratory that provided quantitative phenolic profiles of the SL and germander species (25). More importantly, the fingerprinting identification method was able to distinguish between two genera and two species of the same genus.