Simultaneous Determination of Protopseudohypericin, Pseudohypericin, Protohypericin, and Hypericin Without Light Exposure

Abstract

St. John's wort products are commonly standardized to total naphthodianthrones and hyperforin. Determination of these marker compounds is complicated because of the photochemistry of the naphthodianthrones pseudohypericin and hypericin and the instability of hyperforin in solution. Protopseudohypericin and protohypericin have been identified as naturally occurring naphthodianthrones and, when exposed to light, they are converted into pseudohypericin and hypericin, respectively. However, exposure to light and the resulting naphthodianthrone free-radical reactions oxidize hyperforin. A mathematical relationship between the response of the proto compound and the resulting naphthodianthrone can be established by comparing the analytical response of the proto compound in a solution protected from light with the increase in the analytical response of naphthodianthrone in the same solution after exposure to light. By mathematically converting the proto compounds to their respective products, exposure to light can be avoided while still including proto compounds in a single assay. The method presented here details the reporting of all significant naphthodianthrones, including protopseudohypericin and protohypericin, without exposure to light. This approach includes the benefits of improved naphthodianthrone precision and protection of hyperforin from oxidation.

The principal marker compounds used in industry to sandardize St. John's wort products are “total naphthodianthrones” and hyperforin. Pseudohypericin is the predominant naphthodianthrone in most products, and a standard is readily available. Hypericin is the most studied of the napthodianthrones with a standard readily available. Pseudohypericin and hypericin together make up the total naphthodianthrones to which most St. John's wort products are standardized.

Scientists have shown the photochemical conversion of additional naturally occurring naphthodianthrones to pseudohypericin and hypericin. These “proto” naphthodianthrones are termed protopseudohypericin and protohypericin, respectively (1, 2). Pure reference materials for the proto compounds are not available.

Hyperforin is also of commercial importance; many products list hyperforin content on their labels. Hyperforin standards are readily available but are sensitive to oxidation. St. John's wort contains many other compounds, including the flavonoids quercetin, rutin, and quercitrin among others. The principal label claims for St. John's wort products are based on total naphthodianthrones and hyperforin.

Several analytical methods have been published for the analysis of St. John's wort botanical products. Some require exposure to light (3–7), whereas others avoid it (8–11). By forcing the conversion of proto compounds with light, the analytical methods that include exposure to light have higher naphthodianthrone results and less variability (3, 6). Exposure to light causes not only photochemical conversion but also a variety of free-radical and redox reactions, generating reactive species that oxidize hyperforin (6, 12).

When included in methods, exposure to light is applied only to the test solutions, not to the standard solutions (3–6). Shining light on solutions of pure analytical standards causes free radical reactions that initially appear to destroy the naphthodianthrones. In sample extracts of botanical preparations these reactions also occur, but because of additional redox-active compounds such as hyperforin and flavonoids, the reactions are moderated so that the naphthodianthrones are not “destroyed.” This is the reason that solutions of pure hypericin are considered unstable, yet higher results are obtained for methods that expose test solutions to light.

The Institute for Nutraceutical Advancement (INA) has 3 analytical methods available for St. John's wort, 1 with exposure to light (3) and 2 without (8, 9). All are recommended for the most accurate quantitation of naphthodianthrones (3) and other compounds (8), including hyperforin (9). Samples can be analyzed for hyperforin and other compounds, before exposure to light, and for naphthodianthrones, after exposure to light, but a single method would be preferred for all compounds. Mathematical proto compound conversion avoids exposure to light, and the resulting oxidation of hyperforin, therefore, allows determination of all compounds simultaneously in 1 analysis.

Photochemical conversion and the resulting side reactions are not an issue when only naphthodianthrones are determined and reported. The test solutions can be exposed to light for the best naphthodianthrone results without concern for hyperforin. When additional compounds such as hyperforin are included, exposure to light must be avoided.

The naphthodianthrones and proto compounds can be quantified without exposure to light after a mathematical relationship between the responses of the proto compounds and the resulting naphthodianthrones is determined. This mathematical factor is influenced by several instrument-specific parameters and must be determined on each instrument within the laboratory. The mathematical factor should be determined for each instrument, wavelength setting, background wavelength, and slit width. The mathematical conversion factors must be recalculated if any parameters are changed.

To calculate the mathematical response factor, all test solutions are analyzed before and after exposure to light. By tracking the decrease in the response of the proto compounds after exposure to light and the increase in the resulting naphthodianthrone responses, a relative response factor (RRF) can be calculated. Because of the complexity of free-radical reactions, it is best to determine the RRF in as simple a matrix as possible, without the added interactions of product formulations. These reactions can still occur, but avoiding exposure to light and using a mathematical conversion includes the proto compounds without incurring the effects of complicating side reactions during routine analyses.

It would be difficult to determine factors for all sample types, particularly in samples containing transition metals. Transition metals and divalent metal cations can form complexes with naphthodianthrones, affecting solubility, retention time, and absorption maxima while potentially aggravating the effects of oxidation on other sample components. In all cases, it is best to avoid complicating the determination of the response factor and to focus on minimizing exposure to light and the resulting free-radical and redox reactions during analysis.

Experimental

Samples

The test samples listed in Table 1 were prepared and analyzed with and without exposure to light. All are standardized botanical extracts from different sources. All preparations were used for the determination of the mathematical response factors. The calibration extract was prepared, as were the other test samples, at multiple levels. The multiple levels were then used for calibration and for reporting of the laboratory control sample (LCS), sample extract, and extract formulation. A botanical extract calibration material is used in the laboratory because of its greater stability in solution compared with those of pure reference materials. The extract formulation included lecithin and medium-chain triglycerides for inclusion in capsules.

Table 1. Test samples analyzed.

Sample identification	Sample description
Calibration extract	Standardized botanical powder extract
LCS extracta	Standardized botanical powder extract
Sample extract	Standardized botanical powder extract
Extract formulation	Standardized botanical extract suspended in oil matrix

^aLCS = Laboratory control samples.

Sample Preparation

A sample weight equivalent to ca 300 mg naphthodianthrones was placed into a 50 mL centrifuge tube. For a typical 0.3% naphthodianthrone botanical extract, 0.1 g is sufficient. A 25 mL portion of methanol was added to each sample, and the samples were sonicated while covered, at room temperature, for 20 min. The samples were centrifuged, and a portion from each was withdrawn for analysis. Samples were protected from light during preparation, and low-actinic amber autosampler vials and a paper tent covering the autosampler tray were also used.

All test samples were prepared in duplicate and analyzed immediately. Portions of test solution were also placed in amber autosampler vials and refrigerated for 1 week without exposure to light. The remaining test solutions were left sealed on the laboratory bench for 1 week, at room temperature, and were exposed to fluorescent and natural light for the worst-case “exposed” scenario. Refrigerated samples and exposed samples were then analyzed together 1 week after the initial analysis. The LCS extract was later analyzed according to the INA 30 min light-exposure protocol (3) and included for comparison.

Sample Analysis

Naphthodianthrones were determined by using a binary gradient of acetonitrile versus pH 6.9 phosphate salt buffer on a YMC Phenyl, S-3 μm, 120 Å, 3 × 150 mm column. The gradient included an initial acetonitrile content of 30%, which was increased to 70% over 20 min. The column was returned to 30% acetonitrile and equilibrated before the next injection. A UV-Vis detector monitoring at 590 nm was used for detection of naphthodianthrones with a background of 800 nm, or turned off. A background of 800 nm avoids the fluorescence emitted at 650 nm when light is absorbed by the naphthodianthrones. Improper background wavelengths will cause negative or distorted peak shapes because of strong fluorescence.

Results and Discussion

Establishing the Mathematical Conversion Factors

For determination of the mathematical response factor, all test solutions were analyzed before and after exposure to light. Figures 1 and 2 illustrate the effects of exposure to light, clearly showing the conversion of protopseudohypericin to pseudohypericin and protohypericin to hypericin.

Figure 1. Chromatogram of botanical extract preparation before exposure to light.

Figure 2. Chromatogram of botanical extract preparation after exposure to light.

Data used to determine the response of the protopseudohypericin compound relative to that of pseudohypericin are shown in Table 2. The protopseudohypericin area that “disappeared” caused an increase in the pseudohypericin response. The RRF is calculated by dividing the increase in the area of the pseudohypericin peak by the decrease in the area as follows:

Table 2. Calculation of pseudohypericin RRF, using area counts^a.

Sample description	Refrig. proto	Refrig. pseudo	Exposed proto	Exposed pseudo	RRF
Calibration extract	30.417	266.49	1.1399	344.35	2.659
	31.392	275.9	1.3281	361.5	2.847
LCS extractb	62.095	531.41	1.2993	691.73	2.637
	59.263	507.35	1.1612	664.82	2.710
Sample extract	70 59	542.34	1.2638	717.4	2.547
	137	461.03	1.103	611.52	2.593
Extract formulation	82.175	636.14	1.6019	849.17	2.644
	100.96	801.1	2.0856	1050.1	2.518
Avg.					2.645
RSD, %c					3.883

^aRefrig. = Refrigerated; proto = protopseudohypericin; pseudo = pseudohypericin.

^bLCS = Laboratory control samples.

^cRSD = Relative standard deviation.

$$\text{RRF}=(\text{exposed P}-\text{refrig}.\mathrm{P})/(\text{refrig}.\text{PP}-\text{exposed PP})$$

where P = pseudohypericin and PP = protopseudohypericin. The average increase in the pseudohypericin response was 2.645 times larger than the corresponding decrease in the protopseudohypericin response. The mathematical conversion of protopseudohypericin to pseudohypericin generated results that agreed across all test samples, and all data were included. The same approach was applied to the protohypericin photoconversion, and the results are summarized in Table 3.

Table 3. Calculation of hypericin RRF using area counts^a.

Sample description	Refrig. proto	Refrig. hypericin	Exposed proto	Exposed hypericin	RRF
Calibration extract	6.359	190.54	0	207.79	2.712
	6.562	196.76	0	217.4	3.145
LCS extractb	7.753	243.82	0	266.83	2.968
	7.418	231.93	0	255.47	3.173
Sample extract	11.224	255.63	0	284.71	2.591
	9.314	214.85	0	240	2.700
Extract formulation	13.073	301.27	0	340.76	3.021
	16.177	377.78	0	418.23	2.500
Avg.					2.851
RSD, %c					9.043

^aRefrig. = Refrigerated; proto = protohypericin.

^bLCS = Laboratory control samples.

^cRSD = Relative standard deviation.

Although the precision of the protopseudohypericin conversion factor is acceptable, the precision of the protohypericin conversion factor is not, primarily because of the low levels of protohypericin in the botanical extract samples, with many producing a response of <10 area counts. Most samples have very little protohypericin, except for plant materials. Because products are typically marketed with a “Total Naphthodianthrones” label claim, and finished products generally have no protohypericin remaining, the addition of the protohypericin in most samples amounts to <5% of the total naphthodianthrones. This small contribution to the total makes the variability in the protohypericin mathematical response factor insignificant.

Sample Preparation Comparability

Historically it has been difficult to reproduce analytical results for the naphthodianthrones over time, especially under different storage conditions. The uncorrected sample results are summarized in Table 4. Note the large differences (ca 20%) between the results for the test solutions initially protected from light and those exposed to light. These differences can be seen within a laboratory, and between laboratories, depending on the amount of exposure to light a test solution receives on any given day.

Table 4. Summary of uncorrected sample results.

Sample description	Initial, %	Refrigerated, %	Exposed, %	RSD, %a
LCS extractb	0.313	0.307	0.379	12.0
	0.308	0.307	0.380	12.6
Sample extract	0.299	0.291	0.364	12.6
	0.294	0.289	0.362	12.9
Extract formulation	0.157	0.151	0.191	13.0
	0.161	0.156	0.195	12.4
LCS extract–INAc	NAd	NA	0.385	NA
	NA	NA	0.386	NA

^aRSD = Relative standard deviation.

^bLCS = Laboratory control samples.

^cINA = Institute for Nutraceutical Advancements.

^dNA = Not available.

The sample results listed in Table 5 are the mathematically corrected results from the same analytical runs as those represented in Table 4. Protopseudohypericin and protohypericin responses are multiplied by the experimentally derived RRF, and these values are added to the respective naphthodianthrone responses for quantitation using the calibration curve. For example, a sample analyzed has the following responses, in area counts:

Table 5. Summary of mathematically corrected sample results.

Sample description	Initial, %	Refrigerated, %	Exposed, %	RSD, %a
Current LCSb	0.383	0.380	0.380	0.45
	0.380	0.379	0.382	0.40
Sample extract	0.369	0.369	0.365	0.63
	0.366	0.366	0.364	0.32
Extract formulation	0.197	0.191	0.191	1.79
	0.201	0.197	0.195	1.55

^aRSD = Relative standard deviation.

^bLCS = Laboratory control samples.

$$\text{PP}=70,\mathrm{P}=542,\text{PH}=11.2,\text{and}\mathrm{H}=255$$

where PP = protopseudohypericin, P = pseudohypericin, PH = protohypericin, and H = hypericin. The following responses would be used for quantitation:

$$\begin{array}{c}\text{Pseudohypericin}=(542+(70\ast 2.645)\\ \text{and hypericin}=(255+(11.2\ast 2.851)\end{array}$$

The data show that mathematical correction improves the precision between analyses including those across different week-long storage conditions. It is important to note that all preparation and exposure conditions generate results with an RSD of <2% RSD when the mathematical conversion is applied. The data generated by using the INA 30 min light exposure protocol agree with the results from the no-light mathematically corrected analyses to within approximately 1%. The mathematical correction factor approach solves common problems associated with differences in sample storage conditions and is comparabale to industry standard light exposure optimized methods.

Precision

The proposed no-light, mathematically corrected approach was used for the replicate analyses of the sample extract and the extract formulation. Each sample was analyzed 6 times for the determination of precision as % RSD. The results are presented in Tables 6 and 7.

Table 6. Results from replicate analyses of the sample extract.

Sample identification	Total naphthodianthrones, % by wt
Extract replicate 1	0.369
Extract replicate 2	0.366
Extract replicate 3	0.363
Extract replicate 4	0.363
Extract replicate 5	0.365
Extract replicate 6	0.369
Avg.	0.366
RSD, %a	0.75

^aRSD = Relative standard deviation.

Table 7. Results from replicate analyses of the extract formulation.

Sample identification	Total naphthodianthrones, % by wt
Formulation replicate 1	0.197
Formulation replicate 2	0.201
Formulation replicate 3	0.200
Formulation replicate 4	0.202
Formulation replicate 5	0.197
Formulation replicate 6	0.199
Avg.	0.199
RSD, %a	1.06

^aRSD = Relative standard deviation.

Recovery/Solvent Load Study

A placebo was constructed from the extract formulation specifications containing all ingredients, except St. John's wort extract. The sample extract was then added at 3 levels: 50, 100, and 150% of expected levels. These samples were prepared in groups of 3, and the results are summarized in Tables 8–10.

Table 8. Results from the 50% nominal level spike.

Sample identification	Weight of botanical extract added, g	Total naphthodianthrones in extract	Recovery, %
Low level 1	0.1531	0.360	101.4
Low level 2	0.1466	0.363	102.2
Low level 3	0.1489	0.362	102.1
Avg.			101.9
RSD %a			0.42

^aRSD = Relative standard deviation.

Table 10. Results for the 150% nominal level spike.

Sample identification	Weight of botanical extract added, g	Total naphthodianthrones in extract, %	Recovery, %
High level 1	0.4198	0.367	103.4
High level 2	0.4240	0.357	100.5
High level 3	0.4141	0.363	102.3
Avg.			102.1
RSD, %a			1.42

^aRSD = Relative standard deviation.

Conclusions

Multiple products were evaluated for naphthodianthrones before and after exposure to light. These samples showed quantitative conversion of protopseudohypericin and protohypericin to pseudohypericin and hypericin, respectively. The responses of the resulting pseudohypericin and hypericin were 2.65 and 2.85 times greater than those of protopseudohypericin and protohypericin, respectively. Mathematical response factor correction of proto compounds has been shown to reduce variability and provide results in agreement with standardized methods using light conversion, without subjecting the samples to light. This approach reduces laboratory costs associated with multiple analyses, does not depend on exposure to light, and includes all major naphthodianthrone species while protecting hyperforin and flavonoids from oxidation. This approach can also be used under various improved chromatographic conditions, allowing simultaneous determination of naphthodianthrones, hyperforin, and other marker compounds.

Table 9. Results for the 100% nominal level spike.

Sample identification	Weight of botanical extract added, g	Total naphthodianthrones in extract, %	Recovery, %
Nominal 1	0.2929	0.366	103.3
Nominal 2	0.2950	0.362	101.9
Nominal 3	0.2931	0.364	102.6
Avg.			102.6
RSD, %a			0.68

^aRSD = Relative standard deviation.