ANALYTICAL
195,77-85
BIOCHEMISTRY
(19%)
A Method for the Quantitation of Hypericin, an Antiviral Agent, in Biological Fluids by High-Performance Liquid Chromatography’ Leonard Liebes,’ Sandra Mendoza,
Yehuda Brandi
Mazur,* Dalia Freeman,* Levin, Howard Hochster,
David Lavie,* Gad Lavie, and Daniel Meruelo
Neil
Kudler,
New York University Medical Center, New York, New York 10016; and *The Weizmann Institute of Science, Rehovot 76100 Israel
Received
December
17,199O
Hypericin, a polycyclic aromatic dianthroquinone, is a natural plant product with antiviral properties. We report here the development of a methodology for the extraction and quantitation of hypericin from plasma and biological fluids and the adaptation of a sensitive and selective method for detection of the compound by high-performance liquid chromatography. The methodology offers a rapid and specific means of monitoring drug blood levels in clinical and pharmacokinetic studies. The chromatographic procedure utilizes the substantial retentive properties of hypericin on reversephase media and detection by the strong visible absorbance maximum at 590 nm. Verification by the fluorescence spectral properties of hypericin in organic media can also be utilized. The assay is linear over a 3 log concentration range and hypericin is consistently recovered from murine, simian, and human plasma. The methodology was applied to assess the pharmacokinetic properties of hypericin in mice receiving a single bolus injection of 350 pg. A distribution half-life of 2.0 h and an elimination half-life of 38.5 h were calculated. We also discuss the limitations of direct analysis of hypericin by absorbance or fluorescence measurements. 01991 Academic Press,Inc.
Hypericin (HY), 4,5,7,4’,5’,7’-hexahydroxy-2,2’-dimethyl-meso-naphthodianthrone (Fig. lA), is aconstituent of plants of the genus Hypericum where it occurs
together with its congener (analog) pseudohypericin (1). Hypericum plant extracts have long been a subject of medical interest; they have been described in various pharmacopoeiae and have been used as antidepressants (2-4). Hypericin can be obtained in pure form from plant extracts (1,5) as well as by synthesis (6,7). Both hypericin and pseudohypericin possess antiretroviral properties against murine retroviruses in uiuo and in vitro (5) and against the human immunodeficiency virus (HIV) in vitro (8). Among the features which characterize hypericin’s activity in slowing the progression of Friend-virus-induced splenomegaly in BALB/c mice are that its activity is in microgram amounts and that it is administered infrequently or as a single dose. Its potential therapeutic value for human retroviral-induced diseases remains unknown and awaits clinical evalua-
tion. Infections with human immunodeficiency virus type 1 and human immunodeficiency virus type 2, the retroviruses which appear to be involved in development of the acquired immune deficiency syndrome, continue to spread and expand as a global health hazard. The only
agents with proven clinical
leading
0003-2697/91$3.00 Copyright 0 1991 by Academic Press, All rights of reproduction in any form
to incomplete
cDNA
chain termination
(9).
Given the limited benefits and substantial toxicities associated with these drugs as well as the appearance of
resistant 1 Supported by funding from VIMRx Pharmaceuticals, Inc. ’ To whom reprint requests should be addressed at 550 First Avenue, New York, NY 10016. 3Abbreviations used: HY, hypericin; HIV, human immunodeficiency virus; DHHA, desoxohypericin hexoacetate; HSA, human serum albumin.
efficacy in slowing the pro-
gression of AIDS are nucleoside analogs, zidovudine (3’azido-2’,3’-dideoxythymidine), 2’,3’-dideoxycytidine, and 2’,3’-dideoxyinosine, which affect viral RNA reverse transcription during de nouo infection of cells
strains of HIV
(lo), new drugs which act at
other phases of the retroviral infection and replication cycle are desired. This has led to the examination of other compounds for antiretroviral activity, including hypericin and its congener. We have investigated hypericin’s biochemical andbiophysical properties as a means for quantitation and de77
Inc. reserved.
78
LIEBES
ET
AL.
tion coefficient of its absorption peak at X 621 nm (E = 45,000) was used for calibration of the internal standard stock solutions.
Reagents Acetonitrile (J T Baker, Phillipsburg, NJ) and ethyl acetate (Aldrich, St. Louis, MO) were of HPLC grade. All other organic reagents were of analytical grade.
Extraction OH
0
Solid-phase Cl8 extraction columns from JT Baker and Varian (Harbor, CA) were used to evaluate the efficiency of hypericin recovery from plasma samples. Centrifree membrane columns from Amicon (Acton, MA) were also utilized for hypericin recovery determinations.
OH
B
METHODS
CH3CO0
Solubilization of Hypericin Hexaacetate
CH3CO0
CH3
I
I bCOCH,
CH3COd FIG. 1. acetate.
Columns
Structure
of (A) hypericin
and (B)
desoxohypericin
hexa-
termination of its pharmacokinetic characteristics. We report here the methodology for the extraction and quantitation of hypericin from plasma and biological fluids and the adaptation of a sensitive and selective method for detection of the compound by high-performance liquid chromatography. MATERIALS
Standards In this study, hypericin was prepared synthetically (6). In the visible range of the absorption spectrum in ethanol, it exhibits two pronounced peaks at A 547 and 590 nm (Fig. 2). The molar extinction coefficient of the peak at 590 nm, t = 46,000, was used for calibration of the stock solutions employed in the HPLC quantitation analyses. Desoxohypericin hexaacetate (DHHA, Fig. 1B) was chosen as the internal standard based on its structural similarities to hypericin and its chemical stability under the assay conditions. DHHA was prepared as described by Brockmann et al. (6). The molar extinc-
and Desmohypericin
Stock solutions of synthetic hypericin, l-5 mg/ml in absolute ethanol, were prepared by sonication for 4 X 30 s at 30 W power using a microprobe tip (Fisher Sonicator, Fisher Scientific). They were further diluted to working stock solutions of 1, 10, and 100 pg/ml and kept in the dark until used. A stock solution of DHHA was prepared in a manner similar to that of hypericin at a concentration of 1.5 mg/ml in ethyl acetate and further diluted to 3.3 pg/ml with acetone. When protected from evaporation, shielded from light, and subjected to resonication, the HY and DHHA calibration solutions consistently gave the same HPLC response factors for periods of up to 6 months.
Evaluation
of Matrix
Methodologies
From 0.5- to l.O-ml volumes of plasma containing hypericin at concentrations of 0.5 and 1.0 pglml were applied to methanol-washed, preconditioned Cl8 separation columns which contained 500 mg bonded silica. The plasma was washed with two l-ml aliquots of phosphate-buffered saline which were discarded using a Vat-Elute manifold (Analytichem). Hypericin was eluted with three additions of acetonitrile or ethyl acetate (1 ml each) and taken to dryness in a Savant Speed Vat (Farmingdale, NY). One-milliliter samples of hypericin-containing plasma were also applied to Amicon Centrifree columns and centrifuged for 30 min at 10,OOOg.Both the eluates from the organic washes and the recovered fluid passing through the Centrifree membrane were analyzed for hypericin content by HPLC as described below.
CHROMATOGRAPHIC
Extraction
QUANTITATION
from Plasma
Plasma samples containing hypericin were first heated to 58°C in a shaking water bath for 45 min to inactivate HIV. Dimethyl sulfoxide (0.2 ml) was added to 0.25 ml of thawed plasma in a 12 X 75-mm glass centrifuge tube, followed by 0.075 ml of a mixture consisting of acetonitrile/2-butoxy-ethanol, 90110 (V/V), with vortex shaking for 5 s after each addition. The diluted DHHA internal standard (0.3 ml) was added to each 0.25-ml plasma sample to provide an amount equivalent to 1.0 pg/ml in plasma. Two milliliters of ethyl acetate was then added in two separate aliquots of 1.0 ml each followed by vortex shaking for 10 s after each aliquot. The tubes were then incubated at 37°C for 15 min. The mixture was next centrifuged at 7000g for 5 min to establish a defined interface between the organic and aqueous phases. The upper organic layer was removed and the pellet reextracted using the same procedure. The organic layer was again removed and combined with the first extraction product. The organic supernatant was taken to dryness in a Savant Speed Vat. The extracts were stored dry at -20°C and redissolved in 0.30 ml of acetone just prior to analysis by HPLC. Samples of 5-150 ~1 of the concentrate were injected for analysis. Extraction
from
Urine
Determination of hypericin levels in urine samples was performed by extraction of hypericin from urine, and quantitation of the compound by HPLC was performed using the same methods as described for plasma. Human urine samples were similarly subjected to a 45min HIV heat-inactivation step. HPLC
Analysis
Samples were chromatographed by gradient elution on a reverse-phase, pheny14 pm, 0.45 X lo-cm radialpak analytical column (Waters Associates, Milford MA). The chromatographic system consisted of two Knauer Model 64 pumps connected to a dynamic mixer (Sonntek, Woodcliff Lake, NJ) along with a WISP Model 710B autosampler (Waters Associates). The pumping system and WISP were controlled by an Axxiom 747 multisystem controller-data system (Cole Scientific, Calabasas CA). Fluorescence was detected using an Applied Bio-systems Model 970 fluorescence detector (Ramsey, NJ), and visible absorbance at 590 nm was measured by a Linear Model 203 variable uv/vis detector (Reno NV). Mobile phase solution A consisted of a 70% solution of 0.1% ammonium phosphate (adjusted to pH 7.0 with NaOH) and 30% acetonitrile; solution B was 70% acetonitrile-30% water. A linear gradient of 100% A to 100% B was developed over a 15-min interval
OF
79
HYPERICIN
with a flow rate of 1.2 ml/min, followed by 4 min of 100% B. The reequilibration of the column was achieved by a linear change from 100% B to 100% A over the next 4 min followed by 7 min of isocratic A. This resulted in a complete analysis cycle time of 30 min. The fluorescence detection used excitation at 470 nm and emission using a cutoff filter limiting light transmission below 550 nm. Detection in the visible range was at 590 nm. Hypericin eluted with a retention time of 12.9 min while WIS-4 eluted at 22.2 min (Fig. 6). The assay has a lower limit of sensitivity of 250 pg injected onto the column, achievable with both the visible absorbance and fluorescence detection systems (sensitivity setting of 0.02 PA = 1 V full scale on the fluorescent detector and 0.1 A = 1 V full scale for the absorbance detector).
Calibration
Procedure
and Evaluation
of the Method
Standard solutions in ethanol were added to normal donor control plasma with hypericin at concentrations ranging from 0.050 to 40 pg/ml. The back-calculated hypericin concentrations were normalized against the unextracted hypericin standards. A linear regression analysis was used to determine the degree of fit of the data.
Pharmacokinetic
Anulysis
Pharmacokinetic values were obtained by first using the JANA curve stripping program to estimate the initial pharmacokinetic parameters prior to modeling using PCNONLIN (Statistical Consultants, Lexington, KY). The estimates were used as the initial parameters for a two-compartment bolus injection absorption model. The program used these parameters to fit the data to the best curve possible and then applied Lagrange techniques to reduce the total variance of the key pharmacokinetic values (11).
Recovery of Hypericin
from Plasma
Additions of hypericin diluted from a 2 mg/ml stock solution prepared in ethanol were added to murine, simian, and human plasma samples. The samples were mixed on a Vortex Genie, allowed to stand for 30 min at room temperature, and then frozen at -20°C for at least 24 h until subjected to extraction as described above. Efficiency of hypericin recoveries was calculated from the ratio between the amount of hypericin recovered from a standard amount added to plasma and to an equivalent volume of ethanol. The latter samples of hypericin in ethanol were not extracted but were taken to dryness in the Speed Vat and treated identically to the plasma samples.
80
LIEBES
ET
AL. TABLE
2.3
Hypericin
Visible
Ethanol
LIX 8 5 e 8 9
590 547 598 558
-0.1 II
200
I
I
I
I
300
400
500
600
Wavelength FIG. 2. Ultraviolet/visible ethanol. Concentration,
absorbance 22.4 pg/ml.
700
(nm) spectrum
of hypericin
in
Molar
1 Extinction t
Coefficients Water
t
46,000 22,000 6,000 10,060
_
opment of an extraction methodology which allowed dissociation of hypericin from protein complexes and removal of proteins.
Spectroscopic Studies Absorbance properties
of HY in different solvents. Hypericin exhibits a number of absorbance maxima in BALB/c mice (20 g weight) were injected with 350 pg the visible and ultraviolet regions as has been documented (6,7,12). In ethanol, the absorbance maxima are hypericin dissolved in 0.5% aqueous benzyl alcohol from and 234 nm (Fig. a 5 mg/ml stock solution. Animals were injected in pairs seen at 590,547,510,474,383,340,285, 2). In aqueous media, however, substantial differences and each animal was bled 0.5 ml from the retroorbital plexus of the eye into heparinized tubes. One pair of in the visible spectrum occur. The visible maxima at 547 mice was bled once for each of the sampling points (0, and 590 nm are most sensitive to the type of solvent environment and are “red-shifted” in aqueous media to 0.25,0.5,1,2,4,6,8,24,36,72,120,192,216, and 240 h). 558 and 598 nm, respectively, along with considerable The plasma was separated by centrifugation at 1OOOg reductions in the molar extinction coefficients (Table 1, and subjected to the extraction protocol described Fig. 2, and Fig. 3, spectrum A). above. Results for each pair were averaged together for Similarly, with the addition of HSA to a neutral each time point. aqueous suspension of HY of the same concentration as that in Fig. 3, spectrum A, the 598-nm peak is “blueSpectroscopic Analyses shifted” toward 590 nm as well as increased in intensity Hypericin was dissolved by sonication in water or while the 558-nm peak is red-shifted to 560 nm (Fig. 3 ethanol at a concentration of 2 mg/ml as described spectrum B). When compared to that of hypericin in above. A solution containing 20 PM human serum albu- water alone, a more dramatic increase occurs in the size min (HSA) in Hepes buffer, pH 7.0, was mixed with different amounts of HY ranging from 0.5 to 100 PM. Visible range spectral profiles were generated using a Beckman DU7 spectrophotometer (Fullerton, CA). Fluorescence data were obtained using a Hitachi PerkinElmer SPF spectrofluorometer (Norwalk, CT) configured with excitation and emission slits at 1 nm. The excitation wavelength was set at 470 nm for the generation of fluorescence emission scans.
Murine
Pharmacology
RESULTS
The spectral properties of hypericin in aqueous, organic, and physiologic media were analyzed as part of our development of a sensitive method for the quantitation of hypericin in biologic fluids. These properties were found to vary with the type of media and solvent environment in which hypericin was dispersed. In addition, serum proteins avidly associate with hypericin and affect its biophysical properties. This required the devel-
Wavelength FIG. 3. water and persed in with 5 mM The same
(nm)
Ultraviolet/visible absorption spectrum of hypericin (A) in (B) in water in the presence of HSA. Hypericin was diswater by sonication (concentration, 2 mg/ml) and diluted Hepes buffer (pH 7.0); final concentration, 48 pg/ml. (B) solution in the presence of 50 Fg HSA.
CHROMATOGRAPHIC HY: 1:l 1
HSA
Molar
2:l
3:1
I
I
QUANTITATION
Ratios 4:l
5:l I
I
I
0
VM
FIG. 4. Visible absorbance and fluorescence changes as a function of different hypericin/HSA ratios. Fluorescence using 470 nm excitation and 600 nm emission (A) (arbitrary units/lo), absorbance at 560 nm (0), and absorbance at 590 nm (Cl) are plotted as a function of HY concentrations which were added to a fixed concentration of HSA of 20 PM.
of the 590-nm absorbance peak (0.90 A) than is apparent with the 560-nm absorbance increase (0.27 A). The absorbances at both 590 and 560 nm increase in direct proportion to the amounts of hypericin added to a fixed concentration of HSA (20 PM) but at different rates for each of these wavelengths (Fig. 4). For HY:HSA ratios of up to l:l, the increase in absorbance was fit by slopes of 0.0154 and 0.0268 for measurements at 560 and 590 nm, respectively. For HY:HSA ratios between 2:l and 5:l the plots of the absorbance changes were similar (slope = 0.0105 for 560 nm and 0.0108 for 590 nm). Increases in fluorescent emission intensities of HY at 600 nm were linear with respect to the amount of HY up to a protein:hypericin molar ratio of 1:l (slope = 0.129). Further increases in the concentration of HY relative to HSA beyond a 1:l molar ratio are associated with nonlinear decreases in fluorescence. Fluorescence properties. The differences in spectral absorption properties of hypericin in aqueous media in comparison with polar organic solvents are also reflected in fluorescence spectra (Fig. 5). In polar organic solvents, such as ethanol, hypericin gives a brightly red fluorescent solution with emission maxima at 595 and 645 nm (Fig. 5, spectrum C). In aqueous suspensions (Fig. 5, spectrum A), no fluorescence is observed. Upon addition of HSA to the aqueous suspension of hypericin, fluorescence can be detected with emission spectra (Fig. 5, spectrum B) somewhat similar to those obtained for hypericin in ethanol. Chromatographic
Resolution
of Hypericin
and DHHA
Figure 6 shows chromatograms obtained from human plasma to which 1.0 pg/ml hypericin and 4.0 pg/ml
81
HYPERICIN
DHHA have been added as well as a plasma control containing only DHHA. The plasma samples were extracted with organic solvents and detection was by visible absorbance at 590 nm. The chromatographic peak of hypericin resolved well from that of DHHA which elutes 9.3 min after hypericin. The blank extracted plasma control was free of any other plasma components detectable at this range with the exception of the added DHHA (Fig. 6).
6:l
Recoveries
HY Total,
OF
of Hypericin
from Plasma of Different
Species
Hypericin was extracted from the plasma of mice, monkeys, and humans to determine whether any differences in the efficiencies of recovery exist between the various species used in preclinical studies of hypericin. The recovery from plasma samples which contained concentrations of hypericin ranging from 0.125 to 15 pg/ml was compared with direct standards of hypericin in ethanol. Figure 7 shows that these data can be fit by a straight line with a slope of 0.919. A regression analysis of these data yielded a correlation coefficient of 0.991. The regression analysis of individual data for each species were 0.999,0.998, and 0.997 respectively for human, simian, and murine plasma samples. The data as well as those summarized in Table 2 show consistent recoveries of hypericin throughout this wide concentration range that was equivalent for plasma from the different species which were studied. Assay Precision
and Detection
Limits
The detection limits for routine assays were determined at 10 ngfml when l.O-ml samples of plasma were used for analysis. The interassay precision was determined by replicate analyses of plasma to which 1000 (N
600 Wavelength
Sk0 (nm)
FIG. 5. Hypericin fluorescence emission spectra in aqueous and organic solutions. Hypericin dissolved and sonicated in pH 8.0 water at a concentration of 2.0 mg/ml was diluted to 3.3 pg/ml with water (A), water containing 333 pg/ml HSA (B), and ethanol (C). Fluorescence emission spectra were obtained using 470 nm excitation.
82
LIEBES
ET
AL.
I
DHHA
(22.2)
0
Elutlon FIG. 6. HPLC chromatographic (. - + - * ). An extract from human
profile plasma
of human containing
for DHHAIHY
Recovery from Plasma
The recovery of 1 @g/ml of DHHA from human plasma was found to be 96.6% + 5.8% (N = 8) of the direct control which was not subjected to extraction. A combination of DHHA (1 pg/ml) with hypericin throughout a concentration range of 0.05 to 40 pg/ml was used to evaluate the recovery of hypericin in the presence of this internal standard. Figure 8 shows the existence of a high degree of linearity with this combination throughout this concentration range. Recovery
of Hypericin
from Human
Analysis
of Hypericin
plasma
containing
1 pg/ml
hypericin
and 4 pg/ml
Wis 4
in 0.5 ml of 0.5% benzyl alcohol. The animals were bled at time intervals ranging from 10 min to 240 h. The plasma samples obtained were extracted and analyzed as outlined above. The concentration vs time data from this experiment (Fig. 9) show a peak concentration of 27.8 pg/ml at 0.17 h (10 min) and decreasing measurable values up to 240 h (10 rig/ml). The data were well approximated using a two-compartment model contained in the PCNONLIN pharmacokinetic library, allowing for a first-order input. The distribution phase (a halflife) was 2.0 h, while the elimination phase (0 half-life) was 38.5 h. The volume of distribution was 12.6 ml. The kinetic equilibrium constants (Table 5) indicate a strong tissue absorption phenomenon (4, = 0.1 vs k,, = 0.2) that contributes to an extended p half-life of 38.5 h.
Urine
Hypericin recoveries from urine were lower than those obtained with plasma with a range from 63 to 75% (Table 4). This lower recovery range was also reflected with DHHA. The ratio of DHHA to HY serves as an adequate correction for the effects of the various salts present in urine. Pharmacological
(Minutes)
plasma extracts. An extract from only 4 pg/ml DHHA (+++I.
= 7) and 2500 rig/ml (N = 9) of hypericin were added. Table 3 shows that the percentage coefficient of variations for both of these concentration ranges was ~2%. Range of Linearity
Time
in BALBIc
Mice
Pharmacokinetic data were analyzed in mice that received a bolus iv injection of 350 pg (17.5 mg/kg) of HY
DISCUSSION Hypericin has been found to possess antiretroviral properties, acting directly on the retrovirion particle leading to the loss of infectivity (5,B). Any potential clinical applications of hypericin require an assay for determination of levels of drug in blood and other biological fluids. The hydrophobic properties of hypericin and a complex pattern of molecular interactions in aqueous and polar organic solvents necessitated particular care in
CHROMATOGRAPHIC
QUANTITATION
OF
Hypericin Hypericin level (w/ml) 1000
2500
FIG. 7. The plot of hypericin determinations from extracted plasma of three species: murine (O), simian (A), and human (Cl). The data are the mean and standard deviation of two separate plasma extractions. The levels are correlated against hypericin determinations from direct standards and analyzed by a linear regression, yielding a slope of 0.919 and a correlation coefficient of 0.991 for the aggregate data from all species. The correlation coefficients for each of the species were 0.999, 0.998, and 0.997 for human, simian, and murine plasma samples, respectively.
designing this assay. In the course of this study it became apparent that an assay by direct fluorescence measurement, while highly sensitive in the presence of serum or plasma protein components, was not suitable for quantitation of HY. Thus, our data showed an increase in fluorescence intensity at HY:HSA molar ratios of up to l:l, but at higher levels of HY the fluorescence intensities were markedly attenuated. It appears that low ratios of HY to protein yield a linear increase of fluorescence while higher ratios lead to quenching of fluorescence. The use of absorbance measurements directly on samples of plasma and other biological fluids for HY quantitation at either 560 or 590 nm was also not feasible because of nonlinearity observed at different HY/HSA ratios. Similarly, data obtained with HY additions to
TABLE Hypericin
Recovery
83
HYPERICIN
Precision
TABLE
3
Data
from
Calculated average
Human
Plasma
Number of analyses
SD
%CV
7 9
19 39
1.82 1.61
1028 2407
whole plasma have shown nonquantitative absorbance spectral changes (unpublished data). Initial studies to determine recovery efficiencies using either C8 or Cl8 solid-phase extraction columns showed hypericin recovery from plasma to be very poor (O-14% HY recovered, data not shown). The use of membrane centrifugation columns also gave poor results, suggesting that the association of hypericin with proteins is of a strong nature. The quenching of fluorescence at high hypericin to protein ratios, along with the nonlinear, concentration-related spectroscopic changes which occur when HY interacts with serum proteins, renders quantitation of hypericin in biological fluids by direct measurements of fluorescence, or by absorbance changes, inaccurate. We thus developed a protocol which included: (i) The extraction of hypericin from plasma using a solvent mixture which allowed the concomitant precipitation of plasma proteins and the maintenance of HY in a soluble form. (ii) The chromatographic separation of the molecule in a polar organic solvent environment that provides optimal use of absorbance and of fluorescence properties of the molecule for quantitative analysis. The feasibility of the use of HPLC for the analysis of hypericin has previously been shown in the separation
2 in Plasma
HY level (w/ml)
500 (%)
2500 (%)
5000 6)
10,000 (X0)
Human SD Mouse SD Monkey SD
91.4 12.9 92.1 7.4 84.1 1.0
85.3 7.5 83.1 16.4 92.0 3.2
93.4 a.7 100.0 11.6 88.4 0.7
87.3 8.0 100.0 9.0 94.5 11.9
0
10
20
HY
Concentration
30
4 D
(pglml)
FIG. 8. Plot of hypericin/DHHA area ratios from human plasma extracts as a function of hypericin over a range of 0.05 to 40 pg/ml with the DHHA concentration at 1 fig/ml.
a4
LIEBES TABLE
Hypericin HY level (na/ml) Human SD
urine
Recoveries
AL.
4
TABLE
in Human
Urine
500 (%)
2500 (%)
5000 (%)
10,000 60)
63.4 7.6
65.9 21.2
75.2 0.9
65.9 31.0
. 4 I-
O, 200
100
Time,
5
Pharmacokinetic Estimates of iv Administration of 350 pg Hypericin to BALB/c Mice
by reverse-phase chromatography of hypericin and pseudohypericin from Hypericum plant extracts (13,14). These analyses used analytical methodology similar to that reported here but were devised mainly for quantitation of hypericin in various plant preparations by different solvent elution cycles. We have employed a phenyl reverse-phase column to provide an environment which is less hydrophobic than the Cl8 or C8 columns used in these previous studies. The phenyl solid phase allows a substantial retention of hypericin and a reduction in analysis time to a 30-min cycle as compared to the longer retention time reported previously (14). The organic extraction procedure provides the necessary milieu to efficiently extract hypericin from plasma and urine while precipitating the protein components. The recoveries, which range from 85-100% when extractions of hypericin are repeated twice, do not seem to vary among different animal species and are equivalent in reference to the DHHA internal standard. The extracted hypericin dissolves readily in ethanol or acetone and is eluted from the chromatographic col-
0
ET
hr
FIG. 9. Hypericin plasma levels as a function of time. BALB/c mice were injected with a bolus injection of 350 fig hypericin in 0.5% benzyl alcohol. Fit of plasma HY concentration from two-compartment bolus injection model using PCNONLIN (-) as compared with the actual data (*).
Parameter C,, bdml) Volume (ml) k 21 k 10 k k;;-HL (h) AUC (pg/ml/h) (u-HL (h) 8-HL (h)
Estimate 27.6 12.6 0.1 0.4 0.2 7.1 285.1 2.0 36.5
umn in an organic solvent which provides optimal retention times and enables monitoring at 590 nm and confirmation by fluorescence using excitation at 470 nm. This procedure should be readily adaptable to the analysis of hypericin content in other biological material such as feces and tissue with appropriate modifications to allow for adequate dispersion of the material prior to extraction. We have found that the use of simultaneous detection of visible fluorescence emission with a cutoff filter which limits light below 550 nm along with absorbance detection at 590 nm provides additional levels of discrimination for hypericin to aid in discerning the presence of metabolites. The high molar extinction coefficient for hypericin of 46,000 provides a level of detection sensitivity by absorbance that is at least equal to that of fluorescence detection for most compounds and possibly better given the good signal to noise ratios obtainable with modern variable wavelength HPLC detectors. In summary, the association of hypericin with proteins results in nonlinear fluorescence and absorbance changes. A reliable assay for hypericin requires organic extraction, as solid-phase extraction and membrane filtration are not adequate to remove proteins. The recovery using a dual extraction technique ranges from 85100%. The HPLC separation between hypericin and the internal standard DHHA using a phenyl reverse-phase column is greater than 9 min and utilizes a solvent mixture that optimizes detection at the maximum visible extinction coefficient of the molecule. The use of visible detection provides an ample signal to noise ratio and results in no discernable interfering peaks. The preliminary murine pharmacokinetics show detectible levels out to 240 h and we expect that this assay will provide accurate results for the study of human pharmacokinetits in phase I studies. REFERENCES 1. Brockman, H. V., Falkenhausen,E. Budde, G. (1950) Naturwissenckuften
H., Neeff, R., Dorlars,A., 37,520-545.
and
CHROMATOGRAPHIC 2. Buchner, 217. 3. Muldner, 4. Daniel,
A. (1830)
Buchner$
Reperturium
QUANTITATION
Phurmacie
H., and Zoller, M. (1984) Arzneim.-Borsch. K. (1949) Hippocrates 19, 526-530.
5. Meruelo, D., Lavie, USA 85,5230-5234. 6. Brockmann, 90,2302-2318.
G., and Lavie,
H., Kluge,
D. (1988)
F., and Muxfeldt,
PFOC.
IM 34,
Natl.
H. (1957)
34,
918.
Acad. Chem.
Sci. Ber.
7. Lavie, D., Freeman, D., Bock, H., Fleischer, J., Van Kraanenburg, K., Ittah, Y., Mazur, Y., Lavie, G., Liebes, L., and Meruelo, D. J. Pure Appl. Chem. in press. 8. Lavie, G., Valentine, F., Levin, B., Maxur, Y., Gallo, G., Lavie, D.,
OF
HYPERICIN
85
Weiner, D., and Meruelo, D. (1989) PFOC. Natl. Acad. Sci. USA 86,5963-5967. 9. Yarchhoan, R., Mitsuya, H., Myers, C. E., and Broder, S. (1989) N. Engl. J. Med. 321, 726-738. 10. Larder, B. A., Darby, G., and Richman, D. D. (1989) Science 243, 1731-1734. 11. Draper, N. R., and Smith, H. (1981) Applied Regression Analysis, 2nd ed. Wiley, New York. 12. Brockmann and Eggers, H. (1958) Chem. Ber. 91, 81. 13. Eckhardt, V. W. E. (1984) Dtsh. Apoth. Ztg. 124, 2383-2386. 14. Holzl, J., and Ostrowski, E. (1987) Dtsh. Apoth. Ztg. 127, 12271230.
195,77-85
BIOCHEMISTRY
(19%)
A Method for the Quantitation of Hypericin, an Antiviral Agent, in Biological Fluids by High-Performance Liquid Chromatography’ Leonard Liebes,’ Sandra Mendoza,
Yehuda Brandi
Mazur,* Dalia Freeman,* Levin, Howard Hochster,
David Lavie,* Gad Lavie, and Daniel Meruelo
Neil
Kudler,
New York University Medical Center, New York, New York 10016; and *The Weizmann Institute of Science, Rehovot 76100 Israel
Received
December
17,199O
Hypericin, a polycyclic aromatic dianthroquinone, is a natural plant product with antiviral properties. We report here the development of a methodology for the extraction and quantitation of hypericin from plasma and biological fluids and the adaptation of a sensitive and selective method for detection of the compound by high-performance liquid chromatography. The methodology offers a rapid and specific means of monitoring drug blood levels in clinical and pharmacokinetic studies. The chromatographic procedure utilizes the substantial retentive properties of hypericin on reversephase media and detection by the strong visible absorbance maximum at 590 nm. Verification by the fluorescence spectral properties of hypericin in organic media can also be utilized. The assay is linear over a 3 log concentration range and hypericin is consistently recovered from murine, simian, and human plasma. The methodology was applied to assess the pharmacokinetic properties of hypericin in mice receiving a single bolus injection of 350 pg. A distribution half-life of 2.0 h and an elimination half-life of 38.5 h were calculated. We also discuss the limitations of direct analysis of hypericin by absorbance or fluorescence measurements. 01991 Academic Press,Inc.
Hypericin (HY), 4,5,7,4’,5’,7’-hexahydroxy-2,2’-dimethyl-meso-naphthodianthrone (Fig. lA), is aconstituent of plants of the genus Hypericum where it occurs
together with its congener (analog) pseudohypericin (1). Hypericum plant extracts have long been a subject of medical interest; they have been described in various pharmacopoeiae and have been used as antidepressants (2-4). Hypericin can be obtained in pure form from plant extracts (1,5) as well as by synthesis (6,7). Both hypericin and pseudohypericin possess antiretroviral properties against murine retroviruses in uiuo and in vitro (5) and against the human immunodeficiency virus (HIV) in vitro (8). Among the features which characterize hypericin’s activity in slowing the progression of Friend-virus-induced splenomegaly in BALB/c mice are that its activity is in microgram amounts and that it is administered infrequently or as a single dose. Its potential therapeutic value for human retroviral-induced diseases remains unknown and awaits clinical evalua-
tion. Infections with human immunodeficiency virus type 1 and human immunodeficiency virus type 2, the retroviruses which appear to be involved in development of the acquired immune deficiency syndrome, continue to spread and expand as a global health hazard. The only
agents with proven clinical
leading
0003-2697/91$3.00 Copyright 0 1991 by Academic Press, All rights of reproduction in any form
to incomplete
cDNA
chain termination
(9).
Given the limited benefits and substantial toxicities associated with these drugs as well as the appearance of
resistant 1 Supported by funding from VIMRx Pharmaceuticals, Inc. ’ To whom reprint requests should be addressed at 550 First Avenue, New York, NY 10016. 3Abbreviations used: HY, hypericin; HIV, human immunodeficiency virus; DHHA, desoxohypericin hexoacetate; HSA, human serum albumin.
efficacy in slowing the pro-
gression of AIDS are nucleoside analogs, zidovudine (3’azido-2’,3’-dideoxythymidine), 2’,3’-dideoxycytidine, and 2’,3’-dideoxyinosine, which affect viral RNA reverse transcription during de nouo infection of cells
strains of HIV
(lo), new drugs which act at
other phases of the retroviral infection and replication cycle are desired. This has led to the examination of other compounds for antiretroviral activity, including hypericin and its congener. We have investigated hypericin’s biochemical andbiophysical properties as a means for quantitation and de77
Inc. reserved.
78
LIEBES
ET
AL.
tion coefficient of its absorption peak at X 621 nm (E = 45,000) was used for calibration of the internal standard stock solutions.
Reagents Acetonitrile (J T Baker, Phillipsburg, NJ) and ethyl acetate (Aldrich, St. Louis, MO) were of HPLC grade. All other organic reagents were of analytical grade.
Extraction OH
0
Solid-phase Cl8 extraction columns from JT Baker and Varian (Harbor, CA) were used to evaluate the efficiency of hypericin recovery from plasma samples. Centrifree membrane columns from Amicon (Acton, MA) were also utilized for hypericin recovery determinations.
OH
B
METHODS
CH3CO0
Solubilization of Hypericin Hexaacetate
CH3CO0
CH3
I
I bCOCH,
CH3COd FIG. 1. acetate.
Columns
Structure
of (A) hypericin
and (B)
desoxohypericin
hexa-
termination of its pharmacokinetic characteristics. We report here the methodology for the extraction and quantitation of hypericin from plasma and biological fluids and the adaptation of a sensitive and selective method for detection of the compound by high-performance liquid chromatography. MATERIALS
Standards In this study, hypericin was prepared synthetically (6). In the visible range of the absorption spectrum in ethanol, it exhibits two pronounced peaks at A 547 and 590 nm (Fig. 2). The molar extinction coefficient of the peak at 590 nm, t = 46,000, was used for calibration of the stock solutions employed in the HPLC quantitation analyses. Desoxohypericin hexaacetate (DHHA, Fig. 1B) was chosen as the internal standard based on its structural similarities to hypericin and its chemical stability under the assay conditions. DHHA was prepared as described by Brockmann et al. (6). The molar extinc-
and Desmohypericin
Stock solutions of synthetic hypericin, l-5 mg/ml in absolute ethanol, were prepared by sonication for 4 X 30 s at 30 W power using a microprobe tip (Fisher Sonicator, Fisher Scientific). They were further diluted to working stock solutions of 1, 10, and 100 pg/ml and kept in the dark until used. A stock solution of DHHA was prepared in a manner similar to that of hypericin at a concentration of 1.5 mg/ml in ethyl acetate and further diluted to 3.3 pg/ml with acetone. When protected from evaporation, shielded from light, and subjected to resonication, the HY and DHHA calibration solutions consistently gave the same HPLC response factors for periods of up to 6 months.
Evaluation
of Matrix
Methodologies
From 0.5- to l.O-ml volumes of plasma containing hypericin at concentrations of 0.5 and 1.0 pglml were applied to methanol-washed, preconditioned Cl8 separation columns which contained 500 mg bonded silica. The plasma was washed with two l-ml aliquots of phosphate-buffered saline which were discarded using a Vat-Elute manifold (Analytichem). Hypericin was eluted with three additions of acetonitrile or ethyl acetate (1 ml each) and taken to dryness in a Savant Speed Vat (Farmingdale, NY). One-milliliter samples of hypericin-containing plasma were also applied to Amicon Centrifree columns and centrifuged for 30 min at 10,OOOg.Both the eluates from the organic washes and the recovered fluid passing through the Centrifree membrane were analyzed for hypericin content by HPLC as described below.
CHROMATOGRAPHIC
Extraction
QUANTITATION
from Plasma
Plasma samples containing hypericin were first heated to 58°C in a shaking water bath for 45 min to inactivate HIV. Dimethyl sulfoxide (0.2 ml) was added to 0.25 ml of thawed plasma in a 12 X 75-mm glass centrifuge tube, followed by 0.075 ml of a mixture consisting of acetonitrile/2-butoxy-ethanol, 90110 (V/V), with vortex shaking for 5 s after each addition. The diluted DHHA internal standard (0.3 ml) was added to each 0.25-ml plasma sample to provide an amount equivalent to 1.0 pg/ml in plasma. Two milliliters of ethyl acetate was then added in two separate aliquots of 1.0 ml each followed by vortex shaking for 10 s after each aliquot. The tubes were then incubated at 37°C for 15 min. The mixture was next centrifuged at 7000g for 5 min to establish a defined interface between the organic and aqueous phases. The upper organic layer was removed and the pellet reextracted using the same procedure. The organic layer was again removed and combined with the first extraction product. The organic supernatant was taken to dryness in a Savant Speed Vat. The extracts were stored dry at -20°C and redissolved in 0.30 ml of acetone just prior to analysis by HPLC. Samples of 5-150 ~1 of the concentrate were injected for analysis. Extraction
from
Urine
Determination of hypericin levels in urine samples was performed by extraction of hypericin from urine, and quantitation of the compound by HPLC was performed using the same methods as described for plasma. Human urine samples were similarly subjected to a 45min HIV heat-inactivation step. HPLC
Analysis
Samples were chromatographed by gradient elution on a reverse-phase, pheny14 pm, 0.45 X lo-cm radialpak analytical column (Waters Associates, Milford MA). The chromatographic system consisted of two Knauer Model 64 pumps connected to a dynamic mixer (Sonntek, Woodcliff Lake, NJ) along with a WISP Model 710B autosampler (Waters Associates). The pumping system and WISP were controlled by an Axxiom 747 multisystem controller-data system (Cole Scientific, Calabasas CA). Fluorescence was detected using an Applied Bio-systems Model 970 fluorescence detector (Ramsey, NJ), and visible absorbance at 590 nm was measured by a Linear Model 203 variable uv/vis detector (Reno NV). Mobile phase solution A consisted of a 70% solution of 0.1% ammonium phosphate (adjusted to pH 7.0 with NaOH) and 30% acetonitrile; solution B was 70% acetonitrile-30% water. A linear gradient of 100% A to 100% B was developed over a 15-min interval
OF
79
HYPERICIN
with a flow rate of 1.2 ml/min, followed by 4 min of 100% B. The reequilibration of the column was achieved by a linear change from 100% B to 100% A over the next 4 min followed by 7 min of isocratic A. This resulted in a complete analysis cycle time of 30 min. The fluorescence detection used excitation at 470 nm and emission using a cutoff filter limiting light transmission below 550 nm. Detection in the visible range was at 590 nm. Hypericin eluted with a retention time of 12.9 min while WIS-4 eluted at 22.2 min (Fig. 6). The assay has a lower limit of sensitivity of 250 pg injected onto the column, achievable with both the visible absorbance and fluorescence detection systems (sensitivity setting of 0.02 PA = 1 V full scale on the fluorescent detector and 0.1 A = 1 V full scale for the absorbance detector).
Calibration
Procedure
and Evaluation
of the Method
Standard solutions in ethanol were added to normal donor control plasma with hypericin at concentrations ranging from 0.050 to 40 pg/ml. The back-calculated hypericin concentrations were normalized against the unextracted hypericin standards. A linear regression analysis was used to determine the degree of fit of the data.
Pharmacokinetic
Anulysis
Pharmacokinetic values were obtained by first using the JANA curve stripping program to estimate the initial pharmacokinetic parameters prior to modeling using PCNONLIN (Statistical Consultants, Lexington, KY). The estimates were used as the initial parameters for a two-compartment bolus injection absorption model. The program used these parameters to fit the data to the best curve possible and then applied Lagrange techniques to reduce the total variance of the key pharmacokinetic values (11).
Recovery of Hypericin
from Plasma
Additions of hypericin diluted from a 2 mg/ml stock solution prepared in ethanol were added to murine, simian, and human plasma samples. The samples were mixed on a Vortex Genie, allowed to stand for 30 min at room temperature, and then frozen at -20°C for at least 24 h until subjected to extraction as described above. Efficiency of hypericin recoveries was calculated from the ratio between the amount of hypericin recovered from a standard amount added to plasma and to an equivalent volume of ethanol. The latter samples of hypericin in ethanol were not extracted but were taken to dryness in the Speed Vat and treated identically to the plasma samples.
80
LIEBES
ET
AL. TABLE
2.3
Hypericin
Visible
Ethanol
LIX 8 5 e 8 9
590 547 598 558
-0.1 II
200
I
I
I
I
300
400
500
600
Wavelength FIG. 2. Ultraviolet/visible ethanol. Concentration,
absorbance 22.4 pg/ml.
700
(nm) spectrum
of hypericin
in
Molar
1 Extinction t
Coefficients Water
t
46,000 22,000 6,000 10,060
_
opment of an extraction methodology which allowed dissociation of hypericin from protein complexes and removal of proteins.
Spectroscopic Studies Absorbance properties
of HY in different solvents. Hypericin exhibits a number of absorbance maxima in BALB/c mice (20 g weight) were injected with 350 pg the visible and ultraviolet regions as has been documented (6,7,12). In ethanol, the absorbance maxima are hypericin dissolved in 0.5% aqueous benzyl alcohol from and 234 nm (Fig. a 5 mg/ml stock solution. Animals were injected in pairs seen at 590,547,510,474,383,340,285, 2). In aqueous media, however, substantial differences and each animal was bled 0.5 ml from the retroorbital plexus of the eye into heparinized tubes. One pair of in the visible spectrum occur. The visible maxima at 547 mice was bled once for each of the sampling points (0, and 590 nm are most sensitive to the type of solvent environment and are “red-shifted” in aqueous media to 0.25,0.5,1,2,4,6,8,24,36,72,120,192,216, and 240 h). 558 and 598 nm, respectively, along with considerable The plasma was separated by centrifugation at 1OOOg reductions in the molar extinction coefficients (Table 1, and subjected to the extraction protocol described Fig. 2, and Fig. 3, spectrum A). above. Results for each pair were averaged together for Similarly, with the addition of HSA to a neutral each time point. aqueous suspension of HY of the same concentration as that in Fig. 3, spectrum A, the 598-nm peak is “blueSpectroscopic Analyses shifted” toward 590 nm as well as increased in intensity Hypericin was dissolved by sonication in water or while the 558-nm peak is red-shifted to 560 nm (Fig. 3 ethanol at a concentration of 2 mg/ml as described spectrum B). When compared to that of hypericin in above. A solution containing 20 PM human serum albu- water alone, a more dramatic increase occurs in the size min (HSA) in Hepes buffer, pH 7.0, was mixed with different amounts of HY ranging from 0.5 to 100 PM. Visible range spectral profiles were generated using a Beckman DU7 spectrophotometer (Fullerton, CA). Fluorescence data were obtained using a Hitachi PerkinElmer SPF spectrofluorometer (Norwalk, CT) configured with excitation and emission slits at 1 nm. The excitation wavelength was set at 470 nm for the generation of fluorescence emission scans.
Murine
Pharmacology
RESULTS
The spectral properties of hypericin in aqueous, organic, and physiologic media were analyzed as part of our development of a sensitive method for the quantitation of hypericin in biologic fluids. These properties were found to vary with the type of media and solvent environment in which hypericin was dispersed. In addition, serum proteins avidly associate with hypericin and affect its biophysical properties. This required the devel-
Wavelength FIG. 3. water and persed in with 5 mM The same
(nm)
Ultraviolet/visible absorption spectrum of hypericin (A) in (B) in water in the presence of HSA. Hypericin was diswater by sonication (concentration, 2 mg/ml) and diluted Hepes buffer (pH 7.0); final concentration, 48 pg/ml. (B) solution in the presence of 50 Fg HSA.
CHROMATOGRAPHIC HY: 1:l 1
HSA
Molar
2:l
3:1
I
I
QUANTITATION
Ratios 4:l
5:l I
I
I
0
VM
FIG. 4. Visible absorbance and fluorescence changes as a function of different hypericin/HSA ratios. Fluorescence using 470 nm excitation and 600 nm emission (A) (arbitrary units/lo), absorbance at 560 nm (0), and absorbance at 590 nm (Cl) are plotted as a function of HY concentrations which were added to a fixed concentration of HSA of 20 PM.
of the 590-nm absorbance peak (0.90 A) than is apparent with the 560-nm absorbance increase (0.27 A). The absorbances at both 590 and 560 nm increase in direct proportion to the amounts of hypericin added to a fixed concentration of HSA (20 PM) but at different rates for each of these wavelengths (Fig. 4). For HY:HSA ratios of up to l:l, the increase in absorbance was fit by slopes of 0.0154 and 0.0268 for measurements at 560 and 590 nm, respectively. For HY:HSA ratios between 2:l and 5:l the plots of the absorbance changes were similar (slope = 0.0105 for 560 nm and 0.0108 for 590 nm). Increases in fluorescent emission intensities of HY at 600 nm were linear with respect to the amount of HY up to a protein:hypericin molar ratio of 1:l (slope = 0.129). Further increases in the concentration of HY relative to HSA beyond a 1:l molar ratio are associated with nonlinear decreases in fluorescence. Fluorescence properties. The differences in spectral absorption properties of hypericin in aqueous media in comparison with polar organic solvents are also reflected in fluorescence spectra (Fig. 5). In polar organic solvents, such as ethanol, hypericin gives a brightly red fluorescent solution with emission maxima at 595 and 645 nm (Fig. 5, spectrum C). In aqueous suspensions (Fig. 5, spectrum A), no fluorescence is observed. Upon addition of HSA to the aqueous suspension of hypericin, fluorescence can be detected with emission spectra (Fig. 5, spectrum B) somewhat similar to those obtained for hypericin in ethanol. Chromatographic
Resolution
of Hypericin
and DHHA
Figure 6 shows chromatograms obtained from human plasma to which 1.0 pg/ml hypericin and 4.0 pg/ml
81
HYPERICIN
DHHA have been added as well as a plasma control containing only DHHA. The plasma samples were extracted with organic solvents and detection was by visible absorbance at 590 nm. The chromatographic peak of hypericin resolved well from that of DHHA which elutes 9.3 min after hypericin. The blank extracted plasma control was free of any other plasma components detectable at this range with the exception of the added DHHA (Fig. 6).
6:l
Recoveries
HY Total,
OF
of Hypericin
from Plasma of Different
Species
Hypericin was extracted from the plasma of mice, monkeys, and humans to determine whether any differences in the efficiencies of recovery exist between the various species used in preclinical studies of hypericin. The recovery from plasma samples which contained concentrations of hypericin ranging from 0.125 to 15 pg/ml was compared with direct standards of hypericin in ethanol. Figure 7 shows that these data can be fit by a straight line with a slope of 0.919. A regression analysis of these data yielded a correlation coefficient of 0.991. The regression analysis of individual data for each species were 0.999,0.998, and 0.997 respectively for human, simian, and murine plasma samples. The data as well as those summarized in Table 2 show consistent recoveries of hypericin throughout this wide concentration range that was equivalent for plasma from the different species which were studied. Assay Precision
and Detection
Limits
The detection limits for routine assays were determined at 10 ngfml when l.O-ml samples of plasma were used for analysis. The interassay precision was determined by replicate analyses of plasma to which 1000 (N
600 Wavelength
Sk0 (nm)
FIG. 5. Hypericin fluorescence emission spectra in aqueous and organic solutions. Hypericin dissolved and sonicated in pH 8.0 water at a concentration of 2.0 mg/ml was diluted to 3.3 pg/ml with water (A), water containing 333 pg/ml HSA (B), and ethanol (C). Fluorescence emission spectra were obtained using 470 nm excitation.
82
LIEBES
ET
AL.
I
DHHA
(22.2)
0
Elutlon FIG. 6. HPLC chromatographic (. - + - * ). An extract from human
profile plasma
of human containing
for DHHAIHY
Recovery from Plasma
The recovery of 1 @g/ml of DHHA from human plasma was found to be 96.6% + 5.8% (N = 8) of the direct control which was not subjected to extraction. A combination of DHHA (1 pg/ml) with hypericin throughout a concentration range of 0.05 to 40 pg/ml was used to evaluate the recovery of hypericin in the presence of this internal standard. Figure 8 shows the existence of a high degree of linearity with this combination throughout this concentration range. Recovery
of Hypericin
from Human
Analysis
of Hypericin
plasma
containing
1 pg/ml
hypericin
and 4 pg/ml
Wis 4
in 0.5 ml of 0.5% benzyl alcohol. The animals were bled at time intervals ranging from 10 min to 240 h. The plasma samples obtained were extracted and analyzed as outlined above. The concentration vs time data from this experiment (Fig. 9) show a peak concentration of 27.8 pg/ml at 0.17 h (10 min) and decreasing measurable values up to 240 h (10 rig/ml). The data were well approximated using a two-compartment model contained in the PCNONLIN pharmacokinetic library, allowing for a first-order input. The distribution phase (a halflife) was 2.0 h, while the elimination phase (0 half-life) was 38.5 h. The volume of distribution was 12.6 ml. The kinetic equilibrium constants (Table 5) indicate a strong tissue absorption phenomenon (4, = 0.1 vs k,, = 0.2) that contributes to an extended p half-life of 38.5 h.
Urine
Hypericin recoveries from urine were lower than those obtained with plasma with a range from 63 to 75% (Table 4). This lower recovery range was also reflected with DHHA. The ratio of DHHA to HY serves as an adequate correction for the effects of the various salts present in urine. Pharmacological
(Minutes)
plasma extracts. An extract from only 4 pg/ml DHHA (+++I.
= 7) and 2500 rig/ml (N = 9) of hypericin were added. Table 3 shows that the percentage coefficient of variations for both of these concentration ranges was ~2%. Range of Linearity
Time
in BALBIc
Mice
Pharmacokinetic data were analyzed in mice that received a bolus iv injection of 350 pg (17.5 mg/kg) of HY
DISCUSSION Hypericin has been found to possess antiretroviral properties, acting directly on the retrovirion particle leading to the loss of infectivity (5,B). Any potential clinical applications of hypericin require an assay for determination of levels of drug in blood and other biological fluids. The hydrophobic properties of hypericin and a complex pattern of molecular interactions in aqueous and polar organic solvents necessitated particular care in
CHROMATOGRAPHIC
QUANTITATION
OF
Hypericin Hypericin level (w/ml) 1000
2500
FIG. 7. The plot of hypericin determinations from extracted plasma of three species: murine (O), simian (A), and human (Cl). The data are the mean and standard deviation of two separate plasma extractions. The levels are correlated against hypericin determinations from direct standards and analyzed by a linear regression, yielding a slope of 0.919 and a correlation coefficient of 0.991 for the aggregate data from all species. The correlation coefficients for each of the species were 0.999, 0.998, and 0.997 for human, simian, and murine plasma samples, respectively.
designing this assay. In the course of this study it became apparent that an assay by direct fluorescence measurement, while highly sensitive in the presence of serum or plasma protein components, was not suitable for quantitation of HY. Thus, our data showed an increase in fluorescence intensity at HY:HSA molar ratios of up to l:l, but at higher levels of HY the fluorescence intensities were markedly attenuated. It appears that low ratios of HY to protein yield a linear increase of fluorescence while higher ratios lead to quenching of fluorescence. The use of absorbance measurements directly on samples of plasma and other biological fluids for HY quantitation at either 560 or 590 nm was also not feasible because of nonlinearity observed at different HY/HSA ratios. Similarly, data obtained with HY additions to
TABLE Hypericin
Recovery
83
HYPERICIN
Precision
TABLE
3
Data
from
Calculated average
Human
Plasma
Number of analyses
SD
%CV
7 9
19 39
1.82 1.61
1028 2407
whole plasma have shown nonquantitative absorbance spectral changes (unpublished data). Initial studies to determine recovery efficiencies using either C8 or Cl8 solid-phase extraction columns showed hypericin recovery from plasma to be very poor (O-14% HY recovered, data not shown). The use of membrane centrifugation columns also gave poor results, suggesting that the association of hypericin with proteins is of a strong nature. The quenching of fluorescence at high hypericin to protein ratios, along with the nonlinear, concentration-related spectroscopic changes which occur when HY interacts with serum proteins, renders quantitation of hypericin in biological fluids by direct measurements of fluorescence, or by absorbance changes, inaccurate. We thus developed a protocol which included: (i) The extraction of hypericin from plasma using a solvent mixture which allowed the concomitant precipitation of plasma proteins and the maintenance of HY in a soluble form. (ii) The chromatographic separation of the molecule in a polar organic solvent environment that provides optimal use of absorbance and of fluorescence properties of the molecule for quantitative analysis. The feasibility of the use of HPLC for the analysis of hypericin has previously been shown in the separation
2 in Plasma
HY level (w/ml)
500 (%)
2500 (%)
5000 6)
10,000 (X0)
Human SD Mouse SD Monkey SD
91.4 12.9 92.1 7.4 84.1 1.0
85.3 7.5 83.1 16.4 92.0 3.2
93.4 a.7 100.0 11.6 88.4 0.7
87.3 8.0 100.0 9.0 94.5 11.9
0
10
20
HY
Concentration
30
4 D
(pglml)
FIG. 8. Plot of hypericin/DHHA area ratios from human plasma extracts as a function of hypericin over a range of 0.05 to 40 pg/ml with the DHHA concentration at 1 fig/ml.
a4
LIEBES TABLE
Hypericin HY level (na/ml) Human SD
urine
Recoveries
AL.
4
TABLE
in Human
Urine
500 (%)
2500 (%)
5000 (%)
10,000 60)
63.4 7.6
65.9 21.2
75.2 0.9
65.9 31.0
. 4 I-
O, 200
100
Time,
5
Pharmacokinetic Estimates of iv Administration of 350 pg Hypericin to BALB/c Mice
by reverse-phase chromatography of hypericin and pseudohypericin from Hypericum plant extracts (13,14). These analyses used analytical methodology similar to that reported here but were devised mainly for quantitation of hypericin in various plant preparations by different solvent elution cycles. We have employed a phenyl reverse-phase column to provide an environment which is less hydrophobic than the Cl8 or C8 columns used in these previous studies. The phenyl solid phase allows a substantial retention of hypericin and a reduction in analysis time to a 30-min cycle as compared to the longer retention time reported previously (14). The organic extraction procedure provides the necessary milieu to efficiently extract hypericin from plasma and urine while precipitating the protein components. The recoveries, which range from 85-100% when extractions of hypericin are repeated twice, do not seem to vary among different animal species and are equivalent in reference to the DHHA internal standard. The extracted hypericin dissolves readily in ethanol or acetone and is eluted from the chromatographic col-
0
ET
hr
FIG. 9. Hypericin plasma levels as a function of time. BALB/c mice were injected with a bolus injection of 350 fig hypericin in 0.5% benzyl alcohol. Fit of plasma HY concentration from two-compartment bolus injection model using PCNONLIN (-) as compared with the actual data (*).
Parameter C,, bdml) Volume (ml) k 21 k 10 k k;;-HL (h) AUC (pg/ml/h) (u-HL (h) 8-HL (h)
Estimate 27.6 12.6 0.1 0.4 0.2 7.1 285.1 2.0 36.5
umn in an organic solvent which provides optimal retention times and enables monitoring at 590 nm and confirmation by fluorescence using excitation at 470 nm. This procedure should be readily adaptable to the analysis of hypericin content in other biological material such as feces and tissue with appropriate modifications to allow for adequate dispersion of the material prior to extraction. We have found that the use of simultaneous detection of visible fluorescence emission with a cutoff filter which limits light below 550 nm along with absorbance detection at 590 nm provides additional levels of discrimination for hypericin to aid in discerning the presence of metabolites. The high molar extinction coefficient for hypericin of 46,000 provides a level of detection sensitivity by absorbance that is at least equal to that of fluorescence detection for most compounds and possibly better given the good signal to noise ratios obtainable with modern variable wavelength HPLC detectors. In summary, the association of hypericin with proteins results in nonlinear fluorescence and absorbance changes. A reliable assay for hypericin requires organic extraction, as solid-phase extraction and membrane filtration are not adequate to remove proteins. The recovery using a dual extraction technique ranges from 85100%. The HPLC separation between hypericin and the internal standard DHHA using a phenyl reverse-phase column is greater than 9 min and utilizes a solvent mixture that optimizes detection at the maximum visible extinction coefficient of the molecule. The use of visible detection provides an ample signal to noise ratio and results in no discernable interfering peaks. The preliminary murine pharmacokinetics show detectible levels out to 240 h and we expect that this assay will provide accurate results for the study of human pharmacokinetits in phase I studies. REFERENCES 1. Brockman, H. V., Falkenhausen,E. Budde, G. (1950) Naturwissenckuften
H., Neeff, R., Dorlars,A., 37,520-545.
and
CHROMATOGRAPHIC 2. Buchner, 217. 3. Muldner, 4. Daniel,
A. (1830)
Buchner$
Reperturium
QUANTITATION
Phurmacie
H., and Zoller, M. (1984) Arzneim.-Borsch. K. (1949) Hippocrates 19, 526-530.
5. Meruelo, D., Lavie, USA 85,5230-5234. 6. Brockmann, 90,2302-2318.
G., and Lavie,
H., Kluge,
D. (1988)
F., and Muxfeldt,
PFOC.
IM 34,
Natl.
H. (1957)
34,
918.
Acad. Chem.
Sci. Ber.
7. Lavie, D., Freeman, D., Bock, H., Fleischer, J., Van Kraanenburg, K., Ittah, Y., Mazur, Y., Lavie, G., Liebes, L., and Meruelo, D. J. Pure Appl. Chem. in press. 8. Lavie, G., Valentine, F., Levin, B., Maxur, Y., Gallo, G., Lavie, D.,
OF
HYPERICIN
85
Weiner, D., and Meruelo, D. (1989) PFOC. Natl. Acad. Sci. USA 86,5963-5967. 9. Yarchhoan, R., Mitsuya, H., Myers, C. E., and Broder, S. (1989) N. Engl. J. Med. 321, 726-738. 10. Larder, B. A., Darby, G., and Richman, D. D. (1989) Science 243, 1731-1734. 11. Draper, N. R., and Smith, H. (1981) Applied Regression Analysis, 2nd ed. Wiley, New York. 12. Brockmann and Eggers, H. (1958) Chem. Ber. 91, 81. 13. Eckhardt, V. W. E. (1984) Dtsh. Apoth. Ztg. 124, 2383-2386. 14. Holzl, J., and Ostrowski, E. (1987) Dtsh. Apoth. Ztg. 127, 12271230.
