ANALYTICAL
BIOCHEMISTRY
58-68 (1978)
90,
Analysis of Steroidal Carboxylic Acids by High Pressure Liquid Chromatography RENE Hospital for
L.
FARHI
AND
CARL
MONDER
for Joint Diseases und Mediwl Center, Research Institute Skeletomuscular Discuses. 1919 Madison Avenue, Net.t, York. Ne,c, York 10035 Received
October 3 1, 1977
A procedure utilizing high pressure liquid chromatography has been developed for the analysis of steroids containing 17P-carboxylic acid or 20-hydroxy-21-oic acid substituents. Acids were converted to the p-bromophenacyl esters and were separated on octadecylsilyl columns with methanol:water mixtures as mobile phase. Structurally similar steroids, including the 20~ and 2Op- epimers of the 21-oic acids were well separated. The lower limits of detection ranged from 0.25 to 0.5 nM. The method is simple, rapid. quantitative, sensitive, and specific.
The cortoic acids are a group of acidic metabolites of cortisol which are excreted in the urine in amounts that represent, in normal individuals, about 6% of the products of cortisol metabolism (1). In certain diseases, the proportion of cortoic acids in the urine may change up or down relative to the neutral metabolites (2). In order to determine if these changes are of clinical significance, a method is needed for measuring the amounts of the individual cortoic acids excreted. Four quantitatively important representatives of this class of acids have been isolated (Fig. 1). All are C,, hydroxy acids, and are devoid of chromophores which absorb in the near ultraviolet. In order to measure these acids spectrophotometritally by high pressure liquid chromatography (HPLC)’ it was essential to convert them to derivatives that absorb maximally at 254 nm. Formation of the derivatives had to be specific, quantitative, and reproducible and had to yield products which were easily separated and were readily ’ Abbreviations used: HPLC, high pressure liquid chromatography; ODS, octadecylsilyl; DBAP, cY,p-dibromoacetopherone; THFEA, 3a,l lp,l7-trihydroxy-5-androstane-l7P-carboxylic acid; THEEA. 3cy.l7-dihydroxy-1 I-oxo-5/3-pregnane-17p-carboxylic acid: cu-FHA, 11/3,17.20o-trihydroxy-3-oxo-4-pregnen-2l-oic acid; P-FHA, 1lp,l7.20/3-trihydroxy-3-oxo4-pregnen-21-oic acid: cu-EHA.17.20~dihydroxy-3.1 I-dioxo-4-pregnen-21-oic acid; P-EHA. 17,20P-dihydroxy-3.1 I-dioxo-4-pregnen-21.oic acid; FEA, 1l/3.17-dihydroxy-3-oxo-4-androstene-17Pkarboxylic acid; EEA, 17-hydroxy-3,1 I-dioxo-4-androstene-17p-carboxylic acid; a-DOCHA, 20a-hydroxy-3-oxo-4-pregnen-21-oic acid: P-DOCHA, 20P-hydroxy-3. oxo-4-pregnen-21-oic acid; DOCEA, 3-oxo-4-androstene-17@carboxylic acid; AUFS, absorbancy units full scale. 0003-2697/78/0901-0058$02.00/O Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.
58
HPLC
OF STEROIDAL
59
ACIDS
COOH
COOH
H-L-OH
HO-t-H
..,.&pOH
,-..poH H
Cortolonlc
H acid
B-Cortolonlc
acid
COOH
COOH
H-&OH
HO-&H
Ho,,poH
Ho,._‘;:l:_-“’
H Cortolsc
FIG.
H ocld
a-Cortollc
1. Representative
steroidal
21.oic
acid
acids.
detected at low concentrations. In this paper, we describe the development of an analytical technique for measuring the cortoic acids, based on the method of Durst et al. (3) for preparing derivatives of the fatty acids. METHODS
Instrumentation Analyses were performed on a DuPont liquid chromatograph, Model 830, fitted with an oven and a gradient system. Injections were made through a Rheodyne injection valve, Model 71-20. Zorbax octadecylsilyl (ODS) columns, 2.1- or 4.6mm i.d., 250 mm long, were washed with analytical grade methanol until a stable baseline was obtained at high sensitivity. The 254-nm photometer (DuPont Model 842) contained an g-mm flow cell. A Varian A-25 recorder was used. Pressure input was usually 3000 psi; it was raised to 4500 psi when separations required the use of two columns in series. The oven temperature was 40°C. The flow rate was determined by measuring the volume of effluent collected in a given time. Spectra in the ultraviolet range were obtained on a Cary Model 15 spectrophotometer. Chromatographic
Conditions
Early work was done with an injector encompassing a rubber septum. At high instrument sensitivities spurious and irreproducible peaks were seen on the chromatogram as a result of the action of the injected acetonitrile on the septum. This was eliminated by the use of the injection valve. Injection volumes for the 2.1- and 4.6-mm i.d. columns were 10 and 50 ~1, respectively. The wider column gave sharper peaks due to “infinite wall
60
FARHI
AND
MONDER
effect” with a larger flow rate. Under these conditions, decreased by a factor of two.
the sensitivity
was
Solvents Chromatography was performed with mixtures of methanol (Baker, AR) and water redistilled from glass. To prevent the slow formation of turbidity in these mixtures the water, fresh from the still, was immediately sterilized. At the time of use, it was filtered through Millipore GS (0.22 pm) and the methanol was filtered through Millipore FG (0.2 pm) membranes. Handling and filtering of the water and storage of the alcohol:water mixtures were done using sterile glassware. Methanol could not be replaced with acetonitrile or isopropanol as eluting solvent. Chemicals cw,p-Dibromoacetophenone (DBAP) was obtained from Aldrich Chemical Company. Analysis of the reagent as received by HPLC on an ODS column using gradient elution with methanol:water mixtures showed presence of two minor impurities which did not interfere with the analytical procedure. The DBAP was used without further purification. p-Bromoacetophenone (Aldrich Chemical Co.) was used as received. Dicyclohexano-l&crown-6 (crown ether) was used as recommended by Durst et al. (3). Water was removed by azeotropic distillation with benzene. The solvent was eliminated on a rotary evaporator, and the dry crown ether residue was kept in a stoppered flask. Acetonitrile was either “Distilled-in-glass” bought from Burdick and Jackson Company, or was Mallinckrodt Nanograde. Analysis of the solvent on an ODS column with methanol: water (65:35) as eluant showed the presence of a uv-absorbing impurity which was not retained, and did not interfere with the analysis. Other reagents and chemicals were analytical grade. Silica gel was E Merck, range 0.2-0.5 mm. Steroid Acids The 20-hydroxy-21-oic acids were synthesized as described by Lewbart and Mattox (4). The 20/3 epimers were synthesized from the 20j%hydroxy21-aldehydes (5) by enzymatic oxidation of the aldehyde group (6). C,, acids (etienic and etianic acids) were synthesized as described by Mason et af. (7). The acids used in this study were 3a, 1 1/3,17,20a-tetrahydroxyS/3-pregnan-21-oic acid (cy-cortolic acid); 3a-1 l/3,17,20/3-tetrahydroxy-5/?pregnan-2 1-oic acid (/I-cortolic acid); 3a- 17,20a-trihydroxy- l l-0x0-Sppregnan-21-oic acid (cY-cortolonic acid); 3a-17,20/3-trihydroxy-11-0x0-5/3pregnan-21-oic acid (p-cortolonic acid); 3a,l l/3,17-trihydroxy-5-androstane-17@carboxylic acid (THFEA); 3a, 17-dihydroxy-l l-oxo-5@pregnane17/3-carboxylic acid (THEEA); 11/3,17,20c&rihydroxy-3-oxo-4-pregnen-
HPLC OF STEROIDALACIDS
61
21-oic acid (a- FHA); 11/3,17,20fl-trihydroxy-3-oxo-4-pregnen-21-oic acid (/3-FHA); 17,20a-dihydroxy-3,11-dioxo-4-pregnen-21-oic acid (a-EHA); 17,20/3-dihydroxy-3,11-dioxo-4-pregnen-21-oic acid (/3-EHA); 11/3,17-dihydroxy-3-oxo-4-androstene-17/3-carboxylic acid (FEA); 17-hydroxy3,11-dioxo-4-androstene-17/3-carboxylic acid (EEA); 20a-hydroxy-3oxo-4-pregnen-21-oic acid (a-DOCHA); 20/3-hydroxy-3-oxo-4-pregnen21-oic acid (/3-DOCHA); 3-oxo-4-androstene-17/3-carboxylic acid (DOCEA).
Preparation of Reagents (a) Potassium methoxide, 0.5 m. Pellets of 85% pure potassium hydroxide (332 rag) are dissolved in methanol to obtain 5 ml of solution which is then filtered on Millipore FG filter. The reagent need not be standardized and is stored in a tightly capped container. This stock solution is diluted as needed. (b) Crown ether solution. About 12 mg of crown ether is dissolved in 1 ml of acetonitrile and is filtered through a Millipore FG membrane. (c) Alkylating reagent. To 100 mg of DBAP are added 1 ml of crown ether solution and 9 ml of acetonitrile. Molar ratio of DBAP to crown is about 10. Solutions of DBAP in acetonitrile slowly decompose and are kept in the freezer. (d) Standard solutions o f acids. One mg of each acid was dissolved in 0.10 ml of methanol followed by 0.90 ml of acetonitrile. The solution was filtered on a Millipore FG membrane and stored in a freezer. Derivatization Procedure Analytical preparation of derivatives was carried in standard 1-dram screw-cap vials with special caps and Teflon-covered rubber septa (Pierce Chemical Co., Catalog Nos. 13915 and 12412); 10 × 3 mm o.d. Tefloncovered magnets were used for stirring. After each use, the vials were washed with water, 95% ethanol, and acetone, and were then air dried. Steroid acid standard solution in a volume of 1.5 to 250/xl, measured with a Hamilton syringe, was placed in a vial, and acetonitrile was added to a final volume of 0.95 ml. Fifty microliters of 0.01 M potassium methoxide was added slowly with constant magnetic stirring, followed by 40 ~1 of alkylating reagent. The vials were capped and heated in a water bath at 75 to 80°C with stirring for 20 min. The contents were cooled and chromatographed directly. Quantitative analysis of the chromatographic profile was performed after cutting out and weighing the peaks corresponding to the steroid acid derivatives.
Preparative Synthesis of Derivatives Steroid acids were converted to p-bromophenacyl derivatives in milligram amounts in order to confirm structures, study spectrophotometric
62
FARHI
AND
MONDER
properties, and provide analytical standards. A total of 30 to 40 mg of acid was dissolved in 3 to 4 ml of methanol in a lo-ml screw-cap vial. A milligram of phenolphthalein in 0.1 ml of methanol was added, and the solution was titrated to a red color with 0.1 M potassium methoxide. The methanol was evaporated under nitrogen. To the residue were added 0.1 ml of methanol, 4 ml of acetonitrile, 0.75 ml of crown ether solution, and 75 to 100 mg of DBAP dissolved in 2 ml of acetonitrile. The capped vial was heated for 1 to 2 h in a water bath at 75 to 80°C with magnetic stirring. Initially, the liquid was clear, with pink patches along the wall of the vial, but very quickly a turbidity of potassium bromide appeared, while the patches turned white and disappeared. After cooling, the solution was passed through a small column of silica gel (5-6 g) to remove the crown ether. The vial and column were washed with 25 ml of dichloromethane which was added to the eluate. The derivatives were purified by chromatography on a prepacked column of silica gel (EM Catalog No. 10607) with mixtures of methanol and chloroform. Elution of the derivatives was achieved with the proportions of methanol to chloroform indicated in the parentheses: EHA (1:99); cortolonic acid (2:98); EEA (2:98); THEEA (5:95). The derivatives were crystallized from mixtures of chloroform and hexane. Anal. Calcd for-p-bromophenacyl cortolonic acid, &,H3,0,Br: C, 60.31; H, 6.45; Br, 13.84. Found C, 60.07; H, 6.83; Br, 13.48. Anal. Calcd forp-bromophenacyl EEA, C,,H,,O,Br, C, 61.88; H, 5.75; Br, 14.70. Found C, 61.19; H, 5.77; Br, 15.91. Anal. Calcd for-p-bromophenacyl THEEA, C, 61.43; H, 6.44; Br, 14.59; Found C, 61.40; H, 6.65; Br, 14.67. RESULTS
Spectral Characteristics
of the Derivatives
Figure 2 shows the absorption spectrum of a representative steroid acid derivative. In this, and in all other derivatives, the spectral maxima were at 254 nm. The contributions of the phenacyl chromophore and of the 3-oxo-4-ene moiety of the steroid in Fig. 2 are additive. The two chromophoric groups are therefore noninteracting. Where the phenacyl group of the steroid derivative is the sole contributing chromophore, as in bromophenacyl cortolonic acid, the extinction coefficient is the same as that of p-bromoacetophenone. Neutralization
Procedure
Durst et al. (3) had suggested alternate procedures for converting the carboxylic acid to its potassium salt. These included titration of the alcoholic solution of the acid to an indicator end point or stirring the acid
HPLC OF STEROIDAL
63
ACIDS
-
350
FIG. 2. Ultraviolet absorption spectra of steroid acid derivatives. (----) phenone; (- . -) 17,20/3-dihydroxy-3,1I-dioxo-4-pregnen-21-oic acid; (---) tive, calculated; (-----) P-EHA derivative, found. Solvent was methanol.
DibromoacetoP-EHA deriva-
with a three to fivefold excess of potassium bicarbonate. In our hands, these approaches proved to be inadequate. Variations, including adding the base in water, use of other solvents, sonication of the suspension, or use of potassium carbonate as base, were unsuccessful in improving reproducibility. We subsequently found that a fixed molar excess of methanolic potassium hydroxide added as described under Methods gave an acceptable yield of derivative. When steroid hydroxy acid was briefly exposed to 0.01 M potassium methoxide, there was progressive destruction of the steroid side chain and decreased ester formation which was intensified as the ratio of base to steroid increased. In order to avoid this problem, the acid was first diluted with acetonitrile, then slowly neutralized with 50 ~1 of 0.01 M potassium methoxide prior to addition of alkylating reagent. Reagents
A threefold excess of DBAP was needed for complete derivatization as reported by Durst et al. (3) and Fitzpatrick (8). The molar ratio of DBAP to crown ether was 10. Yield of product was not diminished by the presence of 10 ~1 of water or methanol in the reaction mixture.
64
FARHI
Reaction
AND
MONDER
rates
Figure 3 shows that the rate of derivative formation depended on the concentration of steroid acid. The reaction was faster in more dilute solution. The yield was independent of the concentration of steroid after 20 min. Calibration
curves
Dependence of derivative formation on steroid concentration is illustrated in Figs. 4 and 5. The upper range is limited by the amounts of reagents used. The lower range is limited by several factors: Acids with a conjugated double bond system in the A ring have a lower limit of detectability than do acids with a reduced A ring. Other factors are the methanol and DBAP concentrations. Interference by the latter at very high sensitivity results from traces of decomposition products of the reagent which emerge from columns in the general region where acid derivatives are eluted. DBAP is eluted before any of the bromophenacyl esters and therefore does not itself interfere with the emerging steroids. At high sensitivity, however, the trailing foot of the reagent may emerge from the column coincident with some of the faster moving acid esters, making quantitation of the emerging peak difficult in those cases. The CzOetianic acids can be measured down to 0.25 nM. Concentration 100
Reaction FIG. 3. procedure tration of raphy was nohwater
time (minutes)
Yield of bromphenacyl derivative of FHA as function of time. The derivatization was performed as described in the text with various reaction times. Concensteroidal acid used in the reaction: (0) 12.5 pgiml; (A) 100 &ml. Chromatogperformed on a Zorbax ODS column, 4.6 x 250 mm; oven, 40°C; solvent. metha(65:35); pressure, 3000 psi; flow rate, 1.9 ml mitt-‘.
HPLC OF STEROIDAL 4000
3000
65
ACIDS
-
-
1
5
IO
15
20
Amount
25
30
35nM
Injected
FIG. 4. Calibration curve for the p-bromophenacyl ester of 3a,l7-dihydroxy-ll-oxo-5ppregnane-17P-carboxyhc acid (THEEA). Absorption measured at 254 nm in an 8-mm cell. Other conditions were as described under Methods.
Amount
injected
FIG. 5. Calibration curve for thep-bromophenacyl pregnan-21-oic acid (cortolonic acid).
ester of 3a, 17,20-trihydroxy-I
1-oxo-Sp-
66
FARHI AND MONDER
dependence was linear in the entire range of steroid used (Fig. 4). The concentration curves of the 20-hydroxy acids were linear as well over most of their range (Fig. 5). There was a slight curvature at the lowest steroid levels, because of some acid decomposition by excess base in this region. The lower practical limit of measurement was 0.5 nM. The relative standard deviation of measurement varied from about 2 to 3% in the upper concentration range to about 10% in the lower range with all steroid acids. Chromatographic Profiles
Examples of separations are shown in Figs. 6 and 7. It is of interest that steroids epimeric at C,, were well separated. The pairs a-FHA-a-EHA and p-cortolic acid-P-cortolonic acid were incompletely resolved, but use of two columns in series improved the resolution. For all the 20-hydroxy acids studied, the 20a isomer eluted before the 20/3 isomer. Table 1 shows the retention times of various bromphenacyl acids. Complete separations were achieved in most cases. In order to completely elute acids of widely different polarity from the columns, the solvent schedule was adjusted in a stepwise manner. DOCHA and DOCEA required larger methanol concentrations (methanol:water = 75:25) to get reasonable retention times and sharp peaks.
FIG. 6. Separation of a mixture of p-bromophenacyl esters of cortolic and cortolonic acids. Two Zorbax ODS columns, 4.6 x 250 mm, were placed in series. Oven temperature, WC; solvent, methanoLwater; 65:35; pressure, 4500 psi; flow rate, 1.5 ml min-‘; uv detector at 254 nm; 8-mm cell, 4 x 10m2AUFS. The peaks are: 1, a-cortolic acid; 2, a-cortolonic acid; 3, p-cortolic acid; 4, p-cortolonic acid.
HPLC OF STEROIDAL
ACIDS
67
3
AJL 4
I 0
I
IO
20
30
40 Mmutes
50
,
,
60
70
1
80
FIG. 7. Separation of a mixture ofp-bromophenacyl esters of FHA and EHA. Conditions were as described under Fig. 6. The peaks are: 1, cx-cortisolic acid; 2, cr-cortisonic acid; 3, p-cortisolic acid; 4, p-cortisonic acid.
TABLE RETENTION
TIMES
1
OF SOME STEROID
ACIDS”
Time (min) C,, acids 1lp,17,20a-trihydroxy-3-oxo-4-pregnen-2l-oic 1lb. 17,20P-trihydroxy-3-oxo-4-pregnen-21-oic 17.20~dihydroxy-3.1 I-dioxo-4-pregnen-21-oic 17,20/3-dihydroxy-3,l I-dioxo-4-pregnen-21-oic 3a.l lp,l7,20a-tetrahydroxy-5p-pregnan-21-oic 3a,ll/3.17,20~-tetrahydroxy-5~-pregnan-2l-oic 3a.17.20~trihydroxy-11-oxo-5P-pregnan-21-oic 3a,17,20/3-trihydroxy-1 I-oxo-5/3-pregnan-21-oic
acid acid acid acid acid acid acid acid
CzOacids 1lp,l7-dihydroxy-3-oxo-4-androstene-17P-carboxylic acid 17-hydroxy-3.1 I-dioxo-4 androstene-17fi-carboxylic acid 3o.17.1 l/3-trihydroxy-SP-pregnane-17@carboxylic acid 3c~,17-dihydroxy-1 I-oxo-S/3-pregnane-17/3 carboxylic acid
18.8 28.6 20.7 38.7 24.0 48.0 30.9 51.6
29.0 32.4 40.3 60.3
n DuPont Model 830 chromatograph was used with a spectrophotometric detector at 254 nm, 8-mm flow cell; Zorbax ODS column, 4.6 x 250 mm; eluant, methanol:H,O (60:40); flow rate, 2.2 ml/min at 3000 psi: temperature, 40°C. Retention time of Dibromoacetophenone, 7.1 min.
68
FARHI
AND
MONDER
DISCUSSION
In this paper we have shown that the microprocedure of Durst et al. (3) for the formation and analysis of bromphenacyl derivatives of fatty acids can be modified to provide a sensitive and reproducible analytical technique for the measurement of the recently discovered steroidal carboxylic acids (1). Classically, acids have been derivatized by reacting them with alkyl bromides under basic conditions. Recent interest in the analysis of fatty acids and prostaglandins by high pressure liquid chromatography has resulted in the development of a number of methods which are based on the formation of phenacyl acid derivatives catalyzed by tertiary amines (S- 10). To facilitate the reaction of acid with alkyl bromide, Durst et al. (3) suggested the use of an 18-crown-6 ether as catalyst with the potassium salt of the fatty acid in an aprotic solvent. The 18-crown-6 ethers complex the potassium ion, thus enhancing the reactivity of the counter ion. Our earlier experiences with preparing derivatives with substituted triazines (11,12), oxazolines (13), and carbodiimide (14) were disappointing. We were gratified to note that the method of Durst et al. (3), though described as a qualitative procedure, could be utilized as a quantitative technique as well. The procedure we have described is simple, rapid, quantitative, sensitive, and specific. The range of the method permits it to be used for the estimation of steroid acids in human urine. Further improvements in sensitivity by the use of fluorescent chromophores will increase the range of applicability of the method still further. ACKNOWLEDGMENTS This investigation was supported AM 09006, and General Research
by grants from U. S. Public Support RR 5589.
Health
Service:
P30-1494,
REFERENCES 1. Bradlow, H. L., Zumoff, B., Monder, C., Lee, H. J., and Hellman, L. (1973) J. Clin. Endocrinol. Metabol. 37, 811-818. 2. Monder, C., and Bradlow, H. L. (1977) J. Steroid Biochem. 8, 897-908. 3. Durst, H. D., Milano, M., Kikta, E. J., Jr., Connelly, S. A., and Grushka, E. (1975) Anal. Chem. 47, 1797-1801. 4. Lewbart, M. L., and Mattox, V. R. (1963) J. Org. Chem. 28, 2001-2006. 5. Oh, S. W., and Monder, C. (1976) J. Org. Chem. 41, 2477-2480. 6. Martin, K. O., and Monder, C. (1978) J. Steroid Biochem., in press. 7. Mason, H. L., Myers, C. S., and Kendall, E. C. (1936) J. Biol. Chem. 116, 267-276. 8. Fitzpatrick, F. A. (1976) Anal. Chem. 48, 499-502. 9. Cooper, M. J., and Anders, M. W. (1974) Anal. Chem. 46, 1849-1954. 10. Borch, R. F. (1975) Anal. Chem. 47, 2437-2439. 11. Politzer, I. R., Griffin, G. W.. Dowty, B. J., and Laseter, J. L. (1973) Anal. Lett. 6, 539-546. 12. Regis Laboratory Notes, No. 16 (1974). 13. Woodward R. B., and Olofson, R. A. (1961) J. Amer. Chem. Sot. 83, 1007-1009.
BIOCHEMISTRY
58-68 (1978)
90,
Analysis of Steroidal Carboxylic Acids by High Pressure Liquid Chromatography RENE Hospital for
L.
FARHI
AND
CARL
MONDER
for Joint Diseases und Mediwl Center, Research Institute Skeletomuscular Discuses. 1919 Madison Avenue, Net.t, York. Ne,c, York 10035 Received
October 3 1, 1977
A procedure utilizing high pressure liquid chromatography has been developed for the analysis of steroids containing 17P-carboxylic acid or 20-hydroxy-21-oic acid substituents. Acids were converted to the p-bromophenacyl esters and were separated on octadecylsilyl columns with methanol:water mixtures as mobile phase. Structurally similar steroids, including the 20~ and 2Op- epimers of the 21-oic acids were well separated. The lower limits of detection ranged from 0.25 to 0.5 nM. The method is simple, rapid. quantitative, sensitive, and specific.
The cortoic acids are a group of acidic metabolites of cortisol which are excreted in the urine in amounts that represent, in normal individuals, about 6% of the products of cortisol metabolism (1). In certain diseases, the proportion of cortoic acids in the urine may change up or down relative to the neutral metabolites (2). In order to determine if these changes are of clinical significance, a method is needed for measuring the amounts of the individual cortoic acids excreted. Four quantitatively important representatives of this class of acids have been isolated (Fig. 1). All are C,, hydroxy acids, and are devoid of chromophores which absorb in the near ultraviolet. In order to measure these acids spectrophotometritally by high pressure liquid chromatography (HPLC)’ it was essential to convert them to derivatives that absorb maximally at 254 nm. Formation of the derivatives had to be specific, quantitative, and reproducible and had to yield products which were easily separated and were readily ’ Abbreviations used: HPLC, high pressure liquid chromatography; ODS, octadecylsilyl; DBAP, cY,p-dibromoacetopherone; THFEA, 3a,l lp,l7-trihydroxy-5-androstane-l7P-carboxylic acid; THEEA. 3cy.l7-dihydroxy-1 I-oxo-5/3-pregnane-17p-carboxylic acid: cu-FHA, 11/3,17.20o-trihydroxy-3-oxo-4-pregnen-2l-oic acid; P-FHA, 1lp,l7.20/3-trihydroxy-3-oxo4-pregnen-21-oic acid: cu-EHA.17.20~dihydroxy-3.1 I-dioxo-4-pregnen-21-oic acid; P-EHA. 17,20P-dihydroxy-3.1 I-dioxo-4-pregnen-21.oic acid; FEA, 1l/3.17-dihydroxy-3-oxo-4-androstene-17Pkarboxylic acid; EEA, 17-hydroxy-3,1 I-dioxo-4-androstene-17p-carboxylic acid; a-DOCHA, 20a-hydroxy-3-oxo-4-pregnen-21-oic acid: P-DOCHA, 20P-hydroxy-3. oxo-4-pregnen-21-oic acid; DOCEA, 3-oxo-4-androstene-17@carboxylic acid; AUFS, absorbancy units full scale. 0003-2697/78/0901-0058$02.00/O Copyright 0 1978 by Academic Press, Inc. All rights of reproduction in any form reserved.
58
HPLC
OF STEROIDAL
59
ACIDS
COOH
COOH
H-L-OH
HO-t-H
..,.&pOH
,-..poH H
Cortolonlc
H acid
B-Cortolonlc
acid
COOH
COOH
H-&OH
HO-&H
Ho,,poH
Ho,._‘;:l:_-“’
H Cortolsc
FIG.
H ocld
a-Cortollc
1. Representative
steroidal
21.oic
acid
acids.
detected at low concentrations. In this paper, we describe the development of an analytical technique for measuring the cortoic acids, based on the method of Durst et al. (3) for preparing derivatives of the fatty acids. METHODS
Instrumentation Analyses were performed on a DuPont liquid chromatograph, Model 830, fitted with an oven and a gradient system. Injections were made through a Rheodyne injection valve, Model 71-20. Zorbax octadecylsilyl (ODS) columns, 2.1- or 4.6mm i.d., 250 mm long, were washed with analytical grade methanol until a stable baseline was obtained at high sensitivity. The 254-nm photometer (DuPont Model 842) contained an g-mm flow cell. A Varian A-25 recorder was used. Pressure input was usually 3000 psi; it was raised to 4500 psi when separations required the use of two columns in series. The oven temperature was 40°C. The flow rate was determined by measuring the volume of effluent collected in a given time. Spectra in the ultraviolet range were obtained on a Cary Model 15 spectrophotometer. Chromatographic
Conditions
Early work was done with an injector encompassing a rubber septum. At high instrument sensitivities spurious and irreproducible peaks were seen on the chromatogram as a result of the action of the injected acetonitrile on the septum. This was eliminated by the use of the injection valve. Injection volumes for the 2.1- and 4.6-mm i.d. columns were 10 and 50 ~1, respectively. The wider column gave sharper peaks due to “infinite wall
60
FARHI
AND
MONDER
effect” with a larger flow rate. Under these conditions, decreased by a factor of two.
the sensitivity
was
Solvents Chromatography was performed with mixtures of methanol (Baker, AR) and water redistilled from glass. To prevent the slow formation of turbidity in these mixtures the water, fresh from the still, was immediately sterilized. At the time of use, it was filtered through Millipore GS (0.22 pm) and the methanol was filtered through Millipore FG (0.2 pm) membranes. Handling and filtering of the water and storage of the alcohol:water mixtures were done using sterile glassware. Methanol could not be replaced with acetonitrile or isopropanol as eluting solvent. Chemicals cw,p-Dibromoacetophenone (DBAP) was obtained from Aldrich Chemical Company. Analysis of the reagent as received by HPLC on an ODS column using gradient elution with methanol:water mixtures showed presence of two minor impurities which did not interfere with the analytical procedure. The DBAP was used without further purification. p-Bromoacetophenone (Aldrich Chemical Co.) was used as received. Dicyclohexano-l&crown-6 (crown ether) was used as recommended by Durst et al. (3). Water was removed by azeotropic distillation with benzene. The solvent was eliminated on a rotary evaporator, and the dry crown ether residue was kept in a stoppered flask. Acetonitrile was either “Distilled-in-glass” bought from Burdick and Jackson Company, or was Mallinckrodt Nanograde. Analysis of the solvent on an ODS column with methanol: water (65:35) as eluant showed the presence of a uv-absorbing impurity which was not retained, and did not interfere with the analysis. Other reagents and chemicals were analytical grade. Silica gel was E Merck, range 0.2-0.5 mm. Steroid Acids The 20-hydroxy-21-oic acids were synthesized as described by Lewbart and Mattox (4). The 20/3 epimers were synthesized from the 20j%hydroxy21-aldehydes (5) by enzymatic oxidation of the aldehyde group (6). C,, acids (etienic and etianic acids) were synthesized as described by Mason et af. (7). The acids used in this study were 3a, 1 1/3,17,20a-tetrahydroxyS/3-pregnan-21-oic acid (cy-cortolic acid); 3a-1 l/3,17,20/3-tetrahydroxy-5/?pregnan-2 1-oic acid (/I-cortolic acid); 3a- 17,20a-trihydroxy- l l-0x0-Sppregnan-21-oic acid (cY-cortolonic acid); 3a-17,20/3-trihydroxy-11-0x0-5/3pregnan-21-oic acid (p-cortolonic acid); 3a,l l/3,17-trihydroxy-5-androstane-17@carboxylic acid (THFEA); 3a, 17-dihydroxy-l l-oxo-5@pregnane17/3-carboxylic acid (THEEA); 11/3,17,20c&rihydroxy-3-oxo-4-pregnen-
HPLC OF STEROIDALACIDS
61
21-oic acid (a- FHA); 11/3,17,20fl-trihydroxy-3-oxo-4-pregnen-21-oic acid (/3-FHA); 17,20a-dihydroxy-3,11-dioxo-4-pregnen-21-oic acid (a-EHA); 17,20/3-dihydroxy-3,11-dioxo-4-pregnen-21-oic acid (/3-EHA); 11/3,17-dihydroxy-3-oxo-4-androstene-17/3-carboxylic acid (FEA); 17-hydroxy3,11-dioxo-4-androstene-17/3-carboxylic acid (EEA); 20a-hydroxy-3oxo-4-pregnen-21-oic acid (a-DOCHA); 20/3-hydroxy-3-oxo-4-pregnen21-oic acid (/3-DOCHA); 3-oxo-4-androstene-17/3-carboxylic acid (DOCEA).
Preparation of Reagents (a) Potassium methoxide, 0.5 m. Pellets of 85% pure potassium hydroxide (332 rag) are dissolved in methanol to obtain 5 ml of solution which is then filtered on Millipore FG filter. The reagent need not be standardized and is stored in a tightly capped container. This stock solution is diluted as needed. (b) Crown ether solution. About 12 mg of crown ether is dissolved in 1 ml of acetonitrile and is filtered through a Millipore FG membrane. (c) Alkylating reagent. To 100 mg of DBAP are added 1 ml of crown ether solution and 9 ml of acetonitrile. Molar ratio of DBAP to crown is about 10. Solutions of DBAP in acetonitrile slowly decompose and are kept in the freezer. (d) Standard solutions o f acids. One mg of each acid was dissolved in 0.10 ml of methanol followed by 0.90 ml of acetonitrile. The solution was filtered on a Millipore FG membrane and stored in a freezer. Derivatization Procedure Analytical preparation of derivatives was carried in standard 1-dram screw-cap vials with special caps and Teflon-covered rubber septa (Pierce Chemical Co., Catalog Nos. 13915 and 12412); 10 × 3 mm o.d. Tefloncovered magnets were used for stirring. After each use, the vials were washed with water, 95% ethanol, and acetone, and were then air dried. Steroid acid standard solution in a volume of 1.5 to 250/xl, measured with a Hamilton syringe, was placed in a vial, and acetonitrile was added to a final volume of 0.95 ml. Fifty microliters of 0.01 M potassium methoxide was added slowly with constant magnetic stirring, followed by 40 ~1 of alkylating reagent. The vials were capped and heated in a water bath at 75 to 80°C with stirring for 20 min. The contents were cooled and chromatographed directly. Quantitative analysis of the chromatographic profile was performed after cutting out and weighing the peaks corresponding to the steroid acid derivatives.
Preparative Synthesis of Derivatives Steroid acids were converted to p-bromophenacyl derivatives in milligram amounts in order to confirm structures, study spectrophotometric
62
FARHI
AND
MONDER
properties, and provide analytical standards. A total of 30 to 40 mg of acid was dissolved in 3 to 4 ml of methanol in a lo-ml screw-cap vial. A milligram of phenolphthalein in 0.1 ml of methanol was added, and the solution was titrated to a red color with 0.1 M potassium methoxide. The methanol was evaporated under nitrogen. To the residue were added 0.1 ml of methanol, 4 ml of acetonitrile, 0.75 ml of crown ether solution, and 75 to 100 mg of DBAP dissolved in 2 ml of acetonitrile. The capped vial was heated for 1 to 2 h in a water bath at 75 to 80°C with magnetic stirring. Initially, the liquid was clear, with pink patches along the wall of the vial, but very quickly a turbidity of potassium bromide appeared, while the patches turned white and disappeared. After cooling, the solution was passed through a small column of silica gel (5-6 g) to remove the crown ether. The vial and column were washed with 25 ml of dichloromethane which was added to the eluate. The derivatives were purified by chromatography on a prepacked column of silica gel (EM Catalog No. 10607) with mixtures of methanol and chloroform. Elution of the derivatives was achieved with the proportions of methanol to chloroform indicated in the parentheses: EHA (1:99); cortolonic acid (2:98); EEA (2:98); THEEA (5:95). The derivatives were crystallized from mixtures of chloroform and hexane. Anal. Calcd for-p-bromophenacyl cortolonic acid, &,H3,0,Br: C, 60.31; H, 6.45; Br, 13.84. Found C, 60.07; H, 6.83; Br, 13.48. Anal. Calcd forp-bromophenacyl EEA, C,,H,,O,Br, C, 61.88; H, 5.75; Br, 14.70. Found C, 61.19; H, 5.77; Br, 15.91. Anal. Calcd for-p-bromophenacyl THEEA, C, 61.43; H, 6.44; Br, 14.59; Found C, 61.40; H, 6.65; Br, 14.67. RESULTS
Spectral Characteristics
of the Derivatives
Figure 2 shows the absorption spectrum of a representative steroid acid derivative. In this, and in all other derivatives, the spectral maxima were at 254 nm. The contributions of the phenacyl chromophore and of the 3-oxo-4-ene moiety of the steroid in Fig. 2 are additive. The two chromophoric groups are therefore noninteracting. Where the phenacyl group of the steroid derivative is the sole contributing chromophore, as in bromophenacyl cortolonic acid, the extinction coefficient is the same as that of p-bromoacetophenone. Neutralization
Procedure
Durst et al. (3) had suggested alternate procedures for converting the carboxylic acid to its potassium salt. These included titration of the alcoholic solution of the acid to an indicator end point or stirring the acid
HPLC OF STEROIDAL
63
ACIDS
-
350
FIG. 2. Ultraviolet absorption spectra of steroid acid derivatives. (----) phenone; (- . -) 17,20/3-dihydroxy-3,1I-dioxo-4-pregnen-21-oic acid; (---) tive, calculated; (-----) P-EHA derivative, found. Solvent was methanol.
DibromoacetoP-EHA deriva-
with a three to fivefold excess of potassium bicarbonate. In our hands, these approaches proved to be inadequate. Variations, including adding the base in water, use of other solvents, sonication of the suspension, or use of potassium carbonate as base, were unsuccessful in improving reproducibility. We subsequently found that a fixed molar excess of methanolic potassium hydroxide added as described under Methods gave an acceptable yield of derivative. When steroid hydroxy acid was briefly exposed to 0.01 M potassium methoxide, there was progressive destruction of the steroid side chain and decreased ester formation which was intensified as the ratio of base to steroid increased. In order to avoid this problem, the acid was first diluted with acetonitrile, then slowly neutralized with 50 ~1 of 0.01 M potassium methoxide prior to addition of alkylating reagent. Reagents
A threefold excess of DBAP was needed for complete derivatization as reported by Durst et al. (3) and Fitzpatrick (8). The molar ratio of DBAP to crown ether was 10. Yield of product was not diminished by the presence of 10 ~1 of water or methanol in the reaction mixture.
64
FARHI
Reaction
AND
MONDER
rates
Figure 3 shows that the rate of derivative formation depended on the concentration of steroid acid. The reaction was faster in more dilute solution. The yield was independent of the concentration of steroid after 20 min. Calibration
curves
Dependence of derivative formation on steroid concentration is illustrated in Figs. 4 and 5. The upper range is limited by the amounts of reagents used. The lower range is limited by several factors: Acids with a conjugated double bond system in the A ring have a lower limit of detectability than do acids with a reduced A ring. Other factors are the methanol and DBAP concentrations. Interference by the latter at very high sensitivity results from traces of decomposition products of the reagent which emerge from columns in the general region where acid derivatives are eluted. DBAP is eluted before any of the bromophenacyl esters and therefore does not itself interfere with the emerging steroids. At high sensitivity, however, the trailing foot of the reagent may emerge from the column coincident with some of the faster moving acid esters, making quantitation of the emerging peak difficult in those cases. The CzOetianic acids can be measured down to 0.25 nM. Concentration 100
Reaction FIG. 3. procedure tration of raphy was nohwater
time (minutes)
Yield of bromphenacyl derivative of FHA as function of time. The derivatization was performed as described in the text with various reaction times. Concensteroidal acid used in the reaction: (0) 12.5 pgiml; (A) 100 &ml. Chromatogperformed on a Zorbax ODS column, 4.6 x 250 mm; oven, 40°C; solvent. metha(65:35); pressure, 3000 psi; flow rate, 1.9 ml mitt-‘.
HPLC OF STEROIDAL 4000
3000
65
ACIDS
-
-
1
5
IO
15
20
Amount
25
30
35nM
Injected
FIG. 4. Calibration curve for the p-bromophenacyl ester of 3a,l7-dihydroxy-ll-oxo-5ppregnane-17P-carboxyhc acid (THEEA). Absorption measured at 254 nm in an 8-mm cell. Other conditions were as described under Methods.
Amount
injected
FIG. 5. Calibration curve for thep-bromophenacyl pregnan-21-oic acid (cortolonic acid).
ester of 3a, 17,20-trihydroxy-I
1-oxo-Sp-
66
FARHI AND MONDER
dependence was linear in the entire range of steroid used (Fig. 4). The concentration curves of the 20-hydroxy acids were linear as well over most of their range (Fig. 5). There was a slight curvature at the lowest steroid levels, because of some acid decomposition by excess base in this region. The lower practical limit of measurement was 0.5 nM. The relative standard deviation of measurement varied from about 2 to 3% in the upper concentration range to about 10% in the lower range with all steroid acids. Chromatographic Profiles
Examples of separations are shown in Figs. 6 and 7. It is of interest that steroids epimeric at C,, were well separated. The pairs a-FHA-a-EHA and p-cortolic acid-P-cortolonic acid were incompletely resolved, but use of two columns in series improved the resolution. For all the 20-hydroxy acids studied, the 20a isomer eluted before the 20/3 isomer. Table 1 shows the retention times of various bromphenacyl acids. Complete separations were achieved in most cases. In order to completely elute acids of widely different polarity from the columns, the solvent schedule was adjusted in a stepwise manner. DOCHA and DOCEA required larger methanol concentrations (methanol:water = 75:25) to get reasonable retention times and sharp peaks.
FIG. 6. Separation of a mixture of p-bromophenacyl esters of cortolic and cortolonic acids. Two Zorbax ODS columns, 4.6 x 250 mm, were placed in series. Oven temperature, WC; solvent, methanoLwater; 65:35; pressure, 4500 psi; flow rate, 1.5 ml min-‘; uv detector at 254 nm; 8-mm cell, 4 x 10m2AUFS. The peaks are: 1, a-cortolic acid; 2, a-cortolonic acid; 3, p-cortolic acid; 4, p-cortolonic acid.
HPLC OF STEROIDAL
ACIDS
67
3
AJL 4
I 0
I
IO
20
30
40 Mmutes
50
,
,
60
70
1
80
FIG. 7. Separation of a mixture ofp-bromophenacyl esters of FHA and EHA. Conditions were as described under Fig. 6. The peaks are: 1, cx-cortisolic acid; 2, cr-cortisonic acid; 3, p-cortisolic acid; 4, p-cortisonic acid.
TABLE RETENTION
TIMES
1
OF SOME STEROID
ACIDS”
Time (min) C,, acids 1lp,17,20a-trihydroxy-3-oxo-4-pregnen-2l-oic 1lb. 17,20P-trihydroxy-3-oxo-4-pregnen-21-oic 17.20~dihydroxy-3.1 I-dioxo-4-pregnen-21-oic 17,20/3-dihydroxy-3,l I-dioxo-4-pregnen-21-oic 3a.l lp,l7,20a-tetrahydroxy-5p-pregnan-21-oic 3a,ll/3.17,20~-tetrahydroxy-5~-pregnan-2l-oic 3a.17.20~trihydroxy-11-oxo-5P-pregnan-21-oic 3a,17,20/3-trihydroxy-1 I-oxo-5/3-pregnan-21-oic
acid acid acid acid acid acid acid acid
CzOacids 1lp,l7-dihydroxy-3-oxo-4-androstene-17P-carboxylic acid 17-hydroxy-3.1 I-dioxo-4 androstene-17fi-carboxylic acid 3o.17.1 l/3-trihydroxy-SP-pregnane-17@carboxylic acid 3c~,17-dihydroxy-1 I-oxo-S/3-pregnane-17/3 carboxylic acid
18.8 28.6 20.7 38.7 24.0 48.0 30.9 51.6
29.0 32.4 40.3 60.3
n DuPont Model 830 chromatograph was used with a spectrophotometric detector at 254 nm, 8-mm flow cell; Zorbax ODS column, 4.6 x 250 mm; eluant, methanol:H,O (60:40); flow rate, 2.2 ml/min at 3000 psi: temperature, 40°C. Retention time of Dibromoacetophenone, 7.1 min.
68
FARHI
AND
MONDER
DISCUSSION
In this paper we have shown that the microprocedure of Durst et al. (3) for the formation and analysis of bromphenacyl derivatives of fatty acids can be modified to provide a sensitive and reproducible analytical technique for the measurement of the recently discovered steroidal carboxylic acids (1). Classically, acids have been derivatized by reacting them with alkyl bromides under basic conditions. Recent interest in the analysis of fatty acids and prostaglandins by high pressure liquid chromatography has resulted in the development of a number of methods which are based on the formation of phenacyl acid derivatives catalyzed by tertiary amines (S- 10). To facilitate the reaction of acid with alkyl bromide, Durst et al. (3) suggested the use of an 18-crown-6 ether as catalyst with the potassium salt of the fatty acid in an aprotic solvent. The 18-crown-6 ethers complex the potassium ion, thus enhancing the reactivity of the counter ion. Our earlier experiences with preparing derivatives with substituted triazines (11,12), oxazolines (13), and carbodiimide (14) were disappointing. We were gratified to note that the method of Durst et al. (3), though described as a qualitative procedure, could be utilized as a quantitative technique as well. The procedure we have described is simple, rapid, quantitative, sensitive, and specific. The range of the method permits it to be used for the estimation of steroid acids in human urine. Further improvements in sensitivity by the use of fluorescent chromophores will increase the range of applicability of the method still further. ACKNOWLEDGMENTS This investigation was supported AM 09006, and General Research
by grants from U. S. Public Support RR 5589.
Health
Service:
P30-1494,
REFERENCES 1. Bradlow, H. L., Zumoff, B., Monder, C., Lee, H. J., and Hellman, L. (1973) J. Clin. Endocrinol. Metabol. 37, 811-818. 2. Monder, C., and Bradlow, H. L. (1977) J. Steroid Biochem. 8, 897-908. 3. Durst, H. D., Milano, M., Kikta, E. J., Jr., Connelly, S. A., and Grushka, E. (1975) Anal. Chem. 47, 1797-1801. 4. Lewbart, M. L., and Mattox, V. R. (1963) J. Org. Chem. 28, 2001-2006. 5. Oh, S. W., and Monder, C. (1976) J. Org. Chem. 41, 2477-2480. 6. Martin, K. O., and Monder, C. (1978) J. Steroid Biochem., in press. 7. Mason, H. L., Myers, C. S., and Kendall, E. C. (1936) J. Biol. Chem. 116, 267-276. 8. Fitzpatrick, F. A. (1976) Anal. Chem. 48, 499-502. 9. Cooper, M. J., and Anders, M. W. (1974) Anal. Chem. 46, 1849-1954. 10. Borch, R. F. (1975) Anal. Chem. 47, 2437-2439. 11. Politzer, I. R., Griffin, G. W.. Dowty, B. J., and Laseter, J. L. (1973) Anal. Lett. 6, 539-546. 12. Regis Laboratory Notes, No. 16 (1974). 13. Woodward R. B., and Olofson, R. A. (1961) J. Amer. Chem. Sot. 83, 1007-1009.
