1866
Anal. Chem. 1984, 56,1866-1870
Determination of E Prostaglandins by Automated Heteromodal Column Switching High-Performance Liquid Chromatography with Fluorescence Detection Jeffrey
W.Cox* and Robert H.Pullen
The Upjohn Company, Pharmaceutical Research a n d Development, Drug Metabolism Research, Unit 7256-126-2, Kalamazoo, Michigan 49001
An analytical method was developed for the simultaneous quantitation of plcogrbm amounts of (15R )- and (15S)-15methyi-PGE,. The sample and l8,l6-dimethyi-PGE2(the internal standard) were derlvatlzed with panacyi bromide to form panacyi ester derivatives, which were then analyzed directly with an automated, heteromodai, column switching HPLC system. The HPLC system employed a combinationof cyano and silica analytical columns and fluorescence detection. The overall analysis conditlons were mild and did not promote eplmerizatlon or degradatlon of the ( 1 5 R ) OF ( S S ) epimers. Related prostaglandins did not interfere with the analysis. Assay linearity was demonstrated over a range of 1700 to 30 pg for both (15R) and (15s) epimers ( f 1 0.998). There was no significant interday difference in assay results for either epimer over a range of 1000 to 75 pg ( p 5 0.05), and there was no signlflcant assay blas over the same range ( p > 0.05). Pooled estimates of assay preclsion indicated an assay relative standard deviation of 4 % when derivatizing 1000 or 330 pg of either eplmer and 17 % when derivatizing 75 pg. The method quantitation limit (signai-to-noise ratio 10:l) for both ephners was 30 pg derlvatlzed and’l0 pg inJected on-column. The general utility of the method for E prostaglandins was demonstrated for the analysis of PGE,, PGE,, 8-iso-PGE,, and 16,W-dimeth$i-PGE,.
Analytical methods for prostaglandins in physiological fluids must be sensitive and specific. Current methodology for the analysis of picogram amounts of prostaglandins is limited to radioimmunoassay (1) or gas chromatography/mass spectrometry with negative ion chemical ionization (GC/NICIMS) (2-4). Other methods have been shown to have picogram range instrumental sensitivity for the analysis of prostaglandin derivatives, but an inability to reproducibly derivatize subnanogram quantities of prostaglandins precludes their application at this level. Among these methods are capillary gas chromatography with electron capture detection (5-7) and, more recently, high-performance liquid chromatography with fluorescence detection (8-10). Both Tsuchiya et al. (8) and Hatsumi et al. (9) reported low picogram detection limits for the HPLC/fluorescence assay of 4-bromomethyl-7-acetoxycoumarin and 9-anthryldiazomethane prostaglandin derivatives, respectively, but both indicated that the derivatization efficiency was poorer in the low nanogram range and neither reported subnanogram derivatizations. Similarly, Watkins and Peterson (IO)reported a low picogram detection limit for the HPLC/fluorescence assay of prostaglandin panacyl bromide derivatives but did not report the derivatization of picogram quantities of prostaglandins. The present work was initiated to develop an HPLC method for the analysis of picogram amounts of (15R)- and (15s)15-methyl-PGEz for eventual application to the analysis of human plasma samples. (15R)-15-Methyl-PGEz is a pro-drug under evaluation for the treatment of acute upper gastrointestinal hemorrhage and for gastrointestinal cytoprotection.
It is converted in gastric fluid (11,12) to its more biologically active (155‘) epimer (13),and both the (15R) and (15s) epimers are found in human plasma at low picogram per milliliter concentrations following oral dosing (14). An assay method is described for the simultaneous quantitation of picogram levels of both epimers using panacyl bromide to prepare fluorescent derivatives and a normal phase, column switching HPLC technique for the direct analysis of derivatization reaction mixtures. The general utility of this approach is demonstrated for the analysis of several E-type prostaglandins. EXPERIMENTAL SECTION Materials. Arbaprostil ((15R)-15-methylprostaglandinEz), (15S)-15-methylprostaglandinEz, 16,16-dimethylprostaglandin Ez,(15R)-15-methylprostaglandinEz-llp-3H(specific activity of 3.48 mCi/mg), and panacyl bromide (p-(9-anthroyloxy)phenacyl bromide) were supplied by the Pharmaceutical Research & Development Laboratories of The Upjohn Company (Kalamazoo, MI). Hexane, acetonitrile, tetrahydrofuran, methylene chloride, and 2-propanol were W or HPLC grade and were purchased from Burdick & Jackson Laboratories (Muskegon, MI). Liquid scintillation cocktail was Aqueous Counting Scintillant (Amersham Corp., Arlington Heighta, IL). Nfl-Diisopropylethylamine, 98%, was obtairied from Aldrich Chemical Co., Inc. (Milwaukee, WI), and purified by refluxing for 2 h over barium oxide and distilling directly through a 15-cm air-jacketed Vigreaux column. Highpurity grade nitrogen was purchased from Union Carbide Corp., New York, NY. High-Performance Liquid Chromotography. Apparatus. The HPLC apparatus consisted of two isocratic systems (1and 2) linked serially via a 2.2-mL sampling loop which permitted the direct injection of system 1 eluant into system 2 (Figure 1). Samples were injected with the Upjohn pP Autosampler (The Upjohn Co.) equipped with a venting injection needle. Pumps 1and 2 were Beckman 112 solvent delivery modules (Beckman Instruments, Inc., Berkeley, CA), and mobile phase routing was performed with Valco 10-port air-actuated (60psi) switching valves (Valco Instruments, Houston, TX). The guard column (5cm X 4.6 mm i.d.) was dry packed with Whatman C0:PELL PAC fcyano-aminogroups chemically bonded to 30-38 pm glass beads) (Whatman, Inc., Clifton, NJ). Analytical column 1was a Zorbax CN 15 cm x 4.6 mm i.d. cyanopropylsilane bonded phase column and analytical column 2 was a Zorbax SIL 25 cm X 4.6mm i.d. silica column (both packing materials are spherical particles, 6 pm diameter) (DuPont Co., Wilmington, DE). Detector 1 was an LDC UV (254nm) Monitor 111, Model 1203A (Laboratory Data Control, Riviera Beach, FL). Detector 2 was a Perkin-Elmer 650s fluorescence detector (Perkin-Elmer Corp., Norwalk, CT) with the following settings: excitation wavelength, 375 nm; emission wavelength, 470 nm; slit width (excitation and emission), 20 nm; and response time, 6 s. All connections between columns, detectors, and switching valves were made with in. 0.d. X 0.010 in. i.d. stainless steel tubing and low dead volume connectors. Switching Value Actuation. Switching valve positions were controlled by a Beckman 421 system controller, which has built-in external flags for the control of external devices. The system controller was interfaced to the switching valves by a solenoid valve/solid-state relay device (The Upjohn Co.). The file in the system controller was initiated by a pressure impulse from the autoinjector switchingvalve using a Pressure Switch relay (Bulletin 836,Allen-Bradley, Milwaukee, WI) to the controller’s remote
0003-2700/84/0356-186~$01.50/0 0 1984 American Chemlcal Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
-I m- - __I ANALYTICAL
Inn 1
-
RECORDER AN0 DATA SYSTEM
I--
I
@
@
RESEA.
RESEA.
Figure 1. Schematic representatlon of the HPLC apparatus. POSITION I
Waste Waste
POSITION II
POSITION 111
Waste Guard Column
Waste Guard
Detector 1
Detector 1
Detector 1
Waste
Waste
Waste
Analytical Column 2 Detector 2
Analytical Column 2 Detector 2
Analytical Column 2 Detector 2
Waste
Waste
Waste
Guard Column
Waste
Flgure 2. Schematic representation of the switching valve positions
used to complete one analysis cycle. Position I Injects the sample with
injects the guard column retentate onto system 1 with system 1 eluant routed through the sampling loop to waste; position 111 Injects the sample loop contents onto system 2 and refills the injection loop for the next cycle. Program: position I, 0-0.5 min; posltion 11, 0.5-7 min; position 111, 7-42 min.
the guard column vented to waste: position I I
input. Three switching valve configurations were used in the analysis (Figure 2). The guard column venting time was determined and held constant for subsequent studies (see below). The sampling valve switching time was determined daily by injecting a 1-ng (15R)-l&methyl-PGE2panacyl ester standard. The retention time of the derivative on system 1 was determined with the UV detector, and the switching time (typically7.0 min) was established by adding 0.15 min to the time at which the peak returned to base line. Data Acquisition. Collection and analysis of the fluorescence detector output were performed with the UPACS I1 automated chromatography system 3 (The Upjohn Co.) on a Harris 500 computer. Ouput from both the fluorescence and UV detectors was recorded on a strip chart recorder. Mobile Phase Preparation and HPLC Conditions. Mobile phase 1was hexane/methylene chloride/2-propanol(70/30/1, v/v). Mobile phase 2 was hexane/methylene chloride/tetrahydrofuran/2-propanol (60/20/20/1, v/v). All solvents were filtered through 0.2-pm Nylon 66 membrane filters (Rainin Instrument Co., Wobum, MA) prior to mixing. Both mobile phases were degassed and held in the closed system reservoirs described by Senftleber et al. (15).Mobile phase flow rates were 2.0 mL/min and 1.0 mL/min for systems 1and 2, respectively, and both pumps operated continuously throughout the analysis cycle. The guard column and analytical columns were at ambient temperature. The injection volume was 0.1 mL. Standard and Reagent Preparation. Accurately weighed amounts of (15R)-15-methyl-PGE2and ita (15fl-lB-methyl epimer
1867
were dissolved in acetonitrile to produce stock concentrations for each of 900 pg/mL. These solutions were diluted serially to produce working solutions. All prostaglandin solutions were prepared in polypropylene 12 x 75 mm test tubes fitted with polyethylene stopper plugs (Fisher Scientific, Pittsburgh, PA), and all prostaglandin solution transfers were performed with polypropylene-tipped pipettors (Gilson Pipetman (Rainin Scientific, Woburn, MA) or Varimetric Micropipettors (Labmdmtries, Berkeley, CA). The solutions were stored at -20 "C when not in use, and working solutions were prepared fresh daily. The panacyl bromide stock solution (0.5 mg/mL) was prepared in tetrahydrofuran in a glass, screw top culture tube equipped with a Teflon-lined cap. The working derivatizing reagent was prepared by diluting 1.0 mL of the stock solution with 4.0 mL of acetonitrile. Both the stock and working solutions of panacyl bromide were stable for at least 3 weeks when stored at -20 OC. The tetrahydrofuran and acetonitrile solvents for all solutions were dried over Davison molecular sieves, 10-16 mesh, 40-A pore size (Fisher Scientific, Pittsburgh, PA). Derivatization. Reaction conditions were as described by Watkins and Peterson (10).The reaction was performed in polypropylene micro vials (300 pL volume) using aluminum crimp-on seals faced with Teflon film (The Anspec Co., Inc., Ann Arbor, MI). Each reaction vial contained 250 pL of working panacyl bromide solution (25 pg in tetrahydrofuran/acetonitrile,1/4, v/v) and 10 pL of working internal standard solution (640 pg of 16,16-dimethyl-PGE2). Appropriate amounts of the (15R)- and (15S)-methyl epimer working solutions were added to the vials to achieve the desired standard concentrations. The reaction was initiated by adding 3 pL of N,N-diisopropylethylamineto each vial. The vials were capped and allowed to incubate in a 40 OC water bath for 1h. After incubation, the samples were evaporated in the same water bath under a stream of nitrogen. The residue was reconstituted by adding 300 WLof mobile phase 1,capping, and sonicating for 5 min in an immersion bath (Branson Cleaning Equipment, Shelton, CT). With this procedure, 30 samples could be processed in 3 h. Quantitation Method. Data collected by the Harris 500 computer were analyzed using unweighted linear regression best-fit of peak height ratio [(15R)- or (15S)-lbmethyl epimer/IS] vs. quantitation was amount (pg) derivatized. Radioactivity (3H) performed on a Mark I11 Liquid Scintillation System, Model 6880 (G. D. Searle and Co., Des Plaines, IL) using automatic external standardization to correct for quenching.
RESULTS AND DISCUSSION Chromatography. In order to successfully derivatize subnanogram amounts of 15-methyl-PGE, epimers, it was necessary to use a greater than 104-fold excess of panacyl bromide reagent. If injected directly for HPLC analysis on a normal-phase system, the reagent eluted at the solvent front and tailed into the chromatogram, reducing sensitivity and causing double peaking of the PG derivative. Guard column venting of the solvent front did not totally eliminate this problem. Watkins and Peterson recommended a preanalysis cleanup procedure for PG derivatives employing disposable silica columns (IO). Since it was anticipated that a highly specific and flexible chromatography system would be required for the present application, on-line column switching techniques in combination with guard column venting were considered as an alternative to off-line sample cleanup. On-line heteromodal column switching techniques have been shown to improve chromatogram peak capacity (16) and to be helpful in minimizing prechromatography sample cleanup requirements ( 1 7 , I B ) . The main restriction on the application of the technique is that the mobile phases for the primary and secondary chromatographic modes must be compatible (19);i.e., the two mobile phases must be miscible and the collected solvent from column 1must not be a strong solvent on column 2, especially if large volumes are to be injected. Column selection experiments with the panacyl esters of 15-methyl-PGE2indicated that a normal phase combination
1888
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
of a cyano primary column and a silica secondary column could meet the restriction on mobile phase compatibility and that, furthermore, the two columns had different separation characteristics for the PG derivatives. The panacyl esters of (15R)- and (15S)-15-methyl-PGE2and 16,16-dimethyl-PGEz, the internal standard, coeluted on the cyano column but the three were resolvable on the silica column. Both the cyano and silica columns were equally effective in separating the 15-methyl-PGEz derivatives from panacyl bromide, which eluted earlier, and the panacyl ester of naturally occurring PGE2, which eluted later. Thus, a 15-methyl- and 16,16-dimethyl-PGE2 mixture could be initially purified on the cyano column and transferred to the silica column as a single peak for quantitative analysis. Although the transfer volume was large (2.2 mL), its solvent strength was weak enough that it did not contribute to band spreading on the silica column. The mobile phases for the cyano (system 1) and silica (system 2) columns were selected to provide the appropriate separation characteristics as well as to optimize the fluorescence intensity of the panacyl group. The fluorescence intenstiy was increased by decreasing solvent polarity in the order hexane > tetrahydrofuran (THF) > methylene chloride > 2-propanol > methanol. THF was a more powerful solvent for the elution of the panacyl esters on the silica column than predicted from solvent strength calculations (20) based on hexane, methylene chloride, acetonitrile, and methanol data. The difference was sufficiently large that a hexane-based mobile phase containing THF could be substituted for the methylene chloride/methanol mobile phase used by Watkins and Peterson (10) for silica column chromatograpy, thereby improving the fluorescence detector sensitivity. THF was not included in the cyano column mobile phase in order to maximize differences in the separation characteristics of the two columns. Guard column venting was used to reduce the amount of panacyl bromide applied to the cyano column. Several guard column packing materials were tested for this application, including porous-type particles such as Spheri-5 (Brownlee Labs, Inc., Santa Clara, CA) and Zorbax BP-CN (Du Pont Co., Wilmington, DE), but the best compatibility with the Zorbax CN analytical column was achieved with a C0:PELL PAC cyano amino pellicular material (Whatman, Inc., Clifton, NJ). The venting time was established by repeating the injection of a concentrated ( 15R)-15-methyl-PGEzderivative solution standard and gradually increasing the vent time until a significant loss in prostaglandin peak height was observed. A vent time of 0.5 min (2 mL/min flow rate) was sufficient to remove most of the panacyl bromide without loss of 15methyl-PGE,. Transfer of the 15-methyl-PGEz derivative peak from system 1to system 2 was accomplished with a 2.2-mL sampling loop. The peak width of coeluting panacyl esters of (15R)- and (15s)-15-methyl-PGEzand 16,16-dimethyl-PGEz on the cyano column was 1.4 mL (base-line-to-base-line).The switching time for system 2 injection was set on a daily basis by injecting a (15R)-15-methyl-PGEz derivative solution standard and measuring the retention time on system 1. The switching time was adjusted so that the peak was centered in the sampling loop at the time of injection in order to permit maximum variability in the peak retention time without loss of transfer efficiency. With a typical retention time of 6.5 min (13.0 mL elution volume), the oversized sample loop permitted a f 3 % variation in retention time. Representative chromatograms for mixtures of the panacyl esters of (15R)- and (15s)-15-methyl-PGEz and 16,16-dimethyl-PGE, are shown in Figure 3. The silica column exhibited approximately 14000 theoretical plates for each component and had peak asymmetry factors of C1.2. The peaks
I
0
I
I
10
I
I
20
,
,
30
,
,
40
Flgure 3. Representative chromatograms (detector 2) for the analysis of 15-methyCPGE2epimers: peak identification (1) (15R)-15-methyC PGE, panacyl ester, (2) 16,1Mimethyl-PGE2panacyl ester (the internal standard),and (3) (15S)-15-methyl-PGE2panacyl ester; (a) complete chromatogram for the analysis of 600 pg of each 15-methyl-PGE2 epimer (200 pg equivalents of each injected on-column): (b) expanded chromatogram for the analysis of 990 pg of each 15-methyCPGE2 epimer (330 pg equivalents injected on-column);(c) as in (b) for the analysis of 75 pg of each epimer (25 pg equivalents injected oncolumn); (d) as in (b) for the analysis of an internal standard blank. Chromatogram a is taken from the strip chart recorder. Chromatograms b-d were computer smoothed using the Savitsky-Golay leastsquares procedure (25).
were base line resolved over the range studied. Retention times were 32.83 f 0.05, 34.63 i 0.05, and 36.57 f 0.06 min (within run mean f standard deviation, N = 15) for (15R)15-methyl-, 16,16-dimethyl-, and (15S)-15-methyl-PGE2,respectively. Recovery from the chromatographic system was studied using derivatized ( 15R)-15-methyl-PGEz-11~-3H. Approximately 84% of the injected radioactivity for a 380-pg standard was recovered in the region of the chromatogram that would have been injected onto the silica column. Of this radioactivity, 100% was recovered in the (15R)-15-methyl-PGEzpeak on the silica column, indicating minimal R to S inversion or chemical decomposition (11, 21) during the analysis. The overall chromatographic efficiency of 84% is a minimum value,
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
since it assumes that the derivatization efficiency and radiochemical purity are 100%. Derivatization. Radioactive (15R)-15-methyl-PGEz was used to monitor the reaction progress and to determine if sample loss occurred during solution transfers. The results of these studies indicated that all sample manipulations in glassware should be avoided because of adsorptive losses. At subnanogram levels of 15-methyl-PGEz, glass reaction vials and capillary pipets siliconized by either vapor phase (22) or solution treatments were associated with substantial losses of radioactivity. These losses were reduced by (a) using POlypropylene vials and pipets and (b) avoiding unnecessary sample handling by using the same vial for both derivatization and HPLC injection. With these precautions, 75% of the initial radioactivity was available for HPLC injection when a 1-ng sample of (15R)-15-methyl-PGE2 was derivatized. Linearity studies indicated that this loss was constant proportion of the sample mass over a 30 pg to 1.7 ng range. Samples were derivatized as described previously (IO). When derivatizing 3 ng of 15-methyl-PGE, in the presence of 25 hg of panacyl bromide at 40 “C, the reaction was 90% complete after 15 min, and samples were routinely incubated for 60 min with no loss in the peak height of the derivative. The extent of reaction under these conditions was independent of panacyl bromide concentration between 10 and 500 bg/0.25 mL or diisopropylethylamine concentration between 3 and 10 pLI0.25 mL. Intraday consistency in derivatization efficiency was evaluated by derivatizing 16 samples of 16,16-dimethyl-PGEz (640 pg each). The relative standard deviation in the peak height of the resulting silica column peak was 5%. This was similar to the variance expected from the autosampler alone (2.4% RSD), and indicated that the derivatization procedure contributed minimally to assay variability when derivatizing 640 pg of analyte. Linearity. The linearity of fluorescence detector response vs. the amount of 15-methyl-PGE, epimer derivatized was evaluated by analyzing a series of individually derivatized samples. The experiment was repeated on three separate days to evaluate system stability. Exactly one-third of the derivatized sample was injected on-column. Plots of peak height ratio vs. picograms of 1Bmethyl-PGEzderivatized had linear correlation coefficients of 0.998 or better over a range of roughly 1700 to 30 pg. All curves were analyzed with a best fit linear regression model. Four of the six curves showed significantly positive y intercepts although no interferences were detected in reaction blanks. Since samples were individually derivatized for this experiment, the results indicate that the derivatization and sample handling techniques have a constant efficiency that is independent of the amount derivatized over the range 1700 to 30 pg of either (15R)- or (15s)-15-methyl-PGEz. Precision and Accuracy. The intraday and interday precision and accuracy of the assay were evaluated for both (15R)- and (15S)-15-methyl-PGE2 by repeating triplicate assays for each compound at three different levels (high, intermediate, and low) on three successive days. There was no significant interday difference in assay results (p > 0.05); so the data at each level were pooled. The pooled estimate of assay relative standard deviation at the high and intermediate levels (1000 and 330 pg of each epimer derivatized, 330 and 110 pg equivalents injected on-column) was 4% for both epimers. The pooled assay relative standard deviation at the low level (75 pg derivatized, 25 pg equivalents injected on-column) was 16-17% for both epimers. There was no significant assay bias a t any of the three levels (p > 0.05). Sensitivity. The overall method quantitation limit was defined as the amount of analyte which, when derivatized and
1889
Table I. Prostaglandin Panacyl Ester Derivative Elution Volumes
analyte (15R)-15-methyl-PGEz (15S)-15-methyl-PGEz 16,16-dimethyl-PGEz PGEz PGEl 8-iso-PGEz 8-iso-[(15R)-15-methyl-PGE2] PGDz
elution vel,"'*, mL system 1 system 2 (CN) (silica) 13.0 13.0 13.0 21.0 28.0 20.0
14.5 11.0
25.8 29.6 27.6 45.9 52.1 28.1 17.5 16.9
’Elution volume is reckoned from time of injection into either system 1 or 2. bPGA2, PGBZ, TxB2, 6-keto-PGF1,, and 13,14-dihydro-15-keto-PGEz were not detected after derivatization and injection under these chromatographic conditions. analyzed, produced a peak with a signal-to-noise ratio of 101. This corresponded to 33 pg of either (15R)- or (15S)-15methyl-PGEz. The absence of interfering peaks in the region of interest and the constancy of the derivatization efficiency indicated that the sensitivity could be improved 3-fold to match the chromatography system quantitation limit of 11 pg by injecting 100% of the sample. With the current autoinjector, a maximum of 33% of the sample could be injected with acceptable reproducibility. The chromatography system quantitation limit of 11 pg injected on-column compares favorably with the sensitivity of other HPLC/fluorescence assays for PGEz (8-10,23,24). Watkins and Peterson reported a 60-pg on-column detection limit (SIN ratio unspecified) for the analysis of prostaglandin panacyl bromide derivatives (IO); Tsuchiya et al. reported a 3-pg on-column detection limit (SIN ratio unspecified) for bromomethylacetoxycoumarin derivatives (8);and Hatsumi et al. reported a 30-pg on-column detection limit (SIN ratio unspecified) for the analysis of anthryldiazomethane derivatives (9). Similar sensitivity for PGEz was also achieved with capillary GC/EC (6, 7) and GC/NICIMS ( 4 ) , which have reported detection limits of 30-50 pg and 25 pg injected oncolumn, respectively (SIN ratio of 1O:l). As mentioned previously, these investigators were either unable to perform (69)or did not report (4,IO)derivatizations of picogram amounts of E-type prostaglandins and determined picogram instrumental sensitivity by diluting stock derivative solutions. In contrast, the sensitivity of the present method is similar whether expressed as amount derivatized (33 pg) or amount injected on-column (11pg equivalents), and the analytical precision in the subnanogram range is good or acceptable (4% RSD when derivatizing 300 pg, 17% RSD when derivatizing 75 pg). Specificity. Related E-type prostaglandins were derivatized and analyzed in order to determine if they chromatographically interfered with (15R)- or (15s)-15-methyl-PGEz or 16,lG-dimethyl-PGEz. As shown in Table I , PGE,, PGE1, 8-iso-PGEz,and 8-iso-(15R)-15-methy1-PGEz eluted later than lBmethyl-PGE, on system 1and PGDz eluted earlier. None of these peaks was transferred to the silica column unless the peak trapping window was changed from its normal position. PGA2, PGBz, TxB,, 6-keto-PGF1,, and 13,14-dihydro-15keto-PGEz were not detected on system 1after derivatization and injection under these chromatographic conditions. Application to Other Prostaglandins. By use of the same derivatization and chromatographic conditions optimized for the analysis of 15-methyl-PGEz, the method was used to quantify other E-type prostaglandins listed in Table I simply by adjusting the peak trapping window to the appropriate system 1retention time. The system 2 retention times given
1870
Anal. Chem. 1984, 56, 1870-1876
in Table I for these compounds were obtained in this manner. Linearity of the system 2 peak height with amount derivatized was demonstrated for PGE1, PGE2, 8-iso-PGE,, and 16,16dimethyl-PGE, over a range of 150-900 pg. The sensitivity of the method for these prostaglandins was comparable to that observed for the 15-methyl-PGE, epimers. Theoretically, this analysis approach is applicable, after appropriate modification of the mobile phase compositions, to the analysis of other prostaglandins and eicosanoids whose panacyl bromide derivatives elute later (and are generally more polar) than the 15-methyl-PGE, derivatives on HPLC system 1. The analysis of derivatives eluting before 15-methyl-PGE, is complicated by interference from the panacyl bromide derivatizing reagent. Although not designed for prostaglandin profiling applications, this heteromodal column switching technique can be used to analyze a sample for more than one component, provided that the analytes can be forced to elute as a narrow band on HPLC system 1without compromising the sensitivity and specificity of HPLC system 2. This will normally require that the prostaglandin analytes have closely related molecular structures, as was the case for the analysis of (15R)- and (15S)-15-methyl-PGEz and 16,16-dimethylPGE2.
ACKNOWLEDGMENT The authors thank F. A. Fitzpatrick and W. J. Adams for their helpful comments, Dave Gleason for construction of the solid-state and pressure impulse relay devices for switching valve actuation, and Linda Missias for assistance in the preparation of this manuscript. Registry No. PGE1, 745-65-3; PGE2, 363-24-6; (15R)-15methyl-PGE2, 55028-70-1; (15S)-15-methyl-PGE2,35700-27-7; 16,16-dimethyl-PGE2,39746-25-3;8-iso-PGE2,27415-25-4; 8-iso[ (15R)-15-methyl-PGE2],91200-56-5.
LITERATURE CITED (1) Granstrom, E.; Klndahl, H. Adv. Prostaglandin Thromboxane Res. 1978, 5, 119-210, 1 Fischer, C.; Frollch, J. C. Adv. LlpM Res. 1982, 19, 185-202. I Blalr, I. A. Br. M e d . Bull. 1983, 39, 223-226. 1 Waddell, K. A.; Blalr, I . A.; Wellby, J. Blomed. M s s Spectrom. 1983, IO. 03-80. -Fitzpatrick, F. A. Adv. Prostaglandin Thromboxane Res. 1978, 5 , 95-116. FitzpaGick, F. A.; Wynalda, M. A.; Kalser, D. G. Anal. Chem. 1977, 49, 1032-1035. Fltzpatrlck, F. A.; Strlngfellow, D. A.; Maclouf, J.; Rigaud, M. J . Chromatogr. 1979, 177,51-60. Tsuchiya, H.; Hayashi, T.; Naruse, H.; Takagi, M. J . Chromatogr. 1982. 237. 247-254. Hatsuml, M.; Klmata, S. I.; Hlrosawa, K. J . Chromatogr. 1982, 253, 27 1-275. Watkins. W. D.; Peterson, M. B. Anal. Blochem. 1982, 725,30-40. Merrltt, M. V.; Bronson, G. E. J . Am. Chem. SOC. 1978, 100, 1891-1895. Robert, A.; Yankee, E. W. Proc. SOC. Exp. Biol. Med. 1975, 148, 1155-1158. Robert, A.; Magerleln, 8. J. A&. Biosci. 1973, 9 ,247-253. Wickrema Sinha, A. J.; Shaw, S. R.; Thornburgh, B. A. Proceedings of the 33rd National Meeting of the Academy of Pharmaceutlcal Sciences, San Diego, CA, Nov 14-18, 1982; PTOX p-27. Senftleber, F.; Bowling, D.; Stahr, M. S. Anal. &em. 1983, 55, 810-81 2. Freeman, D. H. Anal. Chem. 1981, 53,2-5. Emi, F.; Keller, H. P.; Morln, C.; Schmitt, M. J . Chromatogr. 1981, 204,65-76. Apffel, J. A.; Alfredson, T. V.; Majors, R. E. J . Chromatogr. 1981, 206,43-57. Major, R. E. J . Chromatogr. Sci. 1980, 18,571-579. Snyder, L. R. J . Chromatogr. Scl. 1978, 16,223-234. Stehle, R. G.; Oesterllng, T. 0. J . Pharm. Sci. 1977, 66, 1590-1595. Fenlmore, D. C.; Davis, C. M.; Whitford, J. H.; Harrington, C. A. Anal. Chem. 1976, 48,2289-2290. Turk, J.; Weiss, S. J.; Davis, J. E.;Needleman, P.Prostaglandins 1978, 76, 291-309. Yamada, K.; Onodera, M.; Aizawa, Y. J . Pharmacol. Methods 1983, 9 , 93-100. Savitsky, A.; Golay, J. M. Anal. Chem. 1964, 3 6 , 1627-1639.
~.
RECEIVED for review February 3, 1984. Accepted April 26, 1984.
Direct Comparison of Secondary Ion and Laser Desorption Mass Spectrometry on Bioorganic Molecules in a Moving Belt Liquid Chromatography/Mass Spectrometry System T. P. Fan,E. D. Hardin, and M. L. Vestal* Department of Chemistry, University of Houston, Houston, Texas 77004 The two sofl lonlzatlon/desorptlon technlques, SIMS and LDMS, have been dlrectly compared in our LC/MS system using nonvolatlle blomolecuies as test samples. An ion gun has been Installed in our LC/LDMS Instrument which utlllzes a movlng belt Interface and a thennospray sample deposttlon devlce ( f ). Both secondary Ion mass spectra and laser desorptlon mass spectra can be acqulred under otherwlse Identical condltlons. Some slgniflcant mass spectral differences between SIMS and LDMS have been observed from amino aclds and nucleosides. Surface coverage and dosage effects on sample Ion currents have been studied and are discussed. The advantages and limltatlons of uslng a continuous ion beam vs. a pulsed laser beam In an LC/MS are examlned and evaluated.
Two techniques that have received much interest for molecular weight determinations on nonvolatile, thermally labile biomolecules are secondary ion mass spectrometry (SIMS) and 0003-2700/84/0356-1870$01.50/0
laser desorption mass spectrometry (LDMS). In 1976 Benninghoven and co-workers (2) showed that kiloelectronvolt primary ions at low current densities could be used to desorb intact molecular ions from organic compounds adsorbed on a metal surface with good sensitivity. This technique called “static” SIMS has been used to detect and identify a wide variety of nonvolatile organic samples including amino acids, peptides, vitamins, pharmaceutical compounds, nucleosides, nucleotides, and others ( 3 , 4 ) . Closely related to SIMS is fast atom bombardment (FAB), which uses a neutral primary atom beam and a liquid sample matrix (5, 6). With FAB highperformance magnetic instruments can be used and impressive results have been obtained from a wide variety of difficult samples, including essentially all of the samples determined by SIMS as well as other samples which have not been successfully analyzed before. The important difference between FAB and SIMS appears not to be the charge state of the primary ionizing beam, but rather it is the use of a liquid sample matrix. Laser desorption mass spectrometry (LDMS) of organic 0 1984 Amerlcan Chemical Society