Perfluorosulfonated Ionomer-Modified Polyethylene. A Material for

Anal. Chem. 2001, 73, 3701-3708

Perfluorosulfonated Ionomer-Modified Polyethylene. A Material for Simultaneous Solid-Phase Enrichment and Enhanced Precolumn Dansylation of C-21 Ketosteroids in Human Serum Patrik Appelblad,†,‡ Abdu Ahmed,† Einar Ponteń,†,§ Torbjo 1 rn Ba 1 ckstro 1 m,‡ and Knut Irgum*,†

Department of Chemistry, Umeå University, S-901 87 Umeå, Sweden, and Obstetrics and Gynecology, Department of Clinical Science, University Hospital, S-901 85 Umeå, Sweden

A new derivatization procedure has been developed where solid-phase catalysis is utilized to facilitate the formation of hydrazones in precolumn labeling of keto-containing compounds. This procedure has been implemented on a solid-phase enrichment and enhanced derivatization (SPEED) device, prepared from porous polyethylene that has been coated with Nafion and dansylhydrazine. The SPEED devices have been optimized using experimental design and characterized for dansylation of C-21 ketosteroids by multivariate data analysis, using progesterone as the model compound. The reaction temperature and the molar ratio between the steroid and the derivatization reagent were found to be the factors most strongly affecting the reaction. Faster reaction kinetics were achieved when the molar ratio between dansylhydrazine and the steroid was increased. Mass spectroscopic analysis showed that the four derivative peaks eluting when derivatized progesterone was separated on an octadecyl silica stationary phase were due to the syn and anti mono- and bis(hydrazones) formed in the reaction. Using optimal reaction conditions, the derivatives mainly constitute the syn and anti conformers of bis-derivatives. In contrast to solution-based acid catalysis, the SPEED device was remarkably insensitive to water in the reaction mixture. A sample volume of 400 µL was found to be the maximum, enabling sample enrichment prior derivatization. Using optimal experimental conditions, picomole amounts of ketosteroids could be derivatized in 10 min at room temperature. Analysis of spiked serum samples containing 0.4-2.0 nmol of progesterone showed overall recoveries of 52-63%. The corresponding 3σ detection limit was 1.3 pmol (n ) 4, 100 µL injected), as estimated from calibration curve data. Development of sensitive and robust analytical techniques that allow reliable quantitation of neurosteroids in body fluids has become an increasingly important task over the past decades. Considering the vast number of metabolites naturally occurring * Corresponding author: e-mail: [email protected]. † Umeå University. ‡ University Hospital. § Present address: SeQuant AB, Box 7956, S-90719 Umeå, Sweden. 10.1021/ac010071w CCC: $20.00 Published on Web 06/30/2001

© 2001 American Chemical Society

Figure 1. Structures of the steroids investigated. Progesterone (4pregnene-3,20-dione), 5β-pregnane-3,20-dione (I), 5R-pregnane-3,20-dione (II), 3R-hydroxy-5β-pregnan-20-one (III), 3β-hydroxy-5Rpregnan-20-one (IV), and 20R-hydroxy-4-pregnen-3-one (V).

at very low levels, high flexibility is required in both the sample pretreatment and detection procedures. A major group of neurosteroids with physiological significance1 is the C-21 ketosteroids from the pregnenolone and progesterone metabolic pathways; cf. Figure 1. Several of these metabolites lack functional groups with useful detectable properties attached to their molecular backbone, wherefore only few analytical techniques are applicable. At present, mass spectrometric detection coupled with either gas or liquid chromatographic separation is considered as the state-of-the-art technique.2-4 Although these instrumental systems provide high efficiency and excellent sensitivity, they are expensive and (1) (a) Baulieu, E. E.; Robel, P. J. Steroid Biochem. Mol. Biol. 1990, 37, 395403. (b) Baulieu, E. E. Psychoneuroendocrinology 1998, 23, 963-987. (2) Ma, Y.-C.; Kim, H.-Y. J. Am. Soc. Mass Spectrom. 1997, 8, 1010-1020. (3) Liere, P.; Akwa, Y.; Weill-Engerer, S.; Eychenne, A.; Pianos, B.; Robel, P.; Sjo ¨vall, J.; Schumacher, M.; Baulieu, E. E. J. Chromatogr., B 2000, 739, 301-312. (4) Wolthers, B. G.; Kraan, G. P. J. Chromatogr., A 1999, 843, 247-274.

Analytical Chemistry, Vol. 73, No. 15, August 1, 2001 3701

Figure 2. Schematic drawing showing the formation of mono- and bis-derivatives of progesterone.

complicated. Other techniques with sensitivities sufficient for trace determination are therefore still needed. It has previously been shown that an alternative could be liquid chromatographic separation of precolumn dansylated steroids, coupled with postcolumn peroxyoxalate chemiluminescence detection for quantification of the formed hydrazones.5,6 However, the derivatization procedure is carried out through acid catalysis in the liquid phase, where the reaction kinetics will ultimately depend on the concentration of the analyte. Derivatization and quantitation of trace levels of analyte in complex matrixes is thus not a feasible scheme. Unfortunately, these conditions are prevailing and the only way of circumventing this problem is to use a larger excess of reagent leading to excessive reagent blanks or to employ a more efficient catalyst. It has been shown that the type of acid used to catalyze the reaction influences both the rate and yield of the hydrazone formation reaction.7-10 In this context, we have introduced11 and optimized12 the use of trifluoromethanesulfonic acid (TFMSA) as a catalyst for precolumn dansylation of a number of ketosteroids. Using this catalyst, which is among the strongest known monoprotic Bro¨nsted acids, it is possible to quantitate low-picomole levels of ketosteroids in both plasma and serum.6,12 Unfortunately, this acid spontaneously forms highly toxic and carcinogenic methyl triflate when diluted in anhydrous methanol, a required component of the derivatization reagent. Derivatized samples must consequently be handled as risk waste and TFMSA-catalyzed derivatization is thus somewhat problematic to implement in routine laboratories.13 The formation of multiple derivatives is another problem associated with this precolumn labeling reaction, as illustrated in (5) Higashidate, S.; Hibi, K.; Senda, M.; Kanda, S.; Imai, K. J. Chromatogr. 1990, 515, 577-584. (6) Appelblad, P.; Jonsson, T.; Ba¨ckstro ¨m, T.; Irgum, K. Anal. Chem. 1998, 70, 5002-5009. (7) Koziol, T.; Grayeski, M. L.; Weinberger, R. J. J. Chromatogr. 1984, 317, 355-366. (8) Kawasaki, T.; Maeda, M.; Tsuji, A. J. Chromatogr. 1979, 163, 143-150. (9) Iwata, J.; Suga, T. J. J. Chromatogr. 1989, 474, 363-371. (10) Kawasaki, T.; Maeda, M.; Tsuji, A. J. Chromatogr. 1983, 272, 261-268. (11) Hyytiaïnen, M.; Appelblad, P.; Ponteń, E.; Stigbrand, M.; Jaegfeldt, H.; Irgum, K. J. Chromatogr. 1996, 740, 279-283. (12) Appelblad, P.; Ponteń, E.; Jaegfeldt, H.; Ba¨ckstro¨m, T.; Irgum, K. Anal. Chem. 1997, 69, 4905-4911. (13) Howells, R. D., McCown, J. D. Chem. Rev. 1977, 77, 69-92.

3702

Analytical Chemistry, Vol. 73, No. 15, August 1, 2001

Figure 2.12,14-16 The model compound progesterone is a C-21 ketosteroid that carries two keto groups in the molecular backbone, at the 3- and the 20-positions. There is also a double bond conjugated to the 3-keto group. Formation of the hydrazones is also subject to syn and anti addition, yielding products that tend to separate on achiral HPLC columns. When progesterone is derivatized with dansylhydrazine (DNSH), we could therefore expect up to four hydrazone peaks in chromatograms run on regular C18 columns.12,14,15 It has previously been shown that the 3-keto group is more reactive than the keto groups in the 17- and 20-positions,14 and by using TFMSA as catalyst, we have shown that it is possible to control the hydrazones formed to the syn and anti monoderivatives only.12 However, to improve the sensitivity of this technique, it would be advantageous to attach two dansyl moieties to the molecular backbone, yielding bis-derivatives from progesterone. With two labels attached to the molecular backbone, the derivatives will become more lipophilic and a mobile phase with high eluting strength can be used, thereby reducing the risk of coelution with impurities, or with other labeled metabolites. Provided internal quenching does not take place, the total quantum yield would also be higher from a bis-derivative compared to a monoderivative. Correspondingly, Weinberger et al. showed that it is possible to yield both mono- and bis-derivatives from progesterone by using an “evaporative-derivatization technique”.15 With their scheme, it is possible to complete the reaction within 10 min and to control the reaction in favor of the bis-derivatives, but the experimental conditions that have to be applied are rather harsh. Another strategy to obtain increased reaction kinetics is to utilize heterogeneous catalysis and carry out the reaction on a solid sorbent. This scheme was previously shown for precolumn labeling of amines,16 as well as for dansylation of carbonyl compounds with 2,4-dinitrophenylhydrazine (DNPH).17 In these studies, the derivatization reagent was added to commercial C18 (14) Goehl, T. J.; Sundaresan, G. M.; Prasad, V. K. J. Pharm. Sci. 1979, 68, 1374-1376. (15) Weinberger, R.; Koziol, T.; Millington, G. Chromatographia 1984, 19, 452456. (16) Herraéz-Hernańdez, R.; Campıńs-Falco´, P.; Sevillano-Cabeza, A., Anal. Chem. 1996, 68, 734-739. (17) Vogel, M.; Karst, U. Poster PB8/54, HPLC 99, Granada, Spain.

solid-phase extraction cartridges prior to addition of the sample, and in the latter work, DNPH was added in a mixture with a sulfonated cation-exchange resin to facilitate the catalysis.17 To utilize the beneficial features of heterogeneous catalysis and to combine it with TFMSA catalysis, we have thus investigated Nafion, a perfluorinated polymer that can be considered as a solidphase analogue of TFMSA,18 as a solid-phase catalyst (SPC) for the precolumn dansylation of C-21 ketosteroids. Using Nafion as an SPC, formation of triflic esters will occur on the polymeric solid sorbent, which means that exposure to hazardous chemicals is minimized compared to solution-based TFMSA catalysis. Moreover, Nafion has ion-exchange properties and by dynamically modifying a solid-phase support material with low porosity and with a narrow pore size distribution, the new SPC material can act as a solid-phase enrichment material where sample cleanup and enrichment could be obtained simultaneous with the derivatization. In this study, we have thus developed a solid-phase enrichment and enhanced derivatization (SPEED) device which is basically a polypropylene (PP) syringe reservoir where a porous polyethylene (PE) disk has been inserted. This disk was then coated with known amounts of DNSH and Nafion. Compared to solution-phase derivatization, this device offers a simplified and more easily automated precolumn dansylation reaction with minimized exposure to hazardous compounds. Operational parameters such as reaction temperature and time and concentrations of DNSH and catalyst have been studied using experimental design coupled with multivariate data analysis, and spiked human sera have been analyzed for progesterone to determine repeatability, reproducibility, recovery, and sensitivity of the SPEED device. EXPERIMENTAL SECTION Reagents and Materials. The steroids 4-pregnene-3,20-dione (57-83-0), 5R-pregnane-3,20-dione (566-65-4), 5β-pregnane-3,20-dione (128-23-4), 3R-hydroxy-5β-pregnan-20-one (128-201), 3β-hydroxy-5R-pregnan-20-one (516-55-2), and 20R-hydroxy4-pregnen-3-one (145-14-2) were all purchased from Sigma (St. Louis, MO). Budesonide (11β,21-dihydroxy-16R,17R-[22R,S]-propylmethylenedioxy-pregna-1,4-diene-3,20-dione) was obtained from Draco La¨kemedel AB (Lund, Sweden). Dansylhydrazine (5(dimethylamino)naphthalenesulfonic hydrazide; 97%) was acquired from Fluka Chemie AG (Buchs, Switzerland), and Nafion perfluorinated ion-exchange resin (H+ form; 5 wt % solution in lower aliphatic alcohols and water) (31175-20-9) was from Aldrich (Steinheim, Germany). Ammonium acetate (p.a.) was acquired from Merck (Darmstadt, Germany), while acetonitrile (HPLC grade) and methanol (HPLC grade) were purchased from J.T. Baker (Deventer, The Netherlands). All water used was purified through Super-Q (Millipore, Bedford, MA) equipment and had an electrolytic conductivity of less than 60 nS/cm. Porous PE with a pore size ranging between 15 and 45 µm and 3-mm thickness (material A) was received as a gift from Porex Technologies (Bautzen-Sigwith, Germany). Porous PE materials PPM-T, with an average pore size of 9 µm and 3.2-mm thickness (material B), and PPM-D, with an average pore size of 16 µm and 4.75-mm thickness (material C), were received as gifts from PIAB (Ta¨by, Sweden). Another porous PE material, with an average pore size (18) Waller, F. J.; Van Scoyoc, R. W. CHEMTECH 1987, 17, 438-441.

of 10 µm and 1.5-mm thickness (material D) was received as a gift from ISTsInternational Sorbent Technology Ltd. (Ystrad Mynach, U.K.). Preliminary Experiments Using Tubular Nafion Membrane as Catalyst. Chopped pieces (∼1 mm long) of 811X tubular Nafion membrane (Perma Pure, Toms River, NJ) were placed in a 1.5-mL polypropylene vial, whereafter 40 µL of 1.2 mM budesonide in acetonitrile and 100 µL of 2.3 mM DNSH in methanol were added. The reaction was allowed to proceed for 1 h and finally 60 µL of 0.4 M NaOH was used to strip the formed derivatives off the membrane catalyst. The sample recovered by this extraction was diluted 100 times with eluent and analyzed with reversed-phase liquid chromatography using fluorescence detection. Preparation of SPEED Devices. Initial experiments were carried out with 6.2-mm-diameter frits manually punched from porous PE disks obtained from several manufacturers. These disks were forcibly inserted into 1-mL PP syringes (IST). The porous PE disks were then washed with 2 mL of water followed by 2 mL of methanol and thereafter dried at 85 °C for 10 min. Nafion and dansylhydrazine were then added to the porous PE so that the material was thoroughly and evenly soaked by the catalyst and reagent solutions (exact experimental conditions are given in table and figure footnotes). The devices were then allowed to dry at 85 °C for 10 min or at room temperature for 48 h, wrapped in aluminum foil for protection against light. The final version of the SPEED device utilized commercially available PP syringes equipped with 1.5-mm-thick PE disks of nominal pore size 10 µm (IST; part no. 120-1061-A). These disks were modified by adding 30 µL of Nafion solution, followed by 50 µL of a 3.0 µM methanolic dansylhydrazine solution. The SPEED devices thus produced were heat sealed for storage in aluminum foil/PE laminate with a silica gel sachet. Extraction and Cleanup. Human serum (0.50 mL) was pipetted into a cylindrical flat bottom glass vial of 10-mL volume, to which water (0.5 mL) and diethyl ether (3.0 mL) were subsequently added. The samples were then allowed to swirl gently on an orbital shaker for 10 min. Following the liquid-liquid extraction, the vials were transferred into an ethanol/dry ice bath, where the aqueous phase froze. The ether phase was then decanted and evaporated under a stream of helium. The residue was finally redissolved in 400 µL of a water/methanol mixture (90:10 v/v) prior to analysis. To determine the analytical recovery of the technique, known amounts of progesterone (0.2 and 2.0 nmol) were added to an extraction vial and the solvent was evaporated to dryness at room temperature prior addition of serum. Derivatization of Steroids Using the SPEED Device. Steroid standard solutions and spiked serum samples (400 µL in both cases) were applied directly onto the coated disk of the SPEED device where the derivatization reaction was allowed to proceed for varying amounts of time. Upon completion of the reaction, the formed derivatives were eluted with 300 µL of eluent, of which 100-µL aliquots were injected into the HPLC system. Blanks were prepared as above but contained no steroids. Chromatographic System. The chromatographic system consisted of an LKB 2150 HPLC pump working at 0.5 mL/min, an Hitachi F1000 fluorescence detector (Merck-Hitachi, DarmsAnalytical Chemistry, Vol. 73, No. 15, August 1, 2001

3703

tadt, Germany) equipped with a 12-µL flow cell, and the excitation and emission wavelengths set at 350 and 520 nm, respectively. However, in the experiments evaluating the analytical repeatability, reproducibility, recovery, and sensitivity, a Jasco (Tokyo, Japan) FP-920 fluorescence detector equipped with a 16-µL flow cell was used. Analytical separations were carried out on a Phenomenex (Torrance, CA) Luna C18(2) 5-µm, 150 × 2 mm i.d. column and the detector output displayed in Tables 1-4 and Figures 3 and 4 were recorded on a model 3395 electronic integrator (HewlettPackard, Palo Alto, CA). All other data were recorded on a PCbased DataApex CSW (version 1.7; Prague, The Czech Republic) chromatography workstation. A Rheodyne 7125 injector valve (Cotati, CA) equipped with loops of either 20-, 50-, or 100-µL volume was used in all experiments where derivatizations were carried out manually. In the fully automated derivatization procedure, sample handling was accomplished using a Gilson Aspec Xli autoinjector (Villiers-le-Bel, France) equipped with either a 100- or 168-µL loop. Isocratic separation was carried out with an eluent consisting of acetonitrile and a 50 mM ammonium acetate buffer, pH 4.7 (80:20 v/v). All eluents were filtered through a 0.47-µm PVDF filter (Millipore) and degassed by purging with helium for 15 min prior to use. Mass Spectrometry. The mass spectrometric analyses were performed on an LCQ DUO mass spectrometer with an ESI source (Finnigan/Thermoquest, San Jose, CA). The samples were infused into the ion spray interface at 2.5 µL/min via a 50-µm-i.d. fusedsilica capillary, using an electronically controlled syringe pump integrated in the instrument. An ESI spray voltage of 4.5 kV was applied, the nitrogen sheath gas and auxiliary gas flows were set to 59 and 38 (instrument settings in arbitrary units), respectively, and the heated capillary was operated at a temperature of 200 °C and a voltage of 30.4 V. The tube lens offset voltage was set to 5 V. A maximum injection time of 500 ms and 3 microscans (scan range 150-1000 Da) were used. Experimental Design and Multivariate Data Analysis. Screening and optimization of the derivatization reaction was performed using Modde 4.0 experimental design and multivariate analysis software (Umetrics, Umeå, Sweden). Influence of Water. Seven standard solutions, each containing 50 µM progesterone in varying amounts of water (0-99%) were prepared from progesterone stock solutions A (5104 µM) or B (505.4 µM) in methanol. The 99% aqueous solution was prepared by weighing 1 mL of A and adding to a 100-mL volumetric flask, whereafter water was added to the mark while weighing. For all other standard solutions, a 5-mL quantity of B was taken to a 50mL volumetric flask, whereafter appropriate volumes of methanol (5-45 mL) and water (0-40 mL) were added while weighing the solution. In all experiments, 80 µL of standard solution (4 nmol) was added to a SPEED device, then allowed to react for 10 minutes, and finally eluted from the SPEED device with 300 µL of eluent. The eluate was then homogenized by applying three aspiration/discharge cycles, and finally, 100 µL (1.33 nmol) was injected. Triplicate analyses were carried out at each concentration level. The dotted lines defining the confidence interval in Figure 6 represent the standard error of the estimate. Sample Breakthrough. Aliquots of increasing volume (from 80 to 500 µL) of a standard solution containing 5.0 µM progesterone in 10:90 methanol/water (v/v) were derivatized to determine 3704


the breakthrough volume for the SPEED devices. Quadruplicate experiments were done at all volumes. Evaluation of Repeatability, Reproducibility, Recovery, and Sensitivity. A data set containing 45 unique samples was used to determine the repeatability, reproducibility, recovery, and sensitivity of the technique. The first series consisted of 10 standard solutions whereas the other series contained 12 spiked human serum samples and 23 standard solutions. The evaluated steroid standard solutions contained progesterone ranging in concentration from 0.5 to 5.0 µM (0.2 to 2.0 nmol in the reaction mixture). Retention time repeatability of the chromatographic system was calculated for the entire data set (n ) 45), as were the variabilities within series (n ) 10) and within day (n ) 35). Reproducibility for steroid standard solutions was evaluated by several injections of standards at the 0.5, 1.0, 2.5, and 5.0 µM levels. The relative standard deviation (RSD) was calculated for progesterone based on peak height measurement of the later eluting of the bis-derivatives. The recovery and the detection limit were determined by analyses of aliquots of human serum spiked with known amounts (0.4-2.0 nmol) of steroids. RESULTS AND DISCUSSION The hypothesis that a polymer carrying perfluoroalkylsulfonic acid groups could act as a catalyst and reagent carrier for the derivatization of ketosteroids was envisaged shortly after our discovery that trifluoromethanesulfonic acid has an extraordinary catalytic effect on the reaction of dansylhydrazine with ketosteroids.6,12 An initial experiment was carried out by adding chopped pieces of a tubular Nafion membrane (811X) into a vial, to which a budesonide standard solution and DNSH solution were added. The reaction was then allowed to proceed for 1 h whereafter the polymer was stripped with base and the resulting solution analyzed for formed derivatives by liquid chromatography with fluorescence detection. Peaks appearing at the same retention time as derivatives formed in homogeneous solution clearly showed that hydrazones were formed from the steroid and that this formation took place because of the presence of the Nafion perfluorosulfonic acid, since peaks did not appear in the absence of this catalyst. Although solid Nafion is commercially available in several confections, we found the solution prepared by dissolving the polymer in lower aqueous alcohols19 to be most suitable. The liquid form makes it possible to prepare coatings of varying thicknesses on solid support materials with different surface area and porosity. It is thereby possible to prepare different types of devicessas particles to be packed in columns or as disks incorporated in solidphase extraction reservoirs. The latter proved to be the most feasible approach, as all starting materials are commercially available and the familiar format of the SPE column allows its direct incorporation in existing sample preparation robots. Derivatization of Progesterone Using the SPEED Device. Porous PE, a material that is chemically inert and available with controlled porosity, was considered the most suitable support material for Nafion coating. One-milliliter PP reservoirs were thus prepacked with porous PE disks, loaded with solutions of Nafion and DNSH, and thereafter dried to form a SPEED device. Attempts at labeling progesterone with the SPEED device using reaction (19) Grot, W. G. U.S. Patent 4,433,082, 1984.

Table 1. Derivatization Yield as a Function of the Amount of Catalyst Applied to the PE Supporta) expt

Nafion amt, µg

total peak heightb (units/105)

RSD %

1 2 3 4

0.44 0.88 1.31 2.18

2.13 ( 0.23 2.57 ( 0.06 2.20 ( 0.26 2.47 ( 0.06

10.8 2.2 12.0 2.3

a In each experiment, three 15-µL aliquots of 5 µM progesterone standard were derivatized using 10 µL of a 2 mM DNSH solution. Formed derivatives were eluted with 1 mL of HPLC eluent, and the volume injected in the chromatographic system was 100 µL. b Total peak height corresponds to the integrator response for both syn and anti bis-derivatives of progesterone.

conditions previously found optimal for TFMSA catalysis12 demonstrated that four hydrazones, corresponding to the syn and anti derivatives of both mono- and bis-derivatives, were formed. Mass spectrometric analysis of derivatives in fractions collected from the HPLC separation verified that the two first eluting derivatives had an MH+ of 562.3, whereas the two later eluting derivatives both had an MH+ of 809.6. This correlated well with formation of syn- and anti-conformers of both mono- and bis-derivatives; Figure 2. However, a rather high relative standard deviation was seen between replicate derivatizations (14.6% RSD, n ) 4), and it was noticed that part of the applied sample solution was not retained on the sorbent, since channels were formed between the PE disk and the sample reservoir wall. This was caused by difficulty in punching the in-house made PE disks with smooth rims and inner diameters exactly matching the PP reservoirs. To improve the derivatization reproducibility, disks with a slightly larger diameter were therefore punched out and used. Derivatization Yield as a Function of Catalyst Concentration. Amounts of Nafion ranging from 0.4 to 2.2 µg were added as 5 wt % aqueous alcoholic solution (10-50 µL) to the PE disks and the relative derivatization yield based on the combined peak height of both bis-derivatives was studied; see Table 1. From these experiments, it became evident that there was no significant correlation between the derivatization yield and the amount of Nafion loaded on the PE disks. The variation among all replicates

was found to be 10.4% RSD (n ) 12) while it varied between 2 and 12% at the individual concentration levels. The difference is mainly due to small variations in reaction time and the fact that the derivatization and sample injection were carried out manually. It can thus be concluded that the catalytic action is established already with 0.5 µg of Nafion. Attempts at derivatizing other ketosteroids in addition to progesterone were thereafter done, using compounds containing either one or two keto groups in the molecular backbone as models. On the basis of the number of peaks appearing in the chromatograms, it was concluded that 5R-pregnane-3,20-dione (IV),5β-pregnane-3,20-dione (V) formed bis-derivatives while 20R-hydroxy-4-pregnane-3-one (VI), 3R-hydroxy-5β-pregnane-20-one (IX), and 3β-hydroxy-5R-pregnane-20one (VIII) formed monoderivatives. Consequently, we believe that the derivatization procedure is generally applicable to steroids containing a carbonyl moiety. Experimental Design and Multivariate Data Analysis. The effect of reaction time and temperature on the derivatization yield for the bis-derivatives was evaluated using a centrally circumscribed composite (CCC) experimental design and partial leastsquares (PLS) analysis in the Modde software package; see Table 2. Modeling of data revealed that the reaction time (TI) had no significant effect while the reaction temperature (TE) had a strong influence on the reaction yield for formation of progesterone bisderivatives, in the tested experimental domain. After removing nonsignificant factors and inserting a quadratic term (TE × TE), we obtained R2 ) 0.97 and Q2 ) 0.96, showing that the projected model was valid and that the predictive power was reliable. From Figure 3 it could moreover be concluded that the best derivatization yield was obtained at room temperature, and it should be noted that the lowest yield for progesterone bis-derivatives was obtained at 50 °C. This observation correlates well with the solution-phase TFMSA catalysis, where this temperature was found to give the highest derivatization yield for the monoderivatives of progesterone.12 Automated Precolumn Derivatization. Having established room temperature as the optimal condition for formation of progesterone bis-derivatives, the derivatization procedure was automated in order to improve the reproducibility. Nineteen SPEED devices with varying amounts of Nafion and DNSH were

Table 2. Effect of Temperature and Time on Dansylation of Progesterone According to the CCC Designa variables

responses

expt

TE (°C)

TI (min)

peak 1 height (units/105)


total height (%)

peak 1 area (units/106)


total area (%)

1 2 3 4 5 6 7 8 9 10 11

32 30 28 28 40 40 50 50 25 25 85

6 30 60 37.5 37.5 37.5 15 60 25 60 10

0.74 0.96 1.03 1.00 0.62 0.67 0.52 0.56 1.02 1.09 0.76

0.46 0.55 0.64 0.54 0.39 0.40 0.31 0.34 0.55 0.60 0.54

71 89 99 91 60 63 49 53 92 100 76

1.68 2.17 2.32 2.25 1.41 1.52 1.19 1.38 2.32 2.50 1.83

1.50 1.79 2.13 1.67 1.43 1.34 1.00 1.28 1.82 1.99 2.15

71 88 99 87 63 64 49 59 92 100 88

a The following factors were studied: temperature (TE) and reaction time (TI). Ten microliters of 5 µM progesterone standard, 50 µL of 2 mM DNSH, and 30 µL of Nafion were used in all experiments. The sorbent was then dried with N2, and the derivatized samples were eluted with 1 mL of eluent. The injected sample volume was 100 µL. Experiment 10 showed the highest response; the total area and total height were consequently normalized to this experiment.


3705

Table 3. Results from the 24 Full Factorial Designa variables

responses

experiment

CC (µg)

DN (mM)

TE (°C)

TI (min)



total height (units/105)



total area (units/107)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

0.87 2.18 0.87 2.18 0.87 2.18 0.87 2.18 0.87 2.18 0.87 2.18 0.87 2.18 0.87 2.18 1.53 1.53 1.53

1 1 3 3 1 1 3 3 1 1 3 3 1 1 3 3 2 2 2

25 25 25 25 40 40 40 40 25 25 25 25 40 40 40 40 30 25 25

10 10 10 10 10 10 10 10 50 50 50 50 50 50 50 50 30 30 30

1.49 1.48 4.15 3.05 1.69 1.49 3.88 2.21 1.81 0.90 3.15 2.88 1.81 1.63 2.27 1.77 3.04 3.70 3.69

0.79 0.81 2.30 1.61 1.05 0.93 2.50 1.20 1.05 0.56 1.64 1.51 1.14 1.03 1.35 1.08 1.64 2.17 2.07

2.28 2.28 6.45 4.66 2.73 2.42 6.38 3.40 2.86 1.46 4.79 4.39 2.95 2.66 3.62 2.86 4.68 5.87 5.76

0.30 0.30 0.81 0.59 0.32 0.28 0.61 0.43 0.34 0.17 0.60 0.55 0.35 0.29 0.45 0.35 0.56 0.62 0.68

0.18 0.19 0.51 0.35 0.24 0.20 0.55 0.27 0.24 0.13 0.37 0.32 0.28 0.24 0.31 0.26 0.36 0.48 0.46

0.47 0.50 1.32 0.94 0.56 0.48 1.16 0.70 0.57 0.30 0.97 0.88 0.63 0.53 0.76 0.62 0.93 1.10 1.15

a The following factors were studied: catalyst amount (CC), concentration of the derivatization reagent (DN), reaction temperature (TE) and reaction time (TI). Progesterone (35 µL of a 100 µM solution in 100% methanol) was used as standard and derivatized in all experiments.

Figure 3. Response surface plot showing the dependence of the relative derivatization yield on temperature and time of the reaction. The model was based on combined peak areas for the progesterone bis-derivatives.

Figure 4. Response surface plot showing the dependence of the relative derivatization yield on the amount of catalyst and the concentration of the derivatization reagent. The model was based on peak area for the first-eluting progesterone bis-derivative.

prepared and used for derivatization of progesterone using an Gilson Aspec Xli autoinjector according to a 24 full factorial experimental design; Table 3. The effect of the DNSH (DN) and catalyst amount (CC), TE, and TI were investigated using PLS analysis as above. After removing nonsignificant factors and inserting two quadratic terms (CC × CC) and (DN × DN), we obtained R2 ) 0.91 and Q2 ) 0.80, which shows that the projected model was valid and that the predictive power was adequate. Modeling of experimental data revealed that the derivatization yield increased with the DNSH concentration (see Figure 4), and the highest yield was achieved after 10 min at room temperature using a low amount of Nafion as catalyst. These data correlate well with the above CCC design, and it thus can be concluded that room temperature and a short reaction time are the best conditions for obtaining progesterone bis-derivatives. Comparison of Porous PE Support Materials for the SolidPhase Derivatization Reaction. Four different PE materials with

different thickness and porosity were evaluated as support materials, using derivatization yield as optimization criterion. The data obtained, presented in Table 4, show that there were no significant differences between SPEED devices prepared from materials A, B, and D, while material C produced a device with a significantly lower catalytic ability. This cannot be explained by differences in the nominal porosities of the materials, but since the same amount of desorption solution was used on all devices, it may have been insufficient to quantitatively elute the SPEED device based on C, which was the thickest of the tested supports. Since D is commercially available prepacked in 1-mL PP reservoirs, it was selected for the continued studies. Final Optimization of the SPEED Derivatization. As a provocation test of the derivatization reaction, a final CCC design was carried out, where the experimental domain was expanded around the provisional optimum found in the 24 full factorial design

3706 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001

Table 4. Comparison of PE Sorbents Used as Supports for the Solid-Phase Derivatization Reactiona material characteristics PE material A B C D

responses

pore size (µM)

thickness (mm)

total peak area (units/107)

RSD (%)

n

15-45b

3 3.2 4.75 1.5

1.05 ( 0.10 1.03 ( 0.04 0.70 ( 0.02 1.06 ( 0.08

9.3 3.9 3.6 7.4

4 3c 4 4

9 16 10

a Four PE sorbents with different thickness and porosity were studied: In all experiments, 30 µL of Nafion, 50 µL of a 3 mM DNSH, and 15 µL of a 100 µM progesterone standard solution were used. The reaction time was set to 10 min at room temperature. The formed derivatives were eluted with 1 mL of eluent, and 100 µL was injected in the HPLC system. b The only pore size distribution was given by the vendor. c One of the replicates for material B was found to be an outlier and consequently excluded. However, as the experiment was at the boundary of being excluded, we report all the individual values for this material (1.02, 1.39, 0.99, 1.07).

described above. The derivatization yields obtained in these experiments are presented in Table 5. Interestingly, as visualized in the response plot in Figure 5, the derivatization yield remained constant over a large part of the experimental domain tested. The reaction time thus played a minor role and the yield was nearly constant, as long as the amount of reagent was sufficient. PLS analysis on the experimental data furthermore revealed that an optimum had been reached since no statistically significant factors were found. This is augmented by the low regression prediction coefficients (R2 ) 0.82, Q2 ) 0.66), indicating that the experimental noise was high in comparison to the systematic variations in derivatization efficiency resulting from the provocations. Since the parameters appear to be situated on a plateau, excellent derivatization reproducibility was obtained among replicates (2.3% RSD; n ) 3) when the area response was evaluated for the first-eluting bis-derivative. Consequently, it is recommended that the SPEED device is loaded with 50 µL of a 3.0 mM (150 nmol) DNSH solution and 30 µL of Nafion, and that the derivatization should be allowed to proceed for 10 min at room temperature. Influence of Water on Derivatization Yield. It has been shown in a previous study, where solution-phase TFMSA catalysis was used, that precolumn dansylation of progesterone is sensitive

Figure 5. Response surface plot showing the dependence of the relative derivatization yield on the reaction time and the concentration of the derivatization reagent. The model was based on peak height measurements of the first-eluting progesterone bis-derivative.

toward water in the reaction mixture.12 Experiments with water deliberately added to the reaction mixture were therefore carried out to determine whether the sensitivity toward water in the samples was lower with the solid-phase catalyst, compared to TFMSA catalysis. Seven standard solutions, each containing 50 µM progesterone and a varying proportion of water (0-99%), were derivatized. A remarkable relationship was found, as shown in Figure 6, where a maximum in reaction yield was obtained when the reaction mixture contained 10-20% methanol. Sample breakthrough is a plausible explanation for the inferior recovery at high methanol concentrations, whereas the decrease in catalytic efficiency at low methanol concentrations is supported by an alkylation catalysis mechanism,12 which requires methanol for the perfluoroalkanesulfonic acid to perform its catalytic action. Experiments with a sample size of 40 µL instead of 80 µL resulted in better yield at higher methanol concentrations (data not shown), supporting breakthrough due to self-elution as a probable cause of the low yields at higher methanol concentrations. Maximum sample loadability was thereafter investigated by derivatizingincreasing volumes (80-500 µL) of a 5.0 µM progesterone solution (0.4-2.5 nmol) containing 10% methanol. A linear relationship was

Table 5. Results from the CCC Design Using Commercially Packed Syringesa variables

responses

experiment

DN (mM)

TI (min)





total area (units/107)

1 2 3 4 5 6 7 8 9 10 11

2.46 3.55 2.46 3.55 2.30 3.63 3 3 3 3 3

8 8 12 12 10 10 7.17 12.83 10 10 10

2.26 3.68 1.93 3.56 2.24 3.37 3.18 3.76 3.40 3.15 3.25

1.59 2.50 1.38 2.40 1.77 2.26 2.20 2.45 2.25 2.24 2.27

0.64 1.02 0.54 0.96 0.69 0.96 0.83 0.96 0.93 0.90 0.89

0.49 0.80 0.40 0.75 0.59 0.74 0.63 0.70 0.67 0.74 0.72

1.13 1.83 0.93 1.71 1.28 1.71 1.46 1.66 1.60 1.63 1.61

a The following factors were studied: concentration of the derivatization reagent (DN) and reaction time (TI). Progesterone (40 µL of a 49.1 µM standard solution) was derivatized throughout all experiments. The formed derivatives were eluted with 500 µL of eluent, and 168 µL was injected into the HPLC system.


3707

Figure 6. Influence of water on derivatization yield. Progesterone (80 µL of a 50 µM solution) was derivatized and thereafter eluted with 300 µL of eluent, of which 100 µL was injected in the HPLC system. Error bars indicate standard error of the mean (n ) 3 at each level).

Figure 8. Chromatogram showing a derivatized progesterone standard solution (5.0 µM; solid line) superimposed on a blank run (dotted line).

progesterone, the recoveries were 52 ( 10 and 63 ( 17%, respectively. The limit of detection for progesterone was estimated from calibration curve data to 1.3 pmol (100 µl injected), based on three times the standard deviation of four 0.5 µM standard solutions. Although the recoveries are somewhat dissatisfactory, they are in the same range as in previous studies, where liquidliquid extraction was applied prior to solution phase TFMSA catalyzed dansylation of various C-21 ketosteroids.6,12 We hence conclude that the recovery deficiency of the presented technique may be an effect of the sample pretreatment, rather than an inferior function of the SPEED device.

Figure 7. Derivatization response as a function of sample volume applied to the SPEED device. Varying volumes of a 5 µM progesterone standard solution in 10% methanol were applied to SPEED devices to determine maximum loadability (n ) 4 at each volume). Error lines indicate the standard error of estimation.

seen over the entire tested range; see Figure 7. From the chromatograms, it was also apparent that the formed derivatives were mainly the syn and anti conformers of bis-derivatives, as visualized in Figure 8. Application to Human Serum Samples. The repeatability, the reproducibility, and the analytical recovery were evaluated using standard solutions ranging from 0.5 to 5 µM (0.2 to 2.0 nmol in the reaction mixture) and spiked human serum containing 0.42.0 nmol of progesterone. The analytical system was capable of a retention time repeatability of

Perfluorosulfonated Ionomer-Modified Polyethylene. A Material for

Recommend Documents