- Email: [email protected]
An automated method of sample preparation of biofluids using pierceable caps to eliminate the uncapping of the sample tubes during sample transfer
J. Biochem. Biophys. Methods 45 (2000) 193–204 www.elsevier.com / locate / jbbm
An automated method of sample preparation of biofluids using pierceable caps to eliminate the uncapping of the sample tubes during sample transfer Deborah S. Teitz, Sanaullah Khan, Mark L. Powell, Mohammed Jemal* Bioanalytical Research, Metabolism and Pharmacokinetics, Bristol-Myers Squibb Pharmaceutical Research Institute, P.O. Box 191, New Brunswick, NJ 08903 -0191, USA Received 16 January 2000; received in revised form 15 May 2000; accepted 1 June 2000
Abstract Biological samples are normally collected and stored frozen in capped tubes until analysis. To obtain aliquots of biological samples for analysis, the sample tubes have to be thawed, uncapped, samples removed and then recapped for further storage. In this paper, we report an automated method of sample transfer devised to eliminate the uncapping and recapping process. This sampling method was incorporated into an automated liquid–liquid extraction procedure of plasma samples. Using a robotic system, the plasma samples were transferred directly from pierceable capped tubes into microtubes contained in a 96-position block. The aliquoted samples were extracted with methyl-tert-butyl ether in the same microtubes. The supernatant organic layers were transferred to a 96-well collection plate and evaporated to dryness. The dried extracts were reconstituted and injected from the same plate for analysis by liquid chromatography with tandem mass spectrometry. 2000 Elsevier Science B.V. All rights reserved. Keywords: Automated sample transfer; Pierceable caps; LC / MS / MS; Sample preparation; Liquid–liquid extraction
1. Introduction Technological advancements in combinatorial chemistry have revolutionized drug *Corresponding author. Tel.: 11-732-519-1582; fax: 11-732-519-1557. E-mail address: [email protected] (M. Jemal). 0165-022X / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 00 )00117-2
194
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
discovery and development in the pharmaceutical industry. Consequently, the workload of the scientists in the discovery and development areas has increased with the influx of new compounds as potential drug candidates. To keep up with the increasing demand of the industry, scientists in these areas are actively involved in the laboratory scale automation of bioanalytical methods. The complex nature of biological matrices requires that sample clean-up be an integral part of any bioanalytical method. The sample preparation, with its multistep clean-up procedures, is usually the rate-limiting step in achieving high-throughput bioanalysis. Fortunately, the steps involved in sample preparation can be automated to streamline uninterrupted analysis of biological samples. In our laboratory, we have utilized two approaches for automation of sample preparation for quantitation by liquid-chromatography (LC) with tandem mass spectrometry (MS / MS). The first approach involves the direct injection of the biological samples into an LC / MS / MS system for on-line extraction [1–3]. The sample preparation involves simply aliquotting the biological sample into a 96-well plate, adding the internal standard and centrifuging before the direct injection from the plate. The first two steps of this sample preparation were automated using a robotic liquid handling system to further reduce the hands-on time the analyst spends on sample preparation. The second approach is based on automating off-line solid-phase extraction (SPE) and liquid–liquid extraction (LLE) in a 96-well format using a robotic liquid handling system [3–11]. Both types of automation have drastically reduced not only the length of time required for sample extraction but also the hands-on time spent by the analyst. However, the task of uncapping and recapping the sample tubes still appeared to be the ‘‘bottleneck’’ in sample analysis. It took approximately 30 min to uncap and recap 96 sample tubes. Hence, there was a need to eliminate this task. Another impetus for eliminating the uncapping / recapping process was that it is potentially hazardous as it may expose the analyst to the biological samples via accidental spillage or breakage. Also the manual uncapping and recapping puts physical stress on the human hands that may contribute to occupational injuries such as tendonitis and carpal tunnel syndrome. Here, we report the development and validation of automated LLE for extraction of plasma samples aliquotted by a robotic liquid handling system from sample tubes screw-capped with pierceable caps. The extracted samples were analyzed by LC / MS / MS. The probes of the robot pierced through the caps and transferred aliquots of samples to 96-well formatted microtubes for extraction.
2. Materials and methods
2.1. Reagents The analyte (I) and the 2 H 5 -labeled internal standard (II), shown in Fig. 1, are products of Bristol-Myers Squibb Pharmaceutical Research Institute. High purity (18.2 MV) water was obtained by passing house deionized water through the Milli-Q system
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
195
Fig. 1. Chemical structures of the analyte (I) and the 2 H 5 -labeled internal standard (II).
(Millipore, Bedford, MA). Methyl-tert-butyl ether (MTBE) and acetonitrile were obtained from EM Science (Gibbstown, NJ).
2.2. Equipment A Packard MultiPROBE 204DT robotic system, equipped with the MPTable software (Packard Instrument company, Meriden, CT), was used for the liquid handling operations. Polypropylene sample tubes (56 3 16 mm) and their caps were purchased from Elkay Products (Shrewbury, MA). We modified these caps through Andwin Scientific (Canoga Park, CA) by cutting a portion of the top of each cap and retrofitting it with a non-leaking resealable polymer septum (Fig. 2A). The plasma samples for analysis were contained in the Elkay tubes that were screw-capped with the modified caps. Microtubes, which come interconnected in strips of eight (Fig. 2B), were used for extraction. The eight-microtube strips were placed in 12 rows in a 96-position block (Fig. 2C) to provide 96 extraction microtubes. Plug-caps, interconnected in a strip of eight, were used to seal the microtubes. The microtubes, 96-position block and the plug-cap strips were obtained from Qiagen (Santa Clarita, CA). A special 96-position sample rack was designed in-house and then custom-made by Packard Instrument Company to hold the capped plasma sample tubes (Fig. 2D). The sample rack had a removable 96-hole cover that could be screwed down over the capped plasma sample tubes to hold them in place. This was necessary to prevent the robot probes from lifting the tubes during retraction. Furthermore, the sample rack was screwed down onto the deck of the robot to prevent the entire rack from moving as the robot probes retracted. Square-well 96-well collection plates and sealing mats to cover the plates were obtained from Varian (Harbor City, CA). A Savant Speed-Vac (Savant Instruments, Holbrook, NY) was used for drying the samples by evaporation under vacuum. A Beckman (Columbia, MD) System Gold pump (Model 128) was used for LC solvent delivery. The LC column used was a 2 3 50 mm (5 mm) Hypersil C8 column from Keystone (Bellefonte, PA). A Hewlett-Packard (Palo Alto, CA) HP 1100 column heater was used to control the column temperature. A Gilson (Middleton, WI) 233 XL autoinjector was used for programmed randomized sample injection from a 96-well plate. A Finnigan TSQ-7000 triple quadrupole mass spectrometer (Finnigan MAT, San Jose, CA) equipped with an electrospray interface was used for detection.
196
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 2. (A) A plasma sample tube with a pierceable cap. (B) A set of eight interconnected microtubes. (C) A 96-position block for holding the microtubes. (D) A sample rack for holding the plasma sample tubes on the robot deck.
2.3. Standard and quality control ( QC) preparations Calibration standards were prepared by spiking the appropriate amounts of I into
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
197
human plasma. The calibration curve range was 0.500–250 ng / ml, consisting of ten concentrations, with each in duplicate. The QC samples were also prepared by spiking I into human plasma to obtain concentrations of 1.50, 100, 200 and 2000 ng / ml as the low, mid, high and dilution QC samples, respectively. The standards and QC samples used for analysis were contained in the Elkay tubes with pierceable caps.
2.4. Automated cap piercing As the first step of the robotic sample transfer and extraction method for plasma samples in tubes capped with pierceable caps, the robot probes were programmed to penetrate through the polymer septum of each cap. For this, each probe was programmed to pick-up 100 ml of air and inject only 25 ml into each tube after piercing through the septum. The probes were programmed to specifically dip into the sample tubes just far enough to pierce the septa without coming into contact with the plasma in the tubes. Thus, the probes pierced all the sample tubes without touching the plasma samples. Both the inside and outside of the probes were washed with the system liquid (25% methanol in water) after each piercing step.
2.5. Sample processing After completion of the piercing sequence, aliquots (0.25 ml) of each sample were transferred from the pierced sample tubes to a set of 96-well formatted microtubes. To avoid the potential for carryover at this stage, the robot probes were each dipped deeply into a wash solvent (25% methanol in water) to rinse the entire length of the probes, followed by a thorough wash (inside and outside) with the system liquid after each sample transfer. Internal standard solution (25 ml), 0.5 N HCl solution (0.1 ml) and MTBE (0.5 ml) were then added sequentially by the robot to each microtube. The microtubes were manually sealed with the plug-cap strips and vortex-mixed for 5 min. To break an emulsion that occasionally occurred, the 96-position block containing the microtubes was immersed in an acetone / dry ice bath to freeze the contents of each microtube, sonicated for 3 min and finally centrifuged at 3000 g for 5 min. The organic layer from each microtube was transferred by the robot to a square-well 96-well plate. The plate was then placed in a Savant Speed-Vac for drying by evaporation. A 75 ml volume of the reconstitution solution (1.0 mM formic acid solution in 2:1 water– acetonitrile solvent mixture) was then added to each well by the robot. The 96-well plate was then covered with a sealing mat and transferred to the autoinjector for injection from the plate.
2.6. LC /MS /MS Electrospray ionization mass spectrometry was used in the negative ion mode. The electrospray voltage was 4.5 kV and the capillary temperature was maintained at 2508C. High purity (99.999%) nitrogen was used both as the nebulizing and the auxiliary gas. The conditions for Q1 were optimized to obtain maximum response for the [M–H] 2 precursor ion of the analyte. Argon, used as the collision gas, was set at 2.5 m Torr and
198
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
the collision offset energy (C off ) was optimized to obtain the maximum response of the product ion used for the selected reaction monitoring (SRM). The C off value of 17 eV was typically used for both I and II. The SRM transitions monitored were m /z 463→m /z 377 for I and m /z 468→m /z 382 for II. Unit mass resolution was maintained at both Q1 and Q3. The peak area ratio of the analyte (I) to the internal standard (II) was calculated at each concentration. The data were fitted to 1 /x-weighted quadratic regression model. LC separation was achieved isocratically on a Hypersil 5032 mm C8 column at a flow rate of 0.30 ml / min and a column temperature of 408C. The mobile phase consisted of 60% A and 40% B, where A was a 1.0 mM formic acid solution in 5% acetonitrile in water and B was a 1.0 mM formic acid solution in 95% acetonitrile in water. The effluent was all directed to the mass spectrometer.
3. Results and discussion
3.1. Automated cap piercing and direct sample transfer A significant part of the sample preparation time is normally consumed by repeated uncapping and recapping of biological sample tubes regardless of the sample preparation method. Yet, for decades, in bioanalytical laboratories, LC and GC autosamplers have been used to pierce through caps fitted with pierceable septa and pick up samples for injection. We saw no reason not to apply this type of technology to biological sample tubes. This prompted us to modify the commercially available tube caps by retrofitting them with pierceable septa. We then wanted to use the robot probes of a liquid handling system to pierce the septa of the capped sample tubes. Several problems were encountered. In initial attempts to pierce through the septa the robot was unreliable, giving us frequent z-motor errors. We quickly found out that when a septum is pierced for the first time, it presents the most resistance for the z-motor. Another important factor contributing to a high number of failures was a software shortcoming. In case of an initial penetration failure, the software allows multiple piercing attempts during a ‘‘dispense’’ operation whereas, an error is generated if any of the four probes fails to penetrate a septum during a ‘‘pick-up’’ operation. By incorporating a separate cap-piercing step while ‘‘dispensing’’ air into the tubes, we were in effect letting the robot use multiple attempts to pierce the septa without failing during the sample transfer from the capped tubes. Furthermore, injection of air into each tube also served to partially compensate for the vacuum that could result from pipetting the sample from a closed tube. Between pipetting each sample, we programmed the robot to dip into a wash solution. This was necessary because with the robot piercing through a cap, a large length of probe could come in contact with any sample stuck to the cap septum. The robot wash station, normally used to wash the probes between pipetting steps, is designed to wash only the tip of the probe. Therefore, the extra dip wash was included to clean off what the wash solution could not reach.
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
199
3.2. Resealability of the pierceable caps Before we used the pierceable caps for our study samples, we wanted to be sure that they were leak-proof and would remain leak-proof after piercing. We tested the effectiveness of sealing and resealing of the septa as follows: The caps were screwed onto the Elkay tubes and then submerged into an aqueous dye solution for 5 min. The insides of the caps were then checked for any seepage of the dye around the septa. No failure of the seals was detected. The septa were then pierced with the robot probes and the dye test was repeated. The septa remained leak-proof even after five piercing steps by the robot probes. The same dye test was done with pierceable caps that were stored for 3 months at 2708C to simulate sample storage conditions. No failure of the seal was seen. Similarly, the caps were tested for resealability after freezing and thawing. For this purpose, the tubes were pierced and frozen five separate times and then tested for leakage with the above dye test. The septa proved to be leak-proof under this condition as well, giving us the confidence to use this technology for our study samples.
3.3. LLE of plasma samples Plasma samples were extracted into MTBE after acidification with 0.1 N HCl. The robot-assisted liquid handling operations of the LLE significantly reduced both the analyst’s time and the overall time of the sample preparation. In early investigations, 96-square-well plates, 96-round-well plates and 96-well formatted microtubes (described above) were tested for extraction. The use of the plates or the microtubes, in lieu of the conventionally used individual tubes, eliminated the need for labeling of the individual tubes. However, with the 96-well plates there was always crossover of the extraction solvents between the wells, even if the wells were only partially filled and sealed off with the sealing mat. The microtubes were finally selected for performing the extraction because the crossover of the extract from one tube to another during vortex-mixing did not occur, probably because the microtubes are interconnected in rows of eight with only one point of attachment to the adjacent microtube (Fig. 2B) and the plug-caps did a better job of sealing each microtube. In addition, the volume of each microtube was sufficient to accommodate the plasma sample and the extraction solvents as the inside volume of the microtube was 1.2 ml and the accumulated extraction volume in the microtube was 0.875 ml. The extraction recovery of I from plasma was 45%.
3.4. LC /MS /MS The electrospray negative ion MS and MS / MS product ion spectra of I are shown in Fig. 3. The MS spectrum is dominated by the [M–H] 2 ion at m /z 463. The MS / MS product ion spectrum of the [M–H] 2 ion produced the major product ion at m /z 377. Thus, the SRM used was m /z 463→m /z 377 for the analyte and m /z 468→m /z 382 for the internal standard. The retention time of the analyte and the internal standard was 1.7 min (Fig. 4) and the chromatographic run time was 2.25 min.
200
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 3. (A) Electrospray negative ion MS spectrum of I. (B) Electrospray negative ion MS / MS product ion spectrum of the [M–H] 2 ion of I.
3.5. Method validation 3.5.1. Specificity Six different lots of control human plasma were analyzed with and without fortification with the internal standard in order to determine whether any endogenous plasma constituents interfered with the analyte or the internal standard. The degree of interference was assessed by inspection of the SRM chromatograms. No interfering peaks from the plasma were found at the retention time and in the ion channel of either the analyte or the internal standard. Following the high concentration point of the
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
201
Fig. 4. SRM chromatograms of a standard plasma sample containing 0.500 ng / ml (LLQ) of the analyte and 100 ng / ml of the internal standard: (A) analyte chromatogram; (B) internal standard chromatogram.
standard curve, four control blank plasma samples were placed to monitor the carry-over of the analyte or the internal standard by the robot’s four probes. No carry-over peaks were observed at the retention times and in the ion channels of either the analyte or the internal standard (Fig. 5). The absence of carry-over is due to the thorough washing of
202
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 5. SRM chromatograms of a blank plasma sample not spiked with the analyte or the internal standard: (A) analyte chromatogram; (B) internal standard chromatogram.
the probes with the system liquid at the wash-stations and the dipping of the probes into a wash solution away from the wash-station. The resealable septa may also be helping to reduce the carry-over by wiping off the outside of the probes as they retract.
3.5.2. Lower limit of quantitation Samples containing the analyte at 0.500 ng / ml, the lowest concentration in the standard curve, were used to assess the lower limit of quantitation (LLQ) for the analyte in plasma. Six different lots of control plasma were spiked at the LLQ concentration. The LLQ samples were processed and analyzed with a standard curve and QC samples and their predicted concentrations were determined. The deviations of the predicted concentrations from the nominal values were within 20% for all six LLQ samples. A typical SRM chromatogram at the LLQ concentration is shown in Fig. 4A. 3.5.3. Accuracy and precision The accuracy and precision of the method was determined by analyzing the QC samples at concentrations within the lower, the 2nd, and the upper quartile of the
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
203
Table 1 Accuracy and precision for I in human plasma from three independent runs. Nominal conc. (ng / ml)
Mean observed conc. (ng / ml)
% Dev.
Between run precision (% R.S.D.)
Within run precision (% R.S.D.)
n
1.50 100 200 2000
1.54 99.7 201 1974
2.56 20.28 0.58 21.38
5.81 5.14 2.11 2.55
3.23 3.64 3.86 5.42
15 15 15 15
standard curve. A fourth QC sample, with a concentration higher than the upper limit of the standard curve range, was also analyzed. This QC sample was diluted ten-fold with control plasma, processed and then analyzed. Five replicate samples at each concentration were analyzed in three separate batches. Calculated deviations of the predicted concentrations from their nominal values were used to assess the accuracy of the method. The calculated R.S.D. values were used to assess the repeatability (intra-assay precision) and reproducibility (inter-assay precision) of the method. In each run, the deviations of the predicted concentrations from their nominal values were within 15%. As shown in Table 1, the intra- and inter-assay precisions were each within 6% R.S.D. at all concentrations. The assay accuracy was within 3% of the nominal values.
4. Conclusions We have demonstrated the validity of aliquotting plasma samples directly from capped tubes for quantitative bioanalysis. Nevertheless, the method can be used for any biofluid. The direct transfer step can be incorporated into any automated sample preparation. Direct transfer of samples from capped tubes not only protects the analyst from biohazards and occupational injury but also saves the analyst’s time. Although in the present work the direct transfer from capped tubes was incorporated into LLE, capping and uncapping precedes any bioanalytical method. Thus any bioanalytical method can benefit from such a direct transfer of the samples from the sample tubes without removing the caps.
References [1] Jemal M, Xia Y-Q, Whigan DB. The use of high-flow high performance liquid chromatography coupled with positive and negative ion electrospray tandem mass spectrometry for quantitative bioanalysis via direct injection of the plasma / serum samples. Rapid Commun Mass Spectrom 1998;12:1389–99. [2] Jemal M, Ouyang Z, Xia Y-Q, Powell ML. A versatile system of high-flow high performance liquid chromatography with tandem mass spectrometry for rapid direct-injection analysis of plasma samples for quantitation of a b-lactam drug candidate and its open-ring biotransformation product. Rapid Commun Mass Spectrom 1999;13:1462–71. [3] Studt T. Solving the sample prep bottleneck. Drug Discovery and Development 1999;20–22.
204
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
[4] Jordan L. Sample preparation perspectives. LC-GC 1993;11:634–8. [5] Chen X-H, Franke J-P, Ensing K, Wijsbeek J, de Zeeuw RA. Pitfalls and solutions in the development of a fully automated solid-phase extraction method for drug screening purposes in plasma and whole blood. J Anal Toxico 1993;17:421–6. [6] Parker TD, Wright DS, Rossi DT. Design and evaluation of an automated solid-phase extraction method development system for use with biological fluids. Anal Chem 1996;68:2437–41. [7] Allanson JP, Biddlecombe RA, Jones AE, Pleasance S. The use of automated solid phase extraction in the ‘96 well’ format for high throughput bioanalysis using liquid chromatography coupled to tandem mass spectrometry. Rapid Commun Mass Spectrom 1996;10:811–6. [8] Plump RS, Gray RDM, Jones CM. Use of reduced sorbent bed and disk membrane solid-phase extraction for the analysis of pharmaceutical compounds in biological fluids, with applications in the 96-well format. J Chromatogr B 1997;694:123–33. [9] Janiszewski J, Schneider RP, Hoffmaster K, Swyden M, Wells D, Fouda H. Automated sample preparation using membrane microtiter extraction for bioanalytical mass spectrometry. Rapid Commun Mass Spectrom 1997;11:1033–7. [10] Simpson H, Berthemy A, Buhrman D, Burton R, Newton J, Kealy M, Wells D, Wu D. High throughput liquid chromatography / mass spectrometry bioanalysis using 96-well disk solid phase extraction plate for the sample preparation. Rapid Commun Mass Spectrom 1998;12:75–82. [11] Steinborner S, Henion J. Liquid–liquid extraction in the 96-well plate format with SRM LC / MS quantitative determination of methotrexate and its major metabolite in human plasma. Anal Chem 1999;71:2340–5.
An automated method of sample preparation of biofluids using pierceable caps to eliminate the uncapping of the sample tubes during sample transfer Deborah S. Teitz, Sanaullah Khan, Mark L. Powell, Mohammed Jemal* Bioanalytical Research, Metabolism and Pharmacokinetics, Bristol-Myers Squibb Pharmaceutical Research Institute, P.O. Box 191, New Brunswick, NJ 08903 -0191, USA Received 16 January 2000; received in revised form 15 May 2000; accepted 1 June 2000
Abstract Biological samples are normally collected and stored frozen in capped tubes until analysis. To obtain aliquots of biological samples for analysis, the sample tubes have to be thawed, uncapped, samples removed and then recapped for further storage. In this paper, we report an automated method of sample transfer devised to eliminate the uncapping and recapping process. This sampling method was incorporated into an automated liquid–liquid extraction procedure of plasma samples. Using a robotic system, the plasma samples were transferred directly from pierceable capped tubes into microtubes contained in a 96-position block. The aliquoted samples were extracted with methyl-tert-butyl ether in the same microtubes. The supernatant organic layers were transferred to a 96-well collection plate and evaporated to dryness. The dried extracts were reconstituted and injected from the same plate for analysis by liquid chromatography with tandem mass spectrometry. 2000 Elsevier Science B.V. All rights reserved. Keywords: Automated sample transfer; Pierceable caps; LC / MS / MS; Sample preparation; Liquid–liquid extraction
1. Introduction Technological advancements in combinatorial chemistry have revolutionized drug *Corresponding author. Tel.: 11-732-519-1582; fax: 11-732-519-1557. E-mail address: [email protected] (M. Jemal). 0165-022X / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 00 )00117-2
194
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
discovery and development in the pharmaceutical industry. Consequently, the workload of the scientists in the discovery and development areas has increased with the influx of new compounds as potential drug candidates. To keep up with the increasing demand of the industry, scientists in these areas are actively involved in the laboratory scale automation of bioanalytical methods. The complex nature of biological matrices requires that sample clean-up be an integral part of any bioanalytical method. The sample preparation, with its multistep clean-up procedures, is usually the rate-limiting step in achieving high-throughput bioanalysis. Fortunately, the steps involved in sample preparation can be automated to streamline uninterrupted analysis of biological samples. In our laboratory, we have utilized two approaches for automation of sample preparation for quantitation by liquid-chromatography (LC) with tandem mass spectrometry (MS / MS). The first approach involves the direct injection of the biological samples into an LC / MS / MS system for on-line extraction [1–3]. The sample preparation involves simply aliquotting the biological sample into a 96-well plate, adding the internal standard and centrifuging before the direct injection from the plate. The first two steps of this sample preparation were automated using a robotic liquid handling system to further reduce the hands-on time the analyst spends on sample preparation. The second approach is based on automating off-line solid-phase extraction (SPE) and liquid–liquid extraction (LLE) in a 96-well format using a robotic liquid handling system [3–11]. Both types of automation have drastically reduced not only the length of time required for sample extraction but also the hands-on time spent by the analyst. However, the task of uncapping and recapping the sample tubes still appeared to be the ‘‘bottleneck’’ in sample analysis. It took approximately 30 min to uncap and recap 96 sample tubes. Hence, there was a need to eliminate this task. Another impetus for eliminating the uncapping / recapping process was that it is potentially hazardous as it may expose the analyst to the biological samples via accidental spillage or breakage. Also the manual uncapping and recapping puts physical stress on the human hands that may contribute to occupational injuries such as tendonitis and carpal tunnel syndrome. Here, we report the development and validation of automated LLE for extraction of plasma samples aliquotted by a robotic liquid handling system from sample tubes screw-capped with pierceable caps. The extracted samples were analyzed by LC / MS / MS. The probes of the robot pierced through the caps and transferred aliquots of samples to 96-well formatted microtubes for extraction.
2. Materials and methods
2.1. Reagents The analyte (I) and the 2 H 5 -labeled internal standard (II), shown in Fig. 1, are products of Bristol-Myers Squibb Pharmaceutical Research Institute. High purity (18.2 MV) water was obtained by passing house deionized water through the Milli-Q system
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
195
Fig. 1. Chemical structures of the analyte (I) and the 2 H 5 -labeled internal standard (II).
(Millipore, Bedford, MA). Methyl-tert-butyl ether (MTBE) and acetonitrile were obtained from EM Science (Gibbstown, NJ).
2.2. Equipment A Packard MultiPROBE 204DT robotic system, equipped with the MPTable software (Packard Instrument company, Meriden, CT), was used for the liquid handling operations. Polypropylene sample tubes (56 3 16 mm) and their caps were purchased from Elkay Products (Shrewbury, MA). We modified these caps through Andwin Scientific (Canoga Park, CA) by cutting a portion of the top of each cap and retrofitting it with a non-leaking resealable polymer septum (Fig. 2A). The plasma samples for analysis were contained in the Elkay tubes that were screw-capped with the modified caps. Microtubes, which come interconnected in strips of eight (Fig. 2B), were used for extraction. The eight-microtube strips were placed in 12 rows in a 96-position block (Fig. 2C) to provide 96 extraction microtubes. Plug-caps, interconnected in a strip of eight, were used to seal the microtubes. The microtubes, 96-position block and the plug-cap strips were obtained from Qiagen (Santa Clarita, CA). A special 96-position sample rack was designed in-house and then custom-made by Packard Instrument Company to hold the capped plasma sample tubes (Fig. 2D). The sample rack had a removable 96-hole cover that could be screwed down over the capped plasma sample tubes to hold them in place. This was necessary to prevent the robot probes from lifting the tubes during retraction. Furthermore, the sample rack was screwed down onto the deck of the robot to prevent the entire rack from moving as the robot probes retracted. Square-well 96-well collection plates and sealing mats to cover the plates were obtained from Varian (Harbor City, CA). A Savant Speed-Vac (Savant Instruments, Holbrook, NY) was used for drying the samples by evaporation under vacuum. A Beckman (Columbia, MD) System Gold pump (Model 128) was used for LC solvent delivery. The LC column used was a 2 3 50 mm (5 mm) Hypersil C8 column from Keystone (Bellefonte, PA). A Hewlett-Packard (Palo Alto, CA) HP 1100 column heater was used to control the column temperature. A Gilson (Middleton, WI) 233 XL autoinjector was used for programmed randomized sample injection from a 96-well plate. A Finnigan TSQ-7000 triple quadrupole mass spectrometer (Finnigan MAT, San Jose, CA) equipped with an electrospray interface was used for detection.
196
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 2. (A) A plasma sample tube with a pierceable cap. (B) A set of eight interconnected microtubes. (C) A 96-position block for holding the microtubes. (D) A sample rack for holding the plasma sample tubes on the robot deck.
2.3. Standard and quality control ( QC) preparations Calibration standards were prepared by spiking the appropriate amounts of I into
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
197
human plasma. The calibration curve range was 0.500–250 ng / ml, consisting of ten concentrations, with each in duplicate. The QC samples were also prepared by spiking I into human plasma to obtain concentrations of 1.50, 100, 200 and 2000 ng / ml as the low, mid, high and dilution QC samples, respectively. The standards and QC samples used for analysis were contained in the Elkay tubes with pierceable caps.
2.4. Automated cap piercing As the first step of the robotic sample transfer and extraction method for plasma samples in tubes capped with pierceable caps, the robot probes were programmed to penetrate through the polymer septum of each cap. For this, each probe was programmed to pick-up 100 ml of air and inject only 25 ml into each tube after piercing through the septum. The probes were programmed to specifically dip into the sample tubes just far enough to pierce the septa without coming into contact with the plasma in the tubes. Thus, the probes pierced all the sample tubes without touching the plasma samples. Both the inside and outside of the probes were washed with the system liquid (25% methanol in water) after each piercing step.
2.5. Sample processing After completion of the piercing sequence, aliquots (0.25 ml) of each sample were transferred from the pierced sample tubes to a set of 96-well formatted microtubes. To avoid the potential for carryover at this stage, the robot probes were each dipped deeply into a wash solvent (25% methanol in water) to rinse the entire length of the probes, followed by a thorough wash (inside and outside) with the system liquid after each sample transfer. Internal standard solution (25 ml), 0.5 N HCl solution (0.1 ml) and MTBE (0.5 ml) were then added sequentially by the robot to each microtube. The microtubes were manually sealed with the plug-cap strips and vortex-mixed for 5 min. To break an emulsion that occasionally occurred, the 96-position block containing the microtubes was immersed in an acetone / dry ice bath to freeze the contents of each microtube, sonicated for 3 min and finally centrifuged at 3000 g for 5 min. The organic layer from each microtube was transferred by the robot to a square-well 96-well plate. The plate was then placed in a Savant Speed-Vac for drying by evaporation. A 75 ml volume of the reconstitution solution (1.0 mM formic acid solution in 2:1 water– acetonitrile solvent mixture) was then added to each well by the robot. The 96-well plate was then covered with a sealing mat and transferred to the autoinjector for injection from the plate.
2.6. LC /MS /MS Electrospray ionization mass spectrometry was used in the negative ion mode. The electrospray voltage was 4.5 kV and the capillary temperature was maintained at 2508C. High purity (99.999%) nitrogen was used both as the nebulizing and the auxiliary gas. The conditions for Q1 were optimized to obtain maximum response for the [M–H] 2 precursor ion of the analyte. Argon, used as the collision gas, was set at 2.5 m Torr and
198
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
the collision offset energy (C off ) was optimized to obtain the maximum response of the product ion used for the selected reaction monitoring (SRM). The C off value of 17 eV was typically used for both I and II. The SRM transitions monitored were m /z 463→m /z 377 for I and m /z 468→m /z 382 for II. Unit mass resolution was maintained at both Q1 and Q3. The peak area ratio of the analyte (I) to the internal standard (II) was calculated at each concentration. The data were fitted to 1 /x-weighted quadratic regression model. LC separation was achieved isocratically on a Hypersil 5032 mm C8 column at a flow rate of 0.30 ml / min and a column temperature of 408C. The mobile phase consisted of 60% A and 40% B, where A was a 1.0 mM formic acid solution in 5% acetonitrile in water and B was a 1.0 mM formic acid solution in 95% acetonitrile in water. The effluent was all directed to the mass spectrometer.
3. Results and discussion
3.1. Automated cap piercing and direct sample transfer A significant part of the sample preparation time is normally consumed by repeated uncapping and recapping of biological sample tubes regardless of the sample preparation method. Yet, for decades, in bioanalytical laboratories, LC and GC autosamplers have been used to pierce through caps fitted with pierceable septa and pick up samples for injection. We saw no reason not to apply this type of technology to biological sample tubes. This prompted us to modify the commercially available tube caps by retrofitting them with pierceable septa. We then wanted to use the robot probes of a liquid handling system to pierce the septa of the capped sample tubes. Several problems were encountered. In initial attempts to pierce through the septa the robot was unreliable, giving us frequent z-motor errors. We quickly found out that when a septum is pierced for the first time, it presents the most resistance for the z-motor. Another important factor contributing to a high number of failures was a software shortcoming. In case of an initial penetration failure, the software allows multiple piercing attempts during a ‘‘dispense’’ operation whereas, an error is generated if any of the four probes fails to penetrate a septum during a ‘‘pick-up’’ operation. By incorporating a separate cap-piercing step while ‘‘dispensing’’ air into the tubes, we were in effect letting the robot use multiple attempts to pierce the septa without failing during the sample transfer from the capped tubes. Furthermore, injection of air into each tube also served to partially compensate for the vacuum that could result from pipetting the sample from a closed tube. Between pipetting each sample, we programmed the robot to dip into a wash solution. This was necessary because with the robot piercing through a cap, a large length of probe could come in contact with any sample stuck to the cap septum. The robot wash station, normally used to wash the probes between pipetting steps, is designed to wash only the tip of the probe. Therefore, the extra dip wash was included to clean off what the wash solution could not reach.
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
199
3.2. Resealability of the pierceable caps Before we used the pierceable caps for our study samples, we wanted to be sure that they were leak-proof and would remain leak-proof after piercing. We tested the effectiveness of sealing and resealing of the septa as follows: The caps were screwed onto the Elkay tubes and then submerged into an aqueous dye solution for 5 min. The insides of the caps were then checked for any seepage of the dye around the septa. No failure of the seals was detected. The septa were then pierced with the robot probes and the dye test was repeated. The septa remained leak-proof even after five piercing steps by the robot probes. The same dye test was done with pierceable caps that were stored for 3 months at 2708C to simulate sample storage conditions. No failure of the seal was seen. Similarly, the caps were tested for resealability after freezing and thawing. For this purpose, the tubes were pierced and frozen five separate times and then tested for leakage with the above dye test. The septa proved to be leak-proof under this condition as well, giving us the confidence to use this technology for our study samples.
3.3. LLE of plasma samples Plasma samples were extracted into MTBE after acidification with 0.1 N HCl. The robot-assisted liquid handling operations of the LLE significantly reduced both the analyst’s time and the overall time of the sample preparation. In early investigations, 96-square-well plates, 96-round-well plates and 96-well formatted microtubes (described above) were tested for extraction. The use of the plates or the microtubes, in lieu of the conventionally used individual tubes, eliminated the need for labeling of the individual tubes. However, with the 96-well plates there was always crossover of the extraction solvents between the wells, even if the wells were only partially filled and sealed off with the sealing mat. The microtubes were finally selected for performing the extraction because the crossover of the extract from one tube to another during vortex-mixing did not occur, probably because the microtubes are interconnected in rows of eight with only one point of attachment to the adjacent microtube (Fig. 2B) and the plug-caps did a better job of sealing each microtube. In addition, the volume of each microtube was sufficient to accommodate the plasma sample and the extraction solvents as the inside volume of the microtube was 1.2 ml and the accumulated extraction volume in the microtube was 0.875 ml. The extraction recovery of I from plasma was 45%.
3.4. LC /MS /MS The electrospray negative ion MS and MS / MS product ion spectra of I are shown in Fig. 3. The MS spectrum is dominated by the [M–H] 2 ion at m /z 463. The MS / MS product ion spectrum of the [M–H] 2 ion produced the major product ion at m /z 377. Thus, the SRM used was m /z 463→m /z 377 for the analyte and m /z 468→m /z 382 for the internal standard. The retention time of the analyte and the internal standard was 1.7 min (Fig. 4) and the chromatographic run time was 2.25 min.
200
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 3. (A) Electrospray negative ion MS spectrum of I. (B) Electrospray negative ion MS / MS product ion spectrum of the [M–H] 2 ion of I.
3.5. Method validation 3.5.1. Specificity Six different lots of control human plasma were analyzed with and without fortification with the internal standard in order to determine whether any endogenous plasma constituents interfered with the analyte or the internal standard. The degree of interference was assessed by inspection of the SRM chromatograms. No interfering peaks from the plasma were found at the retention time and in the ion channel of either the analyte or the internal standard. Following the high concentration point of the
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
201
Fig. 4. SRM chromatograms of a standard plasma sample containing 0.500 ng / ml (LLQ) of the analyte and 100 ng / ml of the internal standard: (A) analyte chromatogram; (B) internal standard chromatogram.
standard curve, four control blank plasma samples were placed to monitor the carry-over of the analyte or the internal standard by the robot’s four probes. No carry-over peaks were observed at the retention times and in the ion channels of either the analyte or the internal standard (Fig. 5). The absence of carry-over is due to the thorough washing of
202
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
Fig. 5. SRM chromatograms of a blank plasma sample not spiked with the analyte or the internal standard: (A) analyte chromatogram; (B) internal standard chromatogram.
the probes with the system liquid at the wash-stations and the dipping of the probes into a wash solution away from the wash-station. The resealable septa may also be helping to reduce the carry-over by wiping off the outside of the probes as they retract.
3.5.2. Lower limit of quantitation Samples containing the analyte at 0.500 ng / ml, the lowest concentration in the standard curve, were used to assess the lower limit of quantitation (LLQ) for the analyte in plasma. Six different lots of control plasma were spiked at the LLQ concentration. The LLQ samples were processed and analyzed with a standard curve and QC samples and their predicted concentrations were determined. The deviations of the predicted concentrations from the nominal values were within 20% for all six LLQ samples. A typical SRM chromatogram at the LLQ concentration is shown in Fig. 4A. 3.5.3. Accuracy and precision The accuracy and precision of the method was determined by analyzing the QC samples at concentrations within the lower, the 2nd, and the upper quartile of the
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
203
Table 1 Accuracy and precision for I in human plasma from three independent runs. Nominal conc. (ng / ml)
Mean observed conc. (ng / ml)
% Dev.
Between run precision (% R.S.D.)
Within run precision (% R.S.D.)
n
1.50 100 200 2000
1.54 99.7 201 1974
2.56 20.28 0.58 21.38
5.81 5.14 2.11 2.55
3.23 3.64 3.86 5.42
15 15 15 15
standard curve. A fourth QC sample, with a concentration higher than the upper limit of the standard curve range, was also analyzed. This QC sample was diluted ten-fold with control plasma, processed and then analyzed. Five replicate samples at each concentration were analyzed in three separate batches. Calculated deviations of the predicted concentrations from their nominal values were used to assess the accuracy of the method. The calculated R.S.D. values were used to assess the repeatability (intra-assay precision) and reproducibility (inter-assay precision) of the method. In each run, the deviations of the predicted concentrations from their nominal values were within 15%. As shown in Table 1, the intra- and inter-assay precisions were each within 6% R.S.D. at all concentrations. The assay accuracy was within 3% of the nominal values.
4. Conclusions We have demonstrated the validity of aliquotting plasma samples directly from capped tubes for quantitative bioanalysis. Nevertheless, the method can be used for any biofluid. The direct transfer step can be incorporated into any automated sample preparation. Direct transfer of samples from capped tubes not only protects the analyst from biohazards and occupational injury but also saves the analyst’s time. Although in the present work the direct transfer from capped tubes was incorporated into LLE, capping and uncapping precedes any bioanalytical method. Thus any bioanalytical method can benefit from such a direct transfer of the samples from the sample tubes without removing the caps.
References [1] Jemal M, Xia Y-Q, Whigan DB. The use of high-flow high performance liquid chromatography coupled with positive and negative ion electrospray tandem mass spectrometry for quantitative bioanalysis via direct injection of the plasma / serum samples. Rapid Commun Mass Spectrom 1998;12:1389–99. [2] Jemal M, Ouyang Z, Xia Y-Q, Powell ML. A versatile system of high-flow high performance liquid chromatography with tandem mass spectrometry for rapid direct-injection analysis of plasma samples for quantitation of a b-lactam drug candidate and its open-ring biotransformation product. Rapid Commun Mass Spectrom 1999;13:1462–71. [3] Studt T. Solving the sample prep bottleneck. Drug Discovery and Development 1999;20–22.
204
D.S. Teitz et al. / J. Biochem. Biophys. Methods 45 (2000) 193 – 204
[4] Jordan L. Sample preparation perspectives. LC-GC 1993;11:634–8. [5] Chen X-H, Franke J-P, Ensing K, Wijsbeek J, de Zeeuw RA. Pitfalls and solutions in the development of a fully automated solid-phase extraction method for drug screening purposes in plasma and whole blood. J Anal Toxico 1993;17:421–6. [6] Parker TD, Wright DS, Rossi DT. Design and evaluation of an automated solid-phase extraction method development system for use with biological fluids. Anal Chem 1996;68:2437–41. [7] Allanson JP, Biddlecombe RA, Jones AE, Pleasance S. The use of automated solid phase extraction in the ‘96 well’ format for high throughput bioanalysis using liquid chromatography coupled to tandem mass spectrometry. Rapid Commun Mass Spectrom 1996;10:811–6. [8] Plump RS, Gray RDM, Jones CM. Use of reduced sorbent bed and disk membrane solid-phase extraction for the analysis of pharmaceutical compounds in biological fluids, with applications in the 96-well format. J Chromatogr B 1997;694:123–33. [9] Janiszewski J, Schneider RP, Hoffmaster K, Swyden M, Wells D, Fouda H. Automated sample preparation using membrane microtiter extraction for bioanalytical mass spectrometry. Rapid Commun Mass Spectrom 1997;11:1033–7. [10] Simpson H, Berthemy A, Buhrman D, Burton R, Newton J, Kealy M, Wells D, Wu D. High throughput liquid chromatography / mass spectrometry bioanalysis using 96-well disk solid phase extraction plate for the sample preparation. Rapid Commun Mass Spectrom 1998;12:75–82. [11] Steinborner S, Henion J. Liquid–liquid extraction in the 96-well plate format with SRM LC / MS quantitative determination of methotrexate and its major metabolite in human plasma. Anal Chem 1999;71:2340–5.