Quantitation of Phosphorothioate Oligonucleotides in Human Blood

Martin Maier, Hans Fritz, Michael Gerster, Jens Schewitz, and Ernst Bayer* ... and Peptide Chemisty, University of Tübingen, Auf der Morgenstelle 18,...
2 downloads 0 Views 114KB Size
AC Research

Anal. Chem. 1998, 70, 2197-2204

Accelerated Articles

Quantitation of Phosphorothioate Oligonucleotides in Human Blood Plasma Using a Nanoparticle-Based Method for Solid-Phase Extraction Martin Maier, Hans Fritz, Michael Gerster, Jens Schewitz, and Ernst Bayer*

Research Center for Nucleic Acid and Peptide Chemisty, University of Tu¨ bingen, Auf der Morgenstelle 18, D-72076 Tu¨ bingen, Germany

Based on the application of cationic polystyrene nanoparticles, a novel method for solid-phase extraction of phosphorothioate oligonucleotides from human plasma has been developed. A high binding affinity, which is required for an effective isolation out of complex mixtures, is mediated by hydrophobic and multiple electrostatic interactions between the oligonucleotides and the nanoparticles. The principle of the method is based on a pHcontrolled adsorption/desorption mechanism. Analysis of the extracted samples was performed by capillary gel electrophoresis. Extraction conditions were optimized, providing the isolation of oligonucleotides (g10 nucleotide units) in high yields and purity even at concentrations in the low-nanomolar range (down to 5 nM). The low salt contamination of the samples allows their direct analysis by electrospray mass spectrometry. The combined linearity and accuracy of the assay together with absolute recovery rates in the range of 60-90% indicate that the developed solid-phase extraction method is generally applicable to quantitation of oligonucleotides in human plasma. Further improvement was achieved with an optimized carrier system of 2-fold enlarged particles which reduces the time consumption of the extraction procedure to ∼30 min.

In recent years, the antisense technology has grown to a promising field of pharmaceutical research. The approach is based on the application of modified single-stranded oligonucle* Corresponding author: (phone) +49 7071 29-72437; (fax) +49 7071 29-5034; (e-mail) [email protected]. S0003-2700(98)00097-3 CCC: $15.00 Published on Web 05/05/1998

© 1998 American Chemical Society

otides (ODNs) of a commonly used length ranging from 12 to 25 nucleotide units as a new class of therapeutic agents. These antisense compounds are developed to efficiently inhibit expression of a harmful gene by specific binding to a complementary target sequence located on the mRNA.1-3 Phosphorothioate oligonucleotides (PTOs) with oxygen replaced by sulfur at a nonbridging position of the phosphodiester linkage represent one of the most commonly used derivatives of natural DNA.4,5 A growing number of phosphorothioate oligonucleotides are undergoing preclinical investigations and clinical trials for their application as human therapeutics.6-8 Reliable analytical techniques are required in order to obtain data on the pharmacokinetics and metabolism of the administered antisense compounds and for their quantitation in pharmaceutical formulations. In a number of previously published investigations, capillary gel electrophoresis (CGE) has successfully been used for the analysis of synthetic oligonucleotides.9-11 More recently, (1) Cohen, J. S., Ed. Oligodeoxynucleotides: Antisense Inhibitors of Gene Expression; CRC Press: Boca Raton, FL, 1989; pp 1-6. (2) Stein, C. A.; Cohen, J. S. Cancer Res. 1988, 48, 2659-68. (3) Stein, C. A.; Cheng, Y.-C. Science 1993, 261, 1004-12. (4) Eckstein, F. Angew. Chem. 1983, 95, 431-47; Angew. Chem., Int. Ed. Eng. 22, 423-39. (5) Eckstein, F. Annu. Rev. Biochem. 1985, 54, 367-402. (6) Cowsert, L. M.; Fox, M. C.; Zon, G.; Mirabelli, C. K. Antimicrob. Agents Chemother. 1993, 37, 171-7. (7) Agrawal, S.; Tang, J. Y. Antisense Res. Dev. 1992, 2, 261-6. (8) Bayever, E.; Iversen, P. L.; Bishop, M. R.; Sharp, J. G.; Tweary, H. K.; Arneson, M. A.; Pirrucello, S. J.; Ruddon, R. W.; Kessinger, A.; Zon, G.; Armitage, J. O. Antisense Res. Dev. 1993, 3, 383-90. (9) DeDionisio, L. J. Chromatogr., A 1993, 652, 101-8. (10) Cohen, A. S.; Vilenchik, M.; Dudley, J. L.; Gemborys, M. W.; Bourque, A. J. J. Chromatogr. 1993, 638, 293-301. (11) Bruin, G. J. M.; Bo¨rnsen, K. O.; Hu ¨ sken, D.; Gassmann, E.; Widmer, H. M.; Paulus, A. J. Chromatogr., A 1995, 709, 181-95.

Analytical Chemistry, Vol. 70, No. 11, June 1, 1998 2197

Srivatsa et al.12 demonstrated that CGE fulfills the requirements of linearity, accuracy, precision, and ruggedness for routine analysis of phosphorothioate oligonucleotides in pharmaceutical formulations. Moreover, modern mass spectrometrical techniques such as electrospray mass spectrometry (ES-MS) are frequently used to complement analytical information about metabolic products and cleavage reactions.13,14 However, stringent requirements are placed on the sample matrix since CGE and ES-MS are both sensitive to ionic contaminants present in the sample which can cause a dramatic reduction in the observed detector response.15 Thus, for quantitative analysis of oligonucleotides in a given biological matrix, the influence of a great variety of other compounds present in the medium, such as salts, fats, fatty acids, proteins, etc., has to be excluded. Similarly, an effective isolation of oligonucleotides from complex buffer systems, such as PCR media is required to enable their characterization by sensitive analytical techniques. In a previously published work from Bourque and Cohen,16 the application of anion-exchange chromatography for quantitative analysis of phosphorothioate oligonucleotides in serum and urine was described. Prior to analysis, the relatively laborious procedure with a detection limit of 200 ppb (∼25 nM) utilizes protein digestion in combination with several extraction steps. In an improved approach, extraction of PTOs was performed by anionexchange chromatography.17 The high salt contamination, which is a result of the extraction process, requires an additional desalting step of the samples by membrane dialysis in order to allow their analysis by CGE. Similarly, Leeds et al.18 developed a two-step solid-phase extraction method for the isolation of phosphorothioate oligonucleotides from human plasma. The method involves strong anion-exchange chromatography followed by reversed-phase chromatography for sample desalting. The latter step was found to be necessary to reduce the high salt concentration used for elution from the ion-exchange resins before a final desalting step by membrane dialysis can be performed. This method was shown to be reproducible and reliable but is, however, quite timeconsuming and suffers from relatively low absolute recovery rates of ∼40%. For both techniques utilizing membrane dialysis, quantitation is restricted to oligonucleotides with a length of more than 18 nucleotide units, as shorter oligomers cannot completely be recovered by membrane desalting.11 In recent studies of Fritz et al.,19 a novel model drug delivery system for oligonucleotides was developed. It was demonstrated that natural and modified oligonucleotides exhibit a strong affinity to cationic polystyrene nanoparticles mediated by the combination of hydrophobic and multiple electrostatic interactions. The properties of the nanoparticles were utilized to develop a novel (12) Srivatsa, G. S.; Batt, M.; Schuette, J.; Carlson, R. H.; Fitchett, J.; Lee, C.; Cole, D. L. J. Chromatogr., A 1994, 680, 469-77. (13) Maier, M.; Bleicher, K.; Kalthoff, H.; Bayer, E. Biomed. Pept., Proteins Nucleic Acids 1995, 1, 235-41. (14) Griffey, R. H.; Greig, M. J.; Gaus, H.-J.; Liu, K.; Monteith, D.; Winniman, M.; Cummins, L. L. J. Mass Spectrom. 1997, 32, 305-13. (15) Demorest, D.; Dubrow, R. J. Chromatogr. 1991, 559, 43-56. (16) Bourque, A. J.; Cohen, A. S. J. Chromatogr., B 1993, 617, 43-9. (17) Bourque, A. J.; Cohen, A. S. J. Chromatogr., B 1994, 662, 343-9. (18) Leeds, J. M.; Graham, M. J.; Truong, L.; Cummins, L. L. Anal. Biochem. 1996, 235, 36-43. (19) Fritz, H.; Maier, M.; Bayer, E. J. Colloid Interface Sci. 1997, 195, 272-88.

2198 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

method for solid-phase extraction.20 In the present study, the nanoparticle-based extraction method was applied for the isolation and quantitation of oligonucleotides from human plasma. In addition to the requirements mentioned above, the extraction method should provide the possibility to quantify shorter degradation products of the oligonucleotides investigated in order to make detailed pharmacokinetic and metabolism studies possible. Extraction conditions were optimized to achieve a high recovery from the biological matrix, which is required for the detection and analysis of oligonucleotides even at low-nanomolar concentrations. The linearity, accuracy, and reproducibility of the assay was investigated. In addition, the carrier system was further developed to reduce the time consumption of the extraction process. EXPERIMENTAL SECTION Materials. Chemicals used for solid-phase synthesis of oligonucleotides were purchased from Perseptive Biosystems (Wiesbaden, Germany) and Applied Biosystems (Weiterstadt, Germany). Silica capillaries were obtained from Polymicro Technology (Phoenix, AZ). Chemicals used for capillary gel electrophoresis were purchased from Fluka (Buchs, Switzerland) and Merck (Darmstadt, Germany) in the highest grade available. Human blood plasma from healthy individuals stabilized with citrate, EDTA (potassium salt), and heparin (ammonium salt) was a gift from the Medical Clinic of the University of Tu¨bingen. The plasma samples were stored at -20 °C and thawed quickly prior to use. Purified water was provided by a MilliQ185 Plus water purification system (Millipore). Oligonucleotide Synthesis and Purification. Oligodeoxyribonucleotides were prepared by standard phosphoramidite chemistry on a 394 DNA/RNA synthesizer (Applied Biosystems, Perkin-Elmer Corp., Foster City, CA) using TentaGel (Rapp Polymere, Tu¨bingen, Germany) or controlled-pore glass (Perseptive Biosystems, Wiesbaden, Germany) as solid supports.21 For synthesis of phosphorothioate oligonucleotides, tetraethylthiuram disulfide (TETD, Applied Biosystems) was used as the sulfurizing reagent. ODNs were purified by reversed-phase HPLC using a solvent delivery system S 1000, a low-pressure gradient mixer S 8110 (Sykam, Gilching, Germany), and an UV/visible Spectrometer UVIS 205 (Linear, Reno, NV). Columns (250 × 4.6 mm and 250 × 25 mm) packed with Nucleosil C18, 5 µm (Grom, Herrenberg, Germany) were used for analytical and preparative separations. The concentration of the oligonucleotides was determined spectrophotometrically. The sequence and modification of the oligonucleotides used in this work are summarized in Table 1. Preparation and Characterization of the Nanoparticles. Cationic polystyrene nanoparticles were prepared, purified, and stabilized according to a previously published procedure.19 The shape, size, and size distribution of the nanospheres were determined by scanning electron microscopy (SEM) using a Cambridge Stereoscan 250 MKR electron microscope (Cambridge Instruments, Cambridge, U.K.) and by photon correlation spectroscopy (PCS) using a photocorrelation spectrometer N4 Plus (Coulter Instruments, Krefeld, Germany). Surface charge, ad(20) Bayer, E.; Fritz, H.; Schewitz, J.; Gerster, M.; Maier, M. German Patent Appl. 1973636600, 1997. (21) Bayer, E.; Bleicher, K.; Maier, M. Z. Naturforsch. 1995, 50B, 1096-100.

Table 1. Sequence and Modification of the Oligonucleotides Used name

sequence 5′ f 3′

modifa

ODN-1 ODN-2 ODN-3 ODN-4 ODN-5 ODN-6 T25 T30

TTT TTT TTT TTT TTT TTT TT CTA TTA ACA ACA CAC AAC AG TTC TTG TCT GCT CTT TTT TTT TTT TTT TTT TTT TTT TTT T TTT TT dO(T)25 dO(T)30

PS PS PS PS PS PS PO PO

a

PS, phosphorothioate backbone; PO, phosphodiester backbone.

sorption/desorption characteristics, and loading capacity of the nanoparticles were determined as described in ref 19. Procedure of Solid-Phase Extraction. For conjugate formation, a known amount of phosphorothioate oligonucleotide was added to an aliquot of human plasma (200-400 µL), and the resultant mixture was placed in a 1.5 mL conical tube and diluted with 800 µL of 50 mM Tris-HCl (pH 9) in deionized water. The tube was briefly vortexed prior to the addition of 200 µL of the nanoparticle suspension (solid content, ∼10 mg/mL) and subsequently vortexed again. After 5-10 min of incubation, the suspension was centrifuged (24000g, 10-15 °C) and the supernatant was carefully removed up to a residual volume of 10-20 µL. Supported by brief ultrasonification and vortexing, the particles were resuspended in 1 mL of a freshly prepared solution of 0.5 M acetic acid in deionized water/ethanol (1:1) and subsequently separated from the washing solution by centrifugation. After the supernatant was removed, the particles were resuspended in 1 mL of deionized water and separated by another centrifugation step. Finally, 200 µL of a solution of 150 µM SDS in aqueous ammonia (25%)/acetonitrile (60:40) was added to the nanoparticle-oligonucleotide conjugates, and the released oligonucleotides were separated from the carrier by centrifugation. To exclude a contamination of the samples with residual particles, the supernatant was placed in another 1.5-mL tube and centrifuged again. Subsequently, the samples were dried by rotoevaporation or lyophilization and stored at -20 °C until before analysis. For ES-MS analysis, oligonucleotide release was induced by the addition of 200 µL of 25% aqueous ammonia to the conjugates. The oligonucleotides were separated from the carrier and prepared for analysis as described above. Capillary Gel Electrophoresis and Electrospray Mass Spectrometry. Capillary gel electrophoresis was performed using three different CE systems, a modular Grom CE System in combination with the Sykam Axiom Software, a HP3D capillary electrophoresis system from Hewlett-Packard including the HP3D ChemStation for data analysis, and a Beckman P/ACE system MDQ in combination with the Beckman P/ACE System MDQ Software. Separations were carried out using either fixed PAA gel capillaries (10% T), prepared as described elsewhere,18 or coated capillaries in combination with entangled polymer solutions. In the case of the fixed gel capillaries, analysis was performed in 100 mM Tris/100 mM borate/8.3 M urea buffer at an applied voltage of -12.5 kV (A). The oligonucleotide analysis kit contain-

ing PVA-coated capillaries, polymer solution B, and oligonucleotide buffer was obtained from Hewlett-Packard. Using this kit, measurements were made at an applied voltage of -25 kV and a column temperature of 40 °C (B). The polymer gel was replaced after every run. In the case of PAA-coated capillaries that were prepared as described previously,22 the entangled polymer gel eCAP ssDNA 100-R Gel together with the ssDNA 100 buffer (Trisborate, urea) from Beckman were applied. Analysis was performed at an applied voltage of -12.5 kV, and the column temperature was maintained at 30 °C (C). The polymer gel was replaced after every fifth run. The samples were injected electrokinetically at a reversed injection voltage of 5-12.5 kV for 2040 s depending on the sample concentration. The total capillary length was adjusted to the CGE system used. The effective separation length was held constant at 0.24 m (detection, 260 nm). Electrospray mass spectrometry was performed using an API III TAGA 6000E triple-quadrupole mass spectrometer (Sciex, Perkin-Elmer Corp., Toronto, Canada). The isolated samples were dissolved in deionized water/acetonitrile (1:1) and measured in a mass/charge range of 400-2000 (negative mode). Quantitation of the Analytical Data. Quantitative capillary gel electrophoresis was performed according to a work of Srivatsa et al.12 The amount of phosphorothioate oligonucleotides in the samples (nPTO) is calculated by the following equation:

Std APTO/TPTO nPTO ) nStd PTO AStd/TStd

(1)

where nStd is the amount of standard oligonucleotide added to the sample, Std and PTO are the molar extinction coefficients, and AStd/TStd and APTO/TPTO are the corrected peak areas (quotient of peak area and migration time) of the standard and the analyte, respectively. RESULTS AND DISCUSSION Nanoparticle-Based Extraction Procedure. Solid-phase extraction of oligonucleotides from human plasma was carried out using cationic polystyrene nanoparticles with a mean particle diameter of ∼180 nm. Covalently bound charged end groups located on the particle surface induce a positive surface charge density of 5.4 µC/cm2 and mediate oligonucleotide adsorption via multiple electrostatic interactions.19 The extraction procedure, which is illustrated schematically in Figure 1, can be divided into three major steps: conjugate formation, washing steps, and oligonucleotide release. The conditions were optimized in order to achieve the isolation of PTOs from human plasma in high yields and purity. In the step of conjugate formation, both acidic conditions and the presence of organic solvents, such as acetonitrile or ethanol, were observed to lead to a denaturation of plasma proteins. As a consequence, the oligonucleotides present in the sample could be encapsulated in the precipitated proteins resulting in poor recoveries. Furthermore, the isolated samples were found to contain considerable amounts of plasma contaminants and are thus not appropriate for analysis. To avoid protein precipitation and to reduce the interactions between plasma proteins and oligo(22) Hjerte´n, S. J. Chromatogr. 1985, 347, 191-8.

Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2199

Figure 1. Scheme of the extraction procedure.

nucleotides, the formation of nanoparticle-oligonucleotide conjugates was performed under nondenaturating conditions using a Tris-HCl buffer (pH 9). In contrast to chromatographic extraction techniques, such as anion-exchange chromatography, the separation from coextracted impurities is achieved by simple washing steps. Strikingly, the high attractive forces between nanoparticles and oligonucleotides allow the purification of the conjugates even for the isolation from complex mixtures such as human plasma. The fact that almost all coadsorbed plasma compounds can be removed by the washing procedure can be explained by different pH-dependent binding affinities of plasma compounds and oligonucleotides to the particle surface. However, using this procedure, the coisolation of polyanionic compounds with pH-dependent binding properties similar to oligonucleotides cannot be excluded. Washing conditions were adapted in order to achieve a high sample purity together with high recovery yields. The results indicate that a combination of two purification steps, an initial one using a mixture of acetic acid in H2O/ethanol followed by a second one with purified water, is most favorable for the extraction from human plasma. While the first step appears to be necessary for the removal of residual plasma contaminants, the second is required to desalt the sample. Oligonucleotide release is induced by a pH-controlled mechanism which is based on the presence of pH-sensitive cationic groups on the particle surface. Under strong alkaline conditions, these positively charged groups are transformed to their deprotonated form and the attractive forces between oligonucleotides and nanoparticles are significantly diminished. This leads to an almost complete desorption at pH 12-13. However, in the case of low oligonucleotide concentrations in the samples, it is neces2200 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

sary to diminish not only the electrostatic but also the hydrophobic interactions in order to achieve a maximum recovery. Thus, the release of ODNs is accomplished by the use of a mixture of ammonia and acetonitrile/SDS. Linearity, Accuracy, and Reproducibility of the Assay. In a first set of experiments, the developed assay was investigated for linearity, which is one of the basic requirements for quantitative analysis. As the procedure of solid-phase extraction is carried out prior to analysis, it is crucial that the analyte and the internal standard are isolated equivalently from the plasma samples. If not further specified, the human plasma used in these studies was stabilized with citrate. Two oligonucleotides, a phosphorothioate 15-mer (ODN-3) and a phosphodiester 30-mer (T30), were used simultaneously as internal standards for the quantitation of ODN-1 in human plasma. Premixed samples of the oligonucleotides were added to the plasma, such that the final concentration of the analyte varied between 10 and 500 nM (absolute amounts, 2-100 pmol) whereas the concentration of the internal standards was held constant at 50 nM (absolute amount, 10 pmol). Each plasma sample was extracted triplicately and analyzed by CGE. The amount of extracted ODN-1 was calculated by eq 1. The values were corrected by the corresponding values of the initial amount of ODN-1 (response factors) which were obtained after three injections of an aliquot of the oligonucleotide mixtures not added to the plasma. The standard curves with either ODN-3 or T30 as the internal standard are plotted as the extracted versus the initial amount of ODN-1 (Figure 2). Both standard curves exhibit high linearity in the investigated concentration range with squares of the correlation coefficients (r2) of 0.9991 and 0.9972. These results demonstrate that the analyte and the standard oligonucleotides are extracted proportionally regardless of their different length and backbone modifications (PS and PO). To ensure that the sequence of the oligonucleotide has no considerable influence on the extraction efficiency, another standard calibration curve was generated using a PTO of mixed sequence (ODN-2) together with T30 as the internal standard. The samples were prepared according to the procedure described above. In Figure 3, the standard curve as the amount of extracted ODN-2 versus its initial amount is depicted. The square of the correlation coefficient of 0.9976 obtained after linear regression demonstrates the linearity of the assay and confirms that both oligonucleotides, ODN-2 and T30, are extracted proportionally. Comparing these results with the data obtained with ODN-1, there is no evidence that the oligonucleotide sequence has a considerable influence on the recovery rates. The reproducibility and accuracy of the assay was addressed by the extraction and quantitation of plasma samples containing a mixture of ODN-1 and the standard oligonucleotides ODN-3 and T30 (Table 2). The initial amount of ODN-1 was determined by triplicate analysis of aliquots of the premixed samples not added to the plasma. In the case of ODN-3 as the internal standard, the mean values of extracted ODN-1 are 104.4 and 90.9% of the initial value for the concentrations of 50 and 100 nM, respectively. Based on T30 as the internal standard, mean values of 92.5 (50 nM) and 96.0% (100 nM) were obtained. The values of extracted ODN-1 are all in the range within 10% of the initial values, demonstrating

Figure 3. Linearity of the assay. Standard calibration curves of the amount of ODN-2 extracted from human plasma versus its initial amount (initial concentrations, 10, 25, 50, 100, 250, and 500 nM). T30 was used as the internal standard at a concentration of 50 nM. The data points shown are mean values ( SD of triplicate determinations. The square of the correlation coefficients (r) is 0.9988.

Figure 2. Linearity of the assay. Standard calibration curves of the amount of ODN-1 extracted from human plasma versus its initial amount (initial concentrations, 10, 25, 50, 100, 250, and 500 nM; internal standard concentration, 50 nM) using (a) ODN-3 and (b) T30 as the internal standards. The data points shown are mean values ( SD of triplicate determinations. Linear regression gave correlation coefficients (r) of 0.9996 (a) and of 0.9986 (b).

that the ratio of ODN-1 to internal standard remains unchanged after the extraction. Determination of Absolute Recovery Rates. The overall yields of PTOs recovered from human plasma were determined after triplicate extraction of plasma samples containing either ODN-1 or ODN-2 at four different concentrations (100, 25, 10, and 5 nM). For quantitation, T30 was added to the samples as an external standard after the extraction procedure. The absolute amounts of extracted PTOs were calculated by eq 1. The mean values of the recovery rates range from 60.4 to 76.7% for ODN-1 and from 61.9 to 86.8% for ODN-2 (Table 3). These results demonstrate that both PTOs are recovered in high yields even at low initial concentrations (down to 5 nM). In Figure 4a and b, typical electropherograms of ODN-2 extracted from human plasma at initial concentrations of 100 and 5 nM, respectively, are depicted. In the first case, only a few signals of impurities are visible at low migration times (Figure 4a). The number and intensity of the detected impurities in relation to the ODN signals increase at lower concentrations (Figure 4b). Their electrophoretic behavior and the fact that they are not removed by the washing steps imply that these contaminants have a polyanionic character. However, the signals of

ODN-2 and the standard T30 could be detected unequivocally, indicating that the analysis is not affected by contaminants coisolated from the plasma. To investigate the reproducibility of the assay by considering the absolute values of recovery, six aliquots of ODN-1 in human plasma were extracted at an initial concentration of 100 nM and analyzed as described above using T30 as the external standard. The absolute recovery rates in percentage of the initial amount of ODN-1 range from 66.1 to 75% with a mean recovery rate of 71.7%. A percentage relative error of 3.28% reflects that the overall recovery of oligonucleotides from plasma samples is highly reproducible. As mentioned above, detailed pharmacokinetic and metabolism studies require quantitative analysis not only of intact oligonucleotides but also of shorter oligomers which emerge as products of enzymatic degradation. Thus, we determined the influence of the oligonucleotide length on the recovery rates. Four PTOs of different length, a 20- (ODN-1), a 15- (ODN-4), a 10- (ODN-5), and a 5-mer (ODN-6), were extracted in duplicate from human plasma at an initial concentration of 100 nM. The overall recoveries calculated by eq 1 are summarized in Table 4. All PTOs were isolated in high yields with mean values of the absolute recovery of more than 70% except of the shortest oligomer, ODN6, for which the overall recovery is significantly reduced to a mean value of 4.2%. It appears that a minimum length of 10 nucleotide units is necessary to provide the high attractive forces between oligonucleotides and nanoparticles required for an effective isolation from human plasma. To confirm that the influence of the oligonucleotide length on the recovery can be neglected in a certain range a mixture of polydA ODNs of different length, (dA25-30) was extracted from human plasma and compared to the sample not added to the plasma. The CGE profile of the extracted sample is depicted in Figure 5. For each oligonucleotide, the ratio of the peak area to the total area remains almost unchanged, indicating that the ODNs of the mixture were isolated equivalently. Furthermore, the pattern of shorter oligomers down to a length of 8 nucleotide units visible in front of the main signals appears to be almost identical to the Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2201

Table 2. Extraction of ODN-1 from Human Plasma: Reproducibility and Accuracy of the Assaya initial concn (nM)

ODN-1/S15

extrd ODN-1 (pmol)

mean ( SD (pm)

mean % of init value ( SD

1 2 3 4 5 6 1 2 3 4

11.02 10.43 11.07 8.92 10.72 8.24 17.25 16.34 18.93 20.15

10.07 ( 1.09

104.4 ( 10.9

18.17 ( 0.97

90.9 ( 7.4

50

100

ODN-1/O30

extrd ODN-1 (pmol)

mean ( SD (pm)

mean % of initial value ( SD

1 2 3 4 5 6 1 2 3 4

9.07 8.63 10.27 9.48 10.08 7.98 18.68 18.12 19.34 20.60

9.25 ( 0.80

92.5 ( 8.0

19.18 ( 0.93

96.0 ( 4.7

a The premixed samples of the oligonucleotides and the internal standard T 30 were injected three times each to determine the relative concentrations before extraction. Aliquots of each sample were added to the plasma such that initial amount of the analyte was 10 (concentration, 50 nM) and 20 pm (concentration, 100 nM) and extracted using the solid-phase extraction method.

Table 3. Absolute Recovery Rates of ODN-1 and ODN-2 Extracted from Human Plasmaa ODN-1 conc (nM)

rec (%)

100 100 100 25 25 25 10 10 10 5 5 5

75.6 60.1 72.3 84.2 70.6 75.3 54.1 70.7 64.2 58.1 62.6 --

ODN-2

mean % ( SD 69.4 ( 6.7 76.7 ( 5.6 63.0 ( 6.8 60.4 ( 2.2

conc (nM)

rec (%)

100 100 100 25 25 25 10 10 10 5 5 5

88.5 84.5 87.6 82.7 83.5 83.4 87.8 72.3 74.0 64.0 63.6 58.0

mean % ( SD 86.8 ( 1.7 83.2 ( 0.4 78.0 ( 7.0 61.9 ( 2.7

a Samples were prepared with initial concentrations of 100, 25, 10, and 5 nM. Three aliquots of each sample were extracted using solidphase extraction. All determined values of the recoveries are based on the ratio of oligonucleotide to external standard T30.

profile of the sample not added to the plasma. These results imply that the extraction method is appropriate for investigations of the enzymatic degradation and metabolism of oligonucleotides in biological fluids. Plasma-Specific Variations of the Recovery Rates. In addition to the experiments described above, which were carried out with citrate-stabilized human plasma, we attempted to elucidate the influence of different stabilizing agents on the recovery rates. Three aliquots each of plasma sample stabilized with citrate, EDTA, or heparin containing ODN-2 at a concentration of 100 nM were subjected to the extraction procedure. The absolute amount of extracted oligonucleotide (in percentage of the initial amount) was determined by eq 1 using T30 as the external standard. Comparison of the recovery rates demonstrates that the applied oligonucleotide can be isolated in high yields of 86.8 and 93.6% from citrate- and EDTA-stabilized plasma, respectively. In contrast, the yields of ODN-2 recovered from heparin-stabilized plasma are reduced to 54.7% (Table 5). This indicates that the conjugate formation between oligonucleotides and nanoparticles is diminished to a certain extent due to the presence of heparin in the plasma. Considering the polyanionic nature of heparin, this can be explained by a competitive adsorption of oligonucleotides 2202 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

Figure 4. CGE analysis of ODN-2 extracted from human plasma at initial concentrations of (a) 100 and (b) 5 nM using T30 as the external standard (C).

and heparin on the particle surface leading to a reduced amount of tightly bound oligonucleotides and thus to lower rates of recovery. ES-MS Analysis of Extracted Samples. Using the standard extraction procedure, the coisolation of small amounts of plasma components cannot completely be excluded. This might be due to the hydrophobic nature of the particle surface. As shown in the results described above, CGE analysis of the samples is not affected by these traces of impurities. However, for ES-MS analysis, it turned out to be advantageous to perform the extraction procedure with a slightly modified release step in order to obtain highly pure samples which could

Table 4. Influence of the PTO Length on the Absolute Recovery Ratesa sample

rec (%)

mean % ( SD

ODN-1

74.0 75.0 75.9 78.4 76.9 79.5 7.1 1.2

74.5 ( 0.5

ODN-4 ODN-5 ODN-6

77.2 ( 1.2 78.2 ( 1.2 4.2 ( 3.0

a Samples were prepared with an initial concentration of 100 nM. All determined values of the recoveries are based on the ratio of oligonucleotide to external standard T30.

Figure 6. Analysis of ODN-2 extracted from human plasma using T25 as the external standard: (a) ES-MS spectrum (arrows, ODN2; S, standard T25); (b) CGE profile (A).

Figure 5. CGE analysis of a mixture of dA25-30 oligonucleotides extracted from human plasma (B). Table 5. Influence of Various Plasma Stabilizers on the Absolute Recovery Rates of ODN-2a plasma

rec (%)

mean % ( SD

citrate

88.5 84.5 87.6 92.4 91.1 97.4 53.6 56.2 54.2

86.8 ( 1.7

EDTA heparin

Table 6. Absolute Recovery Rates of ODN-2 Extracted from Human Plasma Using Enlarged Particles conc (nM)

rec (%)

mean % ( SD

100

90.9 86.8 81.2 76.3

88.8 ( 2.1

25

78.7 ( 2.5

a All determined values of the recoveries are based on the ratio of oligonucleotide to external standard T30.

93.6 ( 2.7 54.7 ( 1.1

a Samples were prepared with an initial concentration of 100 nM. All determined values of the recoveries are based on the ratio of oligonucleotide to external standard T30.

be analyzed directly without further purification. This was achieved using aqueous ammonia without acetonitrile or SDS to induce the release of oligonucleotides from the particles. The fact that a small amount of the extracted oligonucleotides remains on the particle surface during the cleavage step can be neglected, since plasma samples with higher initial concentrations of oligonucleotides (1-5 µM) are required for ES-MS analysis. A typical mass spectrum of ODN-1 extracted from human plasma at an initial concentration of 3 µM with T25 as the external standard is depicted in Figure 6a. The main signals can be referred to the full-length oligonucleotides. No significant impurities are visible either in the mass spectrum or in the corresponding electropherogram (Figure 6b).

Reduction of the Time Consumption of the Extraction Procedure. Separation of the nanoparticles from the dispersion medium is performed by centrifugation, which predominantly determines the time consumption of the extraction process. In the studies described above, a total time of ∼2 h is required for the whole procedure of extraction using nanoparticles with a mean diameter of 180 nm. To reduce the time consumption, an attempt was made to improve the separation steps. The goal was to develop nanoparticles with an enlarged mean diameter without changing the surface characteristics of the material. The preparation of particles with a mean diameter of 400 nm was achieved by a variation of the polymerization conditions according to the procedure described elsewhere.19 The overall recoveries of ODN-2 extracted from human plasma using these enlarged particles are summarized in Table 6. Mean values of 88.8 and 78.7% are obtained for initial concentrations of 100 and 25 nM, respectively. The results indicate that the 2-fold increase of the particle size does not affect the recovery of PTOs from human plasma. On the other hand, the total time of the extraction process could significantly be reduced to ∼30 min. Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2203

CONCLUSIONS A novel method for solid-phase extraction of oligonucleotides has been developed and validated for the quantitation of phosphorothioate oligonucleotides in human plasma. The principle of the method is based on the application of cationic polystyrene nanoparticles which are capable of reversibly binding oligonucleotides by a pH-controlled adsorption/desorption mechanism. Optimized extraction conditions provide the isolation of oligonucleotides in high yields and purity and allow the analysis of the extracted samples by CGE and ES-MS. For future studies, we intend to apply on-line coupling techniques in order to provide mass spectrometric analysis of plasma samples containing oligonucleotides in low concentrations. The standard calibration curves exhibit high linearity, indicating that the oligonucleotides used in these studies were extracted equivalently regardless of their length, sequence, and/or backbone modification. The combined linearity, accuracy, and reproducibility of the assay confirm that the developed extraction method is suitable for quantitation of PTOs in human plasma. High absolute recovery rates in the range of 60-90% were obtained for PTOs even at low-nanomolar concentrations. It was shown that the extraction method is not restricted to longer oligonucleotides but can also be applied for the quantitation of oligomers down to a length of 10 nucleotides. However, it was found that plasma-stabilizing agents have a considerable influence on the yields of recovered oligonucleotides. In contrast to citrateand EDTA-stabilized plasma, the recovery rates were significantly reduced in the presence of heparin as the stabilizer. Due to its polyanionic nature, heparin seems to competitively inhibit conjugate formation between the oligonucleotides and the nanoparticles. To further improve the extraction system, particles with an increased mean diameter of 400 nm were developed. Hence, the

2204 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

time consumption of the extraction procedure could be reduced from 2 h to ∼30 min whereas the efficiency of the method was not affected. Principally, these enlarged particles offer the possibility for future developments, for example, the transfer of the extraction procedure to automated filtration systems. While the detection and quantitation of PTOs from human plasma has been addressed in this work, the suitability of the developed method for the extraction of oligonucleotide derivatives from biological fluids will be discussed elsewhere.23 In another approach, selective enrichment of PTOs synthesized by linear PCR has been performed using the nanoparticle-based extraction method.24 The application of the described method for the investigation of the enzymatic degradation and metabolism of various oligonucleotides in human plasma will be addressed in a following publication.25 ACKNOWLEDGMENT The authors are grateful to SKW Trostberg AG (Germany) and to the Bundesministerium fu¨r Bildung und Forschung (BMBF, Germany) for their generous support. We thank N. Go¨tting for her assistance in the characterization of the nanoparticles used in this work.

Received for review January 29, 1998. Accepted March 31, 1998. AC980097W (23) Gerster, M.; Schewitz, J.; Fritz, H.; Maier, M.; Bayer, E., in preparation. (24) Bauer, T. Ph.D. Thesis, University of Tu ¨ bingen, 1997. (25) Schewitz, J.; Maier, M.; Gerster, M.; Fritz, H.; Bayer, E., in preparation.