RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 5, 309-311 (1991)
Electrospray Ionization for Analysis of Platelet-activating Factor Susan T. Weintraub* and R. Neal Pinckard Department of Pathology, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78284-7750, USA
Mark Hail Finnigan-MAT Corporation, San Jose, CA, USA SPONSOR REFEREE: S. J. Gaskell, Baylor College of Medicine, Houston, TX, USA
Platelet-activating factor (PAF) was analyzed by electrospray-ionizationmass spectrometry (ESI-MS) using a single quadrupole mass spectrometer. The positive-ion spectrum was dominated by an ion corresponding to a sodiated molecule when a low potential difference between the capillary exit (nozzle) and the skimmer was employed, but when the capillary exit voltage was increased, fragmentation of PAF was observed. Initial fragmentation involved the loss of the elements of trimethylamine from the sodiated molecule to yield [M+Na-S9]+. An intense ion at mlz147, generated by the loss of trimethylamine from the sodiated phosphocholine portion of the molecule was also detected, along with a lower intensity ion at mlz 184 which is representative of a protonated phosphocholine moiety. With negative-ion detection the major molecular species was [M CI]-. Interpretation of the mass spectral fragments was verified by ESI tandem mass spectrometry on a triple-quadrupoletandem mass spectrometer.
+
Mass spectral analysis of platelet-activating factor (PAF, l-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine; also denoted as AGEPC) has been accomplished by numerous different approaches. Fast-atom bombardment mass spectrometry (FAB-MS) has been utilized for both qualitative' and quantitative' measurements of the various intact molecular species of PAF. After removal of the phosphocholine moiety, either enzymatically by phospholipase C or chemically by treatment with hydrogen fluoride, derivatives such as the trimethylsilyl,3 t-b~tyldimethylsilyl~~~ and pentafluorobenzoy16 can be readily analyzed by gas chromatography/mass spectrometry (GC/MS). Recently, direct derivatization with heptafluorobutyric anhydride has been shown to produce an extremely useful volatile product .7.8 Each of these methods has clear advantages and disadvantages. While FAB-MS does not require substantial sample preparation or derivatization, and provides a wealth of structural information, limitations in sensitivity hamper its usefulness for biologically-derived samples. FAB used with tandem mass spectrometry (MS/MS) on the other hand, has been shown to provide low picogram sensitivity for analysis of specific PAF homo!ogs,9 but the need for a tandem instrument to accomplish the measurement precludes its use in many laboratories. Most of the G U M S procedures are suitable for analysis of very small quantities, but these techniques generally involve multiple preparative steps, often resulting in sample loss. Furthermore, with the G U M S methods, identification of the polar head group is somewhat problematic, requiring substantial additional purification and/or derivatization. We have, therefore, evaluated the use of electrospray ionization (ESI) for analysis of PAF. The present report demonstrates that ESI-MS has the potential to meet both the qualitative and quantitative requirements for analysis of PAF of biological origin. * Author to whom correspondence should be addressed. 0951-41 98/91/070309-03 $05.00
01991 by John Wiley & Sons, Ltd
EXPERIMENTAL The standard PAF homologs and analogs used in this study were prepared by reacting each alkyl-lysoglycerophosphocholine with the appropriate acid anhydride in the presence of perchloric acid, as described recently."' ESI mass spectra were obtained on either a singlequadrupole mass spectrometer (Finnigan MAT SSQ 700 San Jose, CA, USA) or a triple-quadrupole tandem mass spectrometer (Finnigan MAT TSQ 700). Both instruments were equipped with electrospray interfaces (Analytica of Branford, Branford, CT, USA). The design of this interface, which is similar to that originally described by Whitehouse and associates," is shown schematically in Fig. 1. For all experiments, both the electrospray needle and the skimmer were operated at ground potential, while the electrospray chamber and metallized capillary entrance were operated at -3.4kV for positive ions and +3.4kV for negative ions. A separate potential was placed on the metallized exit of the capillary, +90 V for acquisition of positive ions and -90 V for negative ions. For collisionally activated dissociation (CAD) in the ESI interface, ION SOURCE SKIMMER LENSES
CYUNDRICAL ELECTRODE DRYING GAS
J
-
UOUID SAMPLE
ELECT~OSPFLAY NEEDLE ASSEMBLY
DRYING GAS
1
I
TURBO PUMP
I
I
I
NRBO PUMP
I
1
Figure 1. Schematic diagram of electrospray ion source. Received I May 1991 Accepted 2 May I991
ELECTROSPRAY IONIZATION FOR ANALYSIS OF PAF
310
the potential difference between the capillary exit and the skimmer was increased to 195 V. All PAF samples were dissolved in HPLC grade methanol (Alltech Associates, Deerfield, IL, USA) and infused directly into the ESI interface by means of a syringe pump (Harvard Apparatus model 22, South Natich, MA, USA) at a flow rate of 1 pL/min. The temperature of the nitrogen drying gas as it entered the electrospray chamber was approximately 50 "C. All mass spectra were acquired by utilizing a signalaveraging protocol in the profile mode with a scan rate of 300 u/s. Typically, a 1 min period of signal averaging was employed for each spectrum.
RESULTS AND DISCUSSION The goal of this investigation was evaluation of the overall capabilities of ESI-MS for analysis of PAF. Although rigorous determination of detection limits was not made at this time, we did ascertain that fullscan spectra could be obtained readily with excellent signal-to-noise ratios during infusion of samples at concentrations well below 1pmol/pL. The positive-ion ESI mass spectrum of the hexadecyl P A F homolog (C16:O-AGEPC) is shown in Fig. 2. The base peak at mlz546 represents [ M + N a ] + , with [M H]+ at m / z 524 detected at lower intensity. Furthermore, the presence of ions at mlz 808 ([3M 2Na]'+) and m/z 1070 ([2M + Na]+) indicates that significant clustering of the AGEPC occurs in methanol at the 40pmol/yL concentration used to produce this spectrum. It is interesting to note that even though this sample had been thoroughly washed by means of the two-phase system of Bligh and Dyer,'' a protocol that we have routinely used to eliminate unwanted cations prior to FAB-MS,' the predominant species in the ESI analysis was the sodiated molecule. When the potential difference between the capillary exit and the skimmer was increased from 90 V to 195 V, fragmentation of the analyte was observed, as shown in Fig. 3 . However, the fragmentation pattern differed substantially from the one produced in FAB-MS.' With the aid of preliminary triple quadrupole MS/MS analyses we determined that the ion at mlz487 is formed by loss of the elements of trimethylamine from
+
+
546.3
I
I I1
341 0
m/z
Figure 3. ESI positive-ion mass spectrum of C16:O-AGEPC. The capillary exit voltage was 195 V.
+
[ M+ N a ] + . Analogous to the case of the FAB mass spectrum of AGEPC homologs, the ion at m / z 184 represents a protonated phosphocholine moiety; through ESI-MS/MS we found that this ion is produced solely from [M + HI+ and is, in fact, the only product ion of m / z 524. In contrast, the ion at m / z 147 is only derived from [M Na]+ and is equivalent to the loss of trimethylamine from the sodiated phosphocholine moiety. The sodiated analog of the ion at m/z 184 was not detected at mlz206, however, loss of the sodiated phosphocholine component yields the ion at m / z 341. Thus, in agreement with observations initially reported by Smith and co-workers13 and Chait and associates," highly informative C A D spectra can be acquired by increasing the potential difference between the capillary (or nozzle) and the skimmer in the ESI interface. This approach is particularly valuable because it makes it possible to obtain structural information with electrospray ionization even when a mass spectrometer that does not have MS/MS capability (e.g., a singlequadrupole instrument) is used. With negative-ion detection, the predominant species was [M + CI]-, as shown in Fig. 4. By examining an expanded display of the molecular ion region (not shown), it could be verified that the relative intensities
+
5.
inn
[M+No]
o
H,C - 0 - w , ) , , I
II
- cn,
0
CH,-C-O-CH I
H,C-080
II
P-0-CH,-
CH,-
NICH,),
,568 6
ONa 1081 9
I 60
40
20
* 200
8no
600
12b0
m/z
Figure2. ESI positive-ion mass spectrum of CI6:O-AGEPC. The capillary exit voltage was +90 V.
200
400
Figure4. ESI negative-ion mass spectrum of C l h : 0-AGEPC. The capillary exit voltage was -90 V.
ELECTROSPRAY IONIZATION FOR ANALYSIS O F PAF
of ions of mlz558 and 560 were appropriate for the presence of a single chloride. Formate complexes at [M + 451- and [M 891- were also detected. Furthermore, cluster ions were seen at rnlz 1082 ([2M + Cl]-), rnlz 1092 ([2M formatel-) and mlz 1136 ([2M - H + 2formate]-). We surmised that the counter-ions observed in these spectra were present as a consequence of washing t h e syringes and transfer lines leading to the ESI interface with formic acid. In order to further verify the spectral interpretations described above, the propionyl and butyryl PAF analogs (C16 :0-PGEPC and C16: 0-BGEPC, respectively) were analyzed under similar conditions. In each case (data not shown) when the capillary exit voltage was set at +90V, the base peak was due to the sodiated molecule (mlz 560 for C16 :0-PGEPC and mlz 574 for C16 :0-BGEPC). In addition, corresponding cluster ions were observed along with lower intensity [M + HI' ions. Furthermore, the spectra resulting from C AD in the interface were analogous to that observed for C16 :0-AGEPC, with prominent ions at mlz 147 and mlz 184 as well as an intense [M+Na-59]' ion. Finally, the negative-ion spectra were again comprised mainly of [M CI]- along with lesser amounts of the formate complexes. Our results indicate that ESI-MS will provide a powerful option for analysis of PAF. At low capillary exit voltages, only ions representative of intact molecular species are observed. At increased voltages, fragment ions are present which permit identification of the polar head group. Moreover, exceptional sensitivity and ease of coupling ESI with liquid chromatographic systems make ESI particularly attractive for use in the analysis of biologically derived PAF. In future studies we plan to evaluate different ways to control the nature of the P A F counter-ion as well as to determine the limits of detection for this technique. Ultimately, we
+
+
+
31 1
anticipate that ESI-MS will supply the means not only to perform quantitative analysis of P A F samples of biological origin through the use of appropriate stableisotope-labeled internal standards, but also to yield valuable structural information about each molecular species. Acknowledgements This study was supported in part by NIH grants HL-22555 and A1-21818,
REFERENCES 1. S. T. Weintraub, J . C. Ludwig, G . E. Mott, L M. McManus and
R . N. Pinckard, Biochem. Biophys. Res. Cornrnun. 129, 868 (1985). 2. K. L. Clay. D. 0. Stene and R . C. Murphy, Biomed. Mass Spectrorn. 11, 47 (1984). 3. K. L. Clay, R. C. Murphy, J . L. Andres, L. Lynch and P. M. Henson, Biochem. Biophys. Res. Commun. 121, 815 (1984). 4. K. Satouchi, M. Oda, K. Yasunaga and K. Saito, J . Biochem. 94, 2067 (1983). 5. A. Tokurnura, K. Karniyasu, K. Takauchi and H. Tsukatani, Biochem. Biophys. Res. Commun. 145, 414 (1987). 6 . C . S. Ramesha and W. C. Pickett, Biomed. Mass Speclrom. 13, 107 (1986). 7. R . K. Satsangi, J . C. Ludwig, S. T. Wcintrauh and R. N. Pinckard, J . Lipid Res. 30, 929 (1989). 8. S. T. Weintraub, C. L. Lear and R. N. Pinckard, J . Lipid Res. 31, 719 (1990). 9. P. E. Haroldsen and S. J . Gaskell, Biomed. Enoiron. Muss Spertrorn. 18,439 (1989). 10. S. T. Weintraub, R. N. Pinckard, T. G . Heath and D. A . Gage, J . A m . Soc. Mass Spectrom.. in press (1991). 1 1 . C . M. Whitehouse, R . N. Dreyer, M. Yarnashita and J . Fenn. Anal. Chem. 57, 675 (1985). 12. E. G . Bligh and W. J . Dyer, Can. J . Biochem. Physiol. 37, 911 ( 1959). 13. R . D. Smith, J . A . Loo, C . J . Garinaga, C. G. Edrnonds and H. R. Udscth, J . Am. Soc. Mass Specrrom. 1, 53 (1990). 14. V . Katta, S. K. Chowdhury and B. T. Chait, Anal. Chem. 63, 174 (1991).
Electrospray Ionization for Analysis of Platelet-activating Factor Susan T. Weintraub* and R. Neal Pinckard Department of Pathology, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78284-7750, USA
Mark Hail Finnigan-MAT Corporation, San Jose, CA, USA SPONSOR REFEREE: S. J. Gaskell, Baylor College of Medicine, Houston, TX, USA
Platelet-activating factor (PAF) was analyzed by electrospray-ionizationmass spectrometry (ESI-MS) using a single quadrupole mass spectrometer. The positive-ion spectrum was dominated by an ion corresponding to a sodiated molecule when a low potential difference between the capillary exit (nozzle) and the skimmer was employed, but when the capillary exit voltage was increased, fragmentation of PAF was observed. Initial fragmentation involved the loss of the elements of trimethylamine from the sodiated molecule to yield [M+Na-S9]+. An intense ion at mlz147, generated by the loss of trimethylamine from the sodiated phosphocholine portion of the molecule was also detected, along with a lower intensity ion at mlz 184 which is representative of a protonated phosphocholine moiety. With negative-ion detection the major molecular species was [M CI]-. Interpretation of the mass spectral fragments was verified by ESI tandem mass spectrometry on a triple-quadrupoletandem mass spectrometer.
+
Mass spectral analysis of platelet-activating factor (PAF, l-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine; also denoted as AGEPC) has been accomplished by numerous different approaches. Fast-atom bombardment mass spectrometry (FAB-MS) has been utilized for both qualitative' and quantitative' measurements of the various intact molecular species of PAF. After removal of the phosphocholine moiety, either enzymatically by phospholipase C or chemically by treatment with hydrogen fluoride, derivatives such as the trimethylsilyl,3 t-b~tyldimethylsilyl~~~ and pentafluorobenzoy16 can be readily analyzed by gas chromatography/mass spectrometry (GC/MS). Recently, direct derivatization with heptafluorobutyric anhydride has been shown to produce an extremely useful volatile product .7.8 Each of these methods has clear advantages and disadvantages. While FAB-MS does not require substantial sample preparation or derivatization, and provides a wealth of structural information, limitations in sensitivity hamper its usefulness for biologically-derived samples. FAB used with tandem mass spectrometry (MS/MS) on the other hand, has been shown to provide low picogram sensitivity for analysis of specific PAF homo!ogs,9 but the need for a tandem instrument to accomplish the measurement precludes its use in many laboratories. Most of the G U M S procedures are suitable for analysis of very small quantities, but these techniques generally involve multiple preparative steps, often resulting in sample loss. Furthermore, with the G U M S methods, identification of the polar head group is somewhat problematic, requiring substantial additional purification and/or derivatization. We have, therefore, evaluated the use of electrospray ionization (ESI) for analysis of PAF. The present report demonstrates that ESI-MS has the potential to meet both the qualitative and quantitative requirements for analysis of PAF of biological origin. * Author to whom correspondence should be addressed. 0951-41 98/91/070309-03 $05.00
01991 by John Wiley & Sons, Ltd
EXPERIMENTAL The standard PAF homologs and analogs used in this study were prepared by reacting each alkyl-lysoglycerophosphocholine with the appropriate acid anhydride in the presence of perchloric acid, as described recently."' ESI mass spectra were obtained on either a singlequadrupole mass spectrometer (Finnigan MAT SSQ 700 San Jose, CA, USA) or a triple-quadrupole tandem mass spectrometer (Finnigan MAT TSQ 700). Both instruments were equipped with electrospray interfaces (Analytica of Branford, Branford, CT, USA). The design of this interface, which is similar to that originally described by Whitehouse and associates," is shown schematically in Fig. 1. For all experiments, both the electrospray needle and the skimmer were operated at ground potential, while the electrospray chamber and metallized capillary entrance were operated at -3.4kV for positive ions and +3.4kV for negative ions. A separate potential was placed on the metallized exit of the capillary, +90 V for acquisition of positive ions and -90 V for negative ions. For collisionally activated dissociation (CAD) in the ESI interface, ION SOURCE SKIMMER LENSES
CYUNDRICAL ELECTRODE DRYING GAS
J
-
UOUID SAMPLE
ELECT~OSPFLAY NEEDLE ASSEMBLY
DRYING GAS
1
I
TURBO PUMP
I
I
I
NRBO PUMP
I
1
Figure 1. Schematic diagram of electrospray ion source. Received I May 1991 Accepted 2 May I991
ELECTROSPRAY IONIZATION FOR ANALYSIS OF PAF
310
the potential difference between the capillary exit and the skimmer was increased to 195 V. All PAF samples were dissolved in HPLC grade methanol (Alltech Associates, Deerfield, IL, USA) and infused directly into the ESI interface by means of a syringe pump (Harvard Apparatus model 22, South Natich, MA, USA) at a flow rate of 1 pL/min. The temperature of the nitrogen drying gas as it entered the electrospray chamber was approximately 50 "C. All mass spectra were acquired by utilizing a signalaveraging protocol in the profile mode with a scan rate of 300 u/s. Typically, a 1 min period of signal averaging was employed for each spectrum.
RESULTS AND DISCUSSION The goal of this investigation was evaluation of the overall capabilities of ESI-MS for analysis of PAF. Although rigorous determination of detection limits was not made at this time, we did ascertain that fullscan spectra could be obtained readily with excellent signal-to-noise ratios during infusion of samples at concentrations well below 1pmol/pL. The positive-ion ESI mass spectrum of the hexadecyl P A F homolog (C16:O-AGEPC) is shown in Fig. 2. The base peak at mlz546 represents [ M + N a ] + , with [M H]+ at m / z 524 detected at lower intensity. Furthermore, the presence of ions at mlz 808 ([3M 2Na]'+) and m/z 1070 ([2M + Na]+) indicates that significant clustering of the AGEPC occurs in methanol at the 40pmol/yL concentration used to produce this spectrum. It is interesting to note that even though this sample had been thoroughly washed by means of the two-phase system of Bligh and Dyer,'' a protocol that we have routinely used to eliminate unwanted cations prior to FAB-MS,' the predominant species in the ESI analysis was the sodiated molecule. When the potential difference between the capillary exit and the skimmer was increased from 90 V to 195 V, fragmentation of the analyte was observed, as shown in Fig. 3 . However, the fragmentation pattern differed substantially from the one produced in FAB-MS.' With the aid of preliminary triple quadrupole MS/MS analyses we determined that the ion at mlz487 is formed by loss of the elements of trimethylamine from
+
+
546.3
I
I I1
341 0
m/z
Figure 3. ESI positive-ion mass spectrum of C16:O-AGEPC. The capillary exit voltage was 195 V.
+
[ M+ N a ] + . Analogous to the case of the FAB mass spectrum of AGEPC homologs, the ion at m / z 184 represents a protonated phosphocholine moiety; through ESI-MS/MS we found that this ion is produced solely from [M + HI+ and is, in fact, the only product ion of m / z 524. In contrast, the ion at m / z 147 is only derived from [M Na]+ and is equivalent to the loss of trimethylamine from the sodiated phosphocholine moiety. The sodiated analog of the ion at m/z 184 was not detected at mlz206, however, loss of the sodiated phosphocholine component yields the ion at m / z 341. Thus, in agreement with observations initially reported by Smith and co-workers13 and Chait and associates," highly informative C A D spectra can be acquired by increasing the potential difference between the capillary (or nozzle) and the skimmer in the ESI interface. This approach is particularly valuable because it makes it possible to obtain structural information with electrospray ionization even when a mass spectrometer that does not have MS/MS capability (e.g., a singlequadrupole instrument) is used. With negative-ion detection, the predominant species was [M + CI]-, as shown in Fig. 4. By examining an expanded display of the molecular ion region (not shown), it could be verified that the relative intensities
+
5.
inn
[M+No]
o
H,C - 0 - w , ) , , I
II
- cn,
0
CH,-C-O-CH I
H,C-080
II
P-0-CH,-
CH,-
NICH,),
,568 6
ONa 1081 9
I 60
40
20
* 200
8no
600
12b0
m/z
Figure2. ESI positive-ion mass spectrum of CI6:O-AGEPC. The capillary exit voltage was +90 V.
200
400
Figure4. ESI negative-ion mass spectrum of C l h : 0-AGEPC. The capillary exit voltage was -90 V.
ELECTROSPRAY IONIZATION FOR ANALYSIS O F PAF
of ions of mlz558 and 560 were appropriate for the presence of a single chloride. Formate complexes at [M + 451- and [M 891- were also detected. Furthermore, cluster ions were seen at rnlz 1082 ([2M + Cl]-), rnlz 1092 ([2M formatel-) and mlz 1136 ([2M - H + 2formate]-). We surmised that the counter-ions observed in these spectra were present as a consequence of washing t h e syringes and transfer lines leading to the ESI interface with formic acid. In order to further verify the spectral interpretations described above, the propionyl and butyryl PAF analogs (C16 :0-PGEPC and C16: 0-BGEPC, respectively) were analyzed under similar conditions. In each case (data not shown) when the capillary exit voltage was set at +90V, the base peak was due to the sodiated molecule (mlz 560 for C16 :0-PGEPC and mlz 574 for C16 :0-BGEPC). In addition, corresponding cluster ions were observed along with lower intensity [M + HI' ions. Furthermore, the spectra resulting from C AD in the interface were analogous to that observed for C16 :0-AGEPC, with prominent ions at mlz 147 and mlz 184 as well as an intense [M+Na-59]' ion. Finally, the negative-ion spectra were again comprised mainly of [M CI]- along with lesser amounts of the formate complexes. Our results indicate that ESI-MS will provide a powerful option for analysis of PAF. At low capillary exit voltages, only ions representative of intact molecular species are observed. At increased voltages, fragment ions are present which permit identification of the polar head group. Moreover, exceptional sensitivity and ease of coupling ESI with liquid chromatographic systems make ESI particularly attractive for use in the analysis of biologically derived PAF. In future studies we plan to evaluate different ways to control the nature of the P A F counter-ion as well as to determine the limits of detection for this technique. Ultimately, we
+
+
+
31 1
anticipate that ESI-MS will supply the means not only to perform quantitative analysis of P A F samples of biological origin through the use of appropriate stableisotope-labeled internal standards, but also to yield valuable structural information about each molecular species. Acknowledgements This study was supported in part by NIH grants HL-22555 and A1-21818,
REFERENCES 1. S. T. Weintraub, J . C. Ludwig, G . E. Mott, L M. McManus and
R . N. Pinckard, Biochem. Biophys. Res. Cornrnun. 129, 868 (1985). 2. K. L. Clay. D. 0. Stene and R . C. Murphy, Biomed. Mass Spectrorn. 11, 47 (1984). 3. K. L. Clay, R. C. Murphy, J . L. Andres, L. Lynch and P. M. Henson, Biochem. Biophys. Res. Commun. 121, 815 (1984). 4. K. Satouchi, M. Oda, K. Yasunaga and K. Saito, J . Biochem. 94, 2067 (1983). 5. A. Tokurnura, K. Karniyasu, K. Takauchi and H. Tsukatani, Biochem. Biophys. Res. Commun. 145, 414 (1987). 6 . C . S. Ramesha and W. C. Pickett, Biomed. Mass Speclrom. 13, 107 (1986). 7. R . K. Satsangi, J . C. Ludwig, S. T. Wcintrauh and R. N. Pinckard, J . Lipid Res. 30, 929 (1989). 8. S. T. Weintraub, C. L. Lear and R. N. Pinckard, J . Lipid Res. 31, 719 (1990). 9. P. E. Haroldsen and S. J . Gaskell, Biomed. Enoiron. Muss Spertrorn. 18,439 (1989). 10. S. T. Weintraub, R. N. Pinckard, T. G . Heath and D. A . Gage, J . A m . Soc. Mass Spectrom.. in press (1991). 1 1 . C . M. Whitehouse, R . N. Dreyer, M. Yarnashita and J . Fenn. Anal. Chem. 57, 675 (1985). 12. E. G . Bligh and W. J . Dyer, Can. J . Biochem. Physiol. 37, 911 ( 1959). 13. R . D. Smith, J . A . Loo, C . J . Garinaga, C. G. Edrnonds and H. R. Udscth, J . Am. Soc. Mass Specrrom. 1, 53 (1990). 14. V . Katta, S. K. Chowdhury and B. T. Chait, Anal. Chem. 63, 174 (1991).
