Reduction in Matrix-Related Signal Suppression Effects in

Application of on-line orthogonal LC/LC separations can be effective in reducing both causes of matrix-related signal suppression effects i.e., column...
0 downloads 0 Views 127KB Size
Anal. Chem. 2001, 73, 6014-6023

Reduction in Matrix-Related Signal Suppression Effects in Electrospray Ionization Mass Spectrometry Using On-Line Two-Dimensional Liquid Chromatography Rob Pascoe,† Joe P. Foley,† and Arkady I. Gusev*,‡,§

Drexel University, Philadelphia, Pennsylvania 19104, and Rohm and Haas Co., Spring House, Pennsylvania 19477

The effect of liquid chromatographic separation on matrixrelated signal suppression in electrospray ionization mass spectrometry (LC-ESI-MS) was investigated. A method incorporating on-line two-dimensional liquid chromatography mass spectrometry (LC/LC-MS) was developed to compensate for matrix effects and signal suppression in qualitative and quantitative analysis. The LC/LC-MS(MS) approach was successfully applied for single-component and multicomponent analysis in a variety of complex matrixes. It was demonstrated that matrix-related signal suppression could be induced solely by (i) column overload, (ii) matrix component-analyte coelution, or a combination of each. Application of on-line orthogonal LC/LC separations can be effective in reducing both causes of matrix-related signal suppression effects i.e., column overload and matrix-analyte coelution for a variety of LCn/MSn applications. Liquid chromatography-mass spectrometry (LC-MS) is a powerful analytical technique that combines the chromatographic separation power of LC with the sensitivity and selectivity of electrospray ionization mass spectrometry (ESI-MS).1-3 Arguably the most abundant application of LC-ESI-MS is the characterization of sample mixtures in complex matrixes. However, one drawback associated with LC-ESI-MS analysis in complex biological or environmental samples is that it is susceptible to matrixrelated signal suppression effects.4-6 The matrix-related effects could cause serious discrepancies between signal responses obtained from standard and matrix samples, resulting in difficulties in qualitative and quantitative characterization. †

Drexel University. Rohm and Haas Co. § Current address: (phone) (734) 622 1396; (fax) (734)622 5115; (e-mail) [email protected]. (1) Niessen, W. M. A.; Tinke, A. P. J. Chromatogr., A 1995, 703, 37-57. (2) Cole, R. B. Electrospray Ionization Mass Spectrometry: Fundamentals of Instrumentation and Applications; Wiley: New York, 1997. (3) Gaskell, S. J. J. Mass Spectrom. 1997, 32, 677-688. (4) Gelpi, E. J. Chromatogr., A 1995, 703, 59-80. (5) Jemal, M.; Huang, M.; Jiang, X.; Mao, Y.; Powell, M. Rapid Commun. Mass Spectrom. 1999, 13, 2125-2132. (6) Slobodnik, J.; van Baar, B. L. M.; Brinkman, U. A. Th. J. Chromatogr., A 1995, 703, 81-121. ‡

6014 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

In the ESI process, signal suppression effects have been studied extensively.7-14 The matrix-induced signal suppression effects are believed to result from the competition of analyte ions and matrix components for access to the droplet surface for gasphase emission. In the presence of matrix components, the rate of analyte ion formation may differ significantly from that of a standard sample, resulting in variations in signal response. One approach employed to compensate for the signal suppression effects associated with electrospray ionization is the use of buffer additives.15,16 Signal response can be enhanced through the use of additives such as ammonium formate and propionic acid in the mobile phase. Although signal response may be augmented, the use of buffers may cause complications with the chromatographic separation processes.15 If available, internal standards may also be used to ensure accurate analyte quantification.17 However, the most common method to reduce signal suppression is obtained through sample cleanup prior to LC-MS analysis. Both on-line and off-line cleanup procedures have been used in order to obtain accurate and reproducible quantification. Off-line techniques such as liquid/liquid partitioning, open column extraction, and solid-phase extraction (SPE) are available. While effective, some of these techniques may require time-consuming and labor-intensive preparation procedures. Recent breakthroughs in automated and semiautomated methods have been achieved using a 96-well or higher format.18-22 (7) Thomson, B. A.; Irbarne, J. V. J. Chem. Phys. 1976, 64, 2287-2294. (8) Burhman, D. L.; Price, P. I.; Rudewicz, P. J. J. Am. Soc. Mass Spectrom. 1996, 7, 1099-1105. (9) Zell, M.; Husser, C.; Hopfgartner, G. Rapid Commun. Mass Spectrom. 1997, 11, 1107-2319. (10) Matuszewski, B. K.; Costanzer, M. L.; Chavez-Eng, C. M. Anal. Chem. 1998, 70, 882. (11) King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah. T. J. Am. Soc. Mass Specrom. 2000, 11, 942-950. Bonfiglio, R.; King, R.; Olah, T.; Merkle, K. Rapid Commun. Mass Spectrom. 1999, 13, 1175-1185. (12) Kebarle, P.; Tang. L. Anal. Chem. 1993, 65, 972A-986A. (13) Tang, L.; Kebarle, P. Anal. Chem. 1993, 65, 3654-3668. (14) Constantopoulos, T. L.; Jackson, G. S.; Enke, C. G. J. Am. Soc. Mass Spectrom. 1999, 10, 625-634. (15) Temesi, D.; Law, B. LC-GC 1999, 17, 626-632. (16) Apffel, A.; Fischer, S.; Goldberg, G.; Goodley, P. C.; Kuhlmann. F. E. J. Chromatogr.. A 1995, 712, 177-190. (17) Boyd, R. K. Rapid Commun. Mass Spectrom. 1993, 7, 257-271. (18) Campins-Falco, P.; Herraez-Hernandez, R.; Sevillano-Cabeza, A. J. Chromatogr., B 1993, 619, 177. (19) Mathews, C. Z.; Woolf, E. J.; Lin, L.; Fang, W.; Hsieh, J.; Ha, S.; Simpson, R.; Matuszewski, B. K. J. Chromatogr., B 2001, 751, 237-246. 10.1021/ac0106694 CCC: $20.00

© 2001 American Chemical Society Published on Web 11/10/2001

Figure 1. Compound structures.

On-line techniques include valve switching (diverting valve) and on-line SPE. Diverting matrix components to waste before and after compound elution reduces the amount of undesirable and nonvolatile components.23 On-line SPE systems such as the Prospekt (Spark Holland, The Netherlands) can also be effective for routine analysis. Another approach for sample cleanup is off-line and on-line two-dimensional liquid chromatography mass spectrometry (2DLC-MS). The 2D-LC-MS system utilizes the separation power of two LC columns for sample cleanup as well as enhanced separation. The most common and widely used type of off-line 2D-LC analysis is fraction collection. Off-line 2D-LC-MS using a reversed-phase column and fraction collection followed by analysis of fractions has proven to be very effective.24 On-line 2D-LC with UV and MS detectors have also been used successfully in removing unwanted matrix materials from interfering with analytes in samples containing a complex matrix.9,25-28 However, most on-line 2D-LC-MS methods developed employ column back-flushing. The back-flushing technique involves using one LC column to retain the analyte while the matrix components elute. Once some matrix species are removed from the first column, the analytes are back-flushed into a second LC column for chromatographic separation. Although quite universal and (20) Scott, R. J.; Palmer, L. J.; Lewis, I. A.; Pleasance, S. Rapid Commun. Mass Spectrom. 1999, 13, 2305-2319. (21) Ke, J.; Yancey, M.; Zhang, S.; Lowes, S.; Henion, J. J. Chromatogr., B 2000, 742, 369-380. (22) Cannacchi, V.; Baratte, S.; Cicioni, P.; Frigerio, E.; Long, J. James, C. J. Pharm. Biomed. Anal. 2000, 22, 451-460. (23) Wang, Z.; Hop, C. E. C. A.; Leung, K. H.; Pang, J. M. J. Mass Spectrom. 2000, 35, 71-76. (24) Choi, B. K.; Hercules, D. M.; Gusev, A. I. J. Chromatogr., A 2001, 907, 337-342. (25) El Mahjoub, A.; Staub, C. J. Chromatogr., B 2000, 742, 381-390. (26) Hogendoorn, E. A.; van Zoonen, P. J. Chromatogr., A 1995, 703, 149-166. (27) Doerge, D. R.; Churchwell, M. I.; Delcos K. B. Rapid Commun. Mass Spectrom. 2000, 14, 673-678. (28) Swart, R.; Koivisto, P.; Markides, K. E. J. Chromatogr., A 1998, 828, 209218.

capable of multiple analyte analysis, this method only uses a portion of the chromatographic selectivity available, given that the first column is used strictly as a trapping mechanism. Therefore, 2D-LC-MS analysis using heart cutting and column switching can be advantageous for the reason that it effectively utilizes the chromatographic selectivity available from both columns. Similar systems have been employed for the removal of unwanted matrix components in LC/LC-UV analysis.26,29,30 One of the main causes for signal suppression in an LC-MS system is believed to be the coelution of analytes and matrix components, i.e., when matrix component(s) and analyte exhibit the same retentive behavior and elute at the ESI source simultaneously. In addition, it was recently found that column overload could contribute to significant signal suppression.24 Column overload is the result of a large amount of matrix components that exceed the loading capacity of the column and therefore affect the separation ability of the system. It has been confirmed through off-line LC/LC-MS experiments that column overload can be a major factor in matrix-related signal suppression effects in LCESI-MS analysis.24 The purpose of this paper is to investigate on-line 2D-LCMS(MS) to chromatographically compensate for signal suppression in complex environmental and biological matrixes. This paper will focus on the (i) qualitative improvement of mass spectra in full-scan mode, (ii) quantitative aspects of signal suppression in single ion monitoring (SIM) and multiple reaction monitoring modes (MRM) and (iii) investigation of the chromatographic aspects of signal suppression effects, specifically column overload versus matrix-analyte coelution. EXPERIMENTAL SECTION Materials. Hydroxy-fenozide, methoxy-fenozide, G-fenozide, hydroxy-fenbucanozole, fenbucanozole, wheat forage and straw (29) Hagimoto, T.; Okada, J.; Motohashi, M.; Yoshimura, Y. J. Chromatogr., B 1998, 712, 161-167. (30) Motoyama, A.; Suzuki, O.; Shirota, R.; Namba. Rapid Commun. Mass Spectrom. 1999, 13, 2204-2208.

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

6015

Table 1. Column Conditionsa

flow rate (mL/min)

column 1 C18 (25 cm × 3 mm, 5 µm)

0.5

C18-5 cm (5 cm × 2 mm, 3 µm)

0.4

C18-10 cm (10 cm × 2 mm × 3 µm)

0.3

RAM (10 cm × 2 mm, 5 µm)

0.3

C8 (5 cm × 2 mm, 5 µm)

0.6

column 2

flow rate (mL/min)

C18 (25 cm × 3 mm, 5 µm)

0.5

C18 (25 cm × 4.6 mm, 3 µm)

1.0

a

%B

time (min)

20 50 70 70 20 60 90 90 20 60 90 90 20 60 90 90 20 60 90 90

0.0 6.0 10.0 18.0 0.0 4.0 7.0 10.0 0.0 4.0 6.0 14.0 0.0 3.0 5.0 9.0 0.0 4.0 8.0 10.0

%B

time (min)

20 60 90 90 20 90 90

0.0 14.0 17.0 23.0 0.0 13.0 23.0

For column manufacturer, refer to instrumental conditions section.

matrixes, pecan nut matrix, and mouse plasma were obtained from Rohm and Haas Co. (Philadelphia, PA). Structures of the analytes are shown in Figure 1. Solvents used in the extraction and cleanup procedure were HPLC grade and obtained from Fischer Scientific (Pittsburgh, PA). Sample Preparation Procedures. Standard analyte solutions (100 µg/mL) were prepared by dissolving 10 mg of analyte in 100 mL acetonitrile-water (1:1, v/v). Lower concentration standards were prepared by serial dilution. Wheat forage, wheat straw, and pecan nut matrixes were prepared in the following manner.

Figure 2. Column configuration and valve switching. 6016

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

A 2-g sample of matrix was mixed with 150 mL of extraction solvent (methanol-0.1 M HCl, 90:10) and shaken for ∼30 min. The extract was then separated by vacuum filtration, the filter cake was rinsed with 50 mL of extraction solvent, and the resulting filtrate was collected (matrix stock solution). Plasma samples were mixed with methanol (1:4, v/v) and centrifuged; the resulting liquid was decanted and collected (matrix stock solution). To avoid extraction and recovery issues, matrix stock solutions were spiked with the standard analyte solution after extraction procedures. Preparation of the spiked matrix sample involved drying a 20-mL aliquot of the standard analyte solution under nitrogen, followed by redissolving the sample with 2 mL of standard analyte solution. Samples were prepared individually for each experiment; therefore, some intrasample variations were observed. Compound Recovery. The signal suppression was calculated using percent recovery obtained from spiked matrix samples. The percentage of recovery (relative signal response) was calculated by relating the peak areas obtained from standards versus peak areas of matrix samples spiked with the same concentration of analyte. A value of 100% is indicative of no signal suppression. Recovery of 70-120% is generally acceptable. Signal suppression was induced by augmenting the injection volume up to 100 µL. Increase of injection volume will introduce a larger amount of matrix material into the system, resulting in increased signal suppression. Precision, Accuracy, and Sampling. Precision and accuracy of the methods were evaluated by replicate analyses (n ) 4) of the standards and samples. The coefficient of variation of the interval of the mean was calculated within one sample preparation using the equation ∆ ) t(s)/n1/2, where ∆ is the difference in recovery, t is 2.35 from a t-table at the 95% confidence interval, s is the standard deviation of recovery values, and n is the number of samples (analysis). For most experiments, an increase in recovery greater than ∼4-5% was determined to be statistically significant. Samples were run as follows: standard, spiked matrix (in duplicate), standard, and wash in order to alleviate sample and matrix carryover. Instrumental Conditions. (1) LC-MS(MS) and LC/LCMS(MS) Analysis. LC and LC/LC were carried out with two HP1100 systems (Hewlett-Packard, Wilmington, DE); injection volume was 10-100 µL. Analytes were separated with the following columns: Hewlett-Packard Zorbax C18 (25 cm × 3 mm, 5 µm),

Figure 3. (A-D) LC-MS and (E-H) LC/LC-MS analysis in negative ion mode for wheat forage matrix: 1, G-fenozide; 2, hydroxy-fenozide; 3, methoxy-fenozide. (A-D) LC-MS TIC and full scans of 1, 2, and 3, respectively. (E-H) LC/LC-MS TIC and full scans of 1, 2, and 3, respectively. LC-MS analysis consisted of a C-18 column (10 cm × 2 mm, 3 µm). LC/LC-MS analysis incorporated two C18 columns (10 cm × 2 mm, 3 µm) and (25 cm × 3 mm, 3 µm) in series.

Metachem C18 (5 cm × 2 mm, 3 µm), Waters Spherisorb C18 (25 cm × 4.6 mm, 3 µm), Metachem Restricted Access Media (RAM) (10 cm × 2 mm, 5 µm), and Metachem C8 (5 cm × 2 mm, 5 µm). The solvent gradient was composed of water (phase A) and acetonitrile (phase B) unless otherwise specified (see Table 1). 1D- and 2D-LC systems were equilibrated to initial conditions prior to subsequent injections. Columns were used for over 300 matrix injections without evidence of degradation.

Effluent from the LC and LC/LC systems was split to allow a flow rate of ∼100-200 µL/min into the ion source. LC-MS/MS and LC/LC-MS/MS analyses were performed on the API-365 triple quadrupole ESI mass spectrometer (Perkin-Elmer, Foster City, CA) with a turbo ion spray source. Analyses were performed in both negative and positive ion modes using full-scan, SIM, and MRM modes. For negative ion acquisition, the instrument parameters were -3500, -30, and -200 V for the spray, orifice, Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

6017

Figure 4. (A-D) LC-MS and (E-H) LC/LC-MS analysis in positive ion mode for wheat forage matrix: 1, hydroxy-fenbucanozole; 2, fenbucanozole; 3, methoxy-fenozide. (A-D) LC-MS TIC and SIM of 1, 2, and 3, respectively. (E-H) LC/LC-MS TIC and SIM of 1, 2, and 3, respectively. LC-MS analysis consisted of a C-8 column (5 cm × 2 mm, 5 µm). LC/LC-MS analysis incorporated two C18 columns (5 cm × 2 mm, 5 µm) and (25 cm × 3 mm, 3 µm) in series.

and ring voltages, respectively. For positive ion mode, the instrument parameters were 4200, 20, and 180 V, for the spray, orifice, and ring voltages, respectively. (2) Heart Cutting and Column Switching. A schematic representation of the column configuration applied is given in Figure 2. The on-line 2D-LC system consisted of a binary and quaternary pump (HP 1100, Hewlett-Packard) and two columns connected using the HP 1100 six-port valve (valve 1). An additional six-port valve (valve 2) was installed before the ESI-MS to divert the LC flow to MS only during analyte elution. The 2D-LC analysis 6018

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

consisted of four steps: (A) column 1 separation, (B) loading to column 2, (C) column 2 separation, and (D) ESI-MS analysis. During step A (column 1 separation, Figure 2A), Valve 1 is in the closed position, pump 1 directs mobile phase through column 1 and exits into waste, and pump 2 directs mobile phase through column 2 to waste. The analytes are then separated on the first column. During step B (loading to column 2, Figure 2B), valve 1 is in the open position; the analytes from the first column are loaded onto the second column using the heart-cutting method, ∼70% for the wheat forage matrix, although the increase in recovery was not as pronounced. Although different length and column chemistries were utilized, there was not one system that in particular was better than the others. This result supports the hypothesis that the main cause of signal suppression in the wheat forage matrix can be column overload. The only real deviation was with G-fenozide and the RAM/C18 system, which may be the result of the analyte polarity and the column system employed. Table 6 shows results obtained in negative ion mode for wheat straw matrix using the same column combinations as in Table 5. In this data set, it is apparent that the signal suppression is stronger in the straw matrix compared to forage. The greater increase in signal suppression could be the result of a greater amount of interfering constituents in the sample as well as coelution of matrix components. Comparing all four systems, the C8/C18 column combination demonstrated greater absolute recoveries for G-fenozide and hydroxy-fenozide. Therefore, in the wheat straw matrix, the difference in column chemistries proved to be advantageous. It also suggests that column overload as well as matrix-analyte coelution appears to be causing the observed signal suppression in straw matrix.

Table 4. Comparison of Absolute Recoverya and Increase in Recovery (∆) Obtained from 1Db and 2Dc Analysis of Pecan Nut Matrix Using Negative Ion Mode and 0.1 mM Ammonium Formate as a Mobile-Phase Modifier (All Values Given as Percent) analyte

injection vol (µL)

LC-MS

LC/LC-MS



LC-MS/MS

LC/LC-MS/MS



G-fenozide hydroxyfenozide methoxyfenozide

50 50 50

60 60 53

81 75 85

21 15 32

56 55 55

95 87 85

39 32 30

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b C8, 5 cm, 2 mm, 5 µm. c C8, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm.

6020 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

Table 5. Comparison of Absolute Recoverya and Increase in Recovery (∆) Obtained by LC/LC-MS/MS Analysis of Wheat Forage Matrix Sample Using Negative Ion Mode and 0.1 mM Ammonium Formate as a Mobile-Phase Modifier (All Values Given as Percent) analyte G-fenozide hydroxy-fenozide methoxy-fenozide

injection vol (µL)

LC/LC-MS/MSb



LC/LC-MS/MSc



LC/LC-MS/MSd



LC/LC-MS/MSe



10 50 100 10 50 100 10 50 100

84 69 56 104 98 84 105 105 105

0 10 15 0 2 10 0 20 30

97 77 60 85 95 120 88 88 83

12 25 10 1 10 14 3 3 20

66 51 41 100 80 70 88 84 92

15 21 24 8 18 24 3 5 29

93 73 62 100 100 90 94 85 86

0 4 10 0 5 10 0 0 10

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b LC, C18, 5 cm, 2 mm, 5 µm; LC/LC, C18, 5 cm, 2 mm, 5 µm/C18, 25 cm, 4.6 mm, 3 µm. c LC, C18, 10 cm, 2 mm, 3 µm; LC/LC, C18, 10 cm, 2 mm, 3 µm/C18, 25 cm, 3 mm, 3 µm. d LC, RAM, 10 cm, 2 mm, 5 µm; LC/LC, RAM, 10 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm. e LC, C8, 5 cm, 2 mm, 5 µm; LC/LC, C8, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm.

Table 6. Comparison of Absolute Recoverya and Increase in Recovery (∆) Obtained by LC/LC-MS/MS Analysis of Wheat Straw Matrix Sample Using Negative Ion Mode and 0.1 mM Ammonium Formate as a Mobile-Phase Modifier (All Values Given as Percent) analyte G-fenozide hydroxy-fenozide methoxy-fenozide

injection vol (µL)

LC/LC-MS/MSb



LC/LC-MS/MSc



LC/LC-MS/MSd



LC/LC-MS/MSe



10 50 100 10 50 100 10 50 100

71 36 25 94 73 54 104 93 81

15 10 15 0 10 10 20 30 35

79 44 31 86 85 89 81 66 64

15 15 10 0 14 20 0 2 15

49 28 19 85 54 43 83 69 59

7 14 9 28 21 22 9 24 28

87 59 49 105 88 72 73 69 69

3 10 19 10 30 30 0 5 10

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b LC, C18, 5 cm, 2 mm, 5 µm; LC/LC, C18, 5 cm, 2 mm, 5 µm/C18, 25 cm, 4.6 mm, 3 µm. c LC, C18, 10 cm, 2 mm, 3 µm; LC/LC, C18, 10 cm, 2 mm, 3 µm/C18, 25 cm, 3 mm, 3 µm. d LC, RAM, 10 cm, 2 mm, 5 µm; LC/LC, RAM, 10 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm. e LC, C8, 5 cm, 2 mm, 5 µm; LC/LC, C8, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm.

Table 7. Comparison of Absolute Recoverya and Increase in Recovery (∆) Obtained from 1Db and 2Dc Analysis of Wheat Forage and Straw Using Positive Ion Mode and 0.1% Formic Acid Mobile-Phase Modifier (All Values Given as Percent) analyte

matrix

fenbucanozole

forage straw

methoxy-fenozide

forage straw

Table 8. Absolute Recoverya and Increase in Recovery (∆) Obtained from 1Db and 2Dc Analysis of Mouse Plasma Matrix Sample Using Negative Ion Mode (All Values Given as Percent) analyte

injection vol (µL)

LC-MS

LC/LC-MS



10 50 10 50 10 50 10 50

114 90 82 44 97 79 76 45

124 110 97 59 103 94 81 80

10 20 15 15 6 15 5 35

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b LC, C8, 5 cm, 2 mm, 5 µm. c LC/LC, C8, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm.

Table 7 shows results for wheat forage and straw samples analyzed for fenbucanozole and methoxy-fenozide, in positive ion mode at injection volumes of 10 and 50 µL each with 0.1% formic acid in the mobile phase. The one-dimensional system was composed of a C8 column and 2D-LC-MS/MS consisted of C8

G-fenozide hydroxy-fenozide methoxy-fenozide

injection vol (µL)

LC-MS/MS

LC/LC-MS/MS



10 50 100 10 50 100 10 50 100

89 75 64 100 73 37 100 55 80

109 100 94 102 98 97 100 80 90

20 25 30 2 25 60 0 25 25

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b LC, C8, 5 cm, 2 mm, 5 µm. c LC/LC, C8, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 µm.

and C18 columns in series. The increase in percent recovery ranged from about 5 to 35% for fenbucanozole and methoxyfenozide, while the absolute recoveries were all greater than ∼80% for almost all samples. There was a slightly higher improvement (∆) and absolute recovery observed for methoxy-fenozide when compared to analysis in negative ion mode. Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

6021

Table 9. Comparison of Absolute Recoverya and Increase in Recovery (∆) of Single- and Multiple-Compound Analysis of a Wheat Forage Matrix Sample Using Negative Ion Mode and 0.1 mM Ammonium Formate as a Buffer Additiveb analyte

injection vol (µL)

single residue

multiple residue



G-fenozide hydroxy-fenozide methoxy-fenozide

50 50 50

56 87 103

58 73 79

2 14 24

a Signal response is expressed as a percentage of that obtained from standard samples; 100 is indicative of no signal suppression. b RAM, 5 cm, 2 mm, 5 µm/C18, 25 cm, 3 mm, 3 µm.

Signal suppression induced by a biological matrix was also investigated. Table 8 compares one and two-dimensional LC/LCMS/MS for the analysis of G-fenozide, hydroxy-fenozide, and methoxy-fenozide in mouse plasma. The one-dimensional LC system used a C8 column, and the dual LC was composed of a C8/C18 column combination. The absolute recovery for all three compounds was greater than ∼90%, with an average increase of ∼23% using 2D-LC-MS/MS. Single- and Multiple-Compound Analysis. The data in this paper suggest that column overload is one of the major causes of matrix-related signal suppression effects. Therefore, analysis of multiple compounds in one injection will introduce more matrix material onto the second column and should result in increased

Figure 6. Demonstration of matrix-analyte coelution of matrix components leading to signal suppression. (A, B) TIC and 2D-UV chromatogram of (1) hydroxy-fenbucanozole and (2) methoxy-fenozide for wheat forage matrix in positive ion mode. (C, D) TIC and 2D-UV chromatogram of (1) hydroxy fenbucanozole and (2) methoxy-fenozide for wheat straw matrix in positive ion mode. For (A) and (B): column 1, C8 (5 cm × 2 mm, 5 µm); column 2, C18 (25 cm × 3 mm, 3 µm). For (C) and (D): column 1, C18 (5 cm × 2 mm, 5 µm); column 2, C18 (25 cm × 4.6 mm, 3 µm). 6022

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

signal suppression. Table 9 is a comparison of two 2D-LC-MS/ MS experiments: i.e., (i) multiple-compound analysis, in which three compounds were analyzed in one injection and a column switching valve was opened for a total of ∼3 min (∼1 min. each compound) to load each compound into the second column, and (ii) single-compound experiment, where a valve was opened for ∼1 min to allow only the volume of one peak to be loaded onto the second column. Upon inspection of Table 9, it is evident that the multiple-compound analyses had a higher degree of signal suppression than the single-compound experiments, increasing as much as ∼24% for methoxy-fenozide. The large increase in signal suppression for methoxy-fenozide (∼24%) compared to no statistically significant increase for G-fenozide suggests that later eluting compounds in a multiple-analyte experiment experience some residual column leaching from the matrix components loaded from previous analytes. Therefore, in the case of extreme signal suppression, maximum compensation for column overload can be achieved through single-compound analysis. Matrix-Analyte Coelution versus LC Column Overload. To further investigate the role of matrix-analyte coelution and column overload as the cause of matrix-related signal suppression effects, chromatograms from a UV detector placed after the second column were correlated with signal suppression effects. Figure 6 shows the TIC in positive ion mode and UV chromatograms of a 2D-LC-UV-MS/MS analysis of hydroxy-fenbucanozole and methoxy-fenozide. Compounds were analyzed in wheat forage using a C8/C18 column combination and in a straw matrix using a C18/ C18 column combination. Upon inspection of the UV chromatogram for wheat forage matrix (Figure 6B), minimal matrix interference is observed for both compounds. Absolute recoveries of ∼97% and ∼94% were obtained for hydroxy-fenbucanozole and methoxy-fenozide, respectively. The UV chromatogram for the wheat straw matrix showed minimal interference for hydroxyfenbucanozole with an absolute recovery of ∼90%. Alternatively, a large peak was evident on the UV chromatogram for methoxyfenozide, leading to significant signal suppression and an absolute recovery of ∼63% (Figure 6C and D). In the case of the straw matrix, coelution of matrix substituents with methoxy-fenozide resulted in signal suppression and poor recoveries. The analysis in Figure 6C and D was performed using two C18 columns; ultimately, it maybe more advantageous to use a C4,8,cyanopropyl/C18 combination to create a difference in column chemistry. Therefore, in addition to LC column overload, matrix-related signal suppression effects could also be the result of matrix-analyte coelution. Future development could involve the use of orthogonal chromatographic systems such as size exclusion chromatography (SEC) versus reverse-phase chromatography, two different mobile phases for each LC dimension, and systems with greater overall column capacity.

CONCLUSIONS The use of on-line LC/LC-MS(MS) provided an efficient way to compensate for signal suppression and improve qualitative and quantitative data for the analysis of a variety of compounds in complex matrixes. Qualitatively, the 2D-LC-MS technique demonstrated improvements in data quality for full-scan and SIM modes for both negative and positive ion modes. Quantitatively, the 2D-LC-MS(MS) analysis technique significantly increased recovery of compounds tested for wheat forage, straw, pecan nuts, and mouse plasma matrixes. Absolute recoveries obtained with 2D-LC-MS(MS) were on average greater than the desired value of at least ∼70%, therefore confirming that the LC/LC-MS(MS) method is effective for the simultaneous quantification of multiple compounds (i.e., up to three per injection) in a complex sample matrix with minimal sample cleanup. It was also demonstrated that 2D-LC-MS single-compound analysis (i.e., one compound per injection) could be more effective than multiple-compound analysis in reducing matrix-related signal suppression. The study of single- and multiple-compound analysis confirmed that column leaching from prior eluting compounds could augment signal suppression effects of later eluting analytes. Recovery for methoxy-fenozide, the latest eluted compound was increased by additional ∼24% for single-compound 2D-LC-MS(MS) analysis versus multiple-compound analyses. The origins of signal suppression, i.e., column overload versus matrix-analyte coelution, were also investigated. It was demonstrated that matrix-related signal suppression could be induced solely by column overload, matrix-analyte coelution, or a combination of each. Although all column combinations proved to be effective in reducing signal suppression, the C8/C18 system demonstrated the greatest absolute recovery for most compounds and injection volumes. The difference in column chemistries was effective in reducing both causes of matrix-related signal suppression effects, column overload and matrix-analyte coelution. Overall, the LC/LC-MS(MS) method investigated and developed proved to be relatively fast, reproducible, and most importantly effective for analysis of multiple compounds in a complex sample matrix with minimal sample cleanup. ACKNOWLEDGMENT The authors thank Dr. Stanley S. Stavinski (Rohm and Haas Co.), Dr. Daniel W. Choo (Rohm and Haas Co.), and Dr. Bernard K. Choi (Merck Inc.) for their helpful discussions.

Received for review June 15, 2001. Accepted October 3, 2001. AC0106694

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

6023