Anal. Chem. 1996, 68, 1888-1894
Continuous-Flow Isotope Ratio Mass Spectrometry Using the Chemical Reaction Interface with Either Gas or Liquid Chromatographic Introduction Yohannes Teffera, Josef J. Kusmierz, and Fred P. Abramson*
Department of Pharmacology, George Washington University Medical Center, Washington, D.C. 20037
A novel method of sample introduction into an isotope ratio mass spectrometer (IRMS) is described. The technique uses the chemical reaction interface (CRI) to convert samples coming from a gas chromatograph (GC) or high-performance liquid chromatograph (HPLC) into CO2 using a microwave-induced helium plasma. Optimization parameters for both GC/CRI/IRMS and HPLC/ CRI/IRMS are described. In both modes of operation, it was possible to obtain 13CO2/12CO2 ratios with standard deviations less than 1‰. Investigation of HPLC/CRI/ IRMS performance at low and high concentrations (0.510 µg) resulted in no significant deviations of the isotope ratios. The ability to differentiate samples of different biological origins was illustrated using chlorophyll a from spinach and algae, where a large difference was observed but good precision was maintained (SD < 0.60‰). The chemical reaction interface mass spectrometry (CRIMS) concept, developed by Markey and Abramson in 1982,1 has grown into a sensitive and selective method for the analysis of labeled compounds.2 Its quantitative capability and high selectivity are analogous to those of tracer methods utilizing radioactivity. CRIMS has been successfully coupled to both GC and HPLC separation systems3-6 and has been used on different types of mass spectrometers, including a research quadrupole,7 a massselective detector,8 and double-focusing magnetic sector instruments.6,9,10 This enables researchers to utilize the various advantages of each mass spectrometer, such as high resolution to separate isobaric interferences.4,9 A possible advantage of the chemical reaction interface is that its chemistry allows both 13C and 15N to be quantified under identical experimental conditions just by choosing the appropriate masses.5,7 With conventional chemical combustion interfaces, interference between CO+ and N2+ at m/z 28 requires a CO2 trap when 15N detection is desired,
while with CRIMS, the NO+ that is generated can be monitored without interference from CO2. The concept of isotope ratio monitoring that was introduced by Sano et al. in 1976,11 and by Matthews and Hayes in 1978,12 has grown increasingly sophisticated, and today it yields highly accurate and precise data. Using multicollector isotope ratio mass spectrometers (IRMS), precision for measurement of δ13C can go down as low as (0.5‰ for small size samples and (0.1‰ for larger size samples.13 [The δ notation refers to the difference in isotope ratio between a standard material and a test sample, expressed in per mil (‰) notation. See Methods for details.] Both natural variations in 13C abundance and very low dilutions of labels can be quantified.14 While there are commercial GC interfaces for IRMS instruments, a major limitation of IRMS experimentation is the shortage of interfaces for HPLC. In this respect, the recent work by Brenna and Caimi is encouraging.15,16 However, a transport interface may not be the ideal way to produce a sensitive, practical HPLC/IRMS combination, i.e., one that could be used without degrading the chromatographic integrity of the analysis while maximizing sample transmission to the IRMS. For studies in pharmacology and toxicology, high sensitivity and chromatographic resolution will be important when examining lowabundance metabolites, particularly covalently modified molecules that require HPLC. In general, using HPLC introduction would greatly enhance the many areas of research where IRMS is routinely used but where the range of samples is restricted to volatile materials. Such areas include geochemistry, nutrition, biomedical, and plant physiology.14,17-19 The coupling of HPLC to CRIMS has already been accomplished using a thermospray nebulizer followed by a countercurrent gas diffusion cell (GDC20 ). This interface, commercialized as the Vestec Universal Interface,21 has a particularly high capacity for solvent removal. While the nebulizer vaporizes most
(1) Markey, S. P.; Abramson, F. P. Anal. Chem. 1982, 54, 2375. (2) Abramson, F. P. Mass Spectrom. Rev. 1994, 13, 341. (3) Chace, D. H.; Abramson, F. P. In Synthesis and Applications of Isotopically Labelled Compounds 1988; Baillie, T. A., Jones, J. R., Eds.; Elsevier: Amsterdam, 1989; pp 253-258. (4) Chace, D. H.; Abramson, F. P. Anal. Chem. 1989, 61, 2724. (5) Abramson, F. P.; McLean, M.; Vestal, M. In Synthesis and Applications of Isotopically Labelled Compounds 1991; Buncel, E., Kabalka, G. W., Eds.; Elsevier: Amsterdam, 1992, p 133. (6) Teffera, Y.; Abramson, F. P.; Vestal, M. L.; McLean, M. J. Chromatogr. 1993, 620, 89. (7) Kusmierz, J. J.; Abramson, F. P. Biol. Mass Spectrom. 1993, 22, 537. (8) Song, H.; Kusmierz, J.; Abramson, F. P.; McLean, M. J. Am. Soc. Mass Spectrom. 1994, 5, 765. (9) Chace, D. H.; Abramson, F. P. J. Chromatogr. 1990, 527, 1. (10) Morre´, J. T.; Moini, M. Biol. Mass Spectrom. 1992, 21, 693.
(11) Sano, M.; Yotsui, Y.; Abe, H.; Sasaki, S. Biomed. Mass Spectrom. 1976, 3, 1. (12) Matthews, D. E.; Hayes, J. M. Anal. Chem. 1978, 50, 1465. (13) Merritt, D. A.; Hayes, J. M. Anal. Chem. 1994, 66, 2336. (14) Hachey, D. L.; Wong, W. W.; Boutton, T. W.; Klein, B. P. Mass Spectrom. Rev. 1987, 6, 289. (15) Caimi, J.; Brenna, J. T. Anal. Chem. 1993, 65, 3497. (16) Caimi, J.; Brenna, J. T. J. Mass Spectrom. 1995, 30, 466. (17) Freeman, K. H.; Hayes, J. M.; Trendel, J.-M.; Albrecht, P. Nature 1990, 343, 254. (18) O’Leary, M. H. Phytochemistry 1981, 20, 553. (19) Brenna, J. T. Acc. Chem. Res. 1994, 27, 340. (20) McLean, M.; Vestal, M. L.; Teffera, Y.; Abramson, F. P. J. Chromatogr., in press. (21) Vestal, M. L.; Winn, D. H.; Vestal, C. H.; Wilkes, J. G. In Liquid Chromatography/Mass Spectrometry; Brown, M. A., Ed.; ACS Symposium Series 420; American Chemical Society: Washington, DC, 1990; pp 215231.
1888 Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
S0003-2700(95)01255-8 CCC: $12.00
© 1996 American Chemical Society
Figure 1. Schematic for CRI/IRMS configurations. The UI represents the Vestec Universal Interface.
of the liquid solvent in creating a particle beam, the GDC removes the majority of the solvent that remains in the vapor phase by the helium streams that pass in opposite directions in the GDC. This degree of solvent removal is necessary for any interface that involves chemical reactions. Little or no band broadening occurs, because the GDC operates at liter per minute flow rates and the CRI operates under a partial vacuum. In this paper we present our first CRI/IRMS experiments. The instrumentation for both GC and HPLC introduction to CRI/IRMS is described, and performance data are presented. EXPERIMENTAL SECTION Chemicals. The isotopically calibrated GC/CRI/IRMS standards, C16D34, C20D42, and C24D50, were generous gifts from Dr. John Hayes, Indiana University. The undeuterated alkanes, C16H34, C20H42, and C24H50, as well as caffeine, were obtained from Sigma Chemicals (St. Louis, MO). The HPLC solvents acetonitrile and methanol, with nonvolatile impurities of e0.1 ppm, were obtained from EM Science (Gibbstown, NJ), while acetone (impurities e0.3 ppm) was obtained from Fisher Scientific (Fair Lawn, NJ). All the water used in these experiments was doubly distilled and filtered through a 0.45 mm filter. The isotopically calibrated heptadecanoic acid standard was a gift from Dr. Tom Brenna, Cornell University. Ribose, testosterone, and chlorophyll a from spinach and algae were also obtained from Sigma Chemicals. Instrumentation. The general configuration of the apparatus used in these experiments is shown in Figure 1. Most aspects of the components of the system have been described before.2,4,6,20 Although it has not been necessary when using the CRI on conventional MS analyzers, limitations in the tolerated gas load of the IRMS required that a split be effected between the CRI and the MS to reduce the incoming gas flow as much as 1:10. The mass spectrometer used was a Finnigan/MAT Delta S IRMS (San Jose, CA) with differential turbopumping and a triple Faraday cup collector. Even with the 1:10 split, collisional broadening in the source required that the “tight” ion source chamber be opened up by breaking the circular washer that connects the chamber to its mounting plate so that only four fragments remained to insulate the four mounting posts. The reaction gas used in all experiments
was oxygen.22 As occurs when SO2 is the reactant gas,2 elemental carbon from the sample is converted to CO2 and CO. To permit tuning and calibration, we constructed a CO2 inlet system (not shown in the figure). It uses a tank of CO2 whose flow is metered by a Granville-Phillips (Boulder, CO) Series 301 variable leak and switched with a pair of solenoid-actuated bellows valves to either admit the gas into the main inlet system or pump the CO2 out using the same mechanical vacuum pump to which the inlet split and the column bypass valve are attached. GC/CRI/IRMS. The GC experiments were performed using a Varian 3300 GC (Walnut Creek, CA) and a DB-5 ms 0.25 mm i.d. × 30 m, 0.25 µm film thickness fused silica column (J & W Scientific, Folsom, CA). A high-temperature gas switching valve (Valco, Houston, TX) was used to divert the column flow from the mass spectrometer during the period of solvent elution. For 2 min after injection, the output from the GC was diverted to a pump-out port, and an equal flow of helium makeup gas was introduced into the CRI (Vestec Mass Spectrometry Products, PerSeptive Biosystems, Houston, TX). This ensured that the CRI and IRMS maintained a continuous condition, while avoiding the overwhelming chemical content of the solvent peak. The hydrocarbon + caffeine mixtures were separated with a temperature gradient of 130-240 °C at 6 °C/min after a 3 min delay. A split/ splitless (15 s splitless) injection of 2 µL was used for all the GC analyses. Data were acquired using Isodat 5.2 software from Finnigan/MAT (San Jose, CA). Modifications to the software included the ability to use the third (m/z 46) data channel to obtain enrichment-only chromatograms.4 HPLC/CRI/IRMS. The heart of this configuration20 is the Universal Interface (Vestec) that couples the CRI to the IRMS. All HPLC experiments were conducted using a pair of Isco Model 260D syringe pumps (Isco Inc., Lincoln, NE). For the flow injection analyses of ribose, the solvent was 1:1 methanol/water at a flow rate of 1 mL/min. The experiments with testosterone to study the effects of sample size on isotope ratio were made using a 4.6 mm i.d., 150 mm long Kromasil C-18 column (EKA NOBEL, Bohus, Sweden) with 100% methanol at a flow rate of 1 mL/min. For separation of testosterone and heptadecanoic acid, the same Kromasil C-18 column was used with 1:1 acetone/ acetonitrile at a flow rate of 1 mL/min. For the separation of chlorophylls, a Brownlee 4.6 mm i.d., 250 mm long, 300 Å pore size Aquapore C-8 column (Applied Biosystems, Foster City, CA) was used with a solvent system of 7:3 methanol/acetonitrile at a flow rate of 1 mL/min. All samples were dissolved in their respective HPLC solvent systems, filtered, and injected using a Rheodyne 7125 valve with a 20 µL loop. For a condensed phase internal standard,23 we selected testosterone. The δ13CPDB* value (see below) of testosterone was found by using a standard of heptadecanoic acid whose isotope ratio had been determined off-line. Mixtures of testosterone and heptadecanoic acid were analyzed on column, and testosterone was found to have a value of -33.94 ( 0.60 (mean ( SD, n ) 8). Calculations and Standardization. Oxidative CRI chemistry generates some NO2 so that the m/z 46 channel cannot be used to measure the 18O content, from which the 17O content can be computed in order to obtain the true 13C16O16O signal at m/z 45. This correction is needed to obtain a proper δ13CPDB value. (22) Markey, S. P.; Abramson, F. P. In Synthesis and Applications of Isotopically Labelled Compounds; Duncan, W. P., Susan, A. B., Eds.; Elsevier Scientific Publishing Co.: Amsterdam, 1983; pp 291-296. (23) Caimi, R. J.; Houghton, L. A.; Brenna, J. T. Anal. Chem. 1994, 66, 2989.
Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
1889
Without this correction, there will be an unknown error in the accuracy of the isotope ratios we compute. Under circumstances where there is little expectation for fractionation of oxygen, this error is estimated to be 1-2 δPDB. Thus, our calculations use the notation PDB*, where the asterisk denotes the absence of this correction. For hydrocarbons, no such error is possible, as the materials do not themselves contain oxygen. The δ13C value is defined as the relative difference in isotope ratio between the sample and a standard, calculated as
δ13C (‰) ) [(RSPL - RSTD)/RSTD] × 1000 where RSPL refers to the 17O-corrected 13CO2/12CO2 ratio of the sample and RSTD refers to the same corrected ratio for the standard. In a primary measurement, the standard is Pee Dee Belemnite (PDB), and all values are based on its isotope ratio. In each subsequent calculation, the known isotope ratio of the appropriate internal standard was used as RSTD, and the isotope ratio of each unknown peak was calculated on the basis of that known value. The integration of each peak used the Isodat software with a slope sensitivity of 1 mv/s, which we found gave the most reproducible values. RESULTS AND DISCUSSION GC/CRI/IRMS. The first attempts to use CRI as an interface for IRMS involved GC introduction, since this is the type of separation method that existing commercial instruments use and the method from which most data on continuous-flow IRMS measurements have been obtained. In addition, the chemistry of CRI has been most thoroughly studied using this type of system. Using GC/CRI/IRMS, several instrumental parameters, such as the length of the transfer tube, the dimensions of the CRI tube, the split ratio after the CRI, the type of reactant gas and its concentration, and the optimum IRMS pressure, were all investigated. The dimensions of the reaction tube were optimized by observing the IRMS signal with changes in inner diameter and length. For the reactant gas used, oxygen, a smaller inner diameter (1/16 in.) and a shorter length (4.5 in.) appeared to give the best results. The dimensions of the transfer line were also optimized by observing the IRMS signal. Narrow inner diameter transfer lines (
