Anal. Chem. 1998, 70, 3752-3756
Reduction of Fatty Acid Methyl Esters to Fatty Alcohols To Improve Volatility for Isotopic Analysis without Extraneous Carbon Thomas N. Corso, Betty A. Lewis, and J. Thomas Brenna*
Division of Nutritional Sciences, Savage Hall, Cornell University, Ithaca, New York 14853
Carbon in derivatization groups cannot be distinguished from analyte carbon by chromatography-based high-precision compound-specific or position-specific isotope analysis. We report the reduction of fatty acid methyl esters to fatty alcohols to facilitate high-quality chromatographic separation, without addition of extraneous carbon, with subsequent high-precision position-specific isotope analysis. Methyl palmitate is quantitatively reduced to 1-hexadecanol by LiAlH4 in a one-step reaction. Gas-phase pyrolysis of 1-hexadecanol results in a series of monounsaturated alcohols and r-olefins analogous to fragmentation found for methyl palmitate, as well as an additional peak corresponding to the pyrolytic dehydration product, 1-hexadecene. Carbon isotope analysis of the fragments yielded precision of SD(δ13C) < 0.4‰. Results of position-specific analysis of very low enrichment [1-13C]-1hexadecanol (δ13C ) -4.00‰) showed no evidence of scrambling of the C1 position, and isotope ratios in accord with expectations. The pyrolysis product 1-hexadecene was isotopically enriched relative to 1-hexadecanol, which may cause minor depletion of other pyrolysis products that can be taken into account by routine calibration. The procedure is general and can be extended to compoundspecific and position-specific analysis of moderate molecular weight, low-volatility analytes containing acid groups that would otherwise be blocked with methyl, ethyl, acetyl, or trimethyl silyl groups containing extraneous carbon. Continuous flow (CF) methods introduced in recent years have enabled a dramatic increase in high-precision isotope ratio measurements of chemically pure analytes.1 High-precision determination of isotope ratio requires that the analyte element be converted to a single chemical form for analysis, in part to avoid the inherent practical difficulty and imprecision in exhaustively measuring multiple isotopomer abundances. The most prominent CF method is gas chromatography combustion-isotope ratio mass spectrometry (GCC-IRMS) applied to carbon isotope analysis.2,3 * Correspondence: phone (607) 255-9182, fax (607) 255-1033, e-mail
[email protected]. (1) Brenna, J. T.; Corso, T. N.; Tobias, H. J.; Caimi, R. J. Mass Spectrom. Rev. 1997, 94, 1049-1053. (2) Goodman, K. J.; Brenna, J. T. Anal. Chem. 1992, 64, 1088-1095. (3) Freeman, K. H.; Hayes, J. M.; Trendel, J.-M.; Albrecht, P. Nature 1989, 353, 254-256.
3752 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Analytes are separated from complex mixtures by conventional capillary GC and converted to CO2 by online combustion, dried, and admitted to an IRMS for determination of isotope ratio. Combustion effectively converts all carbon into a single species, regardless of the original analyte structure, and permits the IRMS to be optimized for highest precision analysis of that species. A significant disadvantage of this procedure compared to molecular mass spectrometry is that the IRMS cannot distinguish between carbon originating in the target analyte from that of background or contaminants. As a consequence, it is widely appreciated that baseline resolution of analyte is required for accurate isotope ratio results.4 For sources of extraneous carbon that are independent of analyte concentration, such as column bleed, conventional methods of chromatographic background subtraction perform satisfactorily. Most chemical compounds analyzed by GC contain polar functional groups, making satisfactory chromatography for IRMS analysis difficult. High-polarity stationary phases are now available; however, they generally have lower maximum operating temperatures, leading to long analysis times, broad peaks, and relatively poor resolution. For decades, the routine solution has been to convert polar compounds to less polar derivatives, and volumes are available documenting many effective derivatization agents (e.g., ref 5). Nearly all routinely used derivatizing groups contain carbon. In the special case of IRMS analyses, this extraneous carbon is intrinsically unresolvable from the parent analyte, thus requiring procedures to extract the desired isotope ratio. At least two factors preclude the use of a straightforward measurement of the derivatization reagent’s isotope ratio for this purpose. First, equilibrium and kinetic isotope effects alter the isotopic composition of the derivatization group during the reaction, so that the relationship between reagent and product group isotope ratios is usually unknown.6 Second, the reagent may contain carbon that does not transfer to analyte, making characterization of the transferred group impossible by analysis of the reagent species. Thus, these effects must be measured by coderivatization of a isotopically characterized internal standard, with the analyte isotope ratio extracted arithmetically using mass balance.2 A related issue is that commonly used methyl derivative groups (4) Goodman, K. J.; Brenna, J. T. Anal. Chem. 1994, 66, 1294-1301. (5) Klee, M. S. Modern Practice of Gas Chromatography, 2nd ed.; Wiley: New York, 1985. (6) Rieley, G. Analyst 1994, 119, 915-919. S0003-2700(98)00252-2 CCC: $15.00
© 1998 American Chemical Society Published on Web 08/13/1998
derived from petroleum may be of very low isotope ratio compared to modern terrestrial materials, introducing additional error.2 Finally, calculation of isotope ratio from a weighted average is least satisfactory for small molecules because the number of moles of carbon added in the derivatization can be comparable to the number of moles of carbon in the parent analyte.6 In recognition of these considerations, Tetens et al.7 introduced a one-carbon derivatization for high-precision determination of plasma lactate that substituted a single methyl group, adding a single carbon per molecule, for a trimethylsilyl group that added three carbon atoms per molecule, presented previously by Khalfallah et al.8 A separate factor making derivatization undesirable is that the physical organic mechanism of the derivatization reaction may induce isotopic fractionation in the analyte.6 Thus, for highest accuracy, no carbon should be added to analyte, and any other treatments should be quantitative. As yet, no general strategies for conversion of analytes to volatile analogues without addition of extraneous carbon have been presented. We recently introduced an online strategy for the determination of intramolecular carbon isotope ratios using fatty acid methyl esters (FAMEs) as a test case,9 termed position-specific isotope analysis (PSIA). Purified FAMEs are fragmented pyrolytically, and GCC-IRMS is applied to separate and determine isotope ratio for individual fragments. The derivative methyl group adds only a single carbon per fatty acid, or about 5% total carbon. Previous measurements have indicated that the methyl group may be depleted in 13C to a level of about δ13C ) -50‰ which results in an overall depletion in isotope ratio by -1‰ or more, based on mass balance considerations and assuming the fatty acid is biosynthesized from carbon fixed by a modern plant. More dramatic is the estimated effect on pyrolytic fragments of the analyte as small as C3, where addition of a single derivatization carbon represents as much as one-quarter of the total carbon, leading to a depletion of overall isotope ratio by 5‰. Pioneering early MS work conducted on synthetic peptides employed reduction by LiAlD4 to generate the more volatile polyamine analogues for electron impact ionization.10 Although this approach has been superseded by more convenient derivatization reagents, it offers clear advantages for the isotopic analysis. We report here application of this strategy to convert fatty acid methyl esters to their corresponding alcohols prior to IRMS analysis, to provide suitable chromatography while avoiding the addition of extraneous carbon. We also report the feasibility of online pyrolysis of fatty alcohols for PSIA analysis. EXPERIMENTAL SECTION The online PSIA system consists of an in-house-built GC(I)pyrolysis-GC(II) system coupled to a combustion/high-precision IRMS and an ion trap MS for both high-precision 13C/12C measurements and structure identification, as described previously.9 GC(I) is used for separations prior to pyrolysis, and GC(II) separates pyrolytic fragments for analysis. Injections were made in GC(I) using a split/splitless injector, operated in split (7) Tetens, V.; Kristensen, N. B.; Calder, A. G. Anal. Chem. 1995, 67, 858862. (8) Khalfallah, Y.; Normand, S.; Tissot, S.; Pachiaudi, C.; Beylot, M.; Riou, J. P. Biol. Mass Spectrom. 1993, 22, 707-711. (9) Corso, T. N.; Brenna, J. T. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 10491053. (10) Beimann, K.; Gapp, F.; Seible, J. J. Am. Chem. Soc. 1959, 81, 22-74.
mode. A 50-m × 0.32-mm × 0.52-µm film thickness HP-crosslinked-FFAP capillary column (Hewlett-Packard, Wilmington, DE) separated a test analyte, 1-hexadecanol, from the solvent, ethyl ether. Using an automated rotary valve (Valco Instruments, Houston, TX), the precursor molecule may be directed either to a flame ionization detector (FID) for quantitative analysis and methods development or to the pyrolysis furnace for thermolytic dissociation. The in-house-constructed pyrolysis furnace consisted of one continuous piece of deactivated fused silica, 0.25 mm i.d., resistively heated by a ceramic furnace; the pyrolysis zone was 25 cm in length. The pyrolysis temperature is normally set to a single temperature between 500 and 700 °C, (0.5 °C, using a CN9000A series temperature controller (Omega Engineering, Stamford, CT). Pyrolysis products eluting from the furnace were transferred to GC(II) via a heated transfer line and cryofocused at the column top at -40 °C, followed by separation on a 60-m × 0.32-mm × 0.25-µm film thickness HP-1 (cross-linked methyl siloxane) capillary column (Hewlett-Packard). The GC(II) temperature profile consisted of a linear temperature ramp from -40 to 260 °C at 3 °C/min. After separation, the products were directed via a second rotary valve either to a Varian Saturn III QISMS ion-trap mass spectrometer (Walnut Creek, CA) for structural analysis or to an in-house-built combustion/water-trap/ open-split interface and then to a Finnigan-MAT model 252 IRMS (Bremen, Germany) for isotopic analysis. The combustion furnace consisted of deactivated fused silica filled with oxidized Cu metal held at 850 °C.11 Molecular mass spectra were obtained in positive ion electron impact (EI) mode and in chemical ionization (CI) mode using either methane or ethanol as a CI reagent. Conditions common to all analyses were as follows: electron multiplier voltage, 1.6 kV; emission current, 10 mA, manifold (ion trap) temperature ) 100 °C, axial modulation ) 2.4 V, and acquisition time ) 1 s. Spectral interpretation for structural assignment was augmented using the Wiley mass spectra database (Palisades, Newfield, NY). Searches were conducted using the purity search parameter. Natural abundance 1-hexadecanol, palmitic acid (hexadecanoic acid, 16:0), LiAlH4, and BF3 in methanol (14%) were purchased from Sigma Chemical Co. (St. Louis, MO), and [1-13C]-16:0 was purchased from Cambridge Isotope Laboratories (Cambridge, MA). Ethyl ether was obtained from Fisher Scientific (Pittsburgh, PA). All were used without further purification. Palmitic acid was converted to methyl palmitate (Me16:0) by reaction with BF3 in methanol according to methods described previously.12 The quantitative reduction of Me16:0 to hexadecanol was carried out via the following procedure. LiAlH4 (1.5 molar equiv) in the form of dry powder and about 1 mL of peroxide-free ethyl ether were added to about 2.5 mg of neat analyte in a conedbottom test tube. The reaction mixture was refluxed at 50 °C for 45 min and then cooled on ice. Two drops of distilled, deionized water were added to exhaust excess reagent by conversion of LiAlH4 to solid Li and Al hydroxides. Water must be added carefully, as the reaction with LiAlH4 is rapid, exothermic, and evolves H2 gas. The mixture was centrifuged, and the organic layer was removed and filtered through a 0.2-µm Teflon filter (MSI, (11) Goodman, K. J. Anal. Chem. 1998, 70, 833-837. (12) Sheaff, R. C.; Su, H.-M.; Keswick, L. A.; Brenna, J. T. J. Lipid Res. 1995, 36, 998-1008.
Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
3753
Westboro, MA). Appropriate dilutions were made prior to MS analysis. Notation. Isotope ratios were calibrated against a working standard of CO2 gas ultimately calibrated against NIST RM-22, graphite, and converted to the δ notation, given by
δ13C )
[
]
RSPL - RPDB × 103 RPDB
(1)
where Rx ) [13C]/[12C], SPL refers to the sample, and PDB refers to the international standard PeeDee Belemnite, with RPDB ) 0.011 2372. We previously introduced an alternative to the δ13C notation more useful for samples whose isotope ratios deviate significantly from that of PDB.9 φ13C represents the relative isotope fraction compared to an international standard:
φ13CPDB )
[
]
FSPL - FPDB × 103 FPDB
Figure 1. Isotope ratios of the methyl ester product fragments resulting from the pyrolysis of Me16:0 at 550 °C. The shorter chain methyl ester fragments display a depletion in 13C, probably due to the low 13C content of the extraneous carbon of the methyl group.
(2)
The atom fraction of PDB, FPDB ) 0.011 1123, is readily calculated from RPDB ()0.011 2372) via the formula F ) R/(1 + R). Whereas δ13C refers to the relative difference in isotope ratios compared to a standard, φ13C refers to the relative difference in isotope fractions relative to a standard, in the same units (‰) as for δ13C. The advantages of this notation have been reviewed1 and are principally that, unlike δ13C, (1) no cross-term is required for use of the mass balance equation and (2) φ13C is linearly related to isotope concentration and does not go to infinity at a 13C atom fraction of 1. It is most useful for experiments considering isotope ratios more than 100‰ from δ13C for PDB. Here, both values are reported for the convenience. RESULTS AND DISCUSSION Palmitic acid (C16:0) derived from animal tissue was methylated using the BF3/methanol procedure. The mechanism of this reaction is expected to result in minimal isotopic fractionation.6 The resulting methyl palmitate (Me16:0) was pyrolyzed at 550 °C to yield two product series, R-olefins and a methyl ester monounsaturated at the terminal position, as described previously.9,13 Isotope ratios determined for the methyl esters are plotted in Figure 1 in the φ13C and δ13C notations. Short-chain methyl esters show a distinct depletion in 13C. The smallest fragment, methyl 2-proprenate, is about -35‰, corresponding to a depletion of about -8‰ compared to the longer chain fragments. This depletion is likely due to a very low 13C content of the derivative methyl group, which is a large fraction of small fragments. Optimal fragmentation for PSIA is possible when pyrolytic fragmentation proceeds via breaking of a single carbon-carbon bond, because simple product distributions are observed and because the possibility of potential complications from multiple fragmentation channels is minimized. Reduction of fatty acids or FAME to alcohols opens the possibility of dehydration as a reaction mechanism, since low-molecular-weight alcohols dehydrate under high-energy conditions, as, for instance, under EI (13) Levy, E. J.; Paul, D. G. J. Gas Chromatogr. 1967, 5, 136-145.
3754 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Figure 2. GC-pyrolysis-GC-FID pyrogram resulting from the pyrolysis of 1-hexadecanol at 600 °C. Two series of fragments are observed, an R-olefin series and an ω-unsaturated alcohol series, corresponding to cracking from both ends of the parent molecule. A peak with molecular ion corresponding to loss of 18 Da from 1-hexadecanol due to dehydration is also observed.
conditions.14 Thus, the fragmentation of 1-hexadecanol was assessed at 600 °C; this temperature was chosen to induce the minimal amount of fragmentation consistent with adequate signal for IRMS analysis to avoid secondary pyrolytic degradation.15 Figure 2 is a GC-pyrolysis-GC-FID pyrogram showing the series of fragments formed upon the pyrolysis of 1-hexadecanol at 600 °C. Two series of peaks appear; in analogy to results obtained from pyrolysis of the corresponding FAME, we tentatively identify them as a methyl ester series and an R-olefin series. Confirmation of these assignments was made by ethanol CI and EI mass spectra. For the putative alcohol series, CI analysis of the latest eluting pyrolysis product yielded an MH+ ion at m/z 227, corresponding to a loss of 16 Da (CH2 and 2 H) from the unpyrolyzed 1-hexadecanol (molecular weight 242). The highest mass ions of subsequent members of the series appear at incremental losses of 14 Da, consistent with pyrolytic cleavage along the hydrocarbon chain. Figure 3 shows the mass spectrum of a representative pyrolysis fragment, 13:1, obtained in CI mode. The mass spectral peak at [MH - 18]+ corresponds to the dehydrated molecular ion and is consistent with alcohol structures. These data confirm these peaks to be a monounsaturated primary alcohol series with members C3 (2-propen-1-ol)-C15 (14-pentadecaen-1-ol). (14) McLaffery, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed.; University Science Books: Mill Valley, CA, 1993. (15) Corso, T. N. PhD Thesis, Cornell University, 1998.
Figure 3. Mass spectra of the pyrolysis fragment 12:1 acquired in CI mode (CI reagent, ethanol) with the ion trap MS. The small peak at [MH - 18]+ corresponds to dehydration in the trap and is consistent with alcohol structures.
Figure 5. (a) High-precision isotope analysis for the alcohol series, expressed in φ13C, upon the pyrolysis of [1-13C]hexadecanol at 600 °C diluted with natural abundance 1-hexadecanol. The highest isotope ratio is observed for the shortest fragment (5:1) and the lowest isotope for the longest fragment (15:1), consistent with labeling on C1. (b) φ13C plotted versus the reciprocal of the number of carbons per fragment (1/no. of C). A straight line is observed, with an intercept corresponding to the mean isotope ratio of all positions except the C1 position; the sum and slope correspond to the isotope ratio of the C1 position. Figure 4. Representative ethanol CI mass spectra of the C14 R-olefin obtained with ion trap MS. The [M - H]+ ion is observed at m/z 197, followed by [MH - 14]+ for the product ions.
Figure 4 displays a typical CI spectrum of a member of the other major compound in the series. A strong [MH]+ ion at m/z 197 is observed, as well as an ion series separated by 14 Da, which confirm this compound to be an R-olefin, tetradecene, generated by pyrolysis. Members of this series are ethylene to 1-pentadecene and methane. In addition, a major pyrolysis product is observed at the elution time expected for a C16 R-olefin. The peak has a molecular ion at m/z 225 and yields a mass spectrum similar to those of the other R-olefins in the series. The molecular weight corresponds to loss of water during pyrolysis [MH - 18] and is most likely due to dehydration in the pyrolysis furnace. The two fragment series result from C-C bond-breaking along the chain as described previously,13 and the appearance of a double bond including the terminal carbon is well-known as a stabilization mechanism for free radicals.16 Theory indicates that there should be no significant carbon rearrangement due to the pyrolysis,16 as was confirmed recently for FAME.9 High-Precision PSIA of 1-Hexadecanol. A sample of [1-13C]hexadecanol was diluted with natural abundance 1-hexadecanol, yielding an overall isotope ratio of φ13C ) -3.96‰ (δ13C ) -4.00‰). The results of high precision isotope ratio analysis of the pyrolysis products are plotted in Figure 5. The average (16) Kossiakoff, A.; Rice, F. O. J. Am. Chem. Soc. 1943, 65, 590-595.
precision for both the alcohol series and the olefin series was SD(δ13C) < 0.4‰, which is equivalent to that obtained for previous measurements of methyl ester fragments. The values for the olefin series fall in the natural abundance range, with an average of about φ13C ) -28.69‰ (δ13C ) -29.00‰) for the fragments greater than five carbons. Chromatography for IRMS analysis did not produce baseline resolution for the olefin fragments below five carbons. The isotope ratios observed for these fragments is consistent with loss of the label at the hydroxyl carbon and is strong evidence against significant scrambling of the labeled carbon. A plot of the alcohol series isotope ratios resembles a mixing curve, with the highest isotope ratio φ13C ) 81.59‰ (δ13C ) 82.58‰) for the smallest IRMS measurable fragment, five carbons. Longer chain fragments dilute the label and result in an isotope ratio of about φ13C ) 3.57‰ (δ13C ) 3.61‰) for the largest measurable fragment, 15:1. We have previously shown for methyl ester pyrolysis fragments that a plot of δ13C vs the reciprocal chain length should yield a straight line, with an intercept corresponding to the mean the mean isotope ratio of all positions except C1, while the sum and the slope equals the isotope ratio of the of the C1 position.1,9 The analogous plot for alcohols should be possible. This plot is presented in Figure 5, and shows that the intercept is φ13C ) -34.45‰ (δ13C ) -34.83‰). This value is in reasonably good agreement with the mean isotope ratio for the olefin series (greater than five carbons) of φ13C ) -28.69‰ (δ13C ) -29.00‰), after allowing for some fractionation during pyrolysis. Values for FAME fragments published previously were in closer agreement, Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
3755
leading us to speculate that dehydration product may the source of some fractionation. The isotope ratio of the [MH -18] peak is φ13C ) 7.77‰ (δ13C ) 7.86 ( 0.13‰). Isotope fractionation leading to enrichment of this species implies isotopic depletion in the remaining neutral parent; thus, depletion in the products may be due to this additional reaction channel. Summing the slope and the intercept, as described previously,9 we calculate the hydroxyl carbon position to be φ13C(C1) ) 539.83‰ (δ13C ) 549.23‰), which is also in reasonable agreement with that which we expect from the dilution. Using this value, we can construct a weighted average to the compare with the overall isotope ratio as a further check on the method. We use a mass balance equation of the form
φ13C )
(161 )φ
13
CC1 +
(1615)φ
13
CC2-16
(3)
be easily taken into account by normal calibration. CONCLUSION These data show that chemical reduction can be used to prepare fatty alcohols from the corresponding fatty acid methyl esters for isotopic analysis without addition of carbon. Although we have not tested them, the reduction should be equally effective for free fatty acids.17 The method is directly applicable to CSIA of fatty acids, thus avoiding the current method, in which methyl or ethyl esters are analyzed. This procedure is expected to be particularly useful for isotope variability due to natural processes, requiring comparison with international standards, rather than for tracer measurements, where background isotope ratios are routinely subtracted. Finally, these data demonstrate that controlled online pyrolysis of fatty alcohols produces single-bond breakage, stabilization, and isotopically representative fragments for carbon similar to that of methyl ester PSIA analysis.
Application of this equation yields a calculated isotope ratio of ) 1.44‰, again in good agreement with the measured -3.96‰. Since all high-precision mass spectrometry measurements depend on comparison to standards analyzed contemporaneously, the small degree of fractionation we report here should
ACKNOWLEDGMENT This work was supported by NIH Grant GM49209. T.N.C. acknowledges predoctoral support from NIH Grant DK07158.
(17) March, J. Advanced Organic Chemistry, 4 ed.; Wiley: New York, 1992.
AC9802527
φ13C
3756 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Received for review March 5, 1998. Accepted July 8, 1998.
