Negative ion chemical ionization mass spectrometry of pyrrolizidine

Analysis of constituents ofIris rhizomes. Part I: High performance liquid chromatographic-particle beam-mass spectral analysis of iridals fromIris pal...
0 downloads 0 Views 626KB Size
1036

Anal. Chem. 1983, 55, 1036-1040

(8) Straub, K. M.; Burlingame, A. L. Biomed. Mass Spectrom. 1981, 8 , (9) (10) (11) (12)

(13)

431-435. Weber, R.; Levsen, K.; Louter, G. J.; Boerboom, A. J. H.; Haverkamp, J. Anal. Chem. 1982, 54, 1456-1466. Zakett, D.;Schoen, A. E.; Cooks, R. G.; Hemberger, P. H. J. Am. Chem. SOC. 1981. 103. 1295-1297. Thomson, B. A,; Iiibarne, J. V.; Dzledzlc, P. J. Anal. Chem. 1982, 54. 2219-2225. Gross, M. L.; Chess, E. K.; Lyon, P. A.; Crow, F. W.; Evans, S.; Tudge, H. I n t . J . Mass Spectrom. I o n Phys. 1982, 42, 243-254. Hilker, D. R.; Gross, M. L.; Knoche, H. w.; Shlvely, J. M. Bjomed, Mass Spectrom. 1978, 5 , 64-71.

(14) Hilker, D. R.; Knoche, H. W.; Gross, M. L. Biomed. Mass Spectrom. 1979, 6 , 356-358. (16) Shively, J. M.; Knoche, H. W. J. Bacteriol. 1969, 96, 829-830. (16) Knoche, H. W.; Shively, J. M. J. BlOl. Chem. 1969, 224, 4773-4778. (17) Knoche, H. W.; Shively, J. M. J. Bioi. Chem. 1972, 247, 170-178.

RECEIVED for review December 6, 1982. Accepted March 1, 1983. This work was supported by the Midwest Center for Mass Spectrometry, a National Science Foundation Regional Instrumentation Facility (Grant No. CHE 78-18572).

Negative Ion Chemical Ionization Mass Spectrometry of Pyrrolizidine Alkaloids with Hydroxide Reactant Ion Peter A. Drelfuss," Wllllam C. Brumley, and James A. Sphon Division of Chemistry and Physics, Food and Drug Administration, Washington, D.C. 20204

Edward A. Caress Department of Chemistry, George Washington University, Washington, D.C. 20052

Macrocyclic dlester and noncycllc monoester pyrrollzldlne alkaloids were examlned by hydroxlde reactant ion negatlve Ion chemlcal lonlzation mass spectrometry. I n addltlon to abundant (M - H)- ions, the macrocyclic diester alkaloids produced (M OH)- adducts and extenslve fragmentatlon. The fragmentatlon often provlded the masses of the Intact necic acld side chaln and pyrrollzidlne nucleus (neclne). The spectra of the monoester pyrrohldlne alkalolds were slmllar to those of the macrocycllc dlesters, but no (M OH)- Ions were observed. The study was aided by hlgh-resolutlon exact mass measurements, colllslonally Induced dlssoclatlon/mass analyzed ion klnetlc energy spectrometry, and B2/E linked scans.

+

+

Pyrrolizidine alkaloids (PAS) are naturally occurring toxins present in about 3% of the world's flowering plant species (1). These plants are mainly from the families Boraginaceae (all genera), Compositae (tribes Senecioneae and Eupatorieae), and Leguminosae (genus Crotalaria) (2). Major outbreaks of liver disease in South Africa, the Soviet Union, Jamaica, Afghanistan, and India have been correlated with the direct or indirect consumption of these plants ( I ) . PAS have been found in some herbal teas, milk, and honey a t levels equal to, or greater than, those which have induced chronic liver damage and cancer in laboratory animals (2). Chronic effects include carcinogenicity ( 3 ) ,mutagenicity (4),and teratogenicity (5). Perhaps the greatest threats posed by PAS to inhabitants of industrialized countries are the chronic effects that can be produced by continuous low level consumption of PAS. Huxtable et al. (6) have suggested that low level exposure to PAS may be a cause of lung disease. To assess the extent of any public health hazard, methods with high sensitivity and specificity are necessary to detect and identify low levels of PAS (1 ppm and less) present in foods. Mass spectrometry (MS) has played a significant role in the detection and identification of PAS because of the structural information and sensitivity which can be obtained.

Electron ionization (EI) MS (7-10) mainly has been used, but positive ion (PI) chemical ionization (CI) MS employing B2/E linked scans has also been used (11). The ion current produced by E1 and PI CI is mostly restricted to ions characteristic of the pyrrolizidine ring moiety. Therefore, structural elucidation of the side chain(s) is often difficult. To simplify the structure elucidation, PAS are sometimes hydrolyzed or hydrogenolyzed to produce the resulting mono- or dibasic carboxylic acids, known as necic acids, and the pyrrolizidine nucleus amino alcohols, known as necines ( 4 ) . Hydroxide reactant ion (OH-) was first described and applied as a reactant ion for negative ion (NI) CI MS by Smit and Field (12). OH-, a strong Bronsted base, has often been used because of its proton abstraction abilities, but substitution and elimination reactions have also been observed (13). OH- is known to produce a gas phase analogue of base hydrolysis of esters and therefore has significant potential in the structural elucidation of complex esters (12, 14). Here, we present the analysis of PAS by OH- NI CI MS, a single step instrumental alternative to base hydrolysis followed by P I MS techniques. The structural significance of the fragmentation is explained, aided by collisionally induced dissociation/mass analyzed ion kinetic energy (CID/MIKE) spectrometry (15, 16), high-resolution mass measurements, and B 2 / E linked scans (17). EXPERIMENTAL S E C T I O N Standards. All of the PAS were gifts and were used without further purification. Purity was judged to be 99% or higher on the basis of their E1 and OH- NI CI mass spectra. Senecionine was supplied by Russell J. Molyneux, U.S. Department of Agriculture, Western Regional Research Center, Berkeley, CA. Jaconine, jacoline, anacrotine, and lycopsamine were obtained from C. C. J. Culvenor, CSIRO, Parkville, Victoria, Australia. Fulvine, heliotrine, crocandine, and ehretinine were supplied by C. K. Atal, Regional Research Laboratory, Jammu Tawi, India. Heliotrine, rosmarinine, and senecionine were supplied by Abdel-Fattah M. Rizk, National Research Center, Cairo, Egypt. Peak Matching Reference Materials. A low boiling fraction of Fomblin (18) was supplied by Gary A. McCluskey and Mary Shorter of the Frederick Cancer Research Center, Frederick, MD.

0003-2700/83/0355-1036$01.50/00 1983 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 55, NO. 7, JUNE 1983

1037

Table 1. High-Resolution Mass Measurements of Senecionine Ionized by OH- NI CI MS nominal mass 352 215 197 169 154 154f 153 151 135

peak matching reference

m/z 350 (C,F,,) PFK”

exact mass obtained

theoretical exact mass

elemental composition

352.1759 215.0918 197.Of311 169.0873 154.0868 154-0864 153.0908 151.0770 135.0806

352.1760 Cl,H,NO, 215.0919 CIOHlSO, 197.0814 ~lclH13O4 169.0865 C9H1303 154.0868 C,Hl,NOZ 154.0868 C,HlZNO, 153.0915 C9H130, 151.0759 C,HllO, 135.0807 C9HlIO a Perfluorokerosene. Pentafluoroacetophenone. Fomblin oil. Senecionine in these cases was peak matched against itself. e 2.4-Dinitrofluorobenzene. f Duplicate analvsis with different reference. m/z 210 (C,H,F,O) PFAb m / z 185 (C,F,O) FOMC m / z 151 (C,H,,O,) SENd m / z 140 (C,H,FNO,) DNFBe m / z 1 5 1 (C9H,,0,) SEN m / z 1 5 1 (C,H,,O,) SEN m / z 135 (C,H,,O) SEN m / z 135 (C,F,O) FOM

Other reference materials included perfluorokerosene (PCR Research Chemicals, Inc., Gainesville, FL); 2,4-dinitrofluorobenzene (Aldrich Chemical Co., Inc., Milwaukee, WI); and pentafluoroacetophenone (Aldrich). Instrumentation. The low-resolution OH- NI CI mass spectra were obtained on a modified Finnigan MAT 3300F CI mass spectrometer equipped with dual electron multipliers and a conversion dynode (19). Samples were introduced with a direct insertion probe. Methane and nitrous oxide (Matheson Gas Products, East Rutherford, NJ) were 99.99% pure. Nitrous oxide was added to 1.0 torr of methane in the ion source to give a combined source pressure of about 1.2 torr (uncalibrated source pressure gauge). Reagent gases were ionized with a 140-eV electron beam generated from a heated rhenium filament. The conversion dynode voltage was 3600 V. In those cases where conditions of resonance electron capture (REC) were desired (that is, no OH-), the background concentration of OH- reactant ion in the Finnigan 3300F ion source was too high to provide reproducible REC spectra of PAS. Often, the REC spectra were extremely similar to the OH- NI CI spectra and included ions which we consider to be produced only by OHreactant ion. In other instances using nearly identical REC conditions, the characteristic OH- ions were nearly or totally absent and other ions appeared which we did not study. We have attributed these difficulties to the highly reactive nature of the PAS toward OH- and other reactant ion Bpecies, which can be present in the ion source in variable residual quantities. Reproducible IREC spectra were obtained with a VG Analytical ZAB-2F mass spiectrometer. The instrument was equipped with an EI/CI source and an off-axis electron mulltiplier and “deflector”. The deflector for the electron multiplier was set to +4.3 kV and thus served as a conversion dynode. Samlples were introduced with a direct insertion probe. The electron energy was 100 OV and the emission current was 0.5 mA. The resolution was 1000 (5% valley). OH- NI CI high-resolution mass measurements were obtained by use of the VGI Analytical ZAB-2F instrument at 10000 resolution (5% valley) by peak matching. To produce the OH-, nitrous oxide (6 X lo4 mbar) was introduced into the ion source via the capillary gas chromatographic inlet line. Methane was added through the standard CI reagent gas inlet t o produce a combined source pressure of (1.5-2.0) X 10” mbar. The CD/MIKE spectra were obtained on the ZAR-2F instrument by scanning the electric sector with the digital scan unit of the mass spectrometer or with a Finnigan MAT INCOS 2300 data system. Helium collision gas was introduced into the second field free region collison cell, where the pressure was imaintai~nedat 3.5 X lo-’ mbar. B 2 / Elinked scans were obtained with no collision gas in the first field free region.

RESULTS AND DISCUSSION The structures of the 10 alkaloids that were studied are shown in Figure 1. These are representative of the macrocyclic diester (I-VII) and noncyclic monoester (VIII-X) types of PAS. The OH- NI CI spectrum of senecionine (I) is shown in Figure 2 and illustrates many typical spectral features of the PAS. In addition to an abundant (M - H)- ion a t m f z 334 and an ion a t m / z 352 corresponding to (M 17)-, I exhibits

+

I Senecionine. II = ti 11 Anacrotine, R = OH

03

cr3

VIi Fulvine

VI Crocandine

iV Jacollne, R = OH V Jaconine, R = CI

IX Lycopsdmine, R = ti R = CH,

Vlli Ehretinine

X Heliotrine,

Figure 1. Structures of some pyrrolizidine alkaloids. I54

10a 0 7

IM-HI334

100

150

200

250

300

350

niz

Figure 2. OH- NI C I mass spectrum of senecionine. Source temperature was 90 OC.

considerable fragmentation. High-resolution data (see Table I) obtained on m / z 215 were consistent with the formation of the singly ionized dibasic necic acid side chain from which I is composed (see Figure 2). The ion structures shown in

1038

ANALYTICAL CHEMISTRY, VOL. 55, NO. 7, JUNE 1983

Table 11. OH' NI CI Spectra of Pyrrolizidine Alkaloids comsource pound mol temp, no. wt "C 335

I

90

I1

351

95

I11

353

110

IV

369

100

V

387

100

VI

311

100

VI1

309

110

VI11 IX

275 299

100 110

X

313

100

a

Neck acid anion.

(M-H). (M t OH)334 (61.7) 352 (5.3)

m / z (percent relative abundance) NE b other (relative abundance > 1%)

NAa

215 (15.9) 154 (100)

71 (1.5), 73 ( l . O ) , 87 (4.6), 88 (1.5), 116 (5.3),

117 (1.4), 118 (8.4), 135 (34.5), 146 (2.8), 151 (31.0), 152 (2.0), 153 (12.1), 155 (3.4), 169 (6.5), 197 (23.6), 198 (4.7), 199 (1.8), 216 (5.7), 290 (1,3), 306 (1.7), 335 (14.2), 336 (1.6) 350 (8.5) 368 (2.3) 215 (12.0) 170 (100) 59 (12.5), 87 (2.5), 134 (1.9), 1 5 1 (4.0), 153 (9.8), 1 7 1 (7.1), 197 (53.9), 198 (4.6), 199 (1.3), 216 (2.1), 351 (3.4) 352 (27.5) 370 (3.7) 215 (3.1) 172 (100) 59 (1.3), 87 (2.0), 97 (2.0), 98 (1.1), 99 (1.4), 123 (1.2), 135 (10.7), 136 ( l . O ) , 151 (40.8), 153 (47.2), 154 (5.8), 156 (5.0), 169 (85.5), 170 (8.3), 173 (8.1),179 (1.6), 197 (87.5), 198 (9.3), 199 (l.l), 353 (5.4) 368 (17.3) 386 (2.0) 249 (0.3) 154 (100) 60 (l.l), 87 (3.3), 116 (2.5), 118 (2.6), 125 (3.1), 141 (23.2), 155 (7.1), 157 (5.7), 169 (27.3), 170 (2.1), 185 (3.2), 213 (70.7), 214 (7.8), 231 (4.1), 324 (5.5), 369 (5.8), 370 (1.6) 368 (100) 404 (14.3) 267 (0.2) 154 (11.2) 68 ( l . O ) , 79 (20.1), 81 (18.9), 87 (4.0), 118 (2.6), 1 4 1 (1.2), 142 (1.6), 147 (1.9), 169 (6.6), 171 (l.l), 185 (1.8),187 (2.1), 233 (l.l),246 ( l . O ) , 249 (5.2),251 (1.3), 342 (8.3). 343 (2.4), 344 (2.2), 345 (1.5), 350 (24.9), 351 (3.1), 358 (14.2), 360 (3.6), 368 (8.6), 369 (1.3), 387 (58.2), 388 (43.3), 389 (17.8), 390 (1.8), 405 (1.3), 406 (3.8), 407 (1.6), 422 (15.7), 423 (3.1), 424 ( l l . l ) , 425 (1.3), 426 (1.2), 430 (2.1), 432 (2.5) 310 (100) 328 (10.1) 189 (0.4) 62 (2.2), 7 1 (38.3), 73 (12.6), 85 (1.4), 97 (l.l),113 (1.3), 115 (79.8), 116 (2.4), 117 (1.3), 118 (1.2), 127 (17.2), 128 (1.4), 153 (5.8), 171 (4.9), 254 (2.8), 311 (18.1), 312 (1.8),329 (2.1) 308 (65.8) 326 (2.2) 189 (17.9) 154 (2.6) 7 1 (62.8), 72 (2.1), 73 (7.7), 85 (4.1), 87 (7.5), 113 ( l . O ) , 115 (loo), 116 (8.7), 118 (2.6), 125 (1.2), 127 (29.0), 128 (2.1), 153 (8.9), 1 7 1 (1.2), 290 (2.5), 309 (11.5), 310 (1.5) 1 5 1 (100) 140 (1.7) 79 ( 8 . 0 ) , 81 (8.2), 152 (7.7) 274 (2.2) 161 (100) 154 (68.0) 57 ( l . O ) , 99 (1.3), 115 (24.3), 117 (50.1), 298 (9.1) 118 (2.2), 155 (4.9), 162 (5.9), 254 (43.8), 255 (5.6), 299 (1.4) 312 (15.6) 175 (41.6) 154 (100) 115 (2.3), 155 (6.3), 176 (2.5), 313 (3.1) Necine anion.

Figure 2 are intended to demonstrate only the structural moieties of I represented by the fragmentation and not the actual ion structures. To our knowledge this is the first report of the production of an intact side chain ion from a PA or any dilactone with OH- NI CI MS. A CID/MIKE spectrum obtained on the m / z 215 ion produced abundant fragments at energies corresponding to m / z 197, 171, 169, 153, 125, 116, and 87. These data and the high-resolution data in Table I indicate that ions at m / z 215,197,169,153,151,135, and 116 originate from the side chain. The exact mass of m/z 154 (see Table I) was consistent with a necine type anion as depicted in Figure 2. The observation of a necine ion of this type is significant because the hepatotoxicity of the PAS depends on the presence of the double bond in the necine ( 4 ) as occurs in I, 11, IV, V, etc. The production of this ion is unique to OH- since mlz 154 is not observed in the REC spectrum of I (see Figure 3) or other PAS with a similar necine. Under OH- conditions, the extent of fragmentation which normally results from proton abstraction is rather limited (20, 21). This observation has been attributed to the large fraction of the exothermicity of the proton abstraction reaction, which is partitioned into the water molecule which is formed, rather

than into the (M - H)- anion (20). Since I suffers considerable fragmentation under OH- conditions, other types of reactions may also be occurring. The observation of the (M + 17)adduct ion in the OH- NI CI spectrum of I is evidence that this reactant ion can behave as a nucleophile. High-resolution data and CID/MIKE spectra were obtained on the (M + 17)ion. The exact mass (see Table I) was consistent with the elemental composition of an (M + OH)- ion. Major daughter ions in the CID/MIKE spectrum of the (M + 17)- adduct ion at mlz 352 were observed with energies corresponding to m / z 215, 197, 169,154, and 153 and include no significant loss of water from this adduct. These data suggest that the OHinteraction with I is a bond to carbon rather than an attachment to hydrogen or a hydration of the (M - H)- ion. Smit and Field (21) had postulated that an (M + 17)- ion produced from a cyclic compound with OH- NI CI was an (M + OH)adduct, produced by nucleophilic attack of OH- with subsequent ring cleavage. To our knowledge the data reported here are the first verification of this postulate. Several possible sites on I and the other macrocyclic PAS exist for nucleophilic attack, since gas-phase hydrolysis-like reactions of esters may occur either by attack a t a carbonyl carbon or by an S Ntype ~ mechanism (22-24).

ANALYTICAL CHEMISTRY, VOL. 55, NO. 7, JUNE 1983

R"'

+

'1

1039

(M Cl)-, is observed with an isotopic pattern for two chlorines at m/z 422-426. (M C1)- attachment ions are often observed in the NI CI spectra of chlorinated compounds due to the high concentration of C1- present in the source (25). Ions a t m / z 342, 350, and 358 most likely represent losses of CzH40,HCl, and CO from (M - H)-. The ion at m/z 368 could be the result of nucleophilic displacement of a chlorine atom by an oxygen atom anion, (M 0 - C1)- (26). However, a B 2 / E linked scan on m / z 368 has shown that the probable origin of the (M - 19)- ion produced with hydroxide reagent ion is a loss of HC1 from the (M OH)- ion. The ions at m / z 79 and 81 are due to the presence of a bromine-containing impurity. The probable source of this impurity is a bromine-containing homologue of V, as evidenced by a small (M - H)- candidate a t m / z 430-432. The noncyclic monoester PAS are characterized by (M H)-, necine, and monocarboxylate necic acid anions. No (M OH)- adduct ions are observed. For example, ehretinine (VIII) produces an abundant ion at m / z 151 for the monocarboxylate side chain, and ions of lower abundance a t m / z 274 for the proton-abstracted molecular ion and m / z 140 for the necine anion. In the OH- NI CI spectrum of lycopsamine (IX), (M - H).-,the necic acid anion, and the necine anion are present at m / z 298, 161, and 154, respectively. The most abundant decomposition observed in the CID/MIKE spectrum of m/z 161 corresponds to m/z 117. Therefore, the origin of m / z 117, and most likely m/z 254, is the loss of side chain elements totaling 44 amu from the necic acid anion (mlz 161) and (M - H)- ( m / z 298), respectively. A reasonable source of the 44 amu is CzH40,which is lost by cleavage of the glycolic carbon-carbon bond, with hydrogen rearrangement to the charge-retaining fragment. The spectrum of heliotrine (X) is much more simple, consisting basically of (M - H)-, the intact monocarboxylate necic acid anion, and the necine anion at m / z 312, 175, and 154, respectively. There is no loss of 44 amu from either (M - H)- or the necic acid anion because the presence of the methoxy group instead of a hydroxy group blocks this fragmentation pathway. We suggest that an even electron McLafferty type rearrangement may be ocurring in IX. The hydrogen from the secondary glycolic-OH group is rearranged to the carbonyl oxygen with subsequent loss of CH,CHO. This occurs from the even electron (M - H)- and monocarboxylate necic anion. This type of rearrangement would not be possible for X. The application of OH- NI CI to PA MS can provide the basis for new methods of PA analysis. If OH- NI CI MS is combined with chromatographic techniques, the combination may achieve the sensitivity and structural specificity necessary for the detection and confirmation of low levels of PAS anticipated in foods. A simple screening of extracts for the presence of many hepatotoxic monoester PAS and macrocyclic PAS may also be possible since these compounds often produce the ion characteristic of hepatotoxic PAS at m / z 154. CID/MIKE spectrometry and other MS/MS methods can also be important adjuncts to PA structure elucidation with OH- NI CI MS since they can be used to induce and measure subsequent fragmentation from intact necine and necic acid fragments.

+

+

+

'

I , , I:;,,

100

150

, , , ,;I

215 : , 200

l

v-TJ+A

.2 4 6

250

300

350

1111

Figure 3. REC FJI C I mass spectrum of senecionine. Source temperature was 140 OC.

The OH- NI CI spectra of the 10 alkaloids that were studied are presented in Table 11. Compounds 11-VII, like I, are macrocyclic diesters and produce similar OH- NI CI spectra. In addition to (M - H)- and (M + OH)-, all of the macrocyclic diester alkaloids exhibit intact necic acid anions and fragmentation from this part of the molecule. All of these compounds exhibit an ion 18 amu less than the intact necic acid anions. The m / z 197 that was observed in the spectrum of I can arise from a direct loss of water from the m / z 215 side chain. However, pathways from (M + OH)- and (M - H)- also exist, as indicated by the relatively large m / z 197 ion present in the CID/MIKE spectra of these ions from I. The relative abundance of the intact side chain ions is smaller for the macrocyclic PAS that have a saturated necine than for those with an unsaturated necine. For example, the relative abundance of m / z 215 is significantly smaller in the spectrum of rosmarinine (111)than it is in I or anacrotine (II), all of which have the mme necic acid. Similarly, the relative abundance of m / z 189 is much smaller in the spectrum of crocandine (VI) than it is in fulvine (VII). Another structural feature which seems to affect the relative abundance of the intact side chain ion is the presence of an a,@-unsaturatedester in the 12-membered macrocyclic ring PAS. When the a,@-unsaturatedester moiety of I and I1 is replaced with a saturated moiety as in jacoline (IV) and jaconine (V),the relative abundances of the necic acid anions ( m / z 249 and 2!67, respectively) are small. Since the expected mass of the necic acid can usually be determined from the molecular weight, necine mass, and the iiecic acid dehydration fragment, the observation of a low relative abundance for such a n ion may be #animportant clue in structural determinations. The presence of an a,@-unsaturatedester is not a requisite factor for the observation of an intact necic acid side chain in the 11-membered macrocyclic ring alkaloids. The only structural requirement which seems to be necessary for the observation of intact necic acid anions1 in the 11-membered ring alkaloids is an unsaturated necine. For example, the necic acid side chain ion a t m / z 189 is obvious in VI1 but is extremely small in VI. Many of the macrocylic diester PAS lbesides I also produce abundant necine ions. The assignment of m / z 154 as a necine type ion in I, IV, V, and VI1 is further substantiated in the OH- spectra of I1 and 111, where the corresponding ions are now observed a t m/z 170 and 172, respectively. The observation of an abundant necine anion seems to depend on the presence of a L2- rather than an 11-membered macrocyclic ring. For example, tlhe necine anion is absent in VI and is small (2.6%) in VII. The presence of chlorine in V is responsible for some additional spectral features that are worth noting. In addition to (M - H)- and (M f OH)- ions, a chloride attachment ion,

+

ACKNOWLEDGMENT The authors thank those individuals who donated PA standards and reference materials. We also acknowledge the encouragement provided by Samuel Page and Thomas Fazio and the technical assistance of Kevin White and Zhao Min (Visiting Scientist, National Institute of Metrology, Beijing, People's Republic of China). Registry No. I, 130-01-8; 11, 5096-49-1; 111, 520-65-0; IV, 480-76-2; V, 480-75-1; VI, 72855-83-5; VII, 6029-87-4; VIII, 76231-29-3;IX, 10285-07-1;X, 303-33-3.

Anal. Chem. 1983, 55, 1040-1044

LITERATURE CITED Culvenor, C. C. J. I n "Toxicology in the Tropics"; Smith, R. L.; Bababunmi, E. A,, Eds.; Taylor and Francis Ltd.: London, 1980; pp 124-141. Smith, L. W.; Culvenor, C. C. J. J . Naf. Prod. 1981, 4 4 , 129-152. Svoboda, D. J.; Reddy, J. K. Cancer Res. 1972, 32, 908-912. Bull, L. 6.; Culvenor, C. C. J.; Dick, A. T. "The Fyrrollzidine Alkalolds"; North-Holland: Amsterdam, 1968. Green, C. R.; Christie, G. S. Br. J . Exp. fathol. 1981, 42, 369-378. Huxtabie, R.; Ciaramitaro, D.; Eisensteln, D. Mol. fharmacol. 1978 7 4 , 1189-1203. NeunerJehle, N.; Nesvadba, H.; Spiteller, G. Monafsh. Chem. 1965, 96,321-338. Pedersen, E.; Larsen, E. Org. Mass Spectrom. 1970, 4 , 249-256. Rashkes, Y. V.; Abdulaev, U. A.; Yunusov, S. Y. J . Nat. Comp. 1978, 78, 121-135. Deinzer, M. L.; Arbogast, B. L.; Buhler, D. R.; Cheeke, P. R. Anal. Chem. 1982, 5 4 , 1811-1814. Haddon, W. F.; Molyneux, R. J. 28th Annual Conference on Mass Spectrometry and Allied Topics, New York, NY, May 25-30, 1980; Paper No. RAMOA6. Smit, A. L. C.; Field, F. H. J . Am. Chem. Soc. 1977, 99, 6471-6483. Bruins, A. P. "Advances In Mass Spectrometry"; Quayle, A,, Ed.; Heyden: London, 1980; Vol. EA, pp 246-254. Bruins, A. P. Anal. Chem. 1979, 57, 967-972. Kondrat, R. W.; Cooks, R. G. Anal. Chem. 1970, 5 0 , 81A-92A.

(16) Bradley, C. V.; Howe, I.; Beynon, J. H. Biomed. Mass Spectrom. 1981, 8, 85-89. (17) Boyd, R. K.; Beynon, J. H. Org. Mass Spectrom. 1977, 12, 163-165. (18) Bruins, A. P. Biomed. Mass Specfrom. 1980, 7 , 454-456. (19) Hunt, D. F.; Stafford, G. C.; Crow, F. W.; Russell, J. W. Anal. Chem. 1978, 48, 2098-2105. (20) Hunt, D. R.; Shabanowitz, J.; Giordani, A. B. Anal. Chem. 1980, 52, 386-390. (21) Smit, A. L. C.; Field, F. H. Biomed. Mass Spectrom. 1978, 5 , 572-575. (22) Takashima, K.; Riveros, J. M. J . A m . Chem. SOC. 1978, 700, 6128-6132. (23) Bowle, J. H.; Blumenthal. T. J . Am. Chem. Soc. 1975, 9 7 , 2959-2962. (24) Roy, T. A.; Field, F. H.; Lin, Y. Y.; Smith, L. L. Anal. Chem. 1979, 57, 272-278. ~~~. (25) Dougherty, R. C.; Roberts, J. D.; Biros, F. J. Anal. Chem. 1975, 4 7 , 54-59. (26) Dougherty, R. C.; Dalton, J.; Biros, F. J. Org. Mass Spectrom. 1972, 6 . 1171-1181.

RECEIVED for review January 14, 1983. Accepted March 4, 1983. Presented in part at the 30th Annual Conference on Mass Spectrometry and Allied Topics, Honolulu, HI, June 6-11, 1982.

Determination of Picogram Quantities of Uranium in Biological Tissues by Isotope Dilution Thermal Ionization Mass Spectrometry with Ion Counting Detection W. R. Kelly" and J. D. Fassett Center for Analytical Chemistry, National Measurement Laboratory, National Bureau of Standards, Washington, D.C. 20234

A procedure has been developed for the determination of plcogram quantlties of U In blological matrices by Isotope dllutlon mass spectrometry that uses artlflcially produced n3U (SRM 955) as the isotopic splke. The U Is chemically purlfled by anion exchange chromatography and then loaded onto a single anion exchange bead. The bead Is loaded into the mass spectrometer and provides a polnt source for the emlsslon of U+ ions wlth an Ionization efilciency of 0.2%. One-hundred plcograms of U can be measured with a precision of better than 0.5% (2s). At these levels the chemical blank is the llmltlng source of error. For this quantlty of U the blank correctlon and its assoclated uncertainty were typlcally 4 % and 2 % , respectlvely. The procedure was applled to the determination of U In SRM 1577a, bovine h e r , whlch was found to have a mean concentration of 709 f 13 pg of U/g ( t s , 95% confidence ilmit), the lowest certlfled U concentration of any biologlcal SRM.

The presence of parts per million and lower constituents can markedly alter the chemical and physical properties of materials and affect the biochemistry of plants and animals. It is important to develop accurate and precise measurement procedures and to certify standards which can provide the quality assurance to the segments of industry, public health, and environmental monitoring making ultratrace measurements. For this reason, an ultratrace measurement procedure has been developed for the determination of U in biological matrices. The principles utilized here are applicable to other elements and matrices. This article not subject to

U S . Copyright.

Concentration in Reagents Table I. 238U reagent nitric acid (concntd) nitric acid (8 M) nitric acid (0.1 M ) hydrochloric acid (9 M ) water (7/81) water (9/81)

fg/mL 5.4 14 18 20 22 4.5

The toxicology of U has been extensively investigated ( I ) , and in addition to its radiological hazard the element is a potent biotoxin. Dietary intake of U of 1 yg per day (2, 3) and a body burden of 90 M g ( 4 ) have been reported. The current OSHA standard for U or its insoluble compounds is 0.25 mg of U/m3 of air averaged over an 8-h work shift. Because of its biotoxicity, the ability to measure its concentration a t low levels with high precision and accuracy in subgram samples is essential for detecting anthropogenic contamination, establishing natural levels, and detecting small changes in natural levels. High sensitivity measurements of U in environmental and biological matrices have been made by fluorimetry ( 2 , 5 )and neutron activation analysis (3,6, 7). These techniques typically require nanogram amounts of U and have precisions of several percent. A high yield analytical method has been reported for the determination of U in large quantities of bone ash by a spectrometry (8). The mass spectrometric procedure described here has femtogram sensitivity, using thermal ionization with pulse counting detection. The technique is capable of providing simultaneous isotopic and concentration information. The use of high sensitivity mass spectrometry is promoted by the development of clean and efficient dis-

PIublished 1983 by the American Chemical Society