5501
Anal. Chem. 1999, 65, 3501-3504
Development of Techniques for the Isolation of Iron from Biological Material for Measurement of Isotope Ratios by Fast Atom Bombardment Mass Spectrometry David R. Flory, Leland V. Miller, and Paul V. Fennessey' Departments of Pharmacology and Pediatrics, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, C232, Denver, Colorado 80262
Stable isotopes have gained prominence in nutrition and trace element research. Fast atom bombardment mass spectrometry (FAB-MS)measurement of iron isotope ratios has an accuracy of 299.7% with the stable isotopes MFe,68Fe,and bsFe with relativestandard deviations (RSDs)of 10.9%. The isolation of iron from biological matrices can be accomplished in 4 h with almost total removal of isobaricinterferencescaused by s9KOH,"K*H20, r&aOH, and/or 'OCa-H2O. FAB-MS isotope enrichment measurements from this method compare favorably to predicted absorption/enrichment levels.
INTRODUCT10N To study the kinetics of ingested or infused iron, investigations have utilized the radioisotopes of iron, 66Fe and S9Fe. Ethical considerations centered on the use of radioisotopes for human nutrition research have lead investigators to consider the use of stable isotopes to conduct such studies. Fomon and Janghorbani have utilized stable iron isotopes in studies measuring the absorption of orally administered iron and its subsequent incorporation into erythrocytes.112 Their quantitative measurements were performed by comparing the change in the measured ratio of MFePFe to the baseline MFePFe ratio in the feces and blood, after ingestion of the stable isotope. These measurements were done using inductively coupled plasma mass spectrometry (ICPMS), which has become the method of choice for many investigators doing trace metal analysis. Although it is successful in isotope ratio measurements of many minerals, isotope ratio measurements involving MFe are still hampered by unknown isobaric interferences a t atomic weight 58.3 In addition, certain byproducta of the plasma interfere with the measurement of iron isotopes (e.g., ArN, ArO, AraH20) and must be minimized for effective ICP measurement. Thermal ionization mass spectrometry (TIMS) is the method recognized as having the highest precision and accuracy for isotope measurements,' but it is hampered by a slow sample throughput and TI mass spectrometers are expensive dedicated instruments. The technique of fast atom bombardment mass spectrometry (FAB-MS) is an alternative for isotope ratio measure(1) Janghorbani, M.; Ting, B. T. G.; Fomon,
s. J. Am. J . HematoZ.
1986,21,277-288. (2)Fomon,S. J.; Janghorbani,M.;Ting,B.T.G.;Ziegler,E. E.;Rogers, R. R.; Nelson, S. E.; Ostedgarrd, L. S.; Edwards,B. B. Pediatr. Res. 1988, 24 (l), 20-24. (3)Alves, L. C.;Wiederin, D. R.; Houk, R. S. A d . Chem. 1992,64, 1164-1169. (4)Hachey, D.L.; Wong, W. W.; Boutton, T. W.; Klein, P. D. Mass Spectrom. Rev. 1987,6,289-328. 0003-2700/90/0065-3501$04.00/0
Table I. Iron Isotope Abundances of a Natural and an Enriched Sample Fe isotope natural abundance,' % 'We prep abundance) % 54 56 57 58
5.80 91.72 2.20 0.28 55.85c
0.42 15.84 2.05 81.70 67.59
a From ref 13. Values taken from Oak Ridge National Laboratory sample no. 229101. Average atomic weight.
menta. It is an easy-to-use system and can be retrofitted to virtually any mass spectrometer at minimal cost.6 FAB-MS was used, with some initial success,in a past report to measure the enrichment of "Fe in erythrocytes after an oral dose of enriched MFe.6 FAB-MS has been used routinely in our laboratory to accurately measure enrichment of zinc isotopes from fecal, urine, and/or blood samples.7.8 In this paper, we describe the techniques of iron isolation and the results of measuring accurate isotope ratios of iron stable isotopes by FAB-MS. We use as an example the isotope ratio measurement of MFePFe of iron isolated from erythrocytes both before and after a human infusion of enriched MFe.
EXPERIMENTAL SECTION General Procedures. The iron isotope of 58 atomic mass units (We),in the form of isotopically enrichediron oxide (FeOd), was obtained from the Stable Isotope Division of Oak Ridge National Laboratories. The iron preparation used for this work contained 81.7 atom % W e (asreported in Oak Ridge National Laboratoriesdocumentation,batch no. 229101). Values for both the natural isotopeabundancesand enriched isotopeabundances of iron are given in Table I. Hydrochloricand nitric acids used in this work were ultrapure acids purchased from Seastar Chemicals. All nonenriched iron standards were prepared using atomic absorption standard iron from Aldrich Chemicals. Triple-filtereddeionized water (MU-Q filteringsystem, ContinentalWater Systems/MilliporeCorp., El Paso,TX) was used in all stages of material cleaningand reagent preparation. Glassware was not used in this study. All plasticware, sample tubes (LS-4202-1Y),pipet tips (7512-100)(LifeScienceProducts, Denver, CO), and covered tubes (55.524 and 65/793)(Sarsted Inc., Princeton, NJ) were washed with aqua regia (nitric/ hydrochloric acids, 1:3) prior to use. To prepare stock solutions, natural FeClr6H20 and We in the form of FesO,, were dissolved in aqua regia and then diluted with 0.5 N HCl. Enrichment standard solutions were prepared (5) Caprioli, R. M. Anal. Chem. 1990,62,4774\-486A\. (6)Lehmann, W. D.;Fischer, R.; Heinrich, H.C. Anal. Biochem. 1988, 172. 151-169.
(7)Peirce,P.L.;Hambidge,K.M.;Gose,C.H.;Miller,L.V.;Fennessey, P.V. Anal. Chem. 1987,59,2034-2037. (8) English, J. L.; Hambidge, K. M. Clin. Chim. Acta 1988,175,211216. 0 1993 Ametlcan Chemlcal Society
9502
ANALYTICAL CHEMISTRY, VOL. 65, NO. 23, DECEMBER 1, 1993
bon Isolation
le PreDaration Dry 1 ml rbc's at 1lOOC
0.5 ml column of AG-1x8 resin
Wet ash with conc. H N 0 3
Rinse with 3 column volumes' 12N, 6N. 2.5N, and 0.5N HCI with Milli-Q wash Inbetween each normality
1 1
1
1
DW at 6 5 ' ~
Dissolve in 6N HCI
,
Charge resin with 3 volumes 12N HCI
~ p p S ~ aym p l e 4 Wash with 3 volumes 6 N HCI
Extract 2X with ether (Discard water)
Wash with 3 volumes 4N
HCI
Dry ether under N p Elute Fe with O.5N HCI
Dissolve in 6N HCI
1 Apply IOul to FAB probe
*Equal to 1.5 ml
Flgwo 1. Flow chart of the sample preparatlon and iron Isolation
procedure.
by combining the natural iron solution with known amounts of enriched UFe stock. Concentration of standards was confirmed by the u88of flame atomicabsorptionspectrophotometry. Isotope enrichments were confiied by measuring W e P F e ratios and comparing them to the calculated theoretical values. The standards were submitted to the digestion and iron isolation techniques identical to the erythrocyte samples. Human Studies. One subject was studied after informed consent. The W e solution was prepared as iron citrate at a concentration of 38.2 pg/mL,and 11.92 mL was infused into the subject. Bloodsamplee were obtained in heparinizedsyringesat defied times by antecubital venipuncture with a stainless steel needle. The blood was then transferred to hepariniid tubes and spun to separate erythrocytesfrom plasma. If not used on the day of venipuncture,the separated erythrocyteswere stored at -20 OC. Sample preparation and iron isolation are outlined in Figure 1. For iron isolation, 1mL of isolated or thawed erythrocytes was transferred to a polyethylene beaker and dried for 2 h at 110 OC. To the dried erythrocyteswas added 1.5 mL of concentrated HNOs and then the sample was covered and heated at 65 OC for 2 h or until dry. The solid residue was solubilizedby the addition of 1mL of 6 N HC1. Iron was then isolated from other inorganic constituents by f i t concentrating the iron with a diethyl ether extraction@and then by isolation by ion exchange column chromatography.loJ1 Two hundred microliters of the 6 N HCl solution, containing approximately 100 pg of iron (>95% recovery), was used for diethyl ether extraction. Five hundred microliters of diethyl ether were added to the acid solution; the sample was vortexed for 2 min and then spun at 2600 rpm for 5 min. The ether was removed and another extraction was done combining the second ether layer with the fiit. The ether was dried under nitrogen and then the iron was solubilized in 100 pL of 6 N HC1. This solution, containing approximately 95-100 pg of iron (95-100% recovery), was then used for further iron purification. The separation procedure was performed using AG-1x8 resin (acetate form) from Bio Rad Laboratories (Richmond,CA). The resin was soaked in water and then packed into a 0.5-mL column in disposable pipet tips. The packed column was cleaned by (9) Dodson,R. W.; Forney, G. J.; Swift, E.H.J.Am. Chem. Soc. 1936, 176,2573-2577. (10) Kraus, K.A,; Moore, G. E. J. Am. Chem. SOC.19S3,75,1457-1460. (11) Kraus,K.A.; Moore, G. E.J.Am. Chem. Soc. 19113,75,1460-1463.
washing with varying concentrations of HC1 alternated with a water wash until washes were shown to be free from iron by atomic absorption spectrophotometry. The resin was washed with three column volumes of 6 N HCl to charge the resin and then the 100 p L of sample was applied. The columna were then washed with three column volumes of 6 N HC1, followed by three column volumes of 4 N HC1 (both steps to remove other divalent transition metale), and the iron was eluted with 0.5 N HCl. The first 25 pL of the 0.5 N HC1wash was discarded and the next 250 pL, containing 40-60pg of iron (4040% recovery), kept for FAB analysis. Instrumentation. Isotopic analysis was performed on a VG 7070E HF double-focusing maee spectrometer (Fisons VG Analytical, Manchester, U. K.)equipped with an Ion Tech (London,U. K.) atom gun. The standard stainless steel sample target was replaced by one made of pure silver to prevent potential isobaric interferences from the iron, nickel (UNi), and/or chromium (Wr) found in stainless steel. Ten microliters of sample, containing 1.6-2.4 pg of iron, in 0.5 N HC1, was placed on the target and evaporated to dryness with a heat gun. The FAB probe was then inserted into the vacuum lock, left for a few minutes to ensure totd drying, and then introduced into the FAB source. The mass spectrometer was operatedat a resolution of 500 with the f i e tuning done using a standard iron solution monitoring the mFe signal. The atom gun was supplied with xenon gas and operated at from 7.5 to 8.0 kV and 1mA. Ions that were desorbed from the sample surface were accelerated to 6 keV, and masses were analyzed by scanning the magnet in a linear fashion upward over the range of 5140 atomic maas units (amu). For discrimination against polyatomic interferences,'* the secondary ion energy setting on the maee spectrometerwas increased, which decreased the acceleration voltage by 100-125 eV. The ion beam was detected with the electron multiplier operated at 1.6-2.0 kV. The amplified multiplier signal was digitized and averaged with a Tracor Northern TN-1710 digital signal analyzer having an 8200-channel memory. The peak heightsand baseline valuee of all iron isotopesignale were recorded and averaged for 100 scans. After baseline subtraction, ratios were calculated from the peak heighta of W e and q e isotopes. To measure the amount of isotopicallyenriched iron in either infusion or absorption experiments, we use a method similar to that described by Wolfe1' of tracer divided by tracee. In this study we use percent isotope enrichment, which is defied as percent isotope enrichment (% E) = isotopicallyenriched Fe total Fe in the sample "Isotopically enriched Fe" refers to the specific iron isotope preparation added or infused (in our studies UFe). "Total Fe in the sample" is, as implied, the total iron, which includes the enriched iron added. This definition allows us the practical advantage of easily determining both the quantity of the denominator, by measuring the total iron in the sample (using standard AA) and the % E (using a calibration curve and the iron isotope ratio measured by the FAB techniques described in this paper). The fial unknown, "isotopically enriched iron", can then be calculated.
RESULTS AND DISCUSSION Our initial studies, done to determine the accuracy and precision of measuring the different stable isotopes of iron, were hampered by isobaric interferences for the stable isotopee UFe, 57Fe, and MFe in a natural solution. These interferences were shown to be from MKOH, 39K*H20, WaOH, and Wa.H20 leaching out of the glasswareused (Figure 2a). When natural standards (and subsequently ail isolation work) were exposed exclusively to plastic, the calcium and potassium (12) H e m , R. F. K.; P d e n r i e d e r , W. P.; Satkiewicz., F. G. Radiat. Eft. 1973, 18, 199. (13) De BiBvre, P.: Barnes. I. L.Znt. J.Ma88 Spectrom Zon Procecrses 1981,65, 211-230. (14) Wolfe, R. W. Radioactive and Stable Isotope Trrrcers in Biomedicine, 2nd ed.; Wiley-Liae: New York, 1992; Chapter 3.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 23, DECEMBER 1, 1993 100 LL
40
0.28 1
,I,,
ql,,,
70
I
20 10
9503
0.24
Q LL
0.12
\
0.08 -
YI
Q
-
o.;.-/
LL 0.04 a m
0.28 1
0 100
lbl
a c U 0
0.20 -
0.08
-
0.04
-
0.12
50 40 30
ATOMIC MASS UNITS Flguro 2. FAMS spectra of FeCl in 0.5 N HCI In (a)clean laboratory &ssware and (b) clean laboratory plastic. Spectra are averages from 12measurementsof 100scanslmeasurement.Twoto five micrograms of Fe was used for the measurements.
interferences were essentially removed (Figure 2b) and more accurate atomic abundances were seen. The combination of ether extraction and ion exchange chromatography for iron isolation has not been previously cited in the literature. The reasons for combining these methods are that, in our hands, neither technique alone was sufficient to remove all isobaric interferences. The ether extraction is very good for concentrating iron (95-100% extracted); however, with the amounts of calcium and potassium present in blood, enough of both of these can be extracted by the ether to lead to isobaric interferences. The problems are similar with a single ion exchange chromatography step. This technique is very good for separation of iron from other inorganics (Ni, Zn, Cu, etc.) when these ions are in similar concentrations. However, when Ca and K are in excess, complete purification is not obtained. The combination of these two very different chemical separation techniques (partition and ion exchange)leads to samples that are almost completely free of isobaric interferences. Thus, the only limiting step for our isolation and measurement of iron isotope ratios is sample volume and not our analytical techniques. Occasionally, isolated iron samples can still be plagued by low-level Ca and K contamination. In this instance, we can use the instrument to remove the remaining interference. Generally, for discrimination between polyatomic and monatomic ion species, high mass resolution is used. However, with the low amount of analyte generallypresent in biological samples, the loss of sensitivity is often unacceptable. We were able to make accurate (299.7%) and precise (RSDs 50.9%)isotope enrichment measurements with a resolution of only 500. The initial kinetic energy distribution for polyatomicions is a t a lower energy than those of monoatomic ions.I2 When the secondary ion energy setting is increased, the acceleration voltage is decreased independent of the electrostatic sector voltage. Thus, only those ions of higher initial kinetic energy,monatomic ions, will be passed through the energy filter (E sector). This will eliminate any interference of low levels of calcium or potassium polyatomic ions (SgKOH,38K.H20,QOCaOH,and/or @C.HzO)in the mass range of iron while retaining adequate sensitivity.
Table 11. Isotope Fractionation of Iron by FAB-MS isotopes predicted ratio0 mead ratio (range) % difference 0.00305 0.0029 k 0.0004 4.92 @Fe/MFe @Fe/We 0.04838 0.0435 k 0.0002 10.02 MFe/MFe 0.063 24 0.0664 h 0.0004 4.99 a
From ref 13,
Previous studies measuring the ratios of various stable isotopes of iron by FAB-MS have had discrepancies between measured and published isotope ratios for natural iron.8J7 When this has occurred in other laboratories, the differences were generally circumvented by taking advantage of the good precision of the enrichment measurements and the use of standard curves. However, the questions regarding the deviation from the predicted natural ratios remain. There was a consistent discrepancy in our ratio measurementa below what would be expected from a natural and/or enriched sample of iron when the @Fe/MFeratio is measured (Figure 3a). This deviation is reported to be the result of isotopic fractionation, which is inherent to the FAB-SIMS method.15 From data presented on the relationship between fractionation and atomic mass units,'6 iron fractionation would be expected to occur a t 2-3% /amu in ratio measurementa. A study was undertaken to determine the level of fractionation. Iron fractionation was determined to be 2.5 f 0.1 '3% per amu difference in the ratio measured (Table 11). Thus, when 68F e P F e is measured, there should be a 10% decrease from the expected ratio (58 amu - 54 amu = 4 amu, 4 amu X 2.5 '3% / amu = 10%). This 10% correction was applied to measured ratios from our standard curve plotting a @Fe/MFeratio vs percent enrichment (Figure 3b), and the resulting curves of theoretical and corrected measuredvaluesbecome statistically indistinguishable. This correction for fractionation is the same, 2.5 % lamu, at all levels of enrichment and for all iron isotope ratio measurements. The same solutions that were used in measuring ratios for the 58Fe standard curve by FAB-MS were also measured by ICP-MS at Los Alamos National Laboratory. As shown in Table 111,ICP-MS has very good precision when the different standards are measured, but unlike FAB, the ICP-MS data did not show a consistent deviation from the expected (15) Miller, L. V.; Hambidge, K. M.; Fennessey, P. V. J. Micronutr. Anal. 1990, 8,179-197.
3504
ANALYTICAL CHEMISTRY, VOL. 65, NO. 23, DECEMBER 1, 1993
Table 111. Isotope Ratio (UFePFe) Measurements of "Fe Standards by FAB-MS and ICPMS % theor % enrich ratio FAB-MS diff ICPMS % diff 0 0.0455 0.0909 0.5 1.0
0.0483 0.0565 0.0648 0.1350 0.2220
*
0.0435 0.001 -10 0.0513 0.001 -9.3 0.0586*0.0007 -9.6 0.1221 O.OOO9 -9.6 0.2010 0.001 -9.5
*
0.0442 0.0002 -8.5 0.0581 i 0.0003 +2.8 0.0668i0.0004 +3.0 0.1514 i 0.0007 +10.8 0.2662 i 0.002 +16.6
Table IV. FAB-MS Measurements of Fe Isotope Ratios isotope ratio measd ratio % RSD m e a d ratioa/natural ratiob 54/56 54/57 54/58 56/57 56/58 57/58
0.0664 2.256 22.76 34.57 344.8 10.07
0.6 1.0 0.8 0.6 0.9 1.2
1.001 0.781 0.999 0.804 1.003 1.257
*
Adjusted for fractionation. From ref 13.
measurement. Ratio measurements by FAB show the consistent 2.5% fractionation per amu independent of sample concentration, percent enrichment, and day of measurement. The deviations of the ICP-MS measurements ranged from -8.5 to +16.6% over the same range of enrichments. Ratio measurements were done by FAB-MS for all of the stable isotopes of iron (Table IV). Only those measurements involving the stable isotope at atomic mass 57 (S'Fe), were inaccurate when compared to published values. We believe this is the result of a hydride contamination from the "Fe signal ("FeH+) that cannot be correctedby normal published methods.16 To test the accuracy of enrichment measurements,samples of baseline nonenriched erythrocytes (N = 4) were spiked with known amounts of MFe- enriched solution so that the resulting 5% E would be within a range that is physiologically attainable in human infusion studies. The measured ratios (adjusted for fractionation) were not statistically different from the theoretical ratios (data not shown). The ability of our techniquesto confirm previous data from radioisotope infusion studies was tested in a human infusion study. Past studies6 report that, after an iron infusion, the (16)Dagher, F. J.; Lyons, J. H.; Finlayeon, D. C.; Shamsai,J.; Moore, F. D. Adu. Surg. 1965,1,69-109. (17)Self, R.;Fainveather-Tait,S.J.;Portwd,D. E. Anal. Proc. 1987, 24,366-367.
0.040
I-
Z
Human Subject #1
0.032
I
'*
c
0.018
1
I
I
4
;
0.008
w
n 0.000 0
3
6
8
12
15
DAYS POST INFUSION Figure 4. Human s u b j d infused wlth 455.3 pg of -e. Blood (3 rnL) was drawn, and isotope ratlo measurements were done for days 0, 2, 6, 9, and 15 postinfusion. One mHllllter of blood was used for Iron Isolatkn,and ratioswere measuredas In Flgure 1. Predlcted enrichment was 0.019 % E and measured was 0.020% E f 0.002.
maximum amount of iron incorporated in the erythrocyte will be found after about 14 days. Therefore, blood samples were obtained at 0, 2, 6, 9, and 15 days postinfusion. By estimating the iron content in the erythrocyte volume in the subject using previous tested methods,le and by knowing the amount of the MFe infusion, an expected percent enrichment was calculated. As shown in Figure 4, the determined and expected ratios are statistically the same.
CONCLUSIONS We have shown that FAB-MS is a practical, accurate, and precise technique for the measurement of iron stable isotopes in erythrocytes. The overall precision for this technique is 0.9% RSDs, and with the ability to measure enrichments to 0.016% or lower, this technique is applicable to a wide range of studies of iron absorption in humans. The pilot study agreed with published data of iron erythrocyta incorporation in a human subject, and so further work can now be undertaken in studies that previously required radioactive methods.
RECEIVED for review May 19, 1993. Accepted September 10, 1993.
* Abstract published in Advance ACS Abstracts, October 15,1993.
