Species-selective determination of cobalamin analogues by reversedphase HPLC with ICP-MS detection† Alexei Makarov and Joanna Szpunar* CNRS, EP132, Helioparc Pau Pyre�ne�es, 2 av. Pr. Angot, F-64053 Pau Ce� de�x 9, France. E-mail: joanna.szpunar@univ-pau.fr Received 22nd January 1999, Accepted 12th April 1999 The coupling of reversed-phase HPLC with ICP-MS was optimized for the species-selective determination of cobalt complexes with macrocyclic ligands (vitamin B12 and its analogues).The use of high eYciency microcolumns (33–50 mm) with porous and non-porous small diameter (1.5–2 mm) stationary phases was evaluated and compared with conventional (5 mm packing) HPLC. High concentrations of organic solvent (up to 50% of acetonitrile or methanol ) could be introduced into the ICP using aMeinhard-type nebulizer and a cooled spray chamber. The choice of the column was dictated by the compromise among the analysis time, sensitivity and separation eYciency required.The best results in terms of resolution and sensitivity were obtained using a 15 cm×4.6 mm id C8 column. Absolute detection limits of 80–160 pg (4–8 ng ml-1 in the injected solution) were achieved. The method developed was applied to the determination of the active component in pharmaceutical cobalamin preparations and for monitoring its degradation. the resolution of mixtures of compounds having very similar Introduction structures, such as macrocycles with slightly diVerent func- Cobalt is an essential nutrient and a component of vitamin tional groups.The coupling of RP-HPLC with ICP-MS, B12 (cyanocobalamin).1,2 The status of the latter is an import- however, has attracted less attention than size-exclusion chroant parameter that needs to be evaluated in a number of matography,19 apparently because of the need for long runs clinical samples, such as blood plasma, breast milk, foodstuVs using mobile phases rich in an organic solvent diYcult to and pharmaceutical preparations.Methods for performing handle in ICP-MS. Attempts to alleviate this shortcoming routine analyses for vitamin B12 include the determination of included the use of a cooled spray chamber20–22 or an aerosol the total cobalt by AAS3 or chemiluminescence,4 and a radio- desolvation unit.23–26 immunoassay.5 The classical microbiological assay by Recently, the use of microbore RP-HPLC with post-column Ochromonas malhamensis provides the most reliable, sensitive eZuent dilution prior to direct-injection nebulization (DIN) and specific protozoan method.6 These techniques are not into an ICP-MS was proposed as a method for species-selective cyanocobalamin specific; not only are other cobalamins such analysis of cobalamin analogues.27 The method allowed a gain as adenosylcobalamin (coenzyme B12, 5¾-desoxyadenosyl ), in terms of sensitivity of two orders of magnitude over the methylcobalamin (CH3), hydroxocobalamin (OH) and aquo- previous methods.However, the detection limits were still cobalamin (H2O) co-determined with cyanocobalamin, but above values attainable with radioisotope dilution, the duralso their potentially harmful analogues devoid of enzymatic ation of the chromatographic run exceeded 30 min and activity, such as cobamides and cobinamides. expensive equipment was required. Selectivity in terms of cobalamin species is usually obtained The objective of this work was to develop a method for the by the separation of cobalamin, cobamide and cobinamide species-selective determination of cobalamin analogues by analogues by reversed-phase HPLC.This is followed by the ICP-MS using reversed-phase chromatography and the stano V-line determination of cobalt in the relevant fraction of the dard sample introduction system. Particular attention was chromatographic eluate.7 Despite the attractive figures of given to reducing the duration of the chromatographic run merit, this kind of methodology is too tedious to allow and decreasing the concentration of organic solvent by measurements on a routine basis.Other approaches to speci- employing microparticle microcolumn chromatography, and ation determination of cobalamin analogues included electros- to decreasing the detection limits. pray tandem mass spectrometry8 and HPLC with UV–VIS detection,9–12 atomic absorption (AAS)13,14 or ICP emission spectrometry15 or using a nitroxide radical trap.16 The detec- Experimental tion limits of the on-line methods are at the level of Apparatus several mg ml-1 (as compound), which is insuYcient for many practical applications.The ICP mass spectrometer used in this work was an ELAN Inductively coupled plasma mass spectrometry (ICP-MS) is 6000 (Perkin-Elmer SCIEX, Concord, Ontario, Canada). The an established sensitive and element-selective detection method sample introduction system consisted of a Meinhard nebulizer in liquid chromatography.17,18 Reversed-phase (RP) chroma- combined with a cooled spray chamber (Perkin-Elmer, tography oVers an attractive separation mechanism allowing Norwalk, CT, USA).A Minipuls 3 peristaltic pump (Gilson, Villiers-le-Bel, France) was used to provide make-up flow and also for draining the spray chamber. All the connections were †Presented at the 1999 European Winter Conference on Plasma Spectrochemistry, Pau, France, January 10–15, 1999. made of PEEK tubing (0.17 mm id).J. Anal. At. Spectrom., 1999, 14, 1323–1327 1323For HPLC-ICP-MS experiments using the Meinhard identified by comparison of the retention time of a peak in the sample chromatogram with that of a standard compound. nebulizer with the cooled-spray chamber, an HP Series 1100 pump (Hewlett-Packard, Avondale, PA, USA) using a ICP-MS conditions. The rf power used was 0.9 kW. The Rheodyne (Cotati, CA, USA) Model 7010 injection valve with nebulizer gas flow rate was set at 0.6 l min-1.The X–Y position a 20ml injection loop was used. HPLC experiments with of the ICP torch was adjusted first to maximize the signal for microparticle microcolumns were performed using an ABI cobalt. ICP-MS measurement conditions were optimized daily 140C microbore syringe pump and an ABI Model 112A using a standard built-in software procedure. A tuning solution injection module (Applied Biosystems, Foster City, CA, USA) containing 10 mg ml-1 Co2+ in 60% methanol was used.The with a 1 ml sample loop. An ABI Model 785A absorbance optimum nebulizer gas flow rate was 0.6 l min-1 (1.05 l min-1 detector equipped with a microbore cell was used for HPLC for cross-flow nebulizer). No changes in the signal intensity experiments with UV detection. HPLC-ICP-MS results were were observed up to at least 60% methanol. The spray chamber processed using Turbochrom 4 software (Perkin-Elmer). temperature was 4 °C. Three types of HPLC columns were used: (1) C8, 150 mm×4.6 mm id, 5 mm particle size (Vydac, Hesperia, CA, USA), (2) NPS ODS-II, 33 mm×4.6 mm id, 1.5 mm film Results and discussion particle size (Micra, Northbrook, IL, USA); and (3) TSKgel The HPLC column eYciency is known to increase with Super ODS, 50 mm×4.6 mm id, 2 mm particle size increasing homogeneity of the packing and decreasing packing (TosoHaas, Stuttgart, Germany).particle size.28 Consequently, the smaller and the more homogeneous the packing, the shorter the column can be at a Reagents, standards and samples constant resolution, and the shorter the duration of the Analytical-reagent grade ammonium acetate and acetic acid chromatographic run.Reversed-phase chromatography using were purchased from Aldrich (St. Quentin Fallavier, France). microparticle stationary phases thus represents an attractive Methanol and acetonitrile (Sigma–Aldrich) were of LC grade. potential as a sample introduction technique for ICP-MS, Milli-Q water (18MV) (Millipore, Bedford, MA) was used.allowing a decrease in the cost of speciation analysis. Cyanocobalamin (vitamin B12, CN-Cbl), hydroxocobala- Replacing the conventional 5 mm packing by a 3 mm packing min (vitamin B12a, OH-Cbl), 5¾-deoxyadenosylcobalamin has been demonstrated to allow the rapid speciation determi- (coenzyme B12, Ado-Cbl ), and methylcobalamilco- nation of arsenic by RP-HPLC-ICP-MS.29 The possibility of enzyme B12, Me-Cbl ) were obtained from Sigma (St.Louis, the fast separation of metallothionein–cadmium complexes30 MO, USA). Stock standard solutions were prepared from the and cobalamin analogues31 on columns with a 2 mm porous commercial products by dissolving 25 mg of each of them in stationary phase and a 1.5 mm non-porous packing was evoked 25 ml of distilled water under dim light. They were stored in but no applications of microparticle (1–2 mm) chromatography dark bottles at 4 °C.Working standard solutions were obtained with element-selective detection have been reported. by dilution immediately before the measurements and kept in In this work, we attempted to optimize the coupling of the dark.RP-HPLC with ICP-MS detection with particular attention to The buVer solutions were prepared by dissolving 25 mM post-column peak broadening in the spray chamber and to its ammonium acetate in water (or in 50% organic solvent) and eVect on the chromatographic resolution. Spectrophotometric adjusting the pH to 4.0 with acetic acid.detection was used as a reference because of the minimum post-column peak broadening when a microbore cell is used. Instrumental conditions The optimization was aimed at reducing the analysis time and reducing the organic solvent content in the mobile phase while HPLC conditions. To avoid pressure limitations, acetonitrile retaining the baseline separation of the cobalamin analogues. was used as organic solvent in experiments with the 1.5 mm column and methanol in the other cases.Gradient elution was Choice of the separation conditions used in all separations specific for each column. BuVer A was 25 mM ammonium acetate in water (pH 4.0), buVer B was Analytical column (4.6 mm id ) with conventional (5 mm) packing. The major naturally occurring cobalamins have been 25 mM ammonium acetate in 50% methanol–water (pH 4.0) and buVer C was 25 mM ammonia acetate in 50% aceto- reported to be separated isocratically within 40 min using 30% acetonitrile in water.11 Since acetonitrile at this concentration nitrile–water (pH 4.0).The solutions were de-gassed in an ultrasonic bath. The HPLC conditions are summarized in is poorly tolerated by the plasma, attempts were made to optimize the separation with methanol. Fig. 1(a) shows that it Table 1. The length of the peak tubing connecting the column with the nebulizer did not exceed 20 cm. A calibration curve is possible to separate the cobalamin analogues isocratically using a 25% methanol mobile phase but the time of such was established by the injection of a freshly prepared mixture of hydroxocobalamin, cyanocobalamin, adenosylcobalamin analysis is relatively long (30 min).Shorter runs with the baseline resolution preserved turned out to be impossible in and methylcobalamin in amounts of 0.05, 0.2, 1.0, 2.0 and 5.0 mg ml-1 (each). Compounds in the analysed samples were the isocratic mode. Gradient separation which started at 10% Table 1 Optimum chromatographic conditions Parameter Column type, 4. 6 mm i.d. MICRA NPS II Toso ODS Vydac C8 (33 mm, 1.5 mm) (50 mm, 2 mm) (150 mm, 5 mm) Injection volume/ml 1 5 20 Flow rate/ml min-1 1.0 1.0 1.0 Initial buVer composition (I ) 93% A–7% C 90% A–10% B 80% A–20% B Final buVer composition (F) 83% A–17% C 20% A–80% B 100% B Programme 100% I (1 min) to 100% I to 100% F 100% I to 100% F within 5 min 100% F within 1.5 min within 5 min then 100% F for 1 min Duration of run/min 2.5 5 6 1324 J.Anal. At. Spectrom., 1999, 14, 1323–1327Fig. 2 Gradient HPLC separation of four naturally occurring cobalamins (1 ml, each standard at 2 mg ml-1) at pH 4.0. Gradient: 0–5 min, 20–80% B. TSKgel Super ODS column; bold line, ICP-MS detection, light line, spectrophotometric detection. Peaks 1=hydroxocobalamin; 2=cyanocobalamin; 3=adenosylcobalamin; 4=methylcobalamin. Fig. 3 shows that, indeed, a baseline separation of the cobalamin species is possible within 2 min not only with spectrophotometric detection but also with ICP-MS detection.However, the peak broadening by the spray chamber is significant and the resolution with ICP-MS detection is poorer. It should be noted that methanol-containing mobile phases cannot be used because of excessive pressure build-up on the 1.5 mm columns. An acetonitrile buVer was therefore optimized. Separations were obtained for gradients between 3.5 and 8.5% of acetonitrile; these concentrations are lower than for Fig. 1 HPLC separation of four naturally occurring cobalamins. (a) Isocratic elution (20 ml, each standard at 2 mg ml-1) at 25% methanol methanol. They are also lower in comparison with the concen- (50% B), pH 4.0. Vydac C8 column (4.6 mm id ); ICP-MS detection. trations necessary to obtain baseline separation on an (b) Gradient elution (20 ml, each standard at 2 mg ml-1) at pH 4.0. analytical (5 mm) column. Gradient: 0–5 min, 40–100% B.Vydac C8 column (4.6 mm id); ICP-MS detection. Peaks: 1=hydroxocobalamin; 2=cyanocobalamin; Figures of merit 3=adenosylcobalamin; 4=methylcobalamin. Table 2 summarizes the detection limits for the columns investigated with spectrophotometric and ICP-MS detection. methanol (20% B) was optimized to allow the separation of Values reported elsewhere27 for the coupling of microbore the analyte compounds within 7 min [Fig. 1(b)]. Higher HPLC with ICP-MS using a direct injection nebulizer (DIN) concentrations of methanol in the starting buVer resulted are also given for the purpose of comparison.in a lack of resolution between cyanocobalamin and The lowest detection limits are oVered by analytical adenosylcobalamin. RP-HPLC. Despite the fact that ICP-MS detects only Co, which forms 5% of the total molecule, the detection limits Microcolumns (33–50 mm) with reduced particle size obtained by HPLC-ICP-MS are about 500 times lower than (1.5–2 mm) packing. Replacing the conventional stationary those with spectrophotometric detection. In comparison with phase of porous 5 mm C18 particles by a packing of the previously developed microbore HPLC-DIN-ICP-MS, a 2.29±0.27 mm particles allowed a reduction in the column gain in concentration detection limits of 2–8-fold is observed.length required to carry out the separation to 50 mm. The gradient of the mobile phase for the fastest separation was optimized, similarly as in the case of the analytical column, at 10–40% methanol (20–80% B).Fig. 2 compares chromatograms obtained with ICP-MS (bold line) and spectrophotometric detection (dashed line) under the above conditions. The interface between HPLC and ICP-MS is responsible for considerable peak broadening and a distorted peak shape. Consequently, hardly any shortening of the duration of the chromatographic run could be achieved in comparison with the conventional analytical 5 mm column with ICP-MS detection.A decrease in the duration of the chromatographic run can also be achieved by employing a non-porous packing of a smaller diameter. The advantage of a non-porous packing is the absence of pores, which eliminates all intra-particle diVusion. The sensitivity is significantly improved, but at the Fig. 3 Gradient HPLC separation of four naturally occurring cobala- expense of the sample load. A column length of 33 mm filled mins (1 ml, each standard at 2 mg ml-1) at pH 4.0.Gradient: 0–1 min, with 1.50±0.17 mm particles should oVer, according to the LC 7–17% C. Micra NPS II column; bold line, ICP-MS detection, light theory,28 a similar number of theoretical plates to a 50 mm line, spectrophotometric detection. Peaks: 1=hydroxocobalamin; 2= cyanocobalamin; 3=adenosylcobalamin; 4=methylcobalamin. column with a 2 mm packing. J. Anal. At. Spectrom., 1999, 14, 1323–1327 1325Table 2 Comparison of detection limits (defined as three times the standard deviation of the baseline noise) for cobalamin analogues by RP-HPLC-ICP-MS using diVerent columns and diVerent interfaces ICP-MS with Meinhard nebulizer and cooled spray chamberc Spectrophotometric DIN-ICP-MS:b,c detectiona,b (l=550 nm): Micra NPS TosoHaas ODS Vydac C8 microbore (1 mm id) microbore (1 mm id) (3.3 cm, 1.5 mm) (50 mm, 2 mm) (150 mm, 5 mm) (150 mm, 5 mm) (150 mm, 5 mm) Compound ADLd ng DL/mg ml-1 ADL/ng DL/mg ml-1 ADL/ng DL/mg ml-1 ADL/ng DL/mg ml-1 ADL/ng DL/mg ml-1 OH-Cbl 0.01 0.01 0.04 0.04 0.14 0.007 0.10 0.02 2.9 2.9 CN-Cbl 0.008 0.008 0.03 0.03 0.08 0.004 0.25 0.05 1.2 1.2 Ado-Cbl 0.007 0.007 0.06 0.06 0.10 0.005 0.05 0.01 2.9 2.9 Me-Cbl 0.014 0.014 0.05 0.05 0.16 0.008 0.15 0.03 2.9 2.9 aInvestigated range: 0–100 mg ml-1.bFrom ref. 27. cInvestigated range: 0–5 mg ml-1. dADL=absolute detection limit. The absolute detection limit (0.1 ng) of cobalamin remains the same and matches the detection limits normally determined oV-line by radioisotope dilution.HPLC-ICP-MS is thus a potentially attractive technique for the evaluation of the serum cobalamin status. Microcolumns show a generally poorer analytical performance. With spectrophotometric detection (no postcolumn dead volume), the loss of sensitivity due to the very limited amount of sample injected (1 ml for the non-porous 1.5 mm column and 5 ml for the porous 2 mm column) is compensated in the peak height quantification mode by the decrease in peak width.In the case of ICP-MS detection, as can be seen in Fig. 2, the same absolute amount of analyte injected leads to a two fold smaller signal than with a conventional analytical column. In terms of the detectable analyte concentration this means an eight fold decrease in sensitivity in comparison with the analytical column. For the non-porous 1.5 mm, 33 mm long, columns, the absolute detection limits are similar to those in conventional analytical chromatography.In terms of the detectable analyte concentration, however, a 20-fold loss is observed because the peak broadening in the spray chamber eliminates the gain in sensitivity owing to the decreased peak width observed when microparticle packings were used. Calibration curves are linear over three decades of concentration (0.02–20 mg ml-1) and the analytical precision (standard deviation of five consecutive injections) is 1–2% with UV and 3–6% with ICP-MS detection in all cases.Fig. 4 Chromatograms of a pharmaceutical preparation (diluted ) at Analysis of pharmaceutical preparations pH 4.0. Vydac C8 column (4.6 mm id ); ICP-MS detection. (a) Peaks: 1=Co2+; 2=unknown; 3=hydroxycobalamin; 4=cyanocobalamin; Two commercial pharmaceutical preparations with a known 5=adenosylcobalamin; 6=methylcobalamin. (b) Peaks: 1=Co2+; content of hydroxycobalamin were analyzed by analytical 2=unknown; 3=hydroxycobalamin; 4=cyanocobalamin; 5=aden- (4.6 mm id column) HPLC-ICP-MS.The chromatograms are osylcobalamin. shown in Fig. 4(a) and (b). The chromatogram of the first preparation shows a major signal corresponding to hydroxoco- Conclusions balamin and three others identified as cyano-, adenosyl- and Analytical (4.6 mm id column) RP-HPLC interfaced with methylcobalamins. The concentration determined by using the ICP-MS via a Meinhard nebulizer and a cooled spray chamber calibration curve was 57±5 mg ml-1, compared with the value oVers a fast and sensitive technique for the determination of 60 mg ml-1 given by the manufacturer of the preparation.cobalamin analogues. The method developed is ca. 500-fold The chromatogram of the second solution contains a peak more sensitive than HPLC with spectrophotometric detection of Co2+ eluting with the dead volume and three other signals and finds application when the latter fails because of lack of with retention times of two of them corresponding to hydrosensitivity. The advantages in terms of speed of using smaller xocobalamin and adenosylcobalamin.Peak 2 corresponds to particle size microcolumns are negligible when ICP-MS a compound that could not be identified. The concentration detection is used because of the loss of eYciency due to the determined from the calibration curve was 1.1±0.1 mg ml-1 post-column dead volume (spray chamber). for hydroxocobalamin and 0.5±0.1 mg ml-1 for adenosylcobalamin. The concentration of hydroxocobalamin found was in References good agreement with the value of 1 mg ml-1 given for for this compound by the manufacturer; no independent measurement 1 B-12, Vol. 2, Biochemistry and Medicine, ed. D. Dolphin, Wiley, New York, 1982. was available for adenosylcobalamin. 1326 J. Anal. At. Spectrom., 1999, 14, 1323–13272 P. Gimsing and E. Nexo, in The Cobalamins, Churchill 18 P. C. Uden, J. Chromatogr. A, 1995, 703, 393. 19 A. Makarov and J. Szpunar, Analusis, 1998, 26, M44. Livingstone, Edinburgh, 1983, p. 1. 20 N. P. Vela and J. A. Caruso, J. Anal. At. Spectrom., 1993, 8, 787. 3 K. Akatsuka and I. Atsuya, Fresenius¡� Z. Anal. Chem., 1989, 335, 21 H. Ding, L. K. Olson and J. A. Caruso, Spectrochim. Acta, Part B, 200. 1996, 51, 1801. 4 W. Qin, Z. J. Zhang and H. J. Liu, Anal. Chim. Acta, 1997, 22 Z. Zhao, W. B. Jones, K. Tepperman, J. G. Dorsey and 357, 127. R. C. Elder, J. Pharm. Biomed. Anal., 1992, 10, 279. 5 F. Watanabe, Y. Nakano, E. Stupperich, K. Ushikoshi, 23 H.J. Yang, S. J. Jiang, Y. J. Yang and C. J. Hwang, Anal. Chim. S. Ushikoshi, I. Ushikoshi and S. Kitaoka, Anal. Chem., 1993, Acta, 1995, 312, 141. 65, 657. 24 N. Jakubowski, C. Thomas, D. Stu¡§wer, I. DettlaV and J. Schram, 6 US Pharmacopeia, XXII Revision, US Pharmacopeial Convention, J. Anal. At. Spectrom., 1996, 11, 1023. Rockville, MD, 1989. 25 C. M. Andrle, N. Jakubowski and J. A. C. Broekaert, 7 D. Lambert, C. Adjalla, F. Felden, S. Benhayoun, J. P. Nicolas Spectrochim. Acta, Part B, 1997, 52, 189. and J. L. Gueant, J. Chromatogr., 1992, 608, 311. 26 B. Fairman, T. Catterick, B. Wheals and E. Polinina, 8 H. Chassaigne and R. �©obin¢¥ ski, Analyst, 1998, 123, 131. J. Chromatogr. A, 1997, 758, 85. 9 C. C. Jansen and J. P. De Kleijn, J. Chromatogr. Sci., 1990, 28, 42. 27 H. Chassaigne and R. �©obin¢¥ ski, Anal. Chim. Acta, 1998, 359, 227. 10 H. Iwase, J. Chromatogr., 1992, 590, 359. 28 R. P. W. Scott, High Performance Liquid Chromatography: 11 D. W. Jacobsen, R. Green, E. V. Quadros and Y. D. Montejano, Principles and Methods in Biotechnology, ed. E. D. Katz, Wiley, Anal. Biochem., 1982, 120, 394. Chichester, 1996, pp. 25.94. 12 J. Dalbacke and I. Dahlquist, J. Chromatogr., 1991, 54, 383. 29 S. A. Pergantis, E. M. Heithmar and T. A. Hinners, Analyst, 1997, 13 P. Vin. as, N. Campillo, I. Lopez-Garcia and M. Hernandez- 122, 1063. Cordoba, Anal. Chim. Acta, 1996, 318, 319. 30 H. Chassaigne and R. �©obin¢¥ ski, Fresenius J. Anal. Chem., 1999, 14 P. Vin. as, N. Campillo, I. Lopez-Garcia and M. Hernandez- 363, 522. Cordoba, Chromatographia, 1996, 42, 566. 31 R. �©obin¢¥ ski, H. Chassaigne, I. Rodriguez, A. Wasik and J. Szpunar, J. Anal. At. Spectrom., 1998, 13, 859. 15 M. Morita, T. Uehiro and K. Fuwa, Anal. Chem., 1980, 52, 349. 16 R. J. Kelly, Anal. Commun., 1998, 35, 257. 17 L. Moens, Fresenius¡� J. Anal. Chem., 1997, 359, 309. Paper 9/00633H J. Anal. At. Spectrom., 1999, 14, 13