HPLCâICPMS and Stable Isotope-Labeled Approaches To Assess

Anal. Chem. 2008, 80, 8702–8711

HPLC-ICPMS and Stable Isotope-Labeled Approaches To Assess Quantitatively Ti(IV) Uptake by Transferrin in Human Blood Serum Alejandro Sarmiento-Gonza ´ lez,† Jorge Ruiz Encinar,† Alicia M. Cantarero-Rolda ´ n,‡ † ,† Juan M. Marchante-Gayo ´ n, and Alfredo Sanz-Medel* Faculty of Chemistry, Department of Physical and Analytical Chemistry, University of Oviedo, Juliań Claverıá 8, 33006 Oviedo, Spain, and Plasma-Mass Spectrometry Service, Scientific Research Support Services, University of La Corunã, 15071 La Corunã, Spain Little is known about the effects of titanium found in patients wearing prostheses or about the biochemical pathways of this metal when used as an anticancer drug (e.g., titanocene dichloride). In this work, transferrin has been confirmed as the only carrier protein binding Ti in human blood serum samples by making use of different HPLC protein separations followed by element-specific Ti detection by ICPMS. Besides, isotope dilution analysis has been applied to the quantitative speciation of Ti-Tf in standards and human blood serum samples. Speciesunspecific and species-specific isotope dilution modes have been explored. In the first case, very low Ti-Tf results were obtained even using two different chromatographic mechanisms, anion exchange (20-24%) and size exclusion (33-36%). Surprisingly, no major Ti species except Ti-Tf were observed in the chromatograms, suggesting that Ti(IV) hydrolysis and precipitation as inactive titanium oxide species could take place inside the chromatographic columns. These results demonstrate that chemical degradation of metalloproteins during analytical separations could ruin the sought speciation quantitative results. The isotope dilution species-specific mode, much more accurate in such cases, has been instrumental in demonstrating the possibility of gross errors in final metalloprotein quantification. For this purpose, an isotopically enriched standard of 49Ti-Tf was synthesized and applied to the quantitative speciation of Ti-Tf again. Using this species-specific spike, Ti-Tf dissociation inside the chromatographic columns used could be corrected, and thus, quantitative Ti-Tf binding in serum (92-102%) was observed. In other words, the usefulness and potential of a species-specific isotope dilution analysis approach to investigate quantitatively metal-protein associations, which can be dissociated at certain experimental conditions, is demonstrated here for the first time. Titanium has been long considered as an inert and biocompatible element ideal for biomedical applications such as joint * To whom correspondence should be addressed. Tel.: +34-985103474. Fax: +34-985103125. E-mail: [email protected]. † University of Oviedo. ‡ University of La Corun ã.

8702

Analytical Chemistry, Vol. 80, No. 22, November 15, 2008

replacements in surgical uses. At present, 0.16-0.2% per year of worldwide population receives a total hip joint arthroplasty.1 However, increased concentrations of circulating Ti degradation products derived from such orthopedic prostheses have been recently reported.2 Unfortunately and contrary to popular belief, such released Ti may cause harmful reactions in blood, fibrotic tissues, or osteogenic cells after traveling through the circulatory or lymphatic systems.3 In fact, Ti compounds are being thoroughly investigated as antitumor drugs and some already advanced to phase II clinical trials.4 Of course, such cytostatic treatments imply high Ti concentrations in blood serum (above 1 µg mL-1).5 Therefore, the need for a more careful and systematic characterization of the nature, extent, and clinical implications of increased contents of Ti in biological fluids/tissues is apparent these days. Human blood serum transferrin (Tf), the principal Fe transporter in the blood stream, has been found to exhibit two metal binding sites. Since only ∼30% of the binding sites of human blood serum Tf are occupied by Fe(III) under nonpathological conditions,6 it could be expected that the vacant sites could bind another metal ions. Sadler et al.7,8 demonstrated that Tf can bind strongly to Ti(IV), suggesting that this serum transport protein could be responsible for this metal mobilization to the biochemical pathways in living organisms. In fact, this protein, together with albumin, is considered potentially implicated in Ti(IV) transport in human serum, as recently reviewed by Tinoco et al.9 Ti-Tf binding has been shown by the increase of two sharp bands at 241 and 295 nm (characteristic of phenolate groups generated by the binding of metal ions to Tyr residues) and a broader band at 321 nm (ligand to metal charge transfer). Changes in certain (1) Sargeant, A.; Goswami, T. Mater. Des. 2007, 28, 155–171. (2) Jacobs, J. J.; Skipor, A. K.; Patterson, L. M.; Hallab, N. J.; Paprosky, W. G.; Black, J.; Galante, J. O. J. Bone Jt. Surg. 1998, 80-A, 1447–1458. (3) Wang, M. L.; Tuli, R.; Manner, P. A.; Sharkey, P. F.; Hall, D. J.; Tuan, R. S. Orthopedic Res. 2003, 4, 697–707. (4) Caruso, F.; Rossi, M. Mini Rev. Med. Chem. 2004, 4, 49–60. (5) Korfel, A.; Scheulen, M. E.; Schmoll, H.-J.; Gru ¨ ndel, O.; Harstrick, A.; Knoche, M.; Fels, L. M.; Skorzec, M.; Bach, F.; Baumgart, J.; Sab, G.; Seeber, S.; Thiel, E.; Berdel, W. E. Clin. Cancer. Res. 1998, 4, 2701–2708. (6) Van Campenhout, A.; Van Campenhout, C.; Lagrou, A. R.; Manuel-y-Keenoy, B. Clin. Chem. 2004, 50, 1640–1649. (7) Sun, H.; Li, H.; Weir, R.; Sadler, P. J. Angew. Chem., Int. Ed. 1998, 37, 1577–1579. (8) Guo, M.; Sun, H.; McArdle, H. J.; Gambling, L.; Sadler, P. J. Biochemistry 2000, 39, 10023–10033. (9) Tinoco, A. D.; Earnes, E. V.; Valentine, A. M. J. Am. Chem. Soc. 2008, 130, 2262–2270. 10.1021/ac801029p CCC: $40.75  2008 American Chemical Society Published on Web 10/11/2008

regions observed in the corresponding very complex NMR spectra obtained also seemed to support a Ti-Tf binding. Of course, these Ti-Tf binding studies based on molecular techniques8,10 have been only made in “matrix-free” aqueous solutions containing an ApoTf standard, and Ti-Tf binding investigation in real serum samples demands much more specific protocols. To the best of our knowledge, the only work demonstrating the Ti-Tf complex existence in a real sample of rat serum resorted to using radioactive 45Ti detection and a size exclusion radio-FPLC separation to achieve the required specificity.11 Ti-Tf binding in human blood serum has not yet been proved experimentally. As for “in vivo” experiments in humans, a nonradioactive approach would be desirable; ICPMS offers an advantageous alternative, with unique features including high elemental sensitivity, specificity to the heteroatom (metals, semimetals, or nonmetals), direct isotopic information (ideal for isotope dilution analysis, IDA), and easy online coupling with powerful separation techniques such as HPLC.12 As a matter of fact, ICPMS has been used to study binding patterns of metals such as Al,13 V,14 or Mn15 to Tf in human blood serum samples. However, since these metal-Tf associations may be labile at certain conditions, extreme care must be exercised to maintain the original protein conformation and so the metal-Tf bond stability. To make matters worse, Ti(IV) is a hydrolytic unstable cation, which most easily can precipitate as inactive TiO2 species under the given experimental conditions used in the chromatographic process. Different metal-protein associations may react differently under similar experimental conditions because they can be extremely dependent on the physicochemical parameters and conditions of the medium (e.g., ionic strength, pH, etc.). For example, Quintana et al.15 observed that the Mn-Tf binding was mostly broken whereas Mn coordination to other enzymes was preserved during size exclusion chromatography (SEC) experiments. Unfortunately, the lack of metal-Tf standards has hampered any reliable quantitative speciation study, and the scarce quantitative data presented in the literature have been mostly obtained after total metal determinations in the chromatographic fractions collected16 or by direct comparison of the metal-Tf peak areas with the usual sensitivity observed for the inorganic aqueous standards of the elements injected postcolumn.17 In fact, most of such metal-protein speciation data have been reported without any checking for element chromatographic column recovery. Consequently, quantitative speciation information in this field reported so far should be considered very carefully: first, the chromatographic purity of the fractions/peaks can be doubtful (10) Tinoco, A. D.; Valentine, A. M. J. Am. Chem. Soc. 2005, 127 (32), 11218– 11219. (11) Vavere, A. M.; Welch, M. J. J. Nucl. Med. 2005, 46, 683–690. (12) Sanz-Medel, A.; Montes-Bayoń, M.; Fernańdez Sańchez, M. L. Anal. Bioanal. Chem. 2003, 377 (2), 236–247. (13) Soldado-Cabezuelo, A.; Montes Bayoń, M.; Blanco Gonza´lez, E.; Garcıá Alonso, J. I.; Sanz-Medel, A. Analyst 1998, 123, 865–869. (14) Fernańdez, K. G.; Montes Bayoń, M.; Blanco Gonza´lez, E.; del Castillo Busto, E.; No´brega, J. A.; Sanz-Medel, A. J. Anal. At. Spectrom. 2005, 20, 210–215. (15) Quintana, M.; Coluda, A. D.; Gondikas, A.; Ochsenku ¨ hn-Petropoulou, M.; Michalke, B. Anal. Chim. Acta 2006, 57, 3–574, 172-180. (16) Hallab, N. J.; Jacobs, J. J.; Skipor, A.; Black, J.; Mikecz, K.; Galante, J. O. J. Biomed. Mater. Res. 2000, 49, 353–361. (17) Svantesson, E.; Petterson, J.; Markides, K. E. J. Anal. At. Spectrom. 2002, 17, 491–496.

as more than one metal-containing protein could easily coelute;18 second, the protein and the metal can coelute but this is not a proof of their binding. Finally, stationary and mobile phases used could displace the metal from their protein binding sites (e.g., by denaturating them) leading to false metal distribution patterns.15,19 IDA can be successfully applied to the quantitative determination of heteroatom-containing proteins, provided that the monitored element has at least two isotopes (free of spectral interferences).20 The most common mode of IDA-ICPMS for metals’ quantitative speciation, is the species-unspecific mode21 or postcolumn IDA, used after the separation of the proteins by chromatography. An increasing number of publications can be found in the literature regarding the use of postcolumn IDA with HPLC-ICPMS.22-25 Postcolumn IDA allows the accurate and precise determination of elemental species eluting from the chromatographic column even if the structure of the compounds is unknown. However, it does not compensate for any losses before or during the chromatographic separation. Therefore possible errors during sample preparation steps and columns recoveries have to be considered. An alternative IDA strategy for speciation is the use of speciesspecific spikes.26 The main requisite here is to know the composition and structure of the species of interest (i.e., it will be compulsory to synthesize the corresponding isotope-enriched species or purchase it if commercially available). In any case, the appropriate spiking solution is added to the sample in this case at the beginning of the analytical procedure. If complete mix-up and equilibration is guaranteed between the added enriched species and the naturally occurring one, any artifacts during sample handling, separation, or detection will be compensated (that is, natural and isotope-enriched metal-protein complexes will behave identically) as the isotope ratio will remain constant in subsequent processes. The concept of using isotopically labeled metalloproteins for entire protein quantification has been recently proposed by several authors.27-31 Most of such publications are restricted so far to the synthesis and characterization of the isotopically labeled protein standards,27,28,30,31 but their application and validation for (18) Lobinski, R.; Schaumlo ¨ffel, D.; Szpunar, J. Mass Spectrom. Rev. 2006, 25, 255–289. (19) Sanz-Medel, A.; Soldado Cabezuelo, A. B.; Milacic, R.; Bantan Polar, T. Coord. Chem. Rev. 2002, 228, 373–383. (20) Rodrı´guez-Gonza´lez, P.; Marchante-Gayoń, J. M.; Garcıá-Alonso, J. I.; SanzMedel, A. Spectrochim. Acta B 2005, 60, 151–207. (21) Heumann, K. G. Int. J. Mass Spectrom. Ion Processes 1992, 118/119, 575– 592. (22) Goenaga-Infante, H.; Van Campenhout, K.; Schaumlo¨ffel, D.; Blust, R.; Adams, F. C. Analyst 2003, 128, 651–657. (23) Polec-Pawlak, K.; Schaumlo ¨ffel, D.; Szpunar, J.; Prange, A.; Lobinski, R. J. Anal. At. Spectrom. 2002, 17, 908–912. (24) Ferrarello, C. N.; Ruiz Encinar, J.; Centineo, G.; Garc, J. I.; Fernańdezde la Campa, R.; Sanz-Medel, A. J. Anal. At. Spectrom. 2002, 17, 1024–1029. (25) Wang, Z.; Prange, A. Anal. Chem. 2002, 74, 626–631. (26) Rottmann, L.; Heumann, K. G. Anal. Chem. 1994, 66, 3709–3715. (27) Harrington, C. F.; Vidler, D. S.; Watts, M. J.; Hall, J. F. Anal. Chem. 2005, 77, 4034–4041. (28) Hann, S.; Obinger, C.; Stingeder, G.; Paumann, M.; Furtmu ¨ ller, P. G.; Koellensperger, G. J. Anal. At. Spectrom. 2006, 21, 1224–1231. (29) Del Castillo Busto, M. E.; Montes-Bayoń, M.; Sanz-Medel, A. Anal. Chem. 2006, 82, 8218–8226. (30) Deitrich, C. L.; Raab, A.; Pioselli, B.; Thomas-Oates, J. E.; Feldmann, J. Anal. Chem. 2007, 79, 8381–8390. (31) Hoppler, M.; Meile, L.; Walczyk, T. Anal. Bioanal. Chem. 2008, 390, 53– 59.


8703

EXPERIMENTAL SECTION Apparatus. HPLC separations were carried out using a dualpiston liquid chromatographic pump (Shimadzu LC-10AD, Shimadzu Corp., Kyoto, Japan) equipped with a Rheodyne PEEK stator manual sample injector, model 7125 (Cotati, CA), fitted with a 20-µL PEEK injection loop. In order to avoid metallic contamination, all the HPLC metallic connections were replaced by PEEK connections. A Waters M484 UV-vis absorption detector was used on line with the HPLC equipment. Two chromatographic columns were used. A size exclusion Superdex 200 10/300 GL had a mobile-phase composition of 2 mM bicarbonate + 0.2 mM citrate (pH 7.4) and worked in isocratic mode. An anion exchange Mono Q 5/50 GL had the following mobile-phase compositions: (A) 20 mM tris(hydroxymethyl)aminomethane (TRIS, pH 7.4) and (B) 20 mM TRIS + 250 mM ammonium acetate (pH 7.4) and working in gradient mode (0 min; 0% B, 3 min; 30% B, 6 min; 60% B, 16 min; 60% B, 25 min; 100% B). Both chromatographic columns were purchased from GE Healthcare (Piscataway, NJ). Element-specific detection of Ti was performed using a doublefocusing inductively coupled plasma mass spectrometer (DFICPMS) Element 2 (Thermo Fisher Scientific, Inc., Waltham, MA) working at medium resolution (Rs ) 4000) in order to avoid the spectral interferences on Ti determination, especially important when working in human blood serum samples.33 Isotopes measured and main operating conditions used for ICPMS were as follows: 34S, 44Ca, 46Ti, 47Ti, 48Ti, and 49Ti; integration window, 20%; dwell time per sample, 0.01 ms; number of samples per peak, 20. A peristaltic pump Minipuls 2 (Scharlab, Barcelona, Spain) and a T-piece were used to continuously mix the eluent from the chromatographic column with the 49Ti isotope enriched solution in the species-unspecific IDA experiments. Incubations of ApoTf standard and human blood serum samples with natural or 49Ti-enriched Ti were accomplished in a digital control immersion thermostat (model Digiterm 100, J.P. Selecta, Barcelona, Spain). The 49Ti-Tf solution obtained by incubation was thoroughly purified using a Microcon centrifugal filter device (cutoff MW 10000 Da) and a centrifuge MiniSpin Plus from Eppendorf AG (Hamburg, Germany). Before injection in the chromatographic system, all samples were filtered using syringe filters 0.22 µm (Millipore, Bedford, MA). All dilutions were made

using ultrapure water (18.2 MΩ cm) obtained from a Milli-Q water purification system (Millipore). Reagents and Materials. All chemicals were of analytical grade quality. Human ApoTf (g98%), potassium pyrosulfate (99%), sodium citrate, and ammonium bicarbonate were purchased from Sigma-Aldrich (Steinheim, Germany). Ammonium acetate Trace Select (g99.995%) was from Fluka (Buchs, Switzerland). Sodium chloride Suprapur (99.99%), TRIS, glacial acetic acid, ammonia solution (32%), and sulfuric acid (95-97%) as well as a 1000 mg L-1 Ti ICP standard from Merck (Darmstadt, Germany) were also used. Mobile phases for HPLC were prepared by dilution of the solid salts. Human blood serum samples were kindly provided by the Laboratory of Biochemical Analysis of the “Hospital de San Agustin” of Avile´s (Asturias, Spain). In order to load Tf with a physiologically available Ti(IV) species, solid natural Ti-citrate was synthesized from TiCl4 purchased from Fluka, (Buchs, Switzerland). An isotopically enriched 49Ti-citrate solution was also synthesized from 49TiO2 (71.17% enrichment, Cambridge Isotope Laboratories, Andover, MA). Procedures. Synthesis of the Natural Abundance Ti-Citrate. Citrate was selected as chelator because it is already present in human serum samples and its coordination complex with Ti(IV) is known to be stable at physiological conditions.34 A noncommercial solid natural Ti-citrate with a 1:3 stoichiometry, (NH4)[Ti(H2Cit)3] · 3H2O, was synthesized following the Deng et al.35 procedure. For such purpose, 1.9 g (10 mmol) of Ti(IV) tetrachloride and 6.3 g (30 mmol) of citric acid monohydrate were mixed and the pH of the solution was adjusted to 2.0 with the slow addition of dilute ammonia. The solution was continuously stirred and slightly heated until a white microcrystalline material started to form. Then, the precipitate was collected and recrystallized in hot water. The Ti(IV) final reaction yield was 80%, as determined by ICPMS using standard additions. The chemical characterization was performed by NMR, IR (Figures S-1 and S-2 in Supporting Information, respectively) and elemental analysis (Table S-1). The results observed were in complete agreement with those reported by Deng et al.35 Fresh Ti-citrate (pH 7.4) solutions were daily prepared by dissolving an adequate amount of that solid in a physiological medium, which consists of 50 mM TRIS and 150 mM NaCl Suprapur (pH 7.4). ApoTf Standard Incubated with Ti-Citrate. Two different Ti/ Tf molar ratios were initially studied with ApoTf standards: a 1:1 and 2:1 Ti/Tf molar ratio (R ) 1 and R ) 2), where one and the two Tf metal binding sites should be occupied, respectively. For this purpose, a solution of ∼31 µM (typical Tf concentration in human serum samples) ApoTf were diluted 1:1 with a pH 7.4 model solution, which consists of 20 mM ammonium bicarbonate, 150 mM NaCl, and 300 µM sodium citrate. Afterwards, this solution was spiked with the appropriate amount of Ti-citrate. The solution was finally thoroughly stirred and incubated for at least 12 h at 37 °C. Human Blood Serum Sample Incubation with Ti-Citrate. The same procedure used for the incubation of natural Ti-citrate with ApoTf standards was used for the incubation of serum samples

(32) Sanz-Medel, A. Anal. Bioanal. Chem. 2008, 390 (1), 1–2. (33) Sarmiento Gonza´lez, A.; Marchante-Gayoń, J. M.; Tejerina Lobo, J. M.; Paz Jimeńez, J.; Sanz-Medel, A. Anal. Bioanal. Chem. 2005, 382, 1001–1009.

(34) Collins, J. M.; Uppal, R.; Incarvito, C. D.; Valentine, A. M. Inorg. Chem. 2005, 44, 3431–3440. (35) Deng, Y. F.; Zhou, Z. H.; Wan, H. L. Inorg. Chem. 2004, 43, 6266–6273.

real quantitative analysis are still pending. Only recent work using HPLC-ICPMS with both “postcolumn” and “species-specific” isotope dilution addressed a real sample application, demonstrating unprecedented accuracy in the determination of individual glycoforms of Tf in human serum.29 Herein we report the first heteroatom-tagged proteomics32 investigation where the use of an isotopically labeled metalloprotein (49Ti-Tf), previously synthesized, is demonstrated to be essential to assess the efficiency of the whole procedure and reveal artifacts in the analytical steps used. Then, accurate quantitative results of the metallobiocompound can be obtained, and so reliable experimental information of metal uptakes by proteins in real samples can be investigated.

8704


replacing the 0.5 mL of the ApoTf standard by 0.5 mL of a pool of human basal serum samples. Synthesis of the 49Ti Spikes. Synthesis of the 49Ti-Citrate Spike. Isotopically enriched 49Ti was only commercially available as 49 TiO2 powder. In order to obtain the 49Ti-citrate and the 49Ti-Tf spikes, the 49TiO2 powder had to be dissolved. For this purpose, 0.02 g of 49TiO2 was melted with 0.2 g of potassium pyrosulfate in a platinum crucible using a Bunsen burner. Once the crucible cools down, two drops of concentrated sulfuric acid were added and the solid was melted again. This operation was repeated twice. Finally, after the crucible cools down again, it was placed in a boiling solution of 0.42 M citric acid (Ti/citrate ratio in this solution was 1:100) in order to avoid any possible Ti(IV) precipitation. The final solution was filtered through a filter paper to separate any nonmelted 49TiO2 from the enriched 49Ti-citrate solution. Chemical characterization of the synthesized spike was carried out using NMR and IR. Results in agreement with those reported previously by Deng et al.35 were observed. Isotopic characterization of the 49Ti-citrate spike was performed by Ti isotope ratio measurements at medium resolution using the DFICPMS (Rs ) 4000). A 24-ns dead time correction was automatically applied, and a negative mass bias factor of 2% per mass unit was typically obtained (by measuring a 25 µg L-1 natural Ti solution) and corrected as well. Synthesis and Characterization of a 49Ti-Tf Spike. The 500 µL of a solution of ∼31 µM ApoTf was obtained by dissolving the appropriate amount of the solid in a physiological medium, which consists of 50 mM TRIS and 150 mM NaCl Suprapur (pH 7.4). This solution was diluted 1:1 with the model solution used for incubations, and the ApoTf was saturated with the 49Ti-citrate previously synthesized (following the same incubation procedure). After a 12-h incubation at 37 °C, possible low molecular weight impurities were removed by filtering through 0.22-µm syringe filters, and then, 400 µL of the solution was centrifuged through a Microcon centrifugal filter device (YM-10) at 10000g for 15 min. The protein fraction (retained in the upper side of the filter) was reconstituted with 300 µL of the physiological medium previously used in the incubation and then centrifuged again. This operation was repeated two further times. Finally, the 49Ti-Tf complex was reconstituted in the desired volume of the model solution by inversion of the Microcon tube and a brief (3 min) centrifugation at 1000g. Isotopic characterization of the 49Ti-Tf spike so obtained was then performed by isotope ratio ICPMS measurements at medium resolution using the DF-ICPMS (Rs ) 4000). Dead time and mass bias were corrected as explained above. This 49Ti-Tf complex was freshly prepared and characterized every day of analysis. Ti Isotope Ratio Computation during HPLC-ICPMS Analysis. An adequate total analysis time per chromatographic point (0.91 s) was selected in order to monitor correctly the transient signals obtained. Mass bias was computed assuming the exponential model every five samples by nebulizing a Ti ICP standard in order to compensate for possible mass bias drift. Negative mass bias factors in the range 1.5-2.5% per mass unit were usually obtained, and they were stable along the analysis time. Ti-Tf Quantification in Tf Standards and Human Blood Serum Samples by HPLC-ICPMS Using Species-Unspecific IDA. The eluent of the chromatographic column was continuously mixed

with a solution containing the previously synthesized 49Ti-citrate (at a flow of 150 µL min-1). The intensity chromatograms (counts s-1) were converted into mass flow chromatograms (ng min-1) using the online isotope dilution equation described in detail elsewhere.24 The amount of Ti was obtained in each chromatographic peak by integration using Origin 7.5. Excellent agreement was obtained between quantitative results obtained using every Ti isotope ratio (46Ti/49Ti, 47Ti/49Ti, and 48Ti/49Ti) in the mass flow chromatograms. Ti Bound to Tf Quantification in Tf Standards and Human Blood Serum Samples by HPLC-ICPMS Using Species-Specific IDA. Incubated standards and serum samples initially analyzed by postcolumn IDA were also analyzed by species-specific IDA. For this purpose and following similar procedures found in the literature,36 an aliquot of the isotopically enriched 49Ti-Tf standard was thoroughly mixed with 100 µL of the incubated serum sample and let in the thermostatic bath at 37 °C again for 5 min. Finally, the mixture was injected into the HPLC-ICPMS. Isotope ratios were always computed as peak area ratios and were corrected for mass bias. Again, excellent agreement was obtained between the quantitative results obtained using 46Ti, 47Ti, and 48Ti isotopes as reference. RESULTS AND DISCUSSION Uptake of Ti by Tf under Physiological Conditions. The indicated purity (>98%) of the commercial ApoTf standard used in the model solution was confirmed by 1D-GE and ESIMS analysis (see Figures S-3 and S-4 in the Supporting Information). Ti(IV) binding to ApoTf was initially studied in vitro using this model solution containing ApoTf at the concentration level present in human blood serum (∼31 µM). For this purpose, 2 mol equiv of the previously synthesized natural Ti-citrate was added to this model solution at 37 °C. The solution was incubated for 12 h and then injected into the anion exchange column. Protein and Ti profiles were followed monitoring online the UV (280 nm) and the ICPMS (Ti) signals. Anion exchange chromatography was initially selected13 as it allows Tf isolation from the other major proteins present in human blood serum (IgG and albumin) while maintaining a physiological pH. The chromatogram obtained is shown in Figure 1A. As can be observed, Ti and Tf peaks (280 nm) coelute, suggesting Ti to Tf binding. The two peaks observed in the element and molecular profiles correspond to Tf isoforms and have been previously described in the literature for other metal-Tf bindings.13,14 Stoichiometry of Ti-Tf binding was investigated by titration of the Tf solution with Ti-citrate. The incubated samples were injected into the HPLC system provided with integrated elemental (Ti, ICPMS) and molecular (UV, 321 nm, a ligand to metal-transfer band) detection. Results obtained in both experiments are also plotted in Figure 1B. As can be seen, ICPMS and UV data were virtually identical, confirming the binding of 2 Ti(IV) for 1 mol equiv of Tf. These results are in agreement with those reported by Guo et al.8 using UV measurements directly in the incubated samples. Also, the same chromatographic pattern was maintained as the molar ratio Ti/Tf increased. This finding indicates that the two peaks in Figure 1A, as for other metal-Tf complexes,19,29 are due to Tf isoform occurrence (different content of oligosac(36) Ruiz Encinar, J.; MonterdeVillar, M. I.; Gotor Santamarıá, V.; GarcıáAlonso, J. I.; Sanz-Medel, A. Anal. Chem. 2001, 73, 3174–3180.


8705

Figure 1. (A) Anion exchange chromatogram obtained for the Tf standard model solution incubated with Ti-citrate. Solid and dashed lines correspond to 47Ti (ICPMS) and 280 nm (UV), respectively. (B) In vitro study of the Ti/Tf stoichiometry and Tf saturation by Ti(IV) added as natural Ti-citrate. Increases in peak areas are plotted against the Ti/Tf molar ratio. Triangles and circles correspond to peak areas obtained using HPLC with UV (321 nm) and ICPMS (47Ti) detections, respectively.

charides and sialic acids). As expected, the use of HPLC-ICPMS seems most useful to study Ti-Tf binding in real human blood serum samples. That is, the high Ti specificity offered by ICPMS detection is unrivaled by UV-vis spectroscopy detection (where many metals37 can bind to Tf increasing the 321-nm ligand to metal charge-transfer band). Ti-Tf Binding in Human Blood Serum Samples. As Ti concentration in human blood serum samples at basal levels is extremely low,2 the studied human blood serum samples were previously incubated with controlled amounts of natural Ti-citrate (at levels around those expected in implanted or treated with cytostatic drugs patients).5 In order to ascertain if Tf is the only Ti binding protein in human blood serum, Ti speciation was then accomplished by anion exchange chromatography. ICPMS is an adequate detector not only for metals but also for important heteroatoms such as phosphorus and sulfur.38 In this vein, the 34 S of methionine and cysteine residues present in the amino acid (37) Harris, W. R.; Messori, L. Coord. Chem. Rev. 2002, 228, 237–262. (38) Sanz-Medel, A.; Montes-Bayoń, M.; Fernańdez de la Campa, M. R.; Ruiz Encinar, J.; Bettmer, J. Anal. Bioanal. Chem. 2008, 390 (1), 3–16.

8706


chains has been used to follow the general protein profile by ICPMS. As can be seen in the chromatogram obtained (Figure 2A), three different regions can be observed: one corresponding to serum inmunoglobulins (peaks at 1 and 5 min) where no Ti signal was observed. Then, two peaks corresponding to Tf and containing all the Ti present appear in the chromatogram at 8-12 min. It is interesting to note that fractions corresponding to these two Ti-containing peaks were collected and analyzed by MALDIMS. Such spectra were compared with those obtained for the model solution (Figures S5 and S6 in the Supporting Information), demonstrating that Tf was the very dominant protein in both fractions. Finally, no Ti was associated with the large albumin peak eluting at 23-28 min. This Ti profile obtained in real human blood serum samples turned out to be identical to that obtained for the Tf model solutions (see Figure 1A), demonstrating once again the exceptional capabilities of ICPMS for the specific detection of metals even in very complex samples. In principle, Ti-citrate and Tf could simply coelute in the used anion exchange column. To further prove this chemical association, we resorted to an alternative SEC mechanism to perform such analysis at physi-

Figure 2. Chromatograms obtained for an incubated human blood serum sample (Ti-Tf molar ratio 2) using (A) anion exchange chromatography. (B) Size exclusion chromatography. (47Ti signal, dot line; 49Ti signal, solid line; 34S signal, dash line.)

ological pH of 7.4. As can be observed in Figure 2B, most of the Ti eluted in a single broad peak at 10 mL (it is known that Tf cannot be isolated from the other major blood serum proteins using this SEC column). However, this SEC separation allows a very neat separation of the Ti bound to protein (i.e., Ti-Tf species) from the trace of free Ti-citrate observed at 28-mL elution volume. Once demonstrated that Ti(IV) (as citrate) binds to Tf in human blood serum, even displacing originally bound Fe, we evaluated the possibility to carry out Ti-Tf quantifications by IDA HPLC-ICPMS in real serum samples. Ti-Tf Quantification in Spiked Human Blood Serum Samples by Species-Unspecific IDA. As is customary,36 Ti isotope abundances and concentration of the synthesized 49 Ti-citrate were determined previous to its use. Isotopic composition is given in Table 1, which shows that enriched 49Ti abundance (71.24 ± 0.05%) in the spike agrees well with the indicative value provided by the manufacturer (71.17%). This value differs greatly from the natural 49Ti (5.50 ± 0.02%), allowing its direct use in the species-unspecific IDA experiments. Total Ti concentration obtained by reverse IDA using a natural Ti ICP

Table 1. Isotope Composition (%) of the Isotopically Enriched Ti Spikes Synthesized for the Quantitative Analysis of the Ti-Tf Complex by Species-Unspecific and Species-Specific IDA (n ) 3; uncertainty, 1 standard deviation) isotopes 46

Ti Ti Ti 49 Ti 50 Ti 47 48

natural Ti-citrate 8.00 ± 0.03 7.30 ± 0.03 73.80 ± 0.03 5.50 ± 0.02 5.40 ± 0.02

enriched Ti-citrate

49

1.82 ± 0.01 1.80 ± 0.01 22.04 ± 0.05 71.24 ± 0.05 3.10 ± 0.01

enriched 49 Ti-Tf 1.88 ± 0.01 1.83 ± 0.01 23.30 ± 0.12 69.90 ± 0.11 3.13 ± 0.01

standard was 168 ± 2 mg L-1 (n ) 3). This value implies a final synthesis yield for the purified 49Ti-citrate of 90 ± 1%. In separate experiments, 1 and 2 molar equiv of Ti-citrate to Tf were added to human blood serum samples. Then, the samples were incubated for 12 h to allow Ti-Tf complexation and were analyzed by IDA HPLC-ICPMS using the above 49Ti-citrate spike. The first step to carry out such quantitative analysis is to ensure the integrity of the complex all along the speciation analysis Analytical Chemistry, Vol. 80, No. 22, November 15, 2008

8707

Table 2. Titanium Bound to Tf Quantification Using Different Chromatographic Mechanisms and Species-Unspecific IDAa Ti/Tf molar ratio (R)

size exclusion (%)

anion exchange (%)

1 2

ApoTf Standard 42 ± 3 45 ± 2

25 ± 3 27 ± 2

1 2

Human Blood Serum 33 ± 2 36 ± 2

20 ± 2 24 ± 2

a Results are expressed as the ratio (%) between the Ti added in the incubation process and the Ti quantified in the Tf peaks (n ) 3; uncertainty corresponds to 1 standard deviation).

procedure. Species-unspecific IDA would allow eventual quantification of Ti-Tf, provided that it elutes completely from the chromatographic column. In this vein, the employed anion exchange column was evaluated for such intended postcolumn IDA quantification. Typical postcolumn chromatographic profiles and subsequent mass flows are presented in Figures S-7A and S-8A of the Supporting Information showing that Ti-Tf complex was detected as the only Ti-containing species. Surprisingly, as can be seen in Table 2, total Ti quantified by IDA in that double peak only accounted for ∼25% of the Ti injected into the chromatographic system. Interestingly, such low recoveries were obtained for both the Ti-Tf model solution and the investigated human blood serum samples. These results point to a dissociation of the Ti-Tf complex on passing the stationary phase of the column during the salt gradient used. In an attempt to assess and minimize Ti displacement during the selected chromatographic separation, we resorted to SEC. This chromatography is known to secure even easier labile metalprotein bindings39 because the driving force of these separations is the hydrodynamic volume of the molecules (i.e., saline gradients are not required). Thus, the analogous HPLC-ICPMS postcolumn IDA experiments were carried out to quantify Ti-Tf, and typical SEC chromatograms and mass flows are shown in Figures S-7B and S-8B of the Supporting Information. It was observed that, the higher the concentration of salts in the mobile phase, the lower the obtained Ti recoveries. Figure 3 clearly shows this fact for bicarbonate and citrate increasing concentrations. As can be observed in Figure 3A, the bicarbonate content (pH 7.4) had to be reduced to 2 mM. Citrate was also added to the mobile phase in order to prevent possible Ti hydrolysis and subsequent precipitation as TiO2. Its concentration was also optimized and the results observed are shown in Figure 3B. In this case, Ti-Tf recoveries increased with citrate in the mobile phase up to a limit where the high salt concentrations seem to start dissociating the Ti-Tf complex. Finally, a mobile-phase composition of 2 mM bicarbonate and 0.2 mM citrate (see Figure 3), ensuring maximum ICPMS signal, was selected. As can be seen in Table 2, quantitative results using such optimized conditions were significantly higher than those found previously using anion exchange chromatography, both in the model solution and in human serum, but still far from 100%. In light of these results, it seems that Ti-Tf binding is broken in both types of chromatographies. This dissociation finding (39) Lobinski, R.; Schaumlo ¨ffel, D.; Szpunar, J. Mass Spectrom. Rev. 2006, 25, 255–289.

8708


agrees with previous results reported for Mn-Tf complex also using SEC.15 One possible factor accounting for such dissociation during the chromatography could be adsorption, hydrophobic, and ion-exchange interactions, which are also known to take place during SEC separations.40 Tinoco et al.41 recently demonstrated by differential scanning calorimetry that Ti-Tf binding is not under thermodynamic control. Although the formal binding for Ti(Cit)3-8 is high, there is a thermodynamic sink that overall favors Ti hydrolysis and its precipitation as TiO2. In fact, Ti-Tf is a metastable complex with Ti(IV) thermodynamically drawn to fast and irreversible dissociation ending in TiO2. These stability problems could help to understand why Ti-Tf binding is so prone to breakup under the HPLC mechanisms and conditions assayed (as soon as TiO2 is formed, it would immediately precipitate inside the column leading to the low Ti-Tf recovery obtained from the chromatographic columns). In order to investigate species dissociation, an IDA technique able to measure and correct for species degradation along the chromatographic step should be invaluable.26 In this vein, the use of an isotopically enriched 49Ti-Tf spike could shed light on the detected Ti-Tf degradation problems. Ti-Tf Quantification in Spiked Human Blood Serum Samples by Species-Specific IDA. The 49Ti-Tf spike was obtained after saturation of the binding sites of an ApoTf standard with the previously synthesized 49Ti-citrate. The 49Ti-Tf complex was then thoroughly purified by ultrafiltration and isotopically characterized using DF-ICPMS. The obtained isotopic composition is shown in Table 1. As can be seen, the 49Ti enrichment (69.90 ± 0.11) is slightly lower than that of 49Ti-citrate (71.24 ± 0.05), probably due to small contributions of traces of natural Ti present in the reagents used in the Tf incubation. Finally, Ti quantification in the purified 49Ti-Tf spike solution was carried out using reverse IDA and a natural Ti ICPMS standard. Then, a known amount of the 49Ti-Tf spike freshly prepared was added independently to the ApoTf standard model solution and to the human blood serum previously incubated with natural Ti-citrate (and quantified by postcolumn IDA). Samples were vortexed and left at 37 °C during 5 min to guarantee the complete mixing and equilibration of the incurred Ti-Tf and the spiked 49Ti-Tf species. Finally, both model and real human blood serum were analyzed using SEC and anion exchange chromatographies, coupled as before to ICPMS detection. The obtained SEC chromatogram for the spiked blood serum sample is shown in Figure 4A. As can be clearly observed after comparison with Figure 2B, alteration of the “natural” Ti abundances was obtained for every Ti-containing species detected. In fact, Ti isotope ratios obtained were statistically undistinguishable for both Ti-Tf (47Ti/49Ti ) 0.26 ± 0.01 and 48Ti/49Ti ) 2.55 ± 0.11) and the Ti-citrate species (47Ti/49Ti ) 0.28 ± 0.01 and 48Ti/49Ti ) 2.76 ± 0.09) observed in Figure 4A. This is further evidence that complete isotope equilibration took place, and what is more interesting, generation of the free Ti-citrate occurred after the spiking and, therefore, along the chromatographic process. Taking into consideration the low overall Ti recovery obtained from the column (see Table 2), it seems that most of the Ti-citrate released from the Tf binding (40) Harms, A.; vanElteren, J. T.; Claessens, H. A. J. Chromatogr., A 1996, 755, 219–225. (41) Tinoco, A. D.; Incarvito, C. D.; Valentine, A. M. J. Am. Chem. Soc. 2007, 129, 3444–3454.

Figure 3. Influence of the salt concentration present in the mobile phase in the final Ti recovery from the SEC column.

sites precipitated irreversibly as insoluble TiO2 (inside the chromatographic column). Final quantitative results obtained from the Ti isotope ratios measured in the Ti-Tf species are shown in Table 3. These results demonstrate that the ApoTf model solution complexed Ti quantitatively (>95%) at the two Ti concentration levels assayed. Moreover, binding proved to be almost complete in human blood serum as well, showing that Tf can specifically uptake Ti ions, among the other components, present in that complex matrix. Finally, it is interesting to note that this quantitative Ti-Tf binding was also obtained after addition of 2 mol equiv of Ti (Ti/Tf molar ratio 2:1, see Table 3), which implies that Ti could displace the naturally occurring Fe from the Tf binding sites. This finding is in agreement with results obtained by Tinoco et al.10 using isothermal titration calorimetry in Tf standard solutions. The anion exchange chromatogram for a spiked blood serum is given in Figure 4B and demonstrates again that natural Ti isotope ratio abundances were clearly altered in the Ti-Tf complex (for comparison purposes, see Figure 2A). Interestingly, 47 Ti/49Ti isotope ratios measured for the two Ti-Tf peaks at 8 and 12 min were statistically the same, indicating complete species equilibration. IDA quantitative results obtained are shown in Table

3, and they are statistically the same as those with SEC, demonstrating the reliability of the species-specific approach. As a matter of fact, two independent chromatographies with very different Ti recoveries (Table 2) and chromatographic profiles (Figure 4) provided identical corrected quantitative results (Table 3). This is clear evidence that degradations occurring to the Ti-Tf complex during the analytical separation process, independently of its extent, could be completely corrected, and so accurate and reliable results are achieved for both chromatographies used. Interestingly, del Castillo et al.29 compared also the performance of the two isotope dilution modes used here to study human serum transferrin isoforms. For that purpose, they carried out the synthesis of a metalloprotein (57Fe-Tf), and quantitative Fe recovery from the chromatographic column was observed. Therefore, quantitative results for Tf isoforms obtained by speciesunspecific and species-specific modes were identical. These comparative results back the assumption that similar metal-protein interactions, even under similar experimental conditions, may be different. Thus, today there is an urgent need of individual assessing of every metalloprotein complex analysis procedure (if reliable metal bound to protein determinations are aimed at). Analytical Chemistry, Vol. 80, No. 22, November 15, 2008

8709

Figure 4. Species-specific isotope dilution analysis chromatograms for an incubated human blood serum sample (Ti-Tf molar ratio 2) using (A) size exclusion chromatography and (B) anion exchange chromatography (47Ti signal, dot line; 49Ti signal, solid line). Table 3. Titanium Bound to Tf Quantification Using Different Chromatographic Mechanisms and Species-Specific IDAa Ti/Tf molar ratio (R)

size exclusion (%)

anion exchange (%)

1 2

ApoTf Standard 98 ± 3 100 ± 2

102 ± 4 102 ± 3

1 2

Human Blood Serum 93 ± 3 96 ± 3

94 ± 3 92 ± 3

a Results are expressed as the ratio (%) between the Ti added in the incubation process and the Ti quantified in the Tf peaks (n ) 3; uncertainty, 1 standard deviation).

CONCLUSIONS HPLC-ICPMS experiments here have demonstrated the quantitative reaction of Ti(IV) with Tf in human blood serum. The combination of the quantitative nature of the elemental response provided by ICPMS32 in heteroatom-tagged proteins with the use of stable isotope-based approaches has enabled a reliable 8710


quantitative assessment of the original Ti-Tf binding in human blood serum. The accurate quantitative results observed have also shown that Tf affinity for Ti-citrate leads to Ti(IV) occupation of every Tf binding site at molar ratio 2:1 including most of those originally occupied by iron. Moreover, in the light of all data obtained, it can be concluded that quantitative results reported so far in heteroatom-tagged proteomics studies32 should be considered very carefully. Extreme care must be taken to preserve the original heteroatom-protein interaction in the analytical process unless this heteroatom is covalently bound to the primary amino acid chain (i.e., S, Se, P). Herein we have resorted to postcolumn isotope dilution to show that Ti-Tf binding is mostly broken along the protein separation process, even under really mild chromatographic conditions. Fortunately, the 49Ti-Tf spike synthesized in this work has allowed, by using species-specific IDA mode, the accurate determination of the amount of Ti-Tf species originally present in the sample (even if the complex dissociated very heavily during the analytical process). In fact, to the best of our knowledge, this is the first time that an isotope-tagged protein is synthesized and

applied to correct for the likely dissociation of metal-protein coordination complexes occurring in one of the most common analytical steps (chromatographic separation) used for elemental speciation in proteins. As demonstrated here, this tool could pave the way to develop new analytical procedures to achieve reliable metal-protein quantifications. Such validated approaches will become essential to meet the quality assurance requirements urgently demanded in quantitative heteroatom-tagged proteomics in particular, and in quantitative proteomics in general,42 overcoming some of the present problems of transparency of the results reported in literature.38 In both cases, results so far have largely been qualitative and as “mass spectrometry-based proteomics turns quantitative”,42 new approaches, as the one here, to obtain accurate (metallo)protein quantifications in real samples are increasingly important. The combined use of the two IDA modes assayed here might unveil metal-biomolecules dissociations and measure their extent in sample preparation, preconcentration,

separations, etc., commonly used so far in this active analytical field.

(42) Ong, S.-E.; Mann, M. Nat. Chem. Biol. 2005, 1 (5), 252–262.

AC801029P

ACKNOWLEDGMENT Financial support from the MEC (Madrid, Spain, CTQ200602309/BQU) and Applera Hispania (FUO-EM-023-05) is gratefully acknowledged. J.R.E. acknowledges the MEC and European Social Fund for a Ramon y Cajal contract. The authors gratefully acknowledge G. Valca´rcel (“Hospital de San Agustıń”) for providing the serum samples. SUPPORTING INFORMATION AVAILABLE Additional information as described in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review May 20, 2008. Accepted August 28, 2008.


8711

HPLCâICPMS and Stable Isotope-Labeled Approaches To Assess

Recommend Documents

HPLCâICPMS and Stable Isotope-Labeled Approaches To Assess