Characterization of the Glycation of Albumin in Freeze-Dried and

124-298 and 447-548, suggesting glycation. The pres- ence of glycation on fragments 124-298 and 447-548 correlates with the known sites of HSA glycati...
0 downloads 0 Views 379KB Size
Anal. Chem. 1997, 69, 2457-2463

Characterization of the Glycation of Albumin in Freeze-Dried and Frozen Human Serum David M. Bunk*

National Institute of Standards and Technology, Building 222/B208, Gaithersburg, Maryland 20899

Human serum albumin (HSA) in fresh frozen and freezedried serum reference materials was examined by mass spectrometry and a variety of affinity chromatography techniques. The relative molecular mass distribution of HSA in fresh frozen serum was found to be identical to that of an HSA standard. However, the HSA in the freezedried reference serum exhibited a relative molecular mass distribution that was shifted to higher mass, broader, and substantially more heterogeneous than that of HSA in fresh frozen serum. A proteolytic cyanogen bromide digestion of the HSA from freeze-dried serum contained adducts ∼162 u higher in mass than digest fragments 124-298 and 447-548, suggesting glycation. The presence of glycation on fragments 124-298 and 447-548 correlates with the known sites of HSA glycation. Glycation was further confirmed by the mass spectral analysis of the retained and unretained fractions from glycoaffinity chromatography of HSA from freeze-dried serum. The relative molecular weight of the HSA in the retained fraction indicated the presence of a doubly glycated species. The chemical heterogeneity of Cys-34, the site of the only free thiol in HSA, was examined and found not to be a substantial source of molecular mass heterogeneity for HSA from either fresh frozen or freeze-dried serum. Serum reference materials, used as controls for clinical assays, are produced in two forms: frozen and freeze-dried. The biological activities of the clinically relevant species in the serum are, in most cases, not lost upon freezing. But, in contrast to freezedried serum, frozen serum presents more challenges in its distribution from the site of production to clinical laboratories, as well as for long-term storage. Once serum is freeze-dried, it does not require extreme care to maintain quality during distribution and storage. One detrimental aspect of freeze-drying serum is that this process can irreversibly denature lipoproteins, resulting in changes in viscosity, turbidity, pH, and surface tension.1 Additionally and probably of greater concern, the biological activity of clinically relevant species can be lost or substantially reduced upon freeze-drying. Studies done on alkaline phosphatase,2,3 serum lipoprotein a,4 and bovine serum albumin5 have demonstrated activity or concentration loss of these proteins upon freezedrying. To produce a viable freeze-dried serum reference material, * [email protected]. (1) Ferrero, C. S.; Carobene, A, ; Ceriotti, F.; Modenese, A.; Arcelloni, C. Clin. Chem. 1995, 41(4), 575-580. (2) Ford, A. W.; Dawson, P. J. J. Pharm. Pharmacol. 1993, 45, 86-93. (3) Ford, A. W.; Allahiary, Z. J. Pharm. Pharmacol. 1993, 45, 900-906. (4) Sgoutas, D. S.; Tuten, T. Clin. Chem. 1992, 38(7), 1355-1360. (5) Tarelli, E.; Wood, J. M. J. Biol. Stand. 1981, 9, 121-129.

Figure 1. Chemistry of protein glycation.

it is necessary to ensure that all analytes of interest in the serum have not been altered during the preparation. In light of this, and to complement current activities at the National Institute of Standards and Technology (NIST) to develop methods for the quantitation of clinically relevant proteins in reference materials, a characterization of human serum albumin (HSA) in frozen and freeze-dried serum reference materials was undertaken. The chief biological functions of HSA, the most abundant protein in adult plasma, are to transport and store a chemically diverse collection of both endogenous and exogenous ligands and to maintain plasma osmotic pressure.6 In addition to the noncovalent binding of ligands such as calcium, nonpolar compounds like bilirubin and long-chain fatty acids, hormones such as thyroxine or cortisol, and amino acids, albumin also undergoes several types of covalent modification during its lifetime in the bloodstream. Of the 35 cysteine residues in this 585-residue protein, 34 are cross-linked to form disulfide bonds, leaving one free thiol on residue 34. This free thiol can form mixed-disulfide bonds with other thiol-containing compounds in blood, such as cysteine and glutathione. Recently, HSA has been shown to have substantial antioxidant activity which has been attributed to a glutathione linkage.7 Chromatographic analysis of HSA derived from healthy subjects indicated that about 70-75% of the protein contained a free thiol,8 but the free thiol content dropped to below 50% for patients with liver diseases.9 Although albumin is not glycosylated upon biosynthesis, the primary amino groups on the protein can react with glucose to become glycated. The chemistry of glycation, as shown in Figure 1, begins with the condensation of glucose with a primary amine on the protein, such as those found on the side chains of lysine residues or the N-terminus. The Schiff base (aldimine) intermedi(6) Min He, X.; Carter, D. C. Nature, 1992, 358, 209-215. (7) Cha, M.-K.; Kim, I.-H. Biochem. Biophys. Res. Commun. 1996, 222, 619625. (8) Etoh, T.; Miyazaki, M.; Harada, K.; Nakayama, M.; Sugii, A. J. Chromatogr. 1992, 578, 292-296. (9) Sogami, M.; Era, S.; Nagaoka, S.; Kuwata, K.; Kida, K.; Shigemi, J.; Miura, K.; Suzuki, E.; Muto, Y.; Tomita, E.; Hayano, S.; Swada, S.; Noguchi, K.; Miyata, S. J. Chromatogr. 1985, 19-27.

S0003-2700(96)01205-X This article not subject to U.S. Copyright. Publ. 1997 Am. Chem. Soc.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997 2457

ate that is formed can irreversibly rearrange to form the Amadori product, a ketoamine. For people whose blood glucose is maintained at stable and normal levels, approximately 10-12% of circulating albumin is glycated,10 although some analytical methods place the concentration of glycated albumin as low as 0.61.8%.11 Elevated levels of glycated HSA are observed in people afflicted with diabetes, and the measurement of glycated HSA, along with glycated hemoglobin, is clinically relevant for the diagnosis and monitoring of this disease. Posttranslational modifications, such as mixed-disulfide formation and glycation, can have a profound impact on the measurement of proteins. Modification of an antigenic protein can alter the binding efficiency to an antibody, such as those used in immunoassays, if the modification occurs at the epitope, the binding site of the antibody. If the form or degree of modification is different between samples and standards, immunoassay results may be compromised. To ensure the accuracy of test results and the quality of subsequent diagnostic, it is essential that there be little difference between samples and standards used in clinical immunoassays or that the impact of any difference between samples and standards on assay results be evaluated. With the advent of powerful protein analysis techniques such as electrospray ionization mass spectrometry (ESI-MS) and matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS), posttranslational modifications of proteins can be readily detected with high sensitivity. For the measurement of proteins in a complex matrix such as serum, ESI-MS is particularly well-suited because this technique can be directly coupled to separation techniques such as liquid chromatography and capillary electrophoresis. EXPERIMENTAL SECTION Materials and Equipment. Freeze-dried serum (samples from 1990 and 1994) and fresh frozen serum (produced in 1994) were obtained from a commercial source. The freeze-dried serum was reconstituted with HPLC grade water. An HSA standard (Lot 74147) was obtained from United States Biochemical (Cleveland, OH). Solvents for chromatography were HPLC grade from J. T. Baker (Phillipsburg, NJ). The trifluoroacetic acid (TFA) used for the reversed-phase chromatography was from Pierce (Rockford, IL). Mass spectra were recorded on a Finnigan MAT (San Jose, CA) TSQ-70 triple quadrupole mass spectrometer upgraded with a TSQ-700 operating system and equipped with an Analytica (Branford, CT) electrospray source. In the positive ion mode, mass spectra were obtained by scanning the first quadrupole from m/z 1000 to 2000 in 2 s at a sampling rate of five data points per mass unit. The charge state ion distributions observed in ESI mass spectra were deconvoluted using Finnigan’s BIOMASS software. Samples were introduced into the electrospray source by capillary reversed-phase HPLC using a 0.18 × 500 mm column packed with Vydac C18, 30 nm pore size (LC Packings, San Francisco, CA). An Ultra-Plus binary gradient micro-LC pump (Microtech Scientific, CA) was used to produce the solvent gradients needed for the capillary reversed-phase HPLC. The pump was operated at 40 µL/min, but a column flow rate of 1-2 µL/min was obtained by preinjector flow splitting. (10) Garlick, R. L.; Mazer, J. S. J. Biol. Chem. 1983, 258(10), 6142-6146. (11) Silver, A. C.; Lamb, E.; Cattell, W. R.; Dawnay, A. B. St. J. Clin. Chim. Acta 1991, 202, 11-22.

2458

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

Serum was freeze-dried using a Virtis Genesis series 25 freezedryer (Gardiner, NY). The serum sample (5 mL) was dried at -10 °C under a vacuum of 2.7 Pa for 14 h before increasing the temperature to +10 °C for an additional 4 h of drying. Antibody Affinity Chromatography. An Immunopure protein G IgG orientation kit from Pierce was used to produce an antiHSA antibody affinity chromatography column. Approximately 1.5 mg of mouse monoclonal anti-HSA antibody (Pierce, lot 94101204) was coupled to the protein G stationary phase according to the procedure supplied by Pierce; the bound monoclonal antibodies are covalently cross-linked to the protein G using the dimethyl pimelimidate supplied with the kit. The volume of the anti-HSA affinity column was 2 mL. Extraction of HSA from serum, freeze-dried or fresh frozen, was performed by first diluting the serum 100-fold with 10 mmol/L tris(hydroxymethyl)aminomethane (Tris) buffer, pH 7.5. The diluted serum (250 µL) was applied to the anti-HSA affinity column and equilibrated for 30 min. Unbound serum constituents were removed from the column bed by washing with four column volumes (8 mL) of 10 mmol/L Tris buffer, pH 7.5. The HSA bound to the column was eluted with 0.1 mol/L glycine hydrochloride, pH 2.8. Cyanogen Bromide Digest. Using anti-HSA antibody affinity chromatography as outlined above, ∼0.8 mg of HSA was extracted from serum. Reduction of the protein’s disulfide bonds was accomplished by dissolving the extracted protein in 6 mol/L guanidine hydrochloride, pH 7.5, containing 0.13 mol/L Tris hydrochloride, 0.1 mg/mL ethylenediaminetetraacetic acid (EDTA), and 3 mmol/L dithiothreitol and incubating overnight at room temperature.12 The liberated cysteine residues were alkylated by adding 4-vinylpyridine (Sigma, St. Louis, MO) to the reduced protein solution in a 3-fold molar excess over the dithiothreitol concentration and reacting at room temperature for 2 h; the reaction was quenched by acidifying with TFA. The reduced and alkylated HSA was desalted by ultrafiltration using a Microcon30 (Amicon, Beverly, MA) 30 000 molecular weight cutoff centrifugal ultrafiltration tube and then lyophilized to dryness. Cyanogen bromide was dissolved in 75% TFA in water at a concentration of 10 mg/mL; the dried reduced and alkylated HSA was redissolved in the CNBr solution, using 1 mL of CNBr solution for each milligram of protein.13 The cleavage of HSA by CNBr was allowed to proceed overnight at room temperature, in the dark. The reaction was then quenched by diluting the reaction solution 10-fold with water. This solution was then lyophilized prior to being redissolved in water for analysis by LC/MS. Following this same procedure, 0.6 mg of HSA standard was reduced, alkylated, and proteolytically digested with CNBr. Glycoaffinity Chromatography. Approximately 0.2 mg of HSA was extracted from freeze-dried serum using anti-HSA antibody affinity chromatography as outlined above. The extracted HSA was concentrated and exhaustively dialyzed against 0.25 mol/L ammonium acetate containing 0.05 mol/L magnesium chloride, pH 8.5. The sample (100 µL) was applied to a 1 mL immobilized boronate column (GlycoGel II, Pierce) and incubated at room temperature for 90 min. Unretained protein was removed by washing the boronate resin with 13 column volumes of the ammonium acetate buffer. Retained protein was eluted using 0.2 (12) Darbre, A., Ed. Practical Protein ChemistrysA Handbook; John Wiley and Sons: Chichester, U.K., 1986; p 75. (13) Darbre, A., Ed. Practical Protein ChemistrysA Handbook; John Wiley and Sons: Chichester, U.K., 1986; pp 83-88.

Figure 2. Positive ion ESI mass spectrum of HSA from fresh frozen serum. The insert shows the deconvoluted mass spectrum in the mass range of 66 000-67 250 u.

mol/L sorbitol in 0.1 mol/L Tris buffer, pH 8.5. Retained and unretained protein fractions were desalted and concentrated prior to analysis by LC/MS. Biotinylation of HSA. Approximately 0.4 mg of HSA was affinity extracted from both freeze-dried and fresh frozen serum as outlined above. The extracted protein was dialyzed against 0.1 mol/L phosphate buffer, pH 7.0, containing 5 mmol/L EDTA. A sulfhydryl-reactive biotinylating reagent, 1-biotinamido-4-[4(maleimidomethyl)cyclohexanecarboxamido]butane (biotin-BMCC) from Pierce, was dissolved in dimethyl sulfoxide at a concentration of 4.5 mg/mL. Eight microliters of the biotinBMCC solution was added to the HSA solution (200 µL), and the reaction was allowed to proceed for 2 h at room temperature. Excess biotin-BMCC was removed by ultrafiltration. Biotinylated HSA was isolated from native protein using avidin affinity chromatography with an Ultralink immobilized monomeric avidin column (Pierce). RESULTS AND DISCUSSION The relative molecular mass (Mr) of HSA, calculated from the cDNA sequence14 of the protein, is 66 438 ( 3 (the uncertainty is the standard deviation of the calculated relative molecular mass resulting from the uncertainty in the value of the isotopic abundance of each element15). By combining Edman degradation, proteolytic cleavage, and ESI-MS, Clerc et al.16 verified 99% of the amino acid sequence of a recombinant HSA derived from the cDNA sequence, which had an observed Mr of 66 460, slightly (14) Lawn, R. M.; Adelman, J.; Bock, S. C.; Franke, A. E.; Houck, C. M. Najarian, R. C.; Seeburg, P. H.; Wion, K. L. Nucleic Acids Res. 1981, 9, 6103-6114. (15) Commission on Atomic Weights and Isotopic Abundances, International Union of Pure and Applied Chemistry. Pure Appl. Chem. 1991, 63(7), 975990. (16) Clerc, F. F.; Monegier, B.; Faucher, D.; Cuine, F.; Pourcet, C.; Holt, J. C.; Tang, S.-Y.; Van Dorsselaer, A.; Becquart, J.; Vuilhorgne, M. J. Chromatogr. B 1994, 662, 245-259.

higher than expected, which the authors attribute to salt impurities. The positive ion ESI mass spectrum of HSA from fresh-frozen serum (FF-HSA) is shown in Figure 2. This mass spectrum was obtained by averaging the ion signal under the peak attributed to HSA in the capillary reversed-phase LC/MS chromatogram from an injection of diluted serum. Ions with charge states ranging from +34 to +58 are observed in the m/z range of 1100-2000. The deconvoluted spectrum obtained from this charge state distribution is shown in the inset of Figure 2. Inspection of the deconvoluted relative molecular mass distribution indicates some asymmetry, with additional contribution observed at the high-mass side of the distribution. Although the width of the molecular mass distribution of FF-HSA was highly reproducible, its exact shape was not. From one analysis to the next, the molecular mass distribution’s maximum showed variability, sometimes being observed at the center of the molecular mass distribution, sometimes occurring toward the lower mass side, as in Figure 2, and sometimes to the higher mass side. The observed Mr of FFHSA was 66 486 ( 11 (the values of observed Mr reported in this paper are the means of the relative molecular masses calculated from each peak in the electrospray charge state distribution along with the standard deviations of the means). This distribution was nearly identical to that obtained from the Sigma HSA standard (data not shown), which was observed with an Mr of 66 486 ( 14. The mass distribution of the HSA standard exhibited a similar variability in its shape as the HSA from fresh-frozen serum; therefore, it is unlikely that this variability is the result of contributions from the serum matrix. It seems more likely that instrument instabilities contributed to variability in the observed mass distributions. Although comparable results were obtained for FF-HSA and the HSA standard, the mass distribution of HSA from freeze-dried serum (FD-HSA) was observed at higher mass, substantially Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2459

Table 1. Relative Molecular Masses of the CNBr Digest Fragments of an HSA Standard and HSA Extracted from Freeze-Dried Reference Serum observed Mr CNBr fragment 1-87 88-123 124-298 + adduct 299-329 330-446 447-548 + adduct 549-585 Figure 3. Comparison of the relative mass distributions of HSA from fresh frozen and freeze-dried serum reference materials.

broader, and exhibited much more heterogeneity. The mass distributions of FF-HSA and FD-HSA, obtained from the deconvolution of their charge state distributions, are compared in Figure 3. Two samples produced in the same year (1994) and one older sample (1990) of freeze-dried serum were examined and found to be identical with respect to the relative molecular mass distribution of HSA. Although the serum matrix for both the FFand FD-HSA samples should be similar, it was initially suspected that the freeze-dried serum may contain something that forms adducts with the HSA, such as hydrophobic fatty acids. To test this hypothesis, the freeze-dried serum was delipidated with the standard methods of solvent extraction17 and treatment with activated charcoal18 prior to analysis by LC/MS. Additionally, the HSA from freeze-dried serum was extracted using anti-HSA monoclonal antibody affinity chromatography, delipidated using solvent extraction, activated charcoal, or lipidex chromatography,19 and then analyzed by mass spectrometry. All of these attempts to delipidate the FD-HSA resulted in no noticeable change in the molecular mass distribution of the protein. Therefore, it was concluded that the increase in mass, width of the mass distribution, and the heterogeneity of FD-HSA were not the result of noncovalent adduct formation between the protein and other serum constituents, but the result of a covalent modification to this protein. Because the different HSA chemical forms contributing to the heterogeneous molecular mass distribution of FD-HSA could not be resolved with the quadrupole mass analyzer used for these analyses, the mass spectrum of the intact protein did not provide sufficient information on the nature of the covalent modification(s). To get more information of the nature and sites of protein modification, the HSA was extracted from freeze-dried serum by anti-HSA antibody affinity chromatography and subjected to proteolytic cleavage using CNBr. For comparison, a sample of the HSA standard was also digested with CNBr. The fragments obtained from the CNBr digest were analyzed by reversed-phase LC/MS, and masses of the fragments observed for both the HSA standard and FD-HSA are listed in Table 1. Of the two possible C-terminal structures that result from CNBr cleavage, the observed relative molecular mass of most of the digest fragments (17) Folch, J.; Lees, M.; Sloane-Stanley, G. A. J. Biol. Chem. 1957, 226, 497. (18) Chen, R. F. J. Biol. Chem. 1967, 242(2), 173-181. (19) Glatz, J. F. C.; Veerkamp, J. H. J. Biochem. Biophys. Methods 1983, 8, 5761.

2460 Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

calculated Mr

HSA standarda

HSA extracted from FD serum

10 212 4 478 21 265 21 427b 3 420 14 171 12 224 12 286b 4 239

10 213 ( 2 4 478 ( 1 21 268 ( 8 21 432 ( 10 3 419 ( 1 14 176 ( 4 12 227 ( 3 12 395 ( 5 4 238 ( 1

10 211 ( 3 4 477 ( 1 21 254 ( 7 21 415 ( 10 3 418 ( 1 14 173 ( 3 12 224 ( 3 12 386 ( 4 4 238 ( 1

a The observed M is the mean of the relative molecular masses r calculated from each peak in the charge state distribution of the fragment. The uncertainty reported is the standard deviation of the mean. b Calculated Mr assuming single glycation of fragment.

corresponded to the C-terminal homoserine lactone structure with little to no homoserine observed; only fragment 299-329 exhibited an ion whose relative molecular mass indicated homoserine, although the predominant ion for fragment 299-329 still possessed a relative molecular mass indicating homoserine lactone. All fragments expected from the CNBr digest of HSA were observed for both FD-HSA and the HSA standard and were present with strong ion signals in the mass chromatogram. Only two weak ion signals were observed in the mass chromatogram of the HSA standard for which assignments could not be made and these ion signals were also observed for FD-HSA. The largest difference between the mass chromatograms of FD-HSA and the HSA standard was the appearance of two ion signals present in the mass spectra of fragments 124-298 and 447-548 for FD-HSA but only weakly observed for the HSA standard. Figure 4 compares the mass spectra of fragment 447-548 for (a) the HSA standard and (b) FD-HSA. The insets of Figure 4 provide the deconvoluted mass spectra. Multiply charged ions corresponding to species whose mass is 162 higher than fragment 447-548 are clearly observed in Figure 4b (marked with *) but also in Figure 4a. Although in Figure 4a, for the HSA standard, the intensity of the ions corresponding to the higher mass species are barely discernable above the chemical noise. Similar results were obtained for fragment 124-298. The difference in the relative molecular masses of fragments 124-298 and 447-548 and the corresponding higher mass species coeluting with each fragment of FD-HSA was 161 and 162, respectively. This mass difference suggests that glycation has occurred (see Figure 1); the end product of glycation would have a relative molecular mass 162 higher than the unmodified protein. Evidence of glycation of fragments 124-298 and 447-548 correlates with sites on HSA that are known to be predominantly glycated. Garlick and Mazer10 reported that the principal site for in vivo glycation of HSA is Lys-525. Analysis of HSA glycated in vitro have also identified Lys-199 and Lys-281 as additional sites of glycation.20 Robb et al.21 observed in vivo glycation at the N-terminal R-amino group of HSA but to a lesser extent than the -amino groups of lysine residues. No indications of a covalent (20) Iberg, N.; Fluckiger, R. J. Biol. Chem. 1986, 261(29), 13542-13545. (21) Robb, D. A.; Olufemi, O. S.; Williams, D. A.; Midgley, J. M. Biochem. J. 1989, 261, 871-878.

Figure 4. Positive ion ESI mass spectrum of fragment [447-548] from the cyanogen bromide digest of (a) HSA standard and (b) HSA extracted from freeze-dried serum. The inserts in both (a) and (b) shows the deconvolution of the ESI mass spectrum in the mass range from 12 000 to 12 600 u. The denotes the peaks that produce the peak at Mr ) 12 395 ( 5 in (a) and at Mr ) 12 386 ( 4 in (b), respectively.

*

adduct to CNBr fragment 1-87 was observed in mass chromatogram of FD-HSA or the HSA standard. In addition to the CNBr digest of FD-HSA, a sample of FDHSA was treated to the same conditions (75% TFA overnight) as that used for the CNBr digest but without the addition of CNBr. The purpose of this experiment was to determine whether the chemical modification of the FD-HSA was labile under these harsh conditions and would therefore be lost during the CNBr digest. The result of this study was that FD-HSA treated with 75% TFA had essentially the same molecular mass distribution as FD-HSA,

indicating that the modifying group(s) was stable under the conditions of the CNBr reaction. To provide further confirmation of the presence of glycation on HSA in freeze-dried serum, FD-HSA was analyzed by glycoaffinity chromatography followed, off-line, by mass spectrometry. The immobilized boronate groups on the glycoaffinity stationary phase22 bind cis-diol groups on sugars forming a reversible fivemember ring complex that can be dissociated by another molecule (22) Mallia, A. K.; Hermanson, G. T.; Krohn, R. I.; Fujimoto, E. K.; Smith, R. K. Anal. Lett. 1981, 14, 649-661.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2461

Figure 5. Comparison of the relative molecular mass distributions of the unretained and retained fractions from the glyco-affinity chromatography of HSA extracted from freeze-dried serum. The mass distribution of HSA from freeze-dried serum prior to glyco-affinity chromatography is shown with the dotted line. The zig-zag appearance of the high-mass end of the unretained fraction’s distribution is an artifact of the deconvolution.

containing a cis-diol group, such as sorbitol. HSA extracted from freeze-dried serum by antibody affinity chromatography and extensively dialyzed to remove any residual sugars, was applied to a glycoaffinity column, providing retained and unretained fractions. The relative molecular mass distributions of the protein in the retained and unretained fractions are compared in Figure 5. Also shown in Figure 5 by the dotted line, is the relative molecular mass distribution of FD-HSA prior to glycoaffinity chromatography. The HSA species contained within the unretained fraction has a narrower molecular mass distribution than FD-HSA and is roughly centered over the shoulder on the lowmass side of the relative molecular mass distribution of FD-HSA. While the width of the molecular mass distribution is less than that of FD-HSA, it is still slightly greater than that of HSA from fresh frozen serum. It has been reported that affinity chromatography on boronate columns can exhibit variable capacities, depending on chromatographic conditions.11 It is likely that some glycated HSA remained in the unretained fraction, resulting in the observed high-mass tailing. The detection of an HSA species in the retained fraction confirms that glycated HSA is present in freeze-dried serum. The difference in the relative molecular masses of the HSA species in the retained and unretained fractions, ∆Mr ) 310, indicates the presence of doubly glycated HSA in freeze-dried serum. The molecular mass distribution of the retained fraction is centered over the high-mass shoulder of the FD-HSA molecular mass distribution. The results of the glycoaffinity chromatography of FD-HSA imply that the low-mass shoulder on the molecular mass distribution corresponds to the native, nonglycated HSA, the maximum of the distribution is due to the presence of a singly glycated protein, and the high-mass shoulder is attributable to a doubly glycated species. With the predominance of the doubly glycated species in the retained fraction but the maximum of the relative molecular mass distribution occurring at a mass indicative of singly glycated HSA, it would appear that the glycoaffinity chromatography is more selective toward the doubly glycated HSA than the singly glycated. While the results of the CNBr digest and the glycoaffinity chromatography indicate that the FD-HSA is substantially gly2462 Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

Figure 6. Relative molecular mass distributions of unretained and retained fractions from the biotin-avidin affinity chromatography of biotinylated HSA extracted from fresh frozen serum. The zig-zag appearance of the high-mass end of the unretained fraction’s distribution is an artifact of the deconvolution.

cated, they do not address the possibility that some of the observed molecular mass heterogeneity of FD-HSA could be due to chemical modifications at Cys-34, the site of a reactive free thiol. The difficulty with assessing whether Cys-34 is modified and determining the nature of any modification is that proteolytic cleavage of HSA, to produce peptide fragments that are small enough to provide accurate information on the mass and locations of the modification(s), requires that the 17 disulfides of the protein be reduced. Proteolytic cleavage of this tightly folded protein without prior reduction and denaturation is incomplete and ineffective, yet the chemical methods used to cleave the protein’s intrachain disulfides will most likely also cleave the modifying groups on Cys-34. To gain some insight into the chemical status of Cys-34, a thiol-specific modifying group was used to separate the portion of HSA containing a free thiol from that which does not. Biotin-BMCC is a thiol-specific biotinylating reagent, which, according to its manufacturer (Pierce), is a thousand times more reactive toward a free thiol at pH 7 than to an amine. The biotin group added to the HSA as a result of biotin-BMCC derivatization provides a very effective “handle” through which modified protein can be separated from unmodified using highly selective biotinavidin affinity chromatography,23 a technique that uses the tight binding of biotin to the protein avidin, which is immobilized on the column. Therefore, by coupling derivatization with the thiolspecific reagent biotin-BMCC to biotin-avidin affinity chromatography with off-line mass spectrometry, molecular mass information about the population of HSA containing a free thiol can be obtained. Figure 6 shows the relative molecular mass distributions obtained by ESI-MS of the retained and unretained fractions from biotin-avidin affinity chromatography of biotinylated HSA from fresh frozen serum. For the unretained fraction, which contains HSA without a free thiol and also some HSA with a free thiol but which did not react with the biotin-BMCC, the relative molecular mass, Mr ) 66 592 ( 20, is higher than that obtained for FF-HSA (see Figure 2). The increase in molecular mass is presumably due to the removal of HSA containing the free thiol from the FF(23) Henrickson, K. P.; Allen, S. H. G.; Maloy, W. L. Anal. Biochem. 1979, 94, 366-370.

That glycation is much more prevalent in the freeze-dried serum and not the fresh frozen implies that the freeze-drying process amplifies glycation. In an attempt to reproduce extensive glycation, a sample of fresh frozen serum was freeze-dried using gentle freeze-drying conditions. Unlike the powdered consistency of the commercially prepared freeze-dried serum, the freeze-dried fresh frozen serum had the consistency of a sponge, solid but highly porous. Once reconstituted with water, the freeze-dried fresh frozen serum was analyzed by LC/MS. The relative molecular mass distribution obtained for this sample was identical to that of the original fresh frozen serum and did not exhibit the same increase in molecular mass and molecular mass heterogeneity observed for freeze-dried serum.

Figure 7. Relative molecular mass distributions of unretained and retained fractions from the biotin-avidin affinity chromatography of biotinylated HSA extracted from freeze-dried serum. The zig-zag appearance of the high-mass end of the unretained fraction’s distribution is an artifact of the deconvolution.

HSA, leaving the higher mass, thiol-modified protein. The relative molecular mass of the retained fraction, which contains biotinylated HSA (Mr ) 67 003 ( 18), is close to the calculated relative molecular mass (Mr ) 66 972 ( 3). The width of the relative mass distribution of the retained fraction is nearly the same as that of FF-HSA. As the retained fraction contains biotinylated HSA, which is derived from that portion of FF-HSA containing a free thiol, the fact that the widths of the relative molecular mass distributions of FF-HSA and the retained fraction are similar suggests that molecular mass distribution of FF-HSA is due predominantly to HSA containing a free thiol group. If thiolmodified HSA contributed substantially to the molecular mass distribution of FF-HSA, the width of the mass distribution of the retained fraction would be expected to be less than that of FFHSA. Figure 7 shows the relative mass distributions of the retained and unretained fractions from biotin-avidin affinity chromatography of the biotinylation of HSA from freeze-dried serum. The relative molecular masses and mass distributions are higher for the retained and unretained fractions of FD-HSA than those observed for FF-HSA. The width of the molecular mass distribution of the retained fraction is comparable to that of FD-HSA although slightly higher than that of the unretained fraction. Since the biotinlyated HSA is chemically homogeneous with respect to Cys-34, the width of the molecular mass distribution of the retained fraction must be derived from a source other than thiol modification, which the evidence in this case suggests is glycation.

CONCLUSIONS Substantial differences were observed between the structures of HSA in fresh frozen and freeze-dried serum reference materials. Using proteolytic digests and glycoaffinity chromatography coupled to ESI-MS, it was determined that the albumin in the freeze-dried serum had undergone extensive glycation, something not observed for the fresh frozen serum. While it seems likely that glycation was accelerated or promoted by the freeze-drying process, no changes were observed in the relative molecular mass distribution of HSA in fresh frozen serum after it was freeze-dried at NIST. Therefore, the freeze-drying process itself does not have to be detrimental to the proteins within the serum. These results indicate that it is possible to create a freeze-dried serum whose consituent proteins have not suffered extensive chemical modification. When freeze-drying is used to prepare serum reference materials, structural characterization of the clinically relevant proteins in the serum is essential to ensure the quality of the reference material. With powerful mass spectrometric techniques like ESI-MS and MALDI-MS, detecting structural modifications to proteins is not difficult and should be included in the quality assessment of protein reference materials. ACKNOWLEDGMENT Identification of equipment and materials is meant to accurately report the experimental procedures used for this study and does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the identified equipment or materials are necessarily the best available for the purpose. Received for review November 27, 1996. Accepted April 4, 1997.X AC961205M X

Abstract published in Advance ACS Abstracts, May 15, 1997.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2463