Hemoglobin Raleigh as the cause of a falsely increased hemoglobin A 1C in an automated ion-exchange HPLC method

Transcription

1 Clinical Chemistry 44: (1998) Endocrinology and Metabolism Hemoglobin Raleigh as the cause of a falsely increased hemoglobin A 1C in an automated ion-exchange HPLC method Dan Chen, 1 Dan L. Crimmins, 1 Fong Fu Hsu, 2 Frederik P. Lindberg, 3 and Mitchell G. Scott 1* Irreversible glycation of the hemoglobin A 0 (HbA 0 ) chain leads to the production of HbA 1C, which can be used to monitor long-term blood glucose control in patients with diabetes mellitus. HbA 1C is less positively charged than nonglycated HbA 0, and this decrease in charge is the basis of ion-exchange and electrophoretic methods that measure HbA 1C. We recently identified a sample that appeared to contain 46% HbA 1C by an automated ion-exchange HPLC method (Bio-Rad Variant ) but only 3.8% by an immunoinhibition latex agglutination method. A combination of traditional and mass spectrometric protein analysis and genomic DNA analysis of the Hb chain and genes revealed that the patient was heterozygotic for Hb-Raleigh, a variant containing a valine3alanine substitution at position 1 of the chain. The amino-terminal alanine in this variant Hb is posttranslationally modified by acetylation, leading to a charge difference similar to glycation and making the behavior of HbA 1C and Hb Raleigh virtually identical in the ion-exchange HPLC method. This observation suggests that it is important to confirm HbA 1C values in excess of 15%, especially if they are not consistent with the clinical picture, by an independent HbA 1C method such as immunoassay or boronic acid affinity chromatography. However, for this particular variant Hb, even these latter methods might be misleading, because the acetylated N-terminal amino acid of the Hb-Raleigh chain cannot be glycated. Washington University School of Medicine, 1 Division of Laboratory Medicine; 2 Mass Spectrometry Resource Center, Department of Medicine; and 3 Department of Medicine, Division of Infectious Diseases, St. Louis, MO *Address correspondence to this author at: Washington University School of Medicine, Division of Laboratory Medicine, Box 8118, St. Louis, MO Fax ; mscott@labmed.wustl.edu. Received January 5, 1998; revision accepted March 6, Hemoglobin A 1C (HbA 1C ) is widely accepted as the most reliable marker for monitoring long-term glucose control in diabetes (1). HbA 1C is the result of irreversible glycation of the N-terminal amino acid (valine) of the HbA 0 chain (2, 3). Glycation of the amino group of the N- terminal residue produces a loss of a positive charge, which is the basis for methods that quantitate HbA 1C by ion-exchange chromatography or gel electrophoresis (3). These methods can be affected by other modifications that alter the charge of Hb, such as carbamylation and acetylation (4 6), as well as certain Hb variants (7 12). Other approaches for measuring HbA 1C are based on immunoassays using antibodies specific for the glycated amino terminus of the chain (13) or on the affinity of glycated amino groups for boronate (5, 14, 15). An HPLC ionexchange method was the basis of measurement in the Diabetes Complications and Control Trial (DCCT) and thus has become the reference point for HbA 1C quantification (1). HPLC-based ion-exchange methods for HbA 1C have recently been automated (16), and interferences by Hb species such as HbF and HbS have been minimized. However, several reports have described artificially high HbA 1C results with hemoglobin variants such as Hb- Sherwood Forest (9) (Arg 104 Thr of the chain) and Hb-South Florida (10) (Val 1 Met of the chain) and others (16) when an automated ion-exchange HPLC method is used. We recently observed a HbA 1C value of 46% for a sample that was analyzed on the Bio-Rad Variant automated ion-exchange HPLC method for HbA 1C and investigated this, using a series of protein chemistry and molecular biology techniques. Materials and Methods HbA 1C measurement Whole-blood HbA 1C was measured by two independent methods: (a) cation exchange column-based HPLC on the Bio-Rad Variant analyzer (16) and (b) HbA 1C -specific 1296

2 Clinical Chemistry 44, No. 6, mouse monoclonal antibody-based inhibition of latex agglutinition. For the cation-exchange column-based HPLC, preparation and injection of the sample hemolysate were performed according to the manufacturer s recommendations (Bio-Rad Variant HbA 1C program; Bio- Rad, Diagnostic Group). The HbA 1C specific mouse monoclonal antibody-based inhibition of latex agglutination used a DCA 2000 Hemoglobin A 1C reagent kit (13), as specified by the manufacturer (Bayer Corporation). Whole-blood HbF was measured using the Bio-Rad Variant HPLC -thalassemia Short program. alkaline and citrate agar gel electrophoresis of Hb HB electrophoresis was performed using the Ciba Corning Alkaline and Citrate Systems (Chiron Corporation) according to the manufacturer s instructions. Hb and and chain purification Hb chains were purified by a variation of the procedure published by Bucci (17). Briefly, the globin fraction from the patient s erythrocyte lysate was precipitated at 20 C with 15 volumes of acetone containing 20 ml/l concentrated HCl. The precipitate was washed four times with acetone and lyophilized. Thirty milligrams of the globin fraction were dissolved in 5 ml of buffer (20 mmol/l sodium 2-[N-morpholino]ethanesulfonate, ph 6.6, 8 mol/l urea, 30 mmol/l 2-mercaptoethanol) and applied to a Mono-S FPLC column (Pharmacia) equilibrated with the same buffer. Protein was eluted with a mmol/l NaCl linear gradient in the same buffer, and the absorbance at 280 nm was monitored. Peaks were collected and dialyzed against 10 mmol/l NH 4 HCO 3,pH 7.8. isoelectric focusing Polyacrylamide gel electrophoresis-isoelectric focusing of the various chain preparations was performed using precast IEF 3 10 MiniPlus SepraGels according to the manufacturer s instructions (Integrated Separation Systems). liquid chromatography-mass spectrometry (lc-ms) and digestion of chain Purified Hb chains were digested with sequencing grade modified trypsin (Promega, Madison, WI) by adding 5 L of aqueous 1 mol/l dithiothreitol and 5 L ofa 1 g/l trypsin solution to 100 L of a 1 g/l solution of chain in 100 mmol/l NH 4 HCO 3, ph 7.8. The reaction mixture was incubated for 24 h at 37 C, after which 7.5 L of aqueous 100 ml/l trifluoroacetic acid was added to quench the reaction. Reaction products were stored at 4 C until use. Preparative HPLC was performed on Waters hardware, using Millennium 2.10 software (18). Samples were chromatographed on a Vydac C 18 reversedphase column (218TP54) purchased from the NEST Group (Southborough, MA), using a linear trifluoroacetic acid/ acetonitrile gradient as described previously (19). Amino acid analysis of selected fractions was performed using 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate derivatization after 6 mol/l HCl hydrolysis for 24 h at 110 C as described (20). A Finnigan MAT TSQ-7000 mass spectrometer (San Jose, CA) with an electrospray ionization source was used for molecular mass analysis. Samples were either infused at 5 L/min in a mixture of 500 ml/l methanol, 490 ml/l water, and 10 ml/l acetic acid or analyzed in an LC-MS mode after microbore chromatography on a Michrom Bioresources UMA system on a Reliasil mm C 18 reversed-phase column with minor modifications (21 23). Collision-induced dissociation (CID) experiments were used to obtain partial sequence information on isolated tryptic peptides (23). N-terminal Edman sequence analysis was performed on an ABI Model 477 sequencer (Perkin Elmer). pcr and dna sequencing of gene Genomic DNA was purified from the patient s leukocytes, using sodium dodecyl sulfate-proteinase K digestion (24). A 1.7-kb DNA fragment containing all three coding exons of the Hb chain gene was amplified by PCR (30 cycles of 30 s at 95 C, 30 s at 55 C, 120 s at 72 C). The primers were as follows: 5 of the Hb gene, TGGGCATA- AAAGTCAGGGCA; and 3 of the Hb gene, CAGAT- GCTCAAGGCCCTTCATA. The entire open reading frame was sequenced directly from the purified PCR product using the Perkin Elmer Dye Primer Cycle Sequencing system (Perkin Elmer) and an Applied Biosystems 373 DNA sequencer (25). In addition to the PCR primers, the following oligonucleotides were used in sequencing: EX2 3 CCTTCCTATGACATGAA-CTTA- ACCA; EX3 5 ACAACTACAATCCAGCTACCA; EX1 3 GGTAGA-CCACCAGCAGCCT; and EX2 5 GGCATGTGGAGACAGAGAAGACT. Results and Discussion identification of Hb chain with decreased positive charge HbA 1C analysis was performed using the Bio-Rad Variant method on a whole-blood sample from a 65-year-old white male with a history of diabetes and end-stage renal failure. A large peak constituting 46% of the patient s total Hb was identified as HbA 1C by the instrument software (not shown). Glycated Hb values for this patient at our institution had been 4 8% over the previous 2 years, as measured with a boronate affinity column method (5). The immunoinhibition latex agglutination method (13) gave a result of 3.8% HbA 1C for the current sample. To rule out HbS or HbF as the cause of the unusually high HbA 1C value from the HPLC method, the patient s blood sample was analyzed with the Bio-Rad Variant thalassemia short program, which distinguishes HbF, HbS, and HbA. No HbF or HbS was identified in the patient s Hb, but a large peak eluting earlier than A 0 was evident (not shown). This peak was not identified by the

3 1298 Chen et al.: Falsely increased hemoglobin A 1C -thalassemia Short program, but the accompanying package insert states that peaks in this area are usually HbA 1c. These results clearly demonstrated the presence of a Hb species less positively charged than HbA 0 ; however, the abundance was unlikely for HbA 1c. Electrophoretic analysis at ph 8.6 in a cellulose acetate gel failed to show any difference between normal and patient Hb, whereas electrophoresis at ph 6.2 in a citrate acetate gel detected an abnormal band with mobility similar to HbF (not shown). This also suggested the loss of a positively charged group with a pi 8.6. Such a positive charge loss would most readily be ascribed to modification of either an N-terminal amino-terminal group, a histidine, an arginine, or lysine or to a mutation leading to the loss of a positively charged residue. For example, a similar pattern of Hb electrophoresis and falsely positive HbA 1c was previously reported from a patient with the Hb variant, Hb Sherwood Forest, which has an Arg3Thr mutation in the chain (9). To identify whether the charge alteration resided in the or chain, preparative cation-exchange chromatography was performed under denaturing conditions (6 mol/l urea) to separate the and chains. This procedure revealed three peaks from the patient sample, with one chain peak (y) and one chain peak corresponding to the normal and peaks, but another chain peak (x) eluted earlier than expected (Fig. 1). The earlier eluting chain (x) peak was approximately equal in size to the normal (y) peak, suggesting that the patient is heterozygotic for a chain variant. Isoelectric focusing-polyacrylamide gel electrophoresis (Fig. 1, inset) of the two chain peaks confirmed a charge difference, because the earlier eluting (x) peak has a more acidic pi compared with the normal (y) peak. To further isolate the location of the mutation, the molecular masses of all three fractions were determined by electrospray ionization MS. The values were vs the expected for the chain and vs the expected for (y), but for the (x) peak. Together these results suggested the patient to be heterozygotic for a chain mutation that produces a decreased positive charge and an increased mass of 14 daltons. the mutation is the n-terminal valine to alanine To further identify the site(s) of the mutation, each chain from the patient (peaks x and y) was digested with trypsin, and the resulting peptide mixture chromatographed on reversed-phase media in both a preparative and an LC-MS mode. There was little difference in the preparative chromatogram between the samples except for two peaks, A and B, that eluted between 36 and 37 min (Fig. 2). Peak A was much greater for the variant (x) chain (Fig. 2), whereas the opposite was true for the normal (y) chain digest (not shown). LC-MS yielded molecular mass values of 952 Da for tryptic digest T1 (peak B) and 966 Da for tryptic digest T1 (peak A). Peak B thus represents the T1 peptide of the wild-type (y) chain (residues 1 8) and peak A is the T1 peptide of the mutant (x) chain, which is 14 Da larger. On the basis of the tryptic peptides, over 94.5% of the sequence of normal Fig. 1. Chromatograms of the patient s Hb and chains separated by cation exchange HPLC. Two peaks of similar size, X and Y, are observed along with the chain peak. The (y) and peaks co-elute with normal Hb and chains (not shown). Inset: ph 3 10 isoelectric focusing-polyacrylamide gel of (x) and (y) peaks from preparative HPLC. Lane 1, Hb (x); lane 2, Hb (y) peak, lane 3, Hb (x and y) peaks, and lane S, Hb standards.

4 Clinical Chemistry 44, No. 6, Fig. 2. Preparative C 18 reversed-phase HPLC analysis of the tryptic digest for the mutant Hb chain. Peaks were assigned (after MS analysis) and numbered according to the chronological order of tryptic fragments derived from the known protein sequence. Not identified were T6 VK, T7 AHGK, and T15 YH. Fractions were collected at 0.5-min intervals beginning at 30 min. Peak A of T1 is from Hb (x), whereas peak B of T1 is from Hb (y) contamination of the Hb (x) fraction. globin was found in the map of the (x) chain, virtually eliminating the possibility that other sites in the protein contain a mutation. Thus, the site of mutation in the patient s Hb chain was localized to the N-terminal eight residues. CID spectra for the normal T1 peptide with m/z ion of and the corresponding mutant TI peptide with m/z ion of allows immediate assignment of the mutation site to one of the first two residues (Fig. 3). The b 2 fragment ion [nomenclature from (26)] for the normal peptide (upper panel) gives the expected value of 237 Da, whereas the mutant spectra (lower panel) provides a value of 251 Da, (i.e., 14 Da), for the same N-terminal fragmentation. Because it was not immediately obvious how a 14-Da change within the first two amino acids could lead to a loss of a positive charge, we characterized the mutation at the gene level. The DNA sequence of the patient s Hb gene revealed a single base heterozygous mutation (T3C) at base 2 of the codon for the first amino acid, producing a Val3 Ala substitution. Amino acid analysis for each T1 peptide (containing the first eight amino acids of chains) confirmed that the substitution of the first amino acid valine by alanine in the mutated (x) chain was the only difference between the (x) and (y) chains. Nevertheless, the substitution of valine by alanine does not result in either loss of charge or the 14-Da difference observed between the (x) and (y) chains. Indeed, the mass difference for a Val3Ala change is 28 Da. Interestingly, standard Edman N-terminal sequence analysis of the mutant (x) chain gave no sequence information suggesting that the N terminus of the mutant (x) chain may be blocked because of a posttranslational modification. In fact, Hb-Raleigh has characteristics similar to those of this patient s Hb. In Hb-Raleigh, the chain amino-terminal valine is replaced by acetylalanine (27). Such a mutation is consistent with both the loss of positive charge of chain and the 14-Da difference we observed here. Furthermore, the acetylated N-terminal residue would explain why this protein was refractory to Edman degradation. Thus, the combination of protein biochemical studies and DNA sequence analysis determined that this patient is heterozygous for Hb-Raleigh and that HB-Raleigh produces a falsely increased HbA 1c in this automated HPLC method. Ion-exchange HPLC is a widely used methodology (1) in the measurement of normal and abnormal Hb components, and the difference in charge between HbA and HbA 1C as a result of glycation of the N-terminal amino acid of Hb chain is the basis of the Bio-Rad HPLC HbA 1C method. However, this cation-exchange columnbased method, as well as other similar methods, can

5 1300 Chen et al.: Falsely increased hemoglobin A 1C Fig. 3. CID fragmentation spectra for the unaffected (top panel) and mutant (lower panel) T1 peptides. Instrumental conditions for both spectra: 40 ev collision energy and 3s/scan. CID spectra of the 2 ions for each T1 peptide at voltages from 20 to 30 ev provided complimentary data (not shown). Peak nomenclature is taken from Biemann (26), with peaks assigned according to MS-Product from clearly misidentify as HbA 1C some Hb variants that have a loss of positive charge (7 12, 16), as well as carbamylated Hbs (4 6). In most cases, the markedly high HbA 1C will cause the laboratory to take note and investigate. HPLC and electrophoretic methods are available to clearly identify or rule out HbF or HbS. For variant Hb with decreased charge or carbamylated Hb, alternative glycohemoglobin methods, such as immunoassays that utilize antibodies specific to the glycated N-terminal values of the chain, can provide accurate values (13). Similarly, boronate affinity methods (5) can overcome the falsely increased values in ion-exchange methods caused by variant or carbamylated Hbs. Furthermore, although immunoassays, boronate affinity, and ion-exchange methods detect different modifications of the Hb molecule, extensive work towards standardization (15, 28) has made results reasonably comparable with the gold-standard ionexchange method used in the DCCT trials. Thus, when abnormally high values are identified, it is almost always possible to provide useful clinical information by alternative methods. The rare Hb variant Hb-Raleigh may, however, be an exception because of the chain N-terminal modification. Because this mutation produces an acetylated alanine at the N terminus, Hb-Raleigh chain cannot be glycosylated at the N terminus and thus will not react with the immunoassay methods. Furthermore, it would exhibit decreased reactivity with boronate affinity methods. Therefore, it may not be possible to utilize Hb glycosylation as a means to monitor glucose control in patients with Hb-Raleigh. For such individuals, determination of the glycosylation of other proteins, such as albumin (28, 29), may be necessary. To our knowledge, this conundrum is restricted only to Hb- Raleigh, Hb-Long Island, and HbA 2 -Niigata, which has a chain N-terminal acetylated alanine (30), and which was recently described as having an interference in HbA 1C methods similar to the interference described here (31).

6 Clinical Chemistry 44, No. 6, We thank Michael Berk and David Windus for clinical input and Eva Reyes for technical assistance. References 1. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. New Engl J Med 1993;329: Ladenson JH, Chan KM, Kilzer P. Glycated hemoglobin and diabetes. Clin Chem 1985;31: Bunn H F, Haney D N, Gabbay KH, Gallop PM. Further identification of the nature and linkage of the carbohydrate in hemoglobin A 1c. Biochem Biophys Res Commun 1975;67: Weykamp CW, Penders TJ, Siebelder CW, Muskiet FA, van der Slik W. Interference of carbamylated and acetylated hemoglobins in assays of glycohemoglobin by HPLC, electrophoresis, affinity chromatography, and enzyme immunoassay. Clin Chem 1993;39: Scott MG, Hoffman JW, Meltzer VN, Siegfried BA, Chan K-M. Effects of azotemin on results of the boronate-agarose affinity and ion-exchange methods for glycated hemoglobin. Clin Chem 1984; 30: Lund L. Mourits-Andersen T. Sorensen PJ. Hemoglobin A 1c, uremia [letter]. Clin Nephrol 1988;29; Sosenko JM, Fluckiger R, Platt OS, Gabbay KH. Glycosylation of variant hemoglobins in normal and diabetic subjects. Diabetes Care 1980;3: Puukka R, Puukka M. Effect of hemoglobin F on measurements of hemoglobin A 1c with physicians office analyzers. Clin Chem 1994;40: Schnedl WJ, Reisinger EC, Pieber TR, Lipp RW, Schreiber F, Hopmeier P, Krejs GJ. Hemoglobin Sherwood Forest detected by high performance liquid chromatography. Am J Clin Path 1995; 104: Shah SC, Malone JI, Boisel JP, Kasper TJ. Hemoglobin South Florida, new variant with normal electrophoretic pattern mistaken for glycosylated hemoglobin. Diabetes 1986;35: Blanke S, Johnsen A, Wimberley PD. Hemoglobin Hafnia: alpha 2[beta 116 (G18)His 3 Gln]2; a new hemoglobin variant mistaken for glycosylated hemoglobin. Biochim Biophys Acta 1988; 955: Wajcman H, Blouquit Y, Riou J, Kister J, Poyart C, Soria J, Galacteros F. A new hemoglobin variant found during investigations of diabetes mellitus: Hb Pavie[alpha 135 (H18) Val to Glu]. Clin Chim Acta 1990;188: John WG. Hemoglobin A 1c measurement. new precise immunoassay method involving latex particle agglutination. Clin Chem 1996;42: Lee PD, Sherman LD, O Day MR, Rognerud CL, Ou CN. Comparisons of home blood glucose testing and glycated protein measurements. Diabetes Res Clin Practice 1992;16: Bodor GS, Little RR, Garrett N, Brown W, Goldstein DE, Nahm MH. Standardization of glycohemoglobin determinations in the clinical laboratory: three years of experience. Clin Chem 1992; 38: Halwachs-Baumann G, Katzensteiner S, Schnedl W, Pürstner P, Pieber T, Wilders-Truschnig M. Comparative evaluation of three assay systems for automated determination of hemoglobin A 1C. Clin Chem 1997; Bucci E. Preparation of isolated chains of human hemoglobin. Methods Enzymol 1981;76: Scott MG, Hock KG, Crimmins DL, Fracasso PM. HPLC method for monitoring SDZ PSC 833 in whole blood. Clin Chem 1997;43: Crimmins DL, Saylor M, Rush J, Thoma RS. Facile, in situ matrix-assisted laser desorption ionization-mass spectrometry analysis and assignment of disulfide pairings in heteropeptide molecules. Anal Biochem 1995;226: Crimmins DL, Cherian R. Increasing the sensitivity of 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate amino acid analysis; a simple solution. Anal Biochem 1997;244: Witkowska HE, Bitsch F, Shackleton CH. Expediting rare variant hemoglobin characterization by combined HPLC/electroscopy mass spectrometry. Hemoglobin 1993;17: Shackleton CH, Falick AM, Green BN, Witkowska HE. Electrospray mass spectrometry in the clinical diagnosis of variant hemoglobins. J Chromatograph 1991;562: Wada Y. In: Chapman J, ed. Protein and peptide analysis by mass spectrometry. Towata, NJ: Humana Press, 1996; Strauss WM. Preparation of genetic DNA from mammalian tissue. In: Ausubel FM, Brent R, Kingston RE, Moore DD, Seidel JG, Smith JA, Struhl, eds. Current protocols in molecular biology. New York: John Wiley and Sons, 1996: Rosenthal A, Chanock-Jones DS. New protocols for DNA sequencing with dye terminators. DNA Sequence 1992;3: Biemann K. Contribution of mass spectrometry to peptide and protein structure. Biomed Environment Mass Spect 1988;16: Moo-Penn WF, Bechtel KC, Schmidt RM, Johnson MH, Jue DL, Schmidt DE Jr, et al. Hemoglobin Raleigh(beta valine to acetylalanine). Structural and functional characterization. Biochemistry 1977;16: Goldstein DE, Little RR, Lorenz RA, Malone JI, Nathan D, Peterson CM. Tests of glycemia in diabetes. Diabetes Care 1995;18: Johnson RN, Metcalf PA, Baker JR. Fructosamine: a new approach to the estimation of serum glycoprotein, an index of diabetic control. Clin Chim Acta 1983;127: Harano T, Harano Kushida Y, Ueda S. HbA 2 -Niigata [ 1(NA1)VAL- ALA]: a new chain variant found in the Japanese population. Hemoglobin 1991;15: Watanabe T, Kato K, Yarnada D, Midorikawa S, Wakano S, Shiga M, et al. A non-diabetic case of hemoglobin variant (Hb-Niigata) with inappropriately high and low HbA 1C titers detected by different methods. Clin Chem 44:(in press), 1998.