- Email: [email protected]
Rapid Identification of Hemoglobin Variants by Electrospray Ionization Mass Spectrometry
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Rapid Identification of Hemoglobin Variants by Electrospray Ionization Mass Spectrometry Submitted 03/13/01 (Communicated by H. Ranney, M.D., 05/10/01)
B. J. Wild,1 B. N. Green,2 E. K. Cooper,1 M. R. A. Lalloz,1 S. Erten,1 A. D. Stephens,1 and D. M. Layton3 ABSTRACT: The precise identification of human hemoglobin variants, over 700 human hemoglobin variants are known, is essential for prediction of their clinical and genetic significance. A systematic approach to their rapid identification is described. Traditionally this requires protein or DNA characterization which entails lengthy analytical procedures. To overcome these obstacles a rapid approach to variant hemoglobin identification has been developed using conventional phenotypic methods combined with electrospray ionization–mass spectrometry (ESI–MS). The latter requires only a small amount of whole blood (10 l) but in most cases 2 l would have been sufficient and no preanalytical steps, such as separation of red cells or globin chains, are necessary. Aged, hemolyzed blood samples can also be analyzed. This approach has been used to positively identify 95% of the variants in over 250 samples. The remaining 5% in which a variant was detected by phenotypic techniques were not resolved by mass spectrometry. Ninety-nine different abnormalities comprising 36 ␣-chain variants, 59 -chain variants (including 2 extensions), and 4 hybrid hemoglobins were identified. These include 15 novel variants. The application of ESI–MS described requires approximately 1 h to prepare and analyze each sample and has minimal reagent costs. The turnaround time on a single sample can be as little as 2 h. This technique can now be considered a useful additional tool for reference laboratories. © 2001 Academic Press Key Words: human; hemoglobin; variant; mass spectrometry; peptide sequence.
INTRODUCTION
national guidelines (3–7) it is likely that more abnormal Hbs will be detected. The decision as to whether or not an abnormality is relevant to the patient’s clinical state or future inheritance depends on precise identification and in many situations must be reached quickly. Electrophoresis of Hb at alkaline pH has been the mainstay of first-line screening in most laboratories around the world for the last forty years (initially using paper, then starch gel and finally cellulose acetate as a fine grain, inert support medium). Such electrophoresis had the advantage that, by the time it had been refined by Schneider (8) in 1973, it could detect all the common clinically important Hb variants such as Hb S, C,
Hemoglobin (Hb) variants are most frequently detected during preanesthetic, antenatal, or neonatal screening. In addition, they may be found during the course of clinical investigation of hemolytic anemia, erythrocytosis, methemoglobinemia, or sickle cell and thalassemia syndromes. Occasionally, they are detected because they interfere with the determination of glycated Hb during diabetic monitoring. Currently over 700 Hb variants are characterized (1). With the increasing application of screening programs that has followed International Committee for Standardisation in Haematology recommendations (2), and
Correspondence and reprint requests to: B. J. Wild, Department of Haematological Medicine, King’s College Hospital, Denmark Hill, London SE5 9RS, UK. Fax: ⫹44 (0) 207 346 3514. E-mail: [email protected]. 1 Department of Haematological Medicine, Guy’s, King’s, and St. Thomas’ School of Medicine, Denmark Hill Campus, London SE5 9RS, United Kingdom. 2 Micromass UK Ltd., Tudor Road, Altrincham, Cheshire WA14 5RZ, United Kingdom. 3 Present address: Department of Haematology, Imperial College School of Medicine, Hamersmith Hospital, London W12 ONN, UK. 1079-9796/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved
691
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
D-Punjab, E, O-Arab, and Lepore. Furthermore, it separates these from Hb A and Hb F in less than 30 min. However, as more abnormal Hbs were discovered further tests such as the sickle solubility test and electrophoresis at acid pH in citrate agar were needed to obtain presumptive identification of these abnormal Hbs with sufficient accuracy for clinical purposes. Automated HPLC is now often employed as an alternative to electrophoresis, but in isolation, offers few advantages over electrophoresis at alkaline pH. However, HPLC has the advantage that it only requires a small blood sample (typically 5 l), is automated, and samples can be analyzed overnight and at weekends. Historically definitive characterization required the purification of the globin chains by preparative column chromatography, two-dimensional high-voltage electrophoresis/chromatography or high-performance liquid chromatography (HPLC) of a tryptic digest followed by amino acid analysis of any abnormal spot or peak detected. Even then, digestion with other enzymes, such as chymotrypsin, with or without amino acid sequencing was often needed. The entire procedure could take weeks or months to complete. A comprehensive approach to detection of variant hemoglobins by Riou (9) showed that by using a combination of phenotypic methods and careful calibration against known variants a presumptive identification was possible in approximately 90% of 125 different known variants. However, identification was only presumptive and this approach does not identify novel variants or precisely identify any variant. DNA techniques are also used to identify hemoglobin gene mutations. DNA sequencing is definitive, however, neither the expertise, nor the resources (for either manual or automated DNA sequencing) are widely available. Gene duplication and close homology between globin genes can also prolong localization and identification of mutations by a DNA approach alone. Sensitivity is excellent, but the reagent costs for DNA analysis are considerable. Although mass spectrometry (MS) was invented in the 1920s, the first analysis of the tryptic peptide mixtures of globin chains from human Hb was reported by Matsuo (10) in 1981 using field desorption mass spectrometry. Since then, MS has
Wild et al.
been increasingly applied to the identification of abnormal Hbs. The introduction of fast atom bombardment and liquid secondary ionization MS in 1981 (11) significantly improved the analysis of these mixtures and Rahbar and colleagues developed strategies using these techniques for variant identification (12, 13). However, it was not until the introduction of electrospray ionization mass spectrometry (ESI–MS) in 1988 (14) that it became practicable to determine the molecular weights of intact globin chains with sufficient accuracy to allow MS to be used as part of the procedure to identify abnormal Hbs (15–17). Over the past 10 years, ESI–MS has been developed by several research groups to the stage where it has become a significant adjunct to conventional phenotypic methods in the identification of variant Hbs. A comprehensive review by Shackleton and Witkowska (18) details the progress made by 1994 and gives a full account of the techniques used at that time. This group (19) also suggested that the relatively complicated and labor intensive sample preparation procedures employed prior to MS analysis, might explain why diagnostic laboratories have been reluctant to adopt MS for variant identification. These procedures involved at least some of the following: washing and lysing the red cells, removal of cell debris, removing the heme from the Hb, derivatization or oxidation of the cysteines, overnight digestion of the Hb with trypsin or other enzymes followed by HPLC separation of the tryptic peptides prior to ESI–MS. Considerable simplification of these procedures was achieved by Nakanishi (20) who digested Hb with trypsin without isolating the variant Hb and did not derivatize the cysteines prior to digestion. Furthermore, this group analyzed the mixture of tryptic peptides directly by ESI–MS without prior separation by HPLC. However, they still purified and removed the heme from the Hb prior to digestion. Peptides were observed that covered the whole of the ␣- and -chains, although some of these could not be usefully sequenced by tandem mass spectrometry, particularly the cysteine containing peptides ␣T12 and T10. They observed disulfide bonded dimers of ␣T(12–13) and T12 which partly accounted for 692
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
the poor observance of these peptides. Subsequently, homo- and heterodimers of the cysteine containing T10 and T12 peptides were observed and prevented from forming by oxidizing the cysteine prior to digestion with trypsin (21). Despite this and other reports (22, 23) complex work-up procedures and HPLC separation of the tryptic peptides are still undertaken prior to ESI–MS analysis (24, 25). A different approach using matrix-assisted laser desorption ionization (MALDI) MS and special trypsin-linked probes has recently reduced the digestion time to 15 min, although preparation of the probes is time consuming (26). Peptides were observed covering 96% of both ␣- and -chains. However, sequencing of the peptides by tandem mass spectrometry was not undertaken. This is necessary in order to positively identify variants, since the tryptic peptides from about 50% of samples contain two or more possible mutation sites that can give rise to the observed mass change of the peptide. We describe the development and application of mass spectrometry techniques which can result in unequivocal identification of the ␣ and  chain variants and hybrid hemoglobins such as Hb Lepore in samples found to be abnormal by conventional phenotypic methods. We have also demonstrated the ability of ESI–MS to detect hemoglobin variants such as some unstable variants and M hemoglobins that are “silent” using conventional electrophoretic techniques. Emphasis has been placed on minimizing sample preparation and analysis times by using diluted whole blood without prior purification, derivatization or separation into its constituent chains for the mass spectrometric and enzymatic digest procedures and by analyzing the products of the digest by ESI–MS without prior chromatographic separation.
presence of a variant and often indicates whether the mutation is in the ␣- or -chain. It may also give a presumptive identification. Mass spectrometric analysis is first carried out on diluted whole blood by Step 2 in order to observe the globin chains. As its name implies, mass spectrometry essentially determines the mass (molecular weight) of a peptide or protein and all deductions made from the technique depend on measuring mass or mass difference with high accuracy. In contrast to most phenotypic methods, which analyze Hbs as noncovalently assembled tetramers or dimers, ESI–MS in this application analyses Hb in denaturing solvents, so that the globin chains and heme group are analyzed as separate entities. In essence, the mass spectrometer separates these components according to their masses. Thus, the various chains, their glycated and posttranslationally modified forms are separated by the spectrometer, providing their masses differ sufficiently from one another. In practice, two globin chains that differ in mass by 6 Daltons (Da) or more are resolved by ESI–MS thus enabling the masses of the intact chains to be individually determined with an accuracy of better than ⫾ 0.3 Da. This minimum of 6 Da separation is a characteristic of ESI–MS when measuring intact globin chains. The sequence masses of the principal chains are 15,126.4 Da (␣) and 15,867.2 Da () which differ in mass by approximately 741 Da. Since the highest mass change for a single amino acid mutation is 129 Da (Gly7Trp), the majority of mutations can be unambiguously associated with either the ␣- or -chain from their measured masses and the proportions of variant and normal chains determined. Moreover, again provided the mass difference between the two chains in a heterozygote is at least 6 Da, their masses can be measured without prior separation of the normal and abnormal hemoglobins. From the mass difference between the normal and variant chains, a list of possible variants can then be deduced, based on the single amino acid changes that are created by single nucleotide changes in the coding triplet (Table 1). Of 141 possible mutations shown in Table 1, 127 (90%) give rise to ⬎6 Da mass change. Mutations from isoleucine are ex-
METHODS Optimization of Strategy Figure 1 outlines the essential analytical steps developed to identify a variant hemoglobin. Generally, phenotypic analysis (Step 1) reveals the 693
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
cluded since the normal ␣- and -chains contain no isoleucine. The results from Step 2 divide the samples into two categories, namely those that show both chains in heterozygotes distinctly separated from one another and those that show no detectable separation. The latter samples are then classified as likely to contain two chains that differ in mass by less than 6 Da and need digestion with trypsin (Step 4) before the variant can be detected by MS. Next, tandem mass spectrometry may be carried out on the intact variant chain (Step 3), either from the whole blood sample or from the isolated variant. The masses of characteristic peaks in this spectrum are used to determine the group of amino acids that contains the mutation, e.g., (1– 47), (124 –146), ␣(1–76), ␣(114 –141) and hence limit the number of possible variants that could be present. This procedure is normally undertaken when the variant tryptic peptide is not found in Step 4 or when the variant is suspected to be a hybrid as indicated by the results from Steps 1 and 2. In Step 4, the mixture of peptides produced by digestion of the underivatized Hb with trypsin is analyzed without prior separation by ESI–MS in order to identify the peptide that contains the variant. The mass difference between the normal and variant globin chains determined in Step 2 is used to assist this procedure unless the mass difference is ⬍6 Da when tables listing all possible variants and the masses of their corresponding peptides are used. Hitherto, the cysteines in Hb have often been derivatized by treatment with iodoacetamide (carbamidomethylation) prior to enzymatic digestion, otherwise the cysteine containing tryptic peptides were not detected or were only present in very small amounts. Although derivatization takes only 10 min, removing the excess reagents prior to digestion is labor intensive and time consuming. The recovery of the cysteine containing peptides from the underivatized peptide mixture is significantly improved by simply reducing the latter with dithiothreitol (DTT) before analysis by ESI–MS. At this stage most variants can be detected, including those in heterozygotes in which the two chains differ in mass by as little as 1 Da from one another. Ap-
Wild et al.
proximately 50% of variants can be identified at this stage. These include variants that produce a tryptic peptide containing only one possible mutation that can give rise to the mass change determined in Steps 2 or 4. Others involve mutations to arginine and lysine that produce two “new” peptides with masses characteristic of the mutation. One mutation (Gln3 Lys), in which the mass difference between the normal and variant chain is zero, can be detected and identified in this way. Finally, in Step 5, the variant tryptic peptide is sequenced by tandem MS in order to identify or confirm the mutation unambiguously. Phenotypic Analysis Cellulose acetate electrophoresis (CAM) at alkaline pH and acid agarose electrophoresis were undertaken using Helena equipment and reagents (Helena UK, Newcastle, UK). Sickle solubility tests were undertaken using standard techniques (27). Isoelectric focusing (IEF) was undertaken using the Resolve reagents and Pharmacia LKB Multiphor equipment (Perkin–Elmer Wallac Inc.). Highperformance liquid chromatography (HPLC) was undertaken using the Bio-Rad Variant equipment and “beta thal short” reagents (Bio-Rad, UK; Hemel Hempstead, UK) (28). Isolation of hemoglobin variants such as Hb Lepore-Boston-Washington was undertaken by a modification of the ICSH electrophoretic technique for Hb A2 quantitation (29). Sample Preparation for Mass Spectrometry Step 1. Initially, a stock solution was prepared by diluting 10 l of blood 50-fold with 490 l of HPLC grade water, although later 2 l of blood was diluted with 98 l of HPLC grade water and proportionately less was used in Steps 2–5 below. This stock solution is sufficient for the entire analysis. Working solutions for direct introduction into the mass spectrometer were prepared as follows. Step 2. Twenty microliters of the stock solution was diluted 10-fold by adding 180 l of 1:1 acetonitrile:water containing 0.2% formic acid. Just prior to analysis, the resulting solution was 694
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
(Sigma C-3142) as follows. 2 l of 0.5 mg/ml ␣-chymotrypsin was added to 20 l of the tryptic digest solution, mixed and incubated at 25°C for 30 min. The resulting solution was then diluted with 180 l of 1:1 acetonitrile:water containing 0.2% formic acid to make the working solution for analysis by ESI–MS.
de-salted by manually shaking it for 15–30 s with approximately 5 mg of cation exchange resin beads (Bio-Rad Labs, AG 50W-X8, Cat. No. 1435441) that had been twice washed previously with water and this solution was used for ESI–MS analysis. Step 3. Twenty microliters of the stock solution was diluted 10-fold by adding 180 l of 1:1 acetonitrile:water containing 0.2% formic acid and this solution was used for tandem MS without desalting.
Mass Spectrometry Twenty-microliter aliquots of the working solutions were introduced at 5 l/min into the electrospray source of a Quattro II or Quattro LC tandem mass spectrometer (Micromass UK Ltd., Wythenshawe, Manchester, UK) while scanning and acquiring data over appropriate mass-tocharge ratio (m/z) ranges. These were: m/z 930– 1180 (Step 2), 50 –1500 (Step 3), 150 –2000 (Step 4) and 50 –2000 (Step 5). Acquisition times were 2 min for Step 2, 3 min for Steps 3 and 4, and 5–10 min for Step 5. Mass scale calibration of the spectra obtained in Step 2 employed the multiply charged normal ␣-chain peaks present in each sample. Mass scale calibration of the spectra obtained in the other steps employed the Nan⫹1In peaks from separate introductions of a 0.5 mg/ml solution of NaI in 1:1 2-propanol:water. The baseline subtracted m/z spectra (m/z 970–1180) from Step 2 were deconvoluted to present the spectrum on a true molecular weight scale using the maximum entropy (MaxEnt) based software supplied with the instrument.
Steps 4 and 5. One hundred microliters of the stock solution was first denatured by adding 10 l of 1% aqueous formic acid and 10 l of acetonitrile. After mixing, the solution was allowed to stand at room temperature (25°C) for 5 min. Then 6 l of 1 M aqueous ammonium bicarbonate solution and 5 l of a 5 mg/ml solution of TPCK treated trypsin (Sigma T-8642) was added. The resulting solution, after mixing and centrifuging for approximately 15 s, was incubated at 37°C. A fine precipitate will appear after adding the ammonium bicarbonate, indicating that the hemoglobin has been effectively denatured. This precipitate generally disappears within a minute of adding the trypsin. The solution should not be centrifuged until it has become clear. After 30 min, 30 l of this digest was diluted with 270 l of 1:1 acetonitrile:water containing 0.2% formic acid to give the working solution for ESI–MS analysis. Most variants can be identified from the 30 min (short) digest. The remainder of the digest solution was allowed to incubate for 12–15 h in case the variant was not identified from the short digest. If required, 30 l of the 12- to 15-h digest solution was reduced at room temperature for 10 min by adding 3 l of aqueous 100 mM dithiothreitol (DTT) in order to release the cysteine containing tryptic peptides. The resulting solution was then diluted with 270 l of 1:1 acetonitrile: water containing 0.2% formic acid to give the working solution of the reduced digest for analysis by ESI–MS. To reduce the size of the ␣T9 peptide and hence facilitate identification of variants suspected of being present therein, the tryptic digest was mildly digested with ␣-chymotrypsin
DNA Analysis DNA was prepared from blood by a conventional method and sequence analysis of globin genes was undertaken by cycle sequencing (fmol, Promega). Where possible, novel mutations were confirmed by restriction analysis of DNA amplified by the polymerase chain reaction (PCR) from genomic DNA. Endonucleases (New England Biolabs, MBI Fermentas, Gibco, Promega) were used according to the manufacturer’s instructions. RESULTS Using the strategy depicted in Fig. 1, mutations were positively identified in 95% of over 695
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
FIG. 1. Steps taken to identify a variant Hb by combined phenotypic analysis and mass spectrometry.
250 blood samples found to be abnormal by either phenotypic methods and/or MS. Ninety nine different variants, including 15 novel hemoglobin variants, were identified (Table 2). These comprise 36 ␣-chain variants, 59 -chain variants (including 2 extensions) and 4 hybrid hemoglobins. Figures 2– 4 show typical MS spectra that illustrate how the data from analytical Steps 2–5 logically lead to the positive identification of a -chain variant. In this example, the variant is due to the previously unreported and electrophoretically silent mutation 13 Ala3 Val. MS was undertaken on this sample because it was suspected that the Hb might be abnormal because of a history of neonatal jaundice and hemolysis (the sample was obtained when the individual was 11 months old with an Hb F of 3.1%). Figure 2 shows MaxEnt deconvoluted mass spectra from the diluted blood of (A) a normal adult and (B) the patient with the abnormal Hb (Step 2). The latter spectrum clearly shows two -chain peaks of similar intensity, one at 15,867.3 Da corresponding to the normal -chain (sequence mass 15,867.2 Da) and one at 15,895.4 Da, corresponding to the
variant, thus indicating the presence of a normal/ variant -chain heterozygote. Furthermore, the mass of the variant chain is 28.1 Da higher than the normal -chain. Assuming the mutation is due to a single base change in the coding triplet, there are only three single amino acid changes that can give rise to 28 Da mass increase between normal and variant, namely Ala3 Val, Gln3 Arg and Lys3 Arg (Table 1). It follows that the total number of possible single amino acid mutations that can give a 28 Da increase in mass is 29 at this stage in the analysis, since there are 15 alanine, 3 glutamine and 11 lysine residues in the normal -chain. However, the two mutations involving arginine are unlikely, since these mutations create a charge change resulting in a difference in mobility of the variant hemoglobin which would have been detected by conventional phenotypic means. The results from Step 4 are given in Fig. 3 which shows part of the mass spectra of tryptic digests from (A) a normal control and (B) the sample containing the variant. The two spectra are remarkably similar except for the occurrence of an extra peak at m/z 480.7 in the spectrum from 696
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
FIG. 2. MaxEnt deconvoluted ESI mass spectra of 500-fold diluted and desalted blood obtained under denaturing conditions: (A) from a normal adult sample and (B) from a patient suspected to have a phenotypically silent abnormal Hb. The occurrence of two peaks separated by 28.1 Da clearly indicates the presence of a -chain heterozygote.
from (A) the normal peptide ion and (B) the variant peptide ion. The increase in mass of 28 Da between normal and variant at a particular position (y5⬙ ion) positively identifies the variant as 13 Ala3 Val. This amino acid substitution fully accounts for the mass difference between the normal and variant chains determined in Step 2 (Fig. 2), i.e., the presence of a second mutation in the same chain is unlikely. Thirteen of the 15 novel variants were also examined by DNA techniques and in all cases the results of the DNA analysis agreed with those obtained by mass spectrometry. No material was available for DNA analysis of one novel variant and the other is still under analysis.
the abnormal sample and a reduction in the relative intensity of the peak at m/z 466.7. The m/z difference between these two peaks is 14.0, which corresponds to a mass difference of 28.0 Da because the ions are doubly charged. These data imply that the variant occurs in tryptic peptide T2 since the mass difference between the normal and variant -chains was found in Step 2 to be 28 Da. There are three sites in this peptide that can give rise to 28 Da increase in mass by single amino acid mutations, namely at amino acid numbers 2, 5 and 9 which correspond to 10 Ala3 Val (Hb Iraq-Halabja), 13 Ala3 Val and 17 Lys3 Arg in the -chain respectively. The latter two mutations have not been described previously. Finally, tandem mass spectrometry was undertaken on the normal and variant T22⫹ peptide ions in order to identify the site of the mutation (Step 5). The resulting spectra are shown in Fig. 4
DISCUSSION The results reported here demonstrate that simplified mass spectrometric procedures together 697
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
5) was undertaken at least once on every variant in Table 2 to confirm its identity, even though it may have already been identified in Step 4. The technique used only a small amount of blood (10 l). Although 10 l was used for convenience, in 90% of samples analyzed 2 l would have been sufficient. To minimize sample preparation time, whole blood was used in almost all cases. Stored, aged hemolyzed samples were also successfully analyzed by this technique. In no case was it necessary to wash the red cells prior to analysis by ESI–MS. Although the majority of samples were analyzed from diluted whole blood, it was found that isolation of the variant Hb was necessary in 12% of cases (9% of different variants) in order to detect and identify the variant peptide in the digest mixture. These cases mainly occurred with ␣-chain variants in heterozygotes when the variant was present in only relatively small amounts e.g., Hb Fort Worth (␣27 Glu3 Gly) and the novel variant ␣76 Met3 Arg (10.8% and 9-10% of total Hb respectively). Another case occurred with Hb Q-Thailand (␣74 Asp3 His), which was difficult to distinguish from Hb Q-Iran (␣75 Asp3 His) by tandem MS until isolated. The hybrid Hb Lepore-Boston-Washington (12% of total Hb) was also isolated because its mass is only 2 Da lower than the normal -chain and hence is not resolved from the normal -chain in Step 2. The only -chain variant that needed isolation was Korle-Bu (73 Asp3 Asn) because the variant tryptic peptide appeared to be at much lower abundance in the digest mixture from nonisolated Hb than expected and it was later shown that this was due to a somatic mutation. In such cases, the analysis was undertaken on the variant Hb after preparative isolation using cellulose acetate electrophoresis followed by elution (29). Identification of ␦-chain variants would require isolation of the variant chain before analysis, since they are present at only 1–2% of the total hemoglobin. Samples with ␥-chain variants, except perhaps from newborn children, would also require isolation and concentration before analysis. Sample preparation time was minimized by undertaking digestion with trypsin on diluted whole blood without prior derivatization of the cysteines. Al-
TABLE 1 Nominal Mass and Amino Acid Residue Changes Produced by Single Base Changes in the Nucleotide Coding Triplet Mass change
Amino acid change
Mass change
Amino acid change
0
Gln7Lys Ile7Leu Asn7Asp Gln7Glu Ile7Asn Lys7Glu Lys7Met Pro7Thr Gln7His Ser7Pro Thr7Ile Thr7Asn Asn7Lys Asp7Glu Gly7Ala Ser7Thr Val7Ile Val7Leu Ile7Lys Leu7Gln Ala7Ser Phe7Tyr Pro7Leu Ser7Cys Val7Asp Ile7Met Leu7Met His7Arg Asp7His Asn7His Leu7His Met7Arg Ala7Pro His7Tyr Ser7Ile Ser7Leu Ser7Asn Thr7Lys
28
Ala7Val Gln7Arg Lys7Arg Ala7Thr Arg7Trp Gly7Ser Thr7Met Val7Glu Pro7Gln Val7Met Ile7Phe Leu7Phe Pro7His Gly7Val Ile7Arg Leu7Arg Ala7Asp Cys7Phe Gly7Cys Asp7Tyr Val7Phe Asn7Tyr Cys7Arg Thr7Arg Ala7Glu Gly7Asp Pro7Arg Cys7Tyr Ser7Phe Ser7Arg Gly7Glu Leu7Trp Ser7Tyr Cys7Trp Gly7Arg Ser7Trp Gly7Trp
1
3 4 9 10 12 13 14
15 16
18 19 22 23 24 25 26
27
30
31 32 34 40 42 43 44 46 48 49 53 55 58 59 60 69 72 73 76 83 99 129
Wild et al.
with digestion with trypsin can identify the precise structural abnormality in approximately 95% of the ␣- and -chain variant hemoglobins which had been detected by conventional methods in routine service laboratories. We also detected, and identified, some variants that had not been detected by conventional screening techniques. These were detected in Step 2 while either investigating severe hemolytic anemia (e.g., Hb Southampton) or unexplained erythrocytosis (e.g., Hb Johnstown). Tandem mass spectrometry (Step 698
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
FIG. 3. Part of the mass spectra of tryptic digests from (A) a normal control and (B) the sample containing the variant. The appearance of the peak at m/z 480.7 in the spectrum from the abnormal Hb digest, 28.0 Da higher than the normal T2 peak shows that the variant lies in this peptide, i.e., between 9 and 17 inclusive. The difference on the m/z scale is 14.0 (28.0/2) because the ions are doubly charged.
though for 75% of this work digestion with trypsin was undertaken overnight (12–15 h), the recent introduction of the 5-min denaturing step described under Methods has dramatically reduced the time to produce viable tryptic peptide mixtures. This development significantly improves the reliability of the procedure and undigested ␣- and -chains disappear after incubation for 30 min. Even after 15 min incubation, many variants can be detected and the clinically important variants C, E. D-Punjab, and O-Arab can be positively identified after this very short digest time. Thus same day turn round of urgent samples is now practicable. However, the 12- to 15-h incubation time is still required in a few cases, when reduction with DTT is necessary in order to release the cysteine containing peptides. The cysteine containing peptides are present at high levels in the short digests without reduction. The time to
manually prepare the samples and to acquire and process the mass spectral data to a stage suitable for interpretation was less than 1 h per sample. Interpretation of the data to positively identify the amino acid substitution (or other structural change) although generally rapid, can be time consuming (up to 2 h) and requires considerable experience. However, an experienced analyst can usually identify 10 variants in two to three days. Analysis times are based on manual sample introduction and data processing on a batch basis. A saving of 10 –20 min/sample could be obtained by automating acquisition and print out of the mass spectra in analytical Steps 2, 3, and 4 given additional hardware and software development. Software to list the possible mutations and the corresponding tryptic peptides containing the mutation after Steps 1–3 would also reduce interpretation time. 699
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
FIG. 4. Tandem mass spectra of (A) the normal T22⫹ tryptic peptide ion and (B) the variant T22⫹ tryptic peptide ion. The mass increase of 28.1 Da in the spectrum of the variant at fragment ion y5 positively identifies the variant as 13 (Ala3 Val).
Tryptic peptides covering effectively the whole of the ␣- and -chains were observed in the spectrum of the reduced digest with remarkably little interference, and many mutations that produce a mass change of only 1 Da were identified in heterozygotes from such spectra. Table 2 shows 4 ␣- and 12 -chain variants in this category. All the small peptides (␣T2, ␣T7, ␣T10, ␣T14, T6, T7, and T15) except the two mono-peptides (␣T8 and T8) were observed at good signal-to-background ratio. The latter two were observed in the combinations ␣T(8 –9) and T(8 –9). As observed previously (18), lysine bonds at ␣127 and 95 are incompletely cleaved, leaving significant quantities of ␣T(12–13) and T(10 –11) in the digest mixture. Nevertheless, the ␣T12, ␣T13, T10 and T11 peptides were present at sufficiently high levels to allow viable mass spectra and tandem mass spectra to be acquired from them. An advantage of undertaking
tandem MS on ions selected by the mass spectrometer from the peptide mixture is that viable spectra from low abundance ions can, if necessary, be obtained by acquiring the data over an extended period of time (10 –15 min). This time compares with less than a minute available during elution of a peptide during an HPLC–MS run, and is still considerably shorter than the time to collect fractions from a preparative HPLC run. Although there was no problem observing the large ␣T9 (62–90) and ␣T12 (100 –127) peptides, it was difficult to detect mutations in heterozygotes in these peptides when the mass of the variant differed by only 1 Da from normal. This is mainly because the dominant ions from these peptides carry 3 or 4 charges leading to unresolved isotope peaks. Also, the proportion of the variant is generally 25% or less in many ␣-chain heterozygotes. For the ␣T9 peptide, these difficulties were partly overcome by using the non700
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
tryptic peptides, ␣(62–72) and ␣(81–90) which often occurred at satisfactory levels in the 12- to 15-h tryptic digest. Alternatively, a simple procedure was developed using mild digestion of the tryptic peptide mixture with ␣-chymotrypsin (Step 4). This procedure cleaves the ␣T9 tryptic peptide into two peptides, ␣(62– 80) and ␣(81– 90). These non-tryptic peptides were used to identify several of the variants occurring between ␣62 and ␣90 (Table 2). Reducing the size of the ␣T12 (100 –127) peptide remains one of the problem areas in this simplified approach. It is probable that some of the 13 samples for which no abnormality was detected occur in this region of the ␣-chain since only two variants were identified between ␣100 and ␣127, whereas 13 variants were identified between ␣62 and ␣90 (Table 2). Eight of the 13 samples in which we failed to identify an abnormality occurred in the first 60 abnormal Hbs we analyzed before we started isolating the variant Hb and many of these may therefore have been identified had they been isolated. Unfortunately there is insufficient blood specimen to reanalyze these samples. Except for Lepore-Boston-Washington, positive identification of the 10 known hybrid Hbs in heterozygotes from their masses (Step 2) and by tandem MS (Step 3) without preparative isolation of the variant Hb is feasible. Their masses differ from normal by more than 6 Da and therefore can be separated from the normal -chain by the mass spectrometer. In addition, we predict that they would give distinctive tandem mass spectra. The four hybrids we have so far encountered (Table 2) gave quite characteristic tandem mass spectra (30). However, a compound heterozygote of Lepore-Hollandia with sickle would not be identifiable in this way without separation since the masses of the two variant chains differ from one another by only 1 Da (18, 31). For several of the common clinically important variants (Hb C, E, D-Punjab, and O-Arab) the mutation produces only 1 Da change in the mass of the -chain. Hence, in heterozygotes, the normal and variant chains are not resolved from one another by the mass spectrometer in Step 2 and the measured mass is the abundance-weighted mean of the masses of the two chains. Hence,
when the two chains occur in approximately equal abundance e.g., in Hb C trait (Hb A ⫹ Hb C), the mass change from normal is theoretically 0.5 Da and can just be detected reliably. However, when the variant chain is present at a lower concentration than the normal chain, e.g., in Hb E trait (Hb A ⫹ Hb E), the net mass change can be as low as 0.2 Da, and cannot be detected reliably in Step 2. Consequently, ESI–MS of diluted whole blood without prior separation cannot be used to screen for these variants, and conventional phenotypic methods are still necessary in order to detect them in the first instance. Nevertheless, after digestion with trypsin, all these variants can be detected and identified unambiguously by MS in Step 4 whether present in the homozygous, heterozygous or compound heterozygous state. Therefore, the combination of phenotypic methods and modern MS facilitates both the detection and positive identification of these common variants. In contrast, most uncommon variants can be detected by analyzing diluted whole blood (Step 2), because the mutation produces a mass change greater than 6 Da. Of 141 possible mutations caused by a single base change in the coding triplet, 127 (90%) can be detected in this way. Thus, MS has the potential to detect as well as identify most uncommon variants including those that are phenotypically silent. Therefore, in cases of obvious unexplained hematological abnormality, MS should be considered as a diagnostic procedure to detect the presence of an abnormal hemoglobin. Of the remaining 14 mutations, i.e., those that give ⬍6 Da change by MS (Leu3 Ile, Asn3 Ile, Gln7Lys, Lys7Glu, Asn7Asp, Gln7Glu, Lys7Met, and Pro7Thr), 10 can be easily detected by conventional phenotypic methods. Hence, most variants can be detected by a combination of conventional phenotypic methods and MS. The mutation Leu3 Ile can be neither detected nor identified by MS because it produces neither a mass change nor a new cleavage site after digestion. The analytical power of the present generation of electrospray mass spectrometer instruments has enabled definitive identification of most hemoglobin variants within 24 h requiring only a very small sample (2 l) of whole blood, although 10 701
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
TABLE 2 Variants Identified by Combined Phenotypic Analysis and ESI–MS Mutation
Name
Mass change
Mutation
␣-Chain variants ␣5 Ala3Asp ␣12 Ala3Asp ␣16 Lys3Glu ␣18 Gly3Arg ␣20 His3Gln ␣21 Ala3Pro ␣27 Glu3Ala ␣27 Glu3Gly ␣27 Glu3Val ␣44 Pro3Ala ␣47 Asp3His ␣49 Ser3Arg ␣50 His3Leu ␣54 Gln3Arg ␣57 Gly3Asp ␣63 Ala3Thr ␣64 Asp3His ␣68 Asn3Lys ␣74 Asp3Asn ␣74 Asp3His ␣75 Asp3Tyr ␣76 Met3Arg ␣78 Asn3Lys ␣81 Ser3Cys ␣84 Ser3Arg ␣85 Asp3Asn ␣85 Asp3Tyr ␣87 His3Tyr ␣89 His3Gln ␣90 Lys3Asn ␣94 Asp3Asn ␣94 Asp3Tyr ␣102 Ser3Arg ␣112 His3Asp ␣120 Ala3Glu ␣141 Arg30
J-Toronto J-Paris-I I Handsworth Le Lamentin Fontainebleau Novel Fort Worth Spanish Town Novel Hasharon Savaria Novel Shimonoseki J-Norfolk Novel Q-India G-Philadelphia G-Pest Q-Thailand Winnipeg Novel Stanleyville-II Nigeria Etobicoke G-Norfolk Atago M-Iwate Novel J-Broussais Titusville Setif Manitoba Hopkins-II J-Meerut Koellikera
a
Marseille South Florida Raleigh Novel Deer Lodge Tyne C Sickle G-Siriraj Novel Belfast J-Baltimore D-Ouled Rabah
Posttranslational modification.
Mass change
-Chain Variants 44 44 1 99 ⫺9 26 ⫺58 ⫺72 ⫺30 ⫺26 22 69 ⫺24 28 58 30 22 14 ⫺1 22 48 25 14 16 69 ⫺1 48 26 ⫺9 ⫺14 ⫺1 48 69 ⫺22 58 ⫺156
-Chain Variants (⫺1)Met-(⫹1)Val(⫹2)Pro-Leu(⫺1)Met-(⫹1)Met(⫹2)His-Leu1 Val3Ac-Ala 1 Val3Gly 2 His3Arg 5 Pro3Ser 6 Glu3Lys 6 Glu3Val 7 Glu3Lys 13 Ala3Val 15 Trp3Arg 16 Gly3Asp 19 Asn3Lys
Name
91 163 14 ⫺42 19 ⫺10 ⫺1 ⫺30 ⫺1 28 ⫺30 58 14
20 Val3Met 21 Asp3Val 22 Glu3Ala 22 Glu3Gln 22 Glu3Lys 26 Glu3Lys 27 Ala3Val 30 Arg3Ser 40 Arg3Lys 46 Gly3Glu 51 Pro3His 52 Asp3Asn 56 Gly3Asp 57 Asn3Lys 58 Pro3Arg 69 Gly3Ser 71 Phe3Ser 73 Asp3Asn 74 Gly3Val 82 Lys3Arg 82 Lys3Met 83 Gly3Asp ⫹ 6 Glu3Lys 85 Phe3Ser 90 Glu3Lys 95 Lys3Asn 95 Lys3Glu 98 Val3Met 99 Asp3His 101 Glu3Ala 104 Arg3Thr 106 Leu3Pro 108 Asn3Lys 109 Val3Leu 109 Val3Met 113 Val3Glu 116 His3Tyr 121 Glu3Gln 121 Glu3Lys 124 Pro3Gln 126 Val3Glu 128 Ala3Asp 131 Gln3Glu 133 Val3Ala 134 Val3Glu 143 His3Arg 143 His3Tyr
Olympia Novel G-Coushatta D-Iran E-Saskatoon E Grange-Blanche Tacoma Athens-GA K-Ibadan Novel Osu-Christiansborg J-Bangkok G-Ferrara Dhofar City of Hope Christchurch Korle-Bu Bushwick Novel Helsinki Novel, Pyrgos, & C in same chain Buenos Aires Agenogi Detroit N-Baltimore Ko¨ln Yakima Novel Sherwood Forest Southampton Presbyterian Johnstown San Diego New York Novel D-Los Angeles, D-Punjab O-Arab Ty Gard Hofu J-Guantanamo Camden Novel North Shore Abruzzo Old Dominion
␦22—50 ␦87—116 A ␥81—86 22—␦50
Hybrids Lepore-Hollandia Lepore-Boston-Washington Kenya P-Nilotic
32 ⫺16 ⫺58 ⫺1 ⫺1 ⫺1 28 ⫺69 ⫺28 72 40 ⫺1 58 14 59 30 ⫺60 ⫺1 42 28 3 57 ⫺60 ⫺1 ⫺14 1 32 22 ⫺58 ⫺55 ⫺16 14 14 32 30 26 ⫺1 ⫺1 31 30 44 1 ⫺28 30 19 26 ⫺31 ⫺2 55 88
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
l has often been used for convenience. The initial use of traditional phenotypic methods remains the first line screening approach for detecting some common clinically important variants such as Hb C, E, D-Punjab, and O-Arab which only give a small mass change (1 Da) from normal. Nevertheless, the definitive identification can still usually be obtained within 24 –36 h. Variants present in small amounts relative to the other hemoglobins, such as Hb Lepore or the unstable hemoglobins may require an initial step of purification on cellulose acetate, but this step only adds another 24 h to the procedure. However, as in other fields of laboratory medicine it is important to emphasize that interpretation of ESI–MS data requires considerable expertise. ESI–MS enables the definitive identification of a large proportion of variant hemoglobins detected by conventional separation techniques. Faced with an extensive and growing catalogue of human hemoglobins this facility greatly enhances the ability to predict the clinical and genetic implications of individual variants. ESI–MS constitutes an extremely versatile addition to the techniques at present available to reference laboratories and provides a rapid alternative to DNA analysis.
4.
5.
6.
7.
8.
9.
10.
11.
ACKNOWLEDGMENTS We thank the following for providing some of the abnormal Hb samples mentioned in this paper: Dr. P. Bolton-Maggs (Alder Hey Hospital, Liverpool), Professor T. R. J. Lappin (Royal Victoria Hospital, Belfast), Dr. D. C. Rees (Royal Hallamshire Hospital, Sheffield), Professor T. M. Reynolds (Queen’s Hospital, Burton-onTrent), Dr. N. B. Roberts (Royal Liverpool University Hospital), and Dr. K. Wiener (North Manchester General Hospital).
12.
13.
14.
REFERENCES 1.
International Hemoglobin Information Centre Variant List (1997) Hemoglobin 21(6), 505– 622. 2. International Committee for Standardisation in Haematology (1988) Recommendations for neonatal screening for haemoglobinopathies. Clin. Lab. Haematol. 10, 335–345. 3. British Committee for Standards in Haematology
15.
703
(1988) Guidelines for Haemoglobinopathy Screening. Clin. Lab. Haematol. 10, 87–94. British Committee for Standards in Haematology (1994) Guidelines for fetal diagnosis of globin gene disorders. J. Clin. Pathol. 47, 199 –204. British Committee for Standards in Haematology (1994) Guidelines for the investigation of the alpha and beta thalassaemia traits. J. Clin. Pathol. 47, 289 – 295. British Committee for Standards in Haematology (1998) Guideline, the laboratory diagnosis of haemoglobinopathies. Br. J. Haematol. 101, 783–792. The Standing Medical Advisory Committee on Sickle Cell, Thalassaemia and Other Haemoglobinopathies (1993) Report of a working party of The Standing Medical Advisory Committee on Sickle Cell, Thalassaemia and other Haemoglobinopathies. HMSO, London. Schneider, R. G. (1973) Sickle cell disease. In Developments in Laboratory Diagnosis (Abramson, H., Bertles, J. F., and Wethers, D. I., Eds.), pp. 230 –243. Mosby, St. Louis, MO. Riou, J., Godart, C., Hurtrel, D., Mathis, M., Bimet, C., Bardakdjian-Michau, J., Prehu, C., Wajcman, H., and Galacteros, F. (1997) Cation-exchange HPLC evaluated for presumptive identification of hemoglobin variants. Clin. Chem. 43, 34 –39. Matsuo, T., Matsuda, H., Katakuse, I., Wada, Y., Fujita, T., and Hayashi, A. (1981) Field desorption mass spectra of tryptic peptides of human hemoglobin chains. Biomed. Mass Spectrom. 8, 25–30. Barber, M., Bordoli, R. S., Sedgwick, R. D., and Tyler, A. N. (1981) Fast atom bombardment of solids (FAB): A new ion source for mass spectrometry. J. Chem. Soc. Chem. Commun., 325–327. Rahbar, S., Lee, T. D., Baker, J. A., Rabinowitz, L. T., Asmerom, Y., Legesse, K., and Ranney, H. M. (1986) Reverse phase high-performance liquid chromatography and secondary ion mass spectrometry. A strategy for identification of ten human hemoglobin variants. Hemoglobin 100, 379 – 400. Lee, T. D., and Rahbar, S. (1991) The mass spectral analysis of hemoglobin variants. In Mass Spectrometry of Peptides (Desiderio, D. M., Ed.), pp. 257–274. CRC Press, Boca Raton, FL. Covey, T. R., Bonner, R. F., Shushan, B. I., and Henion, J. (1988) Determination of protein, oligonucleotide and peptide molecular weights by ion-spray mass spectrometry. Rapid Commun. Mass Spectrom. 2, 249 –256. Green, B. N., Oliver, R. W. A., Falick, A. M., Shackleton, C. H. L., Roitman, E., and Witkowska, H. E. (1991) Electrospray MS, LSIMS and MS/MS for the rapid detection and characterization of variant hemoglobins. In Biological Mass Spectrometry (Burlin-
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
16.
17.
18.
19.
20.
21.
22.
23.
game, A. L., and McCloskey, J. A., Eds.), pp. 129 – 146. Elsevier, Amsterdam. Falick, A. M., Shackleton, C. H. L., Green, B. N., and Witkowska, H. E. (1990) Tandem mass spectrometry in the clinical analysis of variant hemoglobins. Rapid Commun. Mass Spectrom. 4, 396 – 400. Witkowska, H. E., et al. (1991) Sickle cell disease in a patient with sickle cell trait and compound heterozygosity for hemoglobin S and hemoglobin QuebecChori. N. Engl. J. Med. 325, 1150 –1154. Shackleton, C. H. L., and Witkowska, H. E. (1994) Mass spectrometry in the characterization of variant hemoglobins. In Mass Spectrometry: Clinical and Biomedical Applications (Desiderio, D. M., Ed.), Vol. 2, pp. 135–199. Plenum Press, New York. Witkowska, H. E., Bitsch, F., and Shackleton, C. H. L. (1993) Expediting rare variant hemoglobin characterization by combined HPLC/electrospray mass spectrometry. Hemoglobin 17(3), 227–242. Nakanishi, T., Kishikawa, M., Shimizu, A., Hayashi, A., and Inoue, F. (1995) Assignment of the ions in the electrospray ionization mass spectra of the tryptic digest of the non-derivatised globin, covering the whole sequence of ␣- and -chains: A rapid diagnosis for haemoglobinopathy. J. Mass Spectrom. 30, 1663– 1670. Nakanishi, T., Miyazaki, A., Kishikawa, M., Shimizu, A., and Yonezawa, T. (1996) Electrospray ionization– tandem mass spectrometry analysis of peptides derived by enzymatic digestion of oxidised globin subunits: An improved method to determine amino acid substitution in the hemoglobin “core.” J. Am. Soc. Mass Spectrom. 7, 1040 –1049. Roberts, N. B., Green, B. N., and Morris, M. (1997) Potential of electrospray mass spectrometry for quantifying glycohemoglobin. Clin. Chem. 43, 771–778. Green, B. N., Morris, M., Wild, B., Layton, M., and Bellingham, A. J. (1997) Developments toward the routine identification of variant hemoglobins by ES–
Wild et al.
24.
25.
26.
27.
28.
29.
30.
31.
704
MS. Presented at the 45th Annual Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, June 1–5. Bisse, E., Zorn, N., Eigel, A., Lizama, M., HuamanGuillen, P., Marz, W., Van Dorsselaer, A., and Wieland, H. (1998) Hemoglobin Rambam (69[E13]Gly3 Asp), a pitfall in the assessment of diabetic control: Characterisation by electrospray mass spectrometry and HPLC. Clin. Chem. 44(10), 2172–2177. Gatlin, C. L., Eng, J. K., Cross, S. T., Detter, J. C., and Yates, J. R., III (2000) Automated identification of amino acid sequence variations in proteins by HPLC/ microspray tandem mass spectrometry. Anal. Chem. 72, 757–763. Houston, C. T., and Reilly, J. P. (1999) Toward a simple, expedient and complete analysis of human haemoglobin by MALDI–TOFMS. Anal. Chem. 71, 3397–3404. Lemann, H., and Huntsman, R. G. (1974) Man’s Haemoglobins, 2nd ed., pp. 388 –391, North Holland, Oxford. Wild, B. J., and Stephens, A. D. (1997) The use of automated HPLC to detect and quantitate haemoglobins. Clin. Lab. Haematol. 19, 171–176. International Committee for Standardisation in Haematology (1978) Recommendations for selected methods for quantitative estimation of Hb A2 and for Hb A2 reference preparation. Br. J. Haematol. 38, 573–578. Wild, B. J., Green, B. N., Lalloz, M. R. A., and Layton, D. M. (1999) Molecular characterisation of hybrid haemoglobins by mass spectrometry. Br. J. Haematol. 105(Suppl. 1), 81, Abstract 212. De Caterina, M., Esposito, P., Grimaldi, E., Di Maro, G., Scopacasa, F., Ferranti, P., Parlapiano, A., Malorni, A., Pucci, P., and Marini, G. (1992) Characterization of hemoglobin Lepore variants by advanced mass-spectrometric procedures. Clin. Chem. 38(8), 1444 –1448.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Rapid Identification of Hemoglobin Variants by Electrospray Ionization Mass Spectrometry Submitted 03/13/01 (Communicated by H. Ranney, M.D., 05/10/01)
B. J. Wild,1 B. N. Green,2 E. K. Cooper,1 M. R. A. Lalloz,1 S. Erten,1 A. D. Stephens,1 and D. M. Layton3 ABSTRACT: The precise identification of human hemoglobin variants, over 700 human hemoglobin variants are known, is essential for prediction of their clinical and genetic significance. A systematic approach to their rapid identification is described. Traditionally this requires protein or DNA characterization which entails lengthy analytical procedures. To overcome these obstacles a rapid approach to variant hemoglobin identification has been developed using conventional phenotypic methods combined with electrospray ionization–mass spectrometry (ESI–MS). The latter requires only a small amount of whole blood (10 l) but in most cases 2 l would have been sufficient and no preanalytical steps, such as separation of red cells or globin chains, are necessary. Aged, hemolyzed blood samples can also be analyzed. This approach has been used to positively identify 95% of the variants in over 250 samples. The remaining 5% in which a variant was detected by phenotypic techniques were not resolved by mass spectrometry. Ninety-nine different abnormalities comprising 36 ␣-chain variants, 59 -chain variants (including 2 extensions), and 4 hybrid hemoglobins were identified. These include 15 novel variants. The application of ESI–MS described requires approximately 1 h to prepare and analyze each sample and has minimal reagent costs. The turnaround time on a single sample can be as little as 2 h. This technique can now be considered a useful additional tool for reference laboratories. © 2001 Academic Press Key Words: human; hemoglobin; variant; mass spectrometry; peptide sequence.
INTRODUCTION
national guidelines (3–7) it is likely that more abnormal Hbs will be detected. The decision as to whether or not an abnormality is relevant to the patient’s clinical state or future inheritance depends on precise identification and in many situations must be reached quickly. Electrophoresis of Hb at alkaline pH has been the mainstay of first-line screening in most laboratories around the world for the last forty years (initially using paper, then starch gel and finally cellulose acetate as a fine grain, inert support medium). Such electrophoresis had the advantage that, by the time it had been refined by Schneider (8) in 1973, it could detect all the common clinically important Hb variants such as Hb S, C,
Hemoglobin (Hb) variants are most frequently detected during preanesthetic, antenatal, or neonatal screening. In addition, they may be found during the course of clinical investigation of hemolytic anemia, erythrocytosis, methemoglobinemia, or sickle cell and thalassemia syndromes. Occasionally, they are detected because they interfere with the determination of glycated Hb during diabetic monitoring. Currently over 700 Hb variants are characterized (1). With the increasing application of screening programs that has followed International Committee for Standardisation in Haematology recommendations (2), and
Correspondence and reprint requests to: B. J. Wild, Department of Haematological Medicine, King’s College Hospital, Denmark Hill, London SE5 9RS, UK. Fax: ⫹44 (0) 207 346 3514. E-mail: [email protected]. 1 Department of Haematological Medicine, Guy’s, King’s, and St. Thomas’ School of Medicine, Denmark Hill Campus, London SE5 9RS, United Kingdom. 2 Micromass UK Ltd., Tudor Road, Altrincham, Cheshire WA14 5RZ, United Kingdom. 3 Present address: Department of Haematology, Imperial College School of Medicine, Hamersmith Hospital, London W12 ONN, UK. 1079-9796/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved
691
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
D-Punjab, E, O-Arab, and Lepore. Furthermore, it separates these from Hb A and Hb F in less than 30 min. However, as more abnormal Hbs were discovered further tests such as the sickle solubility test and electrophoresis at acid pH in citrate agar were needed to obtain presumptive identification of these abnormal Hbs with sufficient accuracy for clinical purposes. Automated HPLC is now often employed as an alternative to electrophoresis, but in isolation, offers few advantages over electrophoresis at alkaline pH. However, HPLC has the advantage that it only requires a small blood sample (typically 5 l), is automated, and samples can be analyzed overnight and at weekends. Historically definitive characterization required the purification of the globin chains by preparative column chromatography, two-dimensional high-voltage electrophoresis/chromatography or high-performance liquid chromatography (HPLC) of a tryptic digest followed by amino acid analysis of any abnormal spot or peak detected. Even then, digestion with other enzymes, such as chymotrypsin, with or without amino acid sequencing was often needed. The entire procedure could take weeks or months to complete. A comprehensive approach to detection of variant hemoglobins by Riou (9) showed that by using a combination of phenotypic methods and careful calibration against known variants a presumptive identification was possible in approximately 90% of 125 different known variants. However, identification was only presumptive and this approach does not identify novel variants or precisely identify any variant. DNA techniques are also used to identify hemoglobin gene mutations. DNA sequencing is definitive, however, neither the expertise, nor the resources (for either manual or automated DNA sequencing) are widely available. Gene duplication and close homology between globin genes can also prolong localization and identification of mutations by a DNA approach alone. Sensitivity is excellent, but the reagent costs for DNA analysis are considerable. Although mass spectrometry (MS) was invented in the 1920s, the first analysis of the tryptic peptide mixtures of globin chains from human Hb was reported by Matsuo (10) in 1981 using field desorption mass spectrometry. Since then, MS has
Wild et al.
been increasingly applied to the identification of abnormal Hbs. The introduction of fast atom bombardment and liquid secondary ionization MS in 1981 (11) significantly improved the analysis of these mixtures and Rahbar and colleagues developed strategies using these techniques for variant identification (12, 13). However, it was not until the introduction of electrospray ionization mass spectrometry (ESI–MS) in 1988 (14) that it became practicable to determine the molecular weights of intact globin chains with sufficient accuracy to allow MS to be used as part of the procedure to identify abnormal Hbs (15–17). Over the past 10 years, ESI–MS has been developed by several research groups to the stage where it has become a significant adjunct to conventional phenotypic methods in the identification of variant Hbs. A comprehensive review by Shackleton and Witkowska (18) details the progress made by 1994 and gives a full account of the techniques used at that time. This group (19) also suggested that the relatively complicated and labor intensive sample preparation procedures employed prior to MS analysis, might explain why diagnostic laboratories have been reluctant to adopt MS for variant identification. These procedures involved at least some of the following: washing and lysing the red cells, removal of cell debris, removing the heme from the Hb, derivatization or oxidation of the cysteines, overnight digestion of the Hb with trypsin or other enzymes followed by HPLC separation of the tryptic peptides prior to ESI–MS. Considerable simplification of these procedures was achieved by Nakanishi (20) who digested Hb with trypsin without isolating the variant Hb and did not derivatize the cysteines prior to digestion. Furthermore, this group analyzed the mixture of tryptic peptides directly by ESI–MS without prior separation by HPLC. However, they still purified and removed the heme from the Hb prior to digestion. Peptides were observed that covered the whole of the ␣- and -chains, although some of these could not be usefully sequenced by tandem mass spectrometry, particularly the cysteine containing peptides ␣T12 and T10. They observed disulfide bonded dimers of ␣T(12–13) and T12 which partly accounted for 692
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
the poor observance of these peptides. Subsequently, homo- and heterodimers of the cysteine containing T10 and T12 peptides were observed and prevented from forming by oxidizing the cysteine prior to digestion with trypsin (21). Despite this and other reports (22, 23) complex work-up procedures and HPLC separation of the tryptic peptides are still undertaken prior to ESI–MS analysis (24, 25). A different approach using matrix-assisted laser desorption ionization (MALDI) MS and special trypsin-linked probes has recently reduced the digestion time to 15 min, although preparation of the probes is time consuming (26). Peptides were observed covering 96% of both ␣- and -chains. However, sequencing of the peptides by tandem mass spectrometry was not undertaken. This is necessary in order to positively identify variants, since the tryptic peptides from about 50% of samples contain two or more possible mutation sites that can give rise to the observed mass change of the peptide. We describe the development and application of mass spectrometry techniques which can result in unequivocal identification of the ␣ and  chain variants and hybrid hemoglobins such as Hb Lepore in samples found to be abnormal by conventional phenotypic methods. We have also demonstrated the ability of ESI–MS to detect hemoglobin variants such as some unstable variants and M hemoglobins that are “silent” using conventional electrophoretic techniques. Emphasis has been placed on minimizing sample preparation and analysis times by using diluted whole blood without prior purification, derivatization or separation into its constituent chains for the mass spectrometric and enzymatic digest procedures and by analyzing the products of the digest by ESI–MS without prior chromatographic separation.
presence of a variant and often indicates whether the mutation is in the ␣- or -chain. It may also give a presumptive identification. Mass spectrometric analysis is first carried out on diluted whole blood by Step 2 in order to observe the globin chains. As its name implies, mass spectrometry essentially determines the mass (molecular weight) of a peptide or protein and all deductions made from the technique depend on measuring mass or mass difference with high accuracy. In contrast to most phenotypic methods, which analyze Hbs as noncovalently assembled tetramers or dimers, ESI–MS in this application analyses Hb in denaturing solvents, so that the globin chains and heme group are analyzed as separate entities. In essence, the mass spectrometer separates these components according to their masses. Thus, the various chains, their glycated and posttranslationally modified forms are separated by the spectrometer, providing their masses differ sufficiently from one another. In practice, two globin chains that differ in mass by 6 Daltons (Da) or more are resolved by ESI–MS thus enabling the masses of the intact chains to be individually determined with an accuracy of better than ⫾ 0.3 Da. This minimum of 6 Da separation is a characteristic of ESI–MS when measuring intact globin chains. The sequence masses of the principal chains are 15,126.4 Da (␣) and 15,867.2 Da () which differ in mass by approximately 741 Da. Since the highest mass change for a single amino acid mutation is 129 Da (Gly7Trp), the majority of mutations can be unambiguously associated with either the ␣- or -chain from their measured masses and the proportions of variant and normal chains determined. Moreover, again provided the mass difference between the two chains in a heterozygote is at least 6 Da, their masses can be measured without prior separation of the normal and abnormal hemoglobins. From the mass difference between the normal and variant chains, a list of possible variants can then be deduced, based on the single amino acid changes that are created by single nucleotide changes in the coding triplet (Table 1). Of 141 possible mutations shown in Table 1, 127 (90%) give rise to ⬎6 Da mass change. Mutations from isoleucine are ex-
METHODS Optimization of Strategy Figure 1 outlines the essential analytical steps developed to identify a variant hemoglobin. Generally, phenotypic analysis (Step 1) reveals the 693
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
cluded since the normal ␣- and -chains contain no isoleucine. The results from Step 2 divide the samples into two categories, namely those that show both chains in heterozygotes distinctly separated from one another and those that show no detectable separation. The latter samples are then classified as likely to contain two chains that differ in mass by less than 6 Da and need digestion with trypsin (Step 4) before the variant can be detected by MS. Next, tandem mass spectrometry may be carried out on the intact variant chain (Step 3), either from the whole blood sample or from the isolated variant. The masses of characteristic peaks in this spectrum are used to determine the group of amino acids that contains the mutation, e.g., (1– 47), (124 –146), ␣(1–76), ␣(114 –141) and hence limit the number of possible variants that could be present. This procedure is normally undertaken when the variant tryptic peptide is not found in Step 4 or when the variant is suspected to be a hybrid as indicated by the results from Steps 1 and 2. In Step 4, the mixture of peptides produced by digestion of the underivatized Hb with trypsin is analyzed without prior separation by ESI–MS in order to identify the peptide that contains the variant. The mass difference between the normal and variant globin chains determined in Step 2 is used to assist this procedure unless the mass difference is ⬍6 Da when tables listing all possible variants and the masses of their corresponding peptides are used. Hitherto, the cysteines in Hb have often been derivatized by treatment with iodoacetamide (carbamidomethylation) prior to enzymatic digestion, otherwise the cysteine containing tryptic peptides were not detected or were only present in very small amounts. Although derivatization takes only 10 min, removing the excess reagents prior to digestion is labor intensive and time consuming. The recovery of the cysteine containing peptides from the underivatized peptide mixture is significantly improved by simply reducing the latter with dithiothreitol (DTT) before analysis by ESI–MS. At this stage most variants can be detected, including those in heterozygotes in which the two chains differ in mass by as little as 1 Da from one another. Ap-
Wild et al.
proximately 50% of variants can be identified at this stage. These include variants that produce a tryptic peptide containing only one possible mutation that can give rise to the mass change determined in Steps 2 or 4. Others involve mutations to arginine and lysine that produce two “new” peptides with masses characteristic of the mutation. One mutation (Gln3 Lys), in which the mass difference between the normal and variant chain is zero, can be detected and identified in this way. Finally, in Step 5, the variant tryptic peptide is sequenced by tandem MS in order to identify or confirm the mutation unambiguously. Phenotypic Analysis Cellulose acetate electrophoresis (CAM) at alkaline pH and acid agarose electrophoresis were undertaken using Helena equipment and reagents (Helena UK, Newcastle, UK). Sickle solubility tests were undertaken using standard techniques (27). Isoelectric focusing (IEF) was undertaken using the Resolve reagents and Pharmacia LKB Multiphor equipment (Perkin–Elmer Wallac Inc.). Highperformance liquid chromatography (HPLC) was undertaken using the Bio-Rad Variant equipment and “beta thal short” reagents (Bio-Rad, UK; Hemel Hempstead, UK) (28). Isolation of hemoglobin variants such as Hb Lepore-Boston-Washington was undertaken by a modification of the ICSH electrophoretic technique for Hb A2 quantitation (29). Sample Preparation for Mass Spectrometry Step 1. Initially, a stock solution was prepared by diluting 10 l of blood 50-fold with 490 l of HPLC grade water, although later 2 l of blood was diluted with 98 l of HPLC grade water and proportionately less was used in Steps 2–5 below. This stock solution is sufficient for the entire analysis. Working solutions for direct introduction into the mass spectrometer were prepared as follows. Step 2. Twenty microliters of the stock solution was diluted 10-fold by adding 180 l of 1:1 acetonitrile:water containing 0.2% formic acid. Just prior to analysis, the resulting solution was 694
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
(Sigma C-3142) as follows. 2 l of 0.5 mg/ml ␣-chymotrypsin was added to 20 l of the tryptic digest solution, mixed and incubated at 25°C for 30 min. The resulting solution was then diluted with 180 l of 1:1 acetonitrile:water containing 0.2% formic acid to make the working solution for analysis by ESI–MS.
de-salted by manually shaking it for 15–30 s with approximately 5 mg of cation exchange resin beads (Bio-Rad Labs, AG 50W-X8, Cat. No. 1435441) that had been twice washed previously with water and this solution was used for ESI–MS analysis. Step 3. Twenty microliters of the stock solution was diluted 10-fold by adding 180 l of 1:1 acetonitrile:water containing 0.2% formic acid and this solution was used for tandem MS without desalting.
Mass Spectrometry Twenty-microliter aliquots of the working solutions were introduced at 5 l/min into the electrospray source of a Quattro II or Quattro LC tandem mass spectrometer (Micromass UK Ltd., Wythenshawe, Manchester, UK) while scanning and acquiring data over appropriate mass-tocharge ratio (m/z) ranges. These were: m/z 930– 1180 (Step 2), 50 –1500 (Step 3), 150 –2000 (Step 4) and 50 –2000 (Step 5). Acquisition times were 2 min for Step 2, 3 min for Steps 3 and 4, and 5–10 min for Step 5. Mass scale calibration of the spectra obtained in Step 2 employed the multiply charged normal ␣-chain peaks present in each sample. Mass scale calibration of the spectra obtained in the other steps employed the Nan⫹1In peaks from separate introductions of a 0.5 mg/ml solution of NaI in 1:1 2-propanol:water. The baseline subtracted m/z spectra (m/z 970–1180) from Step 2 were deconvoluted to present the spectrum on a true molecular weight scale using the maximum entropy (MaxEnt) based software supplied with the instrument.
Steps 4 and 5. One hundred microliters of the stock solution was first denatured by adding 10 l of 1% aqueous formic acid and 10 l of acetonitrile. After mixing, the solution was allowed to stand at room temperature (25°C) for 5 min. Then 6 l of 1 M aqueous ammonium bicarbonate solution and 5 l of a 5 mg/ml solution of TPCK treated trypsin (Sigma T-8642) was added. The resulting solution, after mixing and centrifuging for approximately 15 s, was incubated at 37°C. A fine precipitate will appear after adding the ammonium bicarbonate, indicating that the hemoglobin has been effectively denatured. This precipitate generally disappears within a minute of adding the trypsin. The solution should not be centrifuged until it has become clear. After 30 min, 30 l of this digest was diluted with 270 l of 1:1 acetonitrile:water containing 0.2% formic acid to give the working solution for ESI–MS analysis. Most variants can be identified from the 30 min (short) digest. The remainder of the digest solution was allowed to incubate for 12–15 h in case the variant was not identified from the short digest. If required, 30 l of the 12- to 15-h digest solution was reduced at room temperature for 10 min by adding 3 l of aqueous 100 mM dithiothreitol (DTT) in order to release the cysteine containing tryptic peptides. The resulting solution was then diluted with 270 l of 1:1 acetonitrile: water containing 0.2% formic acid to give the working solution of the reduced digest for analysis by ESI–MS. To reduce the size of the ␣T9 peptide and hence facilitate identification of variants suspected of being present therein, the tryptic digest was mildly digested with ␣-chymotrypsin
DNA Analysis DNA was prepared from blood by a conventional method and sequence analysis of globin genes was undertaken by cycle sequencing (fmol, Promega). Where possible, novel mutations were confirmed by restriction analysis of DNA amplified by the polymerase chain reaction (PCR) from genomic DNA. Endonucleases (New England Biolabs, MBI Fermentas, Gibco, Promega) were used according to the manufacturer’s instructions. RESULTS Using the strategy depicted in Fig. 1, mutations were positively identified in 95% of over 695
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
FIG. 1. Steps taken to identify a variant Hb by combined phenotypic analysis and mass spectrometry.
250 blood samples found to be abnormal by either phenotypic methods and/or MS. Ninety nine different variants, including 15 novel hemoglobin variants, were identified (Table 2). These comprise 36 ␣-chain variants, 59 -chain variants (including 2 extensions) and 4 hybrid hemoglobins. Figures 2– 4 show typical MS spectra that illustrate how the data from analytical Steps 2–5 logically lead to the positive identification of a -chain variant. In this example, the variant is due to the previously unreported and electrophoretically silent mutation 13 Ala3 Val. MS was undertaken on this sample because it was suspected that the Hb might be abnormal because of a history of neonatal jaundice and hemolysis (the sample was obtained when the individual was 11 months old with an Hb F of 3.1%). Figure 2 shows MaxEnt deconvoluted mass spectra from the diluted blood of (A) a normal adult and (B) the patient with the abnormal Hb (Step 2). The latter spectrum clearly shows two -chain peaks of similar intensity, one at 15,867.3 Da corresponding to the normal -chain (sequence mass 15,867.2 Da) and one at 15,895.4 Da, corresponding to the
variant, thus indicating the presence of a normal/ variant -chain heterozygote. Furthermore, the mass of the variant chain is 28.1 Da higher than the normal -chain. Assuming the mutation is due to a single base change in the coding triplet, there are only three single amino acid changes that can give rise to 28 Da mass increase between normal and variant, namely Ala3 Val, Gln3 Arg and Lys3 Arg (Table 1). It follows that the total number of possible single amino acid mutations that can give a 28 Da increase in mass is 29 at this stage in the analysis, since there are 15 alanine, 3 glutamine and 11 lysine residues in the normal -chain. However, the two mutations involving arginine are unlikely, since these mutations create a charge change resulting in a difference in mobility of the variant hemoglobin which would have been detected by conventional phenotypic means. The results from Step 4 are given in Fig. 3 which shows part of the mass spectra of tryptic digests from (A) a normal control and (B) the sample containing the variant. The two spectra are remarkably similar except for the occurrence of an extra peak at m/z 480.7 in the spectrum from 696
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
FIG. 2. MaxEnt deconvoluted ESI mass spectra of 500-fold diluted and desalted blood obtained under denaturing conditions: (A) from a normal adult sample and (B) from a patient suspected to have a phenotypically silent abnormal Hb. The occurrence of two peaks separated by 28.1 Da clearly indicates the presence of a -chain heterozygote.
from (A) the normal peptide ion and (B) the variant peptide ion. The increase in mass of 28 Da between normal and variant at a particular position (y5⬙ ion) positively identifies the variant as 13 Ala3 Val. This amino acid substitution fully accounts for the mass difference between the normal and variant chains determined in Step 2 (Fig. 2), i.e., the presence of a second mutation in the same chain is unlikely. Thirteen of the 15 novel variants were also examined by DNA techniques and in all cases the results of the DNA analysis agreed with those obtained by mass spectrometry. No material was available for DNA analysis of one novel variant and the other is still under analysis.
the abnormal sample and a reduction in the relative intensity of the peak at m/z 466.7. The m/z difference between these two peaks is 14.0, which corresponds to a mass difference of 28.0 Da because the ions are doubly charged. These data imply that the variant occurs in tryptic peptide T2 since the mass difference between the normal and variant -chains was found in Step 2 to be 28 Da. There are three sites in this peptide that can give rise to 28 Da increase in mass by single amino acid mutations, namely at amino acid numbers 2, 5 and 9 which correspond to 10 Ala3 Val (Hb Iraq-Halabja), 13 Ala3 Val and 17 Lys3 Arg in the -chain respectively. The latter two mutations have not been described previously. Finally, tandem mass spectrometry was undertaken on the normal and variant T22⫹ peptide ions in order to identify the site of the mutation (Step 5). The resulting spectra are shown in Fig. 4
DISCUSSION The results reported here demonstrate that simplified mass spectrometric procedures together 697
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
5) was undertaken at least once on every variant in Table 2 to confirm its identity, even though it may have already been identified in Step 4. The technique used only a small amount of blood (10 l). Although 10 l was used for convenience, in 90% of samples analyzed 2 l would have been sufficient. To minimize sample preparation time, whole blood was used in almost all cases. Stored, aged hemolyzed samples were also successfully analyzed by this technique. In no case was it necessary to wash the red cells prior to analysis by ESI–MS. Although the majority of samples were analyzed from diluted whole blood, it was found that isolation of the variant Hb was necessary in 12% of cases (9% of different variants) in order to detect and identify the variant peptide in the digest mixture. These cases mainly occurred with ␣-chain variants in heterozygotes when the variant was present in only relatively small amounts e.g., Hb Fort Worth (␣27 Glu3 Gly) and the novel variant ␣76 Met3 Arg (10.8% and 9-10% of total Hb respectively). Another case occurred with Hb Q-Thailand (␣74 Asp3 His), which was difficult to distinguish from Hb Q-Iran (␣75 Asp3 His) by tandem MS until isolated. The hybrid Hb Lepore-Boston-Washington (12% of total Hb) was also isolated because its mass is only 2 Da lower than the normal -chain and hence is not resolved from the normal -chain in Step 2. The only -chain variant that needed isolation was Korle-Bu (73 Asp3 Asn) because the variant tryptic peptide appeared to be at much lower abundance in the digest mixture from nonisolated Hb than expected and it was later shown that this was due to a somatic mutation. In such cases, the analysis was undertaken on the variant Hb after preparative isolation using cellulose acetate electrophoresis followed by elution (29). Identification of ␦-chain variants would require isolation of the variant chain before analysis, since they are present at only 1–2% of the total hemoglobin. Samples with ␥-chain variants, except perhaps from newborn children, would also require isolation and concentration before analysis. Sample preparation time was minimized by undertaking digestion with trypsin on diluted whole blood without prior derivatization of the cysteines. Al-
TABLE 1 Nominal Mass and Amino Acid Residue Changes Produced by Single Base Changes in the Nucleotide Coding Triplet Mass change
Amino acid change
Mass change
Amino acid change
0
Gln7Lys Ile7Leu Asn7Asp Gln7Glu Ile7Asn Lys7Glu Lys7Met Pro7Thr Gln7His Ser7Pro Thr7Ile Thr7Asn Asn7Lys Asp7Glu Gly7Ala Ser7Thr Val7Ile Val7Leu Ile7Lys Leu7Gln Ala7Ser Phe7Tyr Pro7Leu Ser7Cys Val7Asp Ile7Met Leu7Met His7Arg Asp7His Asn7His Leu7His Met7Arg Ala7Pro His7Tyr Ser7Ile Ser7Leu Ser7Asn Thr7Lys
28
Ala7Val Gln7Arg Lys7Arg Ala7Thr Arg7Trp Gly7Ser Thr7Met Val7Glu Pro7Gln Val7Met Ile7Phe Leu7Phe Pro7His Gly7Val Ile7Arg Leu7Arg Ala7Asp Cys7Phe Gly7Cys Asp7Tyr Val7Phe Asn7Tyr Cys7Arg Thr7Arg Ala7Glu Gly7Asp Pro7Arg Cys7Tyr Ser7Phe Ser7Arg Gly7Glu Leu7Trp Ser7Tyr Cys7Trp Gly7Arg Ser7Trp Gly7Trp
1
3 4 9 10 12 13 14
15 16
18 19 22 23 24 25 26
27
30
31 32 34 40 42 43 44 46 48 49 53 55 58 59 60 69 72 73 76 83 99 129
Wild et al.
with digestion with trypsin can identify the precise structural abnormality in approximately 95% of the ␣- and -chain variant hemoglobins which had been detected by conventional methods in routine service laboratories. We also detected, and identified, some variants that had not been detected by conventional screening techniques. These were detected in Step 2 while either investigating severe hemolytic anemia (e.g., Hb Southampton) or unexplained erythrocytosis (e.g., Hb Johnstown). Tandem mass spectrometry (Step 698
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
FIG. 3. Part of the mass spectra of tryptic digests from (A) a normal control and (B) the sample containing the variant. The appearance of the peak at m/z 480.7 in the spectrum from the abnormal Hb digest, 28.0 Da higher than the normal T2 peak shows that the variant lies in this peptide, i.e., between 9 and 17 inclusive. The difference on the m/z scale is 14.0 (28.0/2) because the ions are doubly charged.
though for 75% of this work digestion with trypsin was undertaken overnight (12–15 h), the recent introduction of the 5-min denaturing step described under Methods has dramatically reduced the time to produce viable tryptic peptide mixtures. This development significantly improves the reliability of the procedure and undigested ␣- and -chains disappear after incubation for 30 min. Even after 15 min incubation, many variants can be detected and the clinically important variants C, E. D-Punjab, and O-Arab can be positively identified after this very short digest time. Thus same day turn round of urgent samples is now practicable. However, the 12- to 15-h incubation time is still required in a few cases, when reduction with DTT is necessary in order to release the cysteine containing peptides. The cysteine containing peptides are present at high levels in the short digests without reduction. The time to
manually prepare the samples and to acquire and process the mass spectral data to a stage suitable for interpretation was less than 1 h per sample. Interpretation of the data to positively identify the amino acid substitution (or other structural change) although generally rapid, can be time consuming (up to 2 h) and requires considerable experience. However, an experienced analyst can usually identify 10 variants in two to three days. Analysis times are based on manual sample introduction and data processing on a batch basis. A saving of 10 –20 min/sample could be obtained by automating acquisition and print out of the mass spectra in analytical Steps 2, 3, and 4 given additional hardware and software development. Software to list the possible mutations and the corresponding tryptic peptides containing the mutation after Steps 1–3 would also reduce interpretation time. 699
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
FIG. 4. Tandem mass spectra of (A) the normal T22⫹ tryptic peptide ion and (B) the variant T22⫹ tryptic peptide ion. The mass increase of 28.1 Da in the spectrum of the variant at fragment ion y5 positively identifies the variant as 13 (Ala3 Val).
Tryptic peptides covering effectively the whole of the ␣- and -chains were observed in the spectrum of the reduced digest with remarkably little interference, and many mutations that produce a mass change of only 1 Da were identified in heterozygotes from such spectra. Table 2 shows 4 ␣- and 12 -chain variants in this category. All the small peptides (␣T2, ␣T7, ␣T10, ␣T14, T6, T7, and T15) except the two mono-peptides (␣T8 and T8) were observed at good signal-to-background ratio. The latter two were observed in the combinations ␣T(8 –9) and T(8 –9). As observed previously (18), lysine bonds at ␣127 and 95 are incompletely cleaved, leaving significant quantities of ␣T(12–13) and T(10 –11) in the digest mixture. Nevertheless, the ␣T12, ␣T13, T10 and T11 peptides were present at sufficiently high levels to allow viable mass spectra and tandem mass spectra to be acquired from them. An advantage of undertaking
tandem MS on ions selected by the mass spectrometer from the peptide mixture is that viable spectra from low abundance ions can, if necessary, be obtained by acquiring the data over an extended period of time (10 –15 min). This time compares with less than a minute available during elution of a peptide during an HPLC–MS run, and is still considerably shorter than the time to collect fractions from a preparative HPLC run. Although there was no problem observing the large ␣T9 (62–90) and ␣T12 (100 –127) peptides, it was difficult to detect mutations in heterozygotes in these peptides when the mass of the variant differed by only 1 Da from normal. This is mainly because the dominant ions from these peptides carry 3 or 4 charges leading to unresolved isotope peaks. Also, the proportion of the variant is generally 25% or less in many ␣-chain heterozygotes. For the ␣T9 peptide, these difficulties were partly overcome by using the non700
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
tryptic peptides, ␣(62–72) and ␣(81–90) which often occurred at satisfactory levels in the 12- to 15-h tryptic digest. Alternatively, a simple procedure was developed using mild digestion of the tryptic peptide mixture with ␣-chymotrypsin (Step 4). This procedure cleaves the ␣T9 tryptic peptide into two peptides, ␣(62– 80) and ␣(81– 90). These non-tryptic peptides were used to identify several of the variants occurring between ␣62 and ␣90 (Table 2). Reducing the size of the ␣T12 (100 –127) peptide remains one of the problem areas in this simplified approach. It is probable that some of the 13 samples for which no abnormality was detected occur in this region of the ␣-chain since only two variants were identified between ␣100 and ␣127, whereas 13 variants were identified between ␣62 and ␣90 (Table 2). Eight of the 13 samples in which we failed to identify an abnormality occurred in the first 60 abnormal Hbs we analyzed before we started isolating the variant Hb and many of these may therefore have been identified had they been isolated. Unfortunately there is insufficient blood specimen to reanalyze these samples. Except for Lepore-Boston-Washington, positive identification of the 10 known hybrid Hbs in heterozygotes from their masses (Step 2) and by tandem MS (Step 3) without preparative isolation of the variant Hb is feasible. Their masses differ from normal by more than 6 Da and therefore can be separated from the normal -chain by the mass spectrometer. In addition, we predict that they would give distinctive tandem mass spectra. The four hybrids we have so far encountered (Table 2) gave quite characteristic tandem mass spectra (30). However, a compound heterozygote of Lepore-Hollandia with sickle would not be identifiable in this way without separation since the masses of the two variant chains differ from one another by only 1 Da (18, 31). For several of the common clinically important variants (Hb C, E, D-Punjab, and O-Arab) the mutation produces only 1 Da change in the mass of the -chain. Hence, in heterozygotes, the normal and variant chains are not resolved from one another by the mass spectrometer in Step 2 and the measured mass is the abundance-weighted mean of the masses of the two chains. Hence,
when the two chains occur in approximately equal abundance e.g., in Hb C trait (Hb A ⫹ Hb C), the mass change from normal is theoretically 0.5 Da and can just be detected reliably. However, when the variant chain is present at a lower concentration than the normal chain, e.g., in Hb E trait (Hb A ⫹ Hb E), the net mass change can be as low as 0.2 Da, and cannot be detected reliably in Step 2. Consequently, ESI–MS of diluted whole blood without prior separation cannot be used to screen for these variants, and conventional phenotypic methods are still necessary in order to detect them in the first instance. Nevertheless, after digestion with trypsin, all these variants can be detected and identified unambiguously by MS in Step 4 whether present in the homozygous, heterozygous or compound heterozygous state. Therefore, the combination of phenotypic methods and modern MS facilitates both the detection and positive identification of these common variants. In contrast, most uncommon variants can be detected by analyzing diluted whole blood (Step 2), because the mutation produces a mass change greater than 6 Da. Of 141 possible mutations caused by a single base change in the coding triplet, 127 (90%) can be detected in this way. Thus, MS has the potential to detect as well as identify most uncommon variants including those that are phenotypically silent. Therefore, in cases of obvious unexplained hematological abnormality, MS should be considered as a diagnostic procedure to detect the presence of an abnormal hemoglobin. Of the remaining 14 mutations, i.e., those that give ⬍6 Da change by MS (Leu3 Ile, Asn3 Ile, Gln7Lys, Lys7Glu, Asn7Asp, Gln7Glu, Lys7Met, and Pro7Thr), 10 can be easily detected by conventional phenotypic methods. Hence, most variants can be detected by a combination of conventional phenotypic methods and MS. The mutation Leu3 Ile can be neither detected nor identified by MS because it produces neither a mass change nor a new cleavage site after digestion. The analytical power of the present generation of electrospray mass spectrometer instruments has enabled definitive identification of most hemoglobin variants within 24 h requiring only a very small sample (2 l) of whole blood, although 10 701
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
Wild et al.
TABLE 2 Variants Identified by Combined Phenotypic Analysis and ESI–MS Mutation
Name
Mass change
Mutation
␣-Chain variants ␣5 Ala3Asp ␣12 Ala3Asp ␣16 Lys3Glu ␣18 Gly3Arg ␣20 His3Gln ␣21 Ala3Pro ␣27 Glu3Ala ␣27 Glu3Gly ␣27 Glu3Val ␣44 Pro3Ala ␣47 Asp3His ␣49 Ser3Arg ␣50 His3Leu ␣54 Gln3Arg ␣57 Gly3Asp ␣63 Ala3Thr ␣64 Asp3His ␣68 Asn3Lys ␣74 Asp3Asn ␣74 Asp3His ␣75 Asp3Tyr ␣76 Met3Arg ␣78 Asn3Lys ␣81 Ser3Cys ␣84 Ser3Arg ␣85 Asp3Asn ␣85 Asp3Tyr ␣87 His3Tyr ␣89 His3Gln ␣90 Lys3Asn ␣94 Asp3Asn ␣94 Asp3Tyr ␣102 Ser3Arg ␣112 His3Asp ␣120 Ala3Glu ␣141 Arg30
J-Toronto J-Paris-I I Handsworth Le Lamentin Fontainebleau Novel Fort Worth Spanish Town Novel Hasharon Savaria Novel Shimonoseki J-Norfolk Novel Q-India G-Philadelphia G-Pest Q-Thailand Winnipeg Novel Stanleyville-II Nigeria Etobicoke G-Norfolk Atago M-Iwate Novel J-Broussais Titusville Setif Manitoba Hopkins-II J-Meerut Koellikera
a
Marseille South Florida Raleigh Novel Deer Lodge Tyne C Sickle G-Siriraj Novel Belfast J-Baltimore D-Ouled Rabah
Posttranslational modification.
Mass change
-Chain Variants 44 44 1 99 ⫺9 26 ⫺58 ⫺72 ⫺30 ⫺26 22 69 ⫺24 28 58 30 22 14 ⫺1 22 48 25 14 16 69 ⫺1 48 26 ⫺9 ⫺14 ⫺1 48 69 ⫺22 58 ⫺156
-Chain Variants (⫺1)Met-(⫹1)Val(⫹2)Pro-Leu(⫺1)Met-(⫹1)Met(⫹2)His-Leu1 Val3Ac-Ala 1 Val3Gly 2 His3Arg 5 Pro3Ser 6 Glu3Lys 6 Glu3Val 7 Glu3Lys 13 Ala3Val 15 Trp3Arg 16 Gly3Asp 19 Asn3Lys
Name
91 163 14 ⫺42 19 ⫺10 ⫺1 ⫺30 ⫺1 28 ⫺30 58 14
20 Val3Met 21 Asp3Val 22 Glu3Ala 22 Glu3Gln 22 Glu3Lys 26 Glu3Lys 27 Ala3Val 30 Arg3Ser 40 Arg3Lys 46 Gly3Glu 51 Pro3His 52 Asp3Asn 56 Gly3Asp 57 Asn3Lys 58 Pro3Arg 69 Gly3Ser 71 Phe3Ser 73 Asp3Asn 74 Gly3Val 82 Lys3Arg 82 Lys3Met 83 Gly3Asp ⫹ 6 Glu3Lys 85 Phe3Ser 90 Glu3Lys 95 Lys3Asn 95 Lys3Glu 98 Val3Met 99 Asp3His 101 Glu3Ala 104 Arg3Thr 106 Leu3Pro 108 Asn3Lys 109 Val3Leu 109 Val3Met 113 Val3Glu 116 His3Tyr 121 Glu3Gln 121 Glu3Lys 124 Pro3Gln 126 Val3Glu 128 Ala3Asp 131 Gln3Glu 133 Val3Ala 134 Val3Glu 143 His3Arg 143 His3Tyr
Olympia Novel G-Coushatta D-Iran E-Saskatoon E Grange-Blanche Tacoma Athens-GA K-Ibadan Novel Osu-Christiansborg J-Bangkok G-Ferrara Dhofar City of Hope Christchurch Korle-Bu Bushwick Novel Helsinki Novel, Pyrgos, & C in same chain Buenos Aires Agenogi Detroit N-Baltimore Ko¨ln Yakima Novel Sherwood Forest Southampton Presbyterian Johnstown San Diego New York Novel D-Los Angeles, D-Punjab O-Arab Ty Gard Hofu J-Guantanamo Camden Novel North Shore Abruzzo Old Dominion
␦22—50 ␦87—116 A ␥81—86 22—␦50
Hybrids Lepore-Hollandia Lepore-Boston-Washington Kenya P-Nilotic
32 ⫺16 ⫺58 ⫺1 ⫺1 ⫺1 28 ⫺69 ⫺28 72 40 ⫺1 58 14 59 30 ⫺60 ⫺1 42 28 3 57 ⫺60 ⫺1 ⫺14 1 32 22 ⫺58 ⫺55 ⫺16 14 14 32 30 26 ⫺1 ⫺1 31 30 44 1 ⫺28 30 19 26 ⫺31 ⫺2 55 88
Wild et al.
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
l has often been used for convenience. The initial use of traditional phenotypic methods remains the first line screening approach for detecting some common clinically important variants such as Hb C, E, D-Punjab, and O-Arab which only give a small mass change (1 Da) from normal. Nevertheless, the definitive identification can still usually be obtained within 24 –36 h. Variants present in small amounts relative to the other hemoglobins, such as Hb Lepore or the unstable hemoglobins may require an initial step of purification on cellulose acetate, but this step only adds another 24 h to the procedure. However, as in other fields of laboratory medicine it is important to emphasize that interpretation of ESI–MS data requires considerable expertise. ESI–MS enables the definitive identification of a large proportion of variant hemoglobins detected by conventional separation techniques. Faced with an extensive and growing catalogue of human hemoglobins this facility greatly enhances the ability to predict the clinical and genetic implications of individual variants. ESI–MS constitutes an extremely versatile addition to the techniques at present available to reference laboratories and provides a rapid alternative to DNA analysis.
4.
5.
6.
7.
8.
9.
10.
11.
ACKNOWLEDGMENTS We thank the following for providing some of the abnormal Hb samples mentioned in this paper: Dr. P. Bolton-Maggs (Alder Hey Hospital, Liverpool), Professor T. R. J. Lappin (Royal Victoria Hospital, Belfast), Dr. D. C. Rees (Royal Hallamshire Hospital, Sheffield), Professor T. M. Reynolds (Queen’s Hospital, Burton-onTrent), Dr. N. B. Roberts (Royal Liverpool University Hospital), and Dr. K. Wiener (North Manchester General Hospital).
12.
13.
14.
REFERENCES 1.
International Hemoglobin Information Centre Variant List (1997) Hemoglobin 21(6), 505– 622. 2. International Committee for Standardisation in Haematology (1988) Recommendations for neonatal screening for haemoglobinopathies. Clin. Lab. Haematol. 10, 335–345. 3. British Committee for Standards in Haematology
15.
703
(1988) Guidelines for Haemoglobinopathy Screening. Clin. Lab. Haematol. 10, 87–94. British Committee for Standards in Haematology (1994) Guidelines for fetal diagnosis of globin gene disorders. J. Clin. Pathol. 47, 199 –204. British Committee for Standards in Haematology (1994) Guidelines for the investigation of the alpha and beta thalassaemia traits. J. Clin. Pathol. 47, 289 – 295. British Committee for Standards in Haematology (1998) Guideline, the laboratory diagnosis of haemoglobinopathies. Br. J. Haematol. 101, 783–792. The Standing Medical Advisory Committee on Sickle Cell, Thalassaemia and Other Haemoglobinopathies (1993) Report of a working party of The Standing Medical Advisory Committee on Sickle Cell, Thalassaemia and other Haemoglobinopathies. HMSO, London. Schneider, R. G. (1973) Sickle cell disease. In Developments in Laboratory Diagnosis (Abramson, H., Bertles, J. F., and Wethers, D. I., Eds.), pp. 230 –243. Mosby, St. Louis, MO. Riou, J., Godart, C., Hurtrel, D., Mathis, M., Bimet, C., Bardakdjian-Michau, J., Prehu, C., Wajcman, H., and Galacteros, F. (1997) Cation-exchange HPLC evaluated for presumptive identification of hemoglobin variants. Clin. Chem. 43, 34 –39. Matsuo, T., Matsuda, H., Katakuse, I., Wada, Y., Fujita, T., and Hayashi, A. (1981) Field desorption mass spectra of tryptic peptides of human hemoglobin chains. Biomed. Mass Spectrom. 8, 25–30. Barber, M., Bordoli, R. S., Sedgwick, R. D., and Tyler, A. N. (1981) Fast atom bombardment of solids (FAB): A new ion source for mass spectrometry. J. Chem. Soc. Chem. Commun., 325–327. Rahbar, S., Lee, T. D., Baker, J. A., Rabinowitz, L. T., Asmerom, Y., Legesse, K., and Ranney, H. M. (1986) Reverse phase high-performance liquid chromatography and secondary ion mass spectrometry. A strategy for identification of ten human hemoglobin variants. Hemoglobin 100, 379 – 400. Lee, T. D., and Rahbar, S. (1991) The mass spectral analysis of hemoglobin variants. In Mass Spectrometry of Peptides (Desiderio, D. M., Ed.), pp. 257–274. CRC Press, Boca Raton, FL. Covey, T. R., Bonner, R. F., Shushan, B. I., and Henion, J. (1988) Determination of protein, oligonucleotide and peptide molecular weights by ion-spray mass spectrometry. Rapid Commun. Mass Spectrom. 2, 249 –256. Green, B. N., Oliver, R. W. A., Falick, A. M., Shackleton, C. H. L., Roitman, E., and Witkowska, H. E. (1991) Electrospray MS, LSIMS and MS/MS for the rapid detection and characterization of variant hemoglobins. In Biological Mass Spectrometry (Burlin-
Blood Cells, Molecules, and Diseases (2001) 27(3) May/June: 691–704 doi:10.1006/bcmd.2001.0430, available online at http://www.idealibrary.com on
16.
17.
18.
19.
20.
21.
22.
23.
game, A. L., and McCloskey, J. A., Eds.), pp. 129 – 146. Elsevier, Amsterdam. Falick, A. M., Shackleton, C. H. L., Green, B. N., and Witkowska, H. E. (1990) Tandem mass spectrometry in the clinical analysis of variant hemoglobins. Rapid Commun. Mass Spectrom. 4, 396 – 400. Witkowska, H. E., et al. (1991) Sickle cell disease in a patient with sickle cell trait and compound heterozygosity for hemoglobin S and hemoglobin QuebecChori. N. Engl. J. Med. 325, 1150 –1154. Shackleton, C. H. L., and Witkowska, H. E. (1994) Mass spectrometry in the characterization of variant hemoglobins. In Mass Spectrometry: Clinical and Biomedical Applications (Desiderio, D. M., Ed.), Vol. 2, pp. 135–199. Plenum Press, New York. Witkowska, H. E., Bitsch, F., and Shackleton, C. H. L. (1993) Expediting rare variant hemoglobin characterization by combined HPLC/electrospray mass spectrometry. Hemoglobin 17(3), 227–242. Nakanishi, T., Kishikawa, M., Shimizu, A., Hayashi, A., and Inoue, F. (1995) Assignment of the ions in the electrospray ionization mass spectra of the tryptic digest of the non-derivatised globin, covering the whole sequence of ␣- and -chains: A rapid diagnosis for haemoglobinopathy. J. Mass Spectrom. 30, 1663– 1670. Nakanishi, T., Miyazaki, A., Kishikawa, M., Shimizu, A., and Yonezawa, T. (1996) Electrospray ionization– tandem mass spectrometry analysis of peptides derived by enzymatic digestion of oxidised globin subunits: An improved method to determine amino acid substitution in the hemoglobin “core.” J. Am. Soc. Mass Spectrom. 7, 1040 –1049. Roberts, N. B., Green, B. N., and Morris, M. (1997) Potential of electrospray mass spectrometry for quantifying glycohemoglobin. Clin. Chem. 43, 771–778. Green, B. N., Morris, M., Wild, B., Layton, M., and Bellingham, A. J. (1997) Developments toward the routine identification of variant hemoglobins by ES–
Wild et al.
24.
25.
26.
27.
28.
29.
30.
31.
704
MS. Presented at the 45th Annual Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, June 1–5. Bisse, E., Zorn, N., Eigel, A., Lizama, M., HuamanGuillen, P., Marz, W., Van Dorsselaer, A., and Wieland, H. (1998) Hemoglobin Rambam (69[E13]Gly3 Asp), a pitfall in the assessment of diabetic control: Characterisation by electrospray mass spectrometry and HPLC. Clin. Chem. 44(10), 2172–2177. Gatlin, C. L., Eng, J. K., Cross, S. T., Detter, J. C., and Yates, J. R., III (2000) Automated identification of amino acid sequence variations in proteins by HPLC/ microspray tandem mass spectrometry. Anal. Chem. 72, 757–763. Houston, C. T., and Reilly, J. P. (1999) Toward a simple, expedient and complete analysis of human haemoglobin by MALDI–TOFMS. Anal. Chem. 71, 3397–3404. Lemann, H., and Huntsman, R. G. (1974) Man’s Haemoglobins, 2nd ed., pp. 388 –391, North Holland, Oxford. Wild, B. J., and Stephens, A. D. (1997) The use of automated HPLC to detect and quantitate haemoglobins. Clin. Lab. Haematol. 19, 171–176. International Committee for Standardisation in Haematology (1978) Recommendations for selected methods for quantitative estimation of Hb A2 and for Hb A2 reference preparation. Br. J. Haematol. 38, 573–578. Wild, B. J., Green, B. N., Lalloz, M. R. A., and Layton, D. M. (1999) Molecular characterisation of hybrid haemoglobins by mass spectrometry. Br. J. Haematol. 105(Suppl. 1), 81, Abstract 212. De Caterina, M., Esposito, P., Grimaldi, E., Di Maro, G., Scopacasa, F., Ferranti, P., Parlapiano, A., Malorni, A., Pucci, P., and Marini, G. (1992) Characterization of hemoglobin Lepore variants by advanced mass-spectrometric procedures. Clin. Chem. 38(8), 1444 –1448.