SEARCH

SEARCH BY CITATION

Keywords:

  • mass spectrometry;
  • haemoglobinopathy;
  • sickle cell disease;
  • population screening

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References

Increasing demand for population screening for the haemoglobinopathies gives rise to a requirement for high throughput systems, which allow for cost effective, rapid, sensitive and specific screening of clinically significant haemoglobins. We have developed a practical and efficient approach using tryptic digestion and electrospray triple quadrupole mass spectrometry–mass spectrometry (MSMS) in multiple reaction monitoring acquisition mode for the identification of the clinically important haemoglobin variants, S, C, DPunjab, OArab, and E. A total of 200 blood samples, comprising 52 haemoglobin AA, 57 AS (sickle cell trait), 44 AC (C trait), 16 SC (SC disease), 14 SS (sickle cell disease), 10 AE (E trait), 2 ADPunjab (DPunjab trait) and 1 each of AOArab (OArab trait), CC (C disease), DPunjabDPunjab (DPunjab disease), OArabOArab (OArab disease), and EE (E disease), have been analysed in parallel with existing phenotype and molecular methods. All haemoglobin variants were correctly identified by MSMS, with no false positives or false negatives. The system detects both heterozygotes and homozygotes and has potential applications in neonatal and antenatal screening.

In man, mutations in the globin genes and the resulting haemoglobinopathies are the commonest inherited disorders and constitute a significant healthcare problem. Haemoglobin variants are usually detected during routine diagnostic follow-up of haematological problems but, more commonly, during preanaesthetic screening or as a result of neonatal and antenatal screening programmes. Over 800 haemoglobin variants have been characterised (Huisman et al, 1998; Pennsylvania State University, 2005), but the majority are not clinically significant. Recent health initiatives to improve the clinical and social impact of haemoglobinopathies (Department of Health, UK, 2000; NHS Sickle Cell and Thalassaemia Screening Programme, 2005) have led to a rapid expansion in neonatal and antenatal screening programmes and have dramatically increased the analytical workload. Although components of a complimentary strategy, the two screening programmes have different objectives. In antenatal screening the aim is to identify heterozygote carriers of those haemoglobinopathies that pose a genetic risk to the fetus. Thus, the aim is to detect the presence or absence of sickle haemoglobin and beta thalassaemia trait or one of the haemoglobin variants that interact with them. These comprise haemoglobins C, DPunjab, OArab, Lepore, and E. In addition, three other conditions of potential clinical significance, namely delta beta thalassaemia, hereditary persistence of fetal haemoglobin trait (HPFH) and alpha zero thalassaemia trait are also included. In neonatal screening the aim is early identification of individuals with sickle cell disease and beta thalassaemia major in order to ensure early initiation of treatment.

Classical biochemical diagnosis of haemoglobinopathies depends on phenotypic information generated from charge sensitive methods, either isoelectric focussing (IEF) or cation-exchange chromatography (British Committee for Standards in Haematology, 1998). These methods form the basis of current haemoglobinopathy screening techniques but, in population screening terms, are considered slow, laborious, and non-specific. Positive cases require confirmatory testing and, more importantly, a range of clinically silent variants are detected. Mass spectrometry–mass spectrometry (MSMS) is potentially faster, more specific, and is already used for cost-effective population screening of inherited metabolic disorders (IMD) (Chace et al, 2003). Electrospray mass spectrometry (MS) and MSMS have both been used for the characterisation of haemoglobinopathies. MS-derived whole blood scans have been used to measure the masses of the intact globin chains and, following tryptic digestion, MSMS techniques have been used for peptide analysis (Wild & Bain, 2001). Although the methods are not suitable for population screening it has been possible to unequivocally characterise the majority of globin mutations. We have developed an MSMS procedure to specifically target and identify the clinically important globin variants in a format suitable for population screening. This approach utilises the simple fact that the actual genetic mutations responsible for each of the variants described have already been characterised and, therefore, the amino acid sequences are known. Tryptic digestion of globin chains leads to specific cleavage of the peptide bonds adjacent to lysine and arginine residues and produces a well-characterised series of peptides. MSMS analysis enables the selection of informative peptides and subsequent collision induced fragmentation such that a product ion specific to a given mutation can be detected. Mutations that alter cleavage sites, by adding or replacing arginine and lysine, generate peptides that are specific if only a single mutation is present.

We have devised a series of multiple reaction monitoring (MRM) and ‘pseudo-MRM’ acquisitions to investigate three informative peptides formed during tryptic digestion of beta globin, namely T1, T3 and T13, which include the mutations associated with haemoglobins S, C, DPunjab, OArab, and E. The specific MRM transition for haemoglobin S was targeted, together with the new peptides specific for haemoglobins C, DPunjab, OArab and E. An equivalent wild-type peptide fragmentation product ion for the normal beta T1 chain and ions corresponding to the normal T3 and T13 peptides were also included to act as a quality control check for the analytical procedure and to allow the carrier or disease status of the individual to be determined.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References

MSMS strategy

Although the MSMS strategy is simple in concept, a detailed explanation of the process used to identify the specific MRM transition for haemoglobin S and the ‘pseudo-MRM’ for haemoglobin C provides a model for how to approach any other specified mutation. Sample preparation is described later in the appropriate section.

Haemoglobin S

Haemoglobin S is the product of a substitution of the amino acid valine for the wild-type glutamic acid at position 6 of the beta globin chain. As a result, the mass of the sickle protein is 30 Da less than that of wild-type beta chain. This difference can be readily observed in the re-constructed spectra of whole blood charge series scans (Wild et al, 2001). However, at a whole molecule level, a mass difference of −30 Da could be due not only to a valine for glutamic acid substitution but also for arginine to tryptophan, threonine to methionine, glycine to serine, and alanine to threonine. In addition, measuring the whole protein mass provides no information on the position of the mutation causing the mass shift. At this stage, fragmentation of the beta chain, either during MS analysis or, more effectively for MSMS, by preanalytical proteolysis, usually with trypsin, is required. Trypsin specifically cleaves peptide bonds on the C-terminal side of lysine or arginine. Consequently, the 15 peptides resulting from tryptic digestion of the wild-type beta chain have been well characterised, are predictable, and are referred to as T1–T15 (Wild et al, 2001). The informative peptide for the sickle mutation is T1, which contains the first eight N-terminal amino acids of the beta chain. The amino acid sequence is shown in Table I.

Table I.  Beta-globin T1 peptide amino acid sequence.
PositionAmino acidAbbreviated formOne letter code
1ValineValV
2HistidineHisH
3LeucineLeuL
4ThreonineThrT
5ProlineProlP
6Glutamic AcidGluE
7Glutamic AcidGluE
8LysineLysK

Electrospray ionisation of the complex peptide mixture formed following trypsin digestion of whole blood produces an array of singly- and multiply-charged peptides. The theoretical mass/charge ratio of the singly-charged wild-type beta chain T1 peptide is [M + H]+ 952·5 and is 476·8 for the doubly-charged peptide [M + H]2+. Both species are observed on an m/z 100–1500 scan of an appropriate digest (Fig 1A). In a sample from a patient with sickle cell disease the substitution of the glutamic acid at position 6 with valine produces equivalent charged peptides [M + H]+ 922·5 and [M + H]2+ 461·8 (Fig 1B) and the wild-type ions are absent. Therefore, the 30 Da mass reduction from wild-type beta chain has been isolated to the T1 peptide. This is highly specific, but a further level of specificity and background reduction can be introduced by using collision-induced dissociation (CID) or fragmentation of the T1 peptide. The peptide of interest, either singly or doubly charged, can be isolated in MS1 subjected to CID, and the fragment ions analysed in MS2. Because peptide bonds are relatively weak the fragmentation pattern provides amino acid sequence information. Fragmentation of charged peptides usually results in the production of two complementary peptide ion series, termed the y ions and b ions. The y ion retains a positive charge at its C-terminal end, whilst the b ion retains the positive charge at the N-terminal end. Thus it is possible to sequence a peptide (Liang et al, 2005). Conversely, from knowledge of the amino acid sequence of a peptide, it is possible to calculate the theoretical masses of the resulting ions. These are shown for the wild-type and sickle T1 peptides in Tables II and III.

image

Figure 1. Mass scans m/z 100–1500 of tryptic digests of blood from a control subject with normal beta haemoglobin (A) and a patient with sickle cell disease (B). The insets emphasise the complexity of raw spectra. The arrows reveal zoomed spectra for the informative [M + 2H]2+ and [M + H]+ ions for control (A) and sickle patient (B), respectively. Note the mass shift of −30 Da for the [M + H]+ ions and −15 Da for the [M + 2H]+ ions.

Download figure to PowerPoint

Table II.  Theoretical wild-type beta T1 peptide fragment ions. MRM product ion shown in bold.
PositionWild-type amino acid sequence b ion y ion
1Vb1100·1  
2Hb2237·2y7853·4
3Lb3350·3y6716·3
4Tb4451·3y5603·2
5Pb5548·4y4502·2
6Eb6677·4y3405·1
7Eb7806·4y2276·1
8K  y1147·1
Table III.  Theoretical haemoglobin S beta T1 peptide fragment ions. MRM product ion shown in bold.
PositionSickle amino acid sequence b ion y ion
1Vb1100·1  
2Hb2237·2y7823·5
3Lb3350·3y6686·4
4Tb4451·3y5573·3
5Pb5548·4y4472·3
6Vb6647·5y3375·2
7Eb7776·5y2276·1
8K  y1147·1

Product ion scans of wild-type (Fig 2A) and sickle (Fig 2B) [M + 2H]2+ ions demonstrate the practicality of the approach. Note that the [M + 2H]2+ ions were used because they fragment at lower collision energies (Light-Wahl et al, 1993). The scan of the sickle protein is unique and essentially diagnostic. However, for population screening of multiple mutations targeting an informative ion provides the opportunity to use the instrument in the high sensitivity MRM mode. In the case of sickle protein, the theoretically most specific target is the y3 ion. In practice, because of the proline effect at position 5 (Williams et al, 1982) the y4 ion (m/z 472·3) provides a much more sensitive signal. The equivalent wild-type reference ion is m/z 502·2. The theoretical wild-type MRM is m/z 476·8/502·2 and for sickle protein m/z 461·8/472·3.

image

Figure 2. Product ion scans of control beta haemoglobin T1 peptide [M + 2H]2+, m/z 476·8 (A) and haemoglobin S T1 peptide [M + 2H]2+, m/z 461·8 (B). Note the y and b series fragments that provide unequivocal sequence data and the y4 MRM target ions, m/z 502·3 and 472·5, respectively.

Download figure to PowerPoint

The final method for identification of sickle protein involves tryptic digestion of whole blood, automated direct injection of the digest, two MRM acquisitions, and an inject-to-inject time of approximately 1 min. Unequivocal confirmation can be made using a second injection of the tryptic digest and product ion scanning of the informative peptide to provide sequence information.

Haemoglobin C

Haemoglobin C is the product of a substitution of the amino acid lysine for the wild type glutamic acid at position 6 of the beta globin chain. As a result, the mass of the C protein is 1 Da less than that of wild-type beta chain. This difference is almost impossible to detect in re-constructed spectra of whole blood charge series scans because of the mass resolution of triple quadrupole instruments (Wild et al, 2001), particularly in a true population screening environment. The informative peptide for the C mutation, as for sickle, is T1. However, the substitution generates a new trypsin cleavage site at position 6 and a unique peptide, VHLTPK [M + H]+ 694·4. In this instance, the specificity of the new peptide renders fragmentation unnecessary for screening. The higher stability of singly charged peptides when compared with doubly charged peptides means that, at the collision energy required for the sickle and wild-type MRMs, the C peptide-specific [M + H]+ ion undergoes minimal fragmentation. Thus, a ‘pseudo-MRM’ m/z 694·4/694·4 can be applied.

The same ‘pseudo-MRM’ approach has been used for haemoglobins DPunjab, OArab, and E. In brief, haemoglobin DPunjab is the product of a substitution of the amino acid glutamine for the wild-type glutamic acid at position 121 of the beta globin chain, i.e. a mass shift of −1 Da. The informative peptide is T13 and contains the amino acids 121–132. The wild-type T13 sequence of EFTPPVQAAYQK is altered to QFTPPVQAAYQK. No other single amino acid substitution in this peptide will cause a mass alteration of −1. ‘Pseudo-MRMs’ for the T13 wild-type peptide [M + H]+ 1378·8 Da and DPunjab [M + H]+ 1377·8 are acquired. Haemoglobin OArab is the product of a substitution of the amino acid lysine for the wild-type glutamic acid at position 121 of the beta globin chain. This creates a new trypsin cleavage point at position 121 and a unique peptide FTPPVQAAYQK [M + H]+ 1249·7. Haemoglobin E is the product of a substitution of the amino acid lysine for the wild-type glutamic acid at position 26 of the beta globin chain. The informative peptide is T3 and contains the amino acids 18–30, VNVDEVGGEALGR [M + H]+ 1314·7 Da. The E mutation creates a new trypsin cleavage site and 2 new peptides, VNVDEVGGK [M + H]+ 916·8 and ALGR [M + H]+ 416·3. The higher mass was selected for use on the basis of improved signal to noise ratio at higher masses.

A summary of the target peptide ions and fragment ions is shown in Table IV.

Table IV.  MSMS target peptide and ion masses.
 Target peptide ion (Da)Target fragment ion (Da)
Wild-type T1476·8y4 502·3
Haemoglobin S461·9y4 472·4
Haemoglobin C694·7 
Wild type T131378·8 
Haemoglobin DPunjab1377·8 
Haemoglobin OArab1249·7 
Wild-type T31314·7 
Haemoglobin E916·8 

Patient group

A total of 200 whole blood EDTA samples, consented for haemoglobinopathy diagnosis and selected to provide significant numbers for each of the variants to be tested, were analysed routinely using established methods and then anonymised for MSMS. These comprised 52 haemoglobin AA, 57 AS (sickle cell trait), 44 AC (C trait), 16 SC (SC disease), 14 SS (sickle cell disease), 10 AE (E trait), 2 ADPunjab (DPunjab trait) and 1 each of AOArab (OArab trait), CC (C disease), DPunjabDPunjab (DPunjab disease), OArabOArab (OArab disease), and EE (E disease).

Materials for trypsin digestion and MSMS analysis

Ammonium bicarbonate (A6141), tosyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (T1426), and 88% formic acid (39,938–8) (Sigma Aldrich Co. Ltd, Dorset, UK). High performance liquid chromatography (HPLC) grade acetonitrile (RH1015) (Rathburn Chemicals Ltd, Walkerburn, UK).

Standard methods for haemoglobinopathy characterisation

The Guideline, laboratory diagnosis of haemoglobinopathies (British Committee for Standards in Haematology, 1998) was taken as minimum standard. The initial haemoglobinopathy screen was performed by HPLC using a VariantTM II operating with HbA2/HbA1c Dual Program kit (Bio-Rad Laboratories Ltd, Hemel Hempstead, UK). Confirmatory tests for provisional haemoglobin identification were made using established methods, including sickle solubility test, acid and alkaline gels (Wild & Bain, 2001), polymerase chain reaction (Fodor & Eng, 1999) and beta globin gene sequencing (Mai et al, 2004).

Sample preparation for MSMS

Following the method described (Wild et al, 2001), EDTA whole blood (10 μl) was simply diluted in deionised water (490 μl) to create a working solution. Acetonitrile (10 μl) and 1% formic acid (10 μl) were then added to 100 μl of the working solution and mixed in order to denature the haemoglobin. The sample was allowed to stand for 5 min at room temperature before addition of 1 mol/l ammonium bicarbonate (6 μl) and TPCK-treated trypsin (5 μl of 5 mg/ml trypsin in deionised water). Initially the sample appeared cloudy but once the solution had cleared, it was pulse centrifuged and incubated for 30 min at 37°C. Following the proteolytic digestion, 40 μl of the solution was diluted in 360 μl of acetonitrile:water (1:1) with 0·2% formic acid, transferred to a 96 deep-well polypropylene plate (Semat International Ltd, St Albans, UK), and loaded onto an CTC Analytics HTS PAL refrigerated autosampler (Presearch Ltd, Hitchin, UK) for MSMS analysis.

Mass spectrometry

Samples (2 μl) were automatically introduced into a continuous solvent stream of acetonitrile:water (1:1) containing 0·025% formic acid flowing at 75 μl/min (Agilent 1100 series) into a SCIEX API 4000 (Applied Biosystems, Warrington, UK) triple quadrupole MSMS with an electrospray source in positive ion mode at 5500 V and 250°C. The interface heater was on, declustering potential 81·0 V, and entrance potential 10 V. The collision gas setting (6·0), collision energy (30 V) and exit potential (15·0 V) were constant for three MSMS experiments. The first experiment, targeted wild-type beta globin T1 and the T1 variants, S and C, the second, wild-type T13 and the T13 variants, DPunjab and OArab, and the third, wild-type T3 and the T3 variant, E. The actual MRMs and ‘pseudo-MRMs’ are shown in Table IV; dwell time 150 ms for each transition. The total acquisition time was 60 s.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References

In practice, four panels of data were presented to the operator for each sample: the total ion chromatogram and then the T1, T13, and T3 peptides. For simplicity and to demonstrate the data effectively, T1 MRM data for samples from subjects with haemoglobins AA, AS, and SS are shown in Fig 3. Note the virtual absence of S signal in the AA, resulting in a very high signal to noise ratio and hence sensitivity for S. Equivalent data for C, DPunjab, OArab, and E are presented in Figs 4–7, respectively. Even using ‘pseudo-MRMs’ the background signals were relatively small and simple visual inspection could identify heterozygote and homozygote status for all the mutations targeted. There was an apparently high wild-type signal for the T13 peptide, even in patients homozygous for DPunjab, because of the 1 Da shift. However, because of the relative ratios of the m/z 1377·8/1378·8 signals, detection of heterozygous and homozygous DPunjab were unaffected. The predicted result patterns for each of the haemoglobin phenotypes are shown in Table V.

image

Figure 3. T1 peptide of beta globin – extracted MRM transitions m/z 476·9/502·3 (wild-type beta haemoglobin) and m/z 461·9/472·4 (haemoglobin S) for control, sickle cell trait, and sickle cell disease.

Download figure to PowerPoint

image

Figure 4. T1 peptide of beta globin – extracted MRM transitions m/z 476·9/502·3 (wild-type beta haemoglobin) and m/z 694·7/694·7 (haemoglobin C) for control, C trait, and C disease.

Download figure to PowerPoint

image

Figure 5. T13 peptide of beta globin – extracted MRM transitions m/z 1378·8/1378·8 (wild-type beta haemoglobin) and m/z 1377·8/1377·8 (haemoglobin DPunjab) for control, DPunjab trait, and DPunjab disease.

Download figure to PowerPoint

image

Figure 6. T13 peptide of beta globin – extracted MRM transitions m/z 1378·8/1378·8 (wild-type beta haemoglobin) and m/z 1249·7/1249·7 (haemoglobin OArab) for control, OArab trait, and OArab disease.

Download figure to PowerPoint

image

Figure 7. T3 peptide of beta globin – extracted MRM transitions m/z 1314·7/1314·7 (wild-type beta haemoglobin) and m/z 916·8/916·8 (haemoglobin E) for control, E trait, and E disease.

Download figure to PowerPoint

Table V.  Mutations, numbers tested, and predicted result patterns for samples tested.
Hb phenotypeNumber of samplesWild-type T1 (502·3 Da)Hb S (472·4 Da)Hb C (694·7 Da)Wild-type T13 (1378·8 Da)Hb DPunjab (1377·8 Da)Hb OArab (1249·7 Da)Wild-type T3 (1314·7 Da)Hb E (916·8 Da)
AA52Present  Present  Present 
AS57PresentPresent Present  Present 
AC44Present PresentPresent  Present 
SS14 Present Present  Present 
SC16 PresentPresentPresent  Present 
AE10Present  Present  PresentPresent
ADPunjab2Present  PresentPresent Present 
AOArab1Present  Present PresentPresent 
CC1  PresentPresent  Present 
DPunjabDPunjab1Present   Present Present 
OArabOArab1Present    PresentPresent 
EE1Present  Present   Present

The specimens selected for analysis provided examples of all the haemoglobin variants detectable using the current method. All the variants, initially characterised and identified by conventional techniques, were identified unequivocally using tryptic digestion of the globin chains and rapid, 1-min acquisition, MSMS. Furthermore, as predicted, the wild-type peptide was absent in disease and compound heterozygote states. The exceptions to this were those patients who had undergone transfusion where, as would be expected, the wild-type peptide was present along with that of the variant. Using the approach described, the detection of heterozygote or homozygote haemoglobins S, C, DPunjab, OArab, and E was 100% specific and 100% sensitive, in the 200 samples analysed.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References

The use of mass spectrometry to characterise haemoglobin variants is not a new approach and, with the introduction of robust and simple-to-use electrospray MS instruments, the number of publications is increasing rapidly. However, to the uninitiated, the MS and MSMS of proteins is generally presented in a form that appears highly complex and, although suitable for confirmation of mutations in a small number of samples, is unlikely to impact on routine clinical diagnostics. The starting point for the use of MSMS in the characterisation of haemoglobinopathies has been the detection of charge variants during routine analysis of haemoglobin by cation-exchange HPLC or IEF. As a consequence, MSMS has been used to definitively identify unknown haemoglobin variants (Wild et al, 2001), an effective but relatively time-consuming process. The aim of the present study was to utilise our knowledge of well-characterised beta haemoglobin mutations to develop a simple, rapid, and targeted MSMS approach suitable for high-throughput population screening of the clinically important haemoglobinopathies.

Although both HPLC and IEF can be modified to increase throughput and are currently the methods of choice for population screening (Henthorn et al, 2004), neither is particularly suitable for the purpose. In contrast, the last decade has seen quantitative electrospray MSMS of diagnostic metabolites emerge as the dominant technology in neonatal screening programmes for the detection of IMD (Chace et al, 2003). The sample preparation, particularly using underivatised samples, is very simple, the results, in MRM mode, are accurate and precise, and the worldwide experience has proven just how robust the technology is for population screening. Laboratories providing a neonatal IMD screen, particularly in North America and the UK, are either already performing universal haemoglobinopathy screening or are under pressure to introduce a service. In developing our current approach, the potential for providing both programmes on a single technology platform, allowing more efficient utilisation of MSMS time and/or providing back-up facilities, proved very compelling. Compared with the sample preparation for IMD screening, the tryptic digestion is more complex than the underivatised method but is much simpler, faster, more easily automated than the butylation method, and quality control of the hydrolysis process can be monitored in each sample by including transitions for each wild-type peptide. The solvent system used in the haemoglobinopathy method is the same as that used in IMD screening, so that they can be seamlessly run back-to-back and, on a single instrument, a service covering 50 000 births per annum can be operated for both IMD and haemoglobinopathy programmes, where screening samples are run overnight and the machine is available for other analyses during the day. On its own, the haemoglobinopathy system could be made considerably quicker with a more rapid autosampler: a potential of 4000 samples per day is not unreasonable.

Sensitivity and specificity of the test are essential considerations in any population-screening programme, even though screening may be only the first stage in providing a final diagnosis. The aim of a true screening programme is ‘to identify those individuals within a population who have a specific disorder for which intervention, such as medical treatment, education or counselling, can improve the natural course of the disease’ (Henthorn et al, 2004). The ability of both HPLC and IEF to detect a large number of haemoglobin variants has always been regarded as an advantage in routine haematological practice. However, many variants are not clinically significant and the specific identification of these variants, currently undertaken by many laboratories, could be viewed as an ineffective use of limited resources. At this stage, the identification of the variant, in many cases, is still provisional and requires confirmation by other techniques, including MSMS, before final validation of results (Wild & Bain, 2004). Targeted detection of those haemoglobin variants deemed clinically significant within a screening programme allows time and resources to be totally focussed, results to be produced within a clinically relevant time frame, and treatment/counselling to be implemented early in neonatal/antenatal programmes, respectively.

We have developed a novel strategy using tryptic digestion and MSMS to provide both sensitivity and specificity as a screening method for clinically significant haemoglobin variants. In devising this initial strategy it was decided to focus on haemoglobins S, C, DPunjab, OArab, and E, since these mutations represent a large proportion of the clinically significant haemoglobinopathies (NHS Sickle Cell and Thalassaemia Screening Programme, 2005) and appropriate samples for comparison were readily available within the laboratory. The obvious and simple screening approach would be to look at the mass of intact beta globin chains. Although this is possible for haemoglobin S, where the mass shift from wild-type is −30 Da, it is not suitable for haemoglobins C, DPunjab, OArab, and E, where the mass shift is only 1 Da. The decision to focus on tryptic peptides, despite the extra sample processing involved, was initially based on the need to detect these 1 Da mutations. Targeting tryptic digest peptides not only makes this possible but increases specificity and provides a ready prepared sample with the potential for sequencing and unequivocal identification. The specificity issues were addressed in the MSMS strategy for haemoglobin S (see above). A mass shift of −30 Da for the intact beta globin protein is not specific for haemoglobin S, neither is a mass shift of −30 Da in the T1 peptide (threonine to alanine at position 4 and glutamic acid to valine at position 7 would both give a −30 Da shift), even the MRM transition is theoretically not unique, but in practical terms for population screening it is highly sensitive, requires no interpretation, and positive results can be easily flagged electronically. Actual confirmation requires the product ion scan of the −30 Da T1 peptide and is unequivocal (Fig 2). The amino acid sequence, b series and y series, can be simply read from the scan. When haemoglobin S is detected using the MRM during screening the sample may be re-submitted for a product ion scan and confirmation as S. However, using an MSMS instrument with a linear ion-trap, because of the rapid product ion scanning facility, product ion scans could be routinely collected with each MRM, providing unequivocal sequence data for each sample that could be referred to without the need for re-injection. The ability to generate peptide sequence data on the initial screening sample has significant implications for the necessity/provision of confirmatory testing.

The stated requirements of the United Kingdom National Health Service Haemoglobinopathy neonatal screening programme imply that, as a minimum in the initial screening phase, detection of haemoglobin S and/or demonstrating a deficiency of wild-type beta-globin should be sufficient (NHS Sickle Cell and Thalassaemia Screening Programme, 2005). Heterozygote or homozygote status can be determined and other sickling mutations characterised at the confirmatory testing stage. The confounding problem in neonatal screening is the very low expression of beta-globin in neonates, particularly those that are premature. It is essential, therefore, to have a very sensitive screening method for detecting haemoglobin S. Using the specific MRM transition for haemoglobin S, the non-specific background signal found in a haemoglobin AA subject (Fig 3) indicated that even levels of haemoglobin S representing significantly less than 1% of total haemoglobin will be detectable. Although, not strictly essential, the current approach also enables detection and confirmation of the other sickling mutations in the original tryptic digest.

The success of the MRM fragmentation approach for haemoglobin S implies that it should have provided a model for the other mutations of interest. Initial experiments with haemoglobins C, DPunjab, OArab, and E demonstrated very high specificity using the peptides formed as a result of the new trypsin cleavage sites unique for each mutation. Sensitivity and specificity were not an issue in the 200 samples analysed but, as confirmation of the mutations was not an objective of the study, MSMS fragmentation was not pursued. Interest centred on the use of ‘pseudo-MRMs’. To perform the wild-type T1 and haemoglobin S T1 fragmentations it is obviously necessary to have the collision gas flowing and the collision energy optimised for the fragmentation to be maximally informative, in this case for the doubly charged species 30 V. At the same energy there is virtually no fragmentation of singly charged peptides, allowing unique peptides to be selected in MS1 and then re-selected in MS2, a ‘pseudo-MRM’. In the present study, this approach resulted in sensitive and specific detection of each mutation targeted. In retrospect, because of the sequence information available on the singly charged peptides by increasing the collision energy to between 45 and 60 V, this may have been a better approach and is currently being investigated. However, unequivocal sequence information can still be obtained for haemoglobin C, DPunjab, OArab, and E using a second injection and product ion scanning in the same way as demonstrated for haemoglobin S.

The potential of this method for neonatal haemoglobinopathy screening has already been discussed but this approach may prove even more valuable in antenatal screening programmes. The major difference between neonatal and antenatal screening is that in the latter it is essential that heterozygotes for compound sickling mutations be detected. At present, we have not described a comprehensive system for antenatal screening but we have demonstrated how successful the approach is in detecting heterozygotes for the majority of the compound sickling mutations. This work is in progress and, although the haemoglobin Lepores and beta thalassaemia trait present some interesting challenges for MSMS, we envisage an MSMS system for antenatal screening will be available in the very near future. More generally, the process is not restricted to the beta chain and could be extended to include any other mutations of clinical interest and provide a comprehensive approach to clinical haemoglobinopathy characterisation and diagnosis. Furthermore, when the clinical picture suggests a haemoglobinopathy, but the targeted mutations are all normal, the MSMS is still available to do a classical sequence analysis.

Technically, MSMS is more specific and more sensitive than either HPLC or IEF and offers the possibility of unequivocal haemoglobin characterisation. However, a cost comparison of the methods is more difficult: both HPLC and IEF require relatively small capital investment but suffer from high consumable costs, while the opposite is true for MSMS. The economics are totally dependent on the size of the population screened, with MSMS more cost-effective with larger populations. However, if integrated with IMD screening, there are various changes to screening laboratory configurations and screened populations where MSMS would provide a highly cost-effective solution.

The samples we used to test the methodology provided examples from both heterozygous and homozygous subjects for each of the mutations targeted, including a significant number of haemoglobin SC compound sickling mutations. However, this initial study was limited to only 200 samples. A definitive assessment of the effectiveness, or otherwise, of this targeted MSMS approach to population screening for clinical haemoglobinopathies will require a significantly sized population study.

In conclusion, a series of MSMS MRM and ‘pseudo MRM’ experiments have been developed to target tryptic digest peptides characteristic of haemoglobins S, C, DPunjab, OArab and E. In 200 highly informative blood samples identified by existing techniques the sensitivity and specificity by MSMS were both 100%. The study serves as a model for clinical haemoglobinopathy diagnosis and demonstrates a significant advance in the application of clinical proteomics.

Author contributions

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References

YAD performed the research, analysed the data, and wrote the manuscript. CT designed the research strategy and contributed to the performance of the research, the analysis of the data, and the writing of the manuscript. RMH contributed to the performance of the research and the writing of the manuscript. BJH provided clinical advice and supervision and contributed to the writing of the manuscript. RND designed the research strategy and contributed to the performance of the research, the analysis of the data, and the writing of the manuscript.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Author contributions
  7. References
  • British Committee for Standards in Haematology (1998) Guideline, the laboratory diagnosis of haemoglobinopathies. British Journal of Haematology, 101, 783792.
  • Chace, D.H., Kalas, T.A. & Naylor, E.W. (2003) Use of tandem mass spectrometry for multianalyte screening of dried blood specimens from newborns. Clinical Chemistry, 49, 17971817.
  • Department of Health, UK (2000) The NHS Plan: A Plan for Investment, A Plan for Reform. Her Majesty's Stationery Office, Norwich, UK, Series number Cm 4818-I:112.
  • Fodor, F.H. & Eng, C.M. (1999) Molecular exclusion of haemoglobin SD disease by prenatal diagnosis. Prenatal Diagnosis, 19, 5860.
  • Henthorn, J.S., Almeida, A.M. & Davies, S.C. (2004) Neonatal screening for sickle cell disorders. British Journal of Haematology, 124, 259263.
  • Huisman, T., Carver, M. & Efremov, G. (1998) Human Hemoglobin Variants, 2nd edn. Sickle Cell Anaemia Foundation, Augusta, USA.
  • Liang, C.H., Patel, M. & Vissa, N. (2005) Mass spectrometry tutorial [WWW document], available at: http://www.cmb.usc.edu/cbmp/2002/MassSpectrometry/tutorial.html, accessed on 26 April 2005.
  • Light-Wahl, K.J., Loo, J.A., Edmonds, C.G., Smith, R.D., Witkowska, H.E., Shackleton, C.H.L. & Wu, C.C. (1993) Collisionally activated dissociation and tandem mass spectrometry of intact hemoglobin β-chain variant proteins with electrospray ionisation. Biological Mass Spectrometry, 22, 112120.
  • Mai, M., Hoyer, J.D. & McClure, R.F. (2004) Use of multiple displacement amplification to amplify genomic DNA before sequencing of the α and β haemoglobin genes. Journal of Clinical Pathology, 57, 637640.
  • NHS Sickle Cell and Thalassaemia Screening Programme (2005) NHS Sickle Cell and Thalassaemia Screening Programme [WWW document], available at: http://www.kcl-phs.org.uk/haemscreening, accessed on 26 April 2005.
  • Pennsylvania State University (2005) HbVar: A database of human hemoglobin variants and thalassaemias [WWW document], available at: http://globin.cse.psu.edu, accessed on 26 April 2005.
  • Wild, B.J. & Bain, B.J. (2001) Investigation of abnormal haemoglobins and thalassaemia. In: Dacie and Lewis Practical Haematology, 9th edn. (ed. by S.M.Lewis, B.J.Bain & I.Bates), pp. 231268. Churchill Livingstone, London, UK.
  • Wild, B.J. & Bain, B.J. (2004) Detection and quantitation of normal and variant haemoglobins: an analytical review. Annals of Clinical Biochemistry, 41, 355369.
  • Wild, B.J., Green, B.N., Cooper, E.K., Lalloz, M.R.A., Erten, S., Stephens, A.D. & Layton, D.M. (2001) Rapid identification of hemoglobin variants by electrospray ionization mass spectrometry. Blood Cells, Molecules, and Diseases, 27, 691704.
  • Williams, D.H., Bradley, C.V., Santikarn, S. & Bojesen, G. (1982) Fast-atom-bombardment mass spectrometry. Biochemical Journal, 201, 105117.