Mass spectrometry and data processing methods

Sample Introduction

Two types of sample introduction were employed in these studies; directly introducing the sample into the source of mass spectrometry or after a chromatographic separation step. Direct injection can be implemented via a syringe, using the fluidic system of the instrument or using an automated sample introduction system TriVersa NanoMate (Advion Biosciences Ltd., Ithaca, NY, USA). Chromatographic separation and sample introduction were achieved via the Acquity UPLC system.

Intact globin chain analysis and multistep hemoglobin variant identification The use of a triple quadrupole system for the analysis of intact globin chains is described and optimized for its application in a high-throughput manner to analyse a significant number of samples. For high-throughput applications, sample introduction needs to be performed in an automated manner to allow the measurements of multiple samples to be carried out 24 hours a day.

72 A Xevo-TQ (Waters Corporation, Manchester, UK) equipped with a standard flow ESI source and controlled by MassLynxTM (version 4.1) software (Waters Corporation, Manchester, UK) was used for method development and testing on standard samples. Samples were introduced into the ESI source from a 250 µL syringe at a flow rate of 5 µL/min using a syringe. In other cases the fluidics system of the Xevo TQ instrument was used, and 1 ml of sample prepared in a 1.5 ml vial which was then introduced to the ESI source again at a flow rate of 5 µL/min.

For high-throughput method development the Xevo-TQ (Waters Corporation, Milford, Manchester, UK) equipped with a TriVersa NanoMate (Advion Biosciences Ltd., Ithaca, NY, USA) automated sample introduction system was used to acquire data on clinical samples.

The instrument was operated in ESI positive mode with a capillary voltage of 3 kV, cone voltage 40 V and source temperature 150ºC for all experiments.

The mass analyser was calibrated using the alpha-chain multiple charged peak series of a control blood sample.

Data were acquired through the tune page of the MassLynxTM software for 3 minutes in multi-channel acquisition (MCA) mode at a scan speed of 1 spectrum/sec.

The measured ESI-MS spectrum was background subtracted with a 25-order polynomial to allow 5% of the data to fall below the new baseline, this subtracted spectrum was then deconvoluted onto a true mass scale by use of a maximum entropy based MaxEnt application program, which is available within the MassLynx software. The mass output parameters were set for mass range 14800-16800 Da at a resolution of 0.15 Da/channel. A Gaussian model was used: peak width 0.8 Da, 40% minimum intensity ratios were set and the algorithm was allowed to iterate until convergence. The deconvoluted spectrum obtained from MaxEnt was smoothed and centred, and a mass correction was applied so that the alpha-chain mass was set to the expected 15126.38 Da.

Samples were assessed for the presence of single-point mutation hemoglobin variants, δ-chain and glycation levels.

The TriVersa NanoMate (Advion Biosciences Ltd., Ithaca, NY, USA) automated sample introduction system utilises an ESI micro fluidics chip containing an array of 400 nano-electrospray nozzles. The system was controlled by ChipSoft software. Sample introduction was performed with 5 µL spray volumes and 1 µL air aspirated

73 after the sample, spray gas pressure 0.8 psi (later 1psi), a 1.55 kV (later 2.1 kV) voltage applied to the tip, sample plate temperature was kept at 15ºC. A 25 second delay time was applied for the inlet start sign to be sent to the mass spectrometer allowing a stable spray to form.

Data were acquired in triplicate through use of an automatic sample list. For each sample, a three minute acquisition was performed in multi-channel acquisition (MCA) mode at a scan speed of 1 spectrum/sec.

To interpret and process raw data measured in the high-throughput analysis BioPharmaLynxTM software (version 1.2, Waters Corporation) was used. BioPharmaLynxTM allows one to deconvolute multiple spectral outputs, deconvolution parameters are set up in the processing method, and results are listed in a table format for each processed sample. This way the calculated intensities for the different globin chains can be easily extracted. The extraction process of the required information needs to be automated to fit to the high-throughput manner of the analysis.

The processing parameters in the BioPharmaLynxTM method were the following: A Gaussian damage model was used (an isotope damage model is not available within this software package). The mass accuracy was linked to a lock mass of 1081.46 Da, with a mass tolerance of ±0.5 Da, MS mass match tolerance for deconvoluted mass peaks was 80 ppm. Deconvolution parameters: range 930-1210 Da, protein MW range 14,800-16,800, peak width 0.8 Da. Background subtraction of the deconvoluted data prior to deconvolution with a background threshold of 5% and a polynomial order 25. The deconvoluted mass spectrum was smoothed with the mean algorithm twice with a smoothing window of 6 channels. The spectrum was also centred using the top 90% and the areas or the heights of the deconvoluted peaks were reported.

After analysis the count number for intensities of pre-identified masses (at greater than 10 % intensity) for the α, β, βSICKLE_{, δ and glycated globin chains were listed}

and extracted into Microsoft Excel for three replicates of each measured sample. The percentages of the δ-chain intensity and glycated α- and β-chain intensity were calculated within the software.

The percentage of glycated hemoglobin present was calculated as described by Roberts et al. and shown in Figure 2.1 (Roberts et al. 2001). The determination of

74 glycated hemoglobin level is based on several assumptions. It is assumed that there is no significant contribution to glycosylated hemoglobin from other globin chains, that all α-chain species have the same sensitivity and that all β-chain species have the same sensitivity. The results obtained when these assumptions were applied were shown by Roberts et al. to correlate with those obtained by other techniques (Roberts

et al. 2001). Results shown later in this Chapter suggest that the calculation of only beta-chain glycation might be necessary to obtain an estimation of the overall glycated hemoglobin level.

Figure 2.1.Schematic of the direct infusion method for glycated hemoglobin level quantitation

Assessment of δ-chain levels was achieved by carrying out a ratio calculation of δ- chain intensity to β-chain intensity, based on the assumption that the β- and δ-chains have similar ionisation efficiencies. This approach uses a δ:β ratio of the intact chains to provide a surrogate marker of Hb A2 levels as illustrated by Figure 2.2.

β-chain 15867.24 Da ESI-MS spectra of blood 1/500 in 50% ACN 0.2% HCOOH obtained Deconvoluted onto true mass scale Centred by area %Glycated Hb calculated α-chain 15126.38 Da αG-chain 15288.52 Da βG-chain 16029.38 Da %

75 Figure 2.2.Schematic of the direct infusion method for delta chain quantitation

A summary of the processes and methods is simplified in Figure 2.3.

δ-chain 15924.32 Da β-chain 15867.24 Da ESI-MS spectra of blood 1/500 in 50% ACN 0.2% HCOOH obtained Deconvoluted onto true mass scale Centred by area %δ calculated

76 Figure 2.3.Illustration of the direct infusion method for the intact globin chain analysis and hemoglobin variant identification

Blood Sample

Sample Preparation for Digested Globin

Chain analysis

Mass spectrum of tryptically

digested sample acquired and

compared to that from control

blood sample

ESI-MS spectrum obtained,

Multiple charged ions observed for the different chains

MS/MS measurement performed of the corresponding wild-type and variant tryptic peptides – Position of the mutation determined MS spectrum is deconvoluted onto true

mass scale with MaxEnt - Presence of any additional peak can be observed

mass 14600 14800 15000 15200 15400 15600 15800 16000 16200 16400 16600 % 0 100 15126.38 15867.24 15924.32 α-chain β-chain δ-chain α [M+15H]15+ β [M+15H]15+ α [M+14H]14+ β [M+14H]14+ α [M+13H]13+ β [M+16H]16+ α [M+16H]16+ β [M+17H]17+ α [M+17H]17+ β [M+18H]18+ α [M+18H]18+ α [M+19H]19+ α [M+20H]20+ Data Analysis Sample Preparation

for Intact Globin Chain Analysis 0.5 μL of blood diluted in 250 μL 50 % ACN 0.2% HCOOH 1 μL of blood diluted in 50 μL water and 10 μL 50% ACN 0.5 % HCOOH , and

digested with trypsin

(5mg/ml) Different chains are quantified based on their relative intensities MS Acquisition MS Acquisition

77 MS/MS strategy for MRM method development for clinically significant variants and HbA2 quantitation

UPLC separation for initial MRM-based method (MRM 1)

The MRM approach requires peptide separation prior to mass spectrometric detection. This was performed using an Acquity UPLC system (Waters Corporation, Manchester, UK) coupled to the TQD instrument.

2 µL of diluted peptide mixture was loaded to the Acquity UPLC BEH C18 column (1.7µm, 2.1 × 50mm) equilibrated with 95% solvent A at a flow rate of 0.8 ml/min. The composition of solvent A was 0.3% formic acid in water, solvent B was 0.3% formic acid in acetonitrile. The peptides were resolved by increasing the organic solvent concentration from 5% to 13.5% in the first 0.5 minute, then with a quasi- isocratic step further increased to 13.7% in the following 4 minutes, to 23% in another 0.5 min, and finally up to 32% by 6 minute of the run. In the last two minutes the organic concentration was increased to 99% to elute all the remaining but not monitored components of the mixture and the column was re-equilibrated with 5% B at a constant temperature of 40ºC.

Mass spectrometry method

Samples were tryptically digested to produce the necessary peptides. Preliminary measurements were necessary to choose the unique peptides to be monitored. MS/MS experiments on the chosen peptides were performed to determine the optimal parameters to obtain the highest intensity peaks from MRM measurements for specific fragmentation routes. Figure 2.4 shows possible unique peptides based on sequence differences for the relative quantitation of the HbA/HbA2 ratios. For

clinically significant variant detection and quantification, the target peptides were chosen based on the location of the mutation relevant to the specific variant, as detailed in Table 2.2. For sickle cell hemoglobin the first tryptic peptide of the beta chain was relevant.

78 Figure 2.4. Differences in amino acid sequences between beta and delta globin

chains

Name

Position

(Tryptic peptide affected by mutation)

Mutation Mass Change of the β

chain (Da)

Hb S β6 (βT1) Glu→Val -30

Hb C β6 (βT1 new peptide) Glu→Lys -1

Hb D-Punjab (or Hb D- Los Angeles)

β121 (βT13)

Glu→Gln -1

Hb E β26 (βT3 new peptide) Glu→Lys -1

Hb O-Arab β121 (βT13 new peptide) Glu→Lys -1

Table 2.2. Tryptic peptides affected in clinically significant variant chains and the mass shift associated with the mutations.

In order to develop an MRM-based screening method for relevant hemoglobin variants, suitable transitions needed to be determined. As the transitions are based on fragmentation processes from a precursor ion to a product ion, several MS/MS experiments had to be carried out. The peptide ion of interest must be fragmented in the collision cell and product ion spectrum acquired. During these product ion scan measurements collision energy, collision gas flow, and solvent conditions can be optimized.

Optimization was predominantly carried out on commercially available samples. Lyophilized HbA, HbA2 and HbS were used. The solid Hb samples were diluted in

water to prepare stock solution, and then an appropriate amount of the stock solution was digested by trypsin. When using a sequence specific protease such as trypsin, the series of peptides produced can be predicted based on the sequence of the protein. As the clinically significant variant mutations affect the T1, T3 and T13 tryptic peptides of the beta chain (Table 2.2), and the beta and delta chains differ in the T2, T3, T5, T10, T12, T13 peptides of the beta chain (as shown in Figure 2.4), suitable MRM transitions need to be developed for the peptide ions listed in Table 2.3. As the specific peptide to be selected is known, it is possible to predict the mass- to-charge ratio of the ionised species to be selected by mass spectrometry. To investigate which mass-to-charge values of the predicted ions can be detected, tryptic

VHLTPEEKSAVTALWGKVNVD EVGGEALGRLLVVYPWTQRFFESFGD LS TPDAVMGNPKVKAH GKKVLGAFSD G VHLTPEEKTAVNALWGKVN VD A VGGEA LGRLLVVYPWTQRFFESFGDLSS PD A VM GN PKVKA H GKKVLGAFS DG

LAHLDNLKGTFATLS ELHCDKLHVDPENFRLLGNVLVCVLAHHFGKEF TPP V Q AAYQKVVAGVANALAHKYH LAHLDNLKGTFSQLSELHCDKLHVDPENFRLLGNVLVCV LARNFGKEFTPQMQAAYQKVVAGVANALAHKYH 75 75 146 146 δ 76 δ 1 β 76 β 1 T2 T3 T5 T10 T12 T13

79 digested peptide mixtures were measured, looking for the selected ion peaks. These can be singly or multiply charged peptides. Fragmentation of doubly charged ion species typically require lower collision energies, and these ions are favoured. Doubly charged peptides are characteristic of tryptic peptides. One charge is distributed on lysine/arginine and other charge is on an amino group; occasional additional charges are usually taken up by histidine or happen because of missed cleavages.

50% ACN and 0.2% formic acid was initially used to dilute tryptically digested lyophilized blood samples, increasing the acid content of the solvent up to 0.5% when necessary. Not all targeted peptides were observable in the spectra, but for those with high intensity, tandem MS measurements were performed, and product ion spectra were obtained. The solvent composition was altered from 0.1% formic to higher formic acid content (0.3% and later 0.5%) to increase chances of generating ions for the targeted peptides, to produce doubly charged ions for the peptides of interest, and to influence their retention times facilitating their elution at a later time. The optimal transitions for MRM method were then selected for evaluation using clinical samples.

80

Tryptic peptide

Mutation

correlated Amino acid sequence [M+H]

+ /Da [M+2H] 2+ / Da Wild-type βT1 Hb Sickle, HbC VHLTPEEK 952.51 476.76 HbS βT1 - VHLTPVEK 922.53 461.77 HbC βT1 - VHLTPK EK 694.43 276.16 347.72 138.58 Wild-type βT2

Differs from the

δT2 SAVTALWGK 932.5

Wild-type δT2

Differs from the

βT2 TAVNALWGK 959.5 Wild-type βT3 Hb E, Differs from the δT3 VNVDEVGGEALGR 1314.67 657.84 HbE βT3 - VNVDEVGGK ALGR 916.47 416.26 458.74 208.64 Wild-type δT3

Differs from the

βT3 VNVDAVGGEALGR 628.9

Wild-type βT5

Differs from the

δT5 FFESFGDLSTPDAVMGNPK 1030

Wild-type δT5

Differs from the

βT5 FFESFGDLSSPDAVMGNPK 1023

Wild-type βT13

Hb D-Punjab, Hb O-Arab Differs from the

δT14 EFTPPVQAAYQK 1378.70 689.85 HbO βT13 - K FTPPVQAAYQK 147.11 1249.66 74.06 625.33 HbD βT13 - QFTPPVQAAYQK 1377.72 689.36

Table 2.3. List of tryptic peptides and variant tryptic peptides and expected mass to charge ratios. (Corresponding peptides are clustered based on what part of the amino acid sequence they comprise, first the wild-type peptide present in normal adult hemoglobin)