Deamidation of α-synuclein

Abstract

The rates of deamidation of α-synuclein and single Asn residues in 13 Asn-sequence mutants have been measured for 5 × 10⁻⁵M protein in both the absence and presence of 10⁻²M sodium dodecyl sulfate (SDS). In the course of these experiments, 370 quantitative protein deamidation measurements were performed and 37 deamidation rates were determined by ion cyclotron resonance Fourier transform mass spectrometry, using an improved whole protein isotopic envelope method and a mass defect method with both enzymatic and collision-induced fragmentation. The measured deamidation index of α-synuclein was found to be 0.23 for an overall deamidation half-time of 23 days, without or with SDS micelles, owing primarily to the deamidation of Asn(103) and Asn(122). Deamidation rates of 15 Asn residues in the wild-type and mutant proteins were found to be primary sequence controlled without SDS. However, the presence of SDS micelles slowed the deamidation rates of nine N-terminal region Asn residues, caused by the known three-dimensional structures induced through protein binding to SDS micelles.

Introduction

Proteins involved in human degenerative diseases often form insoluble aggregates and fibrillar structures.¹ Moreover, accumulation of nonfunctional, denatured proteins is a general characteristic of aged cells.² The degree to which postsynthetically altered protein causes, mediates, or is merely symptomatic of disease or aging is unknown.

Insoluble in vivo forms of α-synuclein and its degradation products in the human brain have been of special interest because they are symptomatic of Parkinson's and Alzheimer's diseases and other dementias. Although the protein is largely unstructured in solutions, it adopts specific structures when associated with biological membranes or membrane models, such as sodium dodecyl sulfate (SDS) micelles.^3–7

The membrane-induced structures of α-synuclein have been examined by EPR site-directed spin labeling,³ fluorescence energy-transfer kinetics,⁴ NMR,⁵ limited proteolysis and circular dichroism,⁶ and site-directed fluorescence labeling.⁷ These studies have shown that, when bound to SDS micelles, the protein exhibits four structural regions, including an α-helix, a linker region, a second α-helix, and a disordered region.

Proteins undergo many different in vivo postsynthetic covalent changes in structure, some of which are biologically useful and others are simply degradative. The most ubiquitous of these changes is nonenzymatic deamidation, which occurs at every Asn and Gln residue, with reaction half-times in 0.15M Tris (37.0°C, pH 7.4) ranging from less than 1 day to centuries.^8,9 These half-times are genetically controlled through primary, secondary, tertiary, and quaternary structures. The primary structure-determined components of Asn and Gln deamidation rates have been experimentally measured,^8,10–12 and a means of deriving these rates from theory has also been developed.¹³

Deamidation of Asn and Gln proceeds by both imide and hydrolysis mechanisms, with primary structure-determined half-times varying between about 1 and 20,000 days. These half-times are modulated by secondary, tertiary, and quaternary structures. A semiempirical method has been developed to estimate the three-dimensional structural components of protein deamidation rates and to combine them with the primary-structure components, thus permitting reliable estimation of the individual Asn deamidation half-times of any protein for which the three-dimensional structure is known.¹⁴

There is experimental support⁸ for the proposal^10,15 that nonenzymatic deamidation serves as a genetically adjustable molecular clock for the regulation of biological processes. Nonenzymatic deamidation half-times of Asn and Gln in α-synuclein are of special interest because postsynthetic covalent changes in protein structure could be involved in the formation of the insoluble protein aggregates and fibrils observed in disease pathology.

We have found that the cumulative deamidation half-time of α-synuclein in 0.15M Tris (37.0°C, pH 7.4) with all amide residues combined is 23 days, whether SDS micelles are present or not. More than 5% of α-synuclein molecules acquire at least one additional negative charge after 2 days—an important finding, as it has been shown that 5% deamidation seeds amyloid formation in a fragment of amylin.¹⁶ Moreover, it is known that α-synuclein amyloid fibril formation can be seeded by amyloid from other proteins.¹⁷

In addition to determining the deamidation rates of wild-type amides, we also have examined the effect of α-synuclein structure on deamidation rates, by introducing Asn residues at 13 different positions in the protein through site-directed mutagenesis. When α-synuclein conformational changes are induced by binding to micelles, the deamidation rates change in predictable ways. Here, we demonstrate that deamidation is a powerful probe of three-dimensional structure. This technique, which could be especially useful in complex heterogeneous systems (most especially in vivo), was previously demonstrated with a single, naturally occurring Asn residue in rabbit muscle aldolase.¹⁰

In the course of this work, we developed an improved method for isotopic envelope measurement of protein deamidation. This protein isotopic envelope method, which includes improvements in data accumulation, Fourier transformation, and data analysis, permits quantitative measurement of the extent of deamidation for any protein of known primary structure for which isotopic envelopes for one or more charge states can be reliably measured by high-resolution mass spectrometry (MS). This new method is substantially more convenient and precise than previous procedures.

Results

Deamidation, which is the change from Asn to Asp or Gln to Glu, raises the molecular weight of the deamidating protein molecules by 0.9804 Da. The difference from 1.0000 arises from the different nuclear binding energies of NH₂ vs. OH. Thus, deamidation can be detected and measured quantitatively by observing the shift in the isotopic envelope to higher mass. Alternatively, sufficient MS resolution allows resolution of the deamidated molecules from the undeamidated molecules in the isotopic envelope, which are different in mass by 0.0196 Da. In ordinary circumstances, a 7-T Fourier transform mass spectrometer (FTMS) is capable of reliable quantitative deamidation measurements using the 0.0196 mass defect for peptides of 40 residues or less.

The envelope method has the advantage of making deamidation measurements directly on large molecules. But, when more than one comparable deamidation rate is occurring simultaneously, the overall protein deamidation is difficult to resolve into the individual amide deamidations. The envelope method was developed for peptides¹⁰ and has been extended to proteins by means of high-resolution FTMS.^18–20 The mass defect method has the advantage that direct specific measurement of the deamidated molecules allows greater sensitivity and more accurate measurement of individual amides. However, the disadvantage is that the measurable molecules are smaller. This will diminish as MS resolution is improved. These two methods are illustrated in Figures 1–4 as described below.

Figure 1.

Isotopic envelopes for the +14 charge states of the 0 h (a) and 48 h (b) deamidated AlaAsn(30)Gly mutant of α-synuclein. Deamidation was measured in 0.15M Tris HCl buffer (37°C, pH 7.4).

Figure 4.

Ion cyclotron resonance Fourier transform mass spectrum of the peptide G(84)AGSIAAATGFVKKDQLGKN(103)EEGAPQEGILE(114) derived by enzymatic digestion with GluC of wild-type α-synuclein after incubation for 22 days in 0.15M Tris HCl buffer (pH 7.4, 37°C). The mass-defect-resolved deamidation of Asn(103) is seen in the peaks with mass 0.0196 lower than the ordinary isotope peaks.

As an example of using the envelope method, Figure 1 shows the isotopic envelope of the +14 charge state of the AlaAsn(30)Gly mutant of α-synuclein at (a) the beginning and (b) 48 h into the experiment in 0.15M Tris HCl buffer (pH 7.4, 37°C). The decrease in population of the lower mass isotope peaks and the increase in population of the higher mass isotope peaks is primarily due to the deamidation of Asn(30) and to a lesser extent the other amide residues in the protein.

To illustrate the calculation methodology, Figure 2(a,b) shows the graphs of E_B − T_B vs. T_A − T_B of the isotope peaks for the experimental measurements of the undeamidated and 48-h deamidated +14 charge states of the AlaAsn(30)Gly mutant illustrated in Figure 1. Here, E_B is defined as the experimental peak height of the second of two successive isotope peaks A and B, and T_A and T_B as the corresponding theoretical peak heights. The slope X is the average number of deamidations per molecule. This calculation was performed for all of the well-observed protein charge state isotopic distributions. For the mutant shown in Figures 1–3, there were nine such distributions, +10 through +18. The median of the nine values of fraction deamidated for each of the time points in the deamidation rate experiment was then plotted as a first-order rate curve shown in Figure 3. This method has the advantage that it can be applied to intact proteins and is very accurate for moderate amounts of deamidation. However, the accuracy deteriorates when only minimal deamidation has occurred as the measurement is based on small changes in large peaks. This envelope method gives a combined deamidation rate for all amides in the protein.

Figure 2.

Plotted values for (E_B − T_B) vs. (T_A − T_B) for the +14 charge states of 0 h (a) and 48 h (b) deamidated AlaAsn(30)Gly mutant of α-synuclein shown in Figure 1. The 6.6% deamidation observed at time 0 apparently occurred during synthesis and purification of the mutant. The median values of X for nine such plots were used for each point in Figure 3.

Figure 3.

First-order deamidation rate curve for deamidation of the AlaAsn(30)Gly mutant of α-synuclein in 0.15M Tris HCl buffer at pH 7.4, 37°C, with percentages of deamidation calculated as illustrated in Figures 1 and 2.

It is interesting to note that this same envelope calculation can be performed to determine the relative deamidation percentage of two experimental protein samples, without using the theoretical isotopic distribution. Use of the theoretical distribution reduces the experimental errors, eliminates the need for a second protein sample with a different percentage of deamidation, and eliminates errors from deamidation prior to the experiment.

Figure 4 shows the mass spectrum of peptide G(84)AGSIAAATGFVKKDQLGKN(103)EEGAPQEGILE(114) obtained by enzymatic digestion of wild-type α-synuclein after 22 days in 0.15M Tris (37.0°C, pH 7.4). The molecules deamidated at Asn(103) are seen to be resolved from the ordinary isotope peaks as a result of 0.0196 mass defect. The only other amides in this peptide are Gln(99) and Gln(109), which both have sequence-determined rates under these reaction conditions of about 6000 days. Therefore, they cannot contribute significantly to this mass defect. Since the enzyme is not able to break the peptide bond after Glu(105) subsequent to deamidation of Asn(103), the data shown in Figure 4 required correction for other peptides to obtain the deamidation rate value given in Table I.

Table I.

Amide Half-Times

Mutation	Sequence	t1/2 (days) for median Xxx -XxxAsnYyy-a	t1/2 (days), buffer onlyb	t1/2 (days), 10−2M SDS	SDS to buffer ratioc	t1/2 (days), 10−2M SDS and 10−3M CaCl2	t1/2 (days), 10−2M SDS and 10−2M CaCl2
L8N	GlyAsn(8)Ser	15.6	14	31	2
Q24N	LysAsn(24)Gly	0.84	0.94	15	16
A30N	AlaAsn(30)Gly	0.84	0.78	7.0	9
V40N	TyrAsn(40)Gly	0.84	1.1	10	9
V49N	ValAsn(49)His	10.2	8.7	16	2
V66G	ThrAsn(65)Gly	0.84	0.80	10	13
V66H	ThrAsn(65)His	10.2	8.9	40	4
V66S	ThrAsn(65)Ser	15.6	14	39	3
T72N	ValAsn(72)Gly	0.84	1.0	63	63
L100N	GlnAsn(100)Gly	0.84	1.0	1.2	1	1.2	1.0
Q109N	PheAsn(109)Glu	62.4	68	72	1
I112N, L113G	GlyAsn(112)Gly	0.84	1.1	1.5	1	1.4	1.1
A140NGA	GluAsn(140)Gly	0.84	1.1	1.3	1	1.1	1.3
Wild type	LysGln(24)Gly	660
Wild type	GluGln(62)Val	6600
Wild type	ThrAsn(65)Val	241
Wild type	AlaGln(79)Lys	6100
Wild type	AspGln(99)Leu	6300
Wild type	LysAsn(103)Glu	62.4	81	68	1
Wild type	ProGln(109)Glu	7400
Wild type	AspAsn(122)Glu	62.4	35	47	1
Wild type	TyrGln(134)Asp	7700
α-Synuclein		26.0d	22.6	23.4	1	21.5

^a10⁻³M peptides, 37°C, pH 7.4, 0.15M Tris. Pentapeptide except for AsnGly, which is median of 10 peptides of varying length.⁸

^bSolutions were 5 × 10⁻⁵M protein, 37°C, pH 7.4, 0.15M Tris.

^cRatio of 10⁻²M SDS half-time to buffer only half-time.

^dCalculated for all nine amides in α-synuclein.

Discussion

The results of these experiments are summarized in Table I. The first two columns designate the protein and amide under study, whereas column 3 gives the predicted deamidation half-time for that primary sequence as measured by peptide models.^8,10–12 The half-times listed are for the medians of 18 different Xxx residues.⁸ The measured deamidation half-times for these amides without SDS in all the α-synucleins under study are similar to the calculated sequence controlled half-times.

In buffer alone, all of the samples demonstrated no significant suppression of deamidation rates by three-dimensional structure, which is to be expected for proteins without well-defined three-dimensional structure. For well-structured proteins, most primary-structure deamidation rates are suppressed by three-dimensional structure. Therefore, their deamidation rates are found to be markedly less than their primary-structure rates, except for amides in unstructured regions.^8,14,21

When associated with SDS micelles, α-synuclein becomes helical between residues 3 and 37, followed by a linker region (residues 38–44) and another helix between residues 45 and 92.⁵ The remaining carboxyl region of the protein remains largely disordered.⁵ It has been suggested that the fold in α-synuclein at the linker is caused by the size of SDS micelles, which requires α-synuclein to bend to maximize its binding.

The individual Asn deamidation rates listed in columns 4 through 7 in Table I are the experimental rates, after correction for the 23-day deamidation half-time of wild-type α-synuclein. These results show marked deamidation rate suppression throughout the region from residues 8 to 72 in the presence of SDS micelles. This rate suppression is also observed for Asn(40) in the linker region, which is found by NMR to be well ordered. Interestingly, Asn(8) and Asn(49) show the least suppression because they are near the ends of the helices, where this effect on the deamidation rate has been shown to be reduced.^10,14 Beyond the helical regions, the deamidation measurements show very little three-dimensional rate suppression with SDS, which is in accordance with other α-synuclein studies.^3–7

The slight overall increase in deamidation rate in the four CaCl₂ experiments may be due to increased ionic strength, which accelerates deamidation⁸ by about the observed amount. CaCl₂ is thought to stabilize this region through chelation, but we have found no significant effect on deamidation of Asn(100), Asn(112), or Asn(140) by the introduction of CaCl₂.

The three-dimensional suppression of deamidation in these α-synucleins varies from about twofold to 60-fold depending upon its position. The helical structure reduces deamidation rate through inhibition of imide formation by increased protein backbone stiffness and by the reduced acidity of hydrogen in the peptide bond.⁸ The amount of three-dimensional modulation of deamidation rate observed here for α-synuclein is not unusual. For example, the computed deamidation half-time of Asn(67)Gly in native ribonuclease A is 70 days,¹⁴ whereas the experimental half-time is 67 days by extrapolation.^22,28 On the other hand, the deamidation half-time of Asn(67) of reduced and unfolded ribonuclease A is 1 day.²³ Native folding of ribonuclase A, therefore, increases the deamidation half-time of Asn(67) by about 60-fold. In other cases, three-dimensional modulation of Asn deamidation increases the deamidation half-times by an amount that can be less than twofold or up to greater than 500-fold.⁸

The variations in deamidation modulation by three-dimensional structure observed in the nine Asn mutants are due to differences in their structural circumstances in the α-synuclein-SDS complex, which affect succinimide formation during Asn deamidation.^14,21 On the other hand, the primary structure-determined deamidation of wild-type α-synuclein occurs primarily at Asn(103) and Asn(122). Both of these residues are in the disordered carboxyl region of α-synuclein. No three-dimensional deamidation rate suppression was observed, regardless of whether SDS micelles and/or CaCl₂ were present. The observed deamidation half-times of wild-type α-synuclein in all cases are consistent with the half-time calculated with the assumption of no three-dimensional suppression.

The measurements summarized in Table I were conducted primarily with the envelope method summarized in Figures 1–3, with a few confirmed by collision-induced dissociation (CID) or enzymatic fragmentation followed by mass defect measurements. All CID and enzymatically measured rates were consistent with the envelope rate measurements.

In 0.15M Tris (37.0°C, pH 7.4), 0.05 mM wild-type α-synuclein deamidates with a half-time of 23 days, primarily as a result of the combined deamidations of Asn(103) and Asn(122). Regardless of whether 10 mM SDS and/or 1 mM CaCl₂ are present, this deamidation half-time is not significantly increased. To the extent that these conditions mimic physiological, nonenzymatic postsynthetic in vivo, deamidation would be expected to add at least one negative charge to more than 5% of α-synuclein molecules in about 2 days after synthesis. These postsynthetic changes may be relevant in understanding the aggregation of α-synuclein in vivo.

The deamidation rates of wild-type and 13 mutants of α-synuclein illustrate the use of deamidation as a three-dimensional protein structure probe. In this case, when α-synuclein is bound to SDS micelles, the deamidation rates of Asn residues in known helical regions are markedly reduced. The differences between deamidation rates among the nine Asn residues are due to the differences in α-synuclein secondary and/or tertiary structure. Quaternary structure differences mediated by its associations with SDS could also be a contributing factor to the rate differences.

This work has successfully demonstrated a method by which the percentage of deamidation in any protein can be determined if the primary structure is known and a reliable ICR FTMS high-resolution spectrum of at least one isotopic envelope can be measured. Such envelopes are often measureable by direct FTMS without prior purification in protein mixtures containing as many as 50–100 protein components. This technique can be extended to much more complex mixtures, if purification of protein mixtures is performed.

Originally developed for deamidation measurement of small peptides,¹⁰ the envelope method has been used qualitatively²⁰ and semiquantitatively¹⁸ for proteins. With the improvements reported and illustrated herein, including carefully calculated theoretical isotopic envelopes, improved Fourier transformation, normalization, and use of the new graphical method shown in Figure 2, this method can now be used for high-precision protein deamidation measurements, even when experimental measurements for the undeamidated proteins are unavailable. These improvements permit, for the first time, simple, quick, and reliable quantitative measurement of the extent of deamidation in any protein for which a single isotopic envelope can be observed with high-resolution MS. The 370 quantitative α-synuclein deamidation measurements reported herein would have been very difficult without these improvements. Deamidation measurement by the mass defect and isotopic envelope methods now makes possible a marked increase in the progress of deamidation research.

Materials and Methods

Materials

Protein concentration: QuantiPro™ BCA Assay Kit (Sigma, St. Louis, MO); protein deamidation: SDS ultrapure bioreagent (J.T. Baker, Phillipsburg, NJ), granular calcium chloride dihydrate (J.T. Baker), Riedel-de Haën Water LC-MS Chromasolv, Sigma trizma hydrochloride (>99%; SigmaUltra), Sigma trizma base (>99%; SigmaUltra); protein filtration: Amicon Bioseparations, Microcon Centrifugal, Ultracel YM-10 Filter Devices (Millipore, Billerica, MA), Riedel-de Haën Methanol LC-MS Chromasolv (Riedel-de Haën, Seelze, Germany), formic acid, 98–100% Reagent ACS; protein digestion: LC-MS grade ammonium bicarbonate (Fluka, Sigma Aldrich, St. Louis, MO), Type XVII-B endoproteinase glu-C from Staphylococcus aureus strain V8 (Sigma).

Protein preparation, modification, and characterization

Wild-type human α-synuclein expression plasmid (pRK172) was provided by M. Goedert (Medical Council Research Laboratory of Molecular Biology, Cambridge, UK).²⁴ Twelve mutant α-synucleins (L8N, Q24N, A30N, V40N, V49N, V66G, V66H, V66S, T72N, L100N, Q109N, and I112N-L113G) were prepared. A 13th mutant, A140NGA, was created by introducing A140N and extending the C-terminus by two residues (Gly-Ala). The site-directed mutagenesis reactions were performed by means of a QuikChange kit (Stratagene). All mutations were confirmed by DNA sequencing at the California Institute of Technology, DNA Sequencing Core Facility, and by ion cyclotron resonance FTMS of the purified proteins. The wild-type human α-synuclein amino acid residue sequence is MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA.

Recombinant α-synucleins were expressed and purified following previously published procedures.²⁵ The sequenced plasmids were transformed to BL21(DE3)pLysS competent cells and plated on Luria-Bertini (LB) agar plates with 34 μg/mL of chloramphenicol and 100 μg/mL of ampicillin. Colonies were expressed in LB broth with the same concentrations of antibiotics at 30°C until the absorbance at 600 nm was ∼0.6. Isopropyl β-d-1-thiogalactopyranoside (IPTG) was then added to a final concentration of 0.5 mM; the cells were harvested 6 h after induction and stored at −80°C.

To extract α-synucleins, the harvested cells were boiled and the lysate was subjected to acid precipitation. The supernatant was chromatographed on a Q-Sepharose Fast Flow 16/10 column (Amersham Biosciences) equilibrated with pH 8.0, 20 mM Tris buffer and eluted with a stepwise gradient from 0 to 0.5M NaCl. Fractions containing the α-synucleins were pooled and further purified on a Mono-Q 10/10 column (Amersham Biosciences). Protein concentrations were determined with a QuantiPro BCA Assay Kit (Sigma).

SDS micelle and α-synuclein binding characterization

Circular dichroism spectra were obtained with a Jasco J-815 CD spectrometer. Titration with SDS confirmed at least 95% formation of α-helix, when compared with known CD spectra.⁶ The critical micelle concentration was measured in pure water by conductivity titration and corrected for Tris buffer²⁶ to ascertain formation of SDS micelles. Conductivity measurements were made with a YSI 3200 conductivity instrument, utilizing a YSI 3253, K = 1.0/cm cell.

Protein deamidation and preparation for MS

All deamidation experiments were conducted with 0.05 mM α-synuclein in pH 7.4, 0.15M Tris HCl buffer at 37°C. To some solutions, 10 mM SDS and/or CaCl₂ (1 or 10 mM) were added. The experiments conducted for this work are listed in Table I. Each experiment contained 10 μL of protein solution placed in Thermo Scientific V-bottom 96-well storage plates. Ten plates prepared for various reaction time intervals were incubated at 37°C. At the end of that specific time interval, the plate was flash frozen in liquid nitrogen, and stored at −75°C until the samples were ready to be processed.

Based on preliminary experiments, two time series were adopted for protein incubation. For the mutants with faster deamidation rates, which included Q24N, A30N, V40N, V66G, T72N, L100N, I112N-L113G, and A140NGA in buffer and L100N, I112N-L113G, and A140NGA in SDS or SDS/CaCl₂, the incubation times chosen were 0, 8, 16, 24, 32, 40, 48, 56, 64, and 72 h. For the mutants with slower deamidation rates, time intervals of 0, 4, 8, 12, 18, 22, 26, 30, 34, and 38 days were selected instead. These included buffer controls of wild-type (WT), L8N, V49N, V66S, V66H, and Q109N; SDS or SDS/CaCl₂ experiments for WT; and SDS experiments for L8N, Q24N, A30N, V40N, V49N, V66G, V66S, V66H, T72N, and Q109N.

To prepare the protein solutions for mass spectroscopy, the samples were thawed, 2 μL was pipetted into a YM-10 filter that contained 500 μL of 1 mM Tris (pH 7.4) buffer, and centrifuged at 14,000 rpm in an Eppendorf 5417R centrifuge at 4°C for 35 min. The concentrated protein solution was diluted on the filter and centrifuged. The dilution and centrifugation was done four times, twice with 500 μL of 1 mM Tris (pH 7.4) buffer, and then twice with 500 μL of water. The final concentrated protein solution was then diluted on the filter with 50 μL of 49:1 methanol:formic acid. The filter was inverted over a 1.8-mL Nunc vial that had been rinsed with methanol twice and the concentrated protein solution was transferred to the vial by low-speed centrifugation. This solution was flash frozen in liquid nitrogen and stored at −75°C. Immediately before MS, the frozen sample was thawed. A sample of 20 μL of the protein solution was then diluted 10:1 in the MS injection syringe by addition of 180 μL of 49.5:49.5:1 H₂O:methanol:formic acid before MS injection.

Enzymatic digestion was employed to separately measure the individual rates of LysAsn(103)Glu and AspAsn(122)Glu in WT α-synuclein, which were also determined by CID fragmentation. For enzymatic digestion, the concentrated protein solution was diluted with 10 μL of pH 7, 1M ammonium carbonate buffer and 32 μL of water, instead of methanol:formic acid. After centrifugal recovery in the Nunc vial, 8 μL of 50 units/mL Glu-C enzyme was added, the solution was incubated for 2 h at 37°C, flash frozen in liquid nitrogen, and stored at −75°C until analysis. Immediately before MS analysis, 50 μL of each sample was diluted 4:1 in the MS injection syringe by addition of 50 μL of 49:1 methanol formic acid and 150 μL of 49.5:49.5:1 H₂O:methanol:formic acid before MS injection.

Mass spectrometry

Mass spectrometry was carried out in a Bruker Apex Qh mass spectrometer, with a B-C 70/11 USR 7-T magnet cooled by a Cryomech PT407 refrigerator and PTL407 head. Sample injection was through a Bruker AP2 electrospray source and AP2 ion funnel. Sample flow rates were 200 μL/h. A two mega-sample recording of a 2.7-s transient was made during each scan. A total of 500 scans were recorded and the transients averaged. For the isotopic envelope experiments, 500-scan average transients were Hanning windowed with a sine function and followed by three zero-fills prior to Fourier transformation. The mass defect transients were windowed with a (sine)² function to maximize resolution.

During protein isotopic envelope deamidation measurements, ion accumulation time was 0.5 s, whereas for experiments involving protein digestion or CID, deamidation ion accumulation time was 4 s. In the CID experiments, source skimmer number 1 was set to 80 V when compared with the usual setting of 20 V to fragment the proteins.

Determination of deamidation rates

Deamidation rates were determined by both the isotopic envelope method^10–12 and the mass defect method.^18,19 The envelope method was used for all of the measurements reported in Table I except for the individual deamidation rates of LysAsn(103)Glu and AspAsn(122)Glu in wild-type α-synuclein. These were measured by mass defect after CID and enzymatic breakage, as were a few of the mutant proteins as additional verification of the envelope measurements.

Data processing

To determine the extent of deamidation, the sum of the peak heights in the envelope was first normalized to the sum of the peak heights in the exactly computed envelope for the undeamidated protein. All isotopic envelope spectra were normalized to 1000 as the sum of the peak heights. The computed envelope must be precise, taking into account the exact molecular composition of the protein. For these calculations, the peak heights were used for convenience, neglecting differences in width. Peak areas might give improved results.

The theoretical isotopic envelopes were computed by summing the exact probabilities of occurrence for every possible combination of atoms in the proteins. The contributions at very close mass positions were combined using weighted averages to obtain peak heights and mass positions. These were further combined by replacing the peak at each mass with a Gaussian distribution with resolution of 1 Da. These theoretical distributions were calculated for both the undeamidated and deamidated forms of the protein because these are slightly different. The mass positions and peak heights of these combination spectra were then used for the deamidation calculations.

The theoretical envelopes depend upon the isotopic abundances, which vary with biological or geological source; for a definitive review, see Ref.27. The isotopic abundances chosen were ¹H 0.99986, ²H 0.00014, ¹²C 0.98940, ¹³C 0.01060, ¹⁴N 0.99635, ¹⁵N 0.00365, ¹⁶O 0.99758, ¹⁷O 0.00038, ¹⁸O 0.00204, ³²S 0.9502, ³³S 0.0075, ³⁴S 0.0421, and ³⁶S 0.0002.

To calculate the fraction of deamidated protein (X), the equation (E_B − T_B) = X(T_A − T_B) was employed, where E_B is defined as the experimental peak height of the second of two successive isotope peaks A and B, and T_A and T_B as the computed theoretical peak heights of A and B.