Asparagine Deamidation and the Role of Higher Order Protein

Structure
Jenny Rivers, Lucy McDonald, Ian J. Edwards, and Robert J. Beynon*

Proteomics and Functional Genomics Group, Faculty of Veterinary Science, University of Liverpool, Crown
Street, Liverpool L69 7ZJ, United Kingdom

Received July 11, 2007

The ‘protein world’ exhibits additional complexity caused by post-translational modifications. One such
process is nonenzymic deamidation of asparagine which is controlled partly by primary sequence, but
also higher order protein structure. We have studied the deamidation of an N-terminal peptide in muscle
glyceraldehyde 3-phosphate dehydrogenase to relate three-dimensional structure, proteolysis, and
deamidation. This work has significant consequences for identification of proteins using peptide mass
fingerprinting.

Keywords: Deamidation • proteolysis • protein structure • asparagine • aspartic acid • peptide mass
fingerprinting

Introduction studies of model peptides, the highest rate of deamidation is

obtained when the carboxyl neighbor is glycine, yielding a half-
The emergence of new analytical methods for protein time for deamidation of around 24 h.6 This is probably because
characterization has led to the recognition that there is an the lack of a Cβ atom minimizes steric hindrance and permits
additional dimension of complexity in the protein world created ready formation of the five-membered imide conducive to the
by a wide range of post-translational modifications. Some of deamidation reaction. The N-terminal neighbor has a minor
these modifications are specific and are part of the obligatory effect on the rate of deamidation.6 While most of our under-
maturation process of a protein, such as the removal of standing of rates of peptide deamidation has derived from
propeptides. Other changes are transient, reversible, and may short, model peptides, the same sequences, when incorporated
only operate on a subset of molecules in the protein pool (the into protein structures, might acquire a relatively immobile
best understood is phosphorylation). Other irreversible changes, backbone trajectory that could constrain the sequence to either
such as deamidation or lysine aldehyde mediated cross-linking, favor or disfavor deamidation.
are nonenzymic, and the longevity of the protein may be
The resolution of modern mass spectrometers used routinely
reflected in the accumulation of such changes.
in proteomic analyses permits ready resolution of the monoiso-
Deamidation of the side chain of asparagine residues is a topic peptide–ion from the 13C isotopomer variants, even at
nonenzymic process1 (www.deamidation.org). The conversion charge states of +2 or +3. At this level of resolution, a
of asparagine to aspartic acid or isoaspartic acid elicits a local deamidation event (Asn f Asp) would be readily recognized,
change in charge, and has the potential to impose a self-timer as it elicits a mass shift of +0.985 Da (-NH2 ) 16.03 to -OH )
on protein molecules, altering activity or stability with lifetime 17.01). In circumstances where a peptide exists as a mixture
kinetics.2–5 The ability to include a nonenzymic irreversible of the amide and cognate acid species, a complex mass
change into a protein that elicits a small steric change but a spectrum would ensue that appears as an atypical isotopomer
substantial local alteration in electrostatic potential could distribution for a peptide of that mass. It follows that partial
provide an opportunity to evolve a programmable irreversible deamidation events should be readily observed by examination
change of state into a protein. Most studies on asparagine of the atypical profile, particularly without prior chromato-
deamidation have been conducted with model peptides6 which
graphic separation that would resolve the amide and cognate
are essentially devoid of higher order structure and which
acid in chromatographic space.
permit the peptide backbone and side chain to adopt a
In the course of proteomics studies of soluble proteins in
conformation compatible with the cyclic intermediate that is
skeletal muscle,7 we observed that a peptide from one protein
required for this reaction to take place. Since the flexibility and
in particular exhibited a noticeable and atypical natural isotope
conformational freedom of the peptide is modified by the
distribution profile, consistent with a mixture of an asparagine-
nature of the amino acids, the rate of deamidation of model
containing peptide and the cognate deamidation product. This
peptides is strongly influenced by the flanking residues6 and
peptide was derived from the N-terminus of an abundant
the primary influence on the rate of asparagine deamidation
protein, glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
is the amino acid C-terminal to the asparagine residue. From
We present here a comprehensive analysis that confirms that
* To whom correspondence should be addressed. Phone: +44 151 794 the ‘atypical’ isotope profile is in fact attributable to partial
4312. Fax:+44 151 794 4243. E-mail: r.beynon@liv.ac.uk. deamidation of an asparagine residue. Deamidation of -AsnGly-

10.1021/pr070425l CCC: $40.75  2008 American Chemical Society Journal of Proteome Research 2008, 7, 921–927 921
Published on Web 02/05/2008
research articles Rivers et al.

Figure 1. Atypical peptide mass spectrum consistent with deamidation. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 1 mg/mL
diluted to 0.2 mg/mL with 50 mM ammonium bicarbonate) purified from rabbit skeletal muscle (Sigma, Dorset, U.K.) was digested in solution
with trypsin at a substrate/protease ratio of 100:1 by weight, and the masses of the resultant tryptic peptides were assessed by MALDI-ToF
mass spectrometry; a coverage map is included at the top of the figure, with identified peptides indicated by a shaded block and those
identified as part of a missed cleavage by an open block. The spectrum of a typical partial cleavage tryptic peptide (T10–11, m/z 1615.9) was
compared with the mass spectrum predicted by the MS-Isotope tool (http://prospector.ucsf.edu/). This behavior, common to almost all other
peptides, emphasized the atypical profile observed for the N-terminal partial cleavage peptide (T1–2, m/z 1032.6).

sequences occurs during sample preparation in proteomics,8 were then washed in 50 mM ammonium bicarbonate, which
and proteolysis conducted at lower pH and temperature will was subsequently discarded. The gel was dehydrated using 5
minimize artifactual deamidation.9 Here, we show that dea- µL of ACN, and incubation at 37 °C was resumed for 30 min.
midation is constrained by higher order structure and is Once dry, the gel was rehydrated in 50 mM ammonium
enhanced after release of that conformational restraint by bicarbonate (9 µL) containing trypsin (1µL of 100 ng/µL trypsin
proteolysis. This observation has significance for the identifica- stock reconstituted in 50 mM acetic acid), and digestion was
tion of deamidation events by protein or peptide mass allowed to continue overnight at 37 °C; the digestion was halted
spectrometry10–12 and reinforces the role that protein confor- by the addition of 2 µL of formic acid.
mation can play in this process. MALDI-ToF Mass Spectrometry. Peptides were analyzed by
MALDI-ToF (M@LDI; Waters, Manchester, U.K.) mass spec-
Experimental Section
trometry. For this, 1 µL of digested material was mixed with
Materials and Reagents. Trypsin (sequence grade) was an equal volume of R-cyano-hydroxycinnamic acid in 50% (v/
obtained from Roche Diagnostics (Lewes, U.K.). All other v) ACN and 0.1% (v/v) trifluoroacetic acid. This was allowed
chemicals and solvents (HPLC grade) were purchased from to dry, and peptides were acquired over the range 900–3000
Sigma-Aldrich Company Ltd. (Dorset, U.K.) and VWR Interna-
m/z. For each combined spectrum, 20–30 spectra were ac-
tional Laboratory Supplies (Leicestershire, U.K.).
quired (laser energy typically 30%) with 10 shots per spectrum
One-Dimensional Gel Electrophoresis (1DGE). Purified
and a laser firing rate of 5 Hz. Data were processed using
GAPDH from rabbit skeletal muscle (Sigma, Dorset, U.K.) (10
MassLynx software to subtract background noise using poly-
µg) was electrophoresed through a 12.5% polyacrylamide gel
nomial order 10 with 40% of the data points below this
and visualized with Biosafe Coomassie Brilliant Blue stain (Bio-
Rad, Hemel Hempstead, U.K.). Gels were destained with a 10% polynomial and a tolerance of 0.01. Spectral data were also
acetic acid 10% methanol solution. smoothed by performing two mean smooth operations with a
In-Gel Trypsin Digestion. Gel plugs containing GAPDH window of three channels. To confirm the assumption that both
(identification confirmed by MALDI-ToF MS, results not shown) acid and amide forms of the peptide ionize with equal signal
were excised from 1D gels using a glass pipet and transferred response in MALDI-ToF MS, the synthetic peptide for the
to an Eppendorf tube. To each tube, 25 µL of 50 mM am- amide form was allowed to fully deamidate (by incubation at
monium bicarbonate, pH 8.2, and 50% (v/v) acetonitrile (ACN) 37 °C) and mixed in a known ratio with asparagine-containing
was added and incubated at 37 °C for 20 min. This process peptide in a strong acidic solution to prevent further deami-
was repeated until all of the stain had been removed. The plugs dation. The signal response from the two variants was identical.

922 Journal of Proteome Research • Vol. 7, No. 3, 2008

Asparagine Deamidation/Role of Higher Order Protein Structure research articles

Figure 2. Esterification of acidic residues in the N-terminal peptide of GAPDH. Tryptic peptides recovered from an in-gel tryptic digest
of GAPDH (purified from rabbit skeletal muscle, Sigma, Dorset, U.K.) were reacted with acetyl chloride and methanol to convert acidic
residues to their corresponding methyl esters. The upper mass spectrum is the peptide resulting from partial deamidation of Asn6,
thus, is a mixture of two forms (asparagine containing and aspartic acid containing). The lower spectrum, obtained after esterification
has resolved the peptide into two distinct reaction products at 1046.63 m/z and 1061.65 m/z, consistent with the addition of one and
two methyl groups (+14.03 Da), respectively.

In-Solution Tryptic Digestion. Soluble protein (purified deconvolution tool, implemented as computer hardware in a
GAPDH; 1 mg/mL) was diluted 10-fold with 50 mM ammonium field programmable gate array.13
bicarbonate prior to addition of trypsin (100:1 substrate/ Absolute Quantification of Proteolysis Using a Stable
protease). The reaction mixture was incubated at 37 °C for 24 h, Isotope-Labeled Synthetic Peptide. The N-terminal peptide of
and peptides were analyzed by MALDI-ToF MS. GAPDH, of sequence VKVGVNGFGR and neutral mass 1041.59
Esterification of Peptides. A stock solution of methanol (1 Da, was synthesized by Sigma-Genosys (Dorset, U.K.) and was
mL, previously stored at -20 °C for 15 min) and acetyl chloride labeled at the arginine residue with both [13C6] and [15N4] giving
(150 µL) was prepared. An aliquot (10 µL) of this mixture was a 10 Da mass offset from the analyte peptide. For quantification
then added to a dried portion of the peptide pool recovered of proteolysis, the synthetic peptide was added to digested
after in-gel digestion of the protein. The mixture was incubated material in 10% (v/v) formic acid to stop digestion and deami-
dation. Peptides were analyzed by MALDI-ToF MS, and the
at room temperature for 45 min prior to drying in a vacuum
relative intensities of analyte peptide and internal standard
centrifuge. Esterified peptides were analyzed by MALDI-ToF
were used to quantify the amount of peptide released from the
MS.
protein during incubation with trypsin at 37 °C. As conversion
Monitored Proteolysis of GAPDH. Digestion reaction mix-
of asparagine to aspartic acid alters the isotope envelope of
tures with trypsin were stopped at selected time points after
the analyte peptide, the composite abundance of the entire
addition of enzyme by removing 10 µL and adding to an equal isotopic envelope for both analyte and internal standard
volume of 10% (v/v) formic acid. The fractions were subse- peptide was summed in each case. These data permitted the
quently stored at -20 °C until the end of the time course. kinetics of proteolytic release of the N-terminal peptide from
Peptides were analyzed by MALDI-ToF MS. GAPDH to be calculated and were used along with the kinetics
Data Processing. The natural isotope profile for the acid of deamidation to investigate the interaction between these two
VKVGVDGFGR and amide VKVGVNGFGR variants of the alternative processes. The rate of deamidation was measured
GAPDH N-terminal peptide were predicted using the MSIso- across the time course of digestion by calculating the propor-
tope tool provided online within the Protein Prospector Package tion of acid and amide variants of the peptide at each time
(http://prospector.ucsf.edu/ucsfhtml4.0/msiso.htm). The in- point. This was done during proteolysis of GAPDH and for the
tensities of each isotopomer peak were added, and the com- synthetic peptide, at different temperatures.
bined theoretical spectrum was compared with the intensities
derived from the experimental mass spectrum. The sum of the Results and Discussion
squares of the deviation between predicted and experimental One of the most abundant soluble sarcoplasmic proteins in
data was used to generate the object function, and the sole skeletal muscle is glyceraldehyde 3-phosphate dehydrogenase,
parameter (PA) was the proportion of the acidic component amounting to 11 ( 1% (mean ( SEM, n ) 3) of soluble protein
(by definition, equal to 1 - PN, where PN is the proportion of when resolved by 1D gel electrophoresis (1DGE) and analyzed
amide). The nonlinear optimization function (Solver) within by densitometry (data not shown) and up to 500 ( 50 nmol/g
Excel was used to obtain the best fit value of PA. Additionally, (mean ( SEM, n ) 4) tissue when analyzed using the QconCAT
some samples were analyzed by a high speed spectrum method for absolute quantification.13–15 MALDI-ToF spectra

Journal of Proteome Research • Vol. 7, No. 3, 2008 923

research articles Rivers et al.

Figure 4. Model of proteolysis and deamidation of the N-terminal

peptide of GAPDH. The simultaneous processes of proteolysis
and release of the N-terminal peptide of GAPDH followed by
deamidation of the asparagine residue to aspartic acid were
modelled according to this scheme. The model also included the
subsequent proteolysis of the N-terminal peptide (VKVGVNGFGR
or VKVGVDGFGR) at the internal arginine residue to generate a
dipeptide and a truncated peptide (VK+VGVNGFGR or VK+
VGVDGFGR). In this scheme, we assumed that the rate of
deamidation was the same, whether in the full length or
truncated N-terminal peptide, and that the rate of removal of the
N-terminal dipeptide was independent of the amide/acid variants.
Figure 3. Time course of deamidation of the N-terminal peptide
of GAPDH. Purified rabbit skeletal muscle GAPDH (Sigma, Dorset,
U.K.; 1 mg/mL diluted to 0.2 mg/mL with 50 mM ammonium
1032.58 and a second at a monoisotopic m/z of 1033.58. The
bicarbonate) was digested with trypsin (trypsin/protein 1:100)
higher m/z peptide could have been a contaminant or it could
over 24 h at 37 °C. Proteolysis was stopped at 0, 2, 5, 10, 30, 60,
120, 240, 480, and 1440 min by mixing 10 µL from the digestion have been generated from the peptide at m/z 1032.58. In the
mixture with 10 µL of 10% (v/v) formic acid. The resulting latter case, the most probable explanation for the mass increase
peptides were analysed by MALDI-ToF mass spectrometry, and was deamidation of the asparagine residue, which, by conver-
deamidation was monitored during proteolysis for the N-terminal sion to an aspartate residue, would increase the mass by 0.985
peptide of sequence VKVGVNGFGR at 1032.59 m/z. The propor- Da (-NH2 to -OH). To prove that the atypical profile was a
tion of acid and amide variants was assessed as described in consequence of deamidation, we esterified the peptide mixture
Experimental Section, from peak height data, and plotted as a to convert carboxyl groups to their methyl esters. The mass
function of time (closed squares). Peptide envelopes illustrating shift on esterification would be 14.03 Da. Because the peptide
the conversion of acid to amide form in MALDI-ToF mass spectra
V2KVGVNGFGR10 would possess a single carboxyl group in the
corresponding to time points over 24 h are inserted above the
amide form (the alpha carboxyl group), and two in the acid
data. To compare this with model peptide studies, the N-terminal
peptide of GAPDH, of sequence VKVGVNGFGR and mass 1041.59 form, esterification should therefore deconvolute the atypical
Da, was synthesised by Sigma-Genosys (Dorset, U.K.) and was peptide into two products, one esterified at a single position
labelled at the arginine residue with both [13C6] and [15N4] giving (+14.03 Da), and a second modified in two positions (+28.06
a 10 Da mass offset from the analyte peptide. This peptide was Da). When the peptide mixture was analyzed after esterifica-
incubated in 50 mM ammonium bicarbonate at 37 °C, and a tion, the MALDI-ToF ions in the 1032-1036 m/z region
sample of the peptide was added to an equal volume of 10% disappeared, and two new ions appeared, one representing the
(v/v) formic acid at selected time points. The relative amounts single modified amide (m/z 1032.58 + 14.03 ) 1046.61) and
of acid and amide variants of the peptide were measured using the second reflecting the double modified acid (m/z 1033.58
MALDI-ToF MS, and this was used to calculate the rate of
+ 28.06 ) 1061.64; Figure 2).
deamidation. These data are presented as open circles. The solid
From this analysis, it was not possible to assess whether the
lines are the trajectories taken by first-order decay for the
synthetic peptide and the proteolyzed glyceraldehyde 3-phos- residue had deamidated in vivo or was an artifact of sample
phate dehydrogenase. preparation and processing. To assess the extent of deamida-
tion of this peptide in the native protein, we treated purified
for this protein, isolated by 1DGE and digested with trypsin rabbit GAPDH with trypsin and monitored the proteolysis and
prior to MS analysis are of high quality, give very high the partition between the acid and amide variants of the
probability identification of this protein (not shown), and yield peptide in MALDI-ToF mass spectra (Figure 3; the same
approximately 20 peptides, ranging from 805.5 m/z to 2265.4 experiments were repeated for an in-solution tryptic digest of
m/z. Close inspection of each peptide indicated that for most, chicken skeletal muscle soluble proteins and the same behavior
the observed mass isotopomer distribution was as expected, was apparent, results not shown). The N-terminal peptide of
and was in close agreement to the distribution predicted by GAPDH (VKVGVNGFGR) was released within a few minutes
the MsIsotope program (http://prospector.ucsf.edu/). One pep- and was readily detected as the first analyte ion to appear in
tide in particular (VKVGVNGFGR, [M+H]+ 1032.58 m/z) was the MALDI-ToF spectrum. In the early stages of digestion, the
notably different from the others, inasmuch as the isotope mass spectrum of this peptide was entirely consistent with it
distribution profile was far removed from the predicted profile being exclusively in the amide form. However, as time pro-
(Figure 1). In particular, the relative intensity of the monoiso- gressed during proteolysis, the mass spectrum of the peptide
topic ion was diminished, and of lower intensity than the first showed that the peptide was converted to a mixture of the
[13C] isotopomer, a relative intensity pattern that is unexpected amide and acid variants, and after 10 h of digestion, the peptide
for a peptide of mass 1031.58 Da, given an empirical formula was over 80% in the acid form. The first-order rate constant
of C46H78N15O12. for this process was approximately 0.0017 min-1, which was
The mass isotopomer envelope is consistent with the analyte higher than the value derived from model peptides; for the
being a mixture of two peptides, one of monoisotopic m/z sequence NH2GVNGGOH, the first-order rate constant was

924 Journal of Proteome Research • Vol. 7, No. 3, 2008

Asparagine Deamidation/Role of Higher Order Protein Structure research articles
across acid or amide forms) decreased slowly as digestion
continued. We created a model (Figure 4) that took into
account the sequential first-order processes of proteolysis (k1)
of the native protein (Nnative) to release the amide form of the
peptide (VKVGVNGFGR) followed by deamidation (k2) to
generate the acid form (VKVGVDGFGR).
k1 k2
Nnative 98 VKVGVNGFGR 98 VKVGVDGFGR

Furthermore, the model also included a secondary process

of proteolysis of the released peptide in either the acid or amide
form to release the ValLys dipeptide. The rate of appearance
of the deamidated peptide is given by
k3
VKVGV(N/D)GFGR 98 VK + VGV(N/D)GFGR

We assumed that the rate of deamidation (k2) was indepen-

Figure 5. Absolute quantification of proteolysis of the GAPDH dent of the N-terminal ValLys dipeptide and that the rate of
N-terminus. Purified rabbit skeletal muscle GAPDH (Sigma, tryptic removal of the N-terminal dipeptide (k1) was the same,
Dorset, U.K.; 1 mg/mL diluted to 0.2 mg/mL with 50 mM irrespective of whether the peptide was in acid or amide form.
ammonium bicarbonate) was digested with trypsin (trypsin/ The change in amount (relative to the initial amount of pro-
protein 1:10) over 24 h at 37 °C. The N-terminal peptide of tein, Nnative(t)0)) of the larger peptides (VKVGVNGFGR +
GAPDH, of sequence VKVGVNGFGR and mass 1041.59 Da, was VKVGVDGFGR, N + D) as a function of time, is given by
synthesised by Sigma-Genosys (Dorset, U.K.) and was labelled
at the arginine residue with both [13C6] and [15N4] giving a 10 Da
mass offset from the analyte peptide. For quantification of
proteolysis, the synthetic peptide was added to digested material
N + D ) Nnative ( k1
k3 - k1 )
(e-k1t + e-k3t) (1)

in 10% (v/v) formic acid to stop digestion at selected time points. As part of the same process, the shortened peptide (VGVNGFGR
Peptides were analyzed by MALDI-ToF MS, and the relative + VGVDGFGR, N′ + D′) appears according to

( )
intensities of analyte peptide and internal standard were used
to quantify the amount of peptide released from the protein k3 k3
N ′ + D ′ ) Nnative 1 - e-k1t + e-k3t (2)
during incubation with trypsin at 37 °C. Both the N-terminal k3 - k1 k3 - k1
peptide (VKVGVNGFGR/VKVGVDGFGR; m/z 1032.59 [M+H]+;
closed triangles) and the shorter peptide produced by further Assuming that the rate of tryptic cleavage is consistent for
proteolysis (VGVNGFGR/VGVDGFGR; m/z 805.59 [M+H]+; open both acid and amide variants, from these equations, we were
circles) were monitored. As conversion of asparagine to aspartic able to calculate the second-order rate constants (first-order
acid alters the isotope envelope of the analyte peptide, the rate constant divided by protease concentration) for initial
composite abundance of the entire isotopic envelope for both release of the large peptide (k1) and the rate of proteolysis of
analyte and internal standard peptide was summed in each case. this large peptide (k3) (Figure 5). The value of k1 was estimated
The solid lines reflect the fitted curves for the transient appear- to be 1.22 ( 0.025 min-1 · µM and for k3, 0.50 ( 0.008 min-1 · µM
ance of the N-terminal peptide (VKVGVNGFGR/VKVGVDGFGR) (trypsin ) 0.2 µM). As expected, the endoproteolytic release of
and the truncated product (VGVNGFGR/VGVDGFGR), modelled
the longer peptide is faster than the release of the N-terminal
and fitted as sequential first-order reactions (see text).
dipeptide, as trypsin is known to act poorly as a dipeptidyl
peptidase. However, the release of the longer peptide is likely
previously measured at 0.0004 min-1.16 However, the buffer to be suppressed by the three-dimensional structure of the
conditions for the two experiments are not identical, and pH protein.
has a large affect on deamidation rate. The rate of deamidation To investigate the effects of the higher order structure of
under these buffer conditions was confirmed using a synthetic GAPDH on proteolysis and subsequent deamidation, we ana-
peptide of the same sequence; for this peptide, the rate of lyzed the X-ray crystal structure of rabbit GAPDH (PDB code
deamidation was 0.0023 min-1. The higher rate of deamidation 1J0X.PDB). First, we used the tool NickPred,18 which although
of the synthetic peptide might reflect an association between designed to predict sites of proteolytic attack, can generate a
the partially digested protein and the N-terminal peptide which comprehensive analysis of the environment of every residue
introduced a degree of conformational ‘freezing’ of the peptide, in a protein sequence. The N-terminal region of GAPDH is
diminishing the deamidation rate, but this remains conjecture rather constrained, exhibiting low temperature factors (B-
at present. values) and low protrusion and accessibility (results not shown).
To investigate the kinetics of both deamidation and pro- Close inspection of the structure in the vicinity of Asn6 revealed
teolysis, a synthetic peptide of sequence VKVGVNGFGR, mass this region of the polypeptide chain was folded in an extended
1041.59 Da, was synthesized and was labeled at the arginine β configuration, constrained by 14 hydrogen bonds in a
residue with both [13C6] and [15N4] giving a 10 Da mass offset network that might be expected to constrain main chain
relative to the natural peptide. This peptide, identical to the flexibility and therefore reduce the propensity for asparagine
N-terminal peptide of GAPDH, was used to monitor the be- deamidation (Figure 6). However, once the peptide was re-
havior of the peptide, and for quantification.17 Because the leased by proteolysis, deamidation proceeded at a higher rate
N-terminal peptide itself contains an internal tryptic cleavage than that predicted from model studies. These experiments are
site (VK - VGVDGFGR), the peptide VKVGVDGFGR (summed consistent with the following propositions; that the residue in

Journal of Proteome Research • Vol. 7, No. 3, 2008 925

research articles Rivers et al.

Figure 6. 3D structure of rabbit skeletal muscle GAPDH. X-ray crystal structure of the N-terminal region of rabbit skeletal muscle GAPDH
(PDB code 1J0X) highlighting the Asn6Gly7 deamidation site and the local hydrogen bonded environment. The green dashed lines
denote hydrogen bonds.

contrasted markedly with proteolysis of the native protein,

when the peptide is initially all in the amide form. We attribute
this behavior to the increased conformational flexibility of the
peptide in the heat-treated protein, such that the peptide could
acquire a conformation that allowed deamidation. Further, this
unfolded and flexible component might be expected to be
hypersensitive to proteolysis and to be released first. As the
digestion proceeded, additional peptide in the amide form was
released, and the proportion of amide therefore increased
transiently, until the deamidation reaction dominated the
peptide profile. When the functions derived previously were
used, we obtained a value for deamidation of 0.0023 min-1, in
close agreement with that observed previously. If the heat-
Figure 7. The effect of denaturing protein structure by heating
treated peptide was allowed to incubate at 37 °C for 24 h after
on the rate of deamidation. GAPDH (1 mg/mL diluted to 0.2 mg/
mL with 50 mM ammonium bicarbonate) purified from rabbit the 60 min denaturation period at 60 °C, and then proteolyzed
skeletal muscle (Sigma, Dorset, U.K.) was digested with trypsin with trypsin, the peptide first released was now only 50% in
in solution at a ratio trypsin/protein 1:100 at 37 °C for 24 h. Prior the amide form, consistent with extensive deamidation prior
to digestion, GAPDH was incubated for 1 h at 4 °C (open to proteolysis, consequential to denaturation. Again, as ex-
triangles), 1 h at 60 °C (closed squares), and 1 h at 60 °C followed pected, proteolysis led to the slower release of peptide that was
by 24 h at 37 °C (open squares). For each, deamidation was constrained and unable to deamidate, and there was a transient
monitored over 24 h proteolysis and the proportion of acid and increase in the proportion of amide which again decayed at
amide was calculated from the relative peak intensities of the the same rate as observed previously (k2 ) 0.0024 min-1). The
two ions in MALDI-ToF mass spectra.
behavior of the system was consistent with the GAPDH
the intact protein is exclusively in the amide form, that the preparation being 76% in the amide form, and 26% in a
tryptic fragment containing the amide residue can undergo denatured form that was then rapidly proteolyzed to generate
deamidation, and that deamidation is not an artifact of the the free acid form of the peptide. The effect of denaturation
mass spectrometric analysis. Excision of the peptide from the on the availability of the N-terminal peptide of GAPDH for
GAPDH structure relieves the constraint in the peptide back- deamidation is quite striking and defines the importance of
bone trajectory, permitting the deamidation reaction to take monitoring the two processes of proteolysis and deamidation
place. It followed therefore that prior denaturation of the simultaneously, especially as this effect is only observed upon
protein might permit deamidation prior to digestion with loss of higher order structure, and not upon increasing
trypsin. We conducted experiments in which we denatured concentration of protease (results not shown).
GAPDH by heating to 60 °C for 1 h before proteolysis (Figure
Conclusions
7), a denaturation treatment that was not sufficient to cause
the protein to precipitate. Subsequently, when trypsin was Deamidation is recognized as a potential source of micro-
added, the N-terminal peptide was again released rapidly, and heterogeneity in protein structure, and it may play a significant
the proportion of amide and acid variants of the peptide was role as a biological ‘timer’ that is mediated nonenzymatically.1–5
assessed as previously described. Under these circumstances, Although many studies have emphasized the deamidation of
the peptide first released was approximately 80% amide, with short, flexible peptides, protein deamidation can be limited by
a significant proportion of acid form being measurable. This higher order structure and might only occur at the peptide level

926 Journal of Proteome Research • Vol. 7, No. 3, 2008

Asparagine Deamidation/Role of Higher Order Protein Structure research articles
8 (5) Weintraub, S. J.; Manson, S. R. Asparagine deamidation: a regula-
following proteolytic release. The ease with which some
peptides deamidated could then lead to the erroneous inter- tory hourglass. Mech. Ageing Dev. 2004, 125 (4), 255–257.
(6) Robinson, N. E.; Robinson, A. B.; Merrifield, R. B. Mass spectro-
pretation of a deamidation event as occurring in the intact metric evaluation of synthetic peptides as primary structure
protein. Difficulties of measuring deamidation have been models for peptide and protein deamidation. J. Pept. Res. 2001,
discussed,19 and analyses often use electrospray ionization 57 (6), 483–493.
mass spectrometry6 and reversed-phase chromatographic ma- (7) Doherty, M. K.; McLean, L.; Hayter, J. R.; Pratt, J. M.; Robertson,
D. H.; El-Shafei, A.; Gaskell, S. J.; Beynon, R. J. The proteome of
trices8 to resolve acid and amide variants of a peptide, chicken skeletal muscle: changes in soluble protein expression
precluding analysis of complex mixtures. There is also consid- during growth in a layer strain. Proteomics 2004, 4 (7), 2082–2093.
erable scope for MALDI sample ionization, which, when (8) Krokhin, O. V.; Antonovici, M.; Ens, W.; Wilkins, J. A.; Standing,
coupled with a simple esterification reaction, can clearly K. G. Deamidation of -Asn-Gly- sequeneces during sample prepa-
ration for proteomics: consequences for MALDI and HPLC-MALDI
identify and characterize such deamidation variants. We sug- analysis. Anal. Chem. 2006, 78, 6645–6650.
gest that there may be merit in closer examination of the (9) Stroop, S. D. A modified peptide mapping strategy for quantifying
isotope distribution profile of peptide mass fingerprints, to site-specific deamidation by electrospray time-of-flight mass
search for anomalies such as that noticed here. In particular, spectrometry. Rapid Commun. Mass Spectrom. 2007, 21, 830–836.
(10) Jedrzejewski, P. T.; Girod, A.; Tholey, A.; Konig, N.; Thullner, S.;
it is advantageous to monitor deamidation and proteolysis Kinzel, V.; Bossemeyer, D. A conserved deamidation site at Asn 2
simultaneously when characterizing post-translational behavior in the catalytic subunit of mammalian cAMP-dependent protein
of known proteins and peptides. This will also unravel informa- kinase detected by capillary LC-MS and tandem mass spectrom-
tion about the higher order structure of a protein, the influence etry. Protein Sci. 1998, 7 (2), 457–469.
(11) Nilsson, M. R.; Driscoll, M.; Raleigh, D. P. Low levels of asparagine
of which not only on proteolysis but also on subsequent deamidation can have a dramatic effect on aggregation of amy-
modifications to newly accessible regions of the protein, is loidogenic peptides: implications for the study of amyloid forma-
paramount. tion. Protein Sci. 2002, 11 (2), 342–349.
(12) Cournoyer, J. J.; Lin, C.; O’Connor, P. B. Detecting deamidation
Acknowledgment. We are grateful to Dr. Gary Evans, products in proteins by electron capture dissociation. Anal. Chem.
2006, 78, 1264–1271.
Genus plc, for his interest in this work. This work has been
(13) Bogdan, I.; Coca, D.; Rivers, J.; Beynon, R. J. Hardware acceleration
supported by the BBSRC, EPSRC (EP/D013623) and Genus of processing of mass spectrometic data for proteomics. Bioin-
plc. We are grateful to Dr. D. H. Robertson for instrumenta- formatics 2007, 23 (6), 724–731.
tion support. (14) Pratt, J. M.; Simpson, D. M.; Doherty, M. K.; Rivers, J.; Gaskell,
S. J.; Beynon, R. J. Multiplexed absolute quantification for pro-
teomics using concatenated signature peptides encoded by Qcon-
References CAT genes. Nat. Protoc. 2006, 1 (2), 1029–1043.
(1) Robinson, A. B.; Rudd, C. J. Deamidation of glutaminyl and (15) Rivers, J.; Simpson, D. M.; Robertson, D. H. L.; Gaskell, S. J.;
asparaginyl residues in peptides and proteins. Curr. Top. Cell. Beynon, R. J. Absolute multiplexed quantitative analysis of protein
Regul. 1974, 8 (0), 247–295. expression during muscle development using QconCAT. Mol. Cell.
(2) Geiger, T.; Clarke, S. Deamidation, isomerization, and racemization Proteomics 2007, 6, 1416–1427.
at asparaginyl and aspartyl residues in peptides. Succinimide- (16) Robinson, N. E.; Robinson, Z. W.; Robinson, B. R.; Robinson, A. L.;
linked reactions that contribute to protein degradation. J. Biol. Robinson, J. A.; Robinson, M. L.; Robinson, A. B. Structure-
Chem. 1987, 262 (2), 785–794. dependent nonenzymatic deamidation of glutaminyl and aspar-
(3) Friedman, A. R.; Ichhpurani, A. K.; Brown, D. M.; Hillman, R. M.; aginyl pentapeptides. J. Pept. Res. 2004, 63 (5), 426–436.
Krabill, L. F.; Martin, R. A.; Zurcher-Neely, H. A.; Guido, D. M. (17) Kirkpatrick, D. S.; Gerber, S. A.; Gygi, S. P. The absolute quantifica-
Degradation of growth hormone releasing factor analogs in neutral tion strategy: a general procedure for the quantification of proteins
aqueous solution is related to deamidation of asparagine residues. and post-translational modifications. Methods (Duluth) 2005, 35
Replacement of asparagine residues by serine stabilizes. Int. J. Pept. (3), 265–273.
Protein Res. 1991, 37 (1), 14–20. (18) Hubbard, S. J. The structural aspects of limited proteolysis of native
(4) Deverman, B. E.; Cook, B. L.; Manson, S. R.; Niederhoff, R. A.; proteins. Biochim. Biophys. Acta 1998, 1382 (2), 191–206.
Langer, E. M.; Rosova, I.; Kulans, L. A.; Fu, X.; Weinberg, J. S.; (19) Lindner, H.; Helliger, W. Age-dependent deamidation of asparagine
Heinecke, J. W.; Roth, K. A.; Weintraub, S. J. Bcl-xL deamidation residues in proteins. Exp. Gerontol. 2001, 36 (9), 1551–1563.
is a critical switch in the regulation of the response to DNA
damage. [erratum: Cell 2003, 115 (4), 503] Cell 2002, 111 (1), 51-62. PR070425L

Journal of Proteome Research • Vol. 7, No. 3, 2008 927