Real-time hydrogen/deuterium exchange kinetics via supercharged electrospray ionization tandem mass spectrometry

Article abstract of DOI:10.1021/ac101957x

Amide hydrogen/deuterium exchange (HDX) rate constants of bovine ubiquitin in an ammonium acetate solution containing 1% of the electrospray ionization (ESI) "supercharging" reagent m-nitrobenzyl alcohol (m-NBA) were obtained using top-down, electron transfer dissociation (ETD) tandem mass spectrometry (MS). The supercharging reagent replaces the acid and temperature "quench" step in the conventional MS approach to HDX experiments by causing rapid protein denaturation to occur in the ESI droplet. The higher charge state ions that are produced with m-NBA are more unfolded, as measured by ion mobility, and result in higher fragmentation efficiency and higher sequence coverage with ETD. Single amino acid resolution was obtained for 44 of 72 exchangeable amide sites, and summed kinetic data were obtained for regions of the protein where adjacent fragment ions were not observed, resulting in an overall spatial resolution of 1.3 residues. Comparison of these results with previous values from NMR indicates that the supercharging reagent does not cause significant structural changes to the protein in the initial ESI solution and that scrambling or back-exchange is minimal. This new method for top-down HDX-MS enables real-time kinetic data measurements under physiological conditions, similar to those obtained using NMR, with comparable spatial resolution and significantly better sensitivity.

Full text of DOI:10.1021/ac101957x

Anal. Chem. 2010, 82, 9050–9057
Real-Time Hydrogen/Deuterium Exchange Kinetics
via Supercharged Electrospray Ionization Tandem
Mass Spectrometry
Harry J. Sterling and Evan R. Williams*
Department of Chemistry, University of California, Berkeley, California 94720-1460, United States
6
-9
Amide hydrogen/deuterium exchange (HDX) rate con-
stants of bovine ubiquitin in an ammonium acetate solu-
tion containing 1% of the electrospray ionization (ESI)
spectroscopies and mass spectrometry (MS).
NMR has the
advantage that exchange rate information about individual amides
is obtained, which is the ultimate spatial resolution possible with
HDX. Mass spectrometry has the advantage of high sensitivity,
and it can be applied to much larger proteins. However, the spatial
resolution with which information about HDX is obtained with
mass spectrometry is often limited to relatively large regions of
the protein defined by peptides formed by proteolysis.
“supercharging” reagent m-nitrobenzyl alcohol (m-NBA)
were obtained using top-down, electron transfer dissocia-
tion (ETD) tandem mass spectrometry (MS). The super-
charging reagent replaces the acid and temperature
“quench” step in the conventional MS approach to HDX
experiments by causing rapid protein denaturation to
occur in the ESI droplet. The higher charge state ions that
are produced with m-NBA are more unfolded, as mea-
sured by ion mobility, and result in higher fragmentation
efficiency and higher sequence coverage with ETD. Single
amino acid resolution was obtained for 44 of 72 exchange-
able amide sites, and summed kinetic data were obtained
for regions of the protein where adjacent fragment ions
were not observed, resulting in an overall spatial resolu-
tion of 1.3 residues. Comparison of these results with
previous values from NMR indicates that the supercharg-
ing reagent does not cause significant structural changes
to the protein in the initial ESI solution and that scram-
bling or back-exchange is minimal. This new method for
top-down HDX-MS enables real-time kinetic data mea-
surements under physiological conditions, similar to
those obtained using NMR, with comparable spatial
resolution and significantly better sensitivity.
HDX coupled to MS is frequently performed with the “bottom-
6
,10,11
up” approach,
down” approach appears promising. In the former, fully proto-
nated or deuterated protein is rapidly diluted into either D O or
O, respectively, allowed to exchange for a fixed period of
although the more recently developed “top-
1
2
2
H
2
time, and then “quenched” by lowering the pH (and temper-
ature) into a range where the intrinsic exchange rate is a
1
minimum (pH ∼ 3.0). The protein is then proteolyzed with
an enzyme that is active in acidic solution, such as pepsin, and
the resulting peptides are separated by liquid chromatography
(
LC) and either mass analyzed directly or separated further
in the gas phase and fragmented by tandem MS methods.
During the proteolysis and LC steps, amides that were
protected during the exchange period become exposed to the
bulk solvent, which can cause some back-exchange and
concomitant loss of HDX information. This effect can be
minimized, though not eliminated completely, with online
pepsin columns, short chromatography methods, and/or back-
9,13,14
exchange correction factors.
A significant advantage of the
bottom-up approach is the theoretically unlimited mass range for
The rates of backbone amide hydrogen/deuterium exchange
HDX) in solution can provide a wealth of information about the
15-17
the analyte, including large multimeric assemblies,
although
(
structures and conformational dynamics of proteins and protein
complexes. There is a substantial reduction of the exchange rates
of different amide sites as a result of either intramolecular
hydrogen bonding or sequestration of the amide hydrogen atoms
from bulk solvent, making these exchange rates a useful probe
of higher-order structure and thermodynamics of local or global
(
4) Barth, A. Biochim. Biophys. Acta 2007, 1767, 1073–1101
.
(5) Shashilov, V. A.; Lednev, I. K. J. Raman Spectrosc. 2009, 40, 1749–1758.
(
(
(
6) Englander, S. W. J. Am. Soc. Mass Spectrom. 2006, 17, 1481–1489
7) Konermann, L.; Tong, X.; Pan, Y. J. Mass Spectrom. 2008, 43, 1021–1036
8) Wales, T. E.; Engen, J. R. Mass Spectrom. Rev. 2006, 25, 158–170
.
.
.
(9) Wang, L. T.; Smith, D. L. Anal. Biochem. 2003, 314, 46–53
10) Engen, J. R.; Smith, D. L. Anal. Chem. 2001, 73, 256A–265A
11) Smith, D. L.; Deng, Y. Z.; Zhang, Z. Q. J. Mass Spectrom. 1997, 32, 135–
46
(12) Kaltashov, I. A.; Bobst, C. E.; Abzalimov, R. R. Anal. Chem. 2009, 81, 7892–
7899
13) Zhang, H. M.; Bou-Assaf, G. M.; Emmett, M. R.; Marshall, A. G. J. Am.
Soc. Mass Spectrom. 2009, 20, 520–524
(14) Feng, L. M.; Orlando, R.; Prestegard, J. H. Anal. Chem. 2006, 78, 6885–
6892
(15) Zencheck, W. D.; Xiao, H.; Nolen, B. J.; Angeletti, R. H.; Pollard, T. D.;
Almo, S. C. J. Mol. Biol. 2009, 390, 414–427
(16) Cheng, G. L.; Basha, E.; Wysocki, V. H.; Vierling, E. J. Biol. Chem. 2008,
283, 26634–26642
(17) Kang, S. B.; Prevelige, P. E. J. Mol. Biol. 2005, 347, 935–948
.
(
(
.
1
2
unfolding events. Since the early studies of Linderstrom-Lang,
measurements of HDX have been made using a variety of different
1
.
3
4
5
.
analytical techniques, including NMR, infrared (IR), and Raman
(
.
*
To whom correspondence should be addressed. Phone: (510) 643-7161. Fax:
(510) 642-7714. E-mail: williams@cchem.berkeley.edu.
.
(
(
(
1) Bai, Y. W.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins: Struct. Funct.
Gen. 1993, 17, 75–86
2) Berger, A.; Linderstrom-Lang, K. Arch. Biochem. Biophys. 1957, 69, 106–
18
3) Dempsey, C. E. Prog. Nucl. Magn. Reson. Spectrosc. 2001, 39, 135–170
.
.
1
.
.
.
.
9050 Analytical Chemistry, Vol. 82, No. 21, November 1, 2010
10.1021/ac101957x  2010 American Chemical Society
Published on Web 10/13/2010

3
2
,33
34
there is an irreversible loss of HDX information for two amides
for each proteolytic cleavage.
coverage and capture efficiency in ECD
experiments.
or ETD
1
8
In the top-down approach, the exchange and quench steps are
the same, but rather than proteolysis and LC-MS or LC-MS/MS,
the intact protein is fragmented in the gas phase. A significant
advantage of the top-down approach is the potential for single
amino acid spatial resolution compared to the traditional bottom-
up approach where the spatial resolution is limited by the length
and overlap of the peptides from proteolysis. Slow heating
methods that are commonly used to fragment gas-phase peptides
and proteins, such as collisional activation, may cause scrambling,
where intramolecular proton or hydrogen atom transfer can lead
to loss of information about where exchange originally occurred
Increased analyte charging in ESI can also be obtained by
adding small amounts of some “supercharging” reagents to the
3
5-37
initial solution from which these ions are formed.
example, ESI of a water/methanol/acetic acid solution containing
% m-nitrobenzyl alcohol (m-NBA) and cytochrome c resulted in
For
1
an increase in the average charge state from 17.3+ to 20.8+ and
the maximum observed charge state from 21+ to 24+ compared
36
to the same solution without m-NBA. Originally used with
“
denaturing” solutions consisting of water, methanol, and acetic
35-37
acid,
m-NBA and other supercharging reagents were shown
by Loo and co-workers to increase the charge states of proteins
19-21
in solution.
In contrast, recent work by a number of research-
38,39
and protein complexes formed from aqueous solutions.
ers has shown that scrambling is largely reduced when intact
proteins or large peptides are fragmented with either electron
4
0
Although many factors, including proton transfer reactivity,
41
37
instrumental conditions, surface tension, etc., can affect charg-
ing in ESI, the charge enhancement observed for proteins and
protein complexes in aqueous solutions containing a supercharg-
ing reagent appears to be predominantly a result of chemical and/
or thermal denaturation that occurs in the electrospray droplet
1
8,22-24
capture dissociation
(ECD) or electron transfer dissociation
2
5-27
(ETD).
The minimal scrambling that occurs with either ECD
or ETD may be due to either nonergodic dissociation, where
cleavage occurs before the recombination energy has time to
2
8,29
dissipate in the molecule,
dissociate before scrambling occurs because of low activation
barriers for dissociation.
or the reduced precursors may
4
2,43
prior to ion formation.
Sulfolane, a supercharging reagent
3
8
3
0,31
found to be effective by Loo et al., destabilizes the native form
of myoglobin by 1.5 kcal/mol/M but has little effect on the native
structure in the initial solution, owing to the typically low
In either case, there is a growing
consensus that HDX coupled to ECD or ETD mass spectrometry
can be a viable approach to obtain accurate amide exchange rate
43
1
8,22-27
concentrations used. Because of the high boiling points of these
information
tion possible.
with close to single amino acid spatial resolu-
supercharging reagents compared to water (bp of m-NBA and
4
4
44
sulfolane are 177 °C at 3 Torr and 287 °C at 760 Torr,
With both the bottom-up and top-down approach, the quench
step affects the results of the experiment in multiple ways. As
described above, the primary effect is to reduce the pH to near
the intrinsic exchange rate minimum, although this rate is still
respectively), the concentration of these supercharging reagents
in the ESI droplets increases as water preferentially evaporates
from the droplet and can cause significant unfolding of the proteins
-3
-1
42,43
fast enough at ∼6 × 10 min at 0 °C (attenuated by the
in the droplet due to chemical and/or thermal denaturation.
1
adjacent side chains and local environment) to cause some
It is well known that the structure of the protein in solution can
significantly affect the charging observed in ESI, where more
back-exchange during any delay between the quench step and
transfer to the gas phase. Decreasing the pH can have the
additional advantage of unfolding the protein and increasing
the charge states of ions formed by electrospray ionization
4
5-47
unfolded structures become more highly charged.
(32) Breuker, K.; Oh, H. B.; Horn, D. M.; Cerda, B. A.; McLafferty, F. W. J. Am.
Chem. Soc. 2002, 124, 6407–6420
33) Iavarone, A. T.; Paech, K.; Williams, E. R. Anal. Chem. 2004, 76, 2231–
238
(34) Syka, J. E. P.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F.
Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 9528–9533
(35) Iavarone, A. T.; Jurchen, J. C.; Williams, E. R. Anal. Chem. 2001, 73, 1455–
1460
.
(
ESI), which can dramatically improve both the sequence
(
2
.
(
(
(
18) Pan, J. X.; Han, J.; Borchers, C. H.; Konermann, L. J. Am. Chem. Soc. 2009,
31, 12801–12808
19) Hoerner, J. K.; Xiao, H.; Dobo, A.; Kaltashov, I. A. J. Am. Chem. Soc. 2004,
26, 7709–7717
20) Jørgensen, T. J. D.; Bache, N.; Roepstorff, P.; Gardsvoll, H.; Ploug, M. Mol.
Cell. Proteomics 2005, 4, 1910–1919
21) Ferguson, P. L.; Konermann, L. Anal. Chem. 2008, 80, 4078–4086
22) Charlebois, J. P.; Patrie, S. M.; Kelleher, N. L. Anal. Chem. 2003, 75, 3263–
266
23) Rand, K. D.; Adams, C. M.; Zubarev, R. A.; Jørgensen, T. J. D. J. Am. Chem.
Soc. 2008, 130, 1341–1349
24) Pan, J.; Han, J.; Borchers, C. H.; Konermann, L. J. Am. Chem. Soc. 2008,
30, 11574–11575
25) Zehl, M.; Rand, K. D.; Jensen, O. N.; Jørgensen, T. J. D. J. Am. Chem. Soc.
008, 130, 17453–17459
26) Rand, K. D.; Zehl, M.; Jensen, O. N.; Jørgensen, T. J. D. Anal. Chem. 2009,
1, 5577–5584
27) Abzalimov, R. R.; Kaplan, D. A.; Easterling, M. L.; Kaltashov, I. A. J. Am.
Soc. Mass Spectrom. 2009, 20, 1514–1517
28) Breuker, K.; Oh, H. B.; Lin, C.; Carpenter, B. K.; McLafferty, F. W. Proc.
Natl. Acad. Sci. U.S.A. 2004, 101, 14011–14016
29) Zubarev, R. A.; Kelleher, N. L.; McLafferty, F. W. J. Am. Chem. Soc. 1998,
20, 3265–3266
30) Syrstad, E. A.; Turecek, F. J. Am. Soc. Mass Spectrom. 2005, 16, 208–224
31) Prell, J. S.; O’Brien, J. T.; Holm, A. I. S.; Leib, R. D.; Donald, W. A.; Williams,
E. R. J. Am. Chem. Soc. 2008, 130, 12680–12689
1
.
.
1
.
.
(36) Iavarone, A. T.; Williams, E. R. Int. J. Mass Spectrom. 2002, 219, 63–72
.
.
(37) Iavarone, A. T.; Williams, E. R. J. Am. Chem. Soc. 2003, 125, 2319–2327
(38) Lomeli, S. H.; Yin, S.; Ogorzalek Loo, R. R.; Loo, J. A. J. Am. Soc. Mass
.
(
(
.
Spectrom. 2009, 20, 593–596
(39) Lomeli, S. H.; Peng, I. X.; Yin, S.; Loo, R. R. O.; Loo, J. A. J. Am. Soc. Mass
Spectrom. 2010, 21, 127–131
(40) Iavarone, A. T.; Jurchen, J. C.; Williams, E. R. J. Am. Soc. Mass Spectrom.
2000, 11, 976–985
(41) Page, J. S.; Kelly, R. T.; Tang, K.; Smith, R. D. J. Am. Soc. Mass Spectrom.
2007, 18, 1582–1590
(42) Sterling, H. J.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2009, 20, 1933–
1943
.
3
.
(
(
(
(
(
(
(
.
.
.
1
.
.
2
.
.
8
.
(43) Sterling, H. J.; Daly, M. P.; Feld, G. K.; Thoren, K. L.; Kintzer, A. F.; Krantz,
B. A.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2010, 21, 1762-1774.
(44) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 89th ed.; CRC
Press: Boca Raton, FL, 2008-2009.
.
.
(45) Chowdhury, S. K.; Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1990, 112,
9012–9013
(46) Loo, J. A.; Loo, R. R. O.; Udseth, H. R.; Edmonds, C. G.; Smith, R. D. Rapid
Commun. Mass Spectrom. 1991, 5, 101–105
(47) Konermann, L.; Douglas, D. J. Rapid Commun. Mass Spectrom. 1998, 12,
435–442
.
1
.
(
(
.
.
.
.
Analytical Chemistry, Vol. 82, No. 21, November 1, 2010 9051

Here, we demonstrate that adding 1% m-NBA to aqueous
solutions containing the protein ubiquitin can significantly increase
charging. Traveling wave ion mobility spectrometry (TWIMS)
results show that the higher charge state ions that are formed
with m-NBA are more unfolded than the lower charge state ions
that are formed without this supercharging reagent, and ETD of
these more highly charged and unfolded ions results in signifi-
cantly improved fragment ion abundance and sequence coverage.
When ESI supercharging is combined with top-down HDX-MS,
the quench step and the concomitant loss of information due to
back-exchange can be eliminated, and the experiments can be
performed continuously at physiological pH and temperature, as
is done with NMR, but has the advantage of higher sensitivity.
This new method could make high spatial resolution, top-down
HDX-MS a much simpler and faster experiment to perform,
without HDX information loss due to back-exchange or scrambling.
EXPERIMENTAL SECTION
All solutions of bovine ubiquitin were prepared from lyophilized
solid (Sigma, St. Louis, MO) without additional purification. Mass
spectra for all HDX-MS experiments were acquired using an LTQ-
orbitrap hybrid mass spectrometer (Thermo Fischer Scientific,
Waltham, MA). ETD experiments were performed by reacting
mass-selected precursor ions with fluoranthene anions for 30 ms.
Fully deuterated bovine ubiquitin was prepared (∼99% deuterium
incorporation measured using mass spectrometry). Exchange for
hydrogen was initiated by diluting a 1 µL aliquot of the deuterated
ubiquitin solution into 99 µL of 200 mM ammonium acetate, pH
Figure 1. Plot of the (a) ESI-MS relative charge state abundances
of 10 µM ubiquitin (20 mM aqueous ammonium acetate, pH 7.0)
formed with and without 1% m-NBA and (b) TWIMS reduced arrival
times as a function of charge state, which shows two distinct families
of conformers.
observed at lower charge states and the more unfolded or
elongated conformers observed at higher charge states, with two
distinct conformers, or families of conformations, observed for
the 6+ and 7+ charge states. These data also indicate a trend of
additional unfolding with increasing charge state within these
conformer families. Koeniger and Clemmer measured the absolute
collision cross sections of ubiquitin ions formed by ESI from a
49:49:2 water/acetonitrile/acetic acid solution and found that the
5+-7+ charge states had nearly identical collision cross sections,
)
6.2, containing 1.0% m-nitrobenzyl alcohol (Sigma, St. Louis,
MO). The sample was mixed for a few seconds before a ∼7 µL
aliquot was loaded into a nanospray capillary, ESI was established,
and the first ETD spectrum was acquired within approximately 2
min. The same capillary was used to obtain data up to 93 min,
and new capillaries were used for the 185 and 385 min time points.
Fragment ions were identified using a m/z uncertainty constraint
of ±0.005. ESI mass spectra and TWIMS arrival time distributions
of individual charge states of ubiquitin were acquired using a
hybrid quadrupole/ion mobility/time-of-flight instrument (Synapt
High Definition Mass Spectrometer; Waters, Milford, MA) equipped
with a Z-spray ion source. Experimental procedures are described
in detail in Supporting Information.
49
consistent with a compact conformation or family of conformers.
A transition to multiple partially folded and elongated conforma-
tions was observed for the 8+ and 9+ ions, and multiple elongated
conformations were observed for 10+-13+ ions. Here, the
transition to a more unfolded structure(s) occurs at a lower charge
state (between 6+ and 7+). This may be due to ion heating in
5
0
the traveling wave apparatus that can cause unfolding and/or
due to the different solution compositions or ESI source conditions
used. Nonetheless, these TWIMS results indicate that the higher
charge state ions obtained by supercharging with m-NBA are
significantly more unfolded than the low charge state ions that
are formed with or without this supercharging reagent. Similar
results were obtained for myoglobin ions formed by electrospray
from a purely aqueous solution containing either m-NBA or
RESULTS AND DISCUSSION
Supercharged ESI-TWIMS-MS. Relative abundances of
charge states of ubiquitin formed by ESI (20 mM aqueous
ammonium acetate, pH 7.0) from solutions that contain either 0%
or 1.0% m-NBA are shown in Figure 1a. Without m-NBA, charge
states ranging from 4+ to 7+ are formed. With 1% m-NBA, the
maximum charge state increases from 7+ to 10+, and the average
charge increases from 5.5+ to 7.4+. The corresponding arrival
time distributions obtained by TWIMS for each of these charge
states are used to calculate the reduced arrival times by multiply-
43
sulfolane, suggesting that the supercharging effect is not protein
specific and should be broadly applicable for inducing rapid
protein unfolding (and/or analyte heating) in an ESI droplet. The
ESI mass spectra and TWIMS arrival time distributions from
which the results in Figure 1 are derived are given in the
Supporting Information.
4
8
ing the centroid of a given TWIMS peak by the ion charge.
These reduced arrival times as a function of charge are shown in
Figure 1b. The data are consistent with two generally separate
families of conformers with the more compact conformations
Supercharged ESI-MS/MS. The effects of charge state on
ETD sequence coverage with supercharging reagent were inves-
(
49) Koeniger, S. L.; Clemmer, D. E. J. Am. Soc. Mass Spectrom. 2007, 18, 322–
331
(50) Shvartsburg, A. A.; Smith, R. D. Anal. Chem. 2008, 80, 9689–9699
(48) Shelimov, K. B.; Clemmer, D. E.; Hudgins, R. R.; Jarrold, M. F. J. Am. Chem.
.
Soc. 1997, 119, 2240–2248
.
.
9052 Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

Figure 2. Electron transfer dissociation mass spectra of the isolated (a) 7+ and (b) 10+ ions obtained from a 3.75 µM ubiquitin, 200 mM
ammonium acetate, pH 6.2, solution containing 0% and 1% m-NBA, respectively. In both cases, these were the highest charge states that could
be reliably isolated with sufficient S/N to perform ETD. The reduced precursors are indicated with a *. Insets are expansions of the m/z 827-877
regions to illustrate differences in fragmentation efficiencies.
cleavages between residues 39 and 74, so only limited sequence
information is obtained from the N-terminal end of the protein.
For the 10+ charge state formed with 1.0% m-NBA, there are
•
4
1 c and 47 z ions, with cleavage at 62 of 72 unique sites
resulting in 86% sequence coverage. For the 18 common
fragment ions that are observed in both the 10+ and 7+ ETD
spectra, the abundances, normalized for charge, are ∼12×
greater in the 10+ ETD spectrum, indicating a substantially
higher fragmentation efficiency for the higher charge state ion.
Five of the 23 fragment ions produced by ETD of the 7+ are
not observed in the ETD spectrum of the 10+, but the
sequence coverage obtained with the 10+ is significantly higher
and is more uniformly distributed. The unique fragmentation
observed for the 7+ is likely due to conformer-dependent
dissociation as has been observed previously with ECD of
Figure 3. Sequence coverage of ubiquitin obtained from ETD of
the 7+ and 10+ charge states of 3.75 µM ubiquitin formed by
electrospray from a 200 mM ammonium acetate, pH 6.2, solution
containing 0% and 1% m-NBA, respectively. 25% and 86% sequence
coverage was obtained for the 7+ and 10+, respectively.
51
different ubiquitin conformers separated by ion mobility. The
sequence coverage values obtained here are somewhat higher
than the 17% and 59% sequence coverage for 7+ and 10+ ions,
respectively, obtained by McLafferty and co-workers using
tigated. ETD of the isolated 7+, the maximum charge state
observed without m-NBA, results primarily in formation of the
6
+ and 5+ charge-reduced precursors (Figure 2a). Some c and
3
2
•
ECD.
z
ions are observed, albeit in very low abundance. Charge-
reduced precursors are also formed upon ETD of the isolated
0+, the highest charge state observed with 1.0% m-NBA, but
Even higher charge states can be obtained from solutions
containing organic solvents and acid. For example, the 13+ of
ubiquitin is formed with the “quench” solution conditions used
by Konermann and co-workers in their top-down HDX experiment
1
in striking contrast to the results without m-NBA, significantly
more fragmentation is observed (Figure 2b). These results
clearly show the advantage of combining supercharging reagents,
such as m-NBA, with ETD experiments even when these ions are
formed from buffered aqueous solutions.
2
4
(
1:1 water/acetonitrile with 0.2% formic acid). To determine if
this higher charge state results in greater sequence coverage,
ETD of the 13+ and 10+ were compared, adjusting the fluoran-
thene anion abundances so that the precursors were depleted to
a similar extent. The sequence coverage for the 13+ formed from
•
The sequence coverage results obtained from the c and z ions
formed by ETD of these two charge states are compared in
Figure 3. For the 7+ formed without m-NBA, there are six c and
(
51) Robinson, E. W.; Leib, R. D.; Williams, E. R. J. Am. Soc. Mass Spectrom.
2006, 17, 1469–1479
•
17 z ions resulting in 25% sequence coverage. There are no
.
Analytical Chemistry, Vol. 82, No. 21, November 1, 2010 9053

Figure 4. Partial ETD mass spectra of fully deuterated 10+ ubiquitin showing the c17 (left) and c16 (right) fragment ions obtained at 3, 41, and
2 min during exchange for hydrogen. The monoisotopic peak (m, inset) and each isotope peak in the distributions are labeled. Partial ETD
spectra for the fully protonated c17 and c16 fragment ions obtained without HDX are in the inset.
9
the water/acetonitrile/formic acid solution was 81% compared to
the 86% obtained from the 10+ formed by supercharging from
aqueous solution, consistent with previous results that show
protein shape can play a greater role in fragmentation efficiency
than charge state. The similar sequence coverage for these two
ions formed from solutions where the protein conformations differ
indicates that the extent of unfolding caused by the supercharging
reagent during the electrospray process is adequate for high
sequence coverage with ETD.
Amide HDX Kinetics at Individual Residues. ETD spectra
of fully deuterated 10+ ubiquitin (formed with 1.0% m-NBA)
exchanging with hydrogen as a function of time were obtained
for times up to 385 min using a single nanospray capillary (∼7
µL) to acquire data from 2 to 94 min and two additional capillaries
to obtain data at 185 and 385 min. Amide exchange kinetics for
individual residues were obtained from these kinetic data. An
example of partial ETD mass spectra showing the c17 and c16
fragments measured at 3, 41, and 92 min are shown in Figure
Figure 5a (open squares). It is apparent from these data that this
site is essentially completely exchanged by ∼100 min. The scatter
in these data, as well as the slightly negative values at the two
longer times (185 and 385 min) can be attributed to subtracting
two relatively large numbers corresponding to all the exchange
that occurs in the c17 and c16 fragments, where a small relative
error in either or both average exchange values lead to a large
relative error for the difference in these values. These data can
be reasonably fit with a single exponential function from which
5
1
an HDX rate constant of 4.5 × 10- min is obtained for the
2
-1
amide of Glu18.
Exchange rate constants with single residue resolution were
obtained for 44 of the 72 backbone amide. This resolution is
slightly higher than that obtained by Konermann and co-workers
2
4
(
39 of 72) using ECD of all 5+ through 12+ charge states. The
exchange rate constants for these residues varied between >1.4
and <3.5 × 10- min , which represents a three orders of
magnitude kinetic window for this experiment as implemented
here (Table S-1; Supporting Information). For example, single
residue resolution was obtained for Val5, but in contrast to Glu18,
very little exchange for hydrogen was observed over the entire
3
-1
4
. The isotopic distributions of both ions shift to lower m/z with
longer times as deuterium is exchanged for hydrogen. The
isotopic distributions of the fragment ions in these exchange
experiments are significantly broader than the natural isotopic
distribution without exchange (Figure 4, inset) owing to the
statistical nature of the HDX process. The average deuterium
content for each fragment ion is obtained from the difference in
the average molecular weights of the exchanged and unexchanged
fragments calculated using the intensity-weighted average of the
observed isotopic peaks. The difference in the average deuterium
content in two sequential fragment ions provides information about
the extent of exchange at the individual amide site.
3
85 min experiment (filled squares, Figure 5a) resulting in a lower
-3
-1
limit to the exchange rate constant of 3.5 × 10 min . Significant
protection of this amide from the bulk solvent can be inferred
from this very slow exchange constant and is consistent with
5
2
its protected position within a beta strand. Adjacent to Val5
on both the N- and C-terminal sides, very fast exchange occurs
at residues 1 and 2 and 7-12 where essentially complete
exchange occurs in less than 3 min. The vastly different rates
of exchange at these neighboring amide exchange sites clearly
indicate that neither H/D scrambling nor back-exchange
occurs to any significant extent at these sites.
For example, the difference in the average deuterium content
of the c17 and c16 fragments as a function of time provides
information about the amide exchange kinetics at residue 18
Amide HDX Kinetics at Contiguous Residues. For residues
where sequential fragment ions are not observed, the next nearest
(
Glu18). The average deuterium content for the Glu18 amide
obtained from these data as a function of time is shown in
(52) Apweiler, R.; Martin, M. J.; et al. Nucleic Acids Res. 2010, 38, D142–D148
.
9054 Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

∼
1 deuterium. The data for these three residues are consistent
with one well-protected amide and two amides with similar rates
of exchange. Although three exchange rate constants can be
obtained for these three residues, no information about which of
the three residues is the slow exchanger is obtained. For Gly75
and Gly76 (filled triangles), nearly complete exchange occurred
by the first measurement, indicating both amides exchange on a
time scale near the upper limit of exchange for the experiment
>1.4 min- ).
1
(
There are 12 regions of the protein where summed kinetics
were obtained, with the longest section corresponding to five
residues (Lys48-Asp52). The average spatial resolution observed
with this supercharged ESI HDX-MS approach is calculated as
the average of all the segment lengths (including single residues)
and is equal to 1.3 residues. This is a significantly higher spatial
resolution than was previously obtained for HDX-MS analysis of
ubiquitin using either pepsin proteolysis and direct mass analysis
53
of the proteolytic fragments or collisionally activated dissociation
5
4
of the intact protein and somewhat higher resolution than was
reported for a single time-point analysis of ubiquitin exchange
24
kinetics using an acid-quenched, top-down approach. To the best
of our knowledge, this is the first example of the use of top-down
mass spectrometry to report exchange rate constants for H/D
exchange of a protein, whereas reporting rate constants (or
5
5-58
protection factors) is common with NMR experiments.
Redundant Information to Reduce Uncertainty. Some of
the HDX kinetic data can lead to ambiguities in interpretation,
but there is redundant information in these data that can eliminate
some uncertainty in individual rate constants. For example, Asn25
and Thr55 have similar initial exchange rate constants, but Asn25
appears to be biexponential with ∼40% fully exchanged by ∼200
min (open circles; Figure 5c), with little exchange occurring after
that, whereas Thr55 can be fit to a single exponential function.
The apparent biexponential behavior observed for Asn25 could
be due to scatter in the data and error induced by subtracting
two relatively large numbers, where a small relative uncertainty
in one or both values leads to a large relative uncertainty in the
subtracted value, or it could indicate two or more different
conformations in solution and/or dynamical events that reside in
5
4
the EX2 regime on a much longer time scale than the duration
of this experiment. If these data were truly biexponential, adjacent
residues should have similar offsets. The HDX kinetics for the
residues adjacent to Asn25 do not show a similar ∼0.4 Da offset,
suggesting it is more likely due to a bias in one (or both) sets of
fragment ion data.
Figure 5. Example hydrogen/deuterium exchange kinetics, where
single residue resolution was obtained for (a) Val5 and Glu18 and
Redundant data can also occur in regions of the protein where
(
b) a three-residue sum and a two-residue sum were obtained for
•
both c and z ions overlap. This occurred for four residues,
Leu67, His68, and Leu69, and Gly75 and Gly76, respectively. Single
residue resolution results for (c) Asn25 and Thr55 illustrate decay to
different average deuterium content end points. Exchange kinetics
and statistically indistinguishable rate constants were obtained
•
from both the c and z ions data for Thr12, Lys29, and Leu56
•
for (d) Asp 32 plotted with c ion data (red) and z ion data (black) and
(Table S-1; Supporting Information), whereas different kinetic
as part of a three-residue in silico digestion (Ile30 + Gln31 + Asp32).
Data were fit to single or double exponential functions where
appropriate.
(53) Liuni, P.; Rob, T.; Wilson, D. J. Rapid Commun. Mass Spectrom. 2010, 24,
15–320
54) Hoerner, J. K.; Xiao, H.; Kaltashov, I. A. Biochemistry 2005, 44, 11286–
1294
55) Pan, Y. Q.; Briggs, M. S. Biochemistry 1992, 31, 11405–11412
56) Johnson, E. C.; Lazar, G. A.; Desjarlais, J. R.; Handel, T. M. Struct. Fold.
Des. 1999, 7, 967–976
57) Bougault, C.; Feng, L. M.; Glushka, J.; Kupce, E.; Prestegard, J. H. J. Biomol.
NMR 2004, 28, 385–390
58) Schanda, P.; Forge, V.; Brutscher, B. Proc. Natl. Acad. Sci. U.S.A. 2007,
104, 11257–11262
3
.
(
1
.
neighbor is used to calculate the summed HDX kinetics for the
residues between where cleavages occur. Examples of summed
kinetic data are shown in Figure 5b for a two-amide sum and a
three-amide sum. The summed HDX kinetics of Leu67, His68,
and Leu69 (+ symbol) are best fit to a single exponential with a
half-life of ∼7 min and with a constant offset corresponding to
(
(
.
.
(
(
.
.
Analytical Chemistry, Vol. 82, No. 21, November 1, 2010 9055

56
Figure 6. Comparison of “relative protection” from HDX for individual residues obtained by MS (red diamonds), NMR (black squares), and
NMR57 (black triangles) across the protein sequence. “Relative protection” is calculated as the (negative) logarithm of the exchange rate constants
normalized within a data set.
data was obtained for Asp32 (Figure 5d). An offset value of -0.4
kinetics. For example, the generally faster exchange kinetics
observed here could be due to faster “breathing” dynamics that
occur at 38 °C compared to 30 °C or 25 °C for the two NMR
experiments.
provided a best fit for the c ion data (open triangles), whereas
•
56
57
the z ion data (open square) is biexponential and indicates a
significant fraction of this population does not exchange on
the time frame of this experiment. The discrepancy between
A more meaningful comparison is to compare the relative rates
of exchange at each site in the protein in order to observe the
regions of the protein that are either more or less protected from
exchange. To do this, we normalized the (negative) logarithms
of the rate constants for each of the three data sets and plotted
these “relative protection” values against the amino acid sequence,
where a value of 1 indicates the slowest exchange rate observed
and a value of 0 indicates the fastest exchange rate observed within
a set of experimental data (Figure 6). The absence of a point for
•
the results obtained for these c and z ions at this one site
may be due to different conformations of the protein that are
•
51
probed individually by the respective c and z ions, or it could
be due to the previously discussed error induced by the
subtraction.
One approach to determine if the information obtained from
•
the c and z ions for Asp32 is different is to use summed kinetics
to investigate the region of the primary sequence immediately
adjacent to this residue to see how they are affected. The
summed kinetic data for Ile30 + Gln31 + Asp32 are shown in
Figure 5d. The resulting summed data is analogous to results
obtained from the bottom-up method where proteolytic digestion
defines the length of the fragments. This in silico digestion makes
it possible to choose the residues at which to “digest” the protein
to provide comparable data. Results from in silico digestion at
residues before and after residues 30 and 32, respectively, are
shown in Figure 5d and show that the summed kinetics using c
5
6
the MS data (red diamond) or the NMR data (black square or
5
7
black triangle ) indicates single residue resolution was not
obtained for that amide. Overall, the MS results agree very well
with the NMR results as indicated by comparable “relative
protection” in regions of high protection (e.g., residues 3-6), in
regions of intermediate protection (e.g., residues 54-56), and
regions of low protection (e.g., both termini). Individual discrepant
data points are observed randomly across the protein, but this is
also the case when just the two NMR data sets are compared.
•
and z ions are essentially the same. This result suggests that
•
the different rate constants obtained from the c and z ion data
5
7
For example, results from one NMR experiment show 0.2
relative protection for residue 65, whereas MS and the other NMR
for Asp32 do not reflect different solution-phase conformers.
On the basis of this result, the value we report for Asp32 in
Table S-1 (Supporting Information) is the average of the two
single residue experimental values for this residue.
5
6
experiment show zero relative protection at this site. To more
thoroughly evaluate these comparisons, the relative protection
factors obtained by MS were plotted versus those from the
individual NMR experiments and the two NMR data sets with each
other (Figure S-2, in Supporting Information). The slopes of the
Comparison to NMR Data. Amide HDX rate constants
obtained in these MS experiments and values reported from
56,57
previous NMR experiments
are given in Table S-1 (Supporting
Information) for comparison. On average, the rate constants
obtained from our MS measurements are higher than those
determined by NMR. However, a direct comparison of the rate
constants is complicated by differences in experimental conditions
used to obtain all three sets of data. Slightly different solution
conditions were used in each of the three experiments, which
may affect the absolute exchange rate constants due to local
structural differences influenced by pH, buffer ions, or ionic
strength. Different solution temperatures were also used, which
can affect protein structure and dynamics and the resulting HDX
2
56
two MS versus NMR plots ((slope ) 0.90; R ) 0.84) or (slope
2
57
)
0.90; R ) 0.84) ) are somewhat better than the two NMR
2
data sets against each other (slope ) 0.85; R ) 0.83), and the
correlation coefficients are essentially equal for all three
comparisons. The generally good agreement between the MS
protection factors reported here and those obtained by NMR
indicates that this supercharging MS method provides mean-
ingful HDX values with nearly single amino acid spatial
resolution as is typically achieved with NMR.
9056 Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

CONCLUSIONS
be essentially eliminated. An advantage of the use of a mixing
system is the ability to continuously acquire data for a single
time point so that scan averaging can be used to improve the
signal-to-noise ratios of these measurements. This could result
in more precise rate constants than were obtained in this
experiment, where 2 min of real-time data were averaged for a
given experimental time point. This type of continuous-flow
approach requires much more protein than one based on a
few nanoelectrospray capillaries as demonstrated here, but the
significantly better sensitivity of mass spectrometry compared
to NMR could make this an attractive method when a large
quantity of the protein of interest is not readily available.
When ESI supercharging is combined with top-down HDX-
MS, real-time amide exchange rate constants for individual
residues can be obtained with significantly higher sensitivity than
NMR. The generally good agreement between values obtained
with this MS method and previous NMR values indicates that H/D
scrambling does not occur to any significant extent, and these
results provide additional evidence that the ESI supercharging
reagent does not cause significant changes to the protein structure
in the initial solution but causes proteins to denature in the
electrospray droplet. Because the droplet lifetime is short (ms),
this method significantly reduces the potential for back-exchange
to occur when HDX is performed near physiological pH because
even the fastest intrinsic exchange rates are significantly slower.
The supercharging reagents have the additional advantage of
increasing the charge states formed by ESI, which increases the
fragmentation efficiency and the resulting sequence coverage
obtained using ECD or ETD. Although not done here, including
results from other precursor charge states would likely improve
the spatial resolution of this method, and this may become
important for larger proteins where ETD fragmentation may
become more limited.
ACKNOWLEDGMENT
The authors thank Michael P. Daly, Don Harris, and the Waters
Corporation for generously making available a Synapt High Definition
Mass Spectrometer demonstration instrument for the TWIMS
studies, thank the National Institutes of Health RO1GM064712 and
T32GM08295 (training grant H.J.S.) for financial support, and thank
James S. Prell for helpful discussion.
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
-1
The fastest reported rate constants are ∼1.4 min and were
limited by the long dead time in these initial experiments. Data
at significantly shorter times could be obtained using a mixing
18
apparatus, such as those used by Konermann and co-workers,
Received for review August 2, 2010. Accepted September
to measure significantly faster rate constants. Using a super-
charging reagent in the diluant, only a single mixing “T” is
required to initiate H/D exchange and back-exchange would
28, 2010.
AC101957X
Analytical Chemistry, Vol. 82, No. 21, November 1, 2010 9057