Characterization of Protonated Model Disaccharides from Tandem Mass Spectrometry and Chemical Dynamics Simulations

Article abstract of DOI:10.1002/cphc.201700202

The fragmentation mechanisms of prototypical disaccharides have been studied herein by coupling tandem mass spectrometry (MS) with collisional chemical dynamics simulations. These calculations were performed by explicitly considering the collisions between the protonated sugar and the neutral target gas, which led to an ensemble of trajectories for each system, from which it was possible to obtain reaction products and mechanisms without pre-imposing them. The β-aminoethyl and aminopropyl derivatives of cellobiose, maltose, and gentiobiose were studied to observe differences in both the stereochemistry and the location of the glycosidic linkage. Chemical dynamics simulations of MS/MS and MS/MS/MS were used to suggest some primary and secondary fragmentation mechanisms for some experimentally observed product ions. These simulations provided some new insights into the fundamentals of the unimolecular dissociation of protonated sugars under collisional induced dissociation conditions.

Full text of DOI:10.1002/cphc.201700202

Accepted Article
Title: Characterization of protonated model disaccharides from tandem
mass spectrometry and chemical dynamics simulations
Authors: Estefania Rossich-Molina, Ane Eizaguirre, Violette Haldys,
Dominique Urban, Gilles Doisneau, Yann Bourdreux, Jean-
Marie Beau, Jean-Yves Salpin, and Riccardo Spezia
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemPhysChem 10.1002/cphc.201700202
Link to VoR: http://dx.doi.org/10.1002/cphc.201700202
A Journal of
www.chemphyschem.org

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
Characterization of protonated model disaccharides from tandem
mass spectrometry and chemical dynamics simulations
Estefania Rossich Molina,[a,b] Ane Eizaguirre,[a,b] Violette Haldys,[a,b] Dominique Urban,[c] Gilles
Doisneau,[c] Yann Bourdreux,[c] Jean-Marie Beau,[c] Jean-Yves Salpin*[a,b] and Riccardo Spezia*[a,b]
Abstract: In the present work the fragmentation mechanisms of
prototypical disaccharides have been studied by coupling tandem
mass spectrometry (MS) with collisional chemical dynamics
simulations. These last calculations were performed by explicitly
considering the collisions between the protonated sugar and the
neutral target gas, leading to an ensemble of trajectories for each
system, from which it was possible to obtain reaction products and
mechanisms without pre-imposing them. The β-aminoethyl and
aminopropyl derivatives of cellobiose, maltose and gentiobiose were
studied in order to see the differences in both the stereochemistry and
the location of the glycosidic linkage. Chemical dynamics simulations
of MS/MS and MS/MS/MS are used to suggest some primary and
secondary fragmentation mechanisms for some experimentally
observed product ions. The simulations performed provide some new
insights about the fundamentals of unimolecular dissociation of
protonated sugars under CID conditions.
ionization (ESI)[5] techniques are particularly useful, and have
been extensively used to generate bio-molecular ions in the gas
phase. In combination with these ionization methods, tandem
mass spectrometry (MS/MS) through collisional induced
dissociation of ions (CID) are also particularly helpful to
understand the unimolecular reactivity and to identify
characteristic features of biomolecules, since activated ions get
enough energy to follow extensive fragmentation.[6]
One difficult task of mass spectrometry in sugar chemistry is to
achieve isomeric distinction. However, the possibility of applying
CID to differentiate isomeric sugars is appealing due to the
relative rapidity of such an analytical approach with respect to
more time-consuming spectroscopic methods. Another possibility
of improving separation of isobaric isomers is given by ion mobility,
as notably recently reported by Pagel and co-workers.[7]
In general, the fragmentation mechanisms of saccharides are less
investigated from a fundamental point of view than other
polymeric molecules, as, in particular, peptides.[8] Studies on
fragmentation mechanisms are reported on simple saccharides
(like lactose) pointing out the possibility of primary vs consecutive
fragmentations also using H/D exchange and 18O or 13C labeling
and quantum chemical calculations.[9] Earlier theoretical studies
were reported estimating barriers and mechanisms for other
systems.[10]
To investigate their reactivity under CID conditions we combined
MS/MS experiments with chemical dynamics simulations of
collision. The use of chemical dynamics to model CID
phenomenon was pioneered by Hase and co-workers in studying
CID of different systems,[11] and more recently in our group it was
applied to elucidate reaction mechanisms of a series of organic
and biological molecules, from urea to peptides, from uracil to
doubly charged complexes formed by an ion and an organic
molecule.[12] Concerning sugars, the first, and up to now unique,
system investigated with this approach was galactose 6-sulfate,
in which simulations were able to reproduce the most important
experimental fragmentation pathways and identify the underlying
mechanisms.[13] In fact, direct dynamics are able to provide
information on the structure of the fragment ions obtained after
collision of a gaseous precursor ion with an inert gas, and thus on
the reaction pathways, without pre-imposing any reaction
coordinate.
Introduction
Carbohydrates, as part of glycoproteins and glycolipids involved
in cell signaling, play an important biological role.[1] Besides their
biological relevance, the characterization of both the structure and
reactivity of carbohydrates is per se a challenging topic for
spectrometric identification due to the numerous questions that
have to be addressed, including the nature and position of the
functional groups, the sites and anomeric configuration of the
glycosidic linkages, and the stereochemical characterization of
the different asymmetric centers of the sugar rings.
To study the intrinsic reactivity of carbohydrates by getting rid of
their environment, gas-phase studies are pointed out as a
satisfactory approach. In this respect, fast atom bombardment
(FAB),[2] matrix assisted laser desorption ionization (MALDI),[3]
desorption electrospray ionization (DESI)[4] or electrospray
[a] Dr. E. Rossich Molina, Dr. A. Eizaguirre, Mrs. V. Haldys, Dr. J.-Y.
Salpin, Dr. R. Spezia
LAMBE, Univ Evry, CEA, CNRS, Université Paris-Saclay, F-91025
Evry, France
[b] Dr. E. Rossich Molina, Dr. A. Eizaguirre, Mrs. V. Haldys, Dr. J.-Y.
Salpin, Dr. R. Spezia
The next step of our work on sugars is to check if chemical
dynamics simulations could also provide new insights on the
mechanisms associated with the glycosidic bond cleavages, and
if simulations could achieve isomeric distinction. To this end, we
decided to study the simplest carbohydrates exhibiting an
interglycosidic linkage, namely disaccharides. We have chosen
LAMBE, Université Cergy-Pontoise, Université Paris-Seine, F-91025
Evry, France
E-mail: jeanyves.salpin@univ-evry.fr ; riccardo.spezia@univ-evry.fr
[c] Dr. D. Urban, Dr. G. Doisneau, Dr. Y. Bourdreux, Prof. J.-M. Beau
ICMMO – SM2B, Univ Paris-Sud, Université Paris-Saclay and CNRS
F-91405 Orsay (France)
three isomers, D-maltose, D-cellobiose and D-gentiobiose, which
1
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
differ from each other either by the position or the α,β-
configuration of the glycosidic linkage. It has been shown that
disaccharides could be easily distinguished by tandem mass
spectrometry by performing MS/MS experiments on the [M-H]-
pseudo-molecular ions.[14] However, several theoretical studies
have demonstrated that the different hydroxyl groups of
saccharides are of close acidity in the gas phase.[15] In addition, a
recent report showed that C-deprotronation can also compete
with O-deprotonation.[16] Consequently, the number of
deprotonated forms to be considered in the negative mode being
too high to perform chemical dynamics simulations, we decided
to work in the positive-ion mode. However, in this case chemical
derivatization was mandatory for two reasons. First, there is no
preferred basic site since protonation of any of the O atoms of the
sugars results in similar proton affinities.[17] Second, under
electrospray conditions, [M+H]+ ions of underivatized
disaccharides are very weak as they easily fragment according to
sequential losses of water molecules. Consequently, the three
disaccharides were functionalized at the reducing end with amino-
ethyl and amino-propyl groups (scheme 1), in order to
unambiguously localize the positive charge for chemical dynamic
simulations, and to generate intense [M+H]+ ions that can then be
easily studied by CID.
functionalities by palladium catalyzed hydrogenation reactions in
methanol for two hours at room temperature. After filtration and
evaporation of the solvent, the β-aminoethyl (n = 1) and the β-
aminopropyl (n = 2) derivatives were obtained in yields of 70%
and 90%, respectively. All glycosides were fully characterized and
then lyophilized prior to their study by mass spectrometry. The
synthesized disaccharides are schematically drawn in Scheme 1.
We have used the combination of experiments and chemical
dynamics simulations to propose fragmentation mechanisms of
these model disaccharide isomers and to investigate the
possibility of differentiating them by CID experiments.
Methods
Organic synthesis. Aminoderivatives of cellobiose, maltose and
gentiobiose were prepared following the same synthetic approach,
which relies on β-selective glycosylations using a bromoalcohol,
followed by appropriate functional group transformations. For all
the required glycosides, the aglycone parts were β-selectively
introduced using the common boron trifluoride-mediated
glycosylation conditions,[18] with either 2-bromoethanol or 3-
bromopropanol as the acceptor partner. Azido displacement of
the bromine atom[19] followed by total deacetylation and reduction
of the azido moiety delivered the β-aminoethyl and the β-
aminopropyl glycosides.[19,20] The experimental conditions used
are exemplified using the following preparation of the cellobiose
derivatives in Scheme 2. Commercially available cellobiose was
peracetylated using acetic anhydride in pyridine at room
temperature overnight. The product was thus obtained with a yield
of 95% as a 5:1 (α:β) mixture of isomers. The glycosylation was
then carried out using 6 equivalents of 2-bromoethanol or 3-
bromopropanol in dichloromethane and 6 equivalents of boron
trifluoride etherate at room temperature for one hour. Flash
chromatography purification gave the β-glycosides in 27% (n = 1)
and 54% (n = 2). Next, treatment of the bromide derivatives with
two equivalents of sodium azide in DMF at room temperature
overnight, quantitatively yielded the azido products that were not
purified. Total deacetylation under Zemplén conditions delivered
the O-unprotected glycosides in 99% (n = 1) and 94% (n = 2)
yields. Finally, the azido groups were converted into amino
Scheme 1. Aminoderivatives of Cellobiose (cellobiosides), Maltose
(maltobiosides) and Gentiobiose (gentiobiosides) studied in the present work.
Scheme 2. Synthetic approach followed to prepare aminoderivatives of
cellobiose, maltose and gentiobiose. For simplicity we show the cellobiose case.
OAc
O
HO
OAc
O
OAc
O
OAc
O
Br
Ac2
O
n
AcO
AcO
AcO
AcO
cellobiose
O
O
O
AcO
AcO
AcO
AcO
OAc
AcO
AcO
n
Br
n = 1, 2
OH
O
OAc
O
OH
OAc
O
MeONa
NaN3
HO
HO
AcO
AcO
O
O
O
O
O
HO
AcO
then
OH
AcO
OH
AcO
n
n
NH2
N3
H2, Pd/C
n = 1, 2
The preparation of the other aminoderivatives is similar.
Mass spectrometry experiments. Experiments were performed
using an Applied Biosystems/MDS Sciex API2000 triple-
quadrupole instrument fitted with
a turboionspray source.
Aqueous solutions (10-2 M) of β-aminoethyl and β-aminopropyl
derivatives of Cellobiose β(1,4), Maltose α(1,4), Gentiobiose
β(1,6) (synthetized as reported above) were diluted in Milli-Q
water and were introduced in the source using direct infusion with
a syringe pump at a flow rate of 5 μl/min. Ionization of the sample
was achieved by applying a voltage of 5.5 kV on the sprayer probe
and by the use of a nebulizing gas (GAS1, air) surrounding the
sprayer probe, intersected by a heated gas (GAS2, air) at an
2
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
angle of 90°. The operating pressure of GAS1 and GAS2 are
adjusted to 2.1 bars, by means of an electronic board (pressure
sensors), as a fraction of the air inlet pressure. The curtain gas
(N2), which prevents air or solvent from entering the analyzer
region, was similarly adjusted to a value of 1.4 bar. The
temperature of GAS2 was set to 100°C. CID spectra were
recorded by introducing nitrogen as collision gas in the second
quadrupole.
where VSaccharideH+ is the intramolecular potential of the protonated
saccharide cation and VAr-Saccharide is the Ar/[saccharide]H+
intermolecular potential. The PM3 semi-empirical Hamiltonian
has been used for the intramolecular potential.[23] The reliability of
such semi-empirical Hamiltonian to model CID processes of
carbohydrates was established in previous work on galactose-6-
sulfate.[13] Event if new semi-empirical methods are now available,
our work on sugars, as well as more recently on peptides,[24] have
shown that PM3 is able to correctly describe different systems in
both positive and negative modes. Surely studying the impact of
the method on the fragmentation products will be an interesting
topic, which is, however, beyond the aim of the present work.
The intermolecular potential between the collision gas (Ar) and
the atoms of the saccharide ion is expressed as a sum of two-
body terms as follows:
Low gas pressures were used to limit multiple ion–molecule
collisions. Moreover, the declustering potential (DP), defined as
the difference of potentials between the “orifice plate” and the
skimmer (grounded) was set to 40 V to perform MS/MS
experiments. CID spectra were recorded at different collision
energies ranging from 5 to 30 eV (laboratory frame). Note that
MS/MS spectra are very likely obtained under a multiple-collision
regime, and this increases the internal energy content of the
precursor ion. With the CAD parameter (which controls the
amount of N2 introduced into Q2) set to its minimum value, the
pressure value measured by the ion gauge, located at the vicinity
of Q2, is about 3 × 10-5 Torr, but the actual pressure inside Q2
cannot be determined accurately. However, according to a report
of the manufacturer,[21] this pressure inside Q2 is closer to 10-2
Torr. Given the dimensions of Q2, the mean free path for a moving
N2 molecule, according to the gas kinetic theory, is several mm at
10-2 Torr. Therefore a molecule of N2 may undergo tens of
collisions within Q2. This is a lower limit for the present ions, which
have a larger diameter and, thus, a larger collision cross section.
We also recorded the CID spectrum of several fragment ions
observed on the MS/MS spectra. To perform these MS/MS/MS
“like” experiments, we operated as follows: from the obtained
MS/MS spectrum, we first chose the fragment ions we wanted to
dissociate. Then, we generated these fragment ions in the
interface region of the instrument, by increasing the DP value to
60 V. Each fragment ion was then mass selected by Q1,
transferred into Q2, where it collided with N2 in the same way that
it did for the MS/MS experiment on the precursor ion, with their
resulting fragment ions (second generation) being analyzed by
Q3. Note that at DP=40 V, secondary fragmentations already
occur, however we used 60 V to increase the intensity of the
fragment ions and ultimately the signal-to-noise ratio of the
pseudo-MS3 spectra. During tandem experiments, ions are
collisionally cooled within Q0 and Q1, located before Q2, but
increasing the voltage was not meant to act on the cooling. Using
either DP=40 V or 60 V results in similar pseudo-MS3 spectra.
For the sake of simplicity, we refer to these results as MS/MS/MS
spectra.
(2)
where i runs over all the atoms of the [saccharide]H+ molecular
ion. We used the purely repulsive potential developed and
discussed by Meroueh and Hase,[25] in which A, B and C are
positive.
Note that we used Ar as collision gas in simulations while in
experiments N2 was used instead. Previous works of our group
have shown that the main difference between the two gases is
that the former provides a more efficient energy transfer and
higher reaction probability with respect to the second.[26] Thus, to
increase the reactivity and indirectly the computational efficiency
we used Ar in simulations. Furthermore, we have to take into
account that simulations model a single collision, while in
experiments more collision can occur, thus increasing the
activation energy given to the ion.
Simulations of Ar-[saccharide]H+ collisions were done using as
reference structures the minimum energy structure of each
system. To obtain reliable minimum energy structures we have
first performed a fast conformational search at the PM3 level, then
we have selected the most stable structures and used the
B3LYP/6-31+G(d,p) level of theory[27] to identify the most stable
conformation. Then, harmonic vibrational frequencies were
obtained at the same level to assess that the structures found
corresponded to local minima. Finally, we have re-optimized
these structures with PM3 to obtain a minimum which can be used
as initial structure in dynamics. These initial structures are
reported in the Supporting Information. Optimizations were made
using Gaussian 09 suite of programs.[28]
Initial conditions in position and momenta of the [saccharide]H+
for chemical dynamics were chosen by adding a quasi-classical
300 K Boltzmann distribution of vibrational/rotational energies
about the isomer’s potential energy minima.[29] Energies for the
normal modes of vibration were selected from 300 K Boltzmann
distribution. The resulting normal mode energies were partitioned
between kinetic and potential energies by choosing a random
phase of each normal mode. A 300 K rotational energy of RT/2
was added to each principal axis of rotation for the ion. Vibrational
and rotational energies were transformed into Cartesian
coordinates and momenta following well-known algorithms
implemented in VENUS.[30] The cation was then randomly rotated
Chemical dynamics simulations. QM + MM chemical dynamics
simulations[22] were performed to model CID of protonated
disaccharides. As in our previous work,[12] we have used as
projectile the argon atom. The collision system constituted by the
protonated [saccharide]H+ cation and the Ar atom was described
by the following potential energy function:
(1)
3
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
about its Euler angles to take into account the random direction
of the collision between Ar and the [saccharide]H+ cation. Relative
velocities were then added to the Ar/[saccharide]H+ system
consistently with the center-of-mass collision energy and impact
parameter. Collision energy in the center-of-mass framework of
450 kcal/mol (19.51 eV) was employed in order to have enough
reactive trajectories from which a reasonable sampling of different
pathways was possible. The impact parameter b was randomly
sampled between 0 and 5.0 Å; this latter value was chosen based
on the molecular size and by energy transfer calculations showing
that for b > 5.0 Å the transferred energy is less than 10% of the
collision energy (and the reactivity consequently drops to zero),
as shown in Figure S1 of the supporting information. The
trajectories were calculated using a software package consisting
of the general chemical dynamics computer program
VENUS96[30] coupled to MOPAC.[31] The latter was used to
calculate the potential energy and gradient for [saccharide]H+
intramolecular potential. The classical equations of motion were
integrated using the velocity Verlet algorithm[32] with a time step
of 0.2 fs that gives energy conservation for reactive and
nonreactive trajectories, as for systems we have previously
investigated.[12,13,26] The trajectories were initiated at ion-projectile
distance of 15 Å, large enough to guarantee no interaction
between the ion and the colliding atom, and halted at a distance
of 300 Å to allow substantial intramolecular motion of the sugar
ion. This corresponds to a total integration time of about 5 ps. The
trajectories in which the ions dissociated were also stopped. In
this case, the criterion distance of 7 Å was used to guarantee no
interaction between the fragments. Approximately 8000
trajectories were performed for each system, 4 to 5 % being
reactive, depending on the system investigated.
Scheme 3. a) Atom numbering, and b) bond numbering and fragmentation
paths notation for β-aminoethyl-cellobiose and β-aminoethyl-maltose.
Results and Discussion
Experimental ESI-MS/MS. Fragmentation of sugars under CID
conditions is a very rich phenomenon, since they can fragment
through glycosidic bond or cross-ring cleavages, as well as loss
of neutral molecules. Scheme 3a presents the atoms numbering,
and to describe the bond that are broken, we employ the
nomenclature introduced by Domon and Costello[33] (Scheme 3b).
All the synthesized di-saccharides were electrosprayed and the
ion corresponding to the protonated species was selected for
MS/MS. The experimental ESI-MS/MS spectra exhibit many
product ions even if only one basic site (the terminal amino group)
is probably predominantly protonated, reflecting that different
fragmentation pathways are opened. The MS/MS spectra of β-
aminoethyl-disaccharides are reported in Figure 1 (the MS/MS
spectra obtained for β-aminopropyl derivatives are given in Figure
S2 of the supporting information). Examination of Figure 1 shows
that there are no remarkable m/z differences between these
spectra. The three isomers indeed have the same ESI/MS-MS
spectra in terms of the fragment ions present. Therefore MS/MS
cannot be presently easily used to distinguish the three
disaccharides. Note however that the abundance of several
peaks, and more particularly of m/z 62 and 206 ions, varies
significantly according to the isomer considered, which can be of
particular interest for analytical purpose.
MS/MS/MS fragmentations are obtained as for MS/MS, just using
as initial structure the fragmentation products, which were
optimized in order to have correct quasi-classical initial conditions.
From simulations it is possible to obtain theoretical (time-
dependent) MS/MS and MS/MS/MS spectra just counting the
number of each fragment obtained at the end of the simulations.
Relative intensities are then obtained just setting to 100 the most
intense peak and scaling the other accordingly, as done in MS/MS
and MS/MS/MS experimental spectra.
4
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
fragments obtained from CID chemical dynamics simulations of
β-aminoethyl-cellobiose (a similar pattern is obtained for β-
aminoethyl-maltose, see Figure S4 of the supporting information)
and β-aminoethyl-gentiobiose. Note that in simulations the
fragmentation patterns do not depend on the isomer considered,
but intensities are generally lower for β-aminoethyl-gentiobiose,
similarly to what obtained experimentally. In simulations, the peak
at m/z 62 is too low to be compared but the peak at m/z 224, which
has a relatively high intensity in cellobiose and maltose
derivatives, is smaller in the case of β-aminoethyl-gentiobiose,
reflecting what is observed during experiments.
Figure 2. Fragmentation distribution of product ions obtained from CID chemical
dynamics simulations of β-aminoethyl-cellobiose (upper panel) and β-
aminoethyl gentiobiose (lower panel).
Figure 1. Experimental ESI-MS/MS of β-aminoethyl cellobiose (top), maltose
(middle) and gentiobiose (bottom). Spectra were recorded with a collision
energy of 25 eV in the laboratory framework.
By inspecting the structure of the different simulation products, we
can assign for each peak the corresponding structures and the
bond breaking sites. They were thus reported to the theoretical
fragment m/z distribution of Figure 2. Both glycosidic bond- (B, Y
and Z ions) and cross-ring cleavages (X ions) are observed.
One of the most important peak is m/z 368 that corresponds to
H2O loss. Another characteristic and abundant peak is m/z 224,
likely Y1 ion, associated with the cleavage of the glycosidic bond.
We will now show how chemical dynamics simulations can be
used to identify the product ions and to suggest mechanisms that
may account for each fragment.
Chemical dynamics simulations of MS/MS. Using as starting
structures the minimum energy structures shown in Figure S3, we
have performed collisional dynamics simulations to mimic present
MS/MS conditions. In Figure 2 we show the distribution of
5
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
Table 1. List of MS/MS and MS/MS/MS fragments for β-aminoethyl-cellobiose
obtained in experiments and theory. Precursor ions are labeled as p.i.
First, comparing the theoretical fragmentation pattern with the
experimental one, we can notice that most of the fragments
obtained in simulations are also present in experiments, in
particular the most abundant ones (while the opposite does not
hold, as we will discuss later). The list of ions observed in CID
chemical dynamics is reported in Table 1, where they are
compared to those obtained in experiments. Since experimental
and simulation conditions are not the exactly the same (and in
particular simulations are in the ps time-scale, such that only fast
processes can be probed), the intensities are not comparable. In
fact, in collisional simulations products obtained through a fast
process are overestimated, while slow pathways (generally
obtained through a statistical unimolecular fragmentation) are
less sampled, and they can be not seen at all. For a deeper
discussion on such differences and on the problem of reproducing
intensities, more details are given elsewhere.[12b,34] In general
“shattering” mechanisms are better sampled, and to reproduce
correctly the slow processes, information on transition states are
needed to then study the kinetics via, for example, RRKM analysis
or via post-TS dynamics, as recently discussed for a smaller
system.[35] We now discuss some relevant peaks for which
discrepancies in intensities can be rationalized due to lack of
simulation time. Of course, the intrinsic limitation in using a semi-
empirical Hamiltonian should always be considered as a source
of error. Finally, we should remark that intensities also depend on
other instrumental specificities (ionization process, mass filters
and/or detector, for example) and it is not unusual that the same
ion exhibits different intensities in MS/MS spectra according to the
instrument used.
In the present case, we should notice that the B2 ion (m/z 325) is
intense in simulations, while it is small in the experimental
spectrum. On the other hand, the m/z 224 ion, Y1, shows similar
intensities in experiments and simulations. The other ions, namely
m/z 368 (H2O loss), m/z 266 (0,2X1), m/z 163 (B1), m/z 74 (0,1X0),
m/z 62 (Y0) and m/z 44 (Z0), are present in both experiments and
simulations, and only m/z 356, m/z 252 (1,5X1), m/z 90 (1,5X0) and
m/z 32 are only present in simulations.
Globally, the reasonable agreement between experiments and
simulations in identifying the principal fragmentation pattern (in
particular concerning the heaviest ions) suggests that the
trajectories obtained can be useful to clarify the associated
mechanisms, and in particular the primary fragmentations. The
absence of lighter ions in simulations will be discussed together
with MS/MS/MS experiments and simulations.
Fragment
(m/z)
MS2[a]
MS3
(368)[b]
MS3
(224)[c]
MS2 th.[d]
MS3
th.[e]
(368)
MS3
th[f]
(224)
386
368
356
350
332
325
290
266
252
248
224
206
188
170
163
158
152
145
140
127
104
97
p.i.
✓
-
p.i.
✓
✓
-
p.i.
-
p.i.
-
✓
-
✓
✓
-
-
-
-
✓
-
✓
-
-
✓
-
-
✓
-
✓
✓
-
-
-
-
-
✓
-
-
✓
✓
✓
-
p.i.
✓
✓
✓
-
✓
-
-
p.i.
✓
✓
✓
-
✓
-
✓
-
-
-
-
-
✓
-
✓
-
-
-
✓
-
✓
✓
✓
✓
✓
✓
✓
-
-
-
-
-
-
-
✓
-
✓
✓
✓
✓
✓
-
-
-
-
-
-
-
✓
✓
✓
-
-
-
-
-
-
✓
-
-
-
90
✓
-
-
-
86
-
-
✓
✓
✓
✓
✓
-
-
-
85
✓
✓
✓
✓
-
✓
-
-
-
-
In the following, we describe the mechanisms directly observed
during chemical dynamics simulations of the collision between the
precursor ion, m/z 386, and the projectile. More particularly, we
provide details about the MS/MS spectrum of protonated β-
aminoethyl-cellobiose (for the other sugars the mechanisms are
the same).
74
✓
✓
✓
✓
-
-
62
-
-
✓
-
44
-
-
33
-
-
-
[a] Experimental MS/MS; [b] Experimental MS/MS/MS from m/z 368 precursor
ion; [c] Experimental MS/MS/MS from m/z 224 precursor ion; [d] Theoretical
MS/MS; [e] Theoretical MS/MS/MS from m/z 368 precursor ion; [f] Theoretical
MS/MS/MS from m/z 224 precursor ion.
6
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
a)
during simulations, in agreement with mechanisms proposed in
the literature for glycosidic bond cleavages. [9a,10a,36] Simulations
also lead to the formation of a carbene neutral (Scheme 4b). We
tried to perform isotope labelling (H/D exchange) experiments but
unfortunately, we failed in exchanging all the exchangeable
hydrogen atoms under our triple-quadrupole conditions. On the
other hand, we did the experiment in a quadrupole ion trap (Figure
S5). Experimentally, one can see that the Y1 ion (m/z 224) is
shifted both to m/z 230 (minor) and 231 (prominent). These m/z
ratios are consistent with the mechanisms depicted in schemes
4b-c (6 exchangeable protons) and scheme 4d (7 exchangeable
protons), respectively. We can also note that the Y1 ion (m/z 224)
is less intense for β-aminoethyl-gentiobiose (Figure 1c). The
mechanism suggests the origin of this difference, namely the type
of glycosidic bond (1,6 vs 1,4) and formation of Y1 is due to such
bond cleavage.
As just mentioned, based on our calculations the Y1 ion could be
generated according to three mechanisms, leading to the same
ion but different neutral fragments. We successfully minimized (i.e.
obtaining real frequencies) these three neutrals at both PM3,
B3LYP/6-31+G(d,p) and MP2/6-31+G(d,p) levels of theory. The
alkenic form (Scheme 4c) is the most stable one, followed by the
epoxy one (Scheme 4d, which is 20, 11 and 8 kcal/mol higher in
energy at PM3, B3LYP and MP2 level, respectively), the carbene
(Scheme 4b) being the less stable one (by 36, 45 and 50 kcal/mol
at PM3, B3LYP and MP2 level, respectively). Note that the
abundance obtained does not reflect the energy ordering.
Probably the carbene abundance is overestimated due to the
energy underestimation at PM3 level with respect to B3LYP and
MP2, and, conversely, the epoxy structure is probably
underestimated in simulations. However, these results clearly
show how dynamical effects can be responsible of the formation
of high-energy structures in mass spectra.[37]
OH
OH
O
OH
OH
O
O
O
O
O
NH3
HO
NH3
HO
OH
O
OH
HO
H
HO
O
H
OH
O
OH
OH
OH
m/z 386
OH
O
OH
O
NH3
HO
HO
OH
HO
OH
Y0 m/z 62
OH
b)
OH
OH
OH
OH
H
O
O
H
O
O
O
O
NH3
NH3
HO
HO
HO
HO
O
OH
O
OH
OH
OH
OH
OH
m/z 386
OH
O
OH
O
HO
NH3
HO
OH
O
HO
OH
OH
Y1 m/z 224
c)
OH
O
OH
O
OH
OH
O
O
O
O
H
NH3
NH3
OH HO
HO
HO
OH
O
O
HO
OH
OH
OH
OH
m/z 386
OH
O
OH
O
HO
NH3
HO
OH
HO
O
OH
OH
Y1 m/z 224
In a recent paper by Bythell et al. the mechanisms implying a
proton mobilization for the glycosidic bond cleavage was
suggested from calculations locating minima and TSs.[9-b] In the
present simulations, we could not observe any proton transfer to
the oxygen atom of the glycosidic bond. As already mentioned,
direct collisional simulations provide the amount of energy in one
shock, and trajectories are not obliged to follow the minimum
energy path (i.e. the path connecting minima and TSs). However,
a mobile proton model mechanism is certainly a possible
alternative to generate the fragment ions. We have shown
recently for peptide fragmentation, that internal energy activation
simulations[24] are more prone to provide proton transfers. Surely,
they can be used to consider this possibility to study sugar
fragmentation.
In the case of cross-ring cleavages (X ions), dissociation
mechanisms observed during simulations are also simple: two
bonds are broken and the products are directly obtained. In the
case of 0,2X1 ions (m/z 266) this can happen either by cleaving
sequentially the C2′-C3′ and C4′-C5′ bonds (Scheme 5a) or by first
breaking the C4′-C5′ bond and then, due to a concerted bond
rearrangement, the C2′-C3′ bond (Scheme 5b). In both cases the
same products (both ions and neutrals) are obtained.
d)
OH
OH
O
OH
OH
O
O
O
O
O
H
NH3
NH3
HO
OH
HO
HO
O
O
HO
O
OH
OH
O
OH
OH
m/z 386
OH
OH
O
OH
OH
O
HO
HO
O
O
NH3
NH3
HO
O
HO
O
HO
HO
O
OH
OH
OH
OH
Y1 m/z 224
Scheme 4. Formation mechanisms of Y0 (m/z 62, panel a) and Y1 (m/z 224,
panels b, c and d) ions from CID simulations. The ratio of the three possible
pathways leading to Y1 as obtained by CID simulations is as follows: b) 75%, c)
20% and d) 5%.
The mechanisms observed during simulations are sometimes
very simple: the projectile hits the molecule and quickly after one
bond is broken and the product ion is obtained. This is the case
for the formation of Y0 (m/z 62) and Y1 (m/z 224) ions: after the
C-O bond scission (C1-O1 and C1′-O4, respectively) a proton is
transferred forming the leaving ions, as shown in Scheme 4.
Formation of either an unsaturated ring (Scheme 4c) or an epoxy
form (Scheme 4d) within the leaving neutral fragment is observed
7
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
bond (C1-O1). Experimentally, the time scale being much longer,
alternate mechanisms can occur, and notably the transient
formation of a proton bound dimer after this particular bond
scission. Within this proton bond dimer a proton transfer between
the two fragments may occur, leading to an intense m/z 62 ion
instead of m/z 325. Unfortunately, such proton bound dimers were
never observed in simulations.
a)
OH
O
OH
O
OH
OH
O
O
O
O
NH3
NH3
OH HO
HO
O
OH
HO
O
HO
OH
OH
OH
OH
m/z 386
OH
O
a)
OH
OH
OH
O
OH
OH
HO
O
O
O
O
O
O
NH3
O
HO
HO
HO
O
NH3
NH3
CH2
H2
C
HO
HO
OH
OH
O
HO
OH
HO
O
OH
OH
OH
OH
OH
0,2X1 m/z 266
m/z 386
OH
b)
OH
OH
OH
O
OH
OH
O
OH
O
O
O
O
O
O
O
O
O
O
O
CH2
CH2NH3
HO CH2
CH2NH3
OH
OH
HO
HO
HO
HO
NH3
NH3
OH HO
HO
O
OH HO
HO
O
OH
OH
OH
OH
OH
OH
OH
OH
OH
O
m/z 386
OH
OH
O
O
O
O
OH
O
CH2
CH2NH3
HO
HO
OH
O
HO
NH3
HO
HO
O
HO
OH
OH
OH
B1 m/z 163
0,2X1 m/z 266
b)
OH
OH
O
OH
OH
O
Scheme 5. Formation mechanisms of 0,2X1 ion (m/z 266) from CID simulations.
O
O
O
O
The two pathways are obtained in simulations with equal probability.
NH3
NH3
OH
HO
HO
O
OH
HO
HO
O
OH
OH
OH
OH
OH
OH
OH
m/z 386
OH
O
O
OH
O
O
O
O
O
OH
NH3
NH3
O
CH2 H2
C
O
OHHO
HO
O
HO
OH
HO
O
OH
OH
OH
OH
NH3
OH
HO
HO
O
m/z 386
OH
OH
OH
OH
B1 m/z 163
O
O
O
c)
OH
NH3
O
CH2
H2C
OH
O
OH
O
OHHO
HO
OH
OH
OH
O
O
O
O
B2 m/z 325
NH3
NH3
HO
OHHO
O
O
HO
OH
HO
Scheme 6. Formation mechanisms of B2 ion (m/z 325) from CID simulations.
OH
OH
OH
OH
B1 m/z 163
m/z 386
The simulated mechanism leading to the B2 ion (m/z 325) is a bit
more complex: first the Ca-Cb bond is broken followed by the C1-
O1 bond, with the charge now located on the B2 side, as shown in
Scheme 6. Note that this peak is very abundant in simulations
while it is not in experiments. This can be due to (at least) two
effects: (i) the mechanism observed in simulations is fast and not
statistical (as it often happens for bond cleavages involving
external groups[12d,12f]) and thus chemical dynamics overestimate
its occurrence here. Another possible explanation (ii) is that the
B2 ion formed can further fragment and this process occurs in time
scales longer than what available in simulations, thus increasing
its relative abundance in simulations. Unfortunately, since
experimentally the intensity of m/z 325 is very weak, it was not
possible to record its fragmentation spectrum. One should also
note that while m/z 325 is overestimated in simulations, the
opposite occurs for m/z 62. These discrepancies illustrate the
difference of conditions between experiments and simulations.
These two ions are associated with the cleavage of the same
Scheme 7. Formation mechanisms of B1 ions (m/z 163) from CID simulations.
The ratio of the different pathways is as follows: a) 16.7%, b) 16.7%, c) 33.2%.
The other mechanisms forming B1 are reported in Scheme S1 of the SI and the
ratio is: S1-a) 16.7%, S1-b) 16.7%.
Based on our chemical dynamics simulations, three types of
fragmentations were observed for the formation of the B1 ion: (1)
the Ca-Cb bond is broken just after collision (charge direct
mechanism) followed by cleavage of C1-O1 bond. At this point we
have obtained an unstable B2 ion which further fragments at the
C1′-O4 position, leading to the m/z 163 ion. This mechanism is
shown in Scheme 7a. (2) The second possibility is that a bond of
the reducing ring is first broken, then promoting the cleavage of
the C1′-O4 bond. An example is given in Scheme 7b and others in
Scheme S1 of the supporting information. (3) Finally, the B1
fragment ion can be obtained by direct cleavage of the C1′-O4
bond after collision (see Scheme 7c). Note that we only obtained
8
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
few trajectories leading to B1 fragment and therefore the ratios
here estimated for the formation of B1 do not necessarily reflect
the actual branching ratios.
Two mechanisms are generally proposed in the literature for
water loss, which differ by the origin of the accompanying
hydrogen atom eliminated. The first one implies the elimination of
a hydrogen atom coming from an adjacent hydroxyl group,
leading to an epoxy form. This mechanism has been proposed in
many studies and is supported by isotope labeling experiments.[38]
The second mechanism, implying the elimination of a hydrogen
atom attached to and adjacent carbon atom and leading to the
formation of a double bond,[36d,39] corresponds to what is observed
in the present simulations. The first step presently observed for
the formation of m/z 368 is indeed the cleavage of one various C-
O(H) bond. Then, when the OH- is expelled and takes a proton
from one nearby C-H, producing H2O and an unsaturated moiety
(Scheme 8). Note that in some cases we observed that the
leaving OH- group takes a proton from the same carbon atom to
which it was bound, leading to a carbene structure (see Scheme
S2 of the Supporting Information). PM3 and B3LYP/6-31G(d,p)
calculations done on two final products associated with loss of OH
from C2, and leading either to alkene (Scheme 8b) or carbene
(Scheme S2a), show that the alkene is more stable by 67 and 72
kcal/mol at PM3 and B3LYP level, respectively, but the carbene
structure, even if higher in energy, is a local minimum (i.e. real
frequencies). Finally, though epoxyde formation was not
observed for water loss during our simulations, it cannot be
discarded. H/D exchange experiments (Figure S5) indeed show
that both types of hydrogens (C-H or O-H) can be eliminated
during sequential losses of water.
One example corresponding to the H2O loss is shown in Figure 3
(panel a). In this case the bonds are broken and formed at the
same time as Ar hits the molecule, thus overestimating the
contribution of C-H hydrogen atoms, which are spatially close to
the OH- leaving group.
A typical example of a complex reaction dynamics is reported in
Figure 3-b, here for the formation of m/z 163 (B1 ion): in this case
the bond scission and formation occur at about 300 fs after that
Ar hits the molecule.
We should also notice that derivatization of the anomeric hydroxyl
group prevents the typical dehydration process involving the
anomeric hydroxyl associated with the bond scission of the C1-O1
bond, weakened by anomeric effect. This process presently
formally corresponds to the formation of the m/z 62 ion.
Alternative routes for dehydration are therefore opened, involving
other hydroxyl groups. Such alternative routes have been
reported in different studies, either about deprotonated methyl-
glycosides,[40] protonated lactose[9-b] or dealing with the interaction
of methyl glycosides with metal ions.[38-a,39-b,41]
Figure 3. Evolution in time of some distances relevant for the formation of m/z
368 (panel a) and 163 (panel b).
Experimental MS/MS/MS. In order to gain more insights about the
possible origin of low mass peaks observed experimentally in
MS/MS spectra but not during the corresponding simulations, we
studied the fragmentation of two product ions whose abundance
was sufficient enough to record their fragmentation spectrum: m/z
368 and m/z 224. MS/MS/MS spectra of β-aminoethyl cellobiose
product ions are reported in Figure 4, and those recorded for β-
aminoethyl-maltose are given in Figure S6. From these
MS/MS/MS spectra, we notice that most of the peaks present in
the precursor ion MS/MS spectrum, are also present in
MS/MS/MS spectra, except m/z 325, 266, 224 and 163. These
latter fragments are thus only obtained from the precursor ion
directly. Observation of peaks in both MS/MS and MS/MS/MS
suggests that they are obtained from a secondary fragmentation
of a first generation fragment ion. But this does not preclude the
eventuality that they can also be produced by a primary
fragmentation, as shown by Klassen’s group using resonance
ejection experiments.[9-a] Finally, some peaks present in
MS/MS/MS spectra are not present in the precursor ion spectra:
m/z 290, 248, 170, 158, 152, 140 and 86. Note that the same
features are observed for β-aminoethyl-maltose. Hereafter, we
will again only discuss mechanisms for β-aminoethyl-cellobiose.
9
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
a)
OH
O
OH
O
OH
OH
O
O
O
O
NH3
OH
HO
NH3
HO
O
OHHO
HO
O
H
H
OH
OH
OH
m/z 386
OH
OH
OH
O
O
O
NH3
H2O
HO
OH
HO
O
OH
m/z 368
b)
OH
OH
O
OH
OH
O
O
O
O
O
NH3
OH HO
HO
NH3
O
HO
OH
HO
O
H
OH
OH
OH
m/z 386
OH
OH
OH
O
O
O
NH3
H2O
HO
OHHO
O
OH
m/z 368
c)
OH
OH
O
OH
O
OH
O
O
O
O
NH3
OHHO
HO
NH3
O
OH
HO
O
OH
OH
OH
H
OH
OH
Figure 4. Experimental MS/MS/MS of β-aminoethyl cellobiose from two
different precursor ions: m/z 368, H2O neutral loss (upper panel), and m/z 224,
Y1, ion (lower panel). Spectra were recorded with a collision energy of 25 eV in
the laboratory framework.
m/z 386
OH
OH
O
O
O
NH3
H2
O
OHHO
O
OH
OH
Most of the peaks present in MS/MS/MS spectra are not obtained
in theoretical CID simulations of the precursor ion m/z 386. This
suggests, together with the analysis of MS/MS and MS/MS/MS
experiments, that these ions can be produced by the successive
fragmentation of a first-generation fragment. However, the low
mass peaks (m/z 74, 62 and 44) are obtained in simulations,
MS/MS experiments and MS/MS/MS experiments from m/z 224
ion.
m/z 368
Scheme 8. Formation mechanisms of m/z 368 as obtained from CID simulations.
The ratio of the different pathways obtained from simulations is as follows: a)
27.3%, b) 18.2%, c) 9.0%. The other mechanisms are reported in Scheme S2
of the SI and the ratio is: S2-a) 27.3%, S2-b) 18.2%.
MS/MS/MS Simulations. In order to understand how peaks
present in the MS/MS spectrum can be obtained from secondary
fragmentations, we have performed MS/MS/MS simulations of the
same two ions studied experimentally, namely m/z 368 and m/z
224. Results, in terms of fragments observed, are summarized in
Table 1, from which simulations can be compared with
experiments. The first difference is that, while the experimental
MS/MS/MS spectra present a significant number of peaks,
simulations only show few fragments: only one product ion, m/z
206, for m/z 368, and three product ions, m/z 206, 104 and 62
when the precursor ion is m/z 224.
The mechanism observed during simulations for the formation of
m/z 206 from m/z 368 is reported in Scheme 9. We can see that
10
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
a)
OH
O
OH
O
OH
OH
OH
O
OH
O
O
O
H
O
HO
HO
HO
HO
H
H
N
O
H
N
NH3
NH3
H
H
H
H
HO
OHHO
O
HO
OH HO
O
O
O
OH
OH
m/z 368
O
H
OH
m/z 224
OH
O
OH
O
OH
HO
HO
OH
HO
NH3
O
HO
OH
O
O
H2O
NH2
OH
H2O
HO
HO
O
NH2
m/z 206
O
OH
Scheme 9. Formation mechanisms of ion m/z 206 using as precursor ion m/z
m/z 206
368 as obtained from CID simulations.
b)
the collision induces the cleavage of C1’-O4 bond followed by a
proton transfer on the leaving –O- group and ultimately formation
of m/z 206. This mechanism is similar to one of the mechanism
accounting for the loss of 162 Da from m/z 386 (scheme 4b). H/D
exchange experiments (Figure S5) show that this fragment ion
OH
O
OH
OH
O
HO
HO
HO
HO
O
H2
O
NH3
NH3
NH3
HO
O
HO
OH
O
O
HO
H
H
m/z 224
was shifted to m/z 211 and therefore incorporated
5
OH
O
exchangeable protons. This result is consistent with the “scheme
8b+scheme 9” reaction sequence resulting in a structure with 5
exchangeable protons. Another important fragment observed in
experiments and not in simulations is m/z 188, which may formally
correspond to water loss from m/z 206. This means that m/z 206
may further react, this occurring at a time-scale that cannot be
sampled by dynamics. Experimentally, the m/z 188 ion is shifted
to m/z 191/192 when performing H/D exchanges (Figure S5), thus
indicating both loss of both D2O and HOD from m/z 211. Again,
both types of hydrogens (C-H or O-H), are eliminated together
with the hydroxyl group. It should be noted that it is not easy to
give a quantitative value of the total energy transferred during
MS/MS/MS experiments, since the first step in fact corresponds
to “in source” fragmentation.
OH
HO
HO
H2O
O
NH3
O
HO
NH3
H2O
O
HO
m/z 206
c)
OH
OH
HO
HO
HO
O
O
NH3
NH3
HO
O
O
OH
OH
m/z 224
OH
O
HO
NH3
Concerning the m/z 224 ion, our simulations lead to three peaks:
m/z 206, 104 and 62. Experimentally, there are two other peaks,
m/z 188 and 86, which correspond to H2O loss from m/z 206 and
104, respectively. They are not present in our simulations likely
for the same reasons: they correspond to further fragmentation of
a product that we cannot obtain due to limitation in time-scale.
The mechanisms deduced from our simulations for the formation
of m/z 206, 104 and 62 are shown in Scheme 10. In Schemes 10a
and 10b we report the two possible ways of forming m/z 206
according to H2O loss: the OH- leaving group takes an H either
from the NH3+ tail (Scheme 10a) or from the CH group to which is
attached (Scheme 10b). Note that the m/z 206 ion formed from
m/z 224 has a different structure with respect to the one formed
from m/z 368.
O
OH
HO
m/z 104
d)
OH
O
OH
O
HO
HO
HO
NH3
O
NH3
HO
O
H
H
OH
OH
m/z 224
OH
HO
O
NH3
HO
Formation of m/z 104 (see Scheme 10c) starts by the C2-C3 bond
cleavage, followed by C4-C5 and C1-O5 bond breaking leading to
three fragments. Finally, the first step leading to m/z 62 is the C1-
O1 bond cleavage, followed by a proton transfer from C1 to O1, as
shown in Scheme 10d.
HO
m/z 62
OH
Scheme 10. Fragmentations observed using m/z 224 as precursor ion in CID
simulations.
11
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
[3]
[4]
[5]
M. Karas, D. Bachmann, U. Bahr, F. Hillenkamp, Int. J. Mass Spectrom.
1987, 78, 53-68.
Conclusions
Z. Takáts, J.M. Wiseman, B.Gologan, R.G.Cook, Science 2004, 5695,
471-473.
In this work, we have studied the collision-induced fragmentation
of a series of disaccharides in the positive-ion mode. Sugars were
modified by adding an aminoethyl or aminopropyl aglycone on the
reducing end in order to unambiguously localize the additional
proton and make calculations on a limited number of initial
structures. Experiments show that the different disaccharides do
not have a specific fragmentation pattern, which could be used to
easily characterize them in a mixture. However, the differences in
intensities (in particular the cellobiose/maltose derivatives vs the
gentiobiose one) could be eventually further exploited to design a
possible identification strategy, notably by coupling a liquid
chromatograph to the mass spectrometer. On the other hand, the
coupling between experiments and collisional chemical dynamics
simulations provides a valuable tool to identify the structure of the
product ions and to suggest some fragmentation mechanisms.
Finally, experimental and theoretical MS/MS/MS data provide
some insights about the fragments coming only from the [M+H]+
ion and those which could be also obtained by subsequent
fragmentation of the first generation of fragments.
Some of the simulated dissociation mechanisms are consistent
with the results deduced from H/D exchange. On the other hand,
these latter experiments also indicate that alternate dissociation
paths not sampled by the simulations, clearly occur, probably
because of the short time scale of our simulations favoring fast
processes over statistical ones. The fact that we used a relatively
simple semi-empirical Hamiltonian representation for the sugars’
dynamics, can be an additional source of discrepancy with
experiments. This is notably reflected by differences in peaks
intensity and absence in simulations of some of the fragments,
which are due to both small time-scale sampling and the use of
approximate potential energy surface description. Another
possibility is that some of the peaks present in MS/MS are
generated by secondary fragmentation. This is why we performed
collisional dynamics of two primary product ions providing, for the
first time to our knowledge, theoretical MS/MS/MS spectra.
J.B. Fenn, M. Mann, C.K. Meng, S. F. Wong, C. M. Whitehouse, Science
1989, 246, 64-71.
[6]
[7]
P.M. Mayer, C. Poon, Mass Spectrom. Rev. 2009, 28, 608-639.
W. Hoffmann, J. Hofmann, K. Pagel, J. Am. Soc. Mass Spectrom. 2014,
25, 471-479.
[8]
a) B. Paizs, S. Suhai, Mass Spectrom. Rev. 2004, 24, 508-548; b) R.
Aebersold, D.R. Goodlett, Chem. Rev. 2001, 101, 269-295; c) A. G.
Harrison, Mass. Spectrom. Rev. 2009, 28, 640-654; d) A.L. Patrick, N.C.
Polfer, Top. Curr. Chem. 2015, 364, 153-182.
[9]
a) M.A. Fentabil, R. Daneshfar, E.N. Kitova, J.S. Klassen, J. Am. Soc.
Mass Spectrom. 2011, 22, 2171-2178; b) B.J. Bythell, M.T. Abutokaikah,
A.R. Wagoner, S. Guan, J.M. Rabus, J. Am. Soc. Mass Spectrom. 2017,
28, 688-703; c) G.E. Hofmeister, Z. Zhou, J.A. Leary, J. Am. Chem. Soc.
1991, 113, 5964-5970; d) V.N. Reinhold, B.B. Reinhold, C.E. Costello,
Anal. Chem. 1995, 67, 1772-1784.
[10] a) T. Yamagaki, K. Fukui, K. Tachibana, Anal. Chem. 2006, 78, 1015-
1022; b) H. Suzuki, A. Kameyama, K. Tachibana, H. Narimatsu, K. Fukui,
Anal. Chem. 2009, 81, 1108-1120.
[11] a) P. de Sainte Claire, W.L. Hase, J. Phys. Chem. 1996, 100, 8190-8196;
b) J. Liu, K. Song, W.L. Hase, J. Chem. Phys. 2003, 119, 3040-3050; c)
E. Martinez-Nuñez, A. Fernandez-Ramos, S.A. Vazquez, J.M.C.
Marques, M. Xue, W.L. Hase, J. Chem. Phys. 2005, 123, 154311; d) O.
Meroueh, Y. Wang, W.L. Hase, J. Phys. Chem. A 2002, 106, 9983-9992.
[12] a) R. Spezia, J.-Y. Salpin, M.-P. Gaigeot, W.L. Hase, K. Song, J. Phys.
Chem. A 2009, 113, 13853-13862; b) R. Spezia, S.B. Lee, A. Cho, K.
Song, Int. J. Mass Spectrom. 2015, 392, 125-138; c) R. Spezia, J.
Martens, J. Oomens, K. Song, Int. J. Mass Spectrom. 2015, 388, 40-52;
d) D. Ortiz, P. Martin-Gago, A. Riera, K. Song, J.-Y. Salpin, R. Spezia,
Int. J. Mass Spectrom. 2013, 335, 33-44; e) E. Rossich Molina, D. Ortiz,
J.-Y. Salpin, R. Spezia, J. Mass Spectrom. 2015, 50, 1340-1351; f) R.
Spezia, A. Cimas, M.-P. Gaigeot, J.-Y. Salpin, K. Song, W.L. Hase, Phys.
Chem. Chem. Phys. 2012, 14, 11724-11736; g) A. Martin-Somer, M.
Yañez, M.-P. Gaigeot, R. Spezia, J. Phys. Chem. A 2014, 118, 10882-
10893.
[13] D. Ortiz, J.-Y. Salpin, K. Song, R. Spezia, Int. J. Mass Spectrom. 2014,
358, 25-35.
[14] a) A. Ballistreri, G. Montaudo, D. Garozzo, M. Giuffrida, G Impallomeni,
S. Daolio, Rapid Comm. Mass Spectrom. 1989, 3, 302-304; b) J.W.
Dallinga, W. Heerma, Biol. Mass Spectrom. 1991, 20, 215-231; c) J.A.
Carroll, L. Ngoka, C.G. Beggs, C.B. Lebrilla, Anal. Chem. 1993, 65,
1582-1587; d) D. Garozzo, M. Giuffrida, G. Impallomeni, A. Ballistreri, G.
Montaudo, Anal. Chem. 1992, 62, 279-286; e) J.A. Carroll, D. Willard,
C.B. Lebrilla, Anal. Chim. Acta 1995, 307, 431-447; f) B. Mulroney, J.C.
Traeger, B.A. Stone, J. Mass Spectrom. 1995, 30, 1277-1283; g) R.E.
March, C.J. Stadey, Rapid Comm. Mass Spectrom. 2005, 19, 805-812.
[15] a) B. Mulroney, J.B. Peel, J.C. Traeger, J. Mass Spectrom. 1999, 34,
544-553; b) J.-Y. Salpin, J. Tortajada, J. Mass Spectrom. 2004, 39, 930-
941.
Acknowledgements
We thank LabEx Charm3at for supporting through action ANR-11-
IDEX-0003-02. R.S. and J.-Y.S. also thank ANR DynBioReact
(Grant number ANR-14-CE06-0029-01) for supporting.
[16] G. Yang, C. Zhu, L. Zhou, J. Mol. Model. 2016, 22, 104.
[17] S. Feng, C. Bagia, G. Mpourmpakis, J. Phys. Chem. A 2013, 117, 5211-
5219.
Keywords: Tandem mass spectrometry • Chemical dynamics •
Carbohydrates • Fragmentation mechanisms
[18] J. Dahmén, T. Frejd, G. Grönberg, T. Lave, G. Magnusson, G. Noori,
Carbohydr. Res. 1983, 116, 303-307.
[19] a) E. F. V. Scriven, K. Turnbull, Chem. Rev. 1988, 88, 297-368; b) S. G.
Alvarez, M. T. Alvarez, Synthesis 1997, 413-414; c) K. Yamamoto, T.
Bruun, J. Y. Kim, L. Zhang, M. Lautens, Org. Lett. 2016, 18, 2644-2647.
[20] a) G. Zemplén, E. Pascu, Ber. Dtsch. Chem. Ges. 1929, 62, 1613-1614;
b) T. K. Lindhorst, in Essentials of Carbohydrate Chemistry and
Biochemistry Eds.: Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim,
2007.
[1]
[2]
a) R.A. Dwek, Chem. Rev. 1996, 96, 683-720; b) I. Bucior, M.M. Burger,
Curr. Opin. Struct. Biol. 2004, 14, 631–637; c) K. Ohtsubo, J.D. Marth,
Cell 2006, 126, 855–867; d) G. W. Hart, R. J. Copeland, Cell 2010, 143,
672–676.
H.R. Morris, M.Panico, M.Barber, R.S.Bordoli, R.D.Sedgwick, A.Tyler,
Biochem. Biophys. Res. Commun. 1981, 101, 623-631.
12
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
[21] I.V. Chernushevich, A.V. Loboda, B.A. Thomson, J. Mass Spectrom.
2001, 36, 849-865.
D.-H. Lu, G.H. Peslherbe, K.N. Swamy, S.R. Vande Linde, H. Wang, R.J.
Wolf, QCPE Bull. 1996, 16.
[22] a) A. Rahman, W.L. Hase, K. Song, J. Wang, S.O. Meroueh in Principles
of Mass Spectrometry Applied to Biomolecules (Eds.: J. Laskin, C.
Lifshitz), Wiley, Hoboken NJ, 2006, pp. 379-474; b) K. Bolton, W.L. Hase,
G.Peslherbe in Modern Methods for Multidimensional Dynamics
Computations in Chemistry (Ed.: D.L. Thompson), World Scientific,
London, 1998, pp. 143-189; c) L. Sun, W.L. Hase, Rev. Comp. Chem.
2003, 19, 79-146.
[31] J.J.P. Stewart, J. Comput.-Aid. Mol. Des. 1990, 4, 1-45.
[32] L. Verlet, Phys. Rev. 1967, 159, 98.
[33] B. Domon, C.E. Costello, Glycoconjugate J. 1988, 5, 397-409.
[34] E. Rossich-Molina, J.-Y. Salpin, R. Spezia, E. Martínez-Núñez, Phys.
Chem. Chem. Phys. 2016, 18, 14980-14990.
[35] A. Martin-Somer, R. Spezia, M.Yanez, Phil. Trans. A 2017, 375,
20160196.
[23] a) J.J.P. Stewart, J. Comp. Chem. 1989, 10, 209-220; b) J.J.P. Stewart,
J. Comp. Chem. 1989, 10, 221-264.
[36] a) G.E. Hofmeister, Z. Zhou, J.A. Leary, J. Am. Chem. Soc. 1991, 113,
5964-5970; b) S. Konig, J.A. Leary, J. Am. Soc. Mass Spectrom. 1998,
9, 1125-1134; c) O.M. Saad, J.A. Leary, J. Am. Soc. Mass Spectrom.
2004, 15, 1274-1286; d) S. Mendonca, R.B. Cole, J.H. Zhu, Y. Cai, A.D.
French, G.P. Johnson, R.A. Laine. J. Am. Soc. Mass Spectrom. 2003,
14, 63-78; e) N. G. Karlsson, H. Karlsson, G. C. Hansson. J. Mass
Spectrom. 1996, 31, 560-572; f) W. Tuting, R. Adden, P. Mischnick. Int.
J. Mass Spectrom. 2004, 232, 107-115.
[24] R. Spezia, A. Martin-Somer, V. Macaluso, Z. Homayoon, S. Pratihar, W.L.
Hase, Faraday Discuss. 2016,195, 599-618.
[25] O. Meroueh, W.L. Hase, J. Phys. Chem. A 1999, 103, 3981-3990.
[26] Y. Jeanvoine, M.-P. Gaigeot, W.L. Hase, K. Song, R. Spezia, Int. J. Mass
Spectrom. 2011, 308, 289-298.
[27] a) A.D. Becke, J. Chem. Phys. 1993, 98, 1372-1377; b) C. Lee, W. Yang,
R.G. Parr, Phys. Rev. B 1988, 37, 785-789.
[37] a) R. Spezia, E. Martinez-Nunez, S. Vasquez, W.L. Hase, Phil. Trans. A.
2017, 375, 20170035 (2017); b) V. Brites, A. Cimas, R. Spezia, N.
Sieffert, J.M. Lisy, M.-P. Gaigeot, J. Chem. Theory Comput. 2015, 11,
871-883, c) X. Ma, W.L. Hase, Phil. Trans. R. Soc. A 2017, 375,
20160204.
[28] Gaussian 09, Revision D.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. ꢀScuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone,
B. Men- nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P.
Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M.
Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery,
Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N.
Kudin, V. N. Staroverov, R. Ko- bayashi, J. Normand, K. Raghavachari,
A. Rendell, J. C. Burant, S. S. Iyen- gar, J. Tomasi, M. Cossi, N. Rega,
N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J.
Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R.
Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G.
Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A.
D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cio- slowski, D. J.
Fox, Gaussian, Inc., Wallingford, CT, 2009.
[38] a) J. M. Berjeaud, F. Couderc, J. C. Promé, Org. Mass Spectrom. 1993,
28, 455; b) K.P. Madhusudanan, J. Mass Spectrom. 2006, 41, 1096-
1104; c) K.P. Madhusudanan, S. Kanojiya, B. Kumar. J. Mass Spectrom.
2005, 1044-1054; d) M. R. Asam, G. L. Glish. J. Am. Soc. Mass Spectrom.
1997, 8, 987-995; e) M. D. Leavell, J. A. Leary. Int. J. Mass Spectrom.
2001, 204, 185-196.
[39] a) B. Mulroney, J.B. Peel, J.C. Traeger, J. Mass Spectrom. 1999, 34,
856-871; b) J.-Y. Salpin, L. Boutreau, V. Haldys, J. Tortajada, Eur. J.
Mass Spectrom. 2001, 7, 321-330; c) J. Zhang, J. S. Brodbelt, Anal.
Chem. 2005, 77, 1761, 1770.
[40] J.C. Promé, H. Aurelle, D. Promé, A. Savagnac, Org. Mass Spectrom.
1987, 22, 6.
[29] a) S. Chapman, D.L. Bunker, J. Chem. Phys. 1975, 62, 2890-2899; b)
C.S. Sloan and W.L. Hase, J. Chem. Phys. 1977, 66, 1523-1533.
[30] a) Y.J. Cho, S.R. Vande Linde, L. Zhu, W.L. Hase, J. Chem. Phys. 1992,
96, 8275; b) X. C. Hu, W.L. Hase, T. Pirraglia, J. Comp. Chem. 1991, 12,
1014-1024; c) W.L. Hase, R.J. Duchovic, X. Hu, A. Komornicki, K. Lim,
[41] J.-Y. Salpin, J. Tortajada, J. Mass Spectrom. 2002, 37, 379-388.
13
This article is protected by copyright. All rights reserved.

10.1002/cphc.201700202
ChemPhysChem
ARTICLE
Entry for the Table of Contents (Please choose one layout)
Layout 1:
ARTICLE
By coupling tandem mass
spectrometry experiments with
chemical dynamics simulations it
was possible to understand
mechanisms for primary and
secondary fragmentation of
disaccharides.
E. Rossich Molina, A. Eizaguirre, V.
Haldys, D. Urban, G. Doisneau, Y.
Bourdreux, J.-M. Beau, J.-Y.
Salpin*, R. Spezia*
Page No. – Page No.
Characterization of protonated
model disaccharides from tandem
mass spectrometry and chemical
dynamics simulations
14
This article is protected by copyright. All rights reserved.