Decomposition of N-hydroxylated compounds during atmospheric pressure chemical ionization

Article abstract of DOI:10.1002/jms.1703

N-Hydroxylated polyamine derivatives were found to decompose during the ionization process of liquid chromatography-atmospheric pressure chemical ionization-mass spectrometry (LC-APCI-MS) experiments. The phenomenon was studied with a model compound, a synthetic N-hydroxylated tetraamine derivative. It was found that reduction, oxidation and water elimination occurred during APCI to generate the corresponding amine, N-oxide, and imine. The investigation further revealed that decomposition of hydroxylamines during APCI depends upon the concentration of the analyte and on the acidity of the solution introduced into the ionization source. The pH-dependence of decomposition was utilized for the development of an MS method that allows for the unambiguous identification of N-OH functionalities. This method was applied for the study of natural products including polyamine toxins fromthevenomof thespider Agelenopsis aperta andmayfoline, a cyclic polyamine derivative of the shrub Maytenus buxifolia. Copyright

Full text of DOI:10.1002/jms.1703

Research Article
Received: 15 September 2009
Accepted: 1 November 2009
Published online in Wiley Interscience: 26 November 2009
(www.interscience.com) DOI 10.1002/jms.1703
Decomposition of N-hydroxylated compounds
during atmospheric pressure chemical
ionization
∗
¨
´
Silvan Eichenberger, Michael Meret, Stefan Bienz and Laurent Bigler
N-Hydroxylated polyamine derivatives were found to decompose during the ionization process of liquid chromatography-
atmospheric pressure chemical ionization-mass spectrometry (LC-APCI-MS) experiments. The phenomenon was studied with
a model compound, a synthetic N-hydroxylated tetraamine derivative. It was found that reduction, oxidation and water
elimination occurred during APCI to generate the corresponding amine, N-oxide, and imine. The investigation further revealed
that decomposition of hydroxylamines during APCI depends upon the concentration of the analyte and on the acidity of the
solution introduced into the ionization source. The pH-dependence of decomposition was utilized for the development of an
MS method that allows for the unambiguous identification of N–OH functionalities. This method was applied for the study of
natural products including polyamine toxins from the venom of the spider Agelenopsisaperta and mayfoline, a cyclic polyamine
c
derivative of the shrub Maytenus buxifolia. Copyright ꢀ 2009 John Wiley & Sons, Ltd.
Keywords: high performance liquid chromatography; artifact; alkaloid; mass spectrometry; natural product; atmospheric pressure
chemical ionization
Introduction
spectrometry (LC-APCI-MS) is widely used for the study of natu-
ral products and drug metabolism. Thus, the lack of awareness
of this APCI reduction could broadly lead to wrong conclusions.
To obtain means to recognize and control artifact formation
from N-hydroxylamines by APCI, we investigated this in-source
reaction with a synthetic N-hydroxylated polyamine derivative in
depth.
Electrospray ionization (ESI) and atmospheric pressure chemi-
cal ionization (APCI) are two of the most important ionization
techniques used in mass spectrometry. These methods are of
particular importance for analytical setups in which mass spec-
trometry is online coupled to liquid chromatography (LC-MS).
Since ESI and APCI are mild ionization methods, they typically
produce quasi-molecular ions of the analytes without fragmen-
tation. Hence, they generate ions that usually provide direct
and unequivocal information about the masses of the sample
molecules. Such information, however, is convoluted by analytes
that undergo fragmentations or decompositions prior to the MS
analysis, e.g. before or during the ionization process. Forma-
tion of such artifacts, especially when their source is not known,
might lead to misinterpretation of MS data. An understanding
of potential decomposition reactions that might occur with ana-
lytes prior to their entry into the MS is particularly important for
more complex investigations, e.g. for the investigation of natural
products.
Artifact formation has already been observed with APCI, in
which the analyte solutions are typically heated to 300–400 ◦C
prior to ionization and therefore prone to thermally induced re-
actions. For example aromatic nitro compounds,[1,2] N-oxides[3–7]
and imines[8], showed partial reduction to the corresponding
amines upon APCI. An analogous reaction was also recently ob-
served with N-hydroxylated compounds in the course of our
investigations of spider venoms.[9–12] In this study we report that
N-hydroxylated polyamine derivatives also undergo reduction
to the respective amines too. This reaction has not previously
been reported in literature. These results are significant be-
cause(1) N-hydroxylaminesarewell-knownconstituentsofnatural
samples and also potential metabolites of drugs and (2) liquid
chromatography-atmospheric pressure chemical ionization-mass
Experimental
Chemicals and sample preparation
High performance liquid chromatography (HPLC) supra grade
acetonitrile (MeCN) from Scharlau (Barcelona, Spain), triflu-
oroacetic acid (TFA) and formic acid (HCOOH) from Fluka
(Buchs, Switzerland) and aqueous solution of NH3 (25%) from
Merck (Darmstadt, Germany) were purchased in the respec-
tive highest qualities. HPLC grade H2O (≤5 ppb total or-
ganic content) was obtained by purification of deionized
H2O with a MilliQ gradient apparatus (Millipore, Milford, MA,
USA). [4-Hydroxy-9-(2-nitrobenzenesulfonyl)-4,9-diazadodecane]-
1,12-diphthalimide (1) was synthesized on solid support
and purified by preparative HPLC.[13] Synthetic (−)-(2S)-
9-hydroxy-2-phenyl-1,5,9-triazacyclotridecan-4-one (=mayfoline)
was obtained from Hesse.[14] Lyophilized Agalenopsis aperta
venom was purchased from Fauna Laboratories Ltd. (Al-
maty, Kazakhstan).
∗
Correspondence to: Laurent Bigler, Institute of Organic Chemistry, University of
Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland.
E-mail: lbigler@oci.uzh.ch
InstituteofOrganicChemistry,UniversityofZurich,CH-8057Zurich,Switzerland
c
J. Mass. Spectrom. 2010, 45, 190–197
Copyright ꢀ 2009 John Wiley & Sons, Ltd.

Decomposition of N-hydroxylated compounds during APCI
NO2
High performance liquid chromatography and mass
spectrometry
O
O
SO2
N
General
N
N
N
OH
O
LC-MS analyses were performed on a Hewlett-Packard 1100 HPLC
system (Hewlett-Packard Co., Palo Alto, CA, USA) fitted with a
HTS PAL autosampler (CTC Analytics, Zwingen, Switzerland), a
Hewlett-Packard 1100 binary pump and a Hewlett-Packard 1100
diode array detector (DAD). The reversed-phase column used
was an Interchim Uptisphere RP C18 column (UP3HDO-20QS,
3 µm, 2.3 × 200 mm, Interchim, Montluc¸on, France). Either a step
gradient or isocratic conditions at flow rates between 150 and
180 µl/min were applied with solvents A and B (solvent A: H2O +
0.1% TFA, solvent B: MeCN + 0.1% TFA).
O
1
(a)
[M+H]+
Intens.
x106
664
r = 0.05
ESI-MS
0.8
0.6
0.4
0.2
0.0
[M+H–16]+
648 (4%)
[M+H–18]+
646 (1%)
[M+Na]+
686 (7%)
The LC system was connected to an EsquireLC quadrupole
ion trap mass spectrometer (Bruker Daltonik GmbH, Bremen,
Germany), equipped with either an ESI or APCI Hewlett-Packard
atmospheric pressure ion (API) source. Conditions for ESI were
nebulizer gas (N2, 40 psi), dry gas (N2, 9 l/min), dry temperature
(300 ◦C), HV capillary (4500 V) and HV EndPlate (−600 V). Condi-
tions for APCI were nebulizer gas (N2, 21 psi), dry gas (N2, 7 l/min),
dry temperature (300 ◦C), APCI temperature (300 ◦C), HV corona
(2870 V), HV capillary (3713 V) and HV EndPlate (−600 V). The MS-
parameters (target mass, compound stability and trap drive) were
optimized for each measurement to obtain highest ion response
and minimal in-source fragmentation. The MS acquisitions were
performed in positive ion mode at normal resolution (0.6 u at half
peak height) and under ion charge control conditions (ICC, target:
10 000). Full scan MS and MS/MS were averaged over 5–8 single
spectra and acquired with a mass window between m/z 50 and
1000. For all MS/MS experiments, the isolation width was set to 1
Da, the fragmentation cut-off to ‘fast calc’, and the fragmentation
amplitude to 1 in the ‘SmartFrag’ mode.
High-resolution Fourier transform (FT) mass spectral data were
obtained with a LTQ-Orbitrap XL mass spectrometer (Thermo
Electron, Bremen, Germany) equipped with a standard ESI source.
The parameters were spray voltage (5 kV), tube lens voltage
(120 V),capillaryvoltage(38 V) andtemperature(275 ◦C).Themass
spectrometerwascalibratedformassaccuracyimmediatelybefore
each measurement according to the manufacturer’s instructions,
the relative mass error being typically lower than 3 ppm
(externally). The high-resolution FT-MS data were additionally
calibrated internally during the measurements with established
lock masses (m/z 429.088735 and 445.120025). Data was acquired
withinamassrangeof m/z 150to1000. Theautomaticgaincontrol
(AGC) target setting for FT-MS experiments was set to 50 000.
Spectra were acquired with a resolution of 60 000 (full width at
half-maximum, FWHM) at m/z 400, and 10 spectra were averaged.
620
640
660
680
700 m/z
APCI-MS
(b)
Intens.
x106
2.5
[M+H–16]+
648
r = 0.65
[M+H]+
664 (76%)
2.0
1.5
1.0
0.5
0.0
[M+H–2]+
662 (27%)
[M+H–18]+
646 (18%)
620
640
660
680
700 m/z
Figure 1. Structure and (a) LC-ESI-MS and (b) LC-APCI-MS of N-hydro-
xylated tetraamine derivative 1 in MeCN/H2O (4 : 6) + 0.1% TFA.
injected into the LC-MS system. A linear gradient from 5 to 20%
B over 40 min at a flow rate of 150 µl/min was applied. The
post-column addition of NH3 to the eluent was performed by the
addition of an aqueous solution of NH3 (10%) at a rate of 20 µl/min
through a Tee located in-between the exit of the column and the
entry of the APCI interface.
Mayfoline
Mayfoline (6.34 µg) was dissolved in MeCN/H2O (1 : 4, 1 ml), and
an aliquot of 5 µl was injected into the LC-MS system under
isocratic conditions with 20% B and a flow rate of 180 µl/min.
The post-column addition of TFA to the sample was performed
by the addition of a mixture of MeCN/H2O/TFA (2 : 6 : 2) at a rate
of 40 µl/min through a Tee located in-between the exit of the
column and the entry of the APCI interface.
Results and Discussion
Synthetic compound 1
Investigations with synthetic N-hydroxylated compounds
For LC-MS analyses, 5 µl of a stock solution of 1 (200 µg) in
MeCN/H2O (1 : 1, 1 ml) was injected at isocratic conditions with
40%ofBandaflowrateof0.18 ml/min. DirectinfusionAPCIexperi-
mentswerecarriedoutbypumping200 µl/minofa30-folddiluted
stock solution of 1 into the mass spectrometer with a syringe in-
fusion pump (Cole-Parmer Instrument Company, Vernon Hills, IL,
USA). For FT-MS experiments, a 10-fold diluted stock solution of 1
was introduced at 6 µl/min using the same syringe infusion pump.
Tetraamine derivative 1 (Fig. 1), which was prepared during our
synthetic pursuit of N-hydroxylated polyamine spider toxins,[13]
was chosen as the model compound for our study of the APCI
behavior of N-hydroxylated secondary amines. The compound
readily undergoes the investigated decomposition reactions and
is also accessible in pure form. Compound 1 also contains an
additional nitroaryl group, allowing the concurrent study of the
decomposition of N-hydroxy and aromatic nitro[1] functionalities.
The investigation of the MS behavior of compound 1 started
with two LC-MS runs performed with either an ESI or an APCI
source under conditions previously applied for the analyses of
polyamine spider toxins.[9,10,15,16] The LC-ESI-MS spectrum of the
Spider venom
Crude lyophilized A. aperta venom (100 µg) was dissolved in
MeCN/H2O (1 : 3, 50 µl) + 0.1% TFA, and an aliquot of 5 µl was
c
J. Mass. Spectrom. 2010, 45, 190–197
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

S. Eichenberger et al.
- H2O
+
+
517
499
NO2
NO2
+H
+H
O
O
O
SO2
N
O
SO2
N
N
N
N
N
N
N
OH
[1+H]+
H
[2+H]+
259
O
O
O
O
- H2O
(a)
(b)
188
188
275
257
Intens.
x105
1.5
Intens.
x104
0.8
259
275
257
MS/MS
of m/z 664
1.0
0.5
0.0
0.6
0.4
0.2
0.0
499
MS/MS
of m/z 648
188
188
646
200
300
400
500
NO2
600 m/z
200
300
400
500
600 m/z
+
+
NO2
+H
+H
515
499
O
O
O
SO2
N
O
SO2
N
N
N
N
N
N
N
O
[3+H]+
273
[4+H]+
257
O
O
O
(c)
(d)
O
- H2O
MS/MS
of m/z 662
255
Intens.
x104
Intens.
x104
1.0
499
3
2
1
0
644
0.8
0.6
0.4
0.2
257
255
273
MS/MS
of m/z 646
229
402 458
188
515
500
186
200
300
400
600 m/z
200
300
400
500
600 m/z
Figure 2. LC-APCI-MS/MS of (a) [1+H]+, (b) [1+H−16]+ = [2+H]+, (c) [1+H−2]+ = [3+H]+ and (d) [1+H−18]+ = [4+H]+ ions with proposed
structures and assignments of relevant fragment ions. For the nitrones 3 and the imines 4, only one of the two isomeric forms is shown.
chromatographic peak of 1 (Fig. 1(a)) showed the expected signal
for the protonated molecule [M+H]+ at m/z 664 (base peak) and
a weaker signal assigned to ions of the type [M+Na]+ (m/z 686,
7%). Two additional signals at m/z 648 (4%) and m/z 646 (1%) were
also observed corresponding to ions of the type [M+H−16]+ and
[M+H−18]+, respectively. Often, ion signals with such low relative
intensities are ignored. These signals, however, became relevant
when APCI was used instead of ESI as the ionization method.
Namely, the spectrum obtained by LC-APCI-MS (Fig. 1(b)) showed
[M+H−16]+ at m/z 648 (base peak) and [M+H−18]+ at m/z 646
(18%) with significantly higher relative intensities. A third signal
thatcannotbeignoredwasfoundatm/z 662(27%), corresponding
to [M+H−2]+ ions. Since it was shown by LC-UV (DAD) and by
NMR that the sample compound was pure, these three additional
ions had to be generated by in-source decomposition of 1, and
were not due to impurities.
In Eqn (1), IDP corresponds to signal intensities of the
i
decomposition products and IQMI to signal intensity of the quasi-
molecular ions. Although r is not the actual molecular ratio of the
three decomposition products to the initial amount of 1, it can
still be regarded as a qualitative measure to describe the extent
of decomposition of 1, allowing, therefore, the characterization of
different experiments.
The structures of the three artifactual ions at m/z 648, 662 and
646 were deduced from data acquired by LC-APCI-MS/MS and
from their accurate masses measured by high-resolution ESI-MS of
sample compound 1. The exact masses of [M+H]+, [M+H−16]+,
and [M+H−18]+ revealed that the artifacts were generated by
formal loss of O and H2O from 1, respectively (Table 1). No high-
resolution ESI-MS data was available for the signal at m/z 662
([M+H−2]+) because of its intensity was too low. However, the
loss of H2 from the parent compound 1 appears to be the most
reasonable process that could generate an artifact responsible for
the respective signal.
The structures of the artifacts formed in the ion source were
deduced from their MS/MS data as amine 2, nitrones 3 and imines
4 (in protonated forms, Fig. 2). The loss of oxygen from 1 could
have principally occurred either at the N–OH position or at the
NO2/SO2 groups of the nosyl portion. The data revealed, however,
that reduction only took place at the hydroxylamine functionality,
thus forming product 2. While fragment ions at m/z 275 were
A term r with the following equation is introduced to estimate
the extent of the overall decomposition of 1 occurring in the
different experiments.
ꢀ
IDP
I[M+H−2] + I
+ I
+
+
+
i
[M+H−18]
I[M+H[]M++H−I16]
r =
ꢀ
=
IDP
i
(1)
+
+
[M+H−2]
IQMI
+
+I[M+H−16] + I
+
+
[M+H−18]
c
www.interscience.wiley.com/journal/jms
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 190–197

Decomposition of N-hydroxylated compounds during APCI
mass spectra of differently concentrated analyte solutions can
be measured when reasonably broad chromatographic peaks are
obtained. LC-APCI-MS of 1 (1 µg) afforded a chromatographic
peak sufficiently broad to allow its splitting into several segments
of 0.1 min, in which each is characterized by a different
analyte concentration. The averaged analyte concentration of
the segments was estimated on the basis of the relative segment
areas (Fig. 3). The MS of the two highlighted segments with low
analyte concentration, at the beginning and at the end of the
peak (∼1.8 and 0.8 µM), showed rather high degrees of sample
decomposition (r = 0.86 at tR 20.6–20.7 min and r = 0.90 at
21.7–21.8 min, respectively). Still prominent but significantly less
sample decomposition was observed at the time segment taken
at the peak maximum (r = 0.65 at tR 21.0–21.1 min). Thus, less
decomposition was observed when more highly concentrated
solutions were investigated. This effect is general as it was
also observed in other measurements, e.g. in those performed
with natural samples of polyamine spider toxins. These results
suggest, in accordance with previously described studies, that
APCI decompositions are surface-supported processes, which are
controlled in their extent by the limited surface of the APCI
interface.[8]
Table 1. Exact masses obtained from high-resolution ESI-FT-MS
investigation of 1
Elemental
m/z
m/z
Ion signal
composition
theoretical
measured
ꢀ ppm
[1+H]+
C32H34O9N5S
C32H34O8N5S
C32H32O8N5S
664.20717
648.21226
646.19661
664.20727
648.21248
646.19692
0.1
0.4
0.5
[1+H−O]+
[1+H−H2O]+
observed for compound 1, the respective signal – which should
be the same if deoxygenation would occur at the NO2 or SO2
groups – was not found in the MS/MS of 2 (Fig. 2(b)). Instead,
a signal was registered at m/z 259, which is consistent with an
amine instead of the hydroxylamine functionality. The fact that
no ion response at m/z 275 was observed also revealed that
deoxygenation of NO2 to NO, as observed for aromatic nitro
compounds,[1] was too slow to compete with the deoxygenation
of the hydroxylamines. The eliminations of H2 (formation of 3) and
of H2O (formation of 4) from sample compound 1 also occurred
with the N-OH functional group rather than with the other groups
containedinthemolecule. Analogouslyto2, theMS/MSofartifacts
3 and 4 showed no ion signals at m/z 275 but signals at m/z 273
and m/z 257, respectively, which are diagnostic for decomposition
located in the ‘left-part’ of the molecules (Fig. 2(c) and (d)).
Dependence of the APCI decomposition of hydroxylamines on the
solvent acidity
To study the pH-dependence of the APCI decomposition of
N-hydroxylated secondary amines, compound 1 was dissolved
either in pure MeCN/H2O (1 : 1) or in MeCN/H2O (1 : 1) admixed
with TFA, HCOOH or TFA followed by NH3 until neutralized and
directly introduced into the APCI-MS. The results are shown with
Dependence of the APCI decomposition of hydroxylamines on the
sample concentration
The online coupling of HPLC to MS allows fast acquisition of MS
data of an analyte directly after column chromatography. Thus,
Intens.
x106
2.5
2.0
1.5
1.0
0.5
0.0
Intens.
x105
tR 20.6-20.7 min
tR 21.0-21.1 min
648
648
[M+H]+
664
6
4
2
0
r = 0.86
r = 0.65
662
[M+H]+
664
662
646
646
640
650 m/z 660
670
640
650 m/z 660
Intens.
670
x106
4
tR 21.7-21.8 min
mAU
20
648
av. conc.: 14.3 µM
UV (254 nm)
3
2
1
0
r = 0.90
15
10
5
662
[M+H]+
646
664
640
650 m/z 660
670
av. conc.: 0.8 µM
av. conc.: 1.8 µM
0
20.0
20.5
21.0
21.5
22.0
min
Figure 3. Concentration dependence of the APCI reduction of hydroxylamines shown with segments of a LC peak of 1.
c
J. Mass. Spectrom. 2010, 45, 190–197
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

S. Eichenberger et al.
molecule is rather likely. It should thus be possible to distinguish
between artifacts and real N–OH-containing sample compounds
by inhibition of the APCI decomposition. On the other hand, if no
ions of the type [M+H−H2]+ and [M+H−O]+ can be found for a
compound – not even when the sample solution is acidified – an
N–OH group is most likely not present in the sample molecules.
The pH-dependence of the APCI decomposition of hydroxy-
lamines can thus be used for the unequivocal recognition of the
N–OH functionality as is illustrated below with the analysis of
native polyamine toxins found in spider venom and of mayfoline,
a natural cyclic polyamine derivative.[17]
(a)
Intens.
x106
0.8
MeCN/H2O 1:1
= 0.10
[
M
+H]+
664
r
0.6
0.4
0.2
0.0
662
6%
648
3%
646
2%
(b)
Intens.
MeCN/H2O 1:1 + 0.1% HCOOH
= 0.38
[
M
+H]+
664
x105
4
r
LC-APCI-MSAnalysisoftheVenomfromtheSpiderAgelenopsis
aperta
3
648
29%
662
27%
2
Some years ago, acylpolyamines of the venoms of A. aperta[10]
and of Paracoelotes birulai[9] were characterized by means of LC-
APCI-MS and -MS/MS, and both venoms were found to contain
N-hydroxylated polyamine derivatives. Due to incomplete data,
some of the constituents of A. aperta were not amenable to
complete structural elucidation. They were regarded as structural
isomers of other identified components, but now, with the
knowledge about the in-source decomposition of N-hydroxylated
secondary amines, they could also have represented artifacts.
Thus, the venom of A. aperta was reinvestigated by LC-APCI-MS.
The 2D-plot in Fig. 5(a) summarizes the ion responses of
all constituents of the acylpolyamine fraction of A. aperta
obtained with the LC-MS setup already used previously. The
spots in the chromatogram represent ion responses registered
in dependence on retention times (abscissa) and m/z values
(ordinate). Remarkable is the occurrence of signal doublets and
triplets characterized by the same tR and by m/z values differing
by 16 u. Interpretation of the MS/MS data of the respective
ions revealed that the triplets represent ions of di-, mono-,
and non-N-hydroxylated polyamine derivatives and the doublets
of mono- and non-N-hydroxylated polyamine derivatives, in
each case sharing the polyamine backbones. Applying the
acquired knowledge of the APCI reduction of hydroxylamines,
the signals of the ‘deoxygenated’ structures were most likely
due to APCI artifacts rather than the response of real sample
compounds. For example, the two ion signals at m/z 449 and
433 of the doublet B at tR = 25.9 min can be interpreted
as the protonated molecules and the respective protonated
‘deoxygenated’ decomposition products [M+H−O]+ of the co-
eluting isomeric mono-N-hydroxylated toxins 8 + 9.[10] The signal
at m/z 433 could therefore be generated by partial reduction of 8
+ 9 during the APCI process (Fig. 5(a)).
646
6%
1
0
(c)
Intens.
x105
5
MeCN/H2O 1:1 + 0.1% TFA
[
M
+H]+
664
648
84%
r
= 0.54
4
3
2
1
0
662
21%
646
11%
(d)
Intens.
[
M
+H]+
664
MeCN/H2O 1:1 + 0.1% TFA
+ add. of NH3
x106
1.0
0.5
0.0
r
= 0.04
648
2%
662
1%
646
1%
600 610 620 630 640 650 660 670 680 690 m/z
Figure 4. Directinfusion APCI-MSexperiments performed with compound
1 dissolved (a) in pure MeCN/H2O (1 : 1), or in MeCN/H2O (1 : 1) admixed
with (b) HCOOH, (c) TFA or (d) TFA followed by addition of NH3 until
neutralized.
the spectra in Fig. 4. Increased acidity of the sample solution led to
morepronounceddecomposition.Whilealmostnodecomposition
of compound 1 was observed when the sample was introduced
into the APCI-MS dissolved in the neutral mixture of MeCN/H2O
(1 : 1) (r = 0.1, Fig. 4(a)), the decomposition of 1 increased
markedly in presence of HCOOH or TFA (r = 0.38 and 0.54,
Fig. 4(b) and (c), respectively). Decomposition was highest when
0.1% TFA was used as the additive and therefore under the
standard conditions used for the chromatographic separation of
spider toxins. Decomposition of 1 was inhibited by the addition
of aqueous NH3 (to the initially acidic solution) as is recognized
from spectrum c (r = 0.04, Fig. 4(d)). This latter result confirms the
fact that the decomposition of 1 occurs in the ion source and not
already before its entry into the HPLC.
The pH-dependence of the APCI decomposition of hydroxy-
lamines can be used as a means to identify the N–OH functionality
of a molecule. If a compound shows [M+H−H2]+ and, particu-
larly, [M+H−O]+ signals in APCI-MS spectra ([M+H−H2O]+ and if
the formation of these ions can be enforced or inhibited by the
addition of acid or base to the sample solution prior to its introduc-
tion into the instrument, the presence of the N–OH group in the
To prove that the doublets and triplets recorded in the 2D-
chromatogram arise in fact from mono- or di-N-hydroxylated
parent compounds, an LC-APCI-MS experiment of the toxin mix-
ture was performed under the same optimized chromatographic
conditionsusedbefore(MeCN/H2Ogradient+ 0.1%TFA), butwith
post-column addition of NH3 to inhibit the APCI decompositions.
The respective 2D-plot is shown in Fig. 5(b). It is readily recognized
thatthedoubletsandtripletsfoundinFig. 5(a)largelydisappeared,
which allows the conclusion that the several vanished peaks arose
from artifacts rather than from native compounds.
Hence, also N-hydroxylated acylpolyamines of spider venom
underwent in-source decomposition during LC-APCI-MS experi-
ments analogously to the synthetic N-hydroxylated compound 1.
The re-analysis of the venom of A.aperta revealed that compounds
generated by APCI decomposition were previously misinterpreted
as constituents of the venom.
c
www.interscience.wiley.com/journal/jms
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 190–197

Decomposition of N-hydroxylated compounds during APCI
OH
OH
OH
N
H
H
H
H
H
H
H
NH2
NH2
N
N
N
N
N
N
N
n
m
l
n
m
l
O
O
5 n=m=1, l=2
6 n=l=1, m=2
7 n=2, m=l=1
N
H
N
8 n=m=1, l=2
9 n=l=1, m=2
H
Intens.
[M+H–O]+
433.4
[M+H]+
433.4
Intens.
x105
1.0
x105
4
LC-APCI-MS of 8+9
LC-APCI-MS of 5–7
0.8
0.6
0.4
0.2
0.0
3
2
1
0
RT 25.9 min
RT 23.5 min
[M+H]+
449.3
430
440
450
m/z
420
430
440
m/z
mAU
30
UV (254 nm)
20
10
0
(a) m/z
LC-APCI-MS
t
500
450
400
d
t
d
d
t
d
m/z 449
m/z 433
A
B
d
t
d
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
37.5 min
(b)
m/z
500
LC-APCI-MS
with post-column
addition of NH3
B’
450
400
m/z 449
m/z 433
A’
Figure 5. 2D-plot of an LC-APCI-MS run of A. aperta venom and the corresponding UV-chromatogram detected at λ = 254 nm using (a) MeCN/H2O +
0.1% TFA as the mobile phase and (b) the same conditions but with post-column addition of NH3. d and t designate a signal doublet and a signal triplet,
respectively.
c
J. Mass. Spectrom. 2010, 45, 190–197
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

S. Eichenberger et al.
signal, the [M+H]+, was recorded. Thus, mayfoline represents an
N-hydroxylated compound with little tendency to undergo APCI
reduction.
O
N
H
N
OH
Since APCI deoxygenation was intended to be taken as
a conclusive argument for the identification of the N–OH
functionality within a sample molecule, it was tested if APCI
deoxygenationofmayfolinecanbeenforcedsothattherespective
ions are unquestionably recognized. This, in fact, could be affected
by the post-column addition of a mixture of MeCN/H2O/TFA
(2 : 6 : 2) to the analyte flow. Under these conditions, the MS
revealed signals of the decomposition products at m/z 290
([M+H−H2]+, 51%), m/z 276 ([M+H−O]+, 34%) and m/z 274
([M+H−H2O]+,83%)withsignificantlyhigherintensities(r = 0.63,
Fig. 6(b)). Decomposition was not at all observed with a sample
of synthetic deoxymayfoline,[14] which has been treated the same
way (data not shown).
N
H
Ph
mayfoline
(a)
[M+H]+
292
Intens.
x105
4
LC-APCI-MS
r = 0.21
[M+H–O]+
276 (12%)
[M+H–H2]+
290 (7%)
3
2
1
0
[M+H–H2O]+
274 (8%)
230 240 250 260 270 280 290 300 310 m/z
(b)
[M+H]+
292
Intens. LC-APCI-MS
with p.c. add. of TFA
[M+H–H2O]+
274 (83%)
x105
3
Conclusions
r = 0.63
[M+H–O]+
276 (34%)
2
1
0
The investigation of the various N-hydroxylated amines revealed
thatN–OH-containingcompoundscharacteristicallyformartifacts
upon APCI. The corresponding decomposition reactions are
strongly pH-, and, to a lesser degree, concentration-dependent.
Thedegreeofdecompositionalsodependsontheexactmolecular
structures of the analytes. For all compounds investigated, APCI
decomposition could be enforced by the addition of sufficient
acid to the analyte solution and suppressed by the addition of
base.
[M+H–H2]+
290 (51%)
230 240 250 260 270 280 290 300 310 m/z
Figure 6. LC-APCI-MSofmayfoline(a) withMeCN/H20gradient+0.1%TFA
at 180 µl/min and (b) with MeCN/H20 gradient + 0.1% TFA at 180 µl/min
and post-column addition of MeCN/H2O/TFA (2 : 6:2, 20 µl/min).
Understanding in-source decomposition of N-hydroxylated
amines can help avoid misinterpretation of MS data from samples
that contain N-hydroxylated analytes (particularly of LC-MS data
for which no additional analytic information is available). The pH-
dependence of the APCI decomposition can be applied in two
ways: it can be used (1) to distinguish unavoidable artifacts from
native compounds, as shown with the investigation of the spider
venom of A. aperta, and (2) for the conclusive identification of
N–OH functionalities within a compound.
TheinsightsgainedwiththeMSinvestigationofhydroxylamines
are of relevance for synthetic organic chemistry as well. For
instance, decompositions of hydroxylamines, prepared by Cope
elimination of N-oxides, were largely suppressed by the addition
of base to the reaction medium.[13]
The eye-catching doublet- and triplet-patterns recognized in
the 2D-chromatogram in Fig. 5(a) can be taken as a count for
the number of N–OH groups in a molecule. Thus, two 2D-plots
of N-hydroxylated (or potentially N-hydroxylated) compounds
measured under acidic and basic APCI conditions can reveal
the presence and number of N–OH groups in an unknown
compound. Since C-hydroxy groups are not reduced during APCI-
MS, this ionization mode can be used to distinguish N- from
C-hydroxylation as shown, for instance, with the APCI-MS of the
non-N-hydroxylated derivatives 5–7 that did not show signals
for deoxygenated products. These results are in accordance with
thoseobtainedwiththemethodreportedtodifferentiateN-oxides
from C-hydroxylated compounds.[3]
Acknowledgements
APCI-MS analysis of mayfoline
The authors thank Prof. Dr M. Hesse for the sample of mayfoline,
the Functional Genomics Center Zurich for providing the LTQ-
Orbitrap XL measurements, Prof. Dr Nathan Luedtke for linguistic
help and the Swiss National Science Foundation for their generous
financial support.
Mayfoline (Fig. 6) is a cyclic N-hydroxylated spermidine alkaloid
isolated from the shrub Maytenus buxifolia,[17] and it was
synthesized some years ago by Hesse et al.[14] Due to the N–OH
functionality, mayfoline was expected to show the same type
of APCI decomposition as the model compound 1 and the
polyamine spider toxins described earlier. However, only a little
decompositionwasobserved(r = 0.21, Fig. 6(a))whenasampleof
thenaturalproductwasanalyzedbyLC-APCI-MSunderthenormal
conditions applied for the separation of polyamine derivatives
(MeCN/H2O gradient + 0.1% TFA). The expected signals for the
protonated decomposition products, [M+H−H2]+, [M+H−O]+
and [M+H−H2O]+, were still observed but with unexpected low
relative intensities of 7, 12, and 8% despite the high acidity of
the solvent mixture used for the APCI-MS experiment. When
a sample of mayfoline was brought into the ion source in
neutral solvent (MeCN/H2O 1 : 1), a spectrum with only a single
References
[1] T. Karancsi, P. Slegel. Reliable molecular mass determination of
aromatic nitro compounds: elimination of gas-phase reduction
occurring during atmospheric pressure chemical ionization. J. Mass
Spectrom. 1999, 34, 975.
[2] E.A. Straube, W. Dekant, W. Voelkel. Comparison of electrospray
ionization, atmospheric pressure chemical ionization, and
atmospheric pressure photoionization for the analysis of
dinitropyrene and aminonitropyrene LC-MS/MS. J. Am. Soc. Mass
Spectrom. 2004, 15, 1853.
c
www.interscience.wiley.com/journal/jms
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 190–197

Decomposition of N-hydroxylated compounds during APCI
[3] R. Ramanathan, A.D. Su, N. Alvarez, N. Blumenkrantz, S.K.
Chowdhury, K. Alton, J. Patrick. Liquid chromatography/mass
spectrometry. Methods for distinguishing N-oxides from
hydroxylated compounds. Anal. Chem. 2000, 72, 1352.
[9] S. Chesnov, L. Bigler, M. Hesse. The spider Paracoelotes birulai:
detection and structure elucidation of new acylpolyamines by
on-line coupled HPLC-APCI-MS and HPLC-APCI-MS/MS. Helv. Chim.
Acta 2000, 83, 3295.
[4] W. Tong, S.K. Chowdhury, J.-C. Chen, R. Zhong, K.B. Alton, J.E.
Patrick. Fragmentation of N-oxides (deoxygenation) in atmospheric
pressure ionization: investigation of the activation process. Rapid
Commun. Mass Spectrom. 2001, 15, 2085.
[5] S.-N. Lin, S.L. Walsh, D.E. Moody, R.L. Foltz. Detection and time
course of cocaine N-oxide and other cocaine metabolites in human
plasma by liquid chromatography/tandem mass spectrometry.
Anal. Chem. 2003, 75, 4335.
[6] D.M. Peiris, W. Lam, S. Michael, R. Ramanathan. Distinguishing
N-oxide and hydroxyl compounds: impact of heated
capillary/heated ion transfertube in inducingatmospheric pressure
ionization source decompositions. J. Mass Spectrom. 2004, 39, 600.
[7] S. Ma, S.K. Chowdhury, K.B. Alton. Thermally induced N-to-O
rearrangement of tert-N-oxides in atmospheric pressure chemical
ionization and atmospheric pressure photoionization mass
spectrometry: differentiation of N-oxidation from hydroxylation
and potential determination of N-oxidation site. Anal. Chem. 2005,
77, 3676.
[8] V. Kertesz, G.J. Van Berkel. Surface-assisted reduction of aniline
oligomers, N-phenyl-1,4-phenylenediimine and thionin in
atmospheric pressure chemical ionization and atmospheric
pressure photoionization. J. Am. Soc. Mass Spectrom. 2002, 13,
109.
[10] S. Chesnov, L. Bigler, M. Hesse. The acylpolyamines from the venom
of the spider Agelenopsis aperta. Helv. Chim. Acta 2001, 84, 2178.
[11] S. Chesnov, L. Bigler, M. Hesse. Detection and characterization of
natural polyamines by high-performance liquid chromatography-
atmospheric pressure chemical ionization (electrospray ionization)
mass spectrometry. Eur. J. Mass Spectrom. 2002, 8, 1.
[12] N. Manov, M. Tzouros, S. Chesnov, L. Bigler, S. Bienz. Solid-phase
synthesis of polyamine spider toxins and correlation with the
natural products by HPLC-MS/MS. Helv. Chim. Acta 2002, 85, 2827.
[13] M. Me´ret, S. Bienz. Efficient and flexible solid-phase synthesis of
N-hydroxypolyamine derivatives. Eur. J. Org. Chem. 2008, 33, 5518.
[14] P. Kuehne, A. Linden, M. Hesse. Asymmetric synthesis of the
alkaloids mayfoline and N(1)-acetyl-N(1)-deoxymayfoline. Helv.
Chim. Acta 1996, 79, 1085.
[15] M. Tzouros, S. Chesnov, S. Bienz, M. Hesse, L. Bigler. New linear
polyamine derivatives in spider venoms. Toxicon 2005, 46, 350.
[16] M.Tzouros,N.Manov,S.Bienz,L.Bigler.Tandemmassspectrometric
investigation of acylpolyamines of spider venoms and their 15N-
labeled derivatives. J. Am. Soc. Mass Spectrom. 2004, 15, 1636.
[17] H. Ripperger. Mayfoline, a novel alkaloid from Maytenus buxifolia.
Phytochemistry 1980, 19, 162.
c
J. Mass. Spectrom. 2010, 45, 190–197
Copyright ꢀ 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms