Viscosalines B1,2 and E1,2: Challenging new 3-alkyl pyridinium alkaloids from the marine sponge Haliclona viscosa

Article abstract of DOI:10.1002/chem.201101362

Four new 3-alkyl pyridinium alkaloids, the viscosalines B₁ (1 a), B₂ (1 b), E₁ (2 a), and E₂ (2 b), were isolated from the Arctic sponge Haliclona viscosa. The structure elucidation of these isomeric compounds was challenging due to ambiguous fragments that derive during "standard" mass spectrometric fragmentation experiments. The final structure elucidation relied on the use of a combination of synthesis, liquid chromatography, and mass spectrometry. Three different mass spectrometers were used to differentiate between the synthetic structural isomers: a time-of-flight (TOF) mass spectrometer and two ion-trap mass spectrometers with different ion-transfer technologies (i.e., skimmer versus funnel optics). Although at first none of the spectrometers returned spectra that permitted structure elucidation, all three mass spectrometers provided analysis that successfully differentiated between the isomers after thorough method optimization. The use of in-source collision-induced dissociation (CID) with the ion trap and TOF instrument returned the most interesting results. The mode of fragmentation of the viscosalines under different experimental conditions is described herein. After successful optimization of the mass spectrometric method applied, the chromatographic method was improved to distinguish the previously inseparable isomers. Finally, both the liquid chromatography and mass spectrometric methods were applied to the natural products and the results compared to those from the synthetic compounds. Copyright

Full text of DOI:10.1002/chem.201101362

DOI: 10.1002/chem.201101362
Viscosalines B1,2 and E1,2: Challenging New 3-Alkyl Pyridinium Alkaloids
from the Marine Sponge Haliclona viscosa
Gesine Schmidt, Christoph Timm, Achim Grube, Christian A. Volk, and
Matthias Kçck*[a]
Abstract: Four new 3-alkyl pyridinium
alkaloids, the viscosalines B1 (1a), B2
(1b), E1 (2a), and E2 (2b), were isolat-
ed from the Arctic sponge Haliclona
viscosa. The structure elucidation of
these isomeric compounds was chal-
lenging due to ambiguous fragments
that derive during “standard” mass
spectrometric fragmentation experi-
ments. The final structure elucidation
relied on the use of a combination of
synthesis, liquid chromatography, and
mass spectrometry. Three different
mass spectrometers were used to differ-
entiate between the synthetic structural
isomers: a time-of-flight (TOF) mass
spectrometer and two ion-trap mass
spectrometers with different ion-trans-
fer technologies (i.e., skimmer versus
funnel optics). Although at first none
of the spectrometers returned spectra
that permitted structure elucidation, all
three mass spectrometers provided
analysis that successfully differentiated
between the isomers after thorough
method optimization. The use of in-
source collision-induced dissociation
(CID) with the ion trap and TOF in-
strument returned the most interesting
results. The mode of fragmentation of
the viscosalines under different experi-
mental conditions is described herein.
After successful optimization of the
mass spectrometric method applied,
the chromatographic method was im-
proved to distinguish the previously in-
separable isomers. Finally, both the
liquid chromatography and mass spec-
trometric methods were applied to the
natural products and the results com-
pared to those from the synthetic com-
pounds.
Keywords: alkaloids · b-alanine ·
in-source collision-induced dissocia-
tion · mass spectrometry · natural
products · total synthesis
Introduction
Piperidine and pyridine (pyridinium) alkaloids are widely
distributed in marine sponges of different genera.[1] Espe-
cially, the order Haplosclerida (genera Haliclona, Xesto-
spongia, and Amphimedon) proved to be a rich source of al-
kaloids containing 1,3-dialkyl pyridine (pyridinium) or pi-
peridine motifs.[2] Many of these compounds have rather
complex structures, such as the manzamines[2b,c,3] and kera-
maphidins.[3b,4] Members of this alkaloid class with a lower
structural complexity are the halitoxins,[5] haliclamines,[6]
and cyclostellettamines.[7] In 2004, Volk and Kçck isolated
the first 3-alkyl pyridinium alkaloid containing a b-amino
acid (viscosaline C)[8] from the Arctic sponge Haliclona vis-
cosa.[2g]
Scheme 1. Viscosalines B1 (1a), B2 (1b), E1 (2a), E2 (2b), and C (3).
Haliclona viscosa, posed several challenges due to their iso-
meric constitution (Scheme 1). Recent studies on the frag-
mentation pattern of related 3-alkyl pyridine alkaloids, such
as cyclostellettamines and haliclamines, proved that this
structural class shows a unique fragmentation pattern de-
pending on the ion selected for fragmentation and the mass
spectrometric method used.[9] This knowledge provided the
decisive idea for fragmentation studies and therefore assist-
ed in the structure elucidation of the new compounds. The
study has extended our knowledge of the fragmentation of
different 3-alkyl pyridine compounds to include the viscosa-
lines and aims to facilitate their future identification during
standard liquid-chromatography–mass spectrometry screen-
ing experiments.
During the present study, the isolation and structure eluci-
dation of four new viscosalines B1 (1a), B2 (1b), E1 (2a),
and E2 (2b), which also originated from the Arctic sponge
[a] Dr. G. Schmidt, Dr. C. Timm, Dr. A. Grube, Dr. C. A. Volk,
Priv.-Doz. Dr. M. Kçck
Alfred-Wegener-Institut fꢀr Polar- und Meeresforschung
in der Helmholtz-Gemeinschaft
Am Handelshafen 12, 27570 Bremerhaven (Germany)
Fax : (+49)471-4831-1425
Results and Discussion
Investigation of the nBuOH fraction of the crude extract of
the sponge Haliclona viscosa by HPLC–HRMS initially re-
vealed the compounds haliclamines C–F,[6b,c] haliclocyclins C
E-mail: mkoeck@awi.de
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201101362.
8180
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8180 – 8189

FULL PAPER
and F,[10,11] cyclostellettamine C,[7a] viscosamine C,[2f] and vis-
cosaline C (3).[2g] A re-examination of this nBuOH fraction
resulted in the isolation of two unidentified compounds with
molecular masses of m/z 594.4989 and 622.5320, that is, for
1 and 2, respectively. In addition to the singly charged mo-
lecular ion (termed M1 for all compounds), the full scan
spectrum of the two new compounds 1 and 2 showed doubly
and triply charged pseudomolecular ions M2 and M3, re-
spectively (Table 1). The molecular masses corresponded to
1 (M1: m/z 594.5; fragments at m/z 246.2 and 260.2) and C13
and C14 for 2 (M1: m/z 622.5; fragments at m/z 260.2 and
274.2). By adopting the nomenclature of the cyclostelletta-
mines,[9a] the two compounds were named viscosaline B (1)
and viscosaline E (2). The different chain lengths left two
possibilities for the constitution of the compounds, that is,
with the b-alanine unit connected either to the longer or
shorter alkyl chain; viscosaline B1 (1a; 12/13) with the
longer alkyl chain connected to the b-alanine moiety, visco-
saline B2 (1b; 13/12) with the shorter chain connected to the
b-alanine moiety, and viscosalines E1 (2a; 13/14) and E2
(2b; 14/13) accordingly. NMR spectroscopy was not suffi-
cient to determine the structure of the viscosalines because
the majority of the methylene groups occurred in one signal
in the 1H NMR spectrum due to the long alkyl chains
(Table 2). Thus, differentiation of the individual methylene
groups was impossible by NMR spectroscopy. As shown for
other 3-alkyl pyridinium alkaloids,[9] the application of dif-
ferent mass spectrometric techniques was essential for their
structure elucidation. In the case of 1 and 2, the challenge
was to assign the correct chain length in the natural prod-
ucts to the b-alanine and pyridine moieties, thus determining
which of the isomers was present in the natural extract.
In principle, it should have been possible to differentiate
the two constitutional isomers of either viscosaline com-
pound by collision-induced dissociation mass spectrometry
(CID–MSn), and the fragments should have been predicta-
ble if a similar fragmentation pattern to that observed for
the cyclostellettamines[9a] occurred in the viscosalines. An
onium fragmentation pattern at
Table 1. Molecular ions and most pronounced fragments observed in the
full-scan mass spectra of the viscosalines B (1) and E (2).[a]
Viscosaline
G
Fragments
M1
M2
A
M3
ACHTUNGTRENNUNG
[M+2H]3+
F1 and F2
[M]+
B (1)
E (2)
C (3)
594.5
622.5
608.5
297.8
311.8
304.8
198.8
208.2
203.5
246.2
260.2
260.2
260.2
274.2
260.2
[a] Values are given as m/z.
the empirical formulas of C38H64N3O2 and C40H68N3O2
(Dm=0.7 and 2.2 ppm, respectively). A comparison of the
MS2 spectra of the two new compounds with the MS2 spec-
trum of 3 indicated very similar structures (Figure 1); how-
ever, in contrast to 3, which contains two C13 alkyl chains,
these new compounds had to have unequal alkyl chain
lengths. After an initial analysis of the MS spectra, fragmen-
tation, and comparison with the MS/MS spectrum of 3, it
was very likely that their chain lengths were C12 and C13 for
an early stage of the fragmenta-
tion process would result in two
fragments for each isomer
(Scheme 2), that is, m/z 246.2
and 349.3 for 1a (12/13) or m/
z 260.2 and 335.3 for 1b (13/
12), m/z 260.2 and 363.3 for 2a
(13/14), or m/z 274.2 and 349.3
for 2b (14/13). However, as al-
ready reported in our first in-
vestigation of viscosaline C
(3),[2g] the compound easily
cleaves acrylic acid (m/z 72)
from the b-alanine moiety
during MS2 (Figure 1). This loss
is followed by an immediate
onium fragmentation, as ob-
served for the cyclostellett-
AHCTUNGTERGaNNUN mines, thus still allowing for
differentiation of the isomers.
However, a cleavage of an NH3
group concurs with the cleavage
into the onium fragments F1
and F2 in MS3, where F1 repre-
sents the alkyl pyridine frag-
ment that results from bond
cleavage at the nitrogen atom
Figure 1. Comparison of the MS2 spectra of the original viscosaline C (3, middle) and the two new natural vis-
cosalines B (1, top) and E (2, bottom). The peak pattern indicates the same structural class for the three com-
pounds. In comparison to viscosaline C (3), viscosaline B (1) has one alkyl chain that is shorter by one methyl-
ene group (top) and viscosaline E (2) has an alkyl chain that is longer by one methylene group (bottom). Stars
indicate the precursor ions selected for fragmentation.
Chem. Eur. J. 2012, 18, 8180 – 8189
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8181

M. Kçck et al.
z 260.2).[2g] The new viscosalines result in the same fragmen-
tation pattern of fragments of unequal mass, thus allowing
alkyl chains of 12 or 13 methylene units for 1 and 13 or 14
methylene units for 2 to be calculated. Although fragments
F1 and F2 have disparate masses in 1 (and 2), the fragment
masses are not different between the isomers 1a,b and
2a,2b (Scheme 3). Therefore, the loss of the b-alanine
moiety leads to the loss of structural information, which
hampered the elucidation of the connectivity of the alkyl
chains and accordingly a pronouncement on which of the
isomers was present in the nBuOH fraction.
Table 2. NMR spectroscopic data for the synthetic viscosalines B1 (1a),
B2 (1b), E1 (2a), and E2 (2b).
Position[a]
E1
Chemical shifts[b]
(13C)/d(15N) d(1H)
[ppm][c] [ppm][c]
B1
(1a)
B2
(1b)
E2
(2b)
d
U
ACHTUNGTRENNUNG
A
R
(2a)
N
R
ACHTUNGTRENNUNG
1
1
1
1
G
–
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
143.6
142.9
144.9
127.3
142.0
60.4
9.02
–
8.47
8.07
8.93
4.54
8
8
8
8
30.3
1.90
9–16
17
18
19
20
21
22
23
24
25
26
27–35
36
37
38
39
40
41
9–17
18
19
20
21
22
23
24
25
26
27
28–35
36
37
38
39
40
41
9–17
18
19
20
21
22
23
24
25
26
27
28–37
38
39
40
41
42
43
9–18
19
20
21
22
23
24
25
26
27
28
29–37
38
39
40
41
42
43
24.5–29.5[d]
30.1/30.3
1.15–1.35[d]
1.56/1.57
2.62/2.59
–
Synthesis: Despite the discouraging results from CID–MSn
fragmentation, other 3-alkyl pyridinium alkaloids have for-
merly shown varying fragmentation pathways depending on
the fragmentation mode used. To test the hypothesis that
other fragmentation modes would facilitate differentiation
of the viscosaline isomers 1a–2b, more material was neces-
sary and required synthesis. Our first approach to synthesize
the amino acid-bearing monomer started from an amino
acid ester and an alkyl pyridine moiety, which seemed very
attractive because this synthetic route provided the opportu-
nity to generate a library of compounds with different
amino acid backbones and side chains. However, this ap-
proach failed in numerous ways. The disappointing results
led us to the structurally more limited approach through
a Michael addition reaction followed by a Zincke reaction
of two alkyl pyridine monomers (see Scheme 4 and the Sup-
porting Information), as reported by Baldwin and co-work-
ers.[12] The accordant synthesis provided the desired four iso-
mers 1a,b and 2a,b for the mass-spectrometric fragmenta-
tion and chromatographic separation studies.
31.6/31.7
138.9/137.9
147.3/148.7
8.51/8.43
–
8.49/8.41
7.48/7.35
7.83/7.68
2.78
AHCTUNGTERG(NNUN 319)
145.0/146.1
123.9/123.3
137.7/136.1
31.3
29.5/29.4
24.5–29.5[d]
25.1/25.2
46.6
(44)
42.2
30.1/30.0
171.6
1.63
1.15–1.35[d]
1.54/1.55
2.89
8.50/8.45
3.11/3.12
2.64/2.63
–
[a] Position numbering follows the numbering of viscosaline C (3) as orig-
inally reported.[2g] [b] The first value represents a signal from a viscosali-
nes B isomer (i.e., 1a+1b) and the second value represents a signal from
a viscosaline E isomer (i.e, 2a+2b). When only one value is given, the
chemical shifts are identical. [c] 1H and 13C NMR chemical shifts are ref-
erenced to the [D6]DMSO signal (d=2.50 and 39.5 ppm, respectively) for
1a–2b. 15N NMR chemical shifts are given in parentheses and are not
calibrated with an external calibrant. The d value has an accuracy of
about 1 ppm to NH3 (d =0 ppm). [d] It was not possible to assign single
chemical shifts.
HPLC and mass spectrometric analysis of the synthetic
products: As expected, the CID–MSn spectra of the synthet-
ic compounds 1a–2b led to the ambiguous fragmentation
products described above (Scheme 3), but the isomers
showed disparate signal intensities for the F1 and F2 frag-
ments (Figure 2). In both isomers, the F1 fragment showed
a higher signal intensity than the F2 fragment. This outcome
indicated that the signal intensity may be used to assign the
connectivity. A comparison of the spectra of the natural
compounds with those spectra
of the central pyridinium unit and F2 represents the alkyl
pyridine fragment that formerly carried the b-alanine
moiety (Scheme 3). Both onium fragments F1 and F2 from
3 consist of a pyridine moiety connected to an alkyl chain of
13 methylene groups and thus have the same mass (m/
obtained from the pure isomers
suggested that the two new vis-
cosalines
1 and 2 might be
a mixture of both isomers be-
cause the intensities of frag-
ments F1 and F2 in the frac-
tions of the natural viscosalines
B (1) and E (2) were similar
(Figure 2).
Unfortunately, HPLC analy-
sis of the synthetic compounds
1a–2b initially revealed identi-
cal retention times for each of
the two isomers and precluded
Scheme 2. Theoretical fragmentation of 1a (n=9, m=8), 1b (n=8, m=9), 2a (n=10, m=9), and 2b (n=9,
m=10) according to the fragmentation mechanisms observed in the cyclostellettamines.
8182
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8180 – 8189

Structure Elucidation of Viscosaline Isomers
FULL PAPER
the assignment of the alkyl-
chain position. Accordingly,
possible natural isomers were
inseparable by the method
used. Therefore,
a different
method had to be found to elu-
cidate the structure of 1 and 2.
Derivatization and mass spec-
trometric fragmentation pat-
terns of the derivatized prod-
ucts: To achieve a better chro-
matographic separation of the
isomers, the next attempt was
to derivatize the synthetic iso-
mers, and thereafter derivatize
and separate the natural frac-
tions that contained 1 and 2 as
well. Because of the early loss
of the terminal b-alanine group,
which was expected after deri-
vatization, we decided to func-
tionalize the pyridine nitrogen
atom. Two possibilities were
available to achieve this aim:
1) formation of an N-oxide
functionality and 2) an N-meth-
ylation reaction. The attempt to
Scheme 3. Fragmentation pathway of 2a and 2b under “standard” CID–MS ion-trap conditions. Boxes with
dashed lines indicate the F1 fragment, whereas boxes with solid lines indicate the F2 fragment.
generate pyridine N-oxides by using standard procedures[13]
resulted in complex mixtures and was therefore not further
investigated. By adding an excess of MeI to the synthetic
viscosalines resulted not only in an N-methylated pyridine
unit but in permethylated derivatives. The derivatization
process eliminated the propanoic acid moiety and attached
three methyl moieties to give
a
quaternary amine
(Scheme 5). In the mass spectra, the quaternary amine
group appeared as a triply charged molecular ion M3d (d=
derivative): m/z 193.5 [M]3+ and 202.9 [M]3+ for 1 and 2, re-
spectively. By using synthetic viscosaline E1 (2a) as an ex-
ample to follow the fragmentation pathway (for detailed
fragment masses of the other viscosalines, see Table 3), the
triply charged molecular ion M3d (m/z 202.9 [M]3+) cleaved
a fragment of
m/z 60 (trimethylammonium) in an onium reaction, thus re-
sulting in a doubly charged daughter ion F3d at m/z 274.2
(Scheme 5). Further fragmentation of this daughter ion re-
sulted in a singly charged fragment peak at m/z 274.2. This
peak constituted two fragments, that is, F1d assigned as the
N-methylated alkyl pyridinium moiety and F2 that corre-
sponds to the alkyl pyridinium moiety that formerly carries
the b-alanine unit. The accordant fragment peaks for visco-
saline E2 (2b) appear at m/z 288.2 and 260.2 for F1d and F2,
respectively. The fragmentation mechanisms that lead to the
fragments F1d and F2 were likely to be onium reactions, as
during the fragmentation of the cyclostellettamines.[9a]
Fragmentation of the triply charged molecular ion M3d of
the derivatized viscosaline E1 (2a) led to a second fragment
Scheme 4. Synthesis of the viscosalines. a) Acrylonitrile, Hꢀnig base,
MeOH, RT, 3 days; b) Boc2O, CH2Cl2/sat. NaHCO3, ultrasonic, 15 min
(65% over 2 steps); c) 1-chloro-2,4-dinitrobenzene, MeOH, D, 4 days
(75%); d) Et3N, nBuOH, D (40–70%); e) 3m HCl, D. Residual HCl from
this last reaction step transformed the acids 9 into methyl esters during
lyophilization (ester/acid=ca. 9:1). Boc=tert-butyloxycarbonyl.
Chem. Eur. J. 2012, 18, 8180 – 8189
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8183

M. Kçck et al.
F4d that was present in singly
and doubly charged ions at
m/z 333.3 [M]+ and 167.2 [M+
H]2+
(see
Table 3
and
Scheme 3), whereas the F4d
ions of derivatized viscosaline
2b were recorded at m/z 319.3
[M]+ and 160.2 2 [M+H]2+
.
These fragments corresponded
to the 3-alkyl pyridine that car-
ries the quaternary amine
moiety. Fragmentation of the
doubly charged F4d ion led to
fragment F2 in all the derivat-
ized viscosaline isomers. Finally,
the isolation and fragmentation
of fragment F1d gave a typical
cleavage of alkyl chains of
Dm=14, which had already
been observed during fragmen-
tation of the cyclostellett-
AHCTUNGTRENNUNG
Figure 2. CID–MS3 spectra of the natural viscosaline E (2) and the synthetic viscosalines E1 (2a) and E2 (2b).
The spectra show the stability preference of the alkyl pyridine fragment F1. The disparate fragment intensities
of the synthetic viscosaline isomers in comparison to similar fragment intensities in the natural fraction also in-
dicate that the natural viscosaline E (2) is a mixture of the possible isomers E1 (2a) and E2 (2b). Stars indicate
the precursor ions selected for fragmentation.
mers 1a–2b could be unambig-
uously differentiated by mass-
spectrometric analysis by their
F1d, F2, and quaternary amine
F4d fragments. The presence of
a mixture of the F1d, F2, and
F4d fragments for each isomer
during subsequent permethyla-
tion and mass spectrometric
analyses of the natural H. visco-
sa fractions supported the ob-
servation that both isomers
were present in the natural
fractions.
Mass spectrometric fragmenta-
tion of the underivatized prod-
ucts: While in principle the de-
rivatization approach worked
well for the viscosalines, it is
advantageous to have a method
that avoids derivatization of the
natural products. Because in-
source CID (“skimmer CID”)
had already proved to be useful
during the structure elucidation
of other 3-alkyl pyridinium al-
kaloids,[9] the tandem MS ex-
periments carried out on the
viscosalines were combined
with in-source CID experiments
and resulted in significant first-
generation fragment ions prior
to the isolation and fragmenta-
Scheme 5. Fragments observed during CID–MSn fragmentation of the permethylated viscosaline E2 (2b). The
corresponding fragment masses for viscosalines B1 (1a), B2 (1b), and E1 (2a) are given in Table 3.
8184
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8180 – 8189

Structure Elucidation of Viscosaline Isomers
FULL PAPER
Table 3. Molecular ions and fragments observed during mass spectromet-
ric fragmentation of permethylated viscosaline isomers.[a]
Table 4. Molecular ions and fragments observed during mass-spectromet-
ric fragmentation of synthetic viscosaline isomers.[a]
Compound
MS
MS2
MS3
F2d=F2
[M+H]+
Compound
MS
MS2
MS3
M3d
F3d
F4d
F4d
[M+H]2+
F1d
M1
F3
[M+H]2+
F4
F5
F6
F1
[M]3+
[M]2+
[M]+
G
[M]+
E
[M]+
G
[M+H]2+
[M]+
[M]+
[M]+
B1 (1a)
B2 (1b)
E1 (2a)
E2 (2b)
193.5
193.5
202.9
202.9
260.2
260.2
274.2
274.2
319.3
305.3
333.3
319.3
160.2
153.2
167.2
160.2
260.2
274.2
274.2
288.3
260.2
246.2
274.2
260.2
B1 (1a)
B2 (1b)
E1 (2a)
E2 (2b)
594.5
594.5
622.5
622.5
267.2
267.2
281.3
281.3
253.2
253.2
267.2
267.2
289.3
275.3
303.3
289.3
491.4
491.4
519.5
519.5
246.2
260.2
260.2
274.2
[a] F1m refers to the N-methylated alkyl pyridinium moiety and F2 refers
to the alkyl pyridinium that formerly carried the b-alanine residue
(Scheme 5). Values are given as m/z.
[a] M1, F1, F3, F4, F5, and F6 refer to the molecular and fragment ions
depicted in Scheme 6. Values are given as m/z.
tion of specific ions. This outcome led to the desired distinc-
tion between the underivatized constitutional isomers by
two previously unobserved mass peaks.
which the charge was located at different nitrogen atoms.
Both forms were likely to persist during measurements be-
cause fragment F1 (m/z 260.2) was observed in the MS2
spectrum of m/z 281.8 (F3; Figure 3).
The application of in-source CID to synthetic viscosaline
E1 (2a) in an ion-trap mass spectrometer resulted predomi-
nately in several singly and doubly charged fragment ions,
that is, m/z 260.2 and 274.2 for the F1 and F2 singly charged
ions, respectively, and m/z 281.8 and 267.2 for the F3 and F4
doubly charged ions, respectively (Figure 3). The corre-
sponding values for 1a,b and 2b are shown in Table 4. As
mentioned above, the F1 and F2 fragments were not indica-
tive of the structure. The same was true for the doubly
charged fragment F4 (m/z 267.2), which constituted the mo-
lecular ion that had lost the b-alanine moiety in an onium
reaction (Scheme 6). In contrast, the doubly charged ion F3
(m/z 281.8) was essential for the structure determination.
The fragment F3 could occur in two different forms in
It was remarkable to observe the doubly charged frag-
ment ion F3, which was seemingly generated from a singly
charged precursor/molecular ion. However, in-source CID
affects all incoming ions and the full scan spectrum of the
viscosalines showed doubly and triply charged pseudomolec-
ular ions that may form doubly charged fragments upon
fragmentation. Nevertheless, it is also possible that the mo-
lecular ion M1 (m/z 622.5) occurs as a zwitterion with an ad-
ditional charged pyridinium nitrogen atom and therefore
provides a doubly charged daughter ion after cleavage of
the acetic acid moiety.
Trapping and fragmentation of the doubly charged F3 in
synthetic viscosaline E1 (2a) resulted in fragments at
Figure 3. MS to MS3 spectra of the synthetic constitutional isomers viscosaline E1 (2a) and E2 (2b) by using a combination of in-source CID and intrap
CID. The MS/MS spectra clearly prove the different constitution of both compounds. The fragment numbers are described in the text, and stars indicate
the precursor ions selected for fragmentation.
Chem. Eur. J. 2012, 18, 8180 – 8189
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8185

M. Kçck et al.
Scheme 6. Proposed fragmentation pathway of viscosaline E1 (2a) under in-source CID+MSn ion-trap fragmentation conditions (Figure 3). The frag-
ments that determine the constitution of the isomer are drawn in solid boxes. The corresponding fragment masses for viscosalines B1 (1a), B2 (1b), and
E2 (2b) are given in Table 4.
m/z 260.3 and 519.5 for fragment F5 and at m/z 303.3 for
fragment F6. Due to the low resolution of the ion-trap mass
spectrometer, it was not possible to differentiate whether
the signal at m/z 260.2 represented the doubly charged ana-
logue of F5 (m/z 519.5), the F1 fragment, or an overlay of
both. Fragment F6 (m/z 303.3) was indicative for the isomer-
ic structure of the molecule because this fragment constitut-
ed the imine species of the amino acid-bearing chain. Fur-
thermore, fragmentation of F5 (m/z 519.5) resulted in the
typical onium fragment F1 at m/z 260.2 for 2a, which unam-
biguously defined the alkyl chain between the two pyridine
moieties (Figure 3C). Viscosaline E2 (2b) fragmented ac-
cordingly, and the F1 fragment (m/z 274.2) persisted as the
fragmentation product of the precursor F5.
Thus, the use of in-source CID proved to be the fastest
and least material-intensive approach to differentiate be-
tween the constitutional isomers of viscosalines 1 and 2. To
achieve in-source CID, the voltage difference between the
capillary exit and skimmer 1 in the ion-transfer region of
the mass spectrometer needed to be increased. However,
this increase was only possible in instruments that used
skimmer optics. Newer instruments often use a funnel tech-
nology for ion transfer, in which they have the advantage of
increased sensitivity but lack the skimmers used to perform
in-source fragmentation. To examine the possibilities for in-
source CID on funnel-technology instruments, we compared
our standard ion trap that uses skimmer optics with a new-
generation instrument containing funnel optics. In-source
CID on funnel-optics instruments can be achieved by in-
creasing the voltage of the funnel 1 outlet and lens, respec-
tively. In the case of the viscosalines, an increase of both
voltages by 50 V gave the best results with respect to peak
intensity. The fragmentation results of viscosalines obtained
by a skimmer-optics instrument compared to a funnel-optics
instruments were similar.
Improvement of the chromatographic separation followed
by in-source CID–HRMS: After the mass spectrometric dif-
ferentiation of the two constitutional isomers of viscosalines
B (1) and E (2) was possible, the HPLC conditions were fur-
ther optimized to distinguish between the constitutional iso-
mers presumably present in the natural extract. A change
from an unbuffered to NH4OAc-buffered solvent revealed
two signals for each of 1 and 2. By coupling the chromato-
graphic separation to mass spectrometric analysis, the previ-
ous assumption that a mixture of both isomers was present
for 1 and 2 in the natural extract was confirmed (Figure 4).
Coupled HPLC–HRMS with the use of soft ionization
(ESI) and in-source CID resulted in a distinctive molecular-
ion pattern that clearly differentiated the two isomers of
each viscosaline compound. Although some mass peaks
were overlaid by coeluting non-viscosaline 3-alkyl pyridini-
um alkaloids, the distinction between the isomers was unam-
biguously possible.
The HPLC–(in-source CID) HRMS spectra of the natural
viscosalines were again distinctly different from the corre-
sponding spectra obtained on the ion-trap mass spectrome-
ter. In addition to the singly, doubly, and triply charged mo-
lecular ions of viscosaline E1 (2a) at m/z 622.5315, 311.7703,
and 208.1847 for M1–M3, respectively (Figure 5), from
which the molecular formula C40H68N3O2 [M]+ was assigned,
the spectrum showed several fragments already observed
during the MSn experiments. The doubly charged mass peak
at m/z 260.2362 for F5 derived from a-cleavage or the
McLafferty rearrangement of the b-alanine moiety during
the triply charged state of the compound. This fragment was
8186
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8180 – 8189

Structure Elucidation of Viscosaline Isomers
FULL PAPER
mechanism (i.e., a-fragmenta-
tion and an onium reaction). By
starting at the doubly charged
fragment (m/z 267), a stepwise
loss of a fragment of m/z 14
(typical for an alkyl-chain
cleavage reaction) was clearly
represented by the doubly
charged
fragments
m/z
253.2304, 246.2237, 239.2148,
and so forth. This cleavage of
the alkyl chain could be fol-
lowed, though at lower intensity
than the molecular and main
fragment masses, to the doubly
charged fragment F7 at m/z
176.1472. This fragment corre-
sponded to 3-methyl-1-[13-(pyr-
idinium-1-yl)tridecyl]pyridinium
in
viscosaline
E1
(2a,
Figure 4. HPLC/UV chromatogram at l=260 nm of a mixture of the synthetic viscosalines B1 (1a), B2 (1b)
and E1 (2a), E2 (2b). A XTerra RP18 column (3.0ꢄ150 mm, I.D.=3.5 mm; Waters) was used for separation by
HPLC. The separation was achieved by applying a gradient from 50% acetonitrile/50% NH4OAc (5 mm) to
70% acetonitrile/30% NH4OAc (5 mm) over 10 min.
Scheme 7). The fragment unam-
biguously assigned the alkyl
chain between the two pyridine
moieties.
A cleavage of the
second alkyl chain was not ob-
served under the applied frag-
mentation conditions. A similar
fragment to F7 was observable
under the fragmentation of re-
lated 3-alkyltetrahydropyridine
alkaloids, the haliclamines,[9b]
from which a similar fragment
emerged during MS3 analysis of
the doubly charged pseudomo-
lecular ion.
Although the exclusive pres-
ence of fragment F7 at m/z 176
in the chromatogram peak al-
lowed the presence of 2a to be
stated, it did not exclude the
presence of 2b in the same
peak. The F7 fragment of 2b
would be expected at m/z 183;
furthermore, this signal would
overlap with the fragment ꢂF7+
m/z 14ꢃ of 2a. However, the
chromatographic separation of
the peaks revealed one peak
Figure 5. High-resolution in-source CID LC-MS profiles of viscosalines E1 (2a) and E2 (2b). The area shaded
in gray marks an as-yet unidentified, coeluting compound. The area needed to determine alkyl-chain cleavage
and length is enlarged.
not indicative for the constitution of the compound. The
doubly charged fragment F3 at m/z 281.7587 and its triply
charged analogue m/z 188.1789 derived from the a-cleavage
of acetic acid (m/z 60).
for isomer 2a, in which the cleavage of the alkyl chain
ended in m/z 176.1, and a second chromatogram peak for
isomer 2b, in which the cleavage terminated with a fragment
at m/z 183.1 (i.e., a lack of m/z 176.1; Figure 5).
A third set of fragments, the doubly charged F4 ion at
m/z 267.2457 and its corresponding singly charged ion at
m/z 533.4829, which was present at a very low intensity, was
essential for the assignment of the constitution. These ions
represented the loss of the b-alanine moiety by a two-step
The combination of the molecular-ion and fragment peaks
under in-source CID–HRMS conditions resulted in a distinc-
tive pattern that allowed a visual recognition of the viscosa-
lines in HPLC–HRMS experiments. The recognition of
a fifth new viscosaline compound provided the following ar-
Chem. Eur. J. 2012, 18, 8180 – 8189
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8187

M. Kçck et al.
signment of the structure of the viscosalines: The first
method employed a methylation reaction of the crude mate-
rial followed by a “simple” MS/MS analysis, and the second
method relied on more sophisticated instrumentation, for
example, the implementation of a second fragmentation
method, namely, in-source CID.
A remarkable observation is the difference in fragmenta-
tion that derived from in-source CID on a TOF mass spec-
trometer or an ion-trap mass spectrometer. This difference
must be attributed to different source geometries and other
parameter differences between the instruments. The results
show that in-source CID can be the decisive fragmentation
method in the structure assignment of compounds, whereas
“standard” CID techniques might reveal ambiguous results.
During in-source CID–HRMS, the composition of the visco-
saline fragment peaks in the spectrum combines to give a vis-
ually recognizable pattern. With the advantage of high-reso-
lution masses and the detailed analysis of the derived frag-
ments, an unambiguous assignment of the constitutional iso-
mers is possible.
Experimental Section
General: HRMS and LCMS were recorded on a Bruker Daltonics micro-
TOFLC mass spectrometer. MSn spectra were acquired with a Bruker
Daltonics Esquire 3000plus ion trap equipped with skimmer optics and
a Bruker Daltonics amaZon SL ion trap equipped with funnel optics. All
the mass spectrometers used an ESI source for ionization (see below for
the instrument setting). Analytical HPLC and HPLC–MS of the extracts,
fractions, and the pure synthetic samples were performed on an Agilent
1100 Series HPLC system equipped with a diode array detector (DAD).
Preparative purification of the synthetic products was carried out on
a JASCO 1500 series HPLC system equipped with a JASCO 2000 series
photodiode array detector and a Sedere SEDEX 75 light-scattering de-
tector.
Scheme 7. Fragments observed for viscosaline E1 (2a) during an in-
source CID experiment on a high-resolution mass spectrometer (approxi-
mate values are given; see the text for exact values). The fragment that
can be used to determine the constitution of the isomer is drawn in
a bold box. The corresponding fragment for viscosaline E2 (2b) is m/z
183.1 [M+H]+.
gument: After improvement of the chromatographic separa-
tion, a distinctive mass pattern of the viscosalines was found
in a formerly overlaid chromatogram peak of the crude ex-
tract of H. viscosa. The exact structure of this new viscosa-
line has not yet been fully established, but implementation
of the results obtained for 1 and 2 indicate viscosaline E1
with a double bond in the b-alanine unit that carries an
alkyl chain. Accordingly, the compound was named dehy-
droviscosaline E1 (4). The assignment of the position of the
double bond is in progress.
NMR spectra were recorded on Bruker AM250 (250 MHz) and Bruker
AV400 (400 MHz) spectrometers. The chemical shifts d are quoted in
ppm and are referenced to the appropriate solvent signal. Coupling con-
stants J are quoted in Hz. Data are reported as follows: chemical shift
multiplicity: s=singlet, d=doublet, t=triplet, q=quartet, qui=quintet,
m=multiplet; coupling constants; integration; assignment.
The melting points were obtained using a Kofler melting-point apparatus
and are uncorrected. FTIR spectra were recorded on Perkin–Elmer 1600
series. Absorption maxima are reported in wave numbers (cmÀ1), and the
following abbreviations are used s=strong, m=medium, w=weak. Ele-
mental analysis was performed on a Heraeus CHN Rapid machine. Data
is given as a percentage.
All the solvents were purified through simple distillation, except for
THF, which was distilled from sodium/benzophenone under argon. The
chemicals were used as purchased. Lithium diisopropylamide (LDA) in
solution was purchased from Fluka.
Conclusion
Despite their rather clear structure, viscosalines posed sever-
al challenges during their structure elucidation. The first
challenge was the assignment of the b-alanine and pyridine
moieties to the correct alkyl chain, which could be solved by
sophisticated NMR spectroscopic experiments.[2g] Viscosa-
lines with dissimilar chain lengths exhibited the additional
challenge that “standard” fragmentation experiments result-
ed in ambiguous fragments that hindered the assignment of
the correct chain to the pyridine or b-alanine moieties. How-
ever, we established two methods for the unambiguous as-
Animal material and extraction: Haliclona viscosa was collected off the
coast of Blomstrandhalvøya, near Hansneset, by scuba diving (depth of
15–25 m, June 2000 and 2001) in Kongsfjorden (West coast of Svalbard,
Norway; 798N, 128E). Voucher specimens were deposited under registra-
tion no. ZMA POR. 17008 and no. ZMA POR. 17009 at the Zoçlogisch
Museum (Amsterdam, The Netherlands). Sponge identification was
kindly conducted by W.H. de Weerdt and Dr. R.W.M. van Soest (Insti-
tute for Biodiversity and Ecosystem Dynamics, Zoological Museum, Uni-
versity of Amsterdam, The Netherlands). Samples of Haliclona viscosa
were immediately frozen after collection and kept at À208C until extrac-
tion. Freeze-dried sponge tissue (year 2001) was extracted at room tem-
8188
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8180 – 8189

Structure Elucidation of Viscosaline Isomers
FULL PAPER
perature with methanol/dichloromethane (1:1 v/v, 3ꢄ1000 mL). The re-
sulting green crude extract was partitioned between n-hexane (3ꢄ
150 mL) and methanol (80 mL). The methanol extract was concentrated,
further partitioned between ethyl acetate (3ꢄ150 mL) and water
(80 mL), and the aqueous layer was extracted with n-butanol (3ꢄ
150 mL). The resulting n-butanol phase (397.3 mg) was purified by using
preparative HPLC to yield viscosaline B (1; 46.5 mg) and viscosaline E
(2; 4.0 mg).
and Ellen Lichte for performing the HPLC analysis. We would also like
to thank Dr. Andrea Schneider (Bruker Daltonics, Bremen) for advice
with the mass spectrometer settings.
[1] a) R. J. Andersen, R. W. M. van Soest, F. Kong in Alkaloids: Chemi-
cal and Biological Perspectives, Vol. 10, 1st ed. (Ed.: S. W. Pelletier),
Mass spectrometry: All the mass spectra were acquired in positive mode
and calibrated externally with a sodium formate cluster (scan range=
m/z 50–800). During direct injection MS and LCMS experiments, the
masses were averaged over several milliseconds to seconds in the record-
ing time. The following ESI inlet conditions were applied during direct
injection (LCMS experiments, respectively) on the TOF spectrometer:
dry-gas temperature=1808C; dry-gas flow=5 LminÀ1 (9 LminÀ1); nebu-
lizer pressure=0.5 bar (4 bar); capillary voltage=4500 V; capillary exit=
100 V; skimmer 1=50 V. For in-source CID, the voltage of the capillary
exit was raised to 210 V, whereas the voltage of skimmer 1 remained un-
affected. For MSn fragmentation on the ion-trap mass spectrometer, the
following conditions were applied on the Esquire 6000 spectrometer:
dry-gas temperature=3008C; dry-gas flow=10 LminÀ1; nebulizer pres-
sure=7.35 p.s.i.; capillary voltage=4000 V; capillary exit=200 V; skimm-
er=40 V. During the in-source CID experiments, the capillary exit volt-
age was raised to 280 V. On the amaZon SL ion trap, the following set-
tings were applied: dry-gas temperature=2208C; dry-gas flow=
5 LminÀ1; nebulizer pressure=8 p.s.i.; capillary voltage=4500 V; capilla-
ry exit=140 V; funnel 1 inlet=100 V, outlet=35 V, lens=25 V. For in-
source CID, the voltages of funnel 1 were changed to 85 and 75 V for the
outlet and lens, respectively. To exclude isotope peaks, the peaks were
isolated with an isolation width of <2 Da for singly charged precursor
ions (and <1 Da for doubly charged precursor ions) and fragmented at
an amplitude of 1.0. The instruments averaged 5 single scans per record-
ing.
ˇ ´
Pergamon Press, 1996, pp. 301–355; b) T. Turk, K. Sepcic, I. Manci-
ni, G. Guella in Studies in Natural Products, Vol. 35 (Ed.: Atta-ur-
Rahman), Elsevier, 2008, pp. 355–397.
[2] a) G. Cimino, S. De Stefano, G. Scognamiglio, G. Sodano, E. Trivel-
lone, Bull. Soc. Chim. Belg. 1986, 95, 783–800; b) R. Sakai, T. Higa,
C. W. Jefford, G. Bernardinelli, J. Am. Chem. Soc. 1986, 108, 6404–
6405; c) R. Sakai, S. Kohmoto, T. Higa, C. W. Jefford, G. Bernardi-
nelli, Tetrahedron Lett. 1987, 28, 5493–5496; d) M. T. Davies-Cole-
man, D. J. Faulkner, G. M. Dubowchik, G. P. Roth, C. Polson, C.
ˇ ´
Fairchild, J. Org. Chem. 1993, 58, 5925–5930; e) K. Sepcic, G.
Guella, I. Mancini, F. Pietra, M. Dalla Serra, G. Menestrina, K.
Tubbs, P. Macek, T. Turk, J. Nat. Prod. 1997, 60, 991–996; f) C. A.
Volk, M. Kçck, Org. Lett. 2003, 5, 3567–3569; g) C. A. Volk, M.
Kçck, Org. Biomol. Chem. 2004, 2, 1827–1830.
ˇ
[3] a) T. Ichiba, R. Sakai, S. Kohmoto, G. Saucy, Tetrahedron Lett. 1988,
29, 3083–3086; b) M. Tsuda, N. Kawasaki, J. Kobayashi, Tetrahedron
Lett. 1994, 35, 4387–4388; c) M. Tsuda, K. Inaba, N. Kawasaki, K.
Honma, J. Kobayashi, Tetrahedron 1996, 52, 2319–2324; d) D. Wata-
nabe, M. Tsuda, J. Kobayashi, J. Nat. Prod. 1998, 61, 689–692; e) J.-
F. Hu, M. T. Hamann, R. T. Hill, M. Kelly in The Alkaloids: Chemis-
try and Biology, Vol. 60 (Ed.: G. A. Cordell), Elsevier, 2003,
pp. 207–285; f) J. Peng, K. V. Rao, Y.-M. Choo, M. T. Hamann in
Modern Alkaloids - Structure, Isolation, Synthesis and Biology (Eds.:
E. Fattorusso, O. Taglialatela-Scafati), Wiley-VCH, Weinheim, 2008.
[4] J. Kobayashi, M. Tsuda, N. Kawasaki, K. Matsumoto, T. Adachi, Tet-
rahedron Lett. 1994, 35, 4383–4386.
Chromatography: Column chromatography was performed on silica gel
60 (particle size=0.04–0.063 mm; Merck) or basic aluminium oxide. TLC
analysis was performed on aluminium plates precoated with Merck silica
gel 60. The compounds visualized by UV irradiation (l=254 nm) or
dying with KMnO4 solution (1% KMnO4, 6.6% K2CO3, and 5% NaOH
in 100 mL of water).
[5] F. J. Schmitz, K. H. Hollenbeak, D. C. Campbell, J. Org. Chem.
1978, 43, 3916–3922.
[6] a) N. Fusetani, K. Yasumuro, S. Matsunaga, Tetrahedron Lett. 1989,
30, 6891–6894; b) C. A. Volk, H. Lippert, E. Lichte, M. Kçck, Eur.
J. Org. Chem. 2004, 3154–3158; c) G. Schmidt, C. Timm, M. Kçck,
Org. Biomol. Chem. 2009, 7, 3061–3064; d) C. Cychon, G. Schmidt,
T. Mordhorst, M. Kçck, Z. Naturforsch. B 2012, in press.
[7] a) N. Fusetani, N. Asai, S. Matsunaga, K. Honda, K. Yasumuro, Tet-
rahedron Lett. 1994, 35, 3967–3970; b) N. Oku, K. Nagai, N. Shin-
doh, Y. Terada, R. W. M. van Soest, S. Matsunaga, N. Fusetani,
Bioorg. Med. Chem. Lett. 2004, 14, 2617–2620; c) J. H. H. L. de Oli-
veira, A. Grube, M. Kçck, R. G. S. Berlinck, M. L. Macedo, A. G.
Ferreira, E. Hajdu, J. Nat. Prod. 2004, 67, 1685–1689.
[8] Due to the presence of more than one compound in the structural
class of the viscosalines, the compound was renamed viscosaline C
according to the cyclostellettamine nomenclature (C=13/13).
[9] a) A. Grube, C. Timm, M. Kçck, Eur. J. Org. Chem. 2006, 1285–
1295; b) G. Schmidt, C. Timm, A. Grube, E. Lichte, E. Cuny, M. Re-
ggelin, M. Kçck, unpublished results.
The preparative HPLC column (16ꢄ250 mm, I.D.=10 mm) was prefilled
with Knauer Kromasil RP18 material. The mobile phase consisted of
A) water containing 0.1% trifluoroacetic acid and B) acetonitrile con-
taining 0.1% trifluoroacetic acid. Separation was achieved by applying
a gradient from 5 to 35% B over 40 min with a flow rate of 1 mLminÀ1
.
All HPLC–MS separations used a XTerra RP18 column (3.0ꢄ150 mm,
I.D.= 3.5 mm; Waters) with a flow rate of 0.4 mLminÀ1. The mobile
phase consisted of A) 5 mm NH4OAc in water and B) acetonitrile. A gra-
dient from 50 to 70% B over 10 min was applied followed by the initial
conditions and conditioning for 5 min for the fractions and synthetic com-
pounds. Total analysis time was 15 min. The natural extracts were sub-
jected to a linear gradient at a column-compartment and solvent temper-
ature of 308C with a mobile phase of A) 0.1% HCOOH in water and
B) 0.1% HCOOH in acetonitrile: 0 min: 20% B; 25 min: 55% B;
27 min: 100% B; isocratic for 3 min followed by the initial conditions
and conditioning for 5 min. Total analysis time was 40 min.[15]
[10] G. Schmidt, C. Timm, M. Kçck, Z. Naturforsch. B: J. Chem. Sci.
2011, 66, 745–748.
[11] C. Timm, C. Volk, F. Sasse, M. Kçck, Org. Biomol. Chem. 2008, 6,
4036–4040.
[12] B. J. Shorey, V. Lee, J. E. Baldwin, Tetrahedron 2007, 63, 5587–5592.
[13] a) N. Meyer, W. Wykypiel, D. Seebach, Org. Syn. Coll. Vol. 1988, 6,
342–348; b) S. Youssif, ARKIVOC 2001, 242–268.
Acknowledgements
Sponge collection was performed by the AWI scientific diving team
during the annual summer expeditions of 2000 and 2001 at the AWIPEV
Arctic Research Base, Ny-ꢅlesund (Svalbard, Norway). Sponge identifi-
cation was kindly conducted by W.H. de Weerdt and Dr. R.W.M. van
Soest (Institute for Biodiversity and Ecosystem Dynamics, Zoological
Museum, University of Amsterdam, The Netherlands). We would like to
thank Prof. Michael Gçbel (Universitꢆt Frankfurt/Main, Germany) for
his support of this project, especially for providing the laboratory space,
[14] C. Timm, PhD thesis, Johann Wolfgang Goethe-Universitꢆt (Frank-
furt am Main), 2007.
[15] The MSn spectra and NMR spectra, synthesis of the viscosalines,
and derivatization of the viscosalines are given in the Supporting In-
formation.
Received: May 4, 2011
Revised: December 14, 2011
Published online: May 21, 2012
Chem. Eur. J. 2012, 18, 8180 – 8189
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8189