Pathways and substrate specificity of DMSP catabolism in marine bacteria of the Roseobacter clade

Article abstract of DOI:10.1002/cbic.200900668

The volatiles released by Phaeobacter gallaeciensis, Oceanibulbus indolifex and Dinoroseobacter shibae have been investigated by GC-MS, and several MeSH-derived sulfur volatiles have been identified. An important sulfur source in the oceans is the algal metabolite dimethylsulfoniopropionate (DMSP). Labelled [²H₆]DMSP was fed to the bacteria to investigate the production of volatiles from this compound through the lysis pathway to [²H₆]dimethylsulfide or the demethylation pathway to [²H₃]-3-(methylmercapto)propionic acid and lysis to [ ²H₃]MeSH. [²H₆]DMSP was efficiently converted to [²H₃]MeSH by all three species. Several DMSP derivatives were synthesised and used in feeding experiments. Strong dealkylation activity was observed for the methylated ethyl methyl sulfoniopropionate and dimethylseleniopropionate, as indicated by the formation of EtSH- and MeSeH-derived volatiles, whereas no volatiles were formed from dimethyltelluriopropionate. In contrast, the dealkylation activity for diethylsulfoniopropionate was strongly reduced, resulting in only small amounts of EtSH-derived volatiles accompanied by diethyl sulfide in P. gallaeciensis and O. indolifex, while D. shibae produced the related oxidation product diethyl sulfone. The formation of diethyl sulfide and diethyl sulfone requires the lysis pathway, which is not active for [²H₆]DMSP. These observations can be explained by a shifted distribution between the two competing pathways due to a blocked dealkylation of ethylated substrates.

Full text of DOI:10.1002/cbic.200900668

DOI: 10.1002/cbic.200900668
Pathways and Substrate Specificity of DMSP Catabolism in
Marine Bacteria of the Roseobacter Clade
Jeroen S. Dickschat,* Claudia Zell, and Nelson L. Brock[a]
The volatiles released by Phaeobacter gallaeciensis, Oceanibul-
bus indolifex and Dinoroseobacter shibae have been investigat-
ed by GC-MS, and several MeSH-derived sulfur volatiles have
been identified. An important sulfur source in the oceans is
the algal metabolite dimethylsulfoniopropionate (DMSP). La-
belled [2H6]DMSP was fed to the bacteria to investigate the
production of volatiles from this compound through the lysis
pathway to [2H6]dimethylsulfide or the demethylation pathway
to [2H3]-3-(methylmercapto)propionic acid and lysis to
[2H3]MeSH. [2H6]DMSP was efficiently converted to [2H3]MeSH
by all three species. Several DMSP derivatives were synthesised
and used in feeding experiments. Strong dealkylation activity
was observed for the methylated ethyl methyl sulfoniopropio-
nate and dimethylseleniopropionate, as indicated by the for-
mation of EtSH- and MeSeH-derived volatiles, whereas no vola-
tiles were formed from dimethyltelluriopropionate. In contrast,
the dealkylation activity for diethylsulfoniopropionate was
strongly reduced, resulting in only small amounts of EtSH-de-
rived volatiles accompanied by diethyl sulfide in P. gallaeciensis
and O. indolifex, while D. shibae produced the related oxidation
product diethyl sulfone. The formation of diethyl sulfide and
diethyl sulfone requires the lysis pathway, which is not active
for [2H6]DMSP. These observations can be explained by a shift-
ed distribution between the two competing pathways due to
a blocked dealkylation of ethylated substrates.
Introduction
The Roseobacter clade forms one of the most important
groups of marine microorganisms.[1] They are distributed from
polar to tropical regions, are present in coastal and open-
ocean environments and are often associated with algae.[2] The
main sulfur metabolite produced by algae in high intracellular
concentrations, and the main source of marine organic sulfur,
is dimethylsulfoniopropionate (DMSP).[3] Large amounts of
DMSP are degraded by marine bacteria according to two com-
peting pathways, that is, lysis to dimethyl sulfide (DMS) by a
DMSP lyase or demethylation to 3-(methylmercapto)propionic
acid (MMPA, Scheme 1). Downstream processing of MMPA in-
cludes the elimination of MeSH.[4]
Several lysis mechanisms have been identified. DMSP can be
transformed to its CoA ester, which is cleaved to DMS and
acryloyl-CoA in a process catalysed by the acyl-CoA transferase
DddD first identified in Marinomonas.[5] Alternatively, in Sulfito-
bacter, the DMSP lyase DddL catalyses the formation of DMS
and acrylate from DMSP.[6] The recently identified DMSP lyase
DddP from Roseovarius nubinhibens cleaves DMSP to DMS by
an unknown mechanism.[7] The DMSP demethylase DmdA was
first described from Ruegeria pomeroyi.[8] No genes or enzymes
for the lysis of MMPA to MeSH are known.
DMS is important in the biogeochemical sulfur cycle,[9] in
2À
which SO4 is transported from the continents via the rivers
to the oceans, while the backflow to land requires the partici-
pation of sulfur gases. The microbial lysis of DMSP to DMS
causes an estimated annual sulfur flux from the oceans to the
atmosphere of 13 to 45 million tonnes of sulfur.[10] Atmospheric
2À
DMS is rapidly oxidised to methanesulfonic acid and SO4
,
which rains down to earth to close the cycle. The massive bac-
Scheme 1. DMSP degradation pathways. DMSP can be transformed into its
CoA ester (DMSP-CoA) by the acyl-CoA transferase DddD and subsequently
cleaved to dimethyl sulfide (DMS) and acryloyl-CoA, lytically cleaved to DMS
and acrylate by the DMSP lyase DddL, or degraded to DMS and acrylate by
an unknown mechanism involving the DMSP lyase DddP. Alternatively,
DMSP can be degraded on the demethylation pathway to 3-(methylmercap-
to)propionic acid (MMPA) by the DMSP demethylase DmdA, that may be
transformed into methanethiol (MeSH) and acrylate by an unknown enzyme.
[a] Dr. J. S. Dickschat, C. Zell, N. L. Brock
Institute of Organic Chemistry, Technical University of Braunschweig
Hagenring 30, 38106 Braunschweig (Germany)
Fax: (+49)531-3915272
E-mail: j.dickschat@tu-bs.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.200900668.
ChemBioChem 2010, 11, 417 – 425
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
417

J. S. Dickschat et al.
terial production of DMS is of climatical relevance, because
2À
atmospheric SO4 initiates the formation of aerosol particles
that act as cloud-condensation nuclei. The resulting clouds
control the reflectance, scattering and absorbance of sunlight
and directly influence the global radiation budget.[11] High
DMS flux rates consequently have a cooling effect and lead to
slower growth of the algae, thus limiting DMS production.[12]
The demethylation pathway also limits sulfur flux, because
MeSH is highly soluble in water and reacts to form metal–
MeSH complexes, with an impact on the availability of trace
metals in seawater.[13] The majority of DMSP seems to be de-
methylated by marine bacteria, whereas only a minor portion
is cleaved to DMS; this suggests that its release from the
oceans is tightly controlled by the bacterioplankton communi-
ty.[14]
Scheme 2. Synthesis of the hydrochlorides of [2H6]DMSP and derivatives.
these feeding experiments were collected by use of the
closed-loop stripping apparatus (CLSA) technique, and the
obtained extracts were analysed by GC-MS.[17] To analyse the
highly volatile compounds covered by the solvent peak of the
CLSA headspace extracts, the feeding experiments with
[2H6]DMSP were repeated but with the solvent-free, solid-
phase microextraction (SPME) technique.
This work focuses on three species of the Roseobacter clade
with sequenced genomes: Phaeobacter gallaeciensis DSM
17395T, Oceanibulbus indolifex DSM 14862T and Dinoroseobacter
shibae DSM 16493T (http://img.jgi.doe.gov/).[15] Recently, the
volatiles released by liquid cultures of two of these strains,
P. gallaeciensis and O. indolifex, have been reported,[16] with
sulfur volatiles forming the main compound class present in
the headspace extracts. Here we report on the volatiles re-
leased by these two species and D. shibae from agar plate cul-
tures and fed with DMSP as an important source of organic
sulfur available to bacteria of the Roseobacter clade in the
marine environment. Several DMSP derivatives were fed to the
bacteria, resulting in the production of a variety of novel vola-
tiles. As will be demonstrated in this article, these feeding ex-
periments also gave insight into the distribution between the
two DMSP catabolic pathways and served to test the specifici-
ties of the encoded DMSP-degrading enzymes in vivo.
Volatiles from Phaeobacter gallaeciensis DSM 17395T
The volatiles released by agar plate cultures of P. gallaeciensis
are summarised in Table S1 of the Supporting Information, and
the gas chromatogram is shown in Figure 2, below. As with
the recently reported headspace composition of P. gallaeciensis
grown in liquid medium,[16] several sulfur compounds including
dimethyl disulfide (2), dimethyl trisulfide (5), S-methyl meth-
anethiosulfonate (12) and S-methyl propanethioate (23) were
produced, in addition to ethyl benzoate (29, Figure 1). All iden-
tified sulfur volatiles contained methylthio groups and were
derived from MeSH by known reactions, such as oxidation with
molecular oxygen in the presence of ascorbate and transition
metal ions or charcoal as used in the CLSA filters, nucleophilic
substitution (to 2, 5, and 12), or transesterification of fatty
acyl-CoA derivatives (to 23, Scheme 3).[18]
Results and Discussion
Genome analysis, synthesis of the hydrochlorides of
[2H6]DMSP and DMSP derivatives, and analytical methods
A BLAST search revealed that P. gallaeciensis encodes a DMSP
lyase with high homology to DddP, and D. shibae encodes
DddD and DddL homologues. None of the known DMSP lyases
is present in O. indolifex, whereas all three organisms encode
DmdA homologues (Scheme 1). To test the biosynthetic capaci-
ties of these species to produce sulfur volatiles from DMSP, the
hydrochloride of isotopically labelled [2H6]DMSP was synthes-
ised by the acid-catalysed (HCl) addition of [2H6]DMS to prop-
2-enoic acid (Scheme 2) and fed to agar plate cultures of the
bacteria. Similarly, the hydrochlorides of the DMSP derivatives
diethylsulfoniopropionate (DESP), ethyl methyl sulfoniopropio-
nate (EMSP), dimethylseleniopropionate (DMSeP) and dimeth-
yltelluriopropionate (DMTeP) were synthesised and fed to agar
plate cultures of the bacteria to investigate the substrate
specificities of the involved enzymes in vivo. In all the feeding
experiments described in this article, the hydrochlorides of
[2H6]DMSP and its derivatives were used, but for simplicity and
brevity the terms [2H6]DMSP, DESP, EMSP, DMSeP and DMTeP
will be used. The volatiles released by the bacterial cultures in
Figure 1. Volatiles from bacteria of the Roseobacter clade.
418
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 417 – 425

DMSP Catabolism in Marine Bacteria
the DESP substrate compared to DMSP. In conclusion, the
DMSP demethylase(s) and the proposed MMPA lyase(s) in
P. gallaeciensis are able to transform EMSP into EtSH and EtSH-
derived volatiles, albeit with significantly reduced turnover
rates of the overall pathway.
The feeding experiment with DESP also yielded diethyl sul-
fide (1), covered by 2 (Figure 2), that arose from lysis activity
towards DESP, whereas the respective lysis product, [2H6]DMS,
from [2H6]DMSP could not be detected by the CLSA technique
because the highly volatile [2H6]DMS coelutes with the solvent.
To exclude a misinterpretation of the data obtained in the
feeding experiment with [2H6]DMSP, the respective agar plate
cultures were reanalysed by using solvent-free SPME. The
SPME headspace extract contained [2H6]-2, but no [2H6]di-
methyl sulfide (Figure S2); this supports a DMSP catabolism
that strongly favours the demethylation pathway over the lysis
pathway. The shifted distribution between these two pathways
in the feeding experiment with DESP towards the lysis path-
way, expressed by the production of 1, can be explained by a
blocked dealkylation of this substrate that enables a substan-
tial participation of the lysis pathway.
Scheme 3. Known reactions to sulfur volatiles from MeSH.
In the feeding experiment with EMSP, the majority of sulfur
volatiles were derived from EtSH, represented by 3, 4, ethyl
methyl trisulfide (6), diethyl trisulfide (7), S-ethyl methanethio-
sulfonate (13), S-methyl ethanethiosulfonate (14) and S-ethyl
ethanethiosulfonate (15). The identity of thiosulfonates 13–15
was established by synthesis from the sulfonyl chlorides and
ethanethiol or sodium methanethiolate (Scheme 4, for mass
spectra see Figure S3). The formation of mainly EtSH-derived
sulfur volatiles from EMSP points to an efficient demethylation
to EMPA, the same intermediate as obtained by the de-ethyla-
tion of DESP. This observation is in full accordance with the
postulated narrow substrate specificity of the enzyme machi-
nery for the dealkylation step, acting only efficiently in the
case of a demethylation but not of a de-ethylation step,
whereas the rather broad substrate specificity of the proposed
MMPA lyase(s) catalysing the lytic cleavage of MMPA and EMPA
is supported.
Feeding of [2H6]DMSP to P. gallaeciensis led to rapid degrada-
tion to [2H3]MMPA catalysed by DmdA and possibly by further
as yet unidentified DMSP demethylases. Subsequently
[2H3]MMPA was cleaved to [2H3]MeSH. The incorporation of
labelling into MeSH-derived volatiles 2, 5 and 12 proceeded
with high rates (ꢀ90%, Table S1). These high incorporation
rates demonstrated that the majority of MeSH-derived mole-
cules were produced from [2H6]DMSP and not from alternative
sulfur sources in the medium. Deuterated compounds elute
with shorter GC retention times than their unlabelled counter-
parts because of the shorter CÀD versus CÀH bond length;
this results in a lower polarisability and a weaker (induced)
dipole interaction of the analyte with the stationary phase.[19]
Even if several isotopomers with a different deuterium content
are present in one sample, they can be separated. The mass
spectra of all chromatographically separated isotopomers of 2,
5 and 12 are shown in Figure S1.
Feeding of the DMSP derivatives DESP, EMSP, DMSeP and
DMTeP resulted in all cases in the production of 2, 5 and 12
obviously originating from sulfur sources such as sulfate in the
medium. Additionally, in the feeding experiment with DESP,
the volatiles 23, S-methyl 3-methylbutanethioate (25) and 29,
but only small amounts of the EtSH-derived compound ethyl-
methyldisulfide (3) were found. The formation of 3 from DESP
can be explained by the dealkylation activity of DmdA and
possible further unidentified DMSP demethylase(s) to yield 3-
(ethylmercapto)propionic acid (EMPA), which is cleaved to
EtSH. At least one of the two steps is comparably slow to the
equivalent reactions on DMSP or MMPA. A reduced activity
towards DESP compared to DMSP in the dealkylation of one S-
alkyl group is likely to be more significant than reduced activi-
ty towards EMPA compared to MMPA in the lysis to the respec-
tive alkanethiol, either due to a higher steric hindrance of the
S-ethyl group compared to the S-methyl group in its nucleo-
philic attack, or due to the higher overall steric hindrance of
To further test the in vivo substrate specificities of the en-
zymes of the DMSP degradation pathways, DMSeP was fed to
P. gallaeciensis. Besides dimethyl tetrasulfide (8), large quanti-
ties of MeSeH-derived selenium compounds were emitted (for
mass spectra see Figure S4). The principle component in the
headspace extract was dimethyl diselenide (17) accompanied
by dimethyl triselenide (21). Furthermore, dimethyl selenyl sul-
fide (16), methyl methylseleno disulfide (18), methyl methyl-
thio diselenide (19), bis(methylseleno)sulfide (20) and methyl
methylseleno trisulfide (22) were found (for compound identifi-
cation see discussion in the Supporting Information, especially
Figure S5). Strikingly, this mixture of polychalcogenides is
simply obtained in a chalcogenide-exchange reaction of 5 with
17;[20] this suggests that the volatiles 18–22 are generated non-
enzymatically from 5 and 17.
Another class of selenium compounds comprises the Se-
methyl alkaneselenoates, represented by Se-methyl propanese-
lenoate (26) and Se-methyl 3-methylbutaneselenoate (27).
Their identity has been confirmed by a synthesis from the re-
ChemBioChem 2010, 11, 417 – 425
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
419

J. S. Dickschat et al.
Figure 2. Total ion chromatograms of the CLSA extracts from P. gallaeciensis agar plate cultures. The headspace extracts were obtained from A) MB2216
medium, B) MB2216+1 mm [2H6]DMSP·HCl, C) MB2216+1 mm DESP·HCl, D) MB2216+1 mm EMSP·HCl and E) MB2216+1 mm DMSeP·HCl. The peaks are
marked with the compound numbers used in the text. Signals arising from the medium, artefacts, and unidentified compounds are labelled with asterisks.
spective methyl esters using dimethyl aluminium methanesele-
nolate, which is readily available from trimethyl aluminium and
selenium (Scheme 5).[21] These compounds were probably
formed in a chemical reaction from MeSeH and the corre-
sponding fatty acyl-CoA derivatives.[18]
The presence of all these MeSeH-derived volatiles demon-
strated that DMSeP was efficiently demethylated to 3-(methyl-
seleno)propionic acid (MSePA) and lysed to MeSeH by P. gallae-
ciensis, thus showing that the involved enzymes accept and
Scheme 4. Synthesis of the thiosulfonates 13–15: a) pyridine, b) Et2O.
transform the selenium analogue of DMSP. DMSeP occurs natu-
420
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 417 – 425

DMSP Catabolism in Marine Bacteria
pool (Scheme 6). The volatiles 30 and 31 are likely to be de-
rived from l-methionine that is formed from MeSH and O-
acetyl homoserine under catalysis of the methionine synthase.
Transamination and oxidative decarboxylation lead to 4-(meth-
ylmercapto)-2-oxobutyric acid (38), which can be transformed
into its CoA thioester (39), which serves as a precursor for 30
and 31 by formal transesterification with EtOH or MeSH, re-
spectively.
Scheme 5. Synthesis of selenoate and thioate esters: a) Se, Me3Al, toluene;
b) S, Me3Al, toluene.
rally as a metabolite produced by the plant Spartina alternifolia
2À
when amended with toxic SeO4 and is an intermediate in its
transformation to dimethyl selenide.[22] The DMSP lyases from
Alcaligenes and Oceanimonas doudoriffii have previously been
shown to cleave DMSeP to dimethyl selenide and acrylate.[23]
Although the demethylation of DMSP and subsequent lysis to
MeSH by bacteria is a well-known process, the bacterial deme-
thylation of DMSeP and lysis to MeSeH have not been reported
before.
In contrast to DMSeP, the tellurium analogue DMTeP was
not transformed by P. gallaeciensis to MeTeH, and no tellurium
volatiles were found, but only the sulfur volatile 23. DMTeP
might not be taken up by the bacteria, or the enzymes of the
DMSP demethylation pathway might not be able to demethy-
late DMTeP and/or to lyse the respective demethylation prod-
uct.
Scheme 6. Biosynthesis of sulfur volatiles from methanethiol via the amino
acid pool.
In summary these results suggest that the DMSP demethy-
lase DmdA encoded in the genome of P. gallaeciensis and pos-
sibly additional unidentified DMSP-demethylating enzymes are
able to efficiently demethylate DMSP, EMSP and DMSeP, where-
as DESP is de-ethylated with a strongly reduced activity. The
lysis of the dealkylation products seems to proceed with high
activity in all cases, because the dealkylation of EMSP and
DESP results in the same proposed intermediate, EMPA. How-
ever, the formation of EtSH-derived volatiles from EMSP is effi-
cient, whereas from DESP it is not.
SPME analysis showed that only traces of [2H6]DMS close to
the detection limit were produced in the feeding experiment
with [2H6]DMSP (Figure S2).[25] Feeding of DESP yielded small
amounts of 1, presumably arising from enzymatic lysis. None
of the known DMSP lyases, DddD, DddL or DddP, is encoded in
the genome of O. indolifex. The formation of these lysis prod-
ucts, albeit in low amounts, indicates the existence of a fourth,
as yet unidentified bacterial DMSP lyase, but a slow nonenzy-
matic cleavage of [2H6]DMSP and DESP cannot be ruled out.
Volatiles from Oceanibulbus indolifex DSM 14862T
Volatiles from Dinoroseobacter shibae DSM 16493T
A similar set of feeding experiments was carried out with O. in-
dolifex. The results closely resembled those from P. gallaeciensis
(Table S2 and Figure 3). Additional sulfur volatiles from O. indo-
lifex not found in P. gallaeciensis were methyl methylthiomethyl
disulfide (9), S-methyl butanethioate (24), ethyl 3-(methylthio)-
propionate (30) and S-methyl 3-(methylthio)propanethioate
(31). The identity of 31 was verified by synthesis from methyl
3-(methylthio)propionate (Scheme 5). Feeding of [2H6]DMSP re-
sulted in the efficient incorporation of deuterium into all sulfur
volatiles, thereby demonstrating that DMSP was rapidly con-
verted to MeSH. All the carbons of compound 9 were deuterat-
ed (Figure S6); this is in accordance with its reported formation
by photolysis of 2 (Scheme 3).[24] The labelling was also intro-
duced into the S-methyl groups of the alkanethioates 23–25
and of 30, and into both S-methyl groups of 31 (Figure S7),
thus pointing to an uptake of [2H3]MeSH into the amino acid
The same feeding experiments were carried out with D. shibae
and lead to similar results to those found for P. gallaeciensis
and O. indolifex, but some important differences were observed
(Table S3 and Figure 4). The most striking disparities were the
occurrence of large amounts of diethyl sulfone (11, from DESP)
and small amounts of ethyl methyl sulfone (10, from EMSP),
but no [2H6]dimethyl sulfone from [2H6]DMSP. SPME analysis
demonstrated that only traces of [2H6]DMS close to its detec-
tion limit were formed from [2H6]DMSP (Figure S2). The sul-
fones 10 and 11 arise from oxidation of ethyl methyl sulfide
and 1, respectively.
The genome of Rhodobacter capsulatus 37b4 contains the
dorA gene, which encodes a dimethyl sulfoxide reductase
(DMSOR) that also catalyses the oxidation of DMS to DMSO.[26]
A similar DMSOR is encoded in D. shibae (Dshi_2278, 35%
identity) and might be involved in the oxidation of the dialkyl
ChemBioChem 2010, 11, 417 – 425
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
421

J. S. Dickschat et al.
Figure 3. Total ion chromatograms of the CLSA extracts from O. indolifex agar plate cultures. The headspace extracts were obtained from A) MB2216 medium,
B) MB2216+1 mm [2H6]DMSP·HCl, C) MB2216+1 mm DESP·HCl, D) MB2216+1 mm EMSP·HCl and E) MB2216+1 mm DMSeP·HCl. The peaks are marked with
the compound numbers used in the text. Signals arising from the medium, artefacts, and unidentified compounds are labelled with asterisks.
sulfides via the dialkyl sulfoxides to the sulfones. Such an
enzyme is not encoded in the genomes of P. gallaeciensis and
O. indolifex, and no dialkyl sulfoxides or sulfones were found in
the headspace extracts of these species. The formation of
quite high amounts of 11 from DESP in combination with the
lack of production of [2H6]dimethyl sulfone from [2H6]DMSP in
D. shibae can be explained by a rapid turnover of DMSP
through the dealkylation pathway deviating this substrate
from the lysis pathway under the growth conditions used,
whereas the blocked dealkylation of DESP results in its lysis
and subsequent oxidation to 11.
EMSP feeding showed that this substrate was efficiently de-
methylated to yield large amounts of EtSH-derived volatiles.
However, after feeding of EMSP, 10 was indeed produced; this
points by inversion of the arguments to a blockage of EMSP
dealkylation. Although speculative, this apparent contradiction
can be solved by assuming a rapid enzyme-catalysed dealkyla-
tion of one enantiomer of EMSP, whereas dealkylation of the
other enantiomer is blocked to result in its lysis (Scheme 7).
422
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 417 – 425

DMSP Catabolism in Marine Bacteria
Figure 4. Total ion chromatograms of the CLSA extracts from D. shibae agar plate cultures. The headspace extracts were obtained from A) MB2216 medium,
B) MB2216+1 mm [2H6]DMSP·HCl, C) MB2216+1 mm DESP·HCl, D) MB2216+1 mm EMSP·HCl and E) MB2216+1 mm DMSeP·HCl. The peaks are marked with
the respective compound numbers as used in the text. Signals arising from the medium, artefacts, and unidentified compounds are labelled with asterisks.
Occurrence of the identified volatiles in other natural
sources
Dimethyl disulfide (2) and dimethyl trisulfide (5) are wide-
spread bacterial volatiles.[27] The higher homologue dimethyl
tetrasulfide (8) has been found less frequently, but has been
reported from streptomycetes and myxobacteria.[19,28,29] Di-
methyl polysulfides also occur in plants, such as the impres-
sively flowering Amorphophallus titanum.[30] The EtSH-derived
disulfides 3 and 4 are aroma constituents of the durian fruit
Durio zibethinus, which is distinctive for its unique odour,[31]
but have never been reported from bacteria. Sulfonate 12 has
previously been detected in other bacteria of the Roseobacter
clade and from Stigmatella aurantiaca,[17,32] and is known from
Scheme 7. Proposed different turnover rates for the two enantiomers of
EMSP via the lysis and the dealkylation pathways.
ChemBioChem 2010, 11, 417 – 425
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
423

J. S. Dickschat et al.
Allium cepa, which also produces the related compounds S-
propyl methanethiosulfonate and S-propyl propanethiosulfo-
nate.[33] Meanwhile, the EtSH-derived sulfonates 13–15 and the
sulfones 10 and 11 obtained in our studies from EMSP have
never been reported from natural sources.
supports their importance in tightly controlling the sulfur flux
from the oceans. The available genetic information alone is not
sufficient to predict the used pathways for the bacterial DMSP
catabolism, because P. gallaeciensis and D. shibae both encode
enzymes with high homology to known DMSP lyases, whereas
the occurrence of 1 in the headspace above O. indolifex cultures
after feeding of DESP suggests the presence of an unknown
type of DMSP lyase in this organism. In addition to a careful
genome mining, a chemical analysis of the products of bacterial
DMSP catabolism is crucial. While our investigations clearly
showed the capacity of agar plate cultures of the Roseobacter
clade bacteria to produce MeSH-derived sulfur volatiles from
DMSP through the dealkylation pathway, the pathway distribu-
tion under other (environmental) conditions might be different.
The substrate specificities of the participating enzymes of the
DMSP demethylation pathway, that is, the DMSP demethylase(s)
and the proposed MMPA lyase, seem to be quite variable. Deal-
kylation of DMSP derivatives was especially efficient for methy-
lated substrates presented to DmdA (or, alternatively, as yet un-
known additional enzymes). Even the sulfur atom could be re-
placed by selenium, but not tellurium, without significant loss
of catalytic activity, whereas the dealkylation of DESP proceeded
with significantly reduced activity in all three species. In con-
trast, lysis of the dealkylation products MMPA, EMPA and MSePA
to the alkanethiols or methaneselenol, respectively, was efficient
in all cases, and, in other words, was neither dependent on the
nature of the chalcogenide-bound alkyl groups nor on the
nature of the chalcogenide atom itself. As a consequence of the
reduced dealkylation activity towards the unnatural substrate
DESP, lysis products were formed by all three species, that were
not produced from DMSP. Since the demethylation pathway do-
minated under the experimental conditions used in our studies,
conclusions on the in vivo substrate specificities of the different
DMSP lyases encoded in the genomes of the three species
under investigation are not possible.
Organoselenium compounds are rare in nature. The bacterial
volatiles dimethyl diselenide (17) and dimethyl selenyl sulfide
(16) were previously identified in different Allium spp., accom-
panied by bis(methylthio)selenide, which was not found in our
investigations.[34] In addition, 17 and the mixed selenium–
sulfur compound 16 are produced by different bacteria supple-
mented with sodium selenate.[35] Besides 16 and 17, the tri-
chalcogenide methyl methylseleno disulfide (18), but not its
isomer bis(methylthio)selenide, is produced from selenate or
selenite by genetically modified Escherichia coli expressing the
UbiE methyltransferase from Geobacillus stearothermophilus.[36]
The same engineered E. coli strain is able to biomethylate
potassium tellurate, but not potassium tellurite, to release di-
methyl telluride, dimethyl ditelluride and the tellurium ana-
logon of 16, dimethyl tellurenyl sulfide.[37,38]
The S-methyl alkanethioates 23–25 are known volatiles from
Streptomyces albidoflavus and from cheese-ripening bacteria
such as Brevibacterium linens and Micrococcus luteus,[39,40]
whereas the selenium analogues 26 and 27 are new natural
products. The unusual compound S-methyl 3-(methylthio)pro-
panethioate (31) is a constituent of human urine and its
odour-causing agent after the consumption of asparagus.[41]
The structurally related ester 30 is present in different fruits,
for example, Passiflora edulis.[42] Neither compound has ever
been reported from bacteria.
Conclusions
In summary, under the experimental conditions used in this
study, DMSP was catabolised by the Roseobacter clade bacteria
almost exclusively through the demethylation pathway to MeSH,
and not through the lysis pathway to DMS (Scheme 8), to result
in the release of several MeSH-derived sulfur volatiles, which have
all been identified by synthesis. In contrast, DESP catabolism not
only yielded sulfur volatiles arising from the dealkylation pathway,
but significant amounts of lysis products were found, represented
by 1 in P. gallaeciensis and O. indolifex, whereas the respective oxi-
dation product 11 was emitted by D. shibae.
The observed distribution between dealkylation and lysis of
DMSP in the Roseobacter clade in favour of the dealkylation
pathway, in accordance with previous reports,[14] has a high
impact on the understanding of the role of bacteria from the
Roseobacter lineage for the global sulfur cycle and climate, and
Further experiments on the genetics and enzymology of the
DMSP catabolic pathways will be carried out in our laborato-
ries to investigate the substrate specificities of particular en-
zymes in vitro.
Experimental Section
Strains, culture conditions and feeding experiments: Phaeobacter
gallaeciensis DSM 17395T, Oceanibulbus indolifex DSM 14862T and
Dinoroseobacter shibae DSM 16493T were precultured in marine
broth (MB2216, Roth) at 288C with shaking (180 rpm) until an
OD600 of about 1.0 was reached. For the feeding experiments,
MB2216 agar medium (25 mL) was spiked after autoclavation with
[2H6]DMSP·HCl, DESP·HCl, EMSP·HCl, DMSeP·HCl or DMTeP·HCl
(1 mm) and poured into glass Petri dishes. (Glass Petri dishes were
used to avoid contamination of the headspace extracts with plasti-
cisers.) The agar plates were inoculated with the preculture
(100 mL), incubated for two to three days at 288C and directly sub-
jected to headspace analysis.
Sampling: The volatile compounds emitted by the bacterial agar
plate cultures were collected by using the CLSA headspace tech-
nique. Briefly, in a closed apparatus containing the agar plates, a
circulating air stream was passed through a charcoal filter (Chrom-
Scheme 8. DMSP catabolism by P. gallaeciensis, O. indolifex, and D. shibae is
dominated by the demethylation pathway.
424
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 417 – 425

DMSP Catabolism in Marine Bacteria
[8] E. C. Howard, J. R. Henriksen, A. Buchan, C. R. Reisch, H. Bꢂrgmann, R.
Welsh, W. Ye, J. M. Gonzalez, K. Mace, S. B. Joye, R. P. Kiene, W. B. Whit-
man, M. A. Moran, Science 2006, 314, 649–652.
[9] J. E. Lovelock, R. J. Maggs, R. A. Rasmussen, Nature 1972, 237, 452–453.
[10] A. J. Kettle, M. O. Andreae, J. Geophys. Res. 2000, 105, 26793–26808.
[11] R. J. Charlson, J. E. Lovelock, M. O. Andreae, S. G. Warren, Nature 1987,
326, 655–661.
[12] S. M. Vallina, R. Simꢃ, Science 2007, 315, 506–508.
[13] R. P. Kiene, L. J. Linn, J. A. Bruton, J. Sea Res. 2000, 43, 209–224.
[14] D. C. Yoch, Appl. Environ. Microbiol. 2002, 68, 5804–5815.
[15] I. Wagner-Dçbler, B. Ballhausen, M. Baumgart, T. Brinkhoff, I. Buchholz,
B. Bunk, H. Cypionka, R. Daniel, T. Drepper, G. Gerdts, S. Hahnke, C. Han,
D. Jahn, D. Kalhoefer, H. Kiss, H.-P. Klenk, N. Kyrpides, W. Liebl, H. Liese-
gang, L. Meincke, A. Pati, J. Petersen, T. Piekarski, C. Pommerenke, S.
Pradella, R. Pukall, R. Rabus, E. Stackebrandt, S. Thole, L. Thompson, P.
Tielen, J. Tomasch, M. von Jan, N. Wanphrut, A. Wichels, H. Zech, M.
Simon, ISME J. 2009; DOI: 10.1038/ismej.2009.94
[16] V. Thiel, T. Brinkhoff, J. S. Dickschat, S. Wickel, J. Grunenberg, I. Wagner-
Dçbler, M. Simon, S. Schulz, Org. Biomol. Chem. 2010, 8, 234–246.
[17] J. S. Dickschat, H. B. Bode, S. C. Wenzel, R. Mꢂller, S. Schulz, ChemBio-
Chem 2005, 6, 2023–2033.
[18] a) H.-W. Chin, R. C. Lindsay, Food Chem. 1994, 49, 387–392; b) S. Bash-
kova, A. Bagreev, T. J. Bandosz, Ind. Eng. Chem. Res. 2002, 41, 4346–
4352; c) H.-W. Chin, R. C. Lindsay, J. Agric. Food Chem. 1994, 42, 1529–
1536; d) S. Helinck, H. E. Spinnler, S. Parayre, M. Dame-Cahagne, P. Bon-
narme, FEMS Microbiol. Lett. 2000, 193, 237–241.
[19] J. S. Dickschat, S. C. Wenzel, H. B. Bode, R. Mꢂller, S. Schulz, ChemBio-
Chem 2004, 5, 778–787.
[20] J. Meija, J. A. Caruso, Inorg. Chem. 2004, 43, 7486–7492.
[21] A. P. Kozikowski, A. Ames, Tetrahedron 1985, 41, 4821–4834.
[22] J. H. Ansede, P. J. Pellechia, D. C. Yoch, Environ. Sci. Technol. 1999, 33,
2064–2069.
tech GmbH, Idstein, Precision Charcoal Filter, 5 mg) for 24 h. The
absorbed volatiles were eluted from the charcoal with analytically
pure dichloromethane (30 mL). The obtained solutions were imme-
diately analysed by GC-MS and then stored at À808C. For compari-
son, agar plates containing the medium alone or amended with
[2H6]DMSP·HCl, DESP·HCl, EMSP·HCl, DMSeP·HCl or DMTeP·HCl
were investigated in the same way. None of the volatiles discussed
in the main text or shown in Tables S1–3 was detected in these ex-
periments. SPME sampling was carried out by drilling a small hole
into the (disposable) agar plates and incubating the SPME fibre in
the culture headspace for 1 h. Subsequently, the collected volatiles
were immediately desorbed from the SPME fibre into the injection
port of the GC-MS.
GC-MS: GC-MS analyses were carried out on a HP7890A GC system
connected to a HP5975C mass selective detector fitted with a HP-5
fused silica capillary column (30 mꢁ0.22 mm i.d., 0.25 mm film,
Hewlett–Packard). Conditions: inlet pressure: 67 kPa, He
23.3 mLminÀ1; injection volume: 1 mL; injector: 2508C; transfer
line: 3008C; electron energy: 70 eV. The GC was programmed as
follows: 508C (5 min isothermic), increasing at 58C minÀ1 to 3208C
and operated in splitless mode (60 s valve time); carrier gas (He):
1.2 mLminÀ1. Retention indices were determined from a homolo-
gous series of n-alkanes (C8–C32). Compounds were identified by
comparison of mass spectra to database spectra from the Wiley 7
library, or by comparison to a synthetic reference compound.
Abbreviations
[23] J. H. Ansede, D. C. Yoch, FEMS Microbiol. Ecol. 1997, 23, 315–324.
[24] R. G. Buttery, R. M. Seifert, J. Agric. Food Chem. 1977, 25, 434.
[25] R. P. Kiene, L. J. Linn, J. Gonzalez, M. A. Moran, J. A. Bruton, Appl. Envi-
ron. Microbiol. 1999, 65, 4549–4558.
[26] a) J. Knꢄblein, K. Mann, S. Ehlert, M. Fonstein, R. Huber, F. Schneider, J.
Mol. Biol. 1996, 263, 40–52; b) J. P. Ridge, K.-F. Aguey-Zinsou, P. V. Bern-
hardt, I. M. Brereton, G. R. Hanson, A. G. McEwan, Biochemistry 2002, 41,
15762–15769.
CLSA: closed-loop stripping apparatus, DESP: diethylsulfoniopropi-
onate, DMS: dimethyl sulfide, DMSeP: dimethylseleniopropionate,
DMSOR: dimethyl sulfoxide reductase, DMSP: dimethylsulfoniopro-
pionate, DMTeP: dimethyltelluriopropionate, EMPA: 3-(ethylmer-
capto)propionic acid, EMSP: ethyl methyl sulfoniopropionate,
MMPA: 3-(methylmercapto)propionic acid, MSePA: 3-(methylsele-
no)propionic acid, SPME: solid-phase microextraction.
[27] S. Schulz, J. S. Dickschat, Nat. Prod. Rep. 2007, 24, 814–842.
[28] K. Wilkins, Chemosphere 1996, 32, 1427–1434.
Acknowledgements
[29] J. S. Dickschat, T. Martens, T. Brinkhoff, M. Simon, S. Schulz, Chem. Biodi-
versity 2005, 2, 837–865.
[30] G. C. Kite, W. L. H. Hetterschieid, Phytochemistry 1997, 46, 71–75.
[31] H. Weenen, W. E. Koolhaas, A. Apriyantono, J. Agric. Food Chem. 1996,
44, 3291–3293.
[32] J. S. Dickschat, I. Wagner-Dçbler, S. Schulz, J. Chem. Ecol. 2005, 31, 925–
947.
We thank Prof. Dr. Stefan Schulz for excellent support. Funding
by the Deutsche Forschungsgemeinschaft (DFG) within the Emmy
Noether programme is gratefully acknowledged.
[33] M. Boelens, P. J. de Valois, H. J. Wobben, A. van der Gen, J. Agric. Food
Chem. 1971, 19, 984–991.
[34] X. Cai, P. C. Uden, E. Block, X. Zhang, B. D. Quimby, J. J. Sullivan, J. Agric.
Food Chem. 1994, 42, 2081–2084.
Keywords: dimethylsulfoniopropionate
selenium · sulfur · volatiles
·
Roseobacter
·
[35] T. G. Chasteen, Appl. Organomet. Chem. 1993, 7, 335–342.
[36] J. W. Swearingen Jr. , D. P. Frankel, D. E. Fuentes, C. P. Saavedra, C. C.
Vasquez, T. G. Chasteen, Anal. Biochem. 2006, 348, 115–122.
[37] M. A. Araya, J. W. Swearingen Jr. , M. F. Plishker, C. P. Saavedra, T. G.
Chasteen, C. C. Vasquez, J. Biol. Inorg. Chem. 2004, 9, 609–615.
[38] J. W. Swearingen Jr. , M. A. Araya, M. F. Plishker, C. P. Saavedra, C. C. Vas-
quez, T. G. Chasteen, Anal. Biochem. 2004, 331, 106–114.
[39] A.-L. Sunesson, C.-A. Nielsson, R. Carlson, G. Blomquist, G. Anderson,
Ann. Occup. Hyg. 1997, 41, 393–413.
[40] P. Bonnarme, C. Lapadatescu, M. Yvon, H.-E. Spinnler, J. Dairy Res. 2001,
68, 663–674.
[41] R. H. White, Science 1975, 189, 810–811.
[42] M. J. Jordan, K. L. Goodner, P. E. Shaw, J. Agric. Food Chem. 2002, 50,
1523–1528.
[1] S. J. Giovannoni, U. Stingl, Nature 2005, 437, 343–348.
[2] a) J. M. Gonzalez, R. Simo, R. Massana, J. S. Covert, E. O. Casamayor, C.
Pedros-Alio, M. A. Moran, Appl. Environ. Microbiol. 2000, 66, 4237–4246;
b) M. V. Zubkov, B. M. Fuchs, S. D. Archer, R. P. Kiene, R. Amann, P. H. Bur-
kill, Environ. Microbiol. 2001, 3, 304–311.
[3] U. Karsten, K. Kuck, C. Vogt, G. O. Kirst in Biological and Environmental
Chemistry of DMSP and Related Sulfonium Compounds (Eds.: R. P. Kiene,
P. T. Visscher, M. D. Keller, G. O. Kirst), Plenum Press, New York, 1996,
pp. 143–153.
[4] T. R. Miller, R. Belas, Appl. Environ. Microbiol. 2004, 70, 3383–3391.
[5] J. D. Todd, R. Rogers, L. Y. Gou, M. Wexler, P. L. Bond, L. Sun, A. R. J.
Curson, G. Malin, M. Steinke, A. W. B. Johnston, Science 2007, 315, 666–
669.
[6] A. R. J. Curson, R. Rogers, J. D. Todd, C. A. Brearly, A. W. B. Johnston, En-
viron. Microbiol. 2008, 10, 757–767.
[7] J. D. Todd, A. R. J. Curson, C. L. Dupont, P. Nicholson, A. W. B. Johnston,
Environ. Microbiol. 2009, 11, 1376–1385.
Received: November 4, 2009
Published online on December 30, 2009
ChemBioChem 2010, 11, 417 – 425
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
425