Marine bacteria from the Roseobacter clade produce sulfur volatiles via amino acid and dimethylsulfoniopropionate catabolism

Article abstract of DOI:10.1039/c4ob00719k

Dimethylsulfoniopropionate (DMSP) is a versatile sulfur source for the production of sulfur-containing secondary metabolites by marine bacteria from the Roseobacter clade. ³⁴S-labelled DMSP and cysteine, and several DMSP derivatives with modified S-alkyl groups were synthesised and used in feeding experiments that gave insights into the biosynthesis of sulfur volatiles from these bacteria. the Partner Organisations 2014.

Full text of DOI:10.1039/c4ob00719k

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
COMMUNICATION
Marine bacteria from the Roseobacter clade
produce sulfur volatiles via amino acid and
dimethylsulfoniopropionate catabolism†
Cite this: Org. Biomol. Chem., 2014,
12, 4318
Received 4th April 2014,
Accepted 14th May 2014
Nelson L. Brock, Markus Menke, Tim A. Klapschinski and Jeroen S. Dickschat*
DOI: 10.1039/c4ob00719k
www.rsc.org/obc
Dimethylsulfoniopropionate (DMSP) is a versatile sulfur source for
the production of sulfur-containing secondary metabolites by
marine bacteria from the Roseobacter clade. 34S-labelled DMSP
and cysteine, and several DMSP derivatives with modified S-alkyl
groups were synthesised and used in feeding experiments that
gave insights into the biosynthesis of sulfur volatiles from these
bacteria.
Dimethylsulfoniopropionate is
a compatible solute from
marine algae where it serves as a cryo- and osmoprotectant.
After its release to the oceans, it is taken up by marine bacteria
and degraded via competing pathways to methanethiol (MeSH)
or dimethyl sulfide (DMS, Scheme 1).1,2 Demethylation of
DMSP (1) to methylthiopropionic acid (2) with transfer of the
methyl group to the cofactor tetrahydrofolic acid (FH4), conver-
sion into its CoA thioester 3, oxidation with flavine adenine
dinucleotide (FAD) to 4, hydration, hydrolysis and NAD+ oxi-
dation give MeSH, acetaldehyde (5) and carbon dioxide.3,4
Four enzymes are involved in this process, and the molecular
mechanisms of the first and the last of them were deduced
from the enzymes’ structures.5,6 On the other side, there are
one DMSP hydrolase7 and five different DMSP lyases8–12
known that cleave DMSP to DMS and 3-hydroxypropionate (6)
or acrylate (7), respectively. Recently, the mechanistic details of
DMSP cleavage by one of the lyases, DddQ, were elucidated
with the crystal structure of the enzyme from R. pomeroyi.13
The DMSP degradation pathways are main contributors to
the formation of volatiles by bacteria from the Roseobacter
clade. In previous studies, sulfur volatiles from several Roseo-
bacter clade members were identified by use of a closed-loop
stripping apparatus (CLSA).14–18 Feeding experiments with
deuterated [methyl-2H6]DMSP and [methyl-2H3]methionine
Scheme 1 Enzymatic degradation of DMSP via (A) the demethylation
pathway and (B) the cleavage pathways.
resulted in the incorporation of isotope labelling into methylthio
groups of all sulfur volatiles with high rates,15–17 demonstrat-
ing that both sulfur metabolites are efficiently degraded to
volatile sulfur compounds. In contrast, isotopically labelled
34SO4 was not incorporated. However, these feeding experi-
2−
ments did not allow for unambiguous conclusions on the
sulfur source for volatile sulfur compounds. To address this
problem we have now synthesised [34S]DMSP and [34S]
cysteine19 that were both used in feeding experiments. More-
over, our previous investigations demonstrated that the chalco-
gene analogs of DMSP, dimethylseleniopropionate (DMSeP)
and dimethyltelluriopropionate (DMTeP), and ethylated DMSP
analogs were also catabolised by Roseobacter clade bac-
teria.15,17 In the present study the capability of degrading non-
natural DMSP analogs to sulfur volatiles by dealkylation and
lysis was further explored by feeding experiments with steri-
cally demanding and heterocyclic DMSP derivatives.
Institut für Organische Chemie, Hagenring 30, 38106 Braunschweig, Germany.
E-mail: j.dickschat@tu-bs.de; Tel: +49 531 3915264
†Electronic supplementary information (ESI) available: Incorporation rates of
feeding experiments with 34S-labelled compounds, results of headspace ana-
lyses, synthetic procedures, spectroscopic data, and copies of NMR spectra. See
DOI: 10.1039/c4ob00719k
4318 | Org. Biomol. Chem., 2014, 12, 4318–4323
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Communication
Scheme 2 Synthesis of [
34S]DMSP. (a) KCN, H2O, reflux, 99%; (b)
Scheme 3 Biosynthesis of methionine.
K34SCN, MeCN, reflux, 92%; (c) SmI2, THF–H2O (37 : 1), rt, 87%; (d) MeI,
KOH, MeOH, rt, 63%; (e) Me3O+BF4−, MeNO2, 0 °C, then CF3COOH,
rt, 51%.
S-methyl methanethiosulfonate (25, Scheme 4). Other sulfur
volatiles such as dimethyl tetrasulfide (24), the S-methyl alka-
A robust synthetic approach to DMSP proceeds via Michael nethioates 33–36, methyl (methylthio)methyl disulfide (37), 4-
addition of DMS to acrylate under acidic conditions.15 methylthio-2-butanone (41), and S-methyl 3-(methylthio)propa-
However, this approach is not practicable for the synthesis of nethioate (42), which have been reported previously from
[34S]DMSP, because [34S]DMS is not commercially available Roseobacter clade bacteria,14,15,17,18 also occurred occasionally.
and the conversion of 34S8 into this highly volatile compound The incorporation rates of the 34S-labelling into the sulfur vola-
endangers significant losses of expensive 34S-labelled material. tiles 15, 20, and 25 are shown in Fig. 2 (for other sulfur vola-
Therefore, a route to [34S]DMSP was developed that omits vola- tiles incorporation rates are summarised in Table 1 of ESI†).
tile intermediates (Scheme 2). Along this route 34S8 was trans- Feeding of [34S]DMSP to O. indolifex and P. inhibens resulted in
formed into K34SCN by treatment with a hot aqueous KCN high incorporation rates into 15 and 25 (ca. 60–70% in all
solution. The propionate portion of DMSP was built from tert- cases), while the incorporation rates for 20 were slightly lower
butyl 3-bromopropionate (8). Nucleophilic substitution with (ca. 50%). In contrast, incorporation of labelling from [34S]
K34SCN yielded the thiocyanate 9 that was reduced with samar- cysteine into 15 and 25 was comparably poor (<10%), but sig-
ium diiodide to the thiol 10, followed by two sequential methy- nificantly higher for 20 in both organisms (>20%). These data
lations, first with methyl iodide to the thioether 11 and second reveal that DMSP is efficiently degraded to MeSH via the
with Meerwein’s salt. Deprotection with trifluoroacetic acid demethylation pathway (Scheme 1) that can react to 15 and 25
and ion exchange chromatography resulted in [34S]DMSP by spontaneous oxidations in the presence of air (Scheme 5).
hydrochloride via five steps with an overall yield of 26%.
These oxidations may be catalysed by the charcoal in the CLSA
To investigate the biosynthesis of sulfur volatiles the filters or by ascorbate and transition metal ions.20,21 The for-
labelled compounds [34S]DMSP and [34S]cysteine19 that was mation of 20 by a similar oxidation requires two MeSH units
synthesised previously were fed to liquid cultures of Oceanibul- and one H2S unit that is not formed from DMSP, as indicated
bus indolifex DSM 14862T and Phaeobacter inhibens DSM by the significantly reduced incorporation rates into this vola-
17395. Isotopically labelled cysteine was additionally fed to a tile from [34S]DMSP in comparison to the incorporation rates
P. inhibens ΔpatB mutant that is impaired in the conversion of for 15 and 25. Instead, the additional sulfur in 20 can arise
cysteine via cystathionine into homocysteine, the precursor of from cysteine via transamination by the aspartate aminotrans-
methionine (accession number YP_006571551; PatB functions ferase AatA to mercaptopyruvate (44), followed by desulfurisa-
as cystathionine-β-lyase in the cleavage of cystathionine to tion by the mercaptopyruvate sulfurtransferase MST. MST can
homocysteine, pyruvate and ammonia, Scheme 3). The vola- transfer the sulfur of 44 to a variety of nucleophiles including
tiles that were emitted by the bacteria in these feeding experi- MeSH under formation of the byproduct pyruvate.22 The
ments were collected with a CLSA and analysed by GC-MS. resulting methyl disulfide (45) may then undergo oxidative
The results of the analyses are summarised in Table 1 of ESI† reactions to form dimethyl polysulfides such as 20. An alterna-
and representative total ion chromatograms are depicted tive pathway to H2S is the direct desulfurisation by the cysteine
in Fig. 1.
desulfurase CD. These mechanisms are in agreement with the
The headspace extracts of liquid cultures of O. indolifex and high incorporation rates from [34S]cysteine into 20. The com-
P. inhibens consistently contained the sulfur volatiles dimethyl parably low incorporation rates from [34S]cysteine into 15 and
disulfide (15) as main compound, dimethyl trisulfide (20), and 25 reveal that methylation of H2S or 45 are possible, but less
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4318–4323 | 4319

View Article Online
Communication
Organic & Biomolecular Chemistry
Scheme 4 Sulfur volatiles from O. indolifex and P. inhibens.
Fig. 1 Chromatograms of headspace extracts from (A) O. indolifex DSM
14862T after feeding of [34S]DMSP, (B) O. indolifex DSM 14862T after
feeding of [34S]cysteine, (C) P. inhibens DSM 17395 after feeding of
[
34S]DMSP, (D) P. inhibens DSM 17395 after feeding of [34S]cysteine, and
(E) P. inhibens DSM 17395 ΔpatB after feeding of [34S]cysteine. Peak
numbers refer to compound numbers in Scheme 4. Asterisks indicate
components originating from the medium and contaminants such as
plasticisers.
Fig. 2 Incorporation of 34S-label into the volatiles 15, 20, and 25. The
bars show incorporation rates of sulfur labelling (in %, mean of three
replicates standard deviation).
important. Conclusively, the formation of MeSH from cysteine
via the methionine pathway (Scheme 3) and degradation of clade by feeding of several synthetic DMSP analogs, including
methionine also does not play a major role. This is further DMSeP, DMTeP, ethylmethylsulfoniopropionate (EMSP), and
corroborated by the observed incorporation of labelling into diethylsulfoniopropionate (DESP).15,17 To further extend these
all three sulfur volatiles in the P. inhibens ΔpatB mutant. The investigations three new alkylated sulfoniopropionates,
increased incorporation rates for 15 and 25 demonstrate that methylpropylsulfoniopropionate (MPSP, 46), isopropylmethyl-
blockage of the methionine pathway results in higher turnover sulfoniopropionate (IMSP, 47), and tetramethylenesulfoniopro-
rates of cysteine towards sulfur volatiles along the desulfurisa- pionate (TMSP, 48), were synthesised by adapting a previously
tion pathways.
published procedure via acid-catalysed Michael addition of
In previous studies we have investigated the substrate speci- dialkyl chalcogenides to acrylic acid (Scheme 6).15 All three
ficity of the DMSP catabolism in bacteria from the Roseobacter DMSP derivatives were fed to O. indolifex and P. inhibens and
4320 | Org. Biomol. Chem., 2014, 12, 4318–4323
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Communication
whether the dealkylation pathway as shown in Scheme 1A is
responsible for their formation remains unclear, because only
the demethylase DmdA, but not the other enzymes of the
pathway (DmdBCD) are found within the available (incom-
plete) O. indolifex genome data. The low concentrations of the
propanethiol-derived volatiles reveal that the sterically
demanding MPSP is a possible, but not a good substrate for
the dealkylation pathway in O. indolifex. In particular, its
demethylation is much less efficient than the previously
reported EMSP demethylation.15
A very similar pattern of volatiles was observed in the
feeding experiment of MPSP to P. inhibens (Fig. 3B). Large
amounts of the lysis product 12 and traces of the corres-
ponding sulfone 30 were found that are presumably formed
from MPSP by the DMSP lyase DddP (Table 3 of ESI†). Further-
more, traces of the dealkylation pathway product 16 were
observed. MPSP dealkylation to propanethiol that, based on
genetic information, proceeds via a fully established dmd
pathway (Table 3 of ESI†) was found to be even less efficient
than in O. indolifex, as indicated by the traces of 16, while no
other compounds derived from propanethiol were found.
Feeding of IMSP gave a similar outcome (Fig. 3C). The
experiment with O. indolifex gave large quantities of isopropyl
methyl sulfide (13) and smaller amounts of its oxidation
product isopropyl methyl sulfone (31). IMSP dealkylation
proved to be more efficient as compared to MPSP dealkylation,
as indicated by the comparably large amounts of the propane-
2-thiol derivatives isopropyl methyl disulfide (17) and diisopro-
pyl disulfide (19). Diisopropyl trisulfide (23) and the oxidation
products of 17, S-isopropyl methanethiosulfonate (26) and
S-methyl propane-2-thiosulfonate (27), were also detected in
one sample. The feeding experiment with IMSP and P. inhibens
also resulted in the production of large amounts of 13 and 31,
while the products of IMSP dealkylation were completely
absent (Fig. 3D).
Scheme 5 Formation of sulfur volatiles from DMSP and cysteine
in P. inhibens and O. indolifex. AtaA (accession numbers: P. inhibens
YP_006573529, O. indolifex WP_007119281): aspartate aminotransferase,
MST: mercaptopyruvate sulfurtransferase (P. inhibens YP_006574318,
O. indolifex WP_007118660), CD: cysteine desulfurase (P. inhibens
YP_006572965, O. indolifex WP_007119584). Reactions that are likely
spontaneous oxidations on air are indicated by [O].
Another interesting experiment was the feeding of the
heterocyclic derivative TMSP. We assumed that this compound
cannot undergo a dealkylation, because in the first step of
Scheme 6 Synthesis of DMSP derivatives. (a) 2 M HCl, 80 °C, 54–88%.
the volatiles that were emitted during these experiments were DMSP demethylation a methyl group is transferred to the
collected by CLSA and the extracts were analysed by GC-MS. cofactor FH4 which would lead to formation of a covalent
The results of these analyses are summarised in Table 2 of bond between the substrate and the cofactor in case of TMSP
ESI† and representative chromatograms are shown in Fig. 3.
turnover. Accordingly, only the TMSP lysis product tetra-
Feeding of MPSP to O. indolifex resulted in large amounts hydrothiophene (14) and its oxidation product tetrahydrothio-
of the lysis product methyl propyl sulfide (12) and traces of its phene-1-oxide (32) were observed in the feeding experiments
oxidation product methyl propyl sulfone (30) (Fig. 3A). This with O. indolifex and P. inhibens (Fig. 3E and 3F). In addition,
result is interesting, because neither the DMSP hydrolase the oxidation product dihydro-2(3H)-thiophenone (38) was
DddD nor any of the five known DMSP lyases is encoded in the observed in O. indolifex. The mechanism of formation of 1,2-
genome of O. indolifex (Table 3 of ESI†), suggesting that still dithiane (39) that was observed as one of the major com-
another unidentified enzyme may be present in this organism. pounds in both organisms is less clear. One possibility may be
However, since the genome of this organism is not finished the nucleophilic ring opening of TMSP by attack of hydrogen
yet, it is also possible that the respective information is sulfide to 49, followed by lysis to 1,4-butanedithiol (50) and
missing. Furthermore, traces of methyl propyl disulfide (16), oxidation to 39 (Scheme 7).
its oxidation products S-propyl methanethiosulfonate (28) and
The structures of the thiosulfonate esters 26–29 as detected
S-methyl propanethiosulfonate (29), and dipropyl disulfide in these feeding experiments were suggested from their mass
(18) were found. These volatiles all arise from propanethiol spectra. For unambiguous compound identification the com-
that is itself formed from MPSP via dealkylation. However, pounds 26 and 28 were synthesised from the respective thiols
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4318–4323 | 4321

View Article Online
Communication
Organic & Biomolecular Chemistry
Fig. 3 Sulfur volatiles released by O. indolifex and P. inhibens after feeding of DMSP derivatives. Total ion chromatograms of headspace extracts
from (A) O. indolifex after feeding of MPSP, (B) P. inhibens after feeding of MPSP, (C) O. indolifex after feeding of IMSP, (D) P. inhibens after feeding of
IMSP, (E) O. indolifex after feeding of TMSP, and (F) P. inhibens after feeding of TMSP. Peak numbers refer to compound numbers in Scheme 4. Com-
pounds from other classes, medium compounds and contaminants are not mentioned.
Scheme 8 Synthesis of reference compounds. (a) Pyridine, 12 h,
Scheme 7 Suggested mechanism of formation of 39 from TMSP.
37–46% of thiosulfonate and 8–19% disulfide; (b) KMnO4, CuSO4,
CH2Cl2, 90 min, 17%.
(52, 53) and sulfonyl chloride 51 by esterification (Scheme 8).15
Since methanethiol is gaseous and highly malodorous, this
compound was substituted by sodium methanethiolate (56) in
Conclusions
the synthesis of 29 that was, however, low yielding (cf. ESI†). The biosynthesis of sulphur volatiles in bacteria of the Roseo-
The syntheses of 26 and 28 yielded also minor amounts of the bacter clade was investigated by feeding experiments with 34S-
labelled compounds. It was shown that DMSP degradation via
disulfides 19 and 18 that were likely formed by oxidation of
the thiols. All synthetic thiosulfonates proved to be identical to the demethylation pathway contributes for the most part to
the natural products, both in terms of retention times and the production of high amounts of methanethiol-derived
sulphur compounds such as dimethyl disulphide. On the con-
mass spectra. In addition, compound 39 was synthesised from
50 by oxidation with KMnO4 and CuSO4 as reported by Nourel- trary, the metabolic pathway of cysteine to methanethiol is
din et al.,23 unequivocally establishing the structure of natural only of minor importance as seen by the low incorporation
rates into these volatiles. However, incorporation of sulphur
39 as detected in the feeding experiments with TMSP.
4322 | Org. Biomol. Chem., 2014, 12, 4318–4323
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Communication
from cysteine into the sulphur atom in the middle position of
dimethyl trisulphide demonstrates its potential to transfer an
unmethylated reduced sulphur atom.
7 J. D. Todd, R. Rogers, L. Y. Gou, M. Wexler, P. L. Bond,
L. Sun, A. R. J. Curson, G. Malin, M. Steinke and
A. W. B. Johnston, Science, 2007, 315, 666.
Furthermore, the biosynthetic transformation of sterically
demanding DMSP analogues into propylated sulphur volatiles
proceeded mainly via the cleavage pathways. This revealed the
limits of the substrate promiscuity of the enzymes from the
8 A. R. J. Curson, R. Rogers, J. D. Todd, C. A. Brearley and
A. W. B. Johnston, Environ. Microbiol., 2008, 10, 757.
9 A. R. J. Curson, M. J. Sullivan, J. D. Todd and
A. W. B. Johnston, ISME J., 2011, 5, 1191.
demethylation pathway. The cyclic DMSP derivative TMSP was 10 J. D. Todd, A. R. Curson, C. L. Dupont, P. Nicholson and
degraded mainly by cleavage to tetrahydrothiophene but small A. W. B. Johnston, Environ. Microbiol., 2009, 11, 1376.
amounts of the side-product 1,2-dithiane could also be found. 11 J. D. Todd, A. R. J. Curson, M. Kirkwood, M. J. Sullivan,
A mechanism for the nucleophilic degradation of TMSP to 1,2-
dithiane was proposed. Future work in our laboratory will
R. T. Green and A. W. B. Johnston, Environ. Microbiol.,
2011, 13, 427.
include studies on the conversion of DMSP derivatives by puri- 12 J. D. Todd, M. Kirkwood, S. Newton-Payne and
fied recombinant enzymes.
A. W. B. Johnston, ISME J., 2012, 6, 223.
13 C.-Y. Li, T.-D. Wei, S.-H. Zhang, X.-L. Chen, X. Gao,
P. Wang, B.-B. Xie, H.-N. Su, Q.-L. Qin, X.-Y. Zhang, J. Yu,
H.-H. Zhang, B.-C. Zhou, G.-P. Yang and Y.-Z. Zhang, Proc.
Natl. Acad. Sci. U. S. A., 2014, 111, 1026.
14 V. Thiel, T. Brinkhoff, J. S. Dickschat, S. Wickel,
J. Grunenberg, I. Wagner-Döbler, M. Simon and S. Schulz,
Org. Biomol. Chem., 2010, 8, 234.
Acknowledgements
This work was funded by the Deutsche Forschungs-
gemeinschaft within the transregional collaborative research
centre SFB TR51 “Roseobacter” and a PhD scholarship of the
Fonds der Chemischen Industrie (to NLB).
15 J. S. Dickschat, C. Zell and N. L. Brock, ChemBioChem,
2010, 11, 417.
16 S. Hahnke, N. L. Brock, C. Zell, M. Simon, J. S. Dickschat
and T. Brinkhoff, Syst. Appl. Microbiol., 2013, 36, 39.
Notes and references
1 A. R. J. Curson, J. D. Todd, M. J. Sullivan and 17 N. L. Brock, C. A. Citron, C. Zell, M. Berger, I. Wagner-
A. W. B. Johnston, Nat. Rev. Microbiol., 2011, 9, 849.
2 C. R. Reisch, M. A. Moran and W. B. Whitman, Front. Micro-
biol., 2011, 2, 172.
Döbler, J. Petersen, T. Brinkhoff, M. Simon and
J. S. Dickschat, Beilstein J. Org. Chem., 2013, 9, 942.
18 J. S. Dickschat, I. Wagner-Döbler and S. Schulz, J. Chem.
Ecol., 2005, 31, 925.
3 E. C. Howard, J. R. Henriksen, A. Buchan, C. R. Reisch,
H. Bürgmann, R. Welsh, W. Ye, J. M. Gonzalez, K. Mace, 19 N. Brock, A. Nikolay and J. S. Dickschat, Chem. Commun.,
S. B. Joye, R. P. Kiene, W. B. Whitman and M. A. Moran,
Science, 2006, 314, 649.
4 C. R. Reisch, M. J. Stoudemayer, V. A. Varaljay, I. J. Amster,
M. A. Moran and W. B. Whitman, Nature, 2011, 473, 208.
2014, 50, 5487.
20 S. Bashkova, A. Bagreev and T. J. Bandosz, Ind. Eng. Chem.
Res., 2002, 41, 4346.
21 H.-W. Chin and R. C. Lindsay, Food Chem., 1994, 49, 387.
5 D. J. Schuller, C. R. Reisch, M. A. Moran, W. B. Whitman 22 G. D. Westrop, I. Georg and G. H. Coombs, J. Biol. Chem.,
and W. N. Lanzilotta, Protein Sci., 2012, 21, 289. 2009, 284, 33485.
6 D. Tan, W. M. Crabb, W. B. Whitman and L. Tong, 23 N. A. Noureldin, M. Caldwell, J. Hendry and D. G. Lee,
PLoS One, 2013, 8, e63870.
Synthesis, 1998, 1587.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 4318–4323 | 4323