See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/275399681

Microbial Origins and Consequences of Dimethyl Sulfide

Article  in  Microbe (Washington, D.C.) · April 2012

DOI: 10.1128/microbe.7.181.1

CITATIONS READS
7 109

3 authors:

Andrew W B Johnston Andrew Robert Jonathan Curson

University of East Anglia University of East Anglia
234 PUBLICATIONS   9,239 CITATIONS    43 PUBLICATIONS   1,540 CITATIONS   

SEE PROFILE SEE PROFILE

Jonathan D Todd
University of East Anglia
85 PUBLICATIONS   2,057 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Aspergillosis in Caribbean corals View project

All content following this page was uploaded by Andrew W B Johnston on 30 May 2016.

The user has requested enhancement of the downloaded file.

Microbial Origins and Consequences
of Dimethyl Sulfide
Clouds, excitable penguins, and fun at the seaside– how is that for an
extended phenotype, Dr. Dawkins?
Andrew W. B. Johnston, Jonathan D. Todd, and Andrew R. J. Curson
Then, clean, medicinal and cold—the sea.
“Breathe in the ozone, John. It’s iodine.”
from “Summoned by Bells,”
by John Betjeman
nce we, too, were misled that the
O tangy seaside smell was ozone and
that it was good for us. In reality, it
is mostly a different gas– dimethyl
sulfide (DMS)— but one that ex-
erts remarkable effects, from fluffy clouds to
penguin behavior. No doubt former English
poet laureate John Betjeman (1906 –1984)
might have woven these facts into a folksy hom-
age to what is an intriguing microbiological
story—if only his chemistry had been up to it.
Uncovering the microbiological sources of di-
methyl sulfide goes back at least 80 years, begin-
ning with the discovery of red algal seaweeds
that emit this gas. Two decades later, the source
of this pungent volatile was identified as a
Summary
• Efforts to identify the microbiological sources
of dimethyl sulfide (DMS) gas began decades
ago when researchers realized it is a breakdown
product of dimethylsulfoniopropionate (DMSP).
• On a global scale, DMSP is made in huge quan-
tities in marine environments, and about 50
million tons of DMS are emitted into the atmo-
sphere each year.
• Research into DMSP catabolism reveals a sur-
prising diversity of lyase enzymes involved in
breaking down this compound.
• Although some lyases are associated with par-
ticular clades of bacteria, they are also prone to
long-range horizontal gene transfer.
Downloaded from www.asmscience.org by
On: Fri, 13 May 2016 15:58:11
breakdown product of dimethylsulfoniopropi-
onate (DMSP), a signature molecule for life at
sea. DMSP is made by many marine phyto-
plankton types— coccolithophores, notably
Emiliania huxleyi, which forms enormous
blooms; dinoflagellates such as Symbiodinium,
which interacts with many invertebrates includ-
ing corals; diatoms, especially those in polar
waters; and a few land plants that live by the
shore. In these organisms, DMSP likely serves as
an osmoprotectant, although it might support
other functions as well, such as alleviating oxi-
dative stress or defending against predation.
Dimethyl Sulfide Is a Major Global Player
Whatever the functions of that chemical, the
abundance of the organisms that make it, and its
remarkably high intracellular concentration
(more than 0.4 M in some dinoflagellates)
mean that DMSP is a major global sulfur
player, with about 109 tons being made
annually in the oceans and along their mar-
gins. Further, it is actively broken down by Andrew W. B.
marine microbes, and some of its products Johnston is Profes-
are themselves influential. sor of Biology, Jon-
The important step entails generating athan D. Todd is
DMS following DMSP cleavage (Fig. 1). RCUK Academic
DMS is the primary molecular conduit for Fellow, and Andrew
transferring sulfur from sea to air and, from R. J. Curson is Se-
there, back to the land via precipitation, nior Research Asso-
providing a critical link in the global sulfur ciate at the School
cycle. Before James Lovelock revealed this of Biological Sci-
step in 1972, hydrogen sulfide was consid- ences, University of
ered the key component in this cycle–a be- East Anglia, in the
lief that now seems odd, because the char- Norwich Research
acteristic “rotten egg” smell of that Park, Norwich NR4
molecule was not associated with the seas. 7TJ, England.
Volume 7, Number 4, 2012 / Microbe Y 181
IP: 192.3.182.83
 FIGURE 1

Simplified diagram of some of the catabolic bioconversions of dimethylsulfoniopropionate. The DMSP released by phytoplankton can be
demethylated to MMPA plus a methyl group that is transferred to acceptor molecule “X” (usually tetrahydrofolate) via the action of the
DmdA DMSP demethylase. Five other enzymes, DddL, DddP, DddQ, DddW and DddY, can cleave DMSP into DMS plus acrylate and one,
DddD, forms 3-hydroxypropionate (3HP) as the other C3 compound. Some of the DMS that is formed is liberated to the air, where its
oxidation products act as cloud condensation nuclei and can also be returned to the surface by precipitation.

Although marine microbes catabolize much
of the DMS, about 50 million tons of this
chemical compound escape to the atmosphere.
There it is oxidized to form a range of ions
that enable water molecules to coalesce, act-
ing as cloud condensation nuclei and reducing
solar radiance on the Earth’s surface.
In a very different guise, DMS is an info-
chemical. Thus, many marine animals— cope-
pod crustaceans, seals, and seabirds such as
penguins—are exquisitely sensitive to DMS and
swim, paddle, or fly towards it because it signals
potential food supplies for them.
Early studies of DMSP catabolism focused
on enzymes, known as “DMSP lyases,” in
some algae that also make this molecule.
These enzymes cleave DMSP into DMS plus
acrylate, a pair of molecules that might help to
defend these algae against grazing zooplank-
ton. How this process works at the level of the
gene is not understood for any eukaryotic
plankton—in part, because of difficulties in
obtaining bacteria-free cultures of such organ-
isms. Such purity is important because marine
bacteria catabolize DMSP, and many of them
associate closely with DMSP-producing eu-
karyotes.
The biochemical and physiological proper-
ties of bacterial DMSP lyases are very diverse,
enabling different bacterial species to deal
with DMSP in very different ways (Fig. 1). The
major route, accounting for about 70% of
total DMSP, depends on demethylating this
molecule to form 3-methiolpropionate
(MMPA). This metabolic process, which does
not release DMS, instead generates methane-
thiol, another volatile compound containing
sulfur.

182 Y Microbe / Volume 7, Number 4, 2012 Downloaded from www.asmscience.org by

IP: 192.3.182.83
On: Fri, 13 May 2016 15:58:11
Recent Efforts To Uncover the Molecular
Basis for DMSP Catabolism
The genetic basis for DMSP catabolism emerged
during the past five years, largely through the
efforts of Mary Ann Moran and William Whit-
man at the University of Georgia in Athens and
through research by our group in Norwich,
United Kingdom. Moran and Whitman remain
focused on the biochemical genetics of the
DMSP demethylation pathway and the molecu-
lar ecology of DMSP catabolism, while we
mainly examine the molecular genetics and
genomics of the pathways that liberate DMS
from DMSP.
These studies are revealing a remarkable di-
versity of the process at a molecular level, pro-
viding some surprises along the way. One key
portion of our work, exploring the different
types of the microbial Ddd enzymes (DMSP-
dependent DMS) that liberate DMS from
DMSP, depended on our following a three-part
approach: first, we isolated bacteria that grow
on DMSP as their sole carbon source and then
liberate DMS (Ddd⫹ phenotype), or else ob-
tained such strains from researchers at other
labs or from culture collections. Second, we
made gene libraries from such strains, cloned
them into a wide-host-range vector (usually cos-
mid pLAFR3), and conjugated these cosmids
into a tractable bacterial host such as Esche-
richia coli or Rhizobium that ordinarily does not
catabolize DMSP. We then screened for
transconjugants that grow on DMSP as a sole
carbon source or make DMS. Third, we local-
ized gene(s) by subcloning and analyzed them
experimentally and bioinformatically.
As with buses in London or other cities, after
a long wait to identify one ddd gene, along came
six in short succession for us to identify and
label as D, L, P, Q, W, or Y. The first of this
group, dddD, was identified in a ␥-Proteobacte-
rium, called Marinomonas, which we isolated
from mud around roots of the salt marsh grass
Spartina, one of the very few angiosperms that
make DMSP and which is abundant on the
North Norfolk coast of England. A single Mari-
nomonas gene, which we called dddD, con-
ferred a Ddd⫹ phenotype to E. coli when it was
cloned in an expression plasmid. However, this
Marinomonas strain does not grow on acrylate,
the generally anticipated initial C3 catabolite
formed after the DMS is cleaved from DMSP.
Downloaded from www.asmscience.org by
On: Fri, 13 May 2016 15:58:11
We therefore continued our quest at the Nor-
folk resort of Caister, where Jon Todd’s young
son Harry harvested some Ulva lactuca sea-
weed, which also makes DMSP. From this ma-
terial, we obtained another ␥-Proteobacterium,
Halomonas, which grows on both DMSP and
acrylate. It also contains dddD, but in a different
genomic neighborhood from where that gene is
situated in the genome of Marinomonas.
To explain this difference, we need to say
more about the enzyme DddD. It does not cata-
lyze cleavage of DMSP into DMS plus acrylate;
instead, the C3 product is 3-hydroxypropionate
(3HP). This result should not be too surprising,
because some bacteria yield 3HP, not the con-
ventional acrylate, from DMSP, as shown by
Duane Yoch and colleagues several years ago at
the University of South Carolina, Columbia.
Gene clusters near dddD— one in Halomonas
and the other in Marinomonas—include other
genes encoding enzymes that convert 3HP to
acetyl CoA before its entry to central metabo-
lism. Crucially, Halomonas but not Marinomo-
nas has two additional genes, acuN and acuK, in
the cluster whose encoded products act on acry-
late, also converting it to 3HP. Thus, Halomo-
nas grows on both DMSP and acrylate, and they
are metabolized independently, later converging
at 3HP.
The dddY gene was identified in the ␤-proteo-
bacterium Alcaligenes faecalis, which grows on
both DMSP and acrylate. It was isolated from a
Spartina stand by Yoch and colleagues more
than 20 years ago, and they later determined
that its DMSP lyase cleaves DMSP into DMS
plus acrylate and, importantly, that this enzyme
likely is located at the cell surface rather than the
cytoplasm, where several other DMSP lyases are
found. Its key ddd gene encodes a polypeptide
with a predicted N-terminal leader that tags it
for the periplasm and whose deduced sequence
matches what Yoch and his collaborators hero-
ically obtained by sequencing the N-terminus of
the purified enzyme itself.
We found that E. coli containing cloned dddY
has a Ddd⫹ phenotype and indeed cleaves
DMSP into DMS plus acrylate. Thus, Alcali-
genes has the textbook pathway in which DMSP
is broken down and the fragments then enter
central metabolism. The downstream steps are
catalyzed by products of several genes, acuN
and acuK, dddA, and dddC, that cluster near
dddY along the Alcaligenes genome, similar to
Volume 7, Number 4, 2012 / Microbe Y 183
IP: 192.3.182.83
 the cluster near dddD in Halomonas. This sim-
ilarity is striking in light of the taxonomic dis-
tance between these two strains and the substan-
tial dissimilarity of their DMSP lyases.
Roseobacters and the dddL,
dddP, dddQ and dddW Genes
Roseobacters are very widely distributed marine
␣-proteobacteria, one of whose signature traits
is the ability to catabolize DMSP. Indeed, Mo-
ran and her collaborators showed that several
strains, including Ruegeria pomeroyi, could de-
methylate DMSP and also cleave it to liberate
DMS. They dissected the former pathway, re-
vealing a novel set of biochemical reactions.
Our attention was drawn to the Roseobacters
because several genome-sequenced strains ap-
parently contain neither dddD nor dddY, even
though they have DMSP lyase activity. The ab-
sence of those genes suggested to us that those
strains likely contain other enzymes to cleave
DMSP.
Indeed they do. Different Roseobacter strains
contain one or more of no less than four genes,
dddL, dddP, dddQ, and dddW, each of which
encodes a different DMSP lyase that cleaves
DMSP into acrylate plus DMS. Moreover, we
should not forget that several of them also con-
tain the dmd genes, which encode the demethyl-
ation pathway enzymes. Strikingly, different
Roseobacters have distinct portfolios of ddd
genes. For example, Ruegeria pomeroyi DSS-3
has dddP, dddQ, and dddW, whereas Roseo-
varius nubinhibens ISM has two versions of
dddQ plus dddP.
Bioinformatic Insights from
Genomes and Metagenomes
In a sense, these six ddd genes are all doing much
the same thing and yet they have very different
designs. Thus, the enzyme DddD is in the family
of Class III CoA transferases, DddP looks like a
member of the M24 peptidases even though
DMSP is not a peptide, and DddY has no se-
quence similarity to any polypeptide with
known function. Similarly, DddL, DddQ, and
DddW were classified as belonging to domains
of unknown function. Nonetheless, these last
three share a small, poorly conserved motif that
resembles a cupin pocket, which may be the
184 Y Microbe / Volume 7, Number 4, 2012 Downloaded from www.asmscience.org by
IP: 192.3.182.83
On: Fri, 13 May 2016 15:58:11
catalytic site and may have evolved indepen-
dently.
Bioinformatics provides a facile way of find-
ing which organisms carry ddd genes. Although
particular Ddd lyases are sometimes associated
with particular clades of bacteria, these distribu-
tions follow no unambiguous pattern. For ex-
ample, genes encoding DddL, DddP, DddQ, and
DddW are found in the Roseobacters, but some
of these have traveled further afield, sometimes
spectacularly so. Thus, the DddP lyase also oc-
curs in some marine ␥-proteobacteria and, even
more strikingly, in some ascomycete fungi, sug-
gesting long-range, interdomain horizontal gene
transfers. It is surely no accident that some of
these fungi are pathogens of corals, which are
rich sources of DMSP.
In contrast, dddD is found mostly in various
marine ␥-proteobacteria, often clustered with
ancillary genes that variously encode DMSP
transporters, transcriptional regulators, and en-
zymes that catalyze downstream catabolic steps,
similar to those in Halomonas. However, dddD,
too, has done some taxonomic wandering, and
is found occasionally in strains of ␣- and ␤-pro-
teobacteria, some of which such as Rhizobium
and Burkholderia interact with the roots of an-
giosperms. This finding raises the notion that
these bacteria may associate with unknown
plants that make DMSP. Further, DddY is also
widely dispersed, sporadically, among the sub-
phyla of proteobacteria, including Shewanella,
Arcobacter, and Desulfovibrio, as well as Alcal-
igenes. The common thread for this gene is that
these bacteria live in microaerobic environ-
ments.
Looking beyond genomes to metagenomes,
information about the relative importance of the
different systems can be gleaned by counting the
homologues in metagenomic sequence reads,
most notably those in the Global Ocean Sam-
pling (GOS) database of subgenomic fragments
of marine bacteria. Among these, DddP and
DddQ are relatively abundant, followed by
DddL and DddD. However, DddY is wholly
absent from the GOS, perhaps because it lurks
in dark anaerobic depths, not in sunny surface
waters being sampled as part of this project. In
keeping with the global importance of the de-
methylation pathway, dmdA, encoding DMSP
demethylase, has more homologues than any of
the ddd genes, by some margin. This abundance
is due to its presence, not only in most Roseo-
bacters, but also in the SAR11 clade, the most
abundant bacteria on Earth.
Nonetheless, we need to keep in mind that
traits, including the stability, Km, and turnover
number of the corresponding enzymes may dif-
fer for the various Ddd and Dmd systems. Num-
bers, alone, of the different participants in a
population are not necessarily a true indicator of
how much they contribute to the global sulfur
cycle.
Many Ways To Catabolize DMSP
To account for finding six different ways to
catabolize DMSP, we offer four possible expla-
nations, ranging from prosaic to highly unlikely.
First, some of these enzymes might not really be
DMSP lyases even though some of them act on
this compound. In other words, they act mainly
on other, chemically similar, as yet unidentified
substrates. Countering that view is the fact that
the most of the various ddd and dmd genes are
just where one would expect them to be—in
marine bacteria.
Second, perhaps it is relatively easy for en-
zymes with different but chemically similar sub-
strates to evolve DMSP-cleaving activity from
different ancestral forms. Third, different pro-
teins with lyase activity may suit particular en-
vironments and substrate availability. Thus, in
the GOS, the DddL DMSP lyase was restricted
to a hypersaline lagoon in the Galapagos. Is this
chance or is the DddL protein particularly
adapted to high-salt environments?
Fourth and most speculatively, could DMSP
be a newcomer as a substrate of major environ-
mental influence? If so, then maybe the bacteria
and their enzymes have not had time to evolve
an optimal set of catabolic systems, and so these
are still in the learning stage, prior to the emer-
gence of the ideal version, after due passing of
evolutionary time. We, too, need more time to
unravel some of the questions that continue to
emerge as we study these important biotransfor-
mations.

ACKNOWLEDGMENTS
We are grateful to our colleagues Matt Sullivan, Emily Fowler, Mark Kirkwood, Nefeli Nikolaidou-Katsaridou, and Lei Sun
for their various contributions. Others, including Gill Malin and Michael Steinke, introduced us to the fascinating world of
DMSP. Thanks also to those, notably Mary Ann Moran and Steve Giovannoni, who sent us bacterial strains, allowing us to
clone by phone.
The work was funded largely by grants from the BBSRC and the NERC of the United Kingdom, and JDT was part-funded
by an RCUK Fellowship.

SUGGESTED READING
Curson, A. R. J., M. J. Sullivan, J. D. Todd, and A. W. B. Johnston. 2011. Catabolism of dimethylsulfoniopropionate:
microorganisms, enzymes and genes. Nature Rev. Microbiol. 9:849 – 859.
González, J. M., R. P. Kiene, and M. A. Moran. 1999. Transformation of sulfur compounds by an abundant lineage of marine
bacteria in the alpha-subclass of the class Proteobacteria. Appl. Environ. Microbiol. 65:3810 –3819.
Lovelock, J. E., R. J. Maggs, and R. A. Rasmussen. 1972. Atmospheric dimethyl sulfide and the natural sulfur cycle. Nature
237:452– 453.
Nevitt, G. A. 2011. The neuroecology of dimethyl sulfide: a global-climate regulator turned marine infochemical. Int. Comp.
Biol., in press.
Raina, J. B., E. A. Dinsdale, B. L. Willis, and D. G. Bourne. 2010. Do the organic sulfur compounds DMSP and DMS drive
coral microbial associations? Trends Microbiol. 18:101–108.
Reisch, C. R., M. A. Moran, and W. B. Whitman. 2011. Bacterial catabolism of dimethylsulfoniopropionate (DMSP).
Frontiers Microbiol. 2:172 (Epub).
Rusch, D. B. et al. 2007. The Sorcerer II Global Ocean Sampling expedition: northwest Atlantic through eastern tropical
Pacific. PLoS Biol 5:e77 2007.
Stefels, J., M. Steinke, S. Turner, G. Malin, and S. Belviso. 2007. Environmental constraints on the production and removal
of the climatically active gas dimethylsulphide (DMS) and implications for ecosystem modelling. Biogeochemistry 83:245–
275.
Vallina, S. M., and R. Simó. 2007. Strong relationship between DMS and the solar radiation dose over the global surface
ocean. Science 315:506 –508.
Yoch, D. C. 2002. Dimethylsulfoniopropionate: its sources, role in the marine food web, and biological degradation to
dimethylsulfide. Appl. Environ. Microbiol. 68:5804 –5815.

Downloaded from www.asmscience.org by Volume 7, Number 4, 2012 / Microbe Y 185

IP: 192.3.182.83
View publication stats
On: Fri, 13 May 2016 15:58:11