A new route for synthesis of dimethylsulphoniopropionate in marine algae

Abstract

The 3-dimethylsulphoniopropionate (DMSP) produced by marine algae is the main biogenic precursor of atmospheric dimethylsulphide (DMS)^1,2,3. This biogenic DMS, formed by bacterial and algal degradation of DMSP⁴,⁵, contributes about 1.5 × 10¹³ g of sulphur to the atmosphere annually³, and plays a major part in the global sulphur cycle, in cloud formation and potentially in climate regulation¹,³. Although DMSP biosynthesis has been partially elucidated in a higher plant⁶,⁷, nothing is known about how algae make DMSP except that the whole molecule is derived from methionine^8,9,10,11,12. Here we use in vivo isotope labelling to demonstrate that DMSP synthesis in the green macroalga Enteromorpha intestinalis proceeds by a route entirely distinct from that in higher plants. From methionine, the steps are transamination, reduction and S -methylation to give the novel sulphonium compound 4-dimethylsulphonio-2-hydroxybutyrate (DMSHB), which is oxidatively decarboxylated to DMSP. The key intermediate DMSHB was also identified in three diverse phytoplankton species, indicating that the same pathway operates in other algal classes that are important sources of DMS. The fact that a transamination initiates this pathway could help explain how algal DMSP (and thereby DMS) production is enhanced by nitrogen deficiency¹².

Main

Many marine phytoplankton species and intertidal macroalgae accumulate DMSP as an osmolyte and cryoprotectant, particularly when salinity is high and nitrogen is limiting²,¹²,¹³. We investigated DMSP synthesis first in the green macroalga Enteromorpha intestinalis, which is rich in DMSP (∼20 μmol g⁻¹ fresh weight) when grown in normal sea water. To identify biosynthetic intermediates between methionine (Met) and DMSP, we followed the metabolism of a small dose of ³⁵S-methionine. The main fates of [³⁵S]Met were conversion to DMSP and incorporation into protein (Fig. 1a). Two stable compounds acquired ³⁵S rapidly and lost it as the [³⁵S]Met dose was depleted, as expected for DMSP pathway intermediates; these were 4-methylthio-2-hydroxybutyrate (MTHB) and its S -methylated derivative DMSHB, a new natural product (Fig. 1b). The most plausible route from Met to these compounds is via the unstable 2-oxo acid, 4-methylthio-2-oxobutyrate (MTOB). We therefore tested for ³⁵S incorporation into MTOB by using a gentle extraction method (Table 1, method A) or by converting it to a stable derivative (Table 1, method B). Both approaches confirmed that MTOB acquired and lost ³⁵S in parallel with MTHB. Computer modelling¹⁴ confirmed that the labelling kinetics of MTOB, MTHB and DMSHB were quantitatively consistent with roles as intermediates.

Figure 1: Labelling kinetics of selected small molecules and protein following administration of a tracer [³⁵S]Met dose to E. intestinalis.

Batches of fronds (100 mg fresh weight, FW) were supplied with 5 μCi (0.5 nmol) [³⁵S]Met. a, Free Met, DMSP and protein; b, Potential DMSP pathway intermediates; DMSHB data points are reduced by a factor of 20; MTPA, 3-methylthiopropylamine. No label (⩽0.5 nCi per 100 mg) was detected in 3-dimethylsulphoniopropionaldehyde.

Table 1 Metabolism of [³⁵S]Met by E. intestinalis

None of the other compounds monitored had labelling kinetics expected of intermediates in DMSP synthesis. S -Methylmethionine (SMM) and 3-dimethylsulphoniopropionaldehyde, both known to be DMSP synthesis intermediates in higher plants⁶,⁷, acquired little or no ³⁵S (Fig. 1b). 3-Methylthiopropylamine, which has been proposed to be an intermediate in algae¹¹, labelled as would be expected for a minor end product (Fig. 1b). Labelling of another hypothetical intermediate, 3-methylthiopropionate⁸,¹¹ varied from undetectable to comparable to that of MTOB in experiments with different lots of E. intestinalis. This can be attributed to a catabolic route involving oxidative decarboxylation of MTOB, which occurs in algae and other plants that do not contain DMSP, as well as in animals, and whose activity can vary with nutritional status⁹,^15,16,17.

These data support the pathway Met → MTOB → MTHB → DMSHB → DMSP, but are not consistent with a route involving SMM. To test this, ³⁵S-labelled SMM, MTHB and DMSHB were supplied (Table 2). SMM was scarcely metabolized, providing additional evidence that it is not an intermediate. MTHB was mainly converted to methionine, which in turn was incorporated into protein—a fate of MTHB well known in other plants and in animals¹⁸,¹⁹. Some ³⁵S from MTHB also entered DMSP, consistent with MTHB being an intermediate; however, an indirect route via methionine cannot be excluded. As would be expected for a late intermediate in the pathway, DMSHB was metabolized efficiently to DMSP and to little else.

Table 2 Uptake and metabolism of ³⁵S-precursors by E. intestinalis

Stable isotope labelling was used to confirm and extend the radiotracer findings. We first verified the presence of the novel intermediate DMSHB and tested whether its conversion to DMSP involves an oxygenase reaction, perhaps analogous to that mediated by lactate oxidase²⁰. To do this, E. intestinalis was given [U-¹³C₅]Met in an atmosphere containing ¹⁶O₂ or ¹⁸O₂. Gas chromatography–mass spectrometry (GC-MS) with selected ion monitoring (SIM) revealed a small pool of DMSHB that became labelled with ¹³C in five or six positions, but not with ¹⁸O (Fig. 2a). Fast atom bombardment (FAB)-MS showed that most of the newly synthesized DMSP had four or five ¹³C atoms and that 30–40% of this DMSP became labelled with ¹⁸O (Fig. 2b). Note that recycling of the [¹³C₄]homocysteine moiety formed in methylation reactions would yield Met (and hence DMSHB and DMSP) with an unlabelled methyl group. The presence of such Met was confirmed by GC-MS, and modelling¹⁴ showed the observed proportions of [¹³C₄]- and [¹³C₅]DMSP conformed to expectations for recycling of the [¹³C₄]homocysteine moiety generated by the methylation step in DMSP synthesis. Also, photosynthetically generated ¹⁶O₂ lowers the intracellular abundance of ¹⁸O₂. The ¹³C and ¹⁸O data are thus fully consistent with a pathway in which the last step is an oxygenase-mediated oxidative decarboxylation of DMSHB.

Figure 2: Evidence for conversion of DMSHB to DMSP by an oxygenase reaction in E.intestinalis.

Fronds (100 mg) were incubated for 24 h with L-[U-¹³C₅]Met (5 μmol) in flasks initially containing 21% ¹⁶O₂ or ¹⁸O₂ in N₂; controls received no Met and no ¹⁸O₂. a, SIM analysis of DMSHB. The peak areas in the control fit the expectation for natural-abundance C, S, Si and O isotopes. In the [¹³C]Met/¹⁶O₂ treatment, endogenous DMSHB was replaced by ¹³C₄- and ¹³C₅-labelled species (the S -demethylated derivatives of [¹³C₅]- and [¹³C₆]DMSHB). Exposure to ¹⁸O did not change this pattern, showing that the α-hydroxyl group does not originate from O₂. b, FAB-MS analysis of DMSP. In the [¹³C]Met/¹⁶O₂ treatment, the peaks at m/z 139 and 140 are [¹³C₄]- and [¹³C₅]DMSP. Under ¹⁸O₂, the signals at m/z 141 and 142 ([¹³C₄, ¹⁸O]- and [¹³C₅, ¹⁸O]DMSP) show that 30–40% of the ¹³C-labelled DMSP molecules contain an ¹⁸O atom.

We also used stable isotope labelling to investigate the conversion of methionine to MTOB. E. intestinalis was given [¹⁵N]Met (5 μmol per 100 mg) and GC-MS was used to follow the labelling of amino acids. Glutamate acquired ¹⁵N readily (7.0% abundance at 2 h), as did aspartate and alanine, but the amide group of glutamine did not (<1% abundance at 2 h). This is consistent with transamination of methionine, but not with oxidative deamination: the ¹⁵NH₃ from a deamination reaction would have led to labelling of glutamine amide via the action of glutamine synthetase²¹. The operation of glutamine synthetase in E. intestinalis was confirmed by supplying ¹⁵NH₄⁺(1 μmol per 100 mg) and showing that glutamine amide nitrogen was rapidly labelled (42% ¹⁵N abundance at 2 h). The conversion of MTHB to Met (Table 2) indicates that the Met → MTOB step is reversible, which is consistent with transamination²².

Collectively, the data for the chlorophyte macroalga E. intestinalis indicate that the pathway for synthesis of DMSP is that shown in Fig. 3. We tested for this pathway in marine planktonic species as these are major producers of DMSP and DMS^1,2,3. For this, we chose a prymnesiophyte (Emiliania huxleyi), a diatom (Melosira nummuloides) and a prasinophyte (Tetraselmis sp.). These algae all contained small pools of the key intermediate DMSHB which acquired label from [³⁵S]Met and lost it as the [³⁵S]Met was consumed (Fig. 4). All of them metabolized supplied [³⁵S]DMSHB to [³⁵S]DMSP (Fig. 4, inset). It is therefore likely that they have the same pathway as E. intestinalis.

Figure 3: Proposed pathway of DMSP synthesis in E.intestinalis and other algae.

The second step is shown as reversible because MTHB can be converted to Met (Table 2).

Figure 4: Evidence that DMSHB participates in DMSP synthesis in Emiliania huxleyi (EH), Melosira nummuloides (MN) and Tetraselmis sp. (TS).

The main figure shows conversion of [³⁵S]Met to DMSP and DMSHB. Data are for 10⁷(EH and TS) or 10⁵cells (MN). Cultures were incubated with [³⁵S]Met (8–10 μCi; about 0.5 nmol) for a short time (t₁) to label heavily the pools of free Met and intermediates, and for a long time (t₂) to allow ³⁵S to chase from these pools. Times t₁ and t₂ (h), culture volumes (ml) and cell numbers were: EH, 1.5, 23, 12, 5× 10⁷; MN, 2.5, 23, 1.2, 5× 10⁵; TS, 0.25, 16, 25, 4.4 × 10⁷. The inset shows synthesis of [³⁵S]DMSP (nCi per 10⁷ cells) from [³⁵S]DMSHB (2–3 μCi, 0.2–0.3 nmol). Incubation was for 17–23 h. Culture volumes (ml) and cell numbers were: EH, 10, 5.5 × 10⁷; MN, 1.2, 5× 10⁵; TS, 10, 1.4 × 10⁷.

Our data establish a pathway for DMSP biosynthesis in marine algae. This pathway has no steps in common with that in higher plants, which proceeds via SMM and 3-dimethylsulphoniopropionaldehyde⁶,⁷. DMSP biosynthesis must therefore have evolved independently at least twice. Our results have two other implications. The first stems from the finding that a transaminase reaction stands at the head of the DMSP pathway; this may help explain why nitrogen deficiency enhances DMSP production¹²,²³,³⁰. Depletion of cellular amino acids would favour the transamination reaction, thereby promoting DMSP synthesis when nitrogen is limiting. Second, our results suggest that DMSP may not be the only precursor of the DMS produced by living algae: DMSHB is another potential precursor in vivo. In support of this possibility, we have obtained preliminary evidence for extensive catabolism of supplied DMSHB to DMS in Tetraselmis sp. and E. huxleyi.

Methods

Algae. E. intestinalis was collected in Florida and kept in aerated sea water at 18 °C in continuous fluorescent light (photosynthetic photon flux density, 50 μmol m⁻² s⁻¹). M. nummuloides (CCMP 482) and Tetraselmis sp. were cultured axenically at 25 °C in the above light regime in modified Gooday's medium²⁴ with 1 mM NO₃⁻; for M. nummuloides, 0.1 mM Na₂SiO₃ was added and Tris omitted. E. huxleyi (CCMP 373) was grown in f/2 medium²⁵ in daylight at 22 °C.

Radiochemicals. L-[³⁵S]SMM was synthesized enzymatically from L-[³⁵S]Met²⁶ or by treating L-[³⁵S]Met with 250 mM methanol in 6 M HCl at 110 °C for 4 h (ref. 27), and converted to L-[³⁵S]DMSHB with HNO₂ (ref. 28). [³⁵S]MTOB was made from L-[³⁵S]Met using L-amino acid oxidase, and converted to L-[³⁵S]MTHB using L-lactic dehydrogenase and NADH; [³⁵S]methylthiopropionate was obtained as a byproduct. Compounds were purified by ion exchange and TLE⁶,⁷.

Labelling conditions. E. intestinalis fronds (50 or 100 mg) were incubated in 0.5–2.5 ml sterile sea water; ¹⁸O labelling was carried out in 50-ml flasks. For phytoplankton species, labelled compounds were added to growing cultures. Incubation was at 18–21 °C for E. intestinalis and E. huxleyi and 25 °C for Tetraselmis and M. nummuloides, under fluorescent light and with gentle agitation. Uptake of ³⁵S was estimated from its disappearance from the medium.

Metabolite analysis. Most metabolites were isolated by methanol–chloroform–water extraction, ion exchange, TLE and TLC⁶,⁷. As MTOB broke down in these procedures, forming MTP, these compounds were isolated using 0.1 M HCl followed by ether extraction⁶ (method A) or by using paired samples extracted in 0.3 ml 66 mM NaBH₄ (to reduce MTOB to MTHB) or in pre-neutralized NaBH₄ (to give the endogenous MTHB level) (method B); MTOB was estimated by difference . Method B was also used to isolate 3-dimethylsulphoniopropionaldehyde as its hydroxy derivative⁷. Metabolites were identified by their respective lability and stability in cold 2 M NaOH. ³⁵S data were corrected for recovery, determined using labelled standards, and for products formed during extraction. Protein synthesis was estimated from ³⁵S labelling of the insoluble residue; the ³⁵S was shown to be in protein-bound Met by proteolysis and TLC.

Mass spectrometry. DMSP was analysed without derivatization by FAB-MS²⁹. DMSHB was derivatized as its t -butyldimethylsilyl ester/ether and analysed by GC-MS with SIM after on-column nucleophile-assisted S -demethylation²⁹. Authentic DMSHB⁶ was used to calibrate the SIM parameters. The diagnostic fragment ion cluster at m/z 321 (loss of a t -butyl radical) was monitored at the appropriate retention time. ¹⁵N-amino acids were analysed as described¹⁴, except that amide-¹⁵N abundance was determined using N -ethoxycarbonyl isobutyl esters by electron impact GC-MS³⁰.