Phytoplankton‐derived zwitterionic gonyol and dimethylsulfonioacetate interfere with microbial dimethylsulfoniopropionate sulfur cycling

Abstract The marine sulfur cycle is substantially fueled by the phytoplankton osmolyte dimethylsulfoniopropionate (DMSP). This metabolite can be metabolized by bacteria, which results in the emission of the volatile sulfur species methanethiol (MeSH) and the climate‐cooling dimethylsulfide (DMS). It is generally accepted that bacteria contribute significantly to DMSP turnover. We show that the other low molecular weight zwitterionic dimethylsulfonio compounds dimethylsulfonioacetate (DMSA) and gonyol are also widely distributed in phytoplankton and can serve as alternative substrates for volatile production. DMSA was found in 11 of the 16 surveyed phytoplankton species, and gonyol was detected in all haptophytes and dinoflagellates. These prevalent zwitterions are also metabolized by marine bacteria. The patterns of bacterial MeSH and DMS release were dependent on the zwitterions present. Certain bacteria metabolize DMSA and gonyol and release MeSH, in others gonyol inhibited DMS‐producing enzymes. If added in addition to DMSP, gonyol entirely inhibited the formation of volatiles in Ruegeria pomeroyi. In contrast, no substantial effect of this compound was observed in the DMSP metabolism of Halomonas sp. We argue that the production of DMSA and gonyol and their inhibitory properties on the release of volatiles from DMSP has the potential to modulate planktonic sulfur cycling between species.

. Marine bacteria can sustain up to 95% of their sulfur and 15% of their carbon requirements through the metabolization of DMSP (Zubkov et al., 2001). Two major metabolic pathways for the degradation of DMSP have been reported from bacteria ( Figure 1a). The demethylation/demethiolation pathway initially leads to the formation of 3-(methylthio) propionate that is the substrate for the release of methanethiol (MeSH) (Taylor & Gilchrist, 1991). The first step of this pathway is encoded in the dmdA gene which is widely distributed in marine bacteria (Howard, Sun, Biers, & Moran, 2008;Reisch, Moran, & Whitman, 2008;Varaljay et al., 2012). The DMSP-cleavage pathways to DMS are catalyzed by several different enzymes forming either acrylate (Alcolombri et al., 2015;Dickschat, Rabe, & Citron, 2015) or 3-hydroxypropionate as further reaction products (Todd et al., 2007).

and can be used for osmoregulatory functions in
Escherichia coli (Cosquer et al., 1999).
Until recently, dimethylsulfonio compounds besides DMSP were considered to be rather exotic and were reported from only a few algal species (Gebser & Pohnert, 2013;Nakamura et al., 1993Nakamura et al., , 1997. This view, however, is changing, with new data collected using ultra-high-pressure liquid chromatography-mass spectrometry (UHPLC-MS) for the direct monitoring of low molecular weight zwitterionic metabolites (Spielmeyer et al., 2011a(Spielmeyer et al., , 2011bSpielmeyer & Pohnert, 2010. Using this methodology, not only the widespread distribution of a diverse family of zwitterionic dimethylsulfonio-metabolites could be shown but also new and unexpected metabolites, such as dimethylsulfoxoniumpropionate (DMSOP), a biogenic dimethylsulfoxide precursor, were discovered (Thume et al., 2018).
We surveyed the two globally important microalgae Emiliania huxleyi and Prorocentrum minimum for the regulation of such zwitterionic metabolites during osmoacclimation (Gebser & Pohnert, 2013). Gonyol, previously a metabolite described in the dinoflagellate Lingulodinium (Gonyaulax) polyedra only, was found in both of the tested species and DMSA was detected in P. minimum (Gebser & Pohnert, 2013;Nakamura et al., 1993Nakamura et al., , 1997. This prompted us to undertake a survey of the distribution of these metabolites in a broader screening of phytoplankton species that is presented here.
We selected the prominent genetic model species  Table 1.

| Cultivation of algae
For the quantification of DMSP, DMSA, and gonyol, microalgae were cultured according to a published procedure (Thume et al., 2018 Maier and Calenberg (1994).
Lingulodinium polyedrum CCAP1221/2 was cultivated in L1 medium (Guillard & Hargraves, 1993). The medium for E. huxleyi RCC1217 and RCC1731 was prepared according to (Spielmeyer et al., 2011a(Spielmeyer et al., , 2011b. Cultivation was initiated from stationary phase stock cultures by a 20fold dilution of the cell suspension in 50 ml tissue culture flasks. Before inoculation, the medium was filtered (GF/C grade microfiber filter; GE healthcare) to remove precipitates. All cultures were grown at 12°C (a typical temperature reached in algal blooms in the North Sea  and North Atlantic (Jickells et al., 2008)) except the arctic isolate P. pouchetii, which was kept at 5°C, under a 14:10 light:dark cycle (Osram biolux lamps; 40 µmol/m 2 s -1 between 400 and 700 nm).
Cultures were grown to the exponential phase and then divided into four aliquots of equal volume. These aliquots were 20-fold diluted with fresh medium and cultivated to the exponential phase before harvesting for extraction.

| Algal sample preparation and analysis
Cultures (40 ml) for the screening of dimethylsulfonio-metabolites were filtered under reduced pressure (GF/C grade microfiber filter; GE healthcare) at 400 mbar, and the filter was immediately transferred to 4 ml glass vials containing 1 ml of methanol for extraction.
These samples were vortexed for 1 min before storage at −20°C.
For UHPLC-MS analysis, 50 µl of the extracts was diluted with 90 µl acetonitrile and 10 µl of an aqueous solution of an internal standard mixture (D 6 -DMSP, D 6 -DMSA, and D 3 -gonyol, the concentration of the standards was between 0.1 and 300 µM dependent on the concentration of the analytes that was estimated in a first prescreening).
For quantification of unmetabolized substrates immediately after quantification of DMS and MeSH, 100 µl of bacteria culture inoculations was diluted with 100 µl methanol. These suspensions were stored at −20°C until further analysis. For the gonyol-dependent DMSP-metabolization experiment, aliquots of 500 µl bacteria culture were added to 500 µl methanol in a microcentrifuge tube. After the addition of the isotope-labeled internal standards (D 6 -DMSP, D 6 -DMSA, and D 3 -gonyol) samples were centrifuged for 5 min at 16,100 g, and the supernatant was frozen at −20°C until measurement. An aliquot of the suspensions (50 µl) was diluted with 200 µl acetonitrile/water 9:1 in microcentrifuge tubes.
All samples were centrifuged before the measurement of the supernatant (5 min, 4,500 g). The supernatant was directly injected into an Acquity UHPLC (Waters) equipped with a SeQuant ZIC ® -HILIC column (5 μm, 2.1 × 150 mm, SeQuant, Umeå). A Q-ToF micro mass spectrometer (Waters Micromass) with electrospray ionization in positive ionization mode was used for detection and identification. For separation and quantification, the method of Gebser and Pohnert (2013) was used. The eluent consisted of 2% acetonitrile and 0.1% formic acid in high purity water (solvent A) and 90% acetonitrile with 10% water and 5 mM ammonium acetate (solvent B). The flow rate was set to 0.60 ml/min. The separation was performed at 35°C according to (Spielmeyer et al., 2011a(Spielmeyer et al., , 2011b.
For quantification of zwitterionic substances, relative response factors were determined by the measurement of an equimolar mixture of the analyte and the corresponding isotopically labeled internal standard. Response factors were calculated by comparison of the peak area of the analytes with the peak area of the corresponding internal standard.
Alcaligenes faecalis M3A and Halomonas sp. HTNK1 were cultivated in M9 minimal medium (Sigma-Aldrich). All cultures were grown under gentle shaking at 28°C with the addition of 10 mM sodium succinate as a carbon source. Exponentially growing cultures were selected for experiments on substrate utilization.

| Incubation of bacteria with zwitterionic molecules
Prior to incubation, 3 ml of bacteria cultures were washed three times by centrifugation (5 min at 16,100g) and resuspension in succinatefree medium to remove any excess organic carbon. For incubation experiments, all bacteria cultures were diluted with succinate-free medium to the same optical density (OD 600 ) of 0.1. Aliquots (450 µl) of these cultures were transferred into autoclaved 5 ml screw-cap glass vials with PTFE/silicone septa. Medium without the addition of bacteria was used as control. After the addition of aqueous solutions of TA B L E 1 Enzymes for DMSP-dependent DMS production (Ddd) identified in model organisms used in this work

| Stability of gonyol at alkaline pH
Gonyol (3.3 µM) in 1N NaOH was incubated for 1 hr at 30°C before the quantification of volatiles. Samples were vigorously shaken several times during incubation and prior to the measurements to achieve equilibrium between the liquid and gas phases. Headspace analysis was performed with GC-FPD as outlined below.

| GC-FPD measurement of MeSH and DMS
For quantification of MeSH and DMS in the headspace of the samples, the sealed vials were flushed with oxygen-free nitrogen for 1 min at a flow rate of 60 ml/min and cryogenic enrichment of the samples was carried out according to Franchini and Steinke (2017).
After rapid heating of the sample loop using freshly boiled water, the samples were introduced to a gas chromatograph (GC-2010,

| Quantification of bacterial growth
To determine the effect of the substrates on bacterial growth, stock cultures were cultivated and washed as outlined above. Bacteria cultures were transferred to autoclaved 20 ml headspace vials with cotton stoppers for further cultivation (28°C, shaking). After addition of 3.3 µM DMSP, DMSA, and gonyol, respectively, bacterial growth was monitored for 72 hr by measuring the optical density at 600 nm (OD 600 ) in standard single-use polystyrene cuvettes (Sarstedt AG & Co.) using a two-beam UV-Vis spectrophotometer (Specord M42, Carl Zeiss Jena). For each strain, a control without substrate addition was prepared. Measurements were performed with three biological replicates.

| Bacterial consumption of MeSH and DMS
In order to determine the consumption of volatile DMS and MeSH by the investigated bacteria species, cultures were prepared as men-

| Distribution of dimethylsulfonio-metabolites
We selected 16 phytoplankton species (eight diatoms, three dinoflagellates, a cryptophyte, and four haptophytes) to screen for intracellular DMSA and gonyol concentrations. As reported previously, gonyol was abundant in L. polyedrum, the dinoflagellate from which it was initially isolated (Gebser & Pohnert, 2013;Nakamura et al., 1993Nakamura et al., , 1997, with concentrations exceeding that of DMSP by more than 10-fold. In addition, our screening revealed this metabolite in quantities relative to the abundance of DMSP of ca. 5% in the haptophytes E. huxleyi and Isocrysis galbana and up to 24% in the dinoflagellates (Table 2). Gonyol was below the detection limit in any of the diatoms investigated.

| Metabolization of dimethylsulfoniometabolites
As a consequence of this broad distribution of DMSP, DMSA, and gonyol, the question arises on how these compounds influence and contribute to the marine microbial sulfur cycle. Laboratory experiments that challenged bacteria with pure DMSP showed a significant turnover of this compound Kiene, Linn, Gonzalez, Moran, & Bruton, 1999;Zubkov et al., 2001Zubkov et al., , 2002. Results in Table 2 suggest that these experiments might have been oversimplified, since bacteria will be frequently exposed to a  not pre-exposed to DMSP and, therefore, were not acclimated to utilize this substrate. Other studies on DMSP consumption in bacteria report that cultures that were grown on high concentrations (5 mM) of DMSP maximize the expression of enzymes required for DMSP catabolism Desouza & Yoch, 1995;Todd et al., 2011). We could not demonstrate DMS release from substrates other than DMSP (Figure 2e-H). This indicates that enzymes involved in DMS production in these bacteria are substrate-specific for DMSP. This high specificity might be explained by the enzyme mechanism recently identified for the DMSP lyase DddQ, from R. lacuscaerulensis (Li et al., 2014). This lyase relies on the abstraction of an acidic alpha proton from DMSP resulting in concomitant beta-elimination of DMS and the release of acrylate.
This elimination mechanism is excluded for the shorter chain length homolog DMSA and the longer chain length homolog gonyol due to the lack of an acidic proton in a suitable position of the substrate to support DMS elimination.
In contrast to the pathways leading to DMS, substrate utilization via the demethylation/demethiolation pathway is apparently not limited to DMSP. We detected elevated MeSH release after incubation of R. pomeroyi with DMSA (790 ± 350 nM MeSH; Figure 2e). MeSH release from DMSP is described in several bacteria (Miller & Belas, 2004;Taylor & Gilchrist, 1991), and here we extend this metabolic activity to the substrate DMSA. Ruegeria pomeroyi also releases MeSH from DMSP but the involved enzymes might not be the same since the electronic situation in both substrates is entirely different. While in DMSP enzymatic abstraction of the acidic α-proton facilitates its lysis, this is not possible for DMSA. In this metabolite demethylation and demethiolation by the attack on the C2-position would represent a plausible pathway for MeSH release. This is supported by findings of Reisch et al. (2008)  Statistical evaluation is given in Table 3 bacteria that accepts DMSA as a substrate might thus be responsible for the observed volatile production. Interestingly, this alternative pathway is efficient; MeSH release from DMSA in R. pomeroyi (790 ± 350 nM) was higher than the demethylation/demethiolation activity for DMSP that accounted for only 390 ± 31 nM MeSH release ( Figure 2e). Due to a lower outlier in the DMSA measurements TA B L E 3 Statistical analyses -p values indicate a statistical difference between treatment a and treatment b in the specific bacterial culture, n = 4 independent biological replicates a: p values - Figure 2a- Note: For comparison of two groups, an unpaired two-tailed t-test was performed. All statistical analyses were performed with a 95% confidence interval using Sigma-Plot version 13.0. p > .05 is considered not significantly different.
(Dean-Dixon test, N = 4, α = .1), the difference is not significant (p = .343, without outlier: p ≤ .001). The importance of this newly identified source for MeSH production lies in the high relevance of MeSH for sulfur assimilation by marine bacteria Visscher, Taylor, & Kiene, 1995). In the other three bacteria tested DMSA is not metabolized to any of the two volatiles since their concentrations are not exceeding those in the control or since they are not produced at all. In A. faecalis, MeSH release might be inhibited in the presence of DMSA (Figure 2g).
Even if all bacteria metabolized gonyol, no DMS or MeSH was released from this substrate in R. pomeroyi, Halomonas sp., and Sulfitobacter sp. (Figure 2e, f). This indicates the involvement of an alternative pathway that does not lead to cleavage of the C5-S bond.
As discussed above for the base-mediated transformation of gonyol, the lack of an acidic γ-proton does not allow a DMSP-lyase type pathway (Alcolombri et al., 2015). Alcaligenes faecalis responds to gonyol with a significantly higher MeSH release (140 ± 3 nM) compared to the untreated control (p = .029) (Figure 2g). It remains unclear whether this can be ascribed to a higher demethylation/demethiolation activity or a decrease in MeSH metabolism.
Interestingly, gonyol inhibited the MeSH release in  Statistical evaluation is given in Table 3 good approximation for net production over the incubation period.
In contrast, MeSH as a substrate is metabolized by all bacteria.
Inactivated bacteria (boiled controls) showed comparable concentrations as the medium controls so that nonspecific loss of the reactive MeSH in the presence of organic material can be ruled out (Figures A2 and A3 in Appendix 2). This suggests that the detected net production of MeSH in our experiments underestimates the gross production rate resulting from the close coupling of production and consumption processes.
In certain combinations (gonyol with R. pomeroyi and Sulfitobacter sp. or DMSA with A. faecalis), no net volatile emission was observed despite substantial metabolization of the administered substrates.

| Inhibition of volatile release from DMSP by gonyol and DMSA
The Alcaligenes faecalis with all three osmolytes DMSP, DMSA, and gonyol showed significantly higher net DMS production with 3,020 ± 170 nM than in the treatment with DMSP and gonyol (1970 ± 530; p = .009) and nearly the same as in the treatment containing DMSP or DMSA only (p = .134, Figure 3c). Together with the result that DMSA is not a source for DMS (Figure 2g), this suggests a protective effect of DMSA on the DMS-production activity in A. faecalis.
Halomonas sp. showed a slightly different pattern than the other bacteria ( Figure 3d). as a by-product (Todd et al., 2010(Todd et al., , 2007, whereas all other DMSproducing enzymes of the bacteria tested here co-produce acrylate ( Figure 1) .
Taken together these results show that the phytoplankton-derived zwitterionic dimethylsulfonio compounds DMSA and gonyol can affect the release of MeSH and the climatically active DMS by marine bacteria. Given their wide distribution (Table 2) implications of these findings for the marine sulfur cycle will have to be addressed using natural plankton communities. The additional sources for sulfur-containing volatiles have to be considered as well as possible inhibitory effects that might serve as indirect regulators of the marine sources of volatile sulfur.

| CON CLUS ION
We show that the zwitterionic algal osmolytes DMSA and gonyol are widely distributed in phytoplankton. As a consequence, bacterial communities will often be exposed to mixtures of these structurally related dimethylsulfonio-metabolites. The compounds and their inhibitory effect on the bacterial sulfur metabolism were highly species-specific. All bacteria tested were capable of metabolizing these substrates. However, the involved pathways apparently differed. The enzymatic release of MeSH from DMSA suggests a so far unrecognized demethylation/demethiolation pathway. Furthermore, gonyol strongly interfered with volatile release from DMSP in R. pomeroyi. This suggests that gonyol affects the marine sulfur cycle by modulating the metabolization of other potential substrates including DMSP. Future studies should consider the differential effects of these molecules on purified enzymes as well as in complex plankton samples to further our understanding of the mechanisms in bacterial degradation of DMSP and related substances.