Letters pubs.acs.org/acschemicalbiology
Assigning the Algal Source of Dimethylsulfide Using a Selective Lyase Inhibitor Uria Alcolombri,†,‡ Lei Lei,† Diana Meltzer,† Assaf Vardi,‡ and Dan S. Tawfik*,† †
Department of Biomolecular Sciences and ‡Department of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot, Israel S Supporting Information *
ABSTRACT: Atmospheric dimethylsulfide (DMS) is massively produced in the oceans by bacteria, algae, and corals. To enable identification of DMS sources, we developed a potent mechanismbased inhibitor of the algal Alma dimethylsulfoniopropionate lyase family that does not inhibit known bacterial lyases. Its application to coral holobiont indicates that DMS originates from Alma lyase(s). This biochemical profiling may complement meta-genomics and transcriptomics to provide better understanding of the marine sulfur cycle.
G
enerated in oceans at more than 107 tons annually, dimethyl sulfide (DMS) comprises a key component of the ocean sulfur cycle.1 Its precursor, dimethylsulfoniopropionate (DMSP), is lysed by DMSP lyases, thus releasing DMS. Both DMS and DMSP are key players in the ocean microbial food webs, functioning as infochemicals in trophic-level interactions, 2,3 and possibly playing a role in climate atmosphere−ocean feedback processes.4,5 It has been suggested that DMSP in the ocean is made solely by algae and is subsequently catabolized, mainly by marine bacteria, to produce methanethiol6 or DMS.7,8 However, animals such as corals were also found to make DMSP.9 Recently, the first algal DMSP lyase, Alma1, was identified in Emiliania huxleyi (Ehux),10 a bloom-forming coccolithophore with a profound contribution to the oceanic carbon and sulfur cycles.11 Sequence searches suggested that Ehux-Alma1 is the first characterized member of an entire family of DMSP lyases present in a wide a range of algae from diverse environments including the coral symbiont dinoflagellate algae, Symbiodinium sp., and from its coral host.10 The Alma DMSP lyase family belongs to the Asp/Glu/ hydantoin racemase superfamily12 and characteristically possesses two active site cysteines, one of which seems to act as the primarily base catalyst (Cys265; Figure 1A).10 This family is evolutionary and mechanistically distinct from a long list of previously identified marine bacterial DMSP lyases (Ddd families).13−15 Alma’s identification is a major step toward answering the key question: what is the biological origin of oceanic DMS?8 It remains, therefore, unclear what the relative contribution is of various species, be they bacteria or algae, to the global DMS released from the marine environment. Furthermore, since a large fraction of DMS is released from coral reefs,16 which are holobionts comprised of the coral © XXXX American Chemical Society
animal, the Symbiodinium dinoflagellate algae, and bacterial microbiome, it remains unknown which biological source(s) predominate in DMS release. Large data sets of marine metagenomes17,18 may provide an estimation of the potential capability of oceanic bacterial populations to produce DMS.14 However, DMSP lyase gene counts per se, or even transcript counts, do not report the actual enzymatic activity of the different classes of enzymes. Further, the potential role of yet unidentified DMSP lyases cannot be excluded. As a complementary approach to genomics−transcriptomics, we aimed to use a chemical genetics approach, initially by developing a distinct yet facile biochemical identifier of Alma DMSP lyases. Here, we describe a potent mechanism-based inhibitor, 2-bromo-3-(dimethylsulfonio)-propionate (BrDMSP; Figure 1A and B), that selectively inhibits the Alma enzyme family and does not inhibit any of the other known DMSP lyase families. This inhibitor comprises the first biochemical probe that can be applied to examine the contribution of different DMSP lyase classes in marine environments. Ehux-Alma1 lost its activity upon treatment with thiol alkylating reagents such as iodoacetamide.10 Accordingly, a C265A mutant of Ehux-Alma1 showed a complete loss of function (by comparison, mutating the second active-site cysteine, C108A, only resulted in 70-fold decline in activity). These findings suggested that selective mechanism-based inactivation via alkylation of the active-site cysteine(s) would be possible. We thus synthesized 2-bromo-3-(dimethylsulfoReceived: September 25, 2016 Accepted: November 28, 2016 Published: November 28, 2016 A
DOI: 10.1021/acschembio.6b00844 ACS Chem. Biol. XXXX, XXX, XXX−XXX
Letters
ACS Chemical Biology
Figure 1. Ehux-Alma1’s activity inhibited by Br-DMSP. (A) Ehux-Alma1’s proposed catalytic mechanism: Cysteine 265 acts as a general-base, thus promoting the β-elimination of DMSP to yield DMS and acrylate. (B) The proposed inhibition mechanism by Br-DMSP. (C) Ehux-Alma1 (closed circles) or Sym-Alma (open circles; 1 μM) were incubated for 1 h with varying concentrations of Br-DMSP (0.015−4 μM). The residual DMSP lyase activity is shown relative to the untreated enzyme. Data were fitted to a standard inhibition equation with a variable cooperativity coefficient, relative activity = 1/[1 + ([Br-DMSP])/IC50)n], to give IC50Ehux = 0.09 ± 0.005 μM and IC50Sym 0.13 ± 0.005 μM; n(Ehux) = 1.4 ± 0.1, n(Sym) = 2 ± 0.14, and R2 > 0.99. A close-up on the 0−1 μM concentration range is shown in the embedded box. (D) Inhibition kinetics under second-order reaction conditions ([Br-DMSP]0 ≈ 1 μM, [E]0 ≈ 1.1 μM; n = 2). A fit to a second-order reaction with equal component concentrations (1/[E] vs time) resulted in systematic deviations (data not shown). Thus, data were fitted to an exponential decay with an apparent half-life of 133 s. (E) Michaelis− Menten plot for Sym-Alma. The derived parameters were as follows: kcat = 427 ± 44 s−1; KM = 15.8 ± 2.7 mM; kcat/KM = 2.7× 104 M−1 s−1, [E]0 = 0.4 μM; via a direct fit to the Michaelis−Menten model, R2 = 0.989 and n = 2.
nio)-propionate (Br-DMSP; Supporting Methods), which retains DMSP’s structure yet possesses a bromide leaving group at C2 where deprotonation of DMSP normally occurs. Thus, instead of the β-elimination catalyzed by the thiolate side chain of C265 acting as a general-base, the designed inhibitor, Br-DMSP, is expected to promote a nucleophilic substitution leading to a covalent DMSP-enzyme adduct (Figure 1). Br-DMSP was found to be a highly potent mechanism-based inhibitor, completely inhibiting Ehux-Alma1 at micromolar concentrations and within minutes of adding it to the purified enzyme (Figure 1C,D). Active-site titration with Br-DMSP indicated that Ehux-Alma1’s activity was fully inhibited at an approximately stoichiometric ratio of Br-DMSP to enzyme. Although Br-DMSP comprises two enantiomers, the kinetics of inhibition coincides with a single reaction phase (see below). However, considering the error range in concentrations of enzyme (e.g., absorbance at 280 nm as compared with protein quantification by Bradford) and Br-DMSP (being highly hygroscopic), it is possible that one enantiomer is responsible for the observed rapid inhibition. As expected for a covalent inhibitor, inhibition was retained upon ≥5000-fold dilutions of the inhibited enzyme to a buffer
with no inhibitor and even after overnight dialysis (Figure 2A). The kinetics of inhibition appears to be fast, as demonstrated by full inhibition at 10 μM Br-DMSP within less than 5 min (at 0.5 μM enzyme). Under such pseudo-first-order conditions ([Br-DMSP] ≫ [enzyme]), a time course could not be measured, because measuring DMS release takes about 5 min (using GC-FPD; Methods) during which inhibition progresses considerably. Second-order kinetics could be followed ([BrDMSP] ≈ [enzyme]), indicating an apparent half-life of approximately 2 min (Figure 1D). The comparatively high rate of inhibition is also in agreement with the fact that the substrate, DMSP, even at above KM concentrations, only partially prevented inhibition by Br-DMSP (Figure 2B). Clearly, Br-DMSP’s binding affinity to EhuxAlma1 is higher than DMSP (KM = 9 mM), probably owing to its higher hydrophobicity. Similarly, glutamate racemases, that are also members of the Asp/Glu racemase superfamily, are competitively inhibited by a hydrophobic substrate analogue, gamma-2-naphthylmethyl-D-glutamate.19 However, hydrophobicity per se is insufficientthe sulfonium moiety of Br-DMSP underlines its selectivity to Ehux-Alma1’s active site. Accordingly, iodoacetamide, that is inherently far more reactive than B
DOI: 10.1021/acschembio.6b00844 ACS Chem. Biol. XXXX, XXX, XXX−XXX
Letters
ACS Chemical Biology
Figure 2. Br-DMSP an irreversible inhibitor. (A) Ehux-Alma1 activity 5 min after the addition of 10 μM Br-DMSP and subsequent overnight dialysis against a buffer with no Br-DMSP. Activity is normalized to an identical enzyme sample without Br-DMSP (O.N, overnight dialysis; n = 2). (B) DMSP inhibits Br-DMSP inhibition. The activity of 1 μM Ehux-Alma1 tested 2 min after the addition of 1 μM Br-DMSP in the presence or absence of 20 mM DMSP (n = 2). (C) Inhibition of Ehux-Alma1 and of the cysteine 108 to alanine mutant (C108A; ∼70-fold lower activity compared to the wild-type enzyme samples). Activities are normalized to the same enzyme samples without Br-DMSP (n = 2).
Figure 3. Br-DMSP is an Alma specific inhibitor that can be applied to identify the origin of DMSP lyase activity in environmental samples. (A) Ehux-Alma1 and Sym-Alma (crude lysates, or isolated enzyme) were treated with 10 μM Br-DMSP for 1 h. The bacterial Ddd enzymes were treated with 100 μM Br-DMSP. DddD was supplemented with 0.5 mM Acetyl-CoA, which is essential for its activity.21 The levels of DMSP lyase activity differ between the treatments and are thus presented as relative to the same enzyme without Br-DMSP (n.d, not detected; n = 2; enzyme concentrations: Ehux-Alma1, 8 nM; Sym-Alma, 30 nM; DddP, 200 nM; DddW, 625 nM; DddY, 1 nM; DddD, 22 nM; DddK, 667 nM; all lysate concentration: DddL, 400 μg/mL; DddQ, 20 μg/mL; Sym-Alma lysate, 50 μg/mL; E. huxleyi 373 lysate, 7.5 μg/mL). (B) DMSP lyase activity in crude extracts derived from coral fragments taken from three independent Acropora millepora colonies and one colony of Stylophora (see Methods for reaction conditions). “+Br-DMSP” denoted parallel samples to which Br-DMSP was added to the lysis buffer at the noted concentrations (n = 3). The measured DMSP lyase activity was normalized to total protein concentrations quantified by Bradford.
Br-DMSP, is a less potent inhibitor (e.g., ∼60% inhibition of 12.5 nM Alma1 was observed after 1 h of incubation with 2 μM iodoacetamide),10 and 2-bromoacrylate, a product analogue, does not inhibit Ehux-Alma1 at 100-fold concentration higher than Br-DMSP (Supporting Information Figure 1). To support the hypothesis that C265 serves as nucleophile (Figure 1A), we compared unmodified and Br-DMSP-modified Ehux-Alma1 by shotgun proteomics. Unfortunately, peptides that include the active-site cysteines could not be identified. Electrospray Ionization Mass Spectrometry was also applied to test the expected mass for Ehux-Alma1’s covalent adduct with Br-DMSP yet gave obscure results due to heterogeneity.
Mutating the second active site cysteine, C108, to alanine, reduced Ehux-Alma1’s activity by 70-fold, yet showed similar inhibition to the wild-type (Figure 2C), supporting the hypothesis that inhibition is not mediated by C108 but rather by C265, which is Alma1’s primary catalytic residue (Figure 1C). So far, Ehux-Alma1 is the only biochemically characterized enzyme of an entire enzyme family that includes paralogues and orthologues identified in a range of marine algae and corals. To expand this family’s characterization and validate that Br-DMSP is a generic mechanism-based inhibitor of Alma DMSP lyases, we purified and characterized an Alma-like gene from a coral C
DOI: 10.1021/acschembio.6b00844 ACS Chem. Biol. XXXX, XXX, XXX−XXX
Letters
ACS Chemical Biology symbiont, the dinoflagellate Symbiodinum-A1 (Sym). This putative DMSP lyase, Sym-Alma, exhibits 45% amino acid similarity to Ehux-Alma1 (with both active-site cysteines present) and was confirmed to have DMSP lyase activity in crude E. coli lysates upon heterologous expression.10 However, the originally reported His-Tagged Sym-Alma lost its activity during purification. We noted that, although Ehux-Alma1 showed no inhibition by EDTA, DMSP lyases of other algal species were reported to be metal-dependent (e.g., Polysiphonia paniculata20). Indeed, Sym-Alma could be purified while retaining activity using a Strep-Tag, thus avoiding purification on Ni-NTA (Supporting Information Figure 2). Accordingly, Sym-Alma was inhibited by EDTA (but not by 1,10phenantroline) and regained activity when supplemented with Ca2+ or Mn2+ ions, but not with Zn2+ or Mg2+ (Supporting Information Figure 2). The Ehux-Alma1 and Sym-Alma are highly diverged, and although their overall structure and function are conserved, Sym-Alma seems to have uniquely acquired a metal binding site that is probably crucial for maintaining its structural integrity. Sym-Alma shows high kcat (427 ± 44 s−1), while due to high KM for DMSP (15.8 ± 2.7 mM; Figure 1D), kcat/KM only assumes a moderate value of 2.7 × 104 M−1 S−1 (compared to 8 × 104 M−1S1− for EhuxAlma110). Finally, Br-DMSP was also found to be a potent inhibitor of Sym-Alma (Figure 1C). Thus, this newly characterized Alma member appears to exhibit Ehux-Alma1’s active-site chemistry. We then aimed to examine the potency of Br-DMSP in algal crude extracts and also in living algal cells. A complete inhibition of Ehux-Alma1 DMSP lyase activity was detected in crude lysates of E. huxleyi 373, which has the highest reported E. huxleyi lyase activity (∼10 fmol cell −1 min−1). Similarly, complete inhibition was observed in crude lysates of E. coli in which the Sym-Alma gene was expressed (Figure 3A). Natural isolates often contain high concentrations of DMSP reaching up to 300 mM intracellularly. To mimic this situation, we measured the rate of DMSP cleavage by a continuous colorimetric assay based on the pH-indicator Cresol purple (Methods). At 10 mM DMSP, upon addition of 8 μM BrDMSP, the lyase activity of Ehux-Alma1 came to a full stop within less than 3 min (Supporting Information Figure 3). We also tested inhibition in vivo. Upon adding Br-DMSP (100 μM) to E. huxleyi 373 cultures, ∼75% inhibition of the DMSP lyase activity was observed within 2.5 h. However, although EhuxAlma1 is expressed in exponentially growing E. huxleyi cells, it is largely inactive until cells are disrupted,10 probably due to its storage in a subcompartment that does not contain DMSP. Thus, we could not establish whether inhibition occurred within living cells during incubation, or during the centrifugation steps that were needed to remove the excess of inhibitor. Further experiments are therefore needed to examine if Br-DMSP is incorporated into living cells and is also inhibiting Ehux-Alma1 in intact cells while the enzyme is still inactive. Importantly, while Br-DMSP tightly inhibited both Alma enzymes tested, it had no effect on any of the known bacterial DMSP lyases (Figure 3). A relatively large repertoire of putative bacterial DMSP lyases has been reported: DddD is a bifunctional CoA-transferase/lyase;21 DddP belongs to the metallopeptidase M24;22 DddQ, DddL, DddW, and DddK belong to the highly diverse Cupin superfamily;13−15 and finally, DddY’s phylogeny remains unknown. All these enzymes show DMSP lyase activity, but some (e.g., DddQ, DddP)
appear to exhibit very low activity, possibly indicating a promiscuous rather than native DMSP lyase activity.23 Regardless, representative genes of all these families were overexpressed in E. coli (Supporting Information Methods). All enzymes except DddL and DddQ could be purified while maintaining DMSP lyase activity (typically, in the presence of their preferred metal, Supporting Methods). For unknown reasons, DddQ and DddL lost activity after purification, and therefore, crude lysates were used for testing inhibition by BrDMSP. In contrast to Alma-like enzymes, none of the bacterial DMSP lyases was inhibited by Br-DMSP, not even at 100 μM, i.e., at concentrations that are ∼100-fold higher than those at which Ehux- and Sym-Alma1 showed complete inhibition (Figure 3A compared to Figure 1B). Finally, as a proof of concept, we examined the origins of the DMSP lyase activity in two species of stony corals. Reef building corals rely on a tightly regulated symbiosis between the coral animal, endocellular microalgae, as well as bacterial components, collectively known as a holobiont.24 The holobiont partners maintain a complex network of interactions, including chemical signaling that involves DMS, DMSP, and other metabolite exchanges.24 Corals and their symbiont algae have Alma genes (e.g. Sym-Alma described above) and produce both DMSP and DMS in coastal areas.9 However, corals are also covered with mucus layers containing massive amount of bacteria (108 cell/mL) comprising the genus of Roseobacter and other bacterial clades that possess ddd genes.16,25 Elucidating the origin of DMS by metagenomic analysis per se is therefore impossible. However, the DMSP lyase observed in crude extracts prepared from samples of coral Acropora millepora colonies (40−140 nmol/hour/mg total protein) was ≥90% inhibited in the presence of ≥10 μM Br-DMSP (Figure 3B). It therefore appears that the Alma-like DMSP lyases and, therefore, the eukaryotic component of the Acropora holobiont have a key role in DMS production in corals. In contrast, no inhibition was obtained when 10 μM Br-DMSP was applied to a coral colony of Stylophora, suggesting the origins for the lower DMSP lyase activity ≥1.5 (nmol/h/mg total protein) of this coral relates to bacterial sources. This experiment therefore demonstrates that Br-DMSP can be used to determine the relative contribution of Alma enzymes for DMS production in marine natural habitats. Additionally, the physiological role of DMS in algae and corals remains largely unknown. In principle, Br-DMSP could also be applied in vivo to elucidate the cellular and ecological role of DMS release in algae and other marine organisms that possess Alma DMSP lyases. The vast majority of marine microorganisms (including bacteria, protists, and algae) are not culturable let alone amenable to genetic manipulations. However, key biogeochemical questions, such as the origins of DMS, may still be tackled by combining genomic data with traditional biochemical approaches such as biochemical profiling and mechanism-based inhibitors.
■
METHODS
DMSP Lyase Activity and Inhibition Assays. DMS release was measured as previously described.10 Tris buffer (100 or 200 mM, pH 8.0) with 100 mM NaCl, was used typically in the presence of 10 mM DMSP. Enzyme concentrations were typically as follows: Ehux-Alma1, 0.3 μg/mL; Sym-Alma, 1 μg/mL; DddY, 0.05 μg/mL; DddD, 2 μg/ mL; and DddP and DddW, 10 μg/mL. Crude lysates were used as is, or diluted up to 100-fold, depending on their level of activity. Reactions were incubated while shaking (800 rpm) for 5 min at 30 °C and terminated by 1000-fold dilution into a sealed glass vial containing 30 mL of water. DMS levels were determined using an Eclipse 4660 D
DOI: 10.1021/acschembio.6b00844 ACS Chem. Biol. XXXX, XXX, XXX−XXX
ACS Chemical Biology
■
Purge-and-Trap Sample Concentrator system equipped with an Autosampler (OI Analytical). Separation and detection were done using GC-FPD (HP 5890) equipped with a RT-XL sulfur column (Restek). All measurements were calibrated using DMS standards. DMSP lyase activity was also monitored colorimetrically, in 0.5 mM Bicine buffer containing 5 mM DMSP, 1 mM DTT, and 0.4 mg mL−1 m-cresol purple. The pH was adjusted to 8.1−8.3 using 1 M NaOH. Reactions were performed in 200 μL volume using 96 well plates, and absorption was followed at 577 nm. Purified enzymes and crude lysates used in this assay were applied at low buffer concentrations,