Identification and Stereospecificity of the First Three Enzymes of 3-Dimethylsulfoniopropionate Biosynthesis in a Chlorophyte Alga

Abstract

Many marine algae produce 3-dimethylsulfoniopropionate (DMSP), a potent osmoprotective compound whose degradation product dimethylsulfide plays a central role in the biogeochemical S cycle. Algae are known to synthesize DMSP via the four-step pathway, l-Met → 4-methylthio-2-oxobutyrate → 4-methylthio-2-hydroxybutyrate → 4-dimethylsulfonio-2-hydroxy-butyrate (DMSHB) → DMSP. Substrate-specific enzymes catalyzing the first three steps in this pathway were detected and partially characterized in cell-free extracts of the chlorophyte alga Enteromorpha intestinalis. The first is a 2-oxoglutarate-dependent aminotransferase, the second an NADPH-linked reductase, and the third an S-adenosylmethionine-dependent methyltransferase. Sensitive radiometric assays were developed for these enzymes, and used to show that their activities are high enough to account for the estimated in vivo flux from Met to DMSP. The activities of these enzymes in other DMSP-rich chlorophyte algae were at least as high as those in E. intestinalis, but were ≥20-fold lower in algae without DMSP. The reductase and methyltransferase were specific for the d-enantiomer of 4-methylthio-2-hydroxybutyrate in vitro, and both the methyltransferase step and the step(s) converting DMSHB to DMSP were shown to prefer d-enantiomers in vivo. The intermediate DMSHB was shown to act as an osmoprotectant, which indicates that the first three steps of the DMSP synthesis pathway may be sufficient to confer osmotolerance.

The tertiary sulfonium compound DMSP is synthesized and accumulated by many marine macroalgae and phytoplankton species (Blunden and Gordon, 1986; Keller et al., 1989) and by certain salt-tolerant flowering plants (Hanson and Gage, 1996). DMSP is environmentally significant because it is biodegraded to DMS, an atmospheric gas with major roles in the global S cycle, in cloud formation, and possibly in climate regulation (Charlson et al., 1987; Malin, 1996). The DMSP produced by marine algae is the main biogenic precursor of oceanic DMS, which contributes about 1.5 × 10¹³ g of S to the atmosphere annually (Groene, 1995; Malin, 1996).

Like betaines, of which it is a S analog, DMSP acts as a cytoplasmic compatible solute or osmoprotectant and so has a key physiological function in adaptation to osmotic stress (Kirst, 1990; Hanson and Gage, 1996). It is also an effective cryoprotectant and contributes to the acclimation of polar algae to freezing temperatures (Karsten et al., 1996). Because DMSP has protectant properties comparable to those of betaines and does not contain N, the DMSP biosynthetic pathway is a rational target for metabolic engineering of stress resistance in N-poor, S-rich environments (Hanson and Burnet, 1994; Le Rudulier et al., 1996).

The prospect of engineering this pathway led us to investigate the steps involved in DMSP synthesis from Met in the higher plant Wollastonia biflora and in marine algae. The higher plant pathway proceeds via the intermediates S-methylmethionine and dimethylsulfoniopropionaldehyde (Hanson et al., 1994; James et al., 1995). The route in the marine macroalga Enteromorpha intestinalis and in three phytoplankton species is completely different (Gage et al., 1997) (Fig. 1). The first step in this pathway is loss of the amino group, giving the 2-oxo acid MTOB; ¹⁵N-labeling evidence strongly implies that the amino group is removed via transamination rather than deamination (Gage et al., 1997). The subsequent steps are reduction to MTHB, S-methylation to yield DMSHB, and oxidative decarboxylation to yield DMSP.

Figure 1.

The biosynthesis of DMSP from Met in marine algae. In vivo isotope tracer data indicate that the first two steps are reversible (Gage et al., 1997).

We report here the detection and partial characterization of enzymes catalyzing the first three steps of the DMSP-synthesis pathway in the chlorophyte alga E. intestinalis, together with radiometric methods for their routine assay. We demonstrate that these enzymes are specific to the pathway, and that the d-enantiomers of MTHB and DMSHB are preferred. We also show that the intermediate DMSHB has osmoprotectant properties, as expected from its structure.

MATERIALS AND METHODS

The following algal species were collected from the intertidal zone or from water ≤2 m deep at the sites and times specified: Enteromorpha intestinalis (L.) Link from Marineland in Florida year-round; Caulerpa ashmeadii Harvey and Udotea conglutinata (Ellis et Solander) Lamouroux from Crystal River, FL, July, 1996; Ulva fasciata Delile, Ulva reticulata Forsskal, Enteromorpha flexuosa (Wulfen) J. Agardh, and Halimeda discoidea Decaisne from Oahu, HI, January, 1996. E. intestinalis for in vivo radiolabeling was used after culturing for up to 1 week, as described previously (Gage et al., 1997). For enzyme extraction, algae were frozen in liquid N₂ or dry ice within a few hours of collection and stored at −80°C; this was shown not to affect enzyme activity. DMSP levels were estimated by a GC method (Paquet et al., 1994).

Chemicals

[³⁵S]Met (43.5 TBq mmol⁻¹) and [methyl-¹⁴C]AdoMet (2.15 GBq mmol⁻¹) were purchased from NEN-DuPont and were mixed with unlabeled compounds to give the desired specific radioactivities. AdoMet was obtained from Boehringer Mannheim and was used without further purification. dl-MTHB (Ca²⁺ salt) was obtained from Fluka or Sigma; the Sigma product was recrystallized from aqueous ethanol. DMSP was purchased from Research Plus, Inc. (Bayonne, NJ). MTP was purchased from TCI (Tokyo, Japan) and neutralized with KOH. Ion-exchange resins were purchased from Bio-Rad.

Synthesis of [³⁵S]MTHB, [³⁵S]MTP, and [³⁵S]MTOB

d- and l-[³⁵S]MTHB were prepared by incubating (24 h, 37°C) 15 MBq of [³⁵S]Met [37 kBq nmol⁻¹, treated with AG-1 (formate) to remove anionic impurities] in a 750-μL reaction mixture containing 25 μmol K₂PO₄, pH 7.4, 3 μmol GSH, 20,000 units of catalase, 2 units of l-amino acid oxidase (Sigma A-9378), 1 μmol of NADH, and 40 units of either Staphylococcus epidermidis d-LDH (Sigma L-9636) or rabbit muscle l-LDH (Sigma L-2500). [³⁵S]MTP was a by-product of these reactions. For the synthesis of [³⁵S]MTOB, the NADH and LDH were omitted and the incubation time was cut to 10 h. Products were isolated by acidifying reaction mixtures with 0.1 volume of 12 n HCl and extracting with 3 × 3 mL of ether. The combined ether extracts were back-extracted into 200 μL of 10 mm NaOH, which was concentrated in vacuo and fractionated by TLE as described below. Products were located by autoradiography, eluted with 0.5 mm β-mercaptoethanol, and stored at −80°C; their radiochemical purity was ≥95%, as determined by TLE and TLC. The optical purity of [³⁵S]MTHB enantiomers was ≥95%, as determined by susceptibility to oxidation by d- and l-LDH, as described below.

Synthesis of [³⁵S]DMSHB

d- and l-[³⁵S]DMSHB were prepared by treating 0.93 MBq of d- or l-[³⁵S]MTHB (37 kBq nmol⁻¹) with 50 μmol methanol in 0.4 mL of 6 n HCl for 4 h at 110°C (Lavine et al., 1954). [³⁵S]DMSHB was isolated by ion exchange (James et al., 1995) and TLE as described below; radiochemical purity was ≥99% as determined by TLE and TLC.

Synthesis of MTHB and DMSHB Enantiomers

Unlabeled d- and l-MTHB were synthesized from d- and l-Met by reaction with HNO₂ (Kleemann et al., 1979). Met (5 mmol) was dissolved in 4.25 mL of 0.9 m H₂SO₄ and cooled to 0°C; 1 mL of ice-cold 6.2 m NaNO₂ was added dropwise, and the reaction mixture was then incubated for 2 h at 22°C. The MTHB product was extracted into 4 × 3 mL of ether, dried in a stream of N₂, and dissolved in 1 mL of water. MTHB was then purified by passage onto a 1.25-mL AG-1 (OH⁻) column, from which it was eluted with 6 mL of 2.5 n HCl, extracted again with ether, and lyophilized. Purity was about 98%, as determined by TLC and TLE. d- and l-DMSHB were prepared from 0.12 mmol of the corresponding form of MTHB by heating at 110°C for 2 h with 0.25 mmol of methanol in 0.5 mL of 6 n HCl. After removing the HCl in vacuo, DMSHB was isolated by ion exchange (James et al., 1995). The product was lyophilized and freed of a small amount (10%) of putative dimer by hydrolysis with 0.1 n HCl at 100°C for 2 h. The optical purity of the d and l forms of MTHB and DMSHB was estimated as ≥92% by circular dichroism measurements. dl-DMSHB was synthesized from dl-MTHB by the method of Toennies and Kolb (1945).

TLE and TLC

TLE separations were on glass-backed 0.1-mm cellulose plates (Merck, Darmstadt, Germany) at 1.8 kV for 20 min at 4°C. The buffers were pyridine:glacial acetic acid:water (1:1:38, v/v/v) for thioethers, and 1.5 n formic acid for sulfonium compounds. TLC of thioethers and amino acids was carried out on cellulose plates developed with n-butanol:glacial acetic acid:water (60:20:20, v/v/v); sulfonium compounds were separated on plastic-backed 0.25-mm silica gel G plates (Machery-Nagel, Düren, Germany) developed with methanol:acetone:concentrated HCl (90:10:4, v/v/v). Compounds were visualized using the spray reagents described previously (James et al., 1995).

Enzyme Extraction

Tissue was pulverized in liquid N₂ and extracted with 3.5 volumes of buffer A (50 mm Bis-Tris [bis(2-hydroxyethyl)iminotris(hydroxymethyl)methane]-HCl, pH 8.0, 5 mm DTT, 2 mm K₂EDTA, and 1 mg mL⁻¹ BSA); for MTHB S-methyltransferase in E. intestinalis, the pH was 7.0 and BSA was omitted. Subsequent steps were performed at 4°C. The brei was centrifuged at 10,000g for 10 min, and the supernatant was desalted on Sephadex G-25 equilibrated in buffer A. For Met aminotransferase and MTOB reductase assays the supernatant was then concentrated 10-fold with a Centricon-30 (Amicon). For all three DMSP synthesis enzymes, the desalted supernatants contained ≥70% of the activity present in the brei. Centrifugation at 100,000g for 1 h did not pellet activity of any of the enzymes. Desalted supernatants were in some cases flash-frozen in liquid N₂ and stored at −80°C; this did not affect enzyme activity. Enzyme assays were carried out under conditions in which product formation was linear with respect to time and enzyme concentration.

Enzyme Assays

MDH, Asp Aminotransferase, and Ala Aminotransferase

MDH assays contained 0.1 m Tris-acetate, pH 8.0, 0.2 mm NADH, and 2.5 mm oxaloacetate; oxaloacetate-dependent oxidation of NADH was followed by the fall in A₃₄₀. Asp aminotransferase assays contained 0.1 m potassium phosphate buffer, pH 7.5, 25 mm l-Asp, 5 mm 2-oxoglutarate, 0.2 mm NADH, and 60 units mL⁻¹ of porcine heart MDH (Sigma M-2634); 2-oxoglutarate-dependent NADH oxidation was monitored. Ala aminotransferase was assayed in the same way except that 100 mm l-Ala replaced Asp, and 120 units mL⁻¹ rabbit muscle LDH (Sigma L-2500) was used in the coupling reaction.

Met Aminotransferase and Met Oxidase

Standard aminotransferase assays (60 μL final volume) contained 0.1 m ammediol-HCl, pH 9.1, 100 μm l-Met (7–22 kBq, treated with AG-1 [formate]) and 1 mm 2-oxoglutarate. Met oxidase assays were the same except that 2-oxo acid was excluded. After 1 h at 25°C, reactions were stopped on ice and acidified with 10 μL of 2.5 n HCl after adding carrier Met and MTOB (0.2 μmol). The [³⁵S]MTOB formed was extracted into 0.8 mL of ether, then back-extracted into 100 μL of 10 mm NaOH, of which 80 μL was taken for scintillation counting. Aminotransferase activity with other 2-oxo acids was assayed using a [³⁵S]Met concentration of 25 μm. Aminotransferase and oxidase data were corrected for MTOB recovery; aminotransferase data were corrected for Met oxidase activity. TLC and TLE confirmed that the product of aminotransferase and oxidase reactions was [³⁵S]MTOB. Aminotransferase activity was assayed in the Met-synthesis direction in 30-μL reaction mixtures containing 0.1 m ammediol-HCl, pH 9.1, 3.1 μm [³⁵S]MTOB (0.63 kBq), and 10 mm amino acid. Incubation and carriers were as above. The [³⁵S]Met product was isolated by passage onto 1-mL AG-50 (H⁺) columns, washing with 25 mL of water, and eluting with 5 mL of 2.5 n HCl. The product was shown to be [³⁵S]Met by TLC. Data were corrected for Met recovery and for ³⁵S found in the AG-50 eluate of blank assays lacking amino acid.

MTOB Reductase

Standard assays (final volume 30 μL) contained 0.1 m ammediol-HCl, pH 8.0, 150 μm NADPH, and 30 μm [³⁵S]MTOB (7.4 kBq). After incubation for 1 h at 25°C, 50 nmol MTOB and 100 nmol MTHB carriers were added, followed at once by 50 μL of 10 mm 2,4-dinitrophenylhydrazine in 2.5 n HCl. The samples were then incubated for 1 h at 22°C and extracted with 0.5 mL of ether. The ether phase was back-extracted with 50 μL of 10 mm NaOH containing 3 μL of 1 m acetic acid, which was concentrated in vacuo and subjected to TLC. [³⁵S]MTHB zones were located with iodoplatinate reagent and quantified by scintillation counting. Data were corrected for MTHB recovery and for ³⁵S in the MTHB zone in assays without NADPH. Assays of MTHB oxidation (final volume 50 μL) contained 0.1 m ammediol-HCl, pH 8.0, 30 μm d-[³⁵S]MTHB (46 kBq), and 1 mm NADP. Incubation was for 2 h at 25°C. Oxidation products were measured as described below for determination of [³⁵S]MTHB configuration.

d-MTHB S-Methyltransferase

Standard assays (final volume 50 μL) contained 50 mm Bis-Tris-HCl, pH 7.0, 1 mm AdoMet, and 25 μm d-[³⁵S]MTHB (1.9–3.7 kBq). After incubation at 22°C for 1 h, 1 mL of 0.2 mm DMSHB carrier was added and the mixture was immediately applied to a 1-mL mixed-resin column (AG-1 [OH⁻]:BioRex 70 [H⁺], 2:1, v/v, firmly packed). The [³⁵S]DMSHB product was eluted with 5 mL of water and quantified by scintillation counting. In some cases the enzyme was assayed using unlabeled MTHB and 100 μm [methyl-¹⁴C]AdoMet (3.7 kBq). These reactions were stopped after 1 h by adding 25 μL of 10% (w/v) TCA, 2 μL of 100 mm DMSHB, and 215 μL of an activated charcoal suspension (38 mg mL⁻¹) in 0.1 n acetic acid to bind AdoMet (Cook and Wagner, 1984). After centrifuging for 5 min at 14,000g, [¹⁴C]DMSHB in the supernatant was quantified by scintillation counting. Values were corrected for DMSHB recovery and for blanks lacking the unlabeled substrate. The identities of the ³⁵S- and ¹⁴C-labeled DMSHB reaction products were confirmed by TLE and TLC.

Configuration of MTHB

[³⁵S]MTHB was prepared from [³⁵S]MTOB by scaling up the standard MTOB reductase assay, and then purified by TLE. Samples (2.6 kBq, 110 pmol) were incubated for 20 h in darkness at 30°C in 30-μL reaction mixtures containing 50 mm ammediol-HCl, pH 9.0, 5 mm NAD⁺, either 6 units of S. epidermidis d-LDH (Sigma L-9636) or 3 units of rabbit muscle l-LDH (Sigma L-1254), and 0.3 unit of Photobacterium fischeri NAD(P)H:FMN oxidoreductase. Controls of authentic d- and l-[³⁵S]MTHB were treated the same way. The reactions were stopped with 70 μL of 0.54 n HCl, mixed with carrier MTHB, MTOB, and MTP (100 nmol each), and extracted with 3 × 0.8 mL of ether. The combined ether fraction was back-extracted into 70 μL of 10 mm NaOH, which was concentrated in vacuo and subjected to TLE. [³⁵S]MTHB, [³⁵S]MTOB, and [³⁵S]MTP (a breakdown product of [³⁵S]MTOB) were located by autoradiography and by I₂ staining, and quantified by scintillation counting. Control reactions without LDH showed no [³⁵S]MTHB oxidation.

In Vivo Radiotracer Experiments

Samples (100 mg fresh weight) of tissue cut from the basal 1- to 2-cm region of E. intestinalis fronds were incubated in 0.5 mL of 0.2-μm filtered seawater containing 37 or 74 kBq (1 or 2 nmol) of [³⁵S]MTHB or [³⁵S]DMSHB. For experiments with [³⁵S]MTHB, the pH was lowered to an initial value of 5.7 by adding Mes-KOH (final concentration 100 mm) to the seawater. Incubation was at 22°C in the light, as described previously (Gage et al., 1997); label uptake was monitored by sampling the medium. After incubation, the tissue was rinsed for 5 min in seawater, blotted dry, and extracted using a methanol-chloroform-water procedure (Hanson et al., 1994; James et al., 1995). Water-soluble metabolites were fractionated by ion-exchange chromatography and analyzed by TLC and TLE. Incorporation of ³⁵S into the insoluble fraction was estimated by scintillation counting after suspending samples in Ready Gel (Beckman) containing 50% (v/v) water; the counting efficiency of this system was determined to be 65%.

Bacterial Osmoprotection Experiments

The Escherichia coli strains used were K-10 and FF4169, an otsA (trehalose-deficient) mutant of K-12 (Giæver et al., 1988). Experiments with K-10 were as described by Hanson et al. (1991), except that the medium was that of Neidhardt et al. (1974) and cultures were inoculated with cells growing exponentially in the presence of NaCl. FF4169 was grown at 37°C to stationary phase in M63 medium (Miller, 1972), pH 7.0, containing 0.2% (w/v) Glc. This culture was used to inoculate experimental M63 media (25 mL) containing NaCl and various supplements (final concentration 1 mm). Supplement solutions were adjusted to pH 7.0 and filter-sterilized before addition; care was taken to avoid alkaline pH while neutralizing the DMSP solution. Experimental cultures were incubated at 37°C, and growth was monitored by the change in A₆₀₀.

RESULTS

Estimation of the Rate of DMSP Synthesis in Vivo

To provide a benchmark for the activities of DMSP pathway enzymes in E. intestinalis, we first estimated the in vivo flux of Met to DMSP by computer modeling of published [³⁵S]Met-tracer kinetic data for this alga (Gage et al., 1997). The computer model used was that described by Mayer et al. (1990). The starting free pool size of free Met was taken to be up to 10 nmol g⁻¹ fresh weight, based on published values for the chlorophyte alga Chlorella sorokiniana (Giovanelli et al., 1980). The flux values from Met to DMSP that gave satisfactory fits to the published labeling patterns of MTHB, DMSHB, and DMSP ranged from 1.2 to 4.2 nmol h⁻¹ g⁻¹ fresh weight.

Met Aminotransferase Activity

Met aminotransferase activity was readily detected in assays containing MTOB or 2-oxoglutarate as the amino acceptor for [³⁵S]Met, but the activities found with other 2-oxo acids were far lower (Table I). 2-Oxoglutarate was therefore used to further characterize the activity. Activity was not inhibited by a 5-fold excess of unlabeled d-Met, indicating that it was not due to the tandem action of a racemase and a d-Met aminotransferase. The pH profile showed an optimum at 9.1, and 40 to 60% of optimal activity in the physiological range of 7.5 to 8.0. The affinity for Met was high, as shown by velocity versus Met concentration curves (Fig. 2A). Double-reciprocal plots indicated that half-maximal velocity was reached at 30 μm Met, although they also suggested the presence of a minor activity with much lower affinity (not shown). The 2-oxoglutarate concentration giving half-maximal velocity was 400 μm at 100 μm [Met].

Table I.

Substrate preference of Met aminotransferase activity

Substrate	Relative Activity
	%
2-Oxo acids
MTOB	100
2-Oxoglutarate	81.0
Phenylpyruvate	14.1
Pyruvate	9.7
Oxaloacetate	9.1
Glyoxylate	2.8
Hydroxypyruvate	<1.0
l-Amino acids
Met	100
Glu	94.9
Phe	6.0
Ala	5.3
Asp	1.3
Gly	<0.2
Ser	<0.2
Gln	8.5
Asn	1.2

2-Oxo acid preference was determined in reactions containing desalted extract equivalent to 5.5 mg fresh weight, 25 μm l-[³⁵S]Met, and an optimal concentration of 2-oxo acid (0.1 mm for MTOB and 1 mm for all others). l-Amino acid preference was determined in reactions containing extract equivalent to 0.6 to 5.5 mg fresh weight, 3.1 μm [³⁵S]MTOB, and 10 mm amino acid. Data for 2-oxo acids and amino acids were obtained at pH 9.1 and are expressed relative to the activities obtained with MTOB (21.7 pkat g⁻¹ fresh weight) and Met (13.0 pkat g⁻¹ fresh weight), respectively.

Figure 2.

Plots of velocity versus substrate concentration for enzyme activities implicated in DMSP biosynthesis. Activities were measured using desalted E. intestinalis extract equivalent to 5 to 6.5 mg fresh weight per assay. A, Met aminotransferase, assayed in the presence of 1 mm 2-oxoglutarate. Met oxidase was assayed in the absence of 2-oxo acid. B, MTOB reductase, assayed using 200 μm NADPH. C, d-MTHB S-methyltransferase, assayed using 1 mm AdoMet. Experiments with each activity were done at least twice, with similar results to those shown. The l-Met and d-MTHB concentrations giving half-maximal velocity given in the text were estimated from double-reciprocal plots.

Consistent with the strong preference for 2-oxoglutarate as the amino acceptor, assays in the reverse (Met synthesis) direction using [³⁵S]MTOB showed that l-Glu was by far the best amino donor after Met itself (Table I). All other amino acids tested were ≥10-fold less effective, including Gln and Asn. Met aminotransferase activity was not increased by including pyridoxal-5′-phosphate (0.1 mm) in extraction and assay buffers. This is not unusual, since the pyridoxal-5′-phosphate of plant aminotransferases is typically tightly bound and few are activated by its addition (Ireland and Joy, 1985).

Met Oxidase Activity

E. intestinalis extracts catalyzed the slow conversion of [³⁵S]Met to [³⁵S]MTOB in the absence of 2-oxo acid (Fig. 2A). This activity was ascribed to a nonspecific l-amino acid oxidase because it was decreased by lowering the O₂ concentration and increased by raising it, and because it was strongly suppressed when unlabeled amino acids that are good substrates for other l-amino acid oxidases (Meister, 1965) were present (Table II). Figure 2A shows that oxidase activity was only 8% of aminotransferase activity even at a Met concentration of 200 μm. Since cytoplasmic levels of Met in algae and other plants are most probably ≤200 μm (Giovanelli et al., 1980), the oxidase seemed unlikely to mediate much MTOB synthesis in vivo and was not investigated further.

Table II.

Characteristics of 2-oxo acid-independent Met oxidation

Treatment	Met Oxidation
	pkat g−1 fresh wt
Control (no additions)	1.69  (100)
N2-Purged	0.57  (34)
O2-Purged	2.98  (176)
+5 mm l-Phe	0.15  (9)
+5 mm l-Leu	0.17  (10)

Desalted extract (equivalent to 17 mg fresh weight) was incubated with 100 μm l-[³⁵S]Met for 1 h in 60-μL reaction mixtures in 1-mL tubes. For O₂ and N₂ treatments, the tubes were closed with serum caps and purged for 2 min before injecting l-[³⁵S]Met to start the reaction. For other treatments the tubes contained air. Values in parentheses are relative to the control (=100).

MTOB Reductase Activity

E. intestinalis extracts showed NADPH- and NADH-dependent MTOB reductase activities of similar magnitude at saturating NAD(P)H levels, but half-maximal rates were attained at 30 μm NADPH versus 650 μm NADH (Fig. 3). The two activities were not additive, consistent with their being due to the same enzyme(s) (Fig. 3B). Because NAD(P)H levels in plant cytoplasm are typically not more than 50 to 150 μm (Heber and Santarius, 1965; Hampp et al., 1984), the NADPH-linked activity appeared likely to be the dominant one in vivo. This activity was therefore characterized further. It was highest at pH 8.0, and about 70% as high at pH 7.0. Velocity versus MTOB concentration plots showed clearly that affinity for MTOB was high (Fig. 2B). Since double-reciprocal plots were nonlinear (showing apparent negative cooperativity), the MTOB concentration giving half-maximal velocity could not be determined precisely, but appeared to be approximately 40 μm. LDH and acetohydroxy acid isomeroreductase can both catalyze the reduction of various 2-oxo acids, and LDH is specifically known to attack MTHB (Meister, 1957; Umbarger, 1996). However, MTOB reductase activity was not attributable to either of these enzymes since [³⁵S]MTOB reduction was scarcely inhibited (≤30%) by a 200-fold excess of unlabeled pyruvate or 2-oxoisovalerate (data not shown).

Figure 3.

NADPH-dependent (A) and NADH-dependent (B) MTOB reductase activity. Reaction mixtures contained desalted E. intestinalis extract equivalent to 8.6 mg fresh weight, 30 μm [³⁵S]MTOB, and various concentrations of NADPH or NADH. Data points are means ± se (n = 3). Where error bars are not shown they were smaller than the symbols. The open symbol in B shows the activity given by 1 mm NADH plus 100 μm NADPH.

The configuration of the [³⁵S]MTHB produced by the E. intestinalis MTOB reductase was determined by testing it as a substrate for purified d- and l-LDH; authentic d- and l-[³⁵S]MTHB were included as controls (Table III). The controls behaved as predicted, and the reductase product was attacked only by d-LDH. The in vitro product of E. intestinalis MTOB reductase is therefore d-MTHB; it is also very probable that the d-enantiomer predominates in vivo, as shown below. Knowing that reductase produces d-MTHB, it was of interest to seek the reverse reaction, i.e. NADP-dependent d-[³⁵S]MTHB oxidation. This was readily detectable; in the presence of 35 μm d-[³⁵S]MTHB and 1 mm NADP at pH 8.0, the reaction rate was 0.4% of the forward rate in comparable conditions. Adding NAD(P)H:FMN oxidoreductase plus FMN (to remove NADPH) did not accelerate the reverse reaction.

Table III.

Evidence that the product of MTOB reductase is the d-enantiomer of MTHB

Substrate	Specificity of LDH Used	[35S]MTHB Oxidation
		%
Reductase product	d	15.5
	l	<0.5
d-[35S]MTHB	d	16.0
	l	<0.5
l-[35S]MTHB	d	<0.5
	l	14.0

[³⁵S]MTHB synthesized from [³⁵S]MTOB using desalted E. intestinalis extract was incubated with S. epidermidis d-LDH or rabbit muscle l-LDH. As controls, comparable amounts of authentic d- and l-[³⁵S]MTHB were treated in the same way. Reaction mixtures were fractionated by TLE, and radioactivity in MTHB, MTOB, and MTP was quantified. MTP is formed from MTOB by spontaneous decarboxylation (Gage et al., 1997). [³⁵S]MTHB oxidation was therefore expressed as the percentage of ³⁵S recovered in MTOB and MTP. Corrections were applied for traces (≤1%) of these compounds present in the [³⁵S]MTHB substrates.

MTHB S-Methyltransferase Activity

Consistent with the product of MTOB reductase being the d form of MTHB, E. intestinalis extracts catalyzed AdoMet-dependent S-methylation of d-MTHB but not l-MTHB. This was demonstrated in two types of assays based on enzymatically synthesized d- and l-[³⁵S]MTHB or on chemically synthesized, unlabeled d- and l-MTHB as substrates (Table IV). There was no detectable activity with MTP as substrate (Table IV), which is in accord with the in vivo radiotracer evidence against a role for this compound in DMSP synthesis in E. intestinalis (Gage et al., 1997). Nor did the enzyme attack other naturally occurring thioethers (3-methylthiopropylamine, d- or l-Met, l-S-methylcysteine), as judged from their complete failure to inhibit d-[³⁵S]MTHB methylation when present in unlabeled form in 40-fold excess (data not shown). The d-MTHB methyltransferase activity had a broad pH optimum in the region of 6.5 to 8.0. Plots of velocity versus [ d-MTHB] showed that activity was half-maximal at 8 μm d-MTHB (Fig. 2C); the AdoMet concentration giving half-maximal velocity was 30 μm at 25 μm [d-MTHB].

Table IV.

Enantiomer and substrate specificity of S-methyltransferase activity

Substrate	Enantiomer	Methyltransferase Activity
		pkat g−1 fresh wt
[35S]MTHB	d	2.42
	l	0.03
Unlabeled MTHB	d	2.44
	l	0.11
[35S]MTP	–	<0.02

Desalted extract equivalent to 3.8 mg fresh weight was incubated with 25 μm [³⁵S]MTHB or [³⁵S]MTP and 1 mm AdoMet, or with 25 μm unlabeled MTHB and 0.1 mm [methyl-¹⁴C]Ado-Met. Incorporation of label into methylated product (DMSHB or DMSP) was measured. The l-[³⁵S]MTHB and l-MTHB could have contained small amounts of the d forms (see Methods), which may account for the slight activity observed with the l enantiomers.

Comparative Biochemistry of DMSP-Accumulating and Nonaccumulating Algae

Evidence that the Met aminotransferase, MTOB reductase, and d-MTHB methyltransferase activities found in E. intestinalis are specific to DMSP synthesis was sought by assaying these enzymes in extracts of six other marine chlorophyte algae, three with DMSP and three without (Table V). The activities of three housekeeping enzymes (Asp and Ala aminotransferases and MDH) were also measured as a check on the quality of the extracts. All six algae had housekeeping enzyme activities comparable to those of E. intestinalis (Table V). Those that accumulated DMSP all showed Met aminotransferase, MTOB reductase, and d-MTHB methyltransferase activities at least as high as those of E. intestinalis. In contrast, the algae lacking DMSP had about 30-fold less Met aminotransferase activity than E. intestinalis, ≥20-fold less MTOB reductase, and no detectable methyltransferase. All six algae had very low levels of Met oxidase (0.5–2.3 pkat g⁻¹ fresh weight; not shown), providing further evidence that this activity is not importantly related to DMSP synthesis.

Table V.

Activities of putative enzymes of DMSP synthesis in various marine chlorophyte algae

Species	DMSP Content	Housekeeping Enzyme Activity			DMSP Pathway Enzyme Activity
		MDH	Asp-AT	Ala-AT	Met-AT	MTOB-R	MTHB-MT
	μmol g−1 fresh wt	nkat g−1 fresh wt			pkat g−1 fresh wt
E. intestinalis	19.4	30	4.5	3.33	30.3	3.71	3.44
E. flexuosa	17.5	155	18.0	4.55	87.2	4.08	13.0
U. reticulata	18.5	138	12.4	1.25	29.7	4.44	7.39
U. fasciata	12.5	185	7.9	2.98	35.3	5.22	7.94
H. discoidea	<0.03	71	22.5	0.63	0.9	<0.10	<0.02
C. ashmeadii	<0.03	116	19.7	2.30	1.3	0.20	<0.02
U. conglutinata	<0.03	22	15.0	1.08	1.0	0.14	<0.02

Activities were measured in E. intestinalis, in three other algal species that accumulate DMSP, and in three that do not. Desalted extracts were prepared as described under Methods for E. intestinalis. Asp and Ala aminotransferases (Asp-AT and Ala-AT) and MDH were assayed spectrophotometrically, and Met aminotransferase (Met-AT), MTOB reductase (MTOB-R), and d-MTHB methyltransferase (MTHB-MT) were assayed radiochemically, as described in Methods.

Confirmation of d-Enantiomer Preferences in Vivo

E. intestinalis fronds converted d-[³⁵S]MTHB to DMSP very efficiently (Table VI). As reported by Gage et al. (1997), the l form was also converted to DMSP, but 5-fold less efficiently than the d form. These data confirm the preference of DMSP synthesis for d-MTHB in vivo. Much of the l-MTHB was metabolized to protein-bound Met, and more label from the l form remained in the intermediate DMSHB (Table VI), presumably because the ³⁵S flux was small in relation to the pool size of DMSHB. The conversion of l-MTHB to Met indicates that it can be oxidized to MTOB and then transaminated to yield Met, as occurs in other organisms (Livesey, 1984; Miyazaki and Yang, 1987). MTOB formed in this way could also be converted to d-MTHB, which would explain the small ³⁵S flux from l-MTHB to DMSHB and DMSP.

Table VI.

Metabolism of enantiomers of MTHB and DMSHB by E. intestinalis

35S-Precursor	Precursor Uptake	35S Distribution
		Insoluble	DMSHB	DMSP
	kBq	% of uptake
d-MTHB	9.1	5.1	<0.1	77.2
l-MTHB	8.0	16.4	27.4	16.1
d-DMSHB	29.3	6.9	23.7	69.5
l-DMSHB	34.8	3.6	71.7	8.1

Algal fronds (0.1 g fresh weight) were incubated with 74 kBq (2 nmol) of [³⁵S]MTHB or 37 kBq (1 nmol) of [³⁵S]DMSHB in 0.5 mL of seawater for 6 h in the light. For [³⁵S]MTHB, the initial pH of the seawater was adjusted to 5.7 with Mes-KOH. Precursor uptake was estimated from disappearance from the medium. Radiolabel incorporated into the insoluble fraction was shown to be in protein-bound Met by protease digestion and acid hydrolysis (Gage et al., 1997). The data are representative of three separate experiments.

Since the MTHB methyltransferase produces d-DMSHB, the enzyme(s) converting DMSHB to DMSP might be expected to prefer d-DMSHB. Supplying d- and l-[³⁵S]DMSHB confirmed this expectation; the d form was converted to DMSP 9-fold more efficiently than the l form (Table VI). The modest conversion of the l form to DMSP is in accord with our previous data (Gage et al., 1997).

Osmoprotection of E. coli by DMSHB

E. intestinalis does not accumulate high levels of the intermediate DMSHB (Gage et al., 1997). However, from its structure this compound is predicted to be an effective osmoprotectant, whereas its precursor, MTHB, is not (Yancey, 1994). This prediction was tested using standard osmoprotection bioassays with two E. coli strains; DMSHB and MTHB were compared with DMSP and Gly betaine as benchmarks. Figure 4 shows data for the osmosensitive strain FF4169; results with strain K-10 were similar except for a higher growth rate in the absence of osmoprotectants. Although less effective than Gly betaine and DMSP, dl-DMSHB was much more potent than dl-MTHB, which protected the cells slightly or not at all. Tested separately, the d and l forms of DMSHB and MTHB behaved like the racemic mixtures. The osmoprotective effect of DMSHB involved its intracellular accumulation, but not any metabolism. Thus, when 1 mm d- or l-[³⁵S]DMSHB was included in media containing 0.5 m NaCl, all of the label taken up by K-10 cells was shown by TLE to be in DMSHB; the intracellular concentration of [³⁵S]DMSHB was estimated to be 0.5 to 0.7 m by assuming cell volume to be 0.4 fL (Hanson et al., 1991).

Figure 4.

Osmoprotection by DMSHB. The osmosensitive (trehalose-deficient) E. coli strain FF4169 was grown aerobically at 37°C in M63-Glc medium containing 0.45 m (A) or 0.65 m (B) NaCl alone (control) and with 1 mm dl-DMSHB, dl-MTHB, DMSP, or Gly betaine (Gly Bet). Growth was monitored by A₆₀₀.

DISCUSSION

In the marine alga E. intestinalis, the first three steps of the DMSP biosynthesis pathway are Met → MTOB → MTHB → DMSHB (Fig. 1). We report here enzyme activities that catalyze these steps, and show that the d-enantiomers of the chiral intermediates MTHB and DMSHB are very strongly preferred. As summarized in Table VII, all three enzymes show high affinities for their substrates and have extractable activities 4- to 25-fold above our highest estimate of the in vivo rate of DMSP synthesis (4.2 nmol h⁻¹ g⁻¹ fresh weight). Moreover, much of the activity was probably not extracted because our extracts contained about 0.5 mg protein g⁻¹ fresh weight, whereas the total protein content (Kjeldahl n × 6.25) of E. intestinalis fronds was found to be 5.9 mg g⁻¹ fresh weight, in accord with published values (2–10 mg g⁻¹ fresh weight; Edwards et al., 1988).

Table VII.

Estimated activities and affinities of DMSP-synthesis enzymes

Enzyme	Enzyme Characteristics
	Vmax	S0.5
	nmol h−1 g−1	μm
Met aminotransferase	103  ± 21	30
MTOB reductase	30  ± 4	40a
d-MTHB methyltransferase	19  ± 2	8

Approximate maximal activity (V_max) values are means ± se of three to five experiments with different batches of E. intestinalis, and are given in units of nmol h⁻¹ g⁻¹ fresh weight for comparison with in vivo flux estimates. Each enzyme was assayed at its pH optimum. The substrate concentrations giving half-maximal velocity (S_0.5) were calculated from the data shown in Figure 2 and may be taken as rough estimates of apparent K_m values.

^aApproximate value only (see text). 

The conversion of MTOB to Met via transamination is the last step in the Met salvage pathway whereby the methylthio moiety of 5′-methylthioadenosine is recycled to Met (Cooper, 1996). Since this pathway is ubiquitous, all algae would be expected to show some Met aminotransferase activity, and indeed this was the case (Table V). However, the activity was 30- to 100-fold higher in the DMSP-accumulating algae, indicating that DMSP synthesis involves either overexpression of a housekeeping Met aminotransferase or expression of a novel enzyme. A novel enzyme appears more likely for two reasons.

First, the strong preference for Glu as an amino donor distinguishes the E. intestinalis activity from the amino-transferases known to be involved in Met salvage in plants, animals, and bacteria, which appear to prefer Gln and Asn (Cooper, 1996). Second, the estimated apparent K_m for Met is exceptionally low (30 μm); aminotransferases typically have K_m values for amino acids in the mm range (Jenkins and Fonda, 1985), and Gln and Asn aminotransferases are no exception (Cooper and Meister, 1985). The E. intestinalis enzyme is also clearly distinct from a Met-glyoxylate aminotransferase involved in glucosinolate synthesis in crucifers, which shows little activity with 2-oxoglutarate as amino acceptor (Chapple et al., 1990), and from a nonspecific, low-affinity Met aminotransferase in pea (Kutacek, 1985). Because DMSP represents about 90% of the reduced S in E. intestinalis, a high-affinity aminotransferase may be needed to sustain a high Met flux to DMSP in the face of competition for Met from AdoMet and methionyl-tRNA synthetases, which can have high affinities for Met (Burbaum and Schimmel, 1992; Schröder et al., 1997).

As with Met transamination, a small capacity to convert MTOB to MTHB may be widespread, because MTHB (of undetermined configuration) has been detected as a metabolite of radiotracer Met or MTOB in diverse higher and lower plants, including algae that lack DMSP (Pokorny et al., 1970; Kushad et al., 1983; Miyazaki and Yang, 1987). However, MTHB formation in these cases may be just a reversible side reaction of the Met salvage pathway (Miyazaki and Yang, 1987), which would be expected to carry very little flux compared with DMSP synthesis. Consistent with this interpretation, DMSP-free algae had no more than 5% of the MTOB reductase activity found in DMSP accumulators.

Unlike the Met → MTOB → MTHB reaction sequence, the conversion of MTHB to DMSHB is known only in association with DMSP synthesis, for which it may be the committing step (Gage et al., 1997). The d-MTHB S-methyltransferase catalyzing this step would thus appear to be a novel enzyme of potential regulatory importance. Our finding that its product is an osmoprotectant has both evolutionary and metabolic engineering implications. Supposing that, like many extant plants, ancient algae had some capacity to form MTHB, evolution of an MTHB methyltransferase could have enabled some DMSHB synthesis to occur and so conferred a selective advantage. DMSHB might thus have served as an ancestral sulfonium osmoprotectant from which synthesis of the more potent protectant DMSP later evolved. In this connection it is noteworthy that two other algal sulfonium compounds, gonyauline (cis-2-[dimethylsulfonio]cyclopropanecarboxyl-ate) (Nakamura et al., 1992) and 4-dimethylsulfonio-2-methoxybutyrate (Blunden and Gordon, 1986), are structurally related to DMSHB and that both could hypothetically be derived from it by single-step reactions.

With respect to metabolic engineering of the DMSP pathway in crops or other organisms, it may be necessary to get only as far as DMSHB to achieve a useful degree of osmotic stress resistance. Whereas genes for Met aminotransferase, MTOB reductase, and MTHB methyltransferase might all be required for this, some crucifers have high Met-glyoxylate aminotransferase activities (4–50 pkat g⁻¹ fresh weight; Glover et al., 1988; Chapple et al., 1990) and in such cases just two engineered genes would perhaps suffice.

Abbreviations:

AdoMet

S-adenosyl-l-Met

ammediol

2-amino-2-methyl-1,3-propanediol

DMS

dimethylsulfide

DMSHB

4-dimethylsulfonio-2-hydroxybutyrate

DMSP

3-dimethylsulfoniopropionate

LDH

lactate dehydrogenase

MDH

malate dehydrogenase

MTHB

4-methylthio-2-hydroxybutyrate

MTOB

4-methylthio-2-oxobutyrate

MTP

3-methylthiopropionate

TLE

thin-layer electrophoresis