Characterization and Functional Expression of cDNAs Encoding Methionine-sensitive and -insensitive Homocysteine S-Methyltransferases from Arabidopsis *

Plants synthesize S-methylmethionine (SMM) from S-adenosylmethionine (AdoMet), and methionine (Met) by a unique reaction and, like other organisms, use SMM as a methyl donor for Met synthesis from homocysteine (Hcy). These reactions comprise the SMM cycle. Two Arabidopsis cDNAs specifying enzymes that mediate the SMM → Met reaction (SMM:HcyS-methyltransferase, HMT) were identified by homology and authenticated by complementing an Escherichia coli yagDmutant and by detecting HMT activity in complemented cells. Gel blot analyses indicate that these enzymes, AtHMT-1 and -2, are encoded by single copy genes. The deduced polypeptides are similar in size (36 kDa), share a zinc-binding motif, lack obvious targeting sequences, and are 55% identical to each other. The recombinant enzymes exist as monomers. AtHMT-1 and -2 both utilize l-SMM or (S,S)-AdoMet as a methyl donor in vitro and have higher affinities for SMM. Both enzymes also use either methyl donor in vivo because both restore the ability to utilize AdoMet or SMM to a yeast HMT mutant. However, AtHMT-1 is strongly inhibited by Met, whereas AtHMT-2 is not, a difference that could be crucial to the control of flux through the HMT reaction and the SMM cycle. Plant HMT is known to transfer the pro-R methyl group of SMM. This enabled us to use recombinant AtHMT-1 to establish that the other enzyme of the SMM cycle, AdoMet:MetS-methyltransferase, introduces the pro-Smethyl group. These opposing stereoselectivities suggest a way to measure in vivo flux through the SMM cycle.

a reaction mediated by AdoMet:Met S-methyltransferase (MMT, EC 2.1.1.12) (1)(2)(3). SMM can then serve as a methyl donor for the synthesis of Met from homocysteine (Hcy) catalyzed by Hcy S-methyltransferase (HMT, EC 2.1.1.10). The tandem action of MMT and HMT, plus that of S-adenosylhomocysteine (AdoHcy) hydrolase, constitutes the SMM cycle ( Fig. 1). Although MMT and the SMM cycle are unique to plants, HMT occurs in bacteria, yeast, and mammals, enabling them to catabolize SMM of plant origin and providing an alternative to the methionine synthase reaction as a means to methylate Hcy (4 -7).
In wheat and other plants, SMM is synthesized in leaves and transported via the phloem to developing seeds where it can be used to methylate Hcy (8). SMM is also synthesized by morning glory flower buds and then used to methylate Hcy during blooming (9). The halves of the SMM cycle can thus sometimes be separated in space or time. However, both halves may also operate concurrently in the same tissue, and in these cases the cycle has been hypothesized to remove excess AdoMet (3). Testing this hypothetical homeostatic role, which is analogous to that of the cyclic methylation/demethylation of Gly in mammalian liver (10), requires determination of flux through the SMM cycle in defined tissues in vivo. Methods to do this are lacking.
The first enzyme of the SMM cycle, MMT, has been purified from Wollastonia biflora and barley, and characterized (2,11). MMT cDNAs have been isolated from W. biflora, Arabidopsis, and maize, and the two latter plants have been shown to have one MMT gene (8). Much less is known about plant HMTs, and none has been cloned from plants or other eukaryotes. HMT was partially purified from jack beans and germinating peas (12,13) and shown to be stereoselective for one of the two methyl groups of SMM (the pro-R methyl) (14). The preparations obtained used either SMM or AdoMet as methyl donor; it was not established whether both activities reside on the same protein. These data appear to indicate that plants can bypass SMM by recycling AdoMet methyl groups directly to Met (Fig.  1, dotted arrows). However, the AdoMet substrates used in these experiments most probably contained significant levels of the nonphysiological R,S diastereomer (15), and it has been suggested that this, not the physiological S,S form, is the substrate for HMTs (7). The form of AdoMet that plant HMTs utilize is therefore unclear. Neither jack bean nor pea HMT was strongly inhibited by Met (Յ25% inhibition by 10 mM Met; Refs. 12 and 13), which contrasts with the Met sensitivity of the yeast enzyme (12).
Recently, the Escherichia coli YagD protein was shown to be an HMT, and a similar enzyme, selenocysteine Se-methyltransferase (SeCysMT), was characterized and cloned from the selenium-accumulating plant Astragalus bisulcatus (7,16,17). These enzymes share significant primary sequence homology (7) and have a GGCC motif near the C terminus. The cysteine residues in this motif have been implicated in zinc binding in two other enzymes that catalyze methyl transfers to Hcy, E. coli B 12 -dependent Met synthase and mammalian betaine-Hcy methyltransferase (18,19). The enzyme-bound zinc is required to activate the thiol group of Hcy for nucleophilic attack (18).
In this work, we identified two Arabidopsis homologs of YagD and confirmed that they encode HMTs. The recombinant enzymes were partially characterized, with emphasis on clarifying their substrate specificity and sensitivity to Met. We also surveyed the genomic complexity of HMT genes in Arabidopsis and used the known stereoselectivity of HMT to establish that of the other enzyme of the SMM cycle, MMT. The results of the stereospecificity study suggest a novel approach to determining flux through the SMM cycle in vivo.
Separation of AdoMet Diastereomers-The S,S (biologically active) and R,S (inactive) diastereomers of AdoMet were separated by HPLC essentially as described by Beaudouin et al. (22). Analytical scale separations of [methyl-14 C]AdoMet and unlabeled AdoMet were made on a 1 ϫ 150 mm Reliasil C18 column using a microbore HPLC system (UMA model, Michrom Bioresources, Auburn, CA). Solvent A was water containing 0.1 M sodium acetate, 20 mM citric acid, 0.93 mM octanesulfonic acid, and 0.12 mM EDTA; solvent B was methanol, and the gradient was from 100 -95% solvent A in 45 min. The elution profile was monitored at 258 nm. The [methyl-14 C]AdoMet contained no detectable R,S form (1% or less) and was used without further purification. As previously reported (15), unlabeled AdoMet was found to contain Ϸ15% of the R,S isomer. In the few cases (see "Results") in which unlabeled AdoMet was included in enzyme assays, specific radioactivity calculations were based on its (S,S)-AdoMet content.
cDNA Generation, Sequencing, and Sequence Analysis-Arabidopsis expressed sequence tags, GenBank TM accession numbers T46013 and H37463 (encoding AtHMT-1 and -2, respectively), were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). The Ϸ750base pair insert in H37463, which is truncated at the 5Ј-end, was used to isolate a full-length cDNA from an Arabidopsis (ecotype Landsberg erecta) leaf library in the Uni-Zap XR vector (Stratagene) (provided by T. L. Thomas, Texas A&M University). DNA sequencing procedures were as described (8). Sequence alignments were made using Clustal W 1.7 (27); phylogenetic analysis was carried out using the Darwin system at the ETH server. Homology searches were made using BLAST programs (28).
cDNA Expression in E. coli-HMT coding sequences were amplified from plasmid templates by high fidelity polymerase chain reaction using recombinant Pfu DNA polymerase (Stratagene). The primers included the first or last 6 -7 codons plus restriction sites for cloning into pBluescript SK-and, for the forward primers, a Shine-Delgarno sequence preceded by a stop codon in frame with the LacZ protein encoded by the vector. The AtHMT-1 primers were 5Ј-CGGAATTCTT-GAAGGAAACAGCTATGGTTTTGGAGAAAAAATC-3Ј (forward) and 5Ј-CCCAAGCTTTCATCTTCGTTTCAAATCTC-3Ј (reverse); the AtHMT-2 primers were 5Ј-AAAACTGCAGGTGAAGGAAACAGCTAT-GACCGGAAACTCTTTTAAC-3Ј (forward) and 5Ј-CGGGGTACCCTAA-AGAGATCTGCGGTTGAC-3Ј (reverse). After ligation into pBluescript, constructs were introduced into E. coli strain DH10B by electroporation. Plasmid preparations were sequenced to verify the inserts and used to transform E. coli strain MTD123 by electroporation.
Enzyme Isolation and Molecular Mass Determination-E. coli cultures (50 ml) were grown to an A 600 of 0.6 -1 in LB medium (24) containing 100 g ml Ϫ1 ampicillin and 1 mM isopropyl-1-thio-␤-D-galactopyranoside. Cells were harvested by centrifugation (4000 ϫ g, 10 min, 4°C), washed in buffer A (100 mM Hepes-KOH, pH 7.5, 1 mM DTT, 10% glycerol), recentrifuged, frozen in liquid N 2 , and stored until extraction at Ϫ80°C. Subsequent operations were at 0 -10°C. Cells were resuspended in buffer A (5 ml/50-ml culture) and broken by sonication; the extract was cleared by centrifugation (10,000 ϫ g, 15 min) and used for enzyme assays directly or after desalting on PD-10 columns (Amersham Pharmacia Biotech) equilibrated in buffer A. Extracts were routinely stored at Ϫ80°C after freezing in liquid N 2 ; this was shown not to affect HMT activity. Yeast extracts were prepared as described previously (26) using buffer A. Native molecular masses were estimated using a Waters 626 HPLC system equipped with a Superdex 200 HR 10/30 column (Amersham Pharmacia Biotech); reference proteins were cytochrome c, carbonic anhydrase, bovine serum albumin, and ␤-amylase. Protein was estimated by Bradford's method (29) using bovine serum albumin as standard.
Enzyme Assays-Unless otherwise indicated, assays were made under conditions in which substrates were saturating and product formation was proportional to enzyme level and time. The assays were modifications of that described by Mudd  Electrospray Mass Spectrometry-The [ 13 C]Met formed by the sequential action of MMT and HMT was analyzed on a FinniganMAT LQC (Thermoquest, San Jose, CA) mass spectrometer system. The source voltage was set at 3.5 kV and capillary voltage at 30 V; the capillary temperature was 22°C. Background source pressure was Ϸ1.5 ϫ 10 Ϫ5 torr as read by an ion gauge. The sample flow rate was 10 l min Ϫ1 . The drying gas was N 2 . The LQC was scanned to 2000 atomic mass units. Spectra were acquired for 0.5 min. Samples were dissolved in 50 l of water; 25 l was injected into the mass spectrometer.
DNA Gel Blot Analyses-Arabidopsis genomic DNA was isolated from leaves as described (30). Five-g samples of the isolated DNA were digested, separated in 0.7% agarose gels, and transferred to supported nitrocellulose membrane (Nitropure, MSI) as described (24). Blots were hybridized overnight at 58°C in 5ϫ SSC, 5ϫ Denhardt's solution, 1% SDS, 1 mM EDTA, and 100 g ml Ϫ1 sonicated salmon sperm DNA and washed at low stringency (1ϫ SSC, 0.1% SDS, 37°C). The probes were the full-length AtHMT-1 or -2 cDNAs and were labeled with 32 P by the random primer method. Radioactive bands were detected by autoradiography.

RESULTS
Genomic-based Cloning of HMT cDNAs from Arabidopsis-BLAST searches using the amino acid sequence of E. coli YagD detected two sets of homologous Arabidopsis expressed sequence tags. Sequencing one insert from each set (GenBank TM accession numbers T46013 and H37463) established that they represent two distinct transcripts. The T46013 insert encodes a 326-residue (36.0 kDa) polypeptide, designated AtHMT-1. The H37463 insert encodes only the C-terminal part of a polypep-tide and so was used to isolate the corresponding full-length cDNA from an Arabidopsis leaf library. This cDNA specifies a 333-residue (36.4 kDa) polypeptide, designated AtHMT-2. The deduced AtHMT-1 and -2 proteins are 55% identical to each other, 50 (AtHMT-1) or 68% (AtHMT-2) identical to Astragalus SeCysMT, and 24 -41% identical to YagD and two yeast proteins, Ypl273w and Yll062c, that were shown to be HMTs while our work was in progress (Fig. 2). 2 AtHMT-1 and -2 also share significant sequence identity (20 -26%) with the N-terminal region of E. coli B 12 -dependent Met synthase and with mammalian betaine-Hcy methyltransferase (not shown). AtHMT-1 and -2 both have a GGCC zinc-binding motif near the C terminus, as well as a third conserved cysteine sited 65 residues upstream that may also be a zinc ligand (18,19). Both AtHMT-1 and -2 appear to lack targeting sequences (e.g. chloroplast or mitochondrial transit peptides), indicating that they are cytosolic enzymes.
Complementation of an E. coli yagD Mutant and Detection of HMT Activity-The coding regions of AtHMT-1 and -2 were subcloned into pBluescript SK-. To express the HMTs as native proteins and not LacZ fusions, the coding sequences were preceded by a stop codon in frame with LacZ and a Shine-Delgarno sequence. These constructs were introduced into E. coli strain MTD123 (⌬yagD ⌬metE ⌬metH), which lacks Met synthase and HMT activity, and is consequently a Met auxotroph that cannot grow on SMM (16). Both constructs enabled transformants to grow on SMM (Fig. 3A); no transformants grew on medium without SMM, indicating complementation of the yagD mutation and not the metE or metH mutations (not shown). No complementation was observed with the vector alone (Fig. 3A), and retransforming MTD123 with rescued plasmids containing the AtHMT-1 or -2 cDNAs restored the ability to grow on SMM, showing that the complementation is because of the encoded plant protein. HMT activity was readily detected in extracts of the complemented strains but not, as expected, in cells transformed with the vector alone (Fig. 3B). The specific activity of AtHMT-1 was Ϸ10-fold higher than that of AtHMT-2; this difference was observed consistently in independent experiments. To authenticate the observed activities, the [ 35 S]Met reaction products were verified by TLC (Fig. 3C).
Methyl Acceptor Specificity of AtHMT-1 and -2-We compared the ability of AtHMT-1 and -2 to catalyze methyl transfer from L-SMM to various thiols and related compounds, using L-Hcy as the benchmark (Table I). Both enzymes utilized D-Hcy, although AtHMT-1 showed a marked preference for the L form. A similar lack of stereospecificity toward Hcy has been noted for other HMTs (7, 31). AtHMT-1 showed significant activity with L-and D-cysteine, which is noteworthy as cysteine is not a substrate for E. coli or yeast HMTs (7,31). Neither enzyme attacked DL-selenocysteine (Table I), glutathione, coenzyme A, sulfide, or thiocyanate (not shown).
Methyl Donor Specificity of AtHMT-1 and -2-SMM occurs in plants as the L enantiomer (8) and AtHMT-1 and -2 both proved to be specific for this form: with 20 M D-or L-[1-14 C]SMM and 2 mM L-Hcy as substrates, activities with D-SMM were undetectable (Ͻ3% of those with L-SMM). To compare L-SMM and (S,S)-AdoMet as methyl donors, Michaelis constants and relative V max values were determined for both enzymes (Table II). (S,S)-AdoMet was found to be a methyl donor for both enzymes, but the K m values were higher than for L-SMM (67-fold for AtHMT-1 and 4.5-fold for AtHMT-2) and the V max values were lower. The (S,S)-[methyl-14 C]AdoMet used in these experiments contained no detectable (Յ1%) R,S diastereomer and, in the assay conditions used (pH 7.5, 30 min), Յ0.3% (R,S)-AdoMet is expected to form by racemization (15). (R,S)-AdoMet therefore did not contribute significantly to the observed activities. Because SMM and (S,S)-AdoMet are substrates, K m values for L-Hcy were determined with both (Table II); fairly similar values were obtained with both methyl donors and with both enzymes. To screen for other potential methyl donors, unlabeled compounds were tested for their ability to inhibit methyl transfer from [ 35 S]SMM when added to assays in 5-fold molar excess. Glycine betaine, choline, phosphocholine, DMSP, and 5-methyltetrahydrofolate had little effect on either enzyme (Յ13% inhibition; data not shown), making it unlikely that they are significant methyl donors.
Met Sensitivity and Other Biochemical Properties of AtHMT-1 and -2-With either L-SMM or (S,S)-AdoMet as a methyl donor, AtHMT-1 activity showed strong product inhibition by L-Met, whereas AtHMT-2 did not, being almost unaffected by L-Met concentrations in the physiological range (Յ500 M) (Fig. 4). AtHMT-1 and -2 both showed maximal activity at pH 7.5. Neither was stimulated by Zn 2ϩ (0.1 or 1 mM). AtHMT-1 activity was modestly inhibited by 1 mM EDTA (26%); AtHMT-2 activity was not. The molecular masses of the native AtHMT-1 and -2 enzymes were estimated by size exclusion chromatography to be 36 kDa. This indicates that both enzymes exist as monomers, as do other HMTs and SecysMT (7,17,31).
Complementation of Yeast HMT Mutations-Yeast cells take up SMM and AdoMet and metabolize them to Met via the action of HMT (5,6). Disruption of the open reading frames yll062c (MHT1) and ypl273w (SAM4) (Fig. 2) has demonstrated that they specify HMTs that prefer SMM and AdoMet, respectively. 2 A triple disruptant (CY61-1D) lacking Met synthase as well as both HMTs is consequently a Met auxotroph that cannot use SMM or AdoMet in place of Met. To confirm that AdoMet and SMM serve as methyl donors for AtHMT-1 and -2 in vivo, each was expressed in CY61-1D and the transformants were tested for the ability to grow on SMM or AdoMet (Fig. 5) (this type of experiment cannot be carried out in E. coli because it cannot absorb AdoMet). AtHMT-1 and -2 enabled growth on either compound establishing that SMM and AdoMet are indeed substrates for both enzymes in vivo as well as in vitro. To exclude the possibility that the differing Met sensitivities shown in Fig. 4 are an artifact of expression in E. coli, SMM: Hcy S-methyltransferase activity was assayed in desalted extracts of yeast transformants expressing AtHMT-1 and -2. As with the recombinant enzymes from E. coli, AtHMT-1 was inhibited strongly by L-Met (92% at 500 M L-Met), whereas AtHMT-2 was not. Activities without L-Met were 1.9 and 1.8 nmol min Ϫ1 mg Ϫ1 protein for AtHMT-1 and -2, respectively; these nearly equal values contrast with the Ϸ10-fold difference seen when these enzymes are expressed in E. coli (Fig. 3B). These data suggest that AtHMT-2 may be less stable (or synthesized more slowly) than AtHMT-1 in the bacterial host.
Diastereospecificity of Methyl Transfer in the SMM Cycle-Because HMT is known to transfer the pro-R methyl group of SMM to Hcy (14), we used recombinant AtHMT-1 to determine the diastereospecificity of MMT, the other enzyme of the SMM cycle. To do this, L-[U-13 C 5 ]Met and unlabeled AdoMet were used as substrates for recombinant Arabidopsis MMT; the SMM formed in this reaction was then incubated with unlabeled L-Hcy and AtHMT-1. The resulting 13 C-labeled Met was analyzed by electrospray MS, together with a 1:1 mixture of unlabeled Met and L-[U-13 C 5 ]Met for comparison (Fig. 6). The product of the MMT/HMT reactions gave peaks of almost equal intensity at m/z 151 and 154, corresponding to [ 13 C 1 ]Met and [ 13 C 4 ]Met, and no appreciable signal above that expected for natural abundance 13 C, 15 (18,19). The asterisk marks a third conserved cysteine residue. AtHMT-1 and -2, Arabidopsis HMT-1 and -2; YagD, E. coli YagD (BAA12002); Yll062c, S. cerevisiae Yll062c (S50958); Ypl273w, S. cerevisiae Ypl273w (S65306); SecysMT, A. bisulcatus selenocysteine Se-methyltransferase (CAA10368). 12 C in the original [ 13 C 5 ]Met substrate (Fig. 6B). These data show that MMT introduces a methyl group into the pro-S position of SMM, i.e. that MMT and HMT have opposite stereoselectivities.
Genomic Complexity and Relationships of HMT Genes in Arabidopsis-Southern blot analyses carried out at low stringency indicated that both AtHMT-1 and AtHMT-2 are encoded by single genes (Fig. 7, A and B). Consistent with this result, BLAST searches of the Arabidopsis genome (Ϸ84% complete at the time of searching) revealed a chromosome III sequence specifying AtHMT-1 (AB023041, nucleotides 21893-23610) but no other closely related sequences. Molecular phylogenic analysis (Fig. 7C) of the sequences aligned in Fig. 2 suggests (a) that AtHMT-2 and Astragalus SecysMT belong on a branch distinct from AtHMT-1, and (b) that extant HMTs are derived from a single ancestral gene that existed prior to the divergence of eubacteria and eukaryotes and has undergone independent duplications in plant and yeast lineages.  (1), the met6 ypl273 yll062 mutant CY61-1D (2), and CY61-1D transformed with pVT102-L containing AtHMT-1 (3) or AtHMT-2 (4), or alone (5) were plated on minimal medium containing 0.1 mM Met, L-SMM, or AdoMet. The AdoMet used was not purified to remove the R,S diastereomer, because in the conditions used, this forms continuously in the medium from racemization of (S,S)-AdoMet (15).

DISCUSSION
The identification of cDNAs encoding plant HMTs completes the set of genes required for operation of the SMM cycle, the others being MMT and AdoHcy hydrolase (3). This opens the way to comprehensive studies of the expression of these genes and to the systematic application of reverse genetics to probe the function of SMM and its cycle. Furthermore, extracts of E. coli expressing AtHMT-1 or -2 have specific activities Ն10 2fold higher than those of the best plant sources (12, 13) making them good material for future enzyme purification. More generally, the HMT cDNAs reported here appear to be the first identified from a eukaryote.
AtHMT-1 and -2 resemble HMTs from other organisms in overall primary structure and in being monomeric proteins. They lack obvious targeting sequences and are therefore presumably cytosolic enzymes. HMT has yet to be definitively localized in plant cells, but preliminary work with pea leaves indicates that it is cytosolic, 3 as are other enzymes involved in Met metabolism, i.e. MMT, Met synthase, AdoMet synthetase, and AdoHcy hydrolase (8,32). AtHMT-1 and -2 share with other HMTs and with SecysMT, a GGCC zinc-binding motif (18), plus a third conserved cysteine residue. This strongly suggests that they have a zinc cofactor. Neither enzyme was stimulated by zinc or severely inhibited by EDTA, but this may be because the zinc is tightly bound, as it is in betaine-Hcy methyltransferase (19).
Our results demonstrate that the physiological S,S diastereomer of AdoMet is a substrate for plant HMTs. This indicates that plants have the potential to bypass SMM by transferring methyl groups directly from AdoMet to Hcy (Fig. 1, dotted  arrows), and the complementation experiments with yeast confirm that plant HMTs can mediate this reaction in a foreign host. But how much flux does this bypass actually carry in planta? Kinetic considerations indicate that it may be very little, especially in tissues where AtHMT-1 is the predominant isoform. AtHMT-1 has K m values for SMM and AdoMet of 29 and 1950 M, respectively and the V max value with SMM is 2.8-fold higher. SMM levels are reported to range from about 5 to Ͼ300 nmol g Ϫ1 fresh weight in various tissues, and SMM/ AdoMet ratios are reported to range from Ϸ1 to Ͼ30 (1,8,(33)(34)(35)(36). Some SMM may be sequestered in the vacuole; however, radiotracer kinetic studies indicate that the metabolically active (presumably cytosolic) SMM pool is a large fraction of the total (36). Assuming the cytosol to be Ϸ5% of tissue water volume (37), it follows from these data that typical cytosolic SMM concentrations are likely to be Ն100 M, and AdoMet concentrations are likely to be similar or lower. In such conditions flux through the AdoMet-driven reaction would be Յ3% of that through the SMM-driven reaction. Simply put, a high prevailing SMM concentration can deny AdoMet access to the AtHMT-1 active site and thereby suppress futile cycling of AdoMet.
Our finding that AtHMT-1 is strongly inhibited by Met is 3  AtHMT-2 cDNA (B). Washing was at low stringency. The sizes of hybridizing bands match the cDNAs with respect to the predicted restriction sites. Genomic reconstruction standards were made with AtHMT-1 and -2 cDNAs equivalent to 1 and 5 copies/haploid genome (shown on the left of each panel). Note that AtHMT-1 and -2 cDNAs cross-hybridize only very weakly as they differ at 55% of the base pairs. The positions of DNA size markers (kb, kilobase) are marked. Abbreviations are as in Fig. 2. C shows a molecular phylogenic tree of the protein sequences from Fig. 2. novel, because the plant HMTs so far known are Met-insensitive (12,13). Met sensitivity may be crucial to the control of flux through the HMT reaction and the SMM cycle. A Met-sensitive HMT could stop the cycle turning when Met levels are elevated, whereas a Met-insensitive enzyme could allow SMM 3 Met conversion even when Met levels are high. It is therefore noteworthy that free Met levels in developing seeds can greatly exceed those in other tissues (Ն400 versus 10 -30 nmol g Ϫ1 fresh weight) (1, 35, 38 -40) and that HMTs isolated from seeds are Met-insensitive (12,13). Moreover, DNA array data indicate that the predominant HMT expressed in developing Arabidopsis seeds is the Met-insensitive AtHMT-2. 4 Another difference between the Arabidopsis HMTs is that AtHMT-1 attacks cysteine. This could explain the origin of S-methylcysteine in the Brassicaceae. No enzyme that catalyzes the Smethylation of cysteine has hitherto been demonstrated (1), although radiotracer data show that the reaction occurs in vivo (41).
The SMM cycle has been proposed to rectify overshoots in the conversion of free Met to AdoMet, thereby sustaining a free Met pool for protein synthesis (3). This hypothesis was based largely on data for whole Lemna plantlets (3), and it has since been found that SMM is transported between organs in the phloem (8). This raises the question of whether the SMM was produced and utilized in the same organs in the Lemna experiments and shows that accurate flux measurements are now needed to clarify the functions of the SMM cycle. Only a few such measurements have been made, and these come from unusual plants (W. biflora and Spartina alterniflora) that convert SMM to DMSP. Isotope tracer studies of SMM synthesis and metabolism in leaves of these plants showed that the methyl flux from Met to SMM was high, but there was little or none from SMM to Hcy, i.e. the SMM cycle turned slowly if at all (36,42). The approach used to make these measurements depends on the metabolism of SMM to DMSP and so unfortunately cannot be applied to the great majority of plants that do not synthesize DMSP.
There is thus a need for methods to estimate flux through the SMM cycle in tissues of non-DMSP-accumulating plants. Our finding that the enzymes of the cycle have opposing stereoselectivities suggests a novel way to do this. For example, consider an organ that imports SMM via the phloem and ultimately uses it to produce Met that is used for protein synthesis. If the SMM cycle is not operating, then supplied SMM that has a 13 C label in the C 4 backbone and the pro-R methyl and 2 H 3 label in the pro-S methyl will give rise to only two labeled species of Met in proteins: [methyl-2 H 3 , 13 C 4 ]Met and [methyl-13 C]Met. However, if the SMM cycle is operating, the additional species [methyl-2 H 3 ]Met, [ 13 C 4 ]Met, and [ 13 C 5 ]Met will be found in proteins and will become relatively more abundant with each turn of the cycle.