The sng2 mutant of Arabidopsis is defective in the gene encoding the serine carboxypeptidase-like protein sinapoylglucose:choline sinapoyltransferase


For correspondence (fax: +1 765 496 7213; email:


Serine carboxypeptidase-like (SCPL) proteins have traditionally been assigned roles in the hydrolytic processing of proteins; however, several SCPL proteins have recently been identified as catalysts in transacylation reactions of plant secondary metabolism. The novel functions of these enzymes suggest a catalytic diversity for plant SCPL proteins that extends beyond simple hydrolysis reactions. Characterization of the Arabidopsis sng2 (sinapoylglucose accumulator 2) mutant has identified another SCPL protein involved in plant secondary metabolism. The sng2 mutant was isolated by screening seed extracts for altered levels of sinapate esters, a group of phenylpropanoid compounds found in Arabidopsis and some other members of the Brassicaceae. Homozygous sng2 seeds accumulate sinapoylglucose instead of sinapoylcholine, and have increased levels of choline and decreased activity of the enzyme sinapoylglucose:choline sinapoyltransferase (SCT). Cloning of the SNG2 gene by a combination of map-based and candidate gene approaches demonstrates that SCT is another member of the growing class of SCPL acyltransferases involved in plant secondary metabolism.


Serine carboxypeptidases are often thought to play a role in the processing and degradation of proteins and peptides because they are capable of catalysing the cleavage of C-terminal peptide bonds. As a result, when proteins or cDNAs that exhibit homology to serine carboxypeptidases have been isolated from tissues actively engaged in protein turnover, it has often been assumed that these serine carboxypeptidase-like (SCPL) proteins are playing a role in proteolytic processes (Bamforth et al., 1979; Bradley, 1992; Walker-Simmons and Ryan, 1980; Washio and Ishikawa, 1994). Similarly, the annotation of many SCPL genes throughout the Arabidopsis genome might be taken to imply that their encoded proteins are involved in protein degradation. The sequence similarity between these gene products and genuine serine carboxypeptidases has been shown to be potentially misleading by the recent identification of two SCPL acyltransferases involved in plant secondary metabolism (Lehfeldt et al., 2000; Li and Steffens, 2000) and an SCPL hydroxynitrile lyase involved in cyanogenic glycoside degradation (Wajant et al., 1994). These examples indicate that SCPL enzymes have a broader catalytic capability than is implied by their similarity to serine carboxypeptidases.

With the completion of the Arabidopsis Genome Initiative (AGI), 46 SCPL genes have been annotated in the Arabidopsis genome. The large number of genes in this family and the developing body of literature implicating SCPL proteins in non-traditional reactions suggests that at least some of these 46 Arabidopsis genes may encode proteins that function in reactions other than protein processing. One of these SCPL genes was recently shown to be required for the synthesis of sinapoylmalate, a UV-protective phenylpropanoid accumulated by Arabidopsis and some other members of the Brassicaceae. In contrast to the function suggested by its homology, the encoded enzyme, sinapoylglucose:malate sinapoyltransferase (SMT), catalyses a transesterification reaction (Lehfeldt et al., 2000). A defect in the SMT gene causes the sng1 mutant to accumulate sinapoylglucose in its leaves in place of sinapoylmalate (Lehfeldt et al., 2000; Lorenzen et al., 1996).

In addition to the sinapoylmalate found in leaves, Arabidopsis also accumulates a seed-specific sinapate ester, sinapoylcholine, also known as sinapine. We predicted that the final enzyme of the seed-specific pathway of sinapate ester metabolism, sinapoylglucose:choline sinapoyltransferase (SCT), might also be an SCPL protein because both SMT and SCT use sinapoylglucose as a substrate and catalyse similar transacylation reactions (Figure 1). In this paper, we describe the isolation and characterization of the Arabidopsis sng2 mutant, which accumulates sinapoylglucose in its seeds in place of sinapoylcholine. By cloning the SNG2 gene, we demonstrate that it encodes SCT and that SCT is an SCPL protein.

Figure 1.

The pathway of sinapate ester biosynthesis.

The enzymes required for the conversion of phenylalanine to sinapic acid are phenylalanine ammonia lyase (PAL), cinnamate 4-hydroxylase (C4H), p-coumarate 3-hydroxylase (C3H), caffeic acid/5-hydroxyferulic acid O-methyltransferase (OMT) and ferulate 5-hydroxylase (F5H). The enzymes unique to sinapate ester biosynthesis are UDP-glucose:sinapic acid glucosyltransferase (SGT), sinapoylglucose:malate sinapoyltransferase (SMT), sinapoylglucose:choline sinapoyltransferase (SCT) and sinapoylcholinesterase (SCE). The biochemical block in the sng2 mutant is indicated with a vertical line across the step catalysed by SCT.


Isolation of the sng2 mutant

To identify mutations affecting the accumulation of seed-specific sinapate esters, we screened methanolic seed extracts from 3000 M3 families by TLC. One mutant was identified that showed greatly reduced levels of sinapoylcholine but accumulated high levels of a compound that co-chromatographed with sinapoylglucose (Figure 2a). In crosses to Columbia wild-type, the mutant phenotype segregated as a single, recessive Mendelian gene (Table 1). Quantitative HPLC analysis of Landsberg erecta wild-type and mutant seed extracts confirmed that mutant seeds accumulated reduced levels of sinapoylcholine and instead accumulated a compound that co-chromatographed with sinapoylglucose, in an amount almost equivalent to the amount of sinapoylcholine normally accumulated in the wild-type (Figure 2b; Table 2).

Figure 2.

Analysis of sinapate ester content in wild-type and sng2 seeds.

(a) Seed extracts were prepared from Landsberg erecta wild-type (lane 1) and sng2 (lane 2). A leaf extract from the sinapoylglucose-accumulating sng1-1 mutant (lane 3) was included as a sinapoylglucose standard. Extracts were analysed on silica gel plates with the mobile phase n-butanol:acetic acid:water (5:2:3 v/v/v). Sinapoylcholine (SC) and sinapoylglucose (SG) were visualized under 312 nm UV light. o, origin; sf, solvent front. (b) Seed extracts from Landsberg erecta wild-type (top) and sng2 (bottom) were analysed by HPLC with UV detection at 335 nm.

Table 1.  Segregation of the sng2 allele in an F2 population derived from a sng2 × Columbia cross
  1. The sng2 phenotype was scored in F3 progeny by dissecting individual embryos from the siliques developing on F2 plants and evaluating their sinapate ester content by TLC.

  2. a Values expected for inheritance of a recessive allele of a Mendelian gene (χ2 = 0.7, P > 0.4).

Table 2.  Biochemical characterization of wild-type and sng2 seeds
LineSinapoylglucoseaSinapoylcholineaCholineaSCT activityb
  1. a nmol mg−1 ± SE

  2. b pkat mg−1 ± SE

  3. ND, not detectable

SNG2/SNG2ND33.22 ± 1.0034.06 ± 2.3930.61 ± 0.38
sng2/sng219.30 ± 3.975.16 ± 0.0860.18 ± 8.560.30 ± 0.01

To identify the compound accumulated in the mutant, seed extracts were analysed by LC/MS. The peak that co-chromatographs with sinapoylglucose in the mutant gave the prominent parental (M-H)-ion at m/z 385 expected for sinapoylglucose as well as a fragment ion at m/z 205, consistent with the loss of glucose from the parent molecule. The TLC, HPLC and LC/MS results were consistent with the hypothesis that the mutant accumulates 1-O-sinapoyl-β-d-glucose. Based upon this phenotype, the mutant was designated sng2 for sinapoylglucose accumulator 2.

Sinapate ester profiles of the wild-type and the sng2 mutant

The accumulation of sinapate esters in Arabidopsis seedlings and maturing embryos is developmentally regulated, both qualitatively and quantitatively (Lorenzen et al., 1996; Ruegger et al., 1999). The sinapate ester profiles of mutant and wild-type tissues were examined in order to determine whether the sng2 mutation altered the accumulation of sinapate esters in tissues other than mature seeds. For these experiments, seedlings were grown on modified MS medium that contained nitrate as its sole nitrogen source because ammonia has been previously shown to inhibit sinapoylmalate accumulation and decrease SMT activity in radish seedlings (Dahlbender and Strack, 1984; Strack et al., 1986). In Landsberg erecta wild-type seedlings, sinapoylcholine levels decreased during the first 3 days after imbibition and there was a transient rise in sinapoylglucose levels (Figure 3a). The decrease in sinapoylglucose levels on days 4–6 was accompanied by an increase in sinapoylmalate content. These results were consistent with previous analyses of the Columbia wild-type ecotype (Lorenzen et al., 1996). In the sng2 mutant, the total level of sinapate esters in seeds is nearly equivalent to that of wild-type (Table 2; Figure 3b,d). As mutant seedlings developed, sinapoylglucose levels remained elevated until after day 3 when, as in wild-type, sinapoylmalate levels increased (Figure 3b). These data indicate that the sng2 mutation, and the absence of sinapoylcholine in the sng2 mutant, do not have pleiotropic effects on the developmental regulation of sinapate ester biosynthesis and turnover in seedlings.

Figure 3.

Developmental changes in sinapate ester content in wild-type and sng2 seedlings and embryos.

Sinapate esters from Landsberg erecta wild-type (a) and sng2(b) seedlings grown on modified Murashige and Skoog agar plates (Lorenzen et al., 1996) and Landsberg erecta wild-type (c) and sng2(d) siliques developing on individual inflorescences were analysed by HPLC with UV detection at 335 nm and quantified using the extinction coefficient of sinapic acid. Siliques at the bottom of the inflorescence represent the oldest developmental time point, whereas siliques at the top of the inflorescence contain the most immature embryos. Each point represents the average of three seedlings or siliques ± SE. (▾) sinapoylmalate; (●) sinapoylglucose; (○) sinapoylcholine.

In order to examine the impact of the sng2 mutation on sinapate ester accumulation in maturing embryos, the sinapate ester content of siliques developing on individual inflorescences was quantified; siliques at the bottom of the rachis contain the most mature embryos, whereas siliques at the top of the inflorescence represent the most immature embryos. In the wild-type, sinapoylglucose levels increase during early maturation. As siliques mature, a decrease in sinapoylglucose levels is accompanied by an increase in sinapoylcholine content (Figure 3c). In contrast, in the sng2 mutant, sinapoylcholine levels remain very low throughout embryonic development and sinapoylglucose content increases to near the same level as the sinapoylcholine found in the wild-type (Figure 3d). These results indicate that the sng2 mutant fails to accumulate sinapoylcholine to a significant degree at any developmental time point examined, and that, in the absence of sinapoylcholine, total levels of sinapate esters in the mutant remain essentially unchanged.

Biochemical profiling of the wild-type and sng2 mutant

In an attempt to identify the underlying reason why the sng2 mutant fails to accumulate sinapoylcholine, the free choline levels of the mutant and the wild-type were quantified by plasma desorption mass spectrometry. As choline is a substrate for SCT, the lack of sinapoylcholine accumulation in the mutant could be explained by a defect in choline biosynthesis; however, the free choline content of the mutant was twice that of wild-type, indicating that a lack of choline is not responsible for the sng2 phenotype (Table 2). Interestingly, the excess of free choline in the mutant was equal to the amount of sinapoylcholine normally present in the wild-type. The presence of both sinapoylglucose and choline in sng2 seeds suggests that sng2 seeds should be able to synthesize sinapoylcholine unless they lack a functional SCT enzyme; therefore, the ability of wild-type and mutant enzyme extracts to synthesize sinapoylcholine was quantified. The results of in vitro enzyme assays from crude seed extracts revealed that the specific activity of SCT in the mutant was only 1% that of the wild-type (Table 2). The data from these biochemical analyses suggest that the SNG2 gene is required for SCT expression or activity.

SNG2 encodes a serine carboxypeptidase-like protein

The cloning of the SNG1 gene demonstrated that it encodes SMT, and revealed a role for an SCPL protein in sinapate ester metabolism (Lehfeldt et al., 2000). Because the reactions catalysed by SMT and SCT are similar transacylation reactions that use sinapoylglucose as an activated sinapate donor, we predicted that SCT might also be an SCPL acyltransferase. This hypothesis is supported by experiments in which Arabidopsis seed extracts were treated with PMSF, an inhibitor of enzymes with active site serine residues, including serine carboxypeptidases. PMSF treatment inhibited the activity of the Arabidopsis SCT by 25%, similar to the results previously obtained with the Brassica napus SCT enzyme (Vogt et al., 1993).

With the expectation that the SNG2 locus was likely to encode SCT, we attempted to isolate the SNG2 gene using a combination of map-based cloning and candidate gene approaches. In an initial survey of an sng2 mapping population, the SNG2 gene was mapped near the RFLP marker m217 (one recombinant chromosome out of 38 total chromosomes). In an analysis of a larger mapping population, the SNG2 locus demonstrated linkage to the SSLP marker nga151 (18 recombinant chromosomes out of 388 total chromosomes) and tight linkage to the SSLP marker nga249 (no recombinant chromosomes out of 384 total chromosomes) (Figure 4).

Figure 4.

Mapping of the SNG2 locus.

The SNG2 locus was initially mapped near the RFLP marker m217 (one recombinant chromosome out of 38 total chromosomes). Further mapping was carried out with the SSLP markers nga249 (no recombinant chromosomes out of 384 total chromosomes) and nga151 (18 recombinant chromosomes out of 388 total chromosomes). The region flanking nga249 is covered by the BAC contig illustrated at the bottom of the figure. Two SCPL genes, T22D6.200 and F17I14.170, near the nga249 marker were identified, and the latter has been shown to correspond to the SNG2 locus.

Before the sequencing in this region of chromosome V was complete, a candidate SCPL gene (T22D6.200) was identified on two overlapping bacterial artificial chromosomes (BACs), F24L14 and F27P17 (Figure 4), by probing a BAC library of Col-0 genomic DNA (Mozo et al., 1998) with cDNA from eight SCPL ESTs with no known genomic position. Sequencing of this gene from the sng2 mutant did not reveal an EMS-induced mutation, and transformation of sng2 plants with a binary vector carrying the wild-type allele did not result in complementation of the mutant phenotype (data not shown). After the sequence for this area of chromosome V was released, BAC F17I14, which carried a gene annotated as an SCPL, was identified near BACs F24L14 and F27P17 (Figure 4). BAC F17I14 had not hybridized in the BAC library screening described above.

Several approaches were then taken to determine whether the identified candidate gene corresponded to SNG2. First, RNA gel blot hybridization analysis was used to quantify F17I14.170 transcript from various tissues, including young leaves, mature leaves, senescent leaves, 10-day-old seedlings, stem tissue, siliques, flowers and roots. This experiment indicated that F17I14.170 is expressed highly only in siliques (Figure 5), consistent with the pattern expected for a gene encoding SCT. Expression of F17I14.170 in the sng2 mutant was also examined by quantifying the abundance of the transcript by RNA gel blot hybridization of total silique RNA from both Landsberg erecta wild-type and the sng2 mutant. Using cyclophilin as an internal standard for loading, expression in the mutant was approximately 15% of wild-type (data not shown). As EMS-induced mutations often lead to mRNA destabilization, the failure of F17I14.170 transcript to significantly accumulate in the mutant also suggests that F17I14.170 represents SNG2. Finally, the putative sng2 genomic allele was amplified by PCR and sequenced. A G → A nonsense mutation corresponding to position 1197 of the cDNA was identified that alters Trp399 of the inferred protein to a stop codon, truncating the last 66 amino acids of the protein (Figure 6). It should be noted that sequencing of the RT–PCR product obtained for the putative SNG2 allele revealed that several portions of genomic sequence that had been annotated as introns by the AGI are actually exons (Figure 6). Inclusion of these nucleotide sequences indicated that F17I14.170 shares all three conserved serine, aspartic acid and histidine residues (S178, D388 and H442 in the F17I14.170 sequence) found in carboxypeptidase Y. The amino acids missing from the inferred mutant protein include His442, which aligns with His398 in carboxypeptidase Y, one of the three amino acids that comprise the active site catalytic triad in carboxypeptidase Y. This mutation also introduces an Sau3AI polymorphism in the putative sng2 allele, which was used to generate a cleavable amplified polymorphic sequence marker. DNA amplified from the wild-type was cleaved by Sau3AI, whereas it was not cleaved when mutant DNA was used as the template for PCR (data not shown). This finding unambiguously demonstrates that the mutation identified by sequencing was not introduced by the PCR process.

Figure 5.

Expression analysis of the SNG2 gene.

RNA was prepared from various tissues and probed with the putative SNG2 cDNA (expressed sequence tag 309H12T7; Genbank accession number AA394342) (top). Expression of cyclophilin (bottom) was used as a loading control. Lane 1, young leaves; lane 2, mature leaves; lane 3, senescent leaves; lane 4, 10-day-old seedlings; lane 5, stems; lane 6, siliques; lane 7, flowers; lane 8, roots.

Figure 6.

The SNG2 cDNA and its deduced amino acid sequence.

The nucleotide sequences not found in the cDNA annotation at GenBank for F17I14.170 are underlined. The nucleotide altered in the sng2 mutant is denoted by an asterisk. Mutation of this site from guanine to adenine in the sng2 mutant alters the trytophan residue in bold to a stop codon. The predicted N-terminal amino acid signal sequence is double underlined. Shaded amino acid sequences correspond to tryptic peptide fragments obtained from SCT purified from Brassica napus (Vogt et al., 1993). These residues correspond to the following B. napus fragments: IDGES, NILYLEDPAG, IVPGAAQQILLGNRK and GYVLGNPAV. The boxed amino acids align with the conserved catalytic triad of carboxypeptidase Y. The GenBank accession number for the SNG2 cDNA is AY033947.

The final evidence that F17I14.170 corresponds to the SNG2 locus was provided by complementation of the sng2 mutant phenotype. A genomic clone of F17I14.170 was isolated from BAC F17I14, and was used to generate a binary vector construct with 1.1 kb of promoter sequence, 2.6 kb of genomic sequence, and 0.3 kb of 3′ sequence. The resulting pBI101-SNG2 vector was transformed into sng2 plants by the floral dip method (Clough and Bent, 1998). When sinapate ester levels of T2 embryos from kanamycin-resistant T1 plants were analysed by TLC and HPLC, sinapoylcholine biosynthesis was not restored in transformed lines carrying the pBI101 empty vector; however, 20 independent lines transformed with the pBI101-SNG2 construct contained wild-type levels of sinapoylcholine (Figure 7).

Figure 7.

Complementation of the sng2 mutant.

Seed extracts were prepared from Landsberg erecta wild-type (a), sng2 transformants carrying the pBI101-empty vector (b), and sng2 transformants carrying the pBI101-SNG2 transgene (c) and were analysed by HPLC with UV detection at 335 nm. SG, sinapoylglucose; SC, sinapoylcholine. To test for complementation, T2 families were identified that segregated 3:1 for pBI101-SNG2 or the pBI101 transgene. Seeds from these populations were then analysed for their sinapate ester content. The presence of sinapoylglucose in the pBI101-SNG2 extracts is explained by the percentage of sng2 seeds within the population that have lost the transgene through segregation.

The SNG2 locus encodes SCT

The evidence described above indicates that SNG2 encodes an SCPL protein, which is likely to be SCT. Several years ago, sequences from Brassica SCT tryptic peptides were published (Vogt et al., 1993), but to date have not been used to clone the SCT gene. The F17I14.170 cDNA sequence contained amino acid sequences corresponding to all four of the tryptic peptide fragments, with two amino acid insertions in the B. napus sequence and two amino acid substitutions (Figure 6), strongly suggesting that the SNG2 gene encodes SCT. Nevertheless, to conclusively demonstrate that the SNG2 locus encodes SCT, the function of the SNG2 protein was determined following expression of the SNG2 cDNA in Escherichia coli. As SMT carries an N-terminal pro-peptide (Lehfeldt et al., 2000), the algorithm described by Nielsen et al. (1997) and available at the SignalP website ( was used to predict which amino acids might comprise a similar pro-peptide in SCT (Figure 6). The portion of the SNG2 cDNA corresponding to the inferred mature protein was then expressed under the control of the T7 promoter in pET28A. SNG2 protein did not significantly accumulate in BL21 (DE3) host cells (data not shown). Visible accumulation of protein required the use of BL21-CodonPlusTM (DE3)-RIL host cells that compensate for rare arginine, isoleucine and leucine codons. Using this host, a novel protein with a molecular mass of approximately 45 kDa, in keeping with the mass of 48 kDa expected for the modified SNG2 polypeptide, was visible in the insoluble fraction of cells containing the pET28A-SNG2 vector induced at 37°C. The band was absent in all fractions from induced cells containing the pET28A empty vector. SNG2 protein was not present in the soluble fraction, suggesting that most of the protein was present in inclusion bodies (Figure 8). Unfortunately, when the soluble fraction of these cell extracts was analysed for SCT activity by HPLC, no enzymatic synthesis of sinapoylcholine was observed, probably because little if any protein had folded into its native state under these induction conditions. As induction at lower temperature and the use of lower concentrations of isopropyl-β-d-thiogalactopyranoside also failed to lead to the production of soluble active enzyme, inclusion bodies were isolated, and the recombinant protein was denatured, refolded, and partially purified by anion exchange chromatography. When the renatured protein was used for SCT assays, HPLC analysis revealed that a compound that co-chromatographs with sinapoylcholine and has a spectrum consistent with known standards of sinapoylcholine was formed in the presence of sinapoylglucose and choline when incubated at 30°C (Table 3). In assays lacking enzyme, trace amounts of sinapoylcholine were identified by HPLC and matrix-assisted laser desorption ionization (MALDI)/MS, presumably reflecting spontaneous transacylation. Although the amount of sinapoylcholine produced in SCT assays was low, probably due to inefficient protein refolding, these data provide conclusive proof that the SNG2 gene encodes SCT.

Figure 8.

Analysis of SNG2 expressed in E. coli.

SDS–PAGE analysis of E. coli harbouring pET28a (lanes 1, 3 and 5) and the SNG2 expression vector pET28a-SNG2 (lanes 2, 4 and 6) for total protein (lanes 1 and 2), insoluble fractions (lanes 3 and 4) and soluble fractions (lanes 5 and 6).

Table 3.  Analysis of SCT activity in E. coli expressing the SNG2 gene
Source of proteinSubstrates addedSinapoylcholine formeda
  • a

    pmol ± SE (n = 3)

pET28a-SNG2++560 ± 45
pET28a-empty vector++73 ± 5
Boiled pET28a-SNG2++74 ± 7
pET28a-SNG2+0 ± 0
pET28a-SNG2+0 ± 0
pET28a-SNG20 ± 0
Buffer alone++55 ± 6


The isolation of the SNG2 gene demonstrates that SCT is another example from an emerging class of acyltransferases that are homologous to serine carboxypeptidases. Only two other examples of SCPL proteins that catalyse transacylation reactions in vivo are known. One of these is SMT, the enzyme responsible for the final step in sinapoylmalate biosynthesis (Lehfeldt et al., 2000). The other is an isobutyryl acyltransferase involved in the synthesis of 2,3,4-tri-O-isobutyrylglucose in wild tomato trichomes (Li and Steffens, 2000). Both of these enzymes are similar to SCT in that they use 1-O-acylglucosides as activated acyl donors in the transacylation reactions that they catalyse. Before the genes encoding these enzymes were isolated, it had not been anticipated that these acyltransferases would be similar to serine carboxypeptidases as carboxypeptidases catalyse distinctly different reactions in which they act as hydrolases that use pro teins and peptides as substrates. Interestingly, carboxypeptidase Y has been shown to catalyse transacylation reactions with amino acids or amino acid derivatives as nucleophiles in vitro; however, it can do so only under alkaline conditions that are not physiologically relevant (Breddam et al., 1980; Widmer et al., 1980). Thus, although SCT, SMT and the Lycopersicon pennellii isobutyryl acyltransferase share significant amino acid identity with carboxypeptidase Y, including at least three inferred active site residues, the structures of the plant proteins must differ from classical carboxypeptidases in such a way as to facilitate the binding of their very different substrates and to favour the transacylation that these enzymes catalyse.

A common feature of serine carboxypeptidases as well as other members of the α/β hydrolase superfamily of proteins is the presence of conserved serine, histidine and aspartic acid active site residues that make up the so-called catalytic triad (Bech and Breddam, 1989; Hayashi et al., 1973, 1975; Liao and Remington, 1990; Liao et al., 1992; Ollis et al., 1992). Presumably, SCT also utilizes the three conserved residues for catalysis, perhaps by nucleophilic attack of S178 on the ester bond of sinapoylglucose, thereby generating a sinapoylated enzyme intermediate. This mechanism would be consistent with previous studies on the reaction kinetics of SCT isolated from B. napus which indicated that SCT has a double displacement mechanism of catalysis (Vogt et al., 1993), typical of serine carboxypeptidases. Despite these similarities, the reaction catalysed by SCT is distinct from that catalysed by carboxypeptidase Y in that carboxypeptidase Y uses water as a nucleophile to cleave the acyl intermediate formed in the first phase of the reaction that it catalyses. In contrast, SCT would have to exclude water from its active site in order to use the hydroxyl of choline as a nucleophile. Without the exclusion of water, SCT would be expected to function as a sinapoylglucose esterase instead of an acyltransferase. Thus, if the catalytic mechanism of SCT is similar to that of carboxypeptidase Y, SCT must employ specific residues or perhaps protein conformations that prevent water from rapidly hydrolysing the sinapoylated enzyme intermediate. Alternatively, S178 may act as a base during catalysis, deprotonating the hydroxyl of choline, thereby facilitating its direct nucleophilic attack on the ester linkage of sinapoylglucose, and obviating the need for exclusion of water from the enzyme's active site. Such a mechanism has recently been postulated for MhpC, a C–C hydrolase from E. coli which, like SCT, SMT and carboxypeptidase Y, employs a catalytic triad (Fleming et al., 2000).

Even with the availability of a completely sequenced genome, without the insights provided by the sng1 and sng2 mutant phenotypes, the functions of the genes encoding SMT and SCT would not have been readily apparent. An understanding of the mechanism that this novel group of proteins uses to carry out transacylation will be fundamental to identifying additional SCPL proteins that function as acyltransferases. In the Arabidopsis genome, there are presently 46 genes annotated as encoding SCPL proteins. Preliminary phylogenetic analysis of their sequences does not group SCT, SMT and the Lycopersicon pennellii isobutyryl acyltransferase into a distinct clade of proteins that might include other putative acyltransferases. Thus, the features of SCPL acyltransferases that distinguish them from traditional serine carboxypeptidases may consist of only a few key residues, or may be elements of secondary or tertiary structure that are not revealed by analysis of their primary sequence.

In addition to providing insight into the role of SCPL proteins in plant metabolism, the sng2 mutant also provides an opportunity to study the function of sinapoylcholine accumulation in seeds of the Brassicaceae. Radiotracer feeding experiments with 14C-choline and 14C-ethanolamine indicated that choline released from the hydrolysis of sinapoylcholine in germinating seeds is incorporated into phosphatidylcholine (Strack, 1981). Despite the clear importance of membrane lipid synthesis in developing seedlings, greatly reduced levels of sinapoylcholine in sng2 seeds appear to have no effect on germination or seedling growth, and the accumulation of free choline in the mutant indicates that sinapoylcholine is not an obligatory storage form for choline in Arabidopsis. These findings are consistent with previous studies on the fah1 mutant which completely lacks sinapate esters at all developmental stages, and also accumulates free choline in its seeds instead of sinapoylcholine (Chapple et al., 1992). Interestingly, some residual sinapoylcholine is found in sng2 seeds, despite the fact that the sng2-1 allele is probably null because it lacks the conserved, catalytic histidine residue. This finding may indicate that there are other acyltransferases that are partially redundant with SNG2. On the other hand, in in vitro reactions, we consistently observed the formation of low levels of sinapoylcholine from sinapoylglucose and choline in the absence of enzyme. These data suggest that spontaneous non-enzymatic synthesis may account for the low levels of sinapoylcholine found in sng2 seeds.

The absence of obvious phenotypes in mutants lacking sinapoylcholine is also a finding of potential agronomic importance. Oilseed rape or canola (Brassica sp.) accumulates sinapoylcholine in seeds, and when post-crushing canola meal is used in poultry feed, by-products of sinapoylcholine degradation impart a fishy taint to eggs (Hobson-Frohock et al., 1977). The examination of breeding lines of B. napus and B. campestris for genetic variation in seed sinapoylcholine accumulation has not identified significant variation for the trait (Vogt et al., 1993). The isolation of the sinapoylcholine-deficient sng2 mutant strongly suggests that it should be possible to manipulate sinapoylcholine levels in Brassica crops, and the cloning of the SNG2 gene provides the tools necessary for a genetic engineering approach to this problem using antisense or co-suppression technology.

Experimental procedures

Plant material

Arabidopsis thaliana L. Heynh. ecotypes Landsberg erecta or Columbia were cultivated at a light intensity of 100 µE m−2 sec−1 at 23°C under a photoperiod of 16 h light/8 h dark in Redi-Earth potting mix (Scotts-Sierra Horticultural Products, Marysville, OH, USA). For seedling material to be used in the analysis of sinapate esters, seeds were surface-sterilized for 10 min in a 2:1 v/v mixture of 0.1% Triton-X 100 and household bleach. Seeds were rinsed thoroughly with sterile water and plated onto modified MS medium (Murashige and Skoog, 1962) (ammonia-free medium to which an additional 20.6 mm potassium nitrate was added in place of ammonium nitrate) containing 0.7% agar.

Genetic methods

For mutant screening, seed (5–10 mg) from M3 families of EMS-mutagenized Landsberg erecta populations were extracted in 200 µl of 50% methanol. An 5 µl aliquot of the extract was analysed by silica gel TLC using a mobile phase of n-butanol:acetic acid:water (5:2:3, v/v/v). After development of the chromatograms, sinapate esters were visualized under UV light. Putative mutants were re-tested by the same procedure in the next generation. Homozygous sng2 progeny were back-crossed to wild-type four times to remove unlinked background mutations.

Sinapate ester analysis

For analysis of seed sinapate ester content, Landsberg erecta and sng2 seeds (10 mg) were extracted in 1.5 ml 50% methanol containing 1.5% v/v acetic acid. For analysis of sinapate ester content throughout silique development, single siliques from 5-week-old plants were extracted beginning with the bottom silique that had just started to turn brown and ending with the first expanding silique. Each silique was extracted in 50 µl 50% methanol containing 1.5% v/v acetic acid. For analysis of germinating seedlings, single seedlings were extracted in 50 µl 50% methanol containing 1.5% v/v acetic acid. In each case, a 20 µl aliquot was analysed by HPLC on a PuresilTM C18 column (Waters) (1200 nm pore size, 5 µm particle size) using a 23 min gradient from 1.5% acetic acid, 0.05% SDS to 30% acetonitrile in 1.5% acetic acid, 0.05% SDS at a flow rate of 1 ml min−1 using UV detection at 335 nm. Sinapate esters were quantified using the extinction coefficient of sinapic acid.

Liquid chromatography/mass spectrometry (LC/MS)

The LC/MS analysis was carried out using a Finnigan MAT LCQ mass spectrometer (Thermoquest Corp. San Jose, CA, USA). A 20 µl aliquot of the sample was injected onto a C18 column (Waters) (1200 nm pore size, 5 µm particle size) (flow rate 0.5 ml min−1) using a 1.5% acetic acid/acetonitrile gradient (starting with 95% of 1.5% acetic acid and ramping to 95% of acetonitrile over 32 min). The sample was ionized using negative ion electrospray ionization (ESI) with the LCQ being scanned to 2000 amu. The source voltage was set at 4 kV and the capillary voltage at 20 V. The background source pressure was 1.5 × 10−5 torr as read by an ion gauge. The drying gas used in this study was nitrogen.

Matrix-assisted laser desorption ionization/mass spectrometry (MALDI/MS)

The MALDI/MS results were obtained using a Voyager mass spectrometer (PerSeptive Biosystems, Framingham, MA, USA). This instrument utilizes a nitrogen laser (337 nm UV laser) for ionization with a time-of-flight mass analyser. The sample and matrix were mixed in a ratio of 1 µl to 1 µl on the sample plate. This mixture was allowed to air dry prior to analysis. The matrix used for this sample was α-cyano-4-hydroxy cinnamic acid.

Choline determination

Choline was extracted from Landsberg erecta wild-type and sng2 seeds and analysed by plasma desorption/MS as described previously (Yang et al., 1995).

Enzyme assays

Enzyme extraction and assay conditions were based upon those used for purification and assay of SCT from B. napus (Vogt et al., 1993). For assay from crude seed extracts, Landsberg erecta or sng2 seeds were frozen in liquid nitrogen and ground to a fine powder. The powder was stirred for 20 min in 5 vol 100 mm potassium phosphate buffer (pH 6.8) containing 20 mm sodium chloride and 4% w/v insoluble polyvinylpolypyrrolidone. The samples were filtered through Miracloth (Calbiochem, La Jolla, CA, USA) and centrifuged for 20 min at 13 000 g. The supernatant was made to 0.1% w/v protamine sulphate, stirred for 20 min, and centrifuged for 20 min at 13 000 g. The supernatant was again filtered through Miracloth, and the protein was precipitated by adding ammonium sulphate to 85% saturation, followed by centrifugation for 20 min at 13 000 g. The pellet was resuspended in 100 mm potassium phosphate buffer (pH 7.0) containing 50 mm sodium chloride, desalted using PD-10 Sephadex® G-25M Columns (Supelco, Bellefonte, PA, USA) into 100 mm potassium phosphate buffer (pH 7.0) and used for determination of SCT activity. Each assay contained 40 µl 2.5 mm sinapoylglucose, 8 µl 100 mm choline chloride, and 32 µl protein extract. The assays were incubated for 60 min at 30°C, stopped by the addition of 420 µl cold 50% methanol, and analysed by HPLC. Sinapoylglucose for use in enzyme assays was purified from the sng1 mutant of Arabidopsis (Lorenzen et al., 1996). Protein content was determined by the bicinchonic acid assay procedure (Pierce) with BSA as a standard. For PMSF inhibition assays, 1 µl 0.1 m PMSF in ethanol was added to protein extracts and the reaction was pre-incubated for 30 min at 30°C before addition of substrates.

For assay of SCT activity from heterologously expressed, denatured, and refolded SNG2 in E. coli, each assay contained 75 µl E. coli extract, 5 µl 100 mm choline chloride and 20 µl 2.5 mm sinapoylglucose. The assays were incubated at 30°C for 4 h, stopped by the addition of 75 µl 100% methanol, and analysed by HPLC.

Analysis of nucleic acids

For DNA gel blot analysis, BAC DNA, isolated using a Plasmid Midi Kit for Purification of BAC DNA (Qiagen, Santa Clarita, CA, USA), and DNA extracted from plant material (Doyle and Doyle, 1990) were digested with restriction endonucleases, electrophoretically separated, transferred to Hybond N+ membrane (Amersham) and hybridized according to standard protocols (Sambrook et al., 1989) with DNA probes made using the DECAprime II system (Ambion, Austin, TX, USA). RNA was extracted from tissues (Goldsbrough and Cullis, 1981), electrophoretically separated, transferred to Hybond N+ membrane, and hybridized with radiolabelled probes prepared as described above from expressed sequence tag 309H12T7 (Genbank accession number AA394342). Hybridization to target RNA was quantified using a Typhoon 8600 Variable Mode Imager (Molecular Dynamics, Sunnyvale, CA, USA). Sequencing was performed using a thermosequenase flourescent-labelled primer cycle sequencing kit (Amersham, Piscataway, NJ, USA) with the IRD-800 M13forward and IRD-700 M13reverse primers (LICOR, Lincoln, NB, USA). The reaction products were analysed with the LongReadIR DNA 4200 automated DNA sequencer (LICOR).

Molecular cloning of the SNG2 gene

A mapping population of 228 F2 plants was generated by self-pollination of F1 plants derived from a sng2/sng2× Columbia wild-type cross. The genotype at the SNG2 locus for all individuals in the population was determined by TLC analysis of embryos from the F3 generation derived from self-pollination of the F2 generation. To roughly map the SNG2 locus, 19 of the sng2/sng2 lines were used for analysis with the Arabidopsis RFLP marker set (Fabri and Schäffner, 1994) obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio State University). Subsequent mapping was performed using the simple sequence length polymorphism markers nga249 and nga151 (Bell and Ecker, 1994).

RT–PCR of the sng2 allele

The cDNA for the Landsberg erecta allele of F17I14.170 was amplified from total silique RNA using the Access RT–PCR System (Promega, Madison, WI, USA). Primers designed from the AGI annotated cDNA for F17I14.170 were used for the reaction. The forward primer (5′-CACTAAGAAGAAGCAGAG-3′) was designed 21 bp upstream of the predicted start codon, and the reverse primer (5′-CACAGAATAGCATGTGGC-3′) was designed 45 bp downstream of the predicted stop codon. The RT–PCR product was subcloned into the pGEM-T Easy vector using the pGEM-T Easy Vector System (Promega).

Complementation of the sng2 mutant

Standard techniques were used for DNA manipulations (Sambrook et al., 1989). For construction of pBI101-SNG2, an 8.2 kb SacII fragment of BAC F17I14 containing F17I14.170 was subcloned into the SacII site of pBS KS. The pBI101-SNG2 construct was generated by cloning the 4.0 kb AvrII/SalI fragment of the subcloned BAC fragment into the XbaI/SalI sites of pBI101 (Clontech, Palo Alto, CA, USA). Plant transformation was performed by floral dip (Clough and Bent, 1998), and transformations with the pBI101-empty vector were included as controls. Transformed seedlings (T1) were identified by selection on MS medium containing 50 mg l−1 kanamycin and 200 mg l−1 timentin and were transferred to soil. The sinapoylcholine content of T2 embryos and seeds was determined by TLC and HPLC as previously described.

Constructs for expression of SNG2 in E. coli

Two oligonucleotides were designed to amplify a truncated fragment of the SNG2 cDNA encoding a protein lacking the predicted SNG2 signal peptide. The N-terminal oligonucleotide (5′-CCATGGCTTTGCTAGTGAAGTCTC-3′) incorporated a start codon and the restriction site NcoI (CCATGG) and altered the putative N-terminal serine of the mature protein to an alanine. The C-terminal oligonucleotide (5′-GTCGACTTAGAGAGGTTCTCCATC-3′) incorporated a SalI restriction site after the stop codon. The SNG2 gene was amplified by PCR, subcloned into EcoRV-digested pBS KS, and sequenced. The modified coding sequence was excised by NcoI/SalI digestion and cloned into the NcoI/SalI-digested pET28A vector (Novagen, Madison, WI, USA) to yield pET28A-SNG2. For analysis of SNG2 expression and activity, the E. coli host BL21-CodonPlusTM DE3)-RIL (Stratagene, La Jolla, CA, USA) was transformed with the empty pET28A vector and pET28A-SNG2.

E.coli growth conditions and preparation of E. coli extracts

For heterologous expression of SNG2, an overnight culture of bacteria grown at 37°C was diluted 200-fold into fresh Luria Bertani (LB) medium and grown at 37°C to an OD600 of 0.6. Cells were subsequently induced with 1 mm isopropyl-β-d-thiogalactopyranoside (IPTG) and grown for 6 h at 30°C. Cells were harvested and lysed in 20 ml of 20 mm Tris–HCl, pH 8.0, and 500 mm NaCl using a French press. The cell lysate was cleared by centrifugation at 14 000 g at 4°C for 30 min. Supernatant (soluble protein fraction) and pellet (insoluble protein fraction) were analysed by SDS–PAGE. For heterologous expression of SNG2 for enzyme assays, an overnight culture of bacteria grown at 37°C was diluted 200-fold into 500 ml of fresh LB medium and grown at 37°C until the OD600 of the culture was between 0.4 and 0.6. Cells were subsequently induced with 1 mm IPTG and grown for 3 h at 37°C. The cells were pelleted at 5000 g for 10 min, resuspended in 25 ml lysis buffer (25 mm Tris-acetate (pH 7.5), 1 mm EDTA, 0.1% Triton X-100, 0.1 mg ml−1 lysozyme, 0.01 mg ml−1 RNAase A, 0.05 mg ml−1 DNAase I and 2 mm magnesium chloride) and incubated for 10 min on ice. The insoluble fraction was pelleted at 15 000 g for 10 min and washed three times in 10 ml wash buffer 1 (50 mm TrisHCl (pH 7.7), 0.3 m sodium chloride, 1 mm EDTA, 0.1% Triton X-100). The pellet was washed with 5 ml wash buffer 2 (wash buffer 1 + 5 mm DTT) and finally resuspended in 1.5 ml wash buffer 2 containing 5% glycerol. The protein content of the sample was quantified on an SDS–PAGE gel using serial dilutions of the inclusion body suspension, after which the sample was stored at −80°C. For renaturation of SNG2 protein, isolated E. coli inclusion bodies (approximately 1.2 mg total protein) were resuspended in 0.64 ml 100 mm Tris–HCl (pH 8) containing 8 m urea, 1 mm EDTA and 20 mm DTT. The sample was incubated at room temperature for 2 h with occasional vortex mixing. The solubilized inclusion bodies were then diluted fivefold with 100 mm Tris–HCl (pH 8) containing 8 m urea and 1 mm EDTA to a final protein concentration of approximately 0.4 mg ml−1. Following denaturation, protein folding was initiated using the so-called rapid dilution technique (Rudolph and Lilie, 1996). Eight aliquots (0.4 ml each) of denatured inclusion body protein were slowly added at 15 min intervals to 100 ml of vigorously stirred 100 mm Tris–HCl (pH 8) containing 0.2 mm EDTA, 15% v/v glycerol, 0.01% v/v Tween-20, 3 mm reduced glutathione and 0.6 mm oxidized glutathione at room temperature. The mixture was then incubated for 16 h at room temperature without stirring. Active recombinant SCT was partially purified by anion exchange chromatography by application of the entire refolding mixture to a column containing 2 ml settled bed volume of QAE Sephadex A-50 (Sigma, St Louis, MO, USA) that had been equilibrated with 50 mm Tris–HCl (pH 7.5). The resin was then washed with 3 ml 50 mm Tris–HCl (pH 7.5) and the eluent discarded. The SCT protein was eluted from the column with two 3 ml washes of 50 mm Tris–HCl (pH 7.5) containing 100 mm NaCl.


We thank Dr David Rhodes for his assistance in the choline measurement experiments and Joanne Cusumano for her preparation of the RNA gel blot. This work was supported by grants from the Division of Energy Biosciences, United States Department of Energy, and the National Science Foundation to C.C., and graduate fellowships from the United States Department of Agriculture and Purdue University to A.M.S. This is journal paper number 16609 from the Purdue University Agricultural Experiment Station.

GenBank accession number AY033947 (SNG2 cDNA).