The Plant Journal (2000) 21(3), 239±248

Functional rescue of a bacterial dapA auxotroph with a plant cDNA library selects for mutant clones encoding a feedback-insensitive dihydrodipicolinate synthase Marc Vauterin*, ValeÂrie Frankard and Michel Jacobs Laboratorium voor Plantengenetica, Instituut voor Moleculaire Biologie, Vrije Universiteit Brussel, Paardenstraat 65, Sint Genesius Rode B-1640, Belgium Received 8 October 1999; accepted 9 December 1999. *For correspondence (fax +32 2 3590399; e-mail [email protected]).

Summary Dihydrodipicolinate synthase (DHDPS; EC4.2.1.52) catalyses the ®rst reaction of lysine biosynthesis in plants and bacteria. Plant DHDPS enzymes are strongly inhibited by lysine (I0.5 » 10 mM), whereas the bacterial enzymes are less (50-fold) or insensitive to lysine inhibition. We found that plant dhdps sequences expressing lysine-sensitive DHDPS enzymes are unable to complement a bacterial auxotroph, although a functional plant DHDPS enzyme is formed. As a consequence of this, plant dhdps cDNA clones which have been isolated through functional complementation using the DHDPS-de®cient Escherichia coli strain encode mutated DHDPS enzymes impaired in lysine inhibition. The experiments outlined in this article emphasize that heterologous complementation can select for mutant clones when altered protein properties are requisite for functional rescue. In addition, the mutants rescued by heterologous complementation revealed a new critical amino acid substitution which renders lysine insensitivity to the plant DHDPS enzyme. An interpretation is given for the impaired inhibition mechanism of the mutant DHDPS enzyme by integrating the identi®ed amino acid substitution in the DHDPS protein structure.

Introduction In nature, two different pathways of lysine biosynthesis exist (LeÂJohn, 1971): the a-aminoadipic acid (aAA) pathway and the diaminopimelic acid (DAP) pathway. The aAA pathway derives lysine from the intermediate a-aminoadipic acid and functions in higher fungi, some phycomycetes and euglenids. The DAP pathway produces lysine through synthesis of the intermediate a,e-diaminopimelic acid and operates in bacteria, a few fungi (cellulosecontaining) and higher plants. In both plants and bacteria, the DAP biosynthetic pathway operates in an identical way, in a sequence of seven catalytic reactions generating the same intermediate products (Figure 1). Nevertheless, there are important differences in regulation of particular enzymatic steps. In plants, the DAP biosynthetic pathway contributes solely to the generation of lysine, whereas in bacteria a dual function is assigned to this pathway: lysine synthesis and production of meso-2,6-diaminopimelic acid (mDAP) and dipicolinate (DPA), respectively, for cell wall synthesis and for sporulation. ã 2000 Blackwell Science Ltd

The ®rst enzyme of the DAP pathway, dihydrodipicolinate synthase, catalyses the condensation of pyruvate and aspartate-b-semi-aldehyde into 4-hydroxy-2,3,4,5-tetrahydrodipicolinate, the ®rst intermediate of the pathway. Control of this condensation step is important because the reaction determines the ¯ux of aspartate towards DAP and/or lysine synthesis. For this reason, most organisms using the DAP pathway for lysine synthesis possess a dihydrodipicolinate synthase which is feedback-inhibited by the end-product lysine. Depending on the needs for pathway intermediates, different classes of DHDPS enzymes can be distinguished among groups of organisms. Bacteria ensure a suf®cient production of mDAP and DPA through moderated (e.g. Escherichia coli, I0.5 = 0.4±1.0 mM lysine; Laber et al., 1992; this article) or no inhibition (e.g. Corynebacterium glutamicum; Cremer et al., 1988) of lysine on dihydrodipicolinate synthase. To limit conversion of mDAP into lysine in bacteria, control of the last step of the DAP pathway is through lysine inhibition of the 239

240 Marc Vauterin et al.

Figure 1. Biochemical scheme of the DAP pathway functional in plants and bacteria.

catalysing enzyme, diaminopimelate decarboxylase, and by lysine repression of the encoding lysA gene, a mechanism mediated by the transcriptional factor, lysR. In plants, only one control point in the DAP biosynthetic pathway has been found so far: a strong feedback inhibition of lysine on dihydrodipicolinate synthase (I0.5 = 10 mM; Dereppe et al., 1992; Frisch et al., 1991; Ghislain et al., 1990; Kumpaisal et al., 1987; Mazelis et al., 1977; Wallsgrove and Mazelis, 1981). In plants, the equimolar ¯ux of metabolites through the subsequent catalytic steps of the pathway allows a single control at the ®rst enzymatic reaction. The strong inhibition by free lysine is an important control on lysine biosynthesis in plants. Indeed, the generation of a Nicotiana sylvestris mutant plant expressing a lysine feedback-insensitive DHDPS enzyme (Negrutiu et al., 1984) showed that impaired inhibition of the DHDPS enzyme severely affects regulation of the lysine biosynthesis pathway, resulting in a strong lysine overproduction (Frankard et al., 1992). Consequently, there is a strong interest in feedback-insensitive plant DHDPS enzymes for manipulating lysine biosynthesis in plants (Falco et al., 1995; Karchi et al., 1993; Perl et al., 1992; Shaul and Galili, 1992, 1993) The difference in inhibition properties between plant and bacterial DHDPS enzymes has consequences for expressing plant cDNAs encoding DHDPS enzymes in a bacterial DHDPS auxotrophic host. In this article, we show that wildtype plant dhdps sequences encoding lysine-sensitive DHDPS enzymes are unable to complement a bacterial auxotroph, although a functional protein is formed. The plant dhdps cDNA clones which have been isolated through functional complementation using a DHDPSde®cient bacterial strain, encode mutated DHDPS enzymes

which are totally insensitive or at least less sensitive to lysine inhibition. The observed phenomenon can be interpreted by comparing the different metabolic functions of the DAP pathway in plants versus bacteria. In this study, we show for the ®rst time that heterologous complementation can select for a mutant transgene when the modi®ed properties of the encoded protein generate a functional enzyme rescuing the host organism, whereas a wild-type transgene does not. This report suggests some caution should be taken when critically evaluating conclusions concerning properties and kinetics of proteins expressed in functionally rescued heterologous systems. This article also describes a new mutation rendering the plant DHDPS enzyme lysine-insensitive. Several mutant plant DHDPS enzymes insensitive to lysine inhibition have been reported previously (Ghislain et al., 1995; Shaver et al., 1996), and all mutations are clustered in a small region of 10 amino acids. This region has been proposed to be the allosteric site of the DHDPS enzyme (Shaver et al., 1996). Our new mutation rendering the plant DHDPS enzymes lysine-insensitive is situated outside this region, indicating that locating the allosteric site is more complex. Based on the available Escherichia coli DHDPS protein structure (Blickling et al., 1997b), an explanation is suggested for the role of the identi®ed amino acid substitution in lysine inhibition. Results and Discussion Isolation of mutated Arabidopsis thaliana cDNAs encoding lysine-insensitive DHDPS enzymes An Arabidopsis thaliana dhdps cDNA was previously isolated by screening an Arabidopsis cDNA library using ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

Selection of mutant cDNAs through functional rescue

Figure 2. Inhibition patterns of different plant DHDPS enzymes expressed in the dapA± auxotrophic Escherichia coli strain AT997. Curves show the percentage inhibition of the DHDPS enzymes measured at different concentrations of lysine added to the enzyme assays. (a±d) Inhibition kinetics of sensitive plant DHDPS enzymes; (e) inhibition kinetics of insensitive plant DHDPS enzymes and the Escherichia coli DHDPS enzyme.

the heterologous poplar dhdps cDNA as probe (Vauterin and Jacobs, 1994). The ®rst indication that wild-type plant DHDPS-encoding sequences are unable to rescue a bacterial DHDPS auxotroph was obtained when the construct AT-P419 expressing the wild-type Arabidopsis thaliana DHDPS apoprotein was transformed into the E. coli AT997 dapA± strain. Although this construct expressed a functional Arabidopsis DHDPS protein, as shown by enzyme assays (Figure 2), this clone was unable to rescue the AT997 strain. Transformed cells did not grow well in LB or minimal M9 medium if no DAP was supplemented. Therefore, EMS mutagenesis experiments were carried out with cell cultures of AT-P419 grown in DAP-enriched LB medium and selected for ampicillin resistance. Previous research has shown that organisms expressing a lysine feedback-insensitive DHDPS enzyme can be selected on medium containing toxic concentrations of S-2-aminoethylcysteine (AEC), a lysine analogue. Bacterial strains (Sano and Shiio, 1970; Shaver et al., 1996) and plants (Negrutiu et al., 1984) have been selected by this approach and AEC resistance was shown to be the result of lysine accumulation due to a mutated DHDPS enzyme which was insensitive for lysine inhibition. The toxicity of AEC is believed to be caused by its substitution for lysine in protein synthesis (Bright et al., 1979). The overproduction of lysine probably dilutes the competing AEC in formation ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

241

of loaded lysyl-tRNAs, which reduces synthesis of nonfunctional proteins. In our experiments, EMS-mutagenized cells were selected on minimal M9 medium for resistance against 5 mM AEC. Of the resistant colonies selected, about half showed no altered properties for lysine inhibition of the DHDPS enzyme. These colonies were probably uptake mutants. Seventeen colonies expressing lysine-insensitive DHDPS enzymes were stored for further analysis. One of these clones, AT-6EMS, revealed a single nucleotide transition (T=A ® C º G) at the ®rst position of codon 53 (numbering according the E. coli DHDPS sequence), resulting in a W53 (TGG) to R53 (CGG) substitution. EMS mutagenesis mainly causes G º C ® A=T and T=A ® C º G transitions. In this case, a thymine in the non-transcribed strand had been mutated to a cytosine. The E. coli AT997 dapA± auxotroph was transformed with the vector constructs AT-P419 and AT-6EMS to express, respectively, the wild-type feedback-sensitive and mutant insensitive Arabidopsis DHDPS apoproteins. Enzyme kinetics for lysine inhibition have been determined for these enzymes and the results are shown in Figure 2. The wild-type enzyme (Figure 2b) was 50% inhibited at 3±4 mM lysine and fully inhibited at 50 mM lysine. The mutated Arabidopsis DHDPS enzyme (Figure 2e) was totally insensitive for lysine inhibition and 98% of activity remained in the presence of 10 mM lysine. These results show that the characterized clone, AT-6EMS, encodes a functional Arabidopsis thaliana DHDPS apoprotein which is totally insensitive for lysine feedback inhibition. The newly de®ned mutation changing tryptophan into arginine at position 53, proves that the allosteric inhibition site of the DHDPS protein is not restricted to the 10 amino acid region proposed by Shaver et al. (1996). Populus deltoides 3 Populus trichocarpa lysine-sensitive and insensitive DHDPS enzymes Substitution of W53 by R53 in Arabidopsis DHDPS leads to a total loss of feedback inhibition. Indeed, all known plant dhdps apoprotein sequences have a conserved tryptophan residue at this position (Figure 3). However, one exception is the poplar dhdps sequence where the tryptophan is replaced by an arginine residue. As a tryptophan to arginine substitution at this position leads to a loss of lysine inhibition in the Arabidopsis DHDPS protein, one would expect the encoded poplar DHDPS protein to be lysine-insensitive. Inhibition tests indicated that the poplar dhdps cDNA, which had been isolated through functional complementation in the E. coli AT997 dapA± strain (Vauterin and Jacobs, 1994), encodes an insensitive DHDPS enzyme (Figure 2e). This was surprising as no evidence had been obtained previously for the existence of plant lysine-insensitive DHDPS enzymes. To test whether poplar expresses a natural feedback-insensitive DHDPS

242 Marc Vauterin et al.

Figure 3. Amino acid sequence alignment of different DHDPS proteins. Residues identi®ed (Blickling et al., 1997a) as being important for functioning of the catalytic site or inhibition site are indicated above the alignment. Mutations rendering insensitivity to the plant DHDPS enzymes are marked below the alignment: (1) this article; (2) Ghislain et al. (1995); (3) Shaver et al. (1996). Numbering follows the Escherichia coli DHDPS protein sequence. Sequences from plant DHDPS proteins start at the apoprotein. Aligned sequences are from Bacillus subtilis, Escherichia coli, Arabidopsis thaliana, Populus deltoides 3 trichocarpa, Nicotiana tomentosiformis, Nicotiana sylvestris, Zea mays and Triticum aestivum (one of the two isozymes).

enzyme (or an isozyme), extracts were made from leaves and bark of poplar (P. deltoides 3 P. trichocarpa). No fraction of DHDPS activity could be detected that was not inhibited by low lysine concentrations. Poplar dhdps sequences were isolated from genomic DNA (P. deltoides 3 P. trichocarpa) through PCR ampli®cation. Sequencing of these fragments revealed that the wildtype poplar dhdps sequence encoded a tryptophan at position 53. The wild-type poplar dhdps sequence includes TGG for tryptophan at amino acid position 53, whereas the isolated cDNA includes AGG for arginine. Apart from a T ® A substitution in the non-transcribed strand of the poplar dhdps cDNA, no other differences with the genomic dhdps sequence were detected. Constructs were made expressing the lysine-sensitive (PDT-1G14, genomic PCR) and insensitive (PDT-4C4, cDNA PCR) poplar DHDPS apoproteins. The E. coli AT997 dapA± auxotroph was transformed with these constructs to express, respectively, the wild-type and mutant poplar DHDPS apoproteins. Activity of the wild-type poplar DHDPS enzyme was controlled by lysine, with 50% inhibition at 6 mM lysine and total inhibition at 50 mM lysine (Figure 2a). The mutant

poplar DHDPS apoprotein retained 98% of its activity at up to 1 mM lysine. The enzyme was weakly inhibited at higher lysine concentrations: 90% of activity remained at 5 mM lysine and 85% at 10 mM lysine (Figure 2e). Enzyme assays have been carried out for the original E. coli AT997 dapA± auxotrophic strain complemented with the poplar dhdps full-length cDNA, as isolated through functional complementation. This poplar DHDPS enzyme displayed exactly the same kinetics as the enzyme expressed from the PDT4C4 clone (Figure 2e). By cloning a poplar (P. deltoides 3 P. trichocarpa) dhdps genomic sequence, we showed that the W53 residue was also conserved in the poplar DHDPS protein. Furthermore, inhibition tests on DHDPS protein extracted from leaves and bark of poplar indicated that the poplar DHDPS enzyme displays a normal inhibition pattern with an I0.5 at 6 mM lysine. As expected, poplar DHDPS encoded by the genomic sequence (PDT-1G14) was expressed in AT997 as an inhibition-sensitive enzyme (Figure 2a). These results prove that the wild-type poplar dhdps gene encodes an inhibition-sensitive DHDPS enzyme, whereas the functional complementation technique had selected for poplar ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

Selection of mutant cDNAs through functional rescue

243

Table 1. The constructs used to express the different plant DHDPS enzymes in a dapA± auxotrophic Escherichia coli strain AT997. The relation of the dhdps gene mutations and DHDPS enzyme-insensitivity is indicated Species Arabidopsis thaliana Populus deltoides 3 Populus trichocarpa Zea mays Nicotiana tomentosiformis Nicotiana sylvestris RAEC-1

Clone

Amino acid change

Mutagen

Inhibition

AT-P419 AT-6EMS PDT-1G14 PDT-4C4 cPDT ZM-XS cNT NST-RAEC1

Wild-type W53 (TGG) Ð R53 (CGG) Wild-type W53 (TGG) Ð R53 (AGG) W53 (TGG) Ð R53 (AGG) Wild-type Wild-type N80 (AAC) Ð I80 (ATT)

± EMS ± ? ? ± ± UV

+ Ð + Ð Ð + (Ð) + Ð

dhdps cDNA clones encoding feedback-insensitive DHDPS enzymes.

Functional rescue of a DHDPS-de®cient Escherichia coli strain is only achieved with plant dhdps sequences encoding lysine-insensitive DHDPS enzymes

Nicotiana sp. lysine-sensitive and insensitive DHDPS enzymes

To test this hypothesis, vectors expressing sensitive and corresponding insensitive DHDPS proteins of Arabidopsis thaliana (AT-P419 and AT-6EMS, respectively), of Populus deltoides 3 Populus trichocarpa (PDT-1G14 and PDT-4C4, respectively) and of Nicotiana sp. (cNS and NST-RAEC1, respectively) were constructed (Table 1 and Figure 4). Functional complementation assays in the dapA± AT997 strain showed that only the constructs expressing feedback-insensitive DHDPS enzymes (Arabidopsis AT-6EMS, Populus PDT-4C4, Nicotiana NST-RAEC1, Populus cPDT) rescued the strain on minimal medium without DAP (Figure 5). Expression of inhibition-sensitive DHDPS enzymes in the AT997 strain did not lead to functional rescue of the auxotroph (Figure 5), although DHDPS activity tests showed that these strains produced an active DHDPS protein (Figure 2a,b,d). Growth of these strains (Arabidopsis AT-P419, Populus PDT-1G14, Nicotiana cNT) was obtained only if 100 mM DAP was supplemented to the medium. The phenotype of these strains is thus comparable to the pUC18-transformed AT997 strain. Clones expressing insensitive plant DHDPS enzymes were, in addition to their prototrophy, resistant to 2.5 mM AEC. The lack of functionality of lysine-sensitive plant DHDPS enzymes in bacteria is explained by differences between plants and bacteria in the metabolic functioning of the DAP pathway. mDAP and DPA are required for bacterial cell wall synthesis and for sporulation, whereas in plants, intermediates are fully converted to lysine. Therefore, additional regulation of gene expression and/or enzyme activity is required in bacteria to establish the balance of mDAP, DPA and lysine production. In bacteria, extra control mechanisms are present at the level of the dapB gene, encoding dihydrodipicolinate reductase (DHDPR, reaction 2 in Figure 1), and the lysA gene, encoding diaminopimelate decarboxylase (reaction 7). In E. coli, expression of the dapB gene is repressed by lysine and induced by diaminopimelate. Guanosine

Expression of the coding sequence (cNT) of the Nicotiana tomentosiformis DHDPS protein in the E. coli AT997 dapA± auxotroph produced a functional DHDPS protein which showed 50% inhibition at 6±7 mM lysine. The enzyme was fully inhibited at 50 mM lysine (Figure 2d). A chimeric coding sequence (NST-RAEC1) was generated by fusion of the N. tomentosiformis cDNA wild-type and N. sylvestris raec-1 mutant sequences. This construct was transformed into the E. coli AT997 dapA± auxotroph to express a functional Nicotiana sp. DHDPS protein. Insensitivity is caused by an N80 to I80 substitution (Ghislain et al., 1995). At 10 mM lysine, 98% of the enzyme activity was still measured. Zea mays DHDPS enzyme Vector construct ZM-XS transformed into the E. coli AT997 dapA± auxotroph expressed a functional maize DHDPS apoprotein. The enzyme displayed a normal inhibition pattern with 50% inhibition at 6±7 mM lysine, except that 10% activity of the enzyme always remained, even at concentrations of 10 mM lysine (Figure 2c). DHDPS enzyme activity of the E. coli wild-type XL-Blue-1 and mutant AT997 strains The E. coli AT997 mutant was generated by Bukhari and Taylor (1971) using N-methyl-N¢-nitro-N-nitrosoguanidine, and has been characterized as being a dapA± mutant (Yeh et al., 1988). The strain is unable to grow on minimal medium and DAP is required to restore growth. Reversion of the mutant has not been observed. Protein extracts of an AT997 culture did not show any DHDPS activity. The DHDPS enzyme of the wild-type E. coli XL-Blue-1 strain was 50% inhibited by 360 mM lysine. The enzyme still showed 25% activity at 1 mM lysine (Figure 2e). ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

244 Marc Vauterin et al.

Figure 5. The DHDPS-de®cient Escherichia coli strain AT997 complemented with different plant DHDPS-encoding sequences. (1) Arabidopsis thaliana DHDPS apoprotein, EMS mutant; (2) Arabidopsis thaliana DHDPS apoprotein, wild-type sequence; (3) Populus deltoides 3 P. trichocarpa DHDPS apoprotein, cDNA-derived; (4) Populus deltoides 3 P. trichocarpa DHDPS apoprotein, derived from genomic sequence; (5) pUC18 vector; (6) Zea mays DHDPS apoprotein; (7) Nicotiana tomentosiformis DHDPS apoprotein, cDNA-derived; (8) Nicotiana sylvestris RAEC-1 DHDPS apoprotein; (9) Populus deltoides 3 P. trichocarpa full DHDPS protein expressed from cDNA isolated by functional complementation. Selection is on minimal medium (100 mg ml±1 ampicillin) with addition of 100 mM DAP (left column), no additions (middle column) or 2.5 mM AEC (right column).

Figure 4. Plasmid construction strategies for expression of plant DHDPS enzymes in a bacterial host. (a) Populus deltoides 3 Populus trichocarpa DHDPS expression cassettes; cPDT, Populus cDNA sequence encoding a DHDPS enzyme with transit peptide; PDT-4C4, Populus DHDPS apoprotein derived from cDNA; PDT1G14, Populus DHDPS apoprotein derived from genomic DNA. (b) Arabidopsis thaliana DHDPS expression cassettes: cAT, Arabidopsis cDNA sequence encoding a DHDPS enzyme with transit peptide; ATP419, Arabidopsis DHDPS apoprotein encoding sequence derived from cDNA; AT-6EMS, Arabidopsis DHDPS apoprotein encoding sequence mutagenized with EMS. (c) Nicotiana sp. DHDPS expression cassettes: cNT, Nicotiana tomentosiformis cDNA encoding a DHDPS enzyme with transit peptide; NS-RAEC1, Nicotiana sylvestris partial genomic sequence encoding an insensitive DHDPS; NST-RAEC1, Nicotiana sylvestris/tomentosiformis hybrid sequence encoding an insensitive DHDPS protein with transit peptide. (d) Zea mays DHDPS expression cassette: cZM Zea mays cDNA encoding a DHDPS enzyme with transit peptide; ZM-XS, Zea mays DHDPS apoprotein encoding sequence derived from cDNA.

tetraphosphate also regulates dapB gene expression. No control of the enzyme is present (Bouvier et al., 1984; Patte, 1983). In addition, in Bacillus, the DHDPR enzyme is strongly inhibited by 2,6-dipicolinate (DPA). By inhibiting

DHDPR, DPA forces DPA synthesis (spore) and blocks mDAP synthesis (cell wall, see Figure 1). Possibly even in Bacillus there is stimulation by DPA of the transcriptional activity of the dapA gene through activation of the RNA polymerase (Hogansen and Stahly, 1975a; Hogansen et al., 1975). The lysA gene, both in E. coli and Bacillus, is induced by diaminopimelate and repressed by lysine. This is in agreement with the double metabolic role of the enzyme: anabolic for lysine biosynthesis and catabolic for diaminopimelate degradation. The lysA gene is induced by guanosine tetraphosphate (Williams and Rogers, 1987) and is regulated through a transcriptional activator encoded by the lysR gene which controls its own transcription. The regulation of the lysine biosynthetic branch should be considered separately for Gram-positive and Gramnegative bacteria. Gram-positive bacteria have several layers of peptidoglycan in their cell wall, Gram-negative bacteria have only one. To ful®l their greater need for mDAP, Gram-positive bacteria have, in contrast to Gramnegative bacteria, an insensitive DHDPS enzyme and, in addition, an aspartate kinase with no inhibition characteristics to ensure at all times that suf®cient aspartate semialdehyde is available as substrate for mDAP synthesis. By contrast, E. coli possesses a lysine- and threonine-sensitive aspartate kinase. Furthermore, diaminopimelate decarboxylase of Gram-positive bacteria is, in addition to the lysA gene regulation, also inhibited by lysine. This extra control again prevents conversion of mDAP into lysine. ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

Selection of mutant cDNAs through functional rescue By the moderated or absence of feedback control by lysine on the DHDPS enzyme and the additional control points, as mentioned above, bacteria ensure suf®cient ¯ux of metabolite through the DAP pathway for synthesis of mDAP and DPA. Expression of a highly sensitive plant DHDPS enzyme in a DHDPS-de®cient bacterium does not allow the bacterium to establish adequate production of mDAP in relation to lysine production. It is therefore likely that a lack of mDAP limits growth of the AT997 strain rather than a shortage of lysine. This phenomenon explains why the functional complementation technique used to clone a poplar dhdps cDNA in the AT997 strain resulted in selection of a clone encoding a feedbackinsensitive DHDPS enzyme. The mutation might be spontaneous (developing microcolonies prior to the action of selection) or due to a reverse transcriptase error (about 1 in 500 bp mis-incorporation at high dNTP and Mn2+ concentrations, Maniatis et al., 1982). It should be noted that the maize dhdps construct (ZMXS) which encodes a wild-type DHDPS apoprotein, does rescue the AT997 strain and even allows growth of the complemented bacterium on 2.5 mM AEC-containing medium (Figure 5). According to our hypothesis, this result suggests that a DHDPS enzyme is expressed which is less sensitive than the wild-type maize DHDPS enzyme. Remarkably, our inhibition studies indicated that the maize DHDPS protein investigated showed a leakage of the inhibition mechanism. About 10% enzyme activity remained uninhibited, even at high lysine concentrations

Figure 6. Subunit A of the Escherichia coli DHDPS protein complexed with or without lysine. The inhibitory lysine is shown at the allosteric site. Amino acid residues moving upon lysine binding are depicted twice: grey shaded residues represent positions in the absence of lysine, white residues indicate positions in the presence of the inhibitory lysine. Asn D80 is an asparagine of subunit D, other indicated residues are from subunit A. Upon lysine binding, Glu84 makes a salt bridge with the inhibitory lysine of subunit D. The plot was created using MOLSCRIPT (Kraulis, 1991).

ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

245

(Figure 2c). Indeed, the maize DHDPS enzyme expressed in the E. coli auxotrophic strain has been reported to display 90% activity at 50 mM lysine, and 18% activity at 200 mM (Shaver et al., 1996). In this respect, the monocotyledonous DHDPS enzyme might differ from the dicotyledonous DHDPS enzyme. W53 to R53 substitution affects the allosteric inhibition mechanism The allosteric inhibition mechanism of the E. coli DHDPS enzyme has been described by Blickling et al. (1997a). From their study, it was concluded that the carboxylate group and the a-amino group of the inhibitory lysine are co-ordinated through A49 and Y106 of subunit A (Figure 6), and through E84 (not shown) and N80 of subunit D. The eamino group of the inhibitory lysine is co-ordinated by H53 and H56, and by G78 through a bridging water molecule. Upon binding of the inhibitory lysine, H53 and H56 of the a2 helix of subunit A change conformation and the a2 helix moves towards the inhibitor. This conformational change results in a repositioning of E84 (subunit A), which allows it to interact with the inhibitory lysine molecule bound at the allosteric site of monomer D. Two sites are now thought to play an interactive role between the inhibitory site and the active site. Firstly, Y106 and Y107 (subunit A) form a hydrophobic stack with the corresponding residues of subunit D. The inhibitory lysine interacts with Y106 and thus in¯uences Y107, which is connected via T44 with Y133, an important residue involved in the Schiff base formation of the catalytic reaction. Secondly, R138 co-ordinates the position of L-ASA (L-aspartate-bsemialdehyde). The inhibitory lysine connects with R138 via N80 and Y107, thus preventing ¯exibility of R138 and correct positioning of L-ASA in the reaction. From above structural model (Blickling et al., 1997a), it appears that the conformational changes at H53 upon lysine binding are a primary event in the signal transduction from the allosteric site towards the active site. The importance of H53 in positioning helix a2, and its relation to the functioning of the allosteric inhibition mechanism, is now emphasized through the mutation of W53 into R53 in the plant DHDPS enzyme. The structure of the Nicotiana sylvestris DHDPS protein (Blickling et al., 1997b) reveals that the W53, conserved in all plant enzymes, has a function in positioning of helix a2, which is comparable to the function of H53 in the E. coli DHDPS protein. Movement of the W53 residue in the plant enzyme is even more signi®cant. Both un-ionized H53 from E. coli and W53 from plants can hydrogen bond the e-NH2 of the inhibitory lysine. Substitution of W53 by R53 will reduce the lysine af®nity drastically through electrostatic repulsion with the e-NH3+. As a consequence, this mutation probably eliminates lysine binding at the inhibitory site.

246 Marc Vauterin et al. Experimental procedures Growth of the bacterial dapA auxotroph The E. coli strain AT997 has been characterized as an auxotroph lacking dihydrodipicolinate synthase activity (Yeh et al., 1988). The strain grows on medium supplemented with diaminopimelate (DAP), as this pathway intermediate can serve as precursor for meso-diaminopimelate and lysine synthesis. Medium for strain maintenance consisted of minimal medium (M9 minimal salts, Gibco BRL) supplemented with 100 mM DAP, 0.5 ml l±1 of 1% (w/v) thiamine hydrochloride, IM 1 ml l±1 of MgSO4´7H2O and 10 ml of 20% (w/v) glucose solution. The antibiotic ampicillin was added at 100 mg ml±1 for plasmid selection. Growth of bacterial cultures for protein extraction was done in Luria broth (Gibco BRL) medium with addition of the same supplements and antibiotics. Functional complementation assays were performed as described by Vauterin and Jacobs (1994).

Construction of poplar DHDPS apoprotein encoding sequences PCR ampli®cations were done on the original Populus deltoides 3 Populus trichocarpa dhdps cDNA clone (cPDT) and on a Populus deltoides 3 Populus trichocarpa genomic DNA library (Universiteit Gent, Belgium). The two primers designed were 5¢-CGG AAT TCG CAT CTT CCG ATG CGC AG-3¢ (pPDT5) located near the start of the apoprotein (bold), and 5¢-GC GAA TTC ATC ATG GCA CTG ATC ATC ATT-3¢ (pPDT3) located downstream of the active site of the protein (Figure 4a). These PCR fragments were cloned into the EcoRI site of pUC18, respecting the orientation of the coding sequence. Completion of the COOH end of the protein was done through insertion of a EcoRV±BamHI restriction fragment derived from the cDNA. PCR ampli®cations were sequenced in order to detect possible ampli®cation errors. Construct PDT-4C4 contains the PCR fragment derived from the dhdps cDNA clone and construct PDT-1G14 is based on the PCR fragment ampli®ed from poplar genomic DNA.

Construction of an Arabidopsis thaliana DHDPS apoprotein encoding sequence The Arabidopsis dhdps cDNA (cAT) insert originally cloned in a pBluescript SK+ vector was released by EcoRI±KpnI digestion and subcloned into the EcoRI±KpnI site of pUC18. A PCR fragment starting at the apoprotein coding sequence (bold) was generated using the primers 5¢-CGG AAT TCG GCT GTT GTA CCT AAC TTC CAT C-3¢ (pAT5) and 5¢-AGG CAC ATT GTA TAT AAT-3¢ (pAT3) located at position 457 of the apoprotein coding sequence (Figure 4b). The N-terminal EcoRI±NcoI cDNA fragment was replaced by this PCR fragment digested with EcoRI±NcoI. This expression vector encoding the Arabidopsis thaliana DHDPS apoprotein (AT-P419) was used for EMS mutagenesis, and generated the mutant clone AT-6EMS.

Construction of Nicotiana sp. DHDPS protein encoding sequences The genomic coding sequence (NS-RAEC1) of the lysine-insensitive DHDPS enzyme has been cloned from the Nicotiana sylvestris RAEC mutant (Ghislain et al., 1995). A wild-type Nicotiana tomentosiformis cDNA encoding a sensitive DHDPS enzyme has

also been cloned in our laboratory. A deletion of the 5¢ non-coding region was performed to obtain an in-frame coding sequence of the Nicotiana tomentosiformis DHDPS enzyme (cNT cloned in the EcoRI±HindIII sites of pBluescript SK+). The BclI±HindIII fragment of the Nicotiana tomentosiformis cDNA has been replaced by the BclI±HindIII genomic Nicotiana sylvestris raec-1 fragment containing the raec-1 mutation, which resulted in the clone NST-RAEC1 (Figure 4c).

Construction of a Zea mays DHDPS apoprotein encoding sequence The Zea mays DHDPS encoding cDNA (cZM) was isolated by Frisch et al. (1991) and was graciously provided by Professor B. Gengenbach (University of Minnesota, USA). The cDNA was partially digested with XbaI (three internal XbaI sites in the cDNA) and fully digested with SpeI (located in the 3¢ non-coding region). The ®rst XbaI site is located six bases downstream of the apoprotein start and allows in-frame cloning into the XbaI site of pUC18. This XbaI±SpeI fragment was isolated and subcloned into the XbaI site of pUC18, giving the clone ZM-XS (Figure 4d).

Ethylmethane sulphonate (EMS) mutagenesis An adaptation of the method of Miller (1972) was used. A 50 ml cell culture of the AT997 bacterial strain transformed with the pUC18 vector expressing the Arabidopsis thaliana DHDPS apoprotein was grown in enriched LB medium and harvested in the logarithmic growth phase through centrifugation (2000 g, 10 min, 4°C). Cells were dissolved in 50 ml M9 minimal medium (Gibco BRL) containing 0.2 M Tris±HCl (pH 7.5). To each 2 ml sample of cell suspension, 30 ml ethyl methanesulphonate (0.14 M ®nal concentration) was added. Cultures were incubated at 37°C for 2 h with vigorous shaking. Cells were washed several times with cold water and dissolved in 5 ml water, and 100 ml samples were plated on minimal medium supplemented with thiamine hydrochloride, MgSO4´7H2O, ampicillin and 5 mM aminoethylcysteine (AEC). Colonies appeared after 3 days of incubation at 37°C.

Dihydrodipicolinate synthase extraction Bacterial cultures (200 ml) were grown in LB medium supplemented with DAP, thiamine hydrochloride, MgSO4´7H2O and ampicillin. Cells were pelleted (2000 g, 10 min, 4°C) and dissolved in 6 ml of a phosphate buffer (8 mM KH2PO4, 40 mM K2HPO4, 1 mM EDTA, 20% glycerol, pH 7.6). Lysozyme was added to digest the bacterial cell wall (1 h on ice). Cells were broken by sonication and cell debris was removed by centrifugation (7700 g, 10 min, 4°C). Proteins were precipitated by addition of 60% ammonium sulphate, and were dissolved in 6 ml of phosphate buffer. This step is required to remove lysine derived from the bacterial debris. Pyruvate was added at 50 mM concentration. Most proteins were precipitated by incubation of the suspension at 65°C for 5 min. DHDPS is stabilized by pyruvate and remains soluble whereas denatured proteins can be removed by centrifugation (7700 g, 10 min, 4°C). DHDPS protein extracts can be stored at ±20°C without any loss of activity.

Dihydrodipicolinate synthase activity test Enzyme assays were carried out in 1 ml sample buffer consisting of 100 mM Tris±HCl (pH 8), 35 mM pyruvate, 2 mM neutralized Lã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

Selection of mutant cDNAs through functional rescue aspartic-b-semi-aldehyde, 40 ml DHDPS extract and 35 ml of an oaminobenzaldehyde solution (0.5 mg o-ABA/35 ml ethanol). Reactions were incubated at 37°C for 30±90 min, depending on the DHDPS concentration in the reaction mixtures. The reaction was stopped by adding 200 ml 12% trichloric acid (TCA). In acidic medium, the o-ABA and L-2,3-dihydrodipicolinate produced form a deep purple coloration which is developed maximally 2 h after TCA addition (incubated in the dark). The colour complex is stable for about 6 h and its absorption is measured at 520 nm. ASA was prepared by ozonolysis (Black and Wright, 1954) of L-allylglycine, binding of the produced L-ASA on a DOWEX column (hydrogen form, 200 mesh), washing with water and eluting with 4 N HCl. ASA was kept in 4 N HCl at ±20°C and was neutralized with 4 vol of 1 N NaOH just before use.

Acknowledgements The authors wish to thank Professor Dr R. Huber (Max-Planck Institut fuÈr Biochemie) for providing the PDB ®les for dihydrodipicolinate synthase, Professor Dr J. Steyaert and T. Hamelryck (Instituut Moleculaire Biologie, Vrije Universiteit Brussel) for their assistance in interpreting the DHDPS protein structures and generating Figure 6. The authors wish to thank Dr Hawkesford (IACR Rothamsted, Harpenden, UK) for reading this manuscript. Dr B. Bachmann and the Escherichia coli Genetic Stock Centre are thanked for providing bacterial auxotrophic strains. Sequence analysis was done using the Genebase software (Applied Maths, Kortrijk, Belgium). M.V. is supported by a `Fonds voor Wetenschappelijk Onderzoek' fellowship (G.0242.97) and V.F. is a post-doctoral fellow of the `Fonds voor Wetenschappelijk Onderzoek ± Vlaanderen'.

References Black, S. and Wright, N. (1954) Aspartic b-semialdehyde dehydrogenase and aspartic b-semialdehyde. J. Biol. Chem. 213, 39±50. Blickling, S., Renner, C., Laber, B., Pohlenz, H.-D., Holak, T. and Huber, R. (1997a) Reaction mechanism of Escherichia coli dihydrodipicolinate synthase investigated by X-ray crystallography and NMR spectroscopy. Biochemistry 36, 24±33. Blickling, S., Beisel, H.-G., Bozic, D., KnaÈblein, J., Laber, B. and Huber, R. (1997b) Structure of dihydrodipicolinate synthase of Nicotiana sylvestris reveals novel quaternary structure. J. Mol Biol. 274, 608±621. Bouvier, J., Richaud, C., Richaud, F., Patte, J.-C. and Stragier, P. (1984) Nucleotide sequence and expression of the Escherichia coli dapB gene. J. Biol. Chem, 259, 14829±14834. Bright, S., Featherstone, L. and Mi¯in, B. (1979) Lysine metabolism in a barley mutant resistant to S-(2-aminoethyl) cysteine. Planta, 146, 629±633. Bukhari, A. and Taylor, A. (1971) Genetic analysis of diaminopimelic acid- and lysine-requiring mutants of Escherichia coli. J. Bacteriol. 105, 844±854. Cremer, J., Treptow, C., Eggeling, L. and Sahm, H. (1988) Regulation of enzymes of lysine biosynthesis in Corynebacterium glutamicum. J. Gen. Microbiol. 134, 3221± 3229. Dereppe, C., Bold, G., Ghisalba, O., Ebert, E. and Schar, H.-P. (1992) Puri®cation and characterisation of dihydrodipicolinate synthase from pea. Plant Physiol. 98, 813±821. Falco, S., Guida, T., Locke, M., Mauvais, J., Sanders, C., Ward, R. and Webber, P. (1995) Transgenic canola and soybean seeds with increased lysine. Biotechnology, 13, 577±582. ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248

247

Frankard, V., Ghislain, M. and Jacobs, M. (1992) Two feedbackinsensitive enzymes of the aspartate pathway in Nicotiana sylvestris. Plant Physiol. 99, 1285±1293. Frisch, D., Tommey, A., Gengenbach, B. and Somers, D. (1991) Direct genetic selection of a maize cDNA for dihydrodipicolinate synthase in an Escherichia coli dapA± auxotroph. Mol. Gen. Genet. 228, 287±293. Ghislain, M., Frankard, V. and Jacobs, M. (1990) Dihydrodipicolinate synthase of Nicotiana sylvestris, a chloroplast localised enzyme of the lysine pahway. Planta, 180, 480±486. Ghislain, M., Frankard, V. and Jacobs, M. (1995) A dinucleotide mutation in dihydrodipicolinate synthase of Nicotiana sylvestris leads to lysine overproduction. Plant J. 8, 733±743. Hogansen, D. and Stahly, D. (1975a) Regulation of dihydrodipicolinate synthase during growth and sporulation of Bacillus cereus. J. Bacteriol. 124, 1344±1350. Hogansen, D., Irgens, R., Dol, R. and Stahly, D. (1975b) Bacterial sporulation and regulation of dihydrodipicolinate synthase in ribonucleic acid polymerase mutants of Bacillus subtilis. J. Bacteriol. 124, 1628±1629. Karchi, H., Shaul, O. and Galili, G. (1993) Seed-speci®c expression of a bacterial densensitized aspartate kinase increases the production of seed threonine and methionine in transgenic tobacco. Plant J. 3, 721±727. Kraulis, P. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946±950. Kumpaisal, R., Hashimoto, T. and Yamada, Y. (1987) Puri®cation and characterisation of dihydrodipicolinate synthase from wheat suspension cultures. Plant Physiol. 85, 145±151. Laber, B., Gomis-RuÈth, F., RomaÄo, M. and Huber, R. (1992) Escherichia coli dihydrodipicolinate synthase, identi®cation of the active site and crystallisation. Biochem. J. 288, 691±695. LeÂJohn, H. (1971) Enzyme regulation, lysine pathways and cell wall structures as indicators of major lines in evolution in fungi. Nature, 231, 164±168. Maniatis, T., Fritsch, E. and Sambrook, J. (1982) Section 5.53. Enzymes used in molecular cloning. In Molecular Cloning: A Laboratory Manual. Cold Spring Harbor: Cold Spring Harbor Laboratory Press. Mazelis, M., Whatley, F. and Whatley, J. (1977) The enzymology of lysine biosynthesis in higher plants: the occurrence, characterisation and some regulatory properties of dihydrodipicolinate synthase. FEBS Lett. 84, 236±240. Miller, J. (1972) Experiment 16. In Experiments in Molecular Genetics. Cold Spring Harbor: Cold Spring Harbor Laboratory Press, pp. 135±139. Negrutiu, I., Cattoir-Reynaerts, A., Verbruggen, I. and Jacobs, M. (1984) Lysine overproducer mutants with an altered dihydrodipicolinate synthase from protoplast culture of Nicotiana sylvestris (Spegazzinni and Comes). Theor. Appl. Genet. 68, 11±20. Patte, J.-C. (1983) Diaminopimelate and lysine. In Amino Acids, Biosynthesis and Genetic Regulation (eds Hermann, K. and Somerville, R.). Amsterdam: Addison-Wesley, pp. 213±228. Perl, A., Shaul, O. and Galili, G. (1992) Regulation of lysine synthesis in transgenic potato plants expressing a bacterial dihydrodipicolinate synthase in their chloroplasts. Plant Mol. Biol. 19, 815±823. Sano, K. and Shiio, I. (1970) Microbial production of L-lysine: production by mutants resistant to S-(2-aminoethyl)-L-cysteine. J. Gen. Appl. Microbiol. 16, 373±391. Shaul, O. and Galili, G. (1992) Increased lysine synthesis in

248 Marc Vauterin et al. tobacco plants that express high levels of bacterial dihydrodipicolinate synthase in their chloroplasts. Plant J. 2, 203±209. Shaul, O. and Galili, G. (1993) Concerted regulation of lysine and threonine synthesis in tobacco plants expressing bacterial feedback-insensitive aspartate kinase and dihydrodipicolinate synthase. Plant Mol. Biol. 23, 759±768. Shaver, J., Bittel, D., Sellner, J., Frisch, D., Somers, D. and Gengenbach, B. (1996) Single-amino acid substitutions eliminate lysine inhibition of maize dihydrodipicolinate synthase. Proc. Natl Acad. Sci. USA 93, 1962±1966. Vauterin, M. and Jacobs, M. (1994) Isolation of a poplar and an

Arabidopsis thaliana dihydrodipicolinate synthase cDNA clone. Plant Mol. Biol. 25, 545±550. Wallsgrove, R. and Mazelis, M. (1981) Spinach leaf dihydrodipicolinate synthase: partial puri®cation and characterisation. Phytochemistry, 20, 2651±2655. Williams, M. and Rogers, P. (1987) Expression of arg genes of Escherichia coli during arginine limitation dependent upon stringent control of translation. J. Bacteriol. 169, 1644±1650. Yeh, P., Sicard, A. and Sinskey, A. (1988) General organization of genes speci®cally involved in the diaminopimelate±lysine biosynthetic pathway of Corynebacterium glutamicum. Mol. Gen. Genet. 212, 105±111.

ã Blackwell Science Ltd, The Plant Journal, (2000), 21, 239±248