Background of the Invention Recent advances in gene transfer technology have opened up new possibilities for introducing desir¬ able traits into plants. A number of such genes have been introduced, in order to confer upon the host plant some measure of protection against environmental stresses. Examples include genes conferring tolerance to chemical herbicides such as glyphosate (Comai, Nature, 317, 741-744 (1985) and Shah, Science, 233, 478-481 (1986)), phosphinothriciπ (De Block, EMBO, §_, 2513-2518 (1987)), bromoxynil (Stalker, 1987, Interna- tional Patent Application No. PCT/US87/00044) , and sulfonylureas (Haughn, Mol. Gen. Genet. , 211, 266-271 (1988)). Transgenic plants have also been engineered to resist certain insect pests (Adang, 1985, published European Patent Application No. 142,924 and Vaeck, Nature, 328, 33-37 (1987)), fungal diseases (Taylor, Mol. Gen. Genet., 210, 572-577 (1987)), and viral diseases (Abel, Science, 232, 738-743 (1986) and Nelson, Bio/Technol. , 6_, 403-405 (1988)).

Another area of interest is the design of plants, especially crop plants, with added value traits. An example of such a trait is improved nutri¬ tional quality in food crops. Lysine, an a ino acid essential in the diet of humans and monogastric ani¬ mals, is among the three most limiting nutrients in most of the cereal crops. Consequently, grain-based diets, such as those based on corn, barley, wheat, rice, maize, millet, sorghum, and the like, must be supplemented with more expensive synthetic lysine or with lysine-containing oilseed protein meals. Increas- ing the lysine content of these grains or of any of the feed component crops would result in significant added value. To date, attempts to elevate lysine levels in

plants have relied on conventional breeding methods and, more recently, mutagenesis and cell culture tech¬ nology.

Naturally-occurring high lysine mutants of maize (Mertz, Science, 145, 279 (1964) and Nelson, Science, 150, 1469-1470 (1965)), barley (Munck, Science, 168, 985-987 (1970)), and grain sorghum (Singh et al., Crop Sci., 13, 535-539 (1973)) have been iden¬ tified. In each case, the improved lysine content results not from increased free lysine production, but from shifts in the overall protein profile of the grain: the reduced levels of lysine-deficient endo¬ sperm proteins (prolamines) are complemented by ele¬ vated levels of more lysine-rich proteins (albumins, globulins and glutelins). While nutritionally superior, these mutants are associated with reduced yields and poor grain quality, limiting their agronomic usefulness.

An alternative approach used to improve nutri- tional quality has been _in vitro selection for biochem¬ ical variants having elevated free lysine pools. Lysine is a member of the "aspartate family" of amino acids in higher plants and microorganisms (see Figure 1). As such, the regulation of its biosynthesis is intimately connected to that of the other members of the family: threonine, methionine and isoleucine. Regulation of metabolite flow appears to be chiefly through endproduct feedback inhibition at key enzymatic steps. The first of these steps, the phosphorylation of aspartate catalyzed by aspartate kinase (AK), is common to all four endproducts. A second site of reg¬ ulation is at the branch-point reaction: the condensa¬ tion of pyruvate with aspartyl semialdehyde to form dihydrodipicolinic acid. This reaction is the first one unique to the biosynthesis of lysine and is cata¬ lyzed by dihydrodipicolinic ^' acid synthase (DHDPS), an

enzyme shown to be strongly feedback inhibited by lysine in plants where it has been examined (Wallsgrove et al., Phytochem. , 2_0, 2651-2655 (1981), and Kumpaisal, Plant Physiol. , _85, 145-151 (1987)). There is evidence to suggest the existence of more than one form of AK (Miflin, 1980, in The Biochem¬ istry of Plants, vol. 5, Amino Acids and Derivatives, Stumpf and Conn (eds.) pp 420-426, Academic Press). One form is sensitive to inhibition by threonine, the other to inhibition by lysine. The growth of plant cell cultures is inhibited in the presence of equimolar amounts of lysine plus threonine. This inhibition may be reversed by the addition of ethioniπe or homoserine (which may be readily converted to ethionine) (Green et al., Crop Sci. , l±, 827-830 (1974)). Hibberd

(Planta, 148, 183-187 (1980)) selected stable lines of maize callus that were resistant to this growth inhibi¬ tion. These lines overproduced threonine ( 6-9 fold) and to a lesser extent, methionine, lysine and iso- leucine (2-4 fold). There was evidence that a lysine- tolerant AK was responsible for the changes observed. In the lines that were regenerated to whole, fertile plants, the overproduction was a stable, heritable trait (Hibberd et al., PNAS, 7_9, 559-563 (1982)). Similar selections have been carried out on tobacco (Bourgin, 1982, in Variability in Plants Regenerated from Tissue Culture, Earle and Demarly (eds.), pp 163-174, Praeger, New York), barley (Arruda, Plant Physiol. , 1_6.- 442-446 (1984)), and carrot (Cattoir- Reynaerts, Biochem Physiol. Pfanzen, 178, 81-90 (1983)).

Lysine analogs, in particular S(2-aminoethyl)- cysteine (AEC) have also been used either alone or in conjunction with lysine plus threonine selections in attempts to isolate lysine-overproducing mutants. Sano

et al. (J. Gen. Appl. Microbiol., 16_, 373-391 (1970)) were able to isolate high lysine bacterial mutants using AEC selection and AEC was proposed to act as a false feedback inhibitor of AK or DHDPS or both. Attempts to isolate similar mutants in plants have had mixed results. Widhol (Can. J. Bot., 54, 1523-1529 (1976)) mutagenized tobacco suspension cells and selected AEC-resistant cell lines that overproduced lysine by ten-fold. Pearl millet mutants were isolated that overproduced lysine by 5-7 fold (Boyes et al.,

Plant Sci., 50, 195-203 (1987)). Bright (Planta, 146, 629-633 (1979)) selected AEC-resistant barley lines that did not accumulate lysine in the absence of AEC and were shown to be AEC uptake mutants. There was also evidence that AEC exerted its inhibitory effects by being incorporated into proteins rather than by interfering with lysine biosynthesis. Schaeffer et al. (Plant Physiol., _84, 509-515 (1987)) applied sequential AEC and lysine plus threonine selections to obtain rice mutants that had 14% higher lysine in seed storage pro¬ teins, but not higher free lysine. An AEC-resistant potato culture was selected by Jacαbsen (J. Plant Physiol. , 123, 307-315 (1986)). This mutant had higher overall amino acid levels than control cultures but this was not due to ^' overproduction of lysine, threonine or methionine. Negrutiu (Theor. Appl. Genet., 68, 11-20 (1984)) subjected tobacco protoplasts to mutage- nesis followed by AEC selection. Two resistant cell lines were obtained that overproduced lysine by 10-30 fold. Biochemical and genetic analysis revealed a feedback-insensitive DHDP synthase. The trait was inherited as a single dominant gene.

Heretofore, recombinant DNA and gene transfer technologies have not been applied to the area of increased metabolite production for added value in

plants. However, it is known that the bacterium Escherichia coli synthesizes lysine by a pathway essen¬ tially identical to that of higher plants. Dihydro¬ dipicolinic acid synthase is encoded by the dap A gene of E^ coli and, while sensitive to lysine (Yugari et al., Biochim. Biophys. Acta, 62, 612-614 (1962)), it is at least 200-fold less sensitive to inhibition by lysine in vitro when compared to the same enzyme iso¬ lated from plants. Further, E^ coli cells carrying the dap A gene on a multicopy plasmid accumulate high levels of free lysine (Dauce-LeReverand, Eur. J. Appl. Microbiol. Biotechnol., 15, 227-231 (1982)), suggesting that DHDPS catalyzes a rate-limiting step. The gene has been sequenced and characterized (Richaud, J. Bacteriol. , 166, 297-300 (1986)) and Glassman, 1988,

M.S. Thesis, University of Minnesota, Minneapolis, MN). Considering the relative inability of conven¬ tional breeding and tissue culture technology to readily obtain plants accumulating significantly higher levels of lysine, a need exists to apply recombinant DNA and gene transfer technology to produce such plants.

Summary of the Invention The present invention provides a method of substantially increasing the level of free L-lysine in a plant comprising: (a) introducing a foreign gene into the cells of a plant tissue source, and (b) expressing the foreign gene in said cells, wherein said foreign genβj comprises a DNA sequence which encodes dihydrodipicolinic acid synthase (DHDPS) which is sub¬ stantially resistant to feedback inhibition by endogenously-produced free L-lysine. In the practice of the present method, the foreign gene comprises a second DNA sequence which is attached to the 5 - terminus of the DHDPS DNA sequence, and which encodes a

chloroplast transit peptide (CTP). The CTP localizes the DHDPS in the chloroplasts of said cells, where it can act to enhance the biosynthesis of free L-lysine. Therefore, the present method also provides plant cells which have been transformed by the intro¬ duction of a foreign gene which expresses a form of DHDPS which is substantially-resistant to feedback inhibition by endogenous lysine. Since techniques are known by which cells from a wide variety of plant tissue sources can be regenerated into whole plants, the present method also provides a transformed plant which produces free L-lysine by a biosynthetic pathway employing DHDPS, wherein the DHDPS is the product of a foreign (exogeneous) gene, and is substantially resis- tant to feedback-inhibition by endogeneously-produced lysine. Preferred plants which are transformed by the present method include the graminaceous species, such as those enumerated hereinabove. The edible parts of such transformed plants can have free L-lysine levels which are at least about 50 times higher than the lysine levels in an untransformed plant of the same species.

The foreign gene employed to transform the plant cells is preferably a chimeric-gene expression cassette comprising a gene coding for DHDPS or an enzymatically-functional fragment thereof which is sub¬ stantially resistant to feedback inhibition. The expression cassette also comprises a second DNA sequence which encodes an amino-terminal chloroplast transit sequence (CTS). The DHDPS gene or gene frag¬ ment is joined in correct reading frame at its 5' - terminus to the CTS gene. The foreign gene is prefer¬ ably under the transcriptional and translational con¬ trol of regulatory regions which are functional in the target plant cells. Such regions can be obtained from

the DNA sequences which are transferred to plant cells by the Ti or Ri plasmids of A. tumefaciens, e.g., seg¬ ments of the "T DNA." For example, a useful transcrip- tional initiation region can be isolated from a Ti or Ri plasmid gene encoding octopine synthase, nopaline synthase or mannopine synthase. The expression cass¬ ette preferably further comprises a gene encoding a function that is selectable in plant cells, such as drug or herbicide resistance, so that the transformed cells, and the plants derived therefrom, can be readily identified.

A further embodiment of the present invention comprises an expression vector such as a plasmid, which incorporates the present expression cassette, wherein said plasmid is capable of replication in a bacterium such as E_. coli or Agrobacterium. The foreign gene can be introduced into the genome of the plant cells as "naked" DNA by methods such as electroporation, micro- injection, icroprojectile injection or via liposome encapsulation. Also, a plasmid incorporating the foreign gene can be introduced into the genome of the plant cells by A. tumefaciens-mediated transformation. As used herein with respect to a gene or gene fragment, the term "foreign" means that the gene or gene fragment is obtained from one or more sources other than the genome of the species of plant within which it is ultimately expressed. The source can be natural, e.g., the gene can be obtained from another source of living matter, such as bacteria, yeast, fungi and the like, or a different species of plant. A pre¬ ferred source of a gene encoding functional DHDPS is the E_. coli dap A gene. The source can also be synthe¬ tic, e.g., the gene or gene fragment can be prepared in vitro by chemical synthesis. As used herein with respect to a foreign gene or gene fragment which has been introduced into plant

cells, the term "expresses" means that the gene is stably incorporated into the genome of the cells, so that the product encoded by the gene, e.g., an enzyme such as DHDPS, is produced in a functional form within the cells. For example, a functional form of DHDPS catalyzes a step in the endogenous biosynthesis of lysine.

As used herein with respect to the feedback inhibition of DHDPS by endogenous lysine, the term "substantially resistant" means that the DHDPS remains functional in the presence of endogenous lysine to the extent that the plant accumulates lysine substantially in excess of that accumulated by a plant of the same species which does not synthesize DHDPS which is so resistant. For example, novel plants resulting from the present method contain extractable lysine levels at least about ten times, and preferably, at least about 50 times, e.g., about 50-250 times higher than plants of the same species which contain only native DHDPS. Furthermore, free lysine is not present in levels which are toxic to the particular plant species which has been altered. Also, plant cells or plants which are referred to as "transformed" have the foreign gene or gene fragment stably, functionally and inheritably integrated into their genome.

Brief Description of the Drawings Figure 1 is a schematic depiction of the lysine biosynthetic pathway wherein the following letters indicate the following enzymes: a—aspartate kinase; b— spartyl-semialdehyde dehydrogenase; c— dihydrodipicolinic acid synthase; d—dihydrodipicolinic acid reductasε; e—succinyloxoaminopimelate synthase; f—succinyldiaminopimelate amino transferase; g— succinyldiaminopimelate desuccinylase; h—diaminopimel- ate; and i—meso-diaminopimelate decarboxylase.

Figure 2 schematically depicts the manipula¬ tion of the dap A gene as described in Example I.

Detailed Description of the Invention The present invention relates to a novel method for obtaining plants, such as the grains, that produce elevated levels of free lysine. The overpro¬ duction results from the introduction and expression of an introduced bacterial gene encoding dihydrodipicoli- nic acid (DHDP) synthase, the branch-point enzyme in the biosynthesis of lysine in both plants and bacteria. Native plant DHDP synthase is highly sensitive to feed¬ back inhibition by L-lysine and constitutes a key site of regulation of the pathway. By contrast, DHDP syn- thase isolated from Escherichia coli is active in the presence of at least 200-fold higher levels of L- lysine.

In order for the introduction of a gene encod¬ ing a lysine-tolerant DHDP synthase to result in the accumulation of higher levels of free lysine in a plant, a long and complex chain of events must occur.

In the first place, the bacterial DHDP synthase gene must be modified ij vitro to include regulatory signals required for gene expression in plant cells. Because the lysine biosynthetic pathway in plants is reported to be sequestered in the chloro- plasts, the bacterial gene is preferably modified to add sequences encoding an amino-terminal chloroplast transit peptide sequence, in order to direct the gene product to these organelles.

In order to alter the biosynthesis of lysine, the gene must be introduced into the plant cells and these transformed cells identified. The gene must also be stably incorporated into the plant cell genome. The transcriptional signals of the gene must be recognized

by and be functional in the plant cells. That is, the gene must be transcribed into messenger RNA, and the mRNA must be stable in the plant nucleus and be trans¬ ported intact to the cytoplasm for translation. The gene must have appropriate translational signals to be recognized and properly translated by plant cell ribo- somes. The polypeptide gene product must escape pro- teolytic attack in the cytoplasm and be able to assume a three-dimensional conformation that will confer enzy- mic activity. The bacterial DHDP synthase must further be able to function in the biosynthesis of lysine; that is, it must be localized near the native plant enzymes catalyzing the flanking steps in biosynthesis (presum¬ ably in the chloroplast) in order to obtain the required substrates (aspartyl semialdchyde and pyru- vate) and to pass on the appropriate product (dihydro- picolinic acid).

Even if all these conditions are met, success¬ ful overproduction of lysine is not a predictable event. There must be no other control mechanism com¬ pensating for the reduced regulation at the DHDP syn¬ thase step. This means not only no other inhibition of biosynthesis, but also no mechanism to increase the rate of breakdown of the accumulated lysine. Lysine must be also overproduced at levels which are not toxic to the plant. Finally, the introduced trait must be stable and heritable in order to permit commercial development and use.

Modification of a Bacterial Gene for

Expression and Function in Plant Cells A gene encoding dihydrodipicolinic acid syn¬ thase (DHDPS) can be obtained from any microorganism that synthesizes lysine by the diaminopimelate pathway. A key criterion in the choice of gene is the relative

insensitivity of the gene product (DHDPS) to inhibition by lysine. One such gene, which is a preferred start¬ ing material for use in the present method, is the E. coli dap A gene, which encodes a functional 32 kD species of DHDPS which is highly resistant to feedback inhibition by lysine. The gene may be isolated from the microorganism by methodologies well known in the art. Such methodologies include screening a genomic library for complementation of a known mutant of DHDPS; immunological screening of polypeptides expressed from gene libraries; screening a genomic library for hybri¬ dization to a radiolabelled oligonucleotide probe, and the like.

The oligonucleotide probe may be synthesized based on sequence derived from genes of other species or from reverse translation of the polypeptide sequence of an isolated DHDPS subunit. Alternatively, the gene may be chemically synthesized wholly or in part.

Once isolated, the gene is characterized using standard recombinant DNA manipulations and molecular analyses. Such techniques are well known to those in the art and are outlined in Maniatis et al., Molecular cloning: A laboratory manual, Cold Spring Harbor Laboratory, NY (1982). Typically, the DNA fragment carrying the DHDPS gene is reduced to the smallest functional fragment possible; that is, extraneous DNA flanking the gene is removed by methods such as Bal31 digestion, deletion of a known restriction endonuclease fragment, and the like, to obtain the smallest DNA fragment that will, for example, still complement the DHDPS mutant. This fragment is usually less than three kilobases, more commonly, about one kilobase. The nucleotide sequence of this DNA fragment may then be determined by any of several conventional methods. The open reading frame (coding sequence), the putative RNA

polymerase binding site (promoter), ribosomal binding site, and the transcriptional termination signal sequence are then identified. The transcriptional ini¬ tiation site may be determined by techniques such as Sl-nuclease mapping or primer repair analysis.

Extracts of bacteria (typically E^ coli) incorporating the isolated gene on a plasmid are assayed to confirm the presence of DHDPS.activity and the in vitro rela¬ tive resistance of enzymic activity to added L-lysine. Once characterized, the gene is modified to allow integration, expression, and function of the ^' gene product in host plant cells. The preferred modifica¬ tions are: 1) the addition of a DNA sequence coding for an amino-terminal chloroplast transit peptide to the 5' terminus of the DHDPS coding sequence; 2) the replacement of bacterial 5' and 3' regulatory sequences with 5' and 3' regulatory sequences recognized by and functional in plant cells; and 3) insertion of this plant-specific expression cassette into a vector suit- able for introducing the DHDPS gene into host plant cells, and stably establishing it therein.

Addition of chloroplast transit peptide (CTP) DNA sequence All lysine biosynthetic enzymes studied in plants to date have been localized in the chloroplasts. Thus, to accomplish proper localization of the product of the foreign gene, a DNA fragment encoding a chloro¬ plast transit peptide sequence is attached to the 5' terminus of the DNA sequence coding for DHDPS, in the proper reading frame, whereby a complete transit peptide-DHDPS preprotein will be translated from trans¬ cripts of the gene fusion. Useful transit peptides (typically 40 to 70 amino acids in length) function post-translationally to direct the preprotein to the

chloroplast, where the preprotein is imported in an energy-dependent process. The transit peptide is cleaved either during or just after import to yield the mature polypeptide. The DNA fragment encoding this transit peptide can be obtained from a variety of plant nuclear genes, so long as the products of said genes are expressed as preproteins comprising an amino-terminal transit pep¬ tide and transported into chloroplasts. Examples of plant gene products known to include such transit pep¬ tide. sequences are the small subunit of ribulose bis- phosphate carboxylase, ferredoxin, chlorophyll a/b binding protein, chloroplast ribosomal proteins encoded by nuclear genes, certain heat shock proteins, amiπo acid biosynthetic enzymes such as acetohydroxy acid synthase and 3-enolpyruvylphosphoshikimate synthase, and the like. Alternatively, the DNA fragment coding the transit peptide may be chemically synthesized either wholly or in part from the known sequences of transit peptides, such as those listed hereinabove.

Regardless of the source of the DNA fragment coding the transit peptide, it should include a trans¬ lation initiation codon and encode an amino acid sequence that is recognized by and will function pro- perly in chloroplasts of the host plant. Attention should also be given to the amino acid sequence at the junction between the transit peptide and the mature DHDPS subunit where the preprotein is cleaved to yield mature DHDPS. Certain conserved amino acid sequences have been identified and may serve as a guideline.

Precise fusion of the transit peptide coding sequence with the DHDPS coding sequence may require manipulation of one or both DNA sequences to introduce, for example, a convenient restriction site. This may be accom- plished by methods including site-directed mutagenesis,

insertion of a chemically-synthesized oligonucleotide linker, and the like.

Function of the chloroplast transit peptide may be assayed i_n_ vitro as follows. The transit peptide-DHDPS gene expression cassette may be intro¬ duced into any of a number of plasmid vectors, so that the gene is placed in the proper orientation under the transcriptional control of a bacteriophage promoter such as SP6, T3, T7 and the like. The resultant plasmid is then digested at a unique restriction site 3' to the coding sequence to form a linear DNA mole¬ cule. This DNA molecule is then subjected to a run-off transcription reaction using an RNA polymerase specific for the promoter being used. Yields of such reactions are typically 10-12 micrograms of essentially pure messenger RNA (mRNA) per microgram of linear DNA tem¬ plate. These mRNA transcripts are then translated in the presence of one or more radiolabelled amino acids using i_n_ vitro translation systems such as wheat germ or rabbit reticulocyte lysate, to yield radiolabelled preprotein.

The radiolabelled preprotein is then incubated with isolated intact chloroplasts in the presence of light to assay import efficiency. The chloroplasts are then treated with a protease or combination of pro¬ teases such as therrnolysin, trypsin, chymotrypsin, and the like. This treatment degrades any protein not sequestered within the chloroplasts. After washing in the presence of protease inhibitors, the chloroplaεt.s are lysed using vortexing, freeze/thaw treatment, ] reduced osmoticu or a combination of these. The lysate may be: (a) assayed directly, (b) it may be first subjected to immunoprecipitatiαn using DHDPS subunit antibacterial antiserum generated by standard protocols, (c) it may be centrifuged to separate sfcjro- mal proteins (supernatant) from .proteins localized in

the thylakoid membranes (pellet), (d) or a combination of these processes may be used. SDS-polyacryla ide gel analysis is used to confirm the presence or absence of a radiolabelled protein band corresponding to the mole- cular weight of the mature bacterial DHDPS subunit. The presence of the band indicates that the preprotein has been imported and processed to the correct size.

Addition of Plant-Recognized Regulatory Sequences Expression of the bacterial gene such as DHDPS in plant cells requires regulatory sequences that are recognized by and functional in plant cells. These sequences include a 5' transcriptional initiation region and 3' translational and transcriptional termin- ation regions. The 5 ^* transcriptional initiation region will include the RNA poly erase binding site (promoter). It may also include regions required for regulation of transcription where the regulation is mediated by chemical or physical induction or repression.

Examples of such regulation include light- induced expression of ribulose bisphosphate carboxylase small subunit, heat-induced expression of heat shock proteins, genes regulated by plant hormones or other metabolites, developmentally regulated expression, wound- or stress-induced expression, and the like.

The 5' sequences may also include transcrip¬ tional enhancer sequences. The 5' regions may be native to the host plant, or may be derived from other plants where the sequences are functional in the host plant. Suitable sequences may also be obtained from genes of the Ti plasmid of Agrobacterium tumefaciens, including octopine synthase, nopaline synthase, manno- pine synthase, and the like, or may be obtained from certain viral genes. Alternatively, the transcrip¬ tional initiation region may be chemically synthesized either wholly or in part.

The 3' region will include transcriptional termination sequences and may include polyadenylation signal sequences. This region may be derived from the same gene as the 5' sequences or from a different gene. These sequences may also be chemically synthesized.

The resultant expression cassette will com¬ prise a 5' transcriptional initiation region, a chloro¬ plast transit peptide coding sequence, the bacterial DHDPS subunit coding sequence, a translational stop codon, and a 3' transcriptional termination region.

The cassette will usually include less than five kilo- bases, and preferably will include between two and three kilobases.

Insertion of the Expression Cassette in a Vector for Transformation of Plants The choice of a vector for introducing the bacterial DHDPS expression cassette into plant cells will depend on the choice of the transformation method which will, in turn, depend on the host plant and the choice of plant tissue source. A wide variety of pro¬ tocols are available for the introduction of foreign DNA into the cells of both monocots and dicots, includ¬ ing the use of microprojectiles, microinjection, elec- troporation, incubation with Ca ⁺⁺ -precipitated DNA, incubation with liposomes containing the foreign DNA (preferably in the presence of PVA and Ca ⁺² ), viral systems, and Agrobacterium-mediated transformation. See M. From et al., PNAS USA, J32, 5824 (1985) (elec- troporation), T. M. Klein et al., Nature, 327, 70 (1987) (microprojectiles) P. Lurquin et al., Plant Sci. Lett., 25, 133 (1982) (liposomes) and J. Paszkowski et al., EMBO J., ^~ b_, 2111 (1984) (use of CaMV gene VI expression signals), the disclosures of which are incorporated by reference herein.

The first five methods are carried out with "naked" DNA where the expression cassette may be simply carried on any ^~ ^_ coli cloning vector such as plasmids pBR322, pRK290, pACYC184, and the like. In the case of viral vectors, it is desirable that the system retain replication functions, but lack functions for disease induction.

For A robacterium-mediated transformation, the expression cassette will be included in a vector, and flanked by fragments of the Agrobacterium Ti or Ri plasmid, representing the right and, optionally the left, borders of the Ti or Ri plasmid transferred DNA (T-DNA). This facilitates integration of the foreign gene into the genome of the host plant cell. This vector will also contain sequences that facilitate replication of the plasmid in Agrobacterium cells, as well as in JΞ^ coli cells.

All DNA manipulations are typically carried out in E. coli cells, and the final plasmid bearing the DHDPS expression cassette is moved into Agrobacterium cells by direct DNA transformation, conjugation, and the like. These Agrobacterium cells will contain a second plasmid, also derived from Ti or Ri plasmids. This second plasmid will carry all the vir genes required for transfer of the foreign DNA into plant cells.

Regardless of the choice of vector or trans¬ formation protocol, identification of transformed plant cells is facilitated by including a gene encoding a function that is selectable in plant cells. Preferred genes are those encoding resistance to a chemical nor¬ mally inhibitory to plant cells, such as resistance to hygromycin, kanamycin, methotrexate, and certain herbi¬ cides. The selectable marker may be carried on a separate plasmid that is co-transformed with the DHDPS- bearing plasmid, or may be carried on the same plasmid

as the DHDPS cassette. In the case of Aαrobacterium- mediated transformation, the selectable marker can be contained within the region of the plasmid flanked by the T-DNA border regions. Alternatively, a screenable marker such as the 3-glucuronidase gene may be used in place of a selectable marker. Cells transformed with this gene may be identified by the production of a blue product on treatment with 5-bromo-4-chloro-3-indoyl-β- D-glucuronide (X-Gluc).

Transformation and Regeneration of Plants

The choice of plant tissue source for trans¬ formation will depend on the nature of the host plant and the transformation protocol. Useful tissue sources include callus, suspension culture cells, protoplasts, leaf segments, stem segments, tassels, pollen, embryos, hypocotyls, tuber segments, meristematic regions, and the like. Preferably, the tissue source will retain the ability to regenerate, whole, fertile plants follow- ing transformation.

The transformation is carried out under con¬ ditions directed to the plant tissue of choice. The plant cells or tissue are exposed to the DNA carrying the DHDPS expression cassette for an effective period of time. This may range from a less-than-one-second pulse of electricity for electroporation, to a two-to- three day co-cultivation in the presence of plasmid- bearing Agrobacterium cells. Buffers and media used will also vary with the plant tissue source and trans- formation protocol. Many transformation protocols employ a feeder layer of suspended culture cells (tobacco or Black Mexican Sweet, for example) on the surface of solid media plates, separated by a sterile filter paper disk from the plant cells or tissues being transformed.

Following treatment with the DNA, the plant cells or tissue may be cultivated for varying lengths of time prior to selection, or may be immediately exposed to a selective agent such as those described hereinabove. Protocols involving exposure to Agrobac¬ terium will also include an agent inhibitory to the growth of the Agrobacterium cells. Commonly used com¬ pounds are cefotaxi e and carbenicillin. The media used in the selection may be formulated to maintain transformed callus or suspension culture cells in an undifferentiated state, or to allow production of shoots from callus, leaf or stem segments, tuber disks, and the like.

Cells or callus observed to be growing in the presence of normally inhibitory concentrations of the selective agent are presumed to be transformed and may be subcultured several additional times on the same medium to remove non-resistant sections. The cells or calli can then be assayed for the presence of the bac- terial DHDPS gene cassette, or may be subjected to known plant regeneration protocols. In protocols involving the direct production of shoots, those shoots appearing on the selective media are presumed to be transformed and may be excised and rooted, either on selective medium suitable for the production of roots, or by simply dipping the excised shoot in a root- inducing compound and directly planting it in vermicu- lite. . Characterization of Regenerated

Plants and Their Progeny In order to produce transgenic plants exhibit¬ ing elevated free lysine levels, the bacterial gene must be taken up into the plant cell and stably ^' inte- grated within the plant genome. Plant cells and

tissues selected for their resistance to an inhibitory agent are presumed to have acquired the gene encoding this resistance during the transformation treatment. Since this marker gene is commonly linked to the bac- terial DHDP synthase gene, it can be assumed that the bacterial DHDP synthase gene has similarly been acquired. Southern blot hybridization analysis using a probe specific to the bacterial DHDP synthase gene can then be used to confirm that the foreign gene has been taken up and integrated into the genome of the plant cell. This technique may also give some indication of the number of copies of the gene that have been incor¬ porated. Successful transcription of the foreign gene into mRNA can likewise be assayed using Northern blot hybridization analysis of total cellular RNA and/or cellular RNA that has been enriched in a polyadenylated region.

Once transcribed, the mRNA must be translated by the plant ribosomes into a polypeptide, such as the bacterial DHDP synthase subunit, which must in turn be able to assume a three-dimensional configuration that will confer catalytic activity. Plant cells or tissues shown both to carry and transcribe the bacterial DHDP synthase gene may be further characterized by extrac- tion of the cells or tissues and demonstration of lysine-tolerant DHDP synthase activity. This in vitro lysine tolerance is preferably comparable to that observed in extracts of the microorganism that served as the source of the gene and should be readily dis- tinguishable from the more highly lysine-sensitive native DHDP synthase activity which can be extracted from control plant cells or tissues. Depending on the transcriptional initiation and regulatory sequences used in the construction of the cassette, the activity may be detected in all plant tissues, in selected tissues, or only under selected inducing conditions.

Regardless of the location of the bacterial DHDP synthase activity in the plant, it must be local¬ ized within the plant cell so that it can participate in the biosynthesis of lysine, catalyzing the conver- sion of substrates into product. The ability of the bacterial DHDP synthase to so participate may be assessed by determining the relative tolerance of plant cells or tissues to the lysine analog S-(2-aminoethyl)- cysteine (AEC). Like lysine, AEC is a potent inhibitor of plant DHDP synthase. Plant cells or tissues exposed to inhibitory concentrations of AEC are effectively starved for lysine. DHDP synthase from E^ coli , how- "ever, is considerably less sensitive to this inhibi¬ tion. The ability of plant cells or tissues expressing active bacterial DHDP synthase to tolerate normally inhibitory concentrations of AEC strongly suggests that the bacterial enzyme is functioning properly in the biosynthesis of lysine.

If the bacterial DHDP synthase is contributing to the biosynthesis of lysine, and if no other mechan¬ isms act to regulate the free lysine pool, free lysine may accumulate to levels higher than seen in control plant cells or tissues. Free amino acid levels may be readily measured by techniques such as reverse phase HPLC analysis of trichloroacetic acid (TCA) extracts of transgenic plant cells or tissues.

Plants that accumulate significantly elevated levels of free lysine in accord with these mechanisms are grown to maturity. These plants are allowed to flower and are self-pollinated or crossed to an appro¬ priate parental line to obtain seed. This seed may then be analyzed for inheritance of the desired trait. An initial screen of the seed may be germina¬ tion and seedling growth in the presence of coπcentra- tions of AEC that will inhibit the growth of seedlings

germinated from control seed. Seedlings demonstrating AEC tolerance may be grown up and fully characterized as described hereinabove for the original regenerate plants. Plants that may be improved by such a trans¬ formation include but are not limited to processed plants (soybeans, canola, potatoes, tomatoes, lupins, sunflower and cottonseed), forage plants (alfalfa, clover and fescue), and the grains (corn, wheat, barley, oats, rice, sorghum, millet and rye). The plants or plant parts may be used directly as feed or food or the lysine may be extracted for use as a feed or food additive.

Although the free lysine levels described herein are elevated in the leaves of transformed plants, it is expected that the present method will allow elevated free lysine levels in other plant organs, including tubers and seeds.

The invention will be further described in accord with the following detailed example.

Example I. A. Materials and Methods

Restriction endonucleases, T4 ligase, poly- nucleotide kinase, and calf intestine phosphatase were all used according to manufacturer's recommendations. Standard recombinant DNA techniques, transformation of Escherichia coli cells, and molecular analyses were performed according to Maniatis et al., Molecular Clon- ing; A Laboratory Manual, Cold Spring Harbor Labora¬ tory, New York (1982) .

All _±n_ vitro transcriptions (SP6) and transla¬ tions (rabbit reticulocyte lysate) were done using reagents from Promega Biotec (Madison, WI) and followed the manufacturer's protocols.

Oligonucleotides were chemically synthesized using the phosphoramidite method (Caruthers, Science, 230, 281-285 (1985)).

Agrobacterium tumefaciens strain LBA4404 (pAL4404) (Hoekema et al., Nature, 303, 179-180 (1983)) was used for all plant cell transformations.

The isolation and characterization of the dap A gene has been described in Glassmaπ, "Cloning and Characterization of the dap A Gene of Escherichia coli" (M.S. Thesis, Univ. of Minnesota, Minneapolis, MN (1988)), the disclosure of which is incorporated by reference herein.

The gene was isolated from an Escherichia coli genomic library (Clarke and Carbon Collection of Hybrid Plasmids, Yale University School of Medicine, New

Haven, CT). Fragments of a plasmid (pLC1730) reported to carry the dap A gene were subcloned into cloning vector pBR322. Fragments carrying the dap A gene were identified by complementation of a dap A mutant of E_. coli and confirmed by demonstration of a plasmid- encoded protein of molecular mass 32 kilodaltons (corresponding to the molecular mass of the dihydrodi¬ picolinic acid synthase subunit) and by dihydrodipico¬ linic acid synthase enzymatic activity. The nucleotide sequence for one such plasmid construct was determined, and putative promoter sequences and a coding sequence for the dap A gene were identified. Extraneous flank¬ ing DNA was removed by further subcloning.

Of particular interest are plasmids pDAP1763 and pDAP1205. pDAP1763 carries the entire dap A coding sequence plus 5' and 3' untranslated regions on a 1564 bp Ndel-BstEII fragment. pDAP1205 carries a 1375 bp Ndel-Sphl fragment that is missing the last thirteen bases of the coding sequence and all untranslated 3' sequences. Both plasmids are pBR322 derivatives.

B. Chloroplast Transit Peptide Sequence

A DNA sequence encoding a pea chloroplast transit peptide was constructed by first synthesizing subfragments of the sequence and then combining them. Based on the published sequence for the chloroplast transit peptide of a ribulose bisphosphate carboxylase small subunit gene of pea (Cashmore, in: Genetic Engineering of Plants, A. Hollaender, ed., Plenum Press (1983)), eight oligonucleotides (four per strand) were chemically synthesized. The six internal 5' termini bases were phosphorylated using T4 poly¬ nucleotide kinase, then 100 pmoles of all eight oligo¬ nucleotides were mixed together in a total volume of 70 μl. Ten icroliters of this mixture were combined with 100 ng of Hindlll-Sphl digested pPol26 (Robbins et al. , 3. Virol., 58, 339-347 (1986)), denatured at 95°C for 5 minutes, annealed at room temperature for two hours and then ligated to give pSTP31. The synthetic sequence consisted of a Hindlll 5' overhang, 30 bp of untrans- lated leader, 171 bp encoding the 57-amiπo acid transit peptide (tp), and a 3' Sphl overhang which represents the evolutionarily conserved cleavage site for the transit peptide.

C. Modifying the dap A Gene for Use in Plant Cells

Bacterial DNA external to the coding region of the dap A gene was removed and replaced with sequences recognized and processed by plant cells.

In order to add the transit peptide sequence described supra, the 5' terminus of the dap A coding sequence was modified to create an Sphl site. This was accomplished by replacing the 108 bp Nsil-Nrul fragment in pDAP1763 with a 29 bp synthetic linker which regen¬ erated both the Nsil and the Nrul sites. This proce- dure deleted the bacterial promoter for dap A and

changed the sequence surrounding the start codoπ from CCATGT to GCATGC, thus creating the required Sphl restriction site. This change is predicted to result in a phenylalaπine-to-leucine substitution at position two in the amino acid sequence. This plasmid was designated pDAP1900.

The 865 bp Sphl fragment of pDAP1900 which includes the dap A coding sequence (less the last thir¬ teen base pairs) was ligated into Sphl-digested pSTP31 such that the 5'-terminus of the dap A coding sequence was fused to the 3' -terminus of the synthetic transit peptide sequence and the natural cleavage site of the pea small subunit preprotein was restored. The resultant plasmid was designated pDAP4001. The poly- linker in this plasmid provided a Sad site just 3' to the end of the coding sequence. This site was used to move the transit peptide (tp)::dap A cassette as a Hindlll-Sacl 1070 bp fragment into Hindlll-Sacl cut pGEM3 (Promega Biotec, Madison, WI), generating plasmid pDAP4104.

The 3'-terminus of the dap A coding sequence was similarly modified to do four things: (1) destroy the Sphl site (rendering the site created at the 5'- terminus unique), (2) restore the last four amino acids of the sequence which had been lost in the original construction of pDAP1205, (3) add two stop codons, and (4) create a BamHI site adjacent to the end of the coding sequence.

This was accomplished by digesting plasmid pDAP1205 with Sphl plus BamHI and ligating a 25 bp synthetic linker in place of 147 bp of pBR322 DNA to give plasmid pDAP1252. The destruction of the Sphl site resulted in a codon change from alanine to glycine at position 289 in the deduced amino acid sequence. The newly created BamHI site was used to add transcription termination signals recognized by plant

cells. pDAP1252 was digested with BamHI and Hindlll and ligated to a 260 bp Ba HI-Hindlll fragment from pRLlOl containing the transcription termination and polyadenylation signal sequences from the πopaline synthase (nos) gene of pTiC58 (nucleotides 673 to 422, Bevan et al. NAR, _, U, 369-385 (1983)). This construct was designated pDAP1264.

Finally, the modified dap A termini were com¬ bined to reconstruct the gene. pDAP1264 was digested with BstEII (in the middle of the dap A coding sequence) and EcoRI (just 3' to the end of the nos poly A sequence) and ligated into BstEII-EcoRI-digested pDAP4104. This plasmid, pDAP4201, carried the synthe¬ tic chloroplast transit peptide (STP) sequence fused to the coding sequence of dap A and the nos 3' region, all under the transcriptional control of the SP6 promoter . in pGEM3. This construction was used for in_ vitro ana¬ lyses of transit peptide function, as described in Section 4, hereinbelow. For _in_ vivo studies of gene function in plant cells, the STP :: dap A :: nos 3' cassette described hereinabove was placed under the control of the 35S promoter of CaMV as follows. Plasmid p35-227 was derived from pPol26 by inserting the 450 bp BamHI-Bglll fragment carrying the 35S promoter sequence from pCaMVPRO (a gift from V. Walbot; see Fromm et al., PNAS, J32, 5824-5828 (1985)) into the BamHI site of the pPol26 polyliπker. This construction placed a Hindlll site just 3' to the promoter. The 1340 bp Hindlll fragment of pDAP4201 carrying the entire dap A gene cassette was ligated into this site in the correct orientation to form a functional transcriptional unit for expression in plant cells (plasmid pDAP4307).

D. in vitro Function of Transit Peptide

The ability of the synthetic transit peptide sequence to direct the import and processing of the dap A gene product by intact chloroplasts was assessed using irτ_ vitro chloroplast import assays (della-Cioppa et al., Bio/Technology, 5 , 579-584 (1987)). ^3S S- methionine-labelled preprotein for the assay was pre¬ pared as follows. pDAP4201 DNA was digested with EcoRI and sub- jected to run-off transcription using SP6 polymerase (Melton et al., NAR, VI , 1035-1056 (1984)). After removal of plasmid DNA by digestion with RNase-free DNasel, the mRNA was ethanol-precipitated, resuspended in 25 μl sterile distilled deionized water and quan- titated spectrophotometrically. Eight micrograms of this mRNA were used in an _in_ vitro translation reaction (rabbit reticulocyte lysate) which included 2.5 μCi/μl- ^3s S-methionine (New England Nuclear, Boston, MA) in 145 μl total volume. The reaction mixture was incubated at 30°C for 90 minutes and stopped by placing it on ice.

Active, intact chloroplasts were isolated from deveined leaves of Latuca sativa (romaine lettuce) using a method described for pea seedlings (Bartlett et al., in Methods in Chloroplast Molecular Biology, Edelman et al. , eds, Elsevier Biomedical Press (1982)). Isolated chloroplasts were resuspended in import buffer (50 mM HEPES, pH 7.6, 0.3 M sorbitol), adjusted to 2 mg/ml chlorophyll, and maintained on ice until used.

Fifty microliters of the i ^~ \ vitro translation products (mostly tp :: DHDPS subunit preprotein) ^' were combined on ice with 110 μl import buffer, 25 μl 0.1 M L-methionine, 15 μl 0.1 M ATP, and 100 μl chloroplasts. The import reaction mixture was placed at an angle and rotated gently directly in front of a 150W fiber optic bulb at room temperature. At 5, 10 and 15 minutes, 70

μl samples were removed and pipetted into eppendorf tubes on ice containing 7 μl of 1 mg/ml thermolysin in 10 mM CaCl ₂ (Cline et al., Plant Physiol. , 75, 675-678, (1984)). A second sample taken at 15 minutes was pipetted into a tube on ice containing 7 μl 10 mM CaCl2 as control. All tubes were incubated on ice for 20 minutes, then diluted with 150 μl import buffer con¬ taining 50 mM EDTA to inhibit thermolysin. After cen- trifuging for 10 seconds, the chloroplasts were resus- pended in 50 μl lysis buffer (10 M HEPES pH 7.5, 10 M EDTA, 1 mM PMSF, 30-μg/ml aprotinin) and were lysed by two cycles of freeze/thawing and vαrtexing. The lysates were centrifuged at 14,000 x g for 20 minutes to separate the stromal (supernatant) and thylakoid (pellet) fractions.

Stromal protein samples were electrophoresed through a 12.5% SDS-polyacrylamide gel (Laemmli, Nature, 227, 680 (1970)). The gel was stained, dried and subjected to autoradiography. The control sample (no protease) showed a 42 kilodalton band corresponding to the tp :: DHDPS subunit preprotein. All three thermolysin-treated samples lacked the 42 kilodalton band and showed instead a band migrating at about 32 kilodaltons that was protected from proteolytic digestion and therefore was concluded to be imported into the chloroplast. The 32 kilodalton protein corresponds to the molecular mass of the mature bac¬ terial dihydrodipicolinic acid synthase subunit.

E. Construction of the Binary Vector pBVI

The following section describes the construc¬ tion of the vector used for the introduction of the modified dap A into plant cells. The vector comprises the left and right. T-DNA borders of pTiAch5 to facili- tate integration of the gene into the plant genome, a

selectable marker for E. coli (ampicillin resistance), a selectable marker for Agrobacterium tumefaciens and tobacco, (kanamycin resistance), and origin of replica¬ tion sequences to allow plasmid replication in both E^ coli and Agrobacterium tumefaciens.

1. Subcloning of the T-DNA borders

The left border of the pTiAch5 T ^* _ DNA was obtained by digesting pOTY8 (Hoekema, supra) with BamHI and Clal and then isolating a 1206 bp fragment (nucleotides 1 to 1206 according to Barker et al., Plant Mol. Biol., 2 , 335-350 (1983); see also DeVos et al., Plasmid, 6_, 249-253 (1981)). The fragment was ligated into Bglll-Clal-digested pPol26 to give plasmid pOTBLl.

The right border was similarly obtained from pOTY8 as a 5766 bp Hpal-Xhol fragment inserted into the polylinker region of pPol26 digested with Hpal and Xhol to make pTB11828.

2. Subcloning the pSa origin of replication

The broad host range origin of replication of pSa was isolated by digesting pUCD2 (Close et al., Plasmid, _12_, 111-118 (1984)) with Hindi and BamHI. The 2900 bp ori fragment was ligated with Hpal-Bglll-digested pPol26 to give pPolSa.

3. Construction of plasmid containing left and right T-DNA borders and the pSa origin of replication (pBR322LRSa)

The left and right T-DNA borders and the pSa ori were inserted into pBR322 as follows.

pOTBLl was digested with Clal and Hindlll (nucleotide 602, Barker, supra) to generate a 604 bp fragment which was then made blunt- ended with the Klenow fragment of DNA poly- merase. The DNA fragment was inserted into the PvuII site of pBR322 to give pBR322L. The right border was carried on a 912 bp Clal-BamHI fragment from pTB11828 (nucleotides 13774 to 14686, Barker, supra) and was ligated into Clal-BamHI-digested pBR322L to make

PBR322LR.

The hybrid Clal site of pBR322LR was sensitive to methylation to E_^_ coli strain MC1000 and therefore refractory to digestion by Clal.

Transferring the plasmid to E _ coli strain GM272 (da -, dcm ^~ ) (Marinus et al., J. Bacteriol., 114, 1143-1150 (1973)) allowed the insertion of the 2900 bp EcoRI-Clal pSa frag- ment from pPolSa into EcoRI-Clal-digested pBR322LR. The resulting plasmid was desig¬ nated pBR322LRSa.

4. Construction of a kanamycin resistance marker A selectable marker capable of conferring kanamycin resistance in both Agrobacterium and in plants was constructed by using the pro¬ moter from transcript 24 of the pTiAch5 TR DNA. This promoter region was isolated as a 1503 bp Ecof*I-ClaI fragment from pRK290-EcoΔCla (Gelvin et al. , Mol. Gen. Genet. , 199, 240-248 (1985)) representing nucleotides 21,631 to 20,128 (Barker, supra). The fragment was ligated into pPol26 digested with EcoRI and Clal to make pPolPTR. A BamHI

site in the polylinker just 3' to the trans¬ cript 24 promoter was used to insert a 1500 bp Bglll-BamHI fragment from Tn5 (Rothstein et al., Cell, _19, 795-805 (1980)). This fragment contains the neomycin phosphotransferase II

(NPTII) coding sequence, but lacks the native NPTII promoter. Functional orientation of the fragment with respect to the transcript 24 promoter was confirmed by the ability of the resultant plasmid pPOlNPTII to confer kanamy¬ cin resistance on E^ coli cells.

Polyadenylatiαn signal sequences from the octopine synthetase (ocs) gene of pTiAch5 were isolated as a 707 bp PvuII fragment from pTB11828 (nucleotides 12,5541 to 11,834, Barker, supra). This fragment was ligated into Smal-digested pPolNPTII, replacing 500 bp of extraneous Tn5 sequences 3' to the NPTII coding sequence. Correct orientation of the poly A fragment was confirmed by the fragment pattern generated by digestion with Xmalll. The plasmid was designated pPolNPTII-A.

5. Insertion of the selectable marker into pBR322LRSa

. >

The pTR :: NPTII :: ocs 3 gene cassette was moved as a 3100 bp Xhol fragment from pPolNPTII-A into Sail-digested pBR322LRSa to make pBVl. This 11.5 kilobase plasmid con¬ tained - in a clockwise orientation from the unique EcoRI site - the following components: the broad host range pSa ori, the right border, cloning sites Hpal and BamHI, the

pTR/NPTII/ocs chimeric gene (transcribed clockwise), 1415 bp of pBR322 (Sail to PvuII), the left border, and the PvuII-EcoRI fragment of pBR322 which includes the origin of repli- cation and the ampicillin resistance gene.

F. Construction of transformation vector containing the modified dap A gene (pDPZ4474 and pDAP4511)

The unique Hpal and BamHI sites just inside the right border of pBVl were used to insert the CaMV p35S :: STP31 :: dap A :: nos 3' cassette. Using sites in the polylinker flanking this cassette in pDAP4307, a 1830 bp Smal-Bglll fragment was ligated into Hpal- BamHI-digested pBVl to give pDPZ4474. In this con- struction, the dap A gene is just ahead of the NPTII gene and is transcribed in the same direction as NPTII.

Similarly, a 1830 bp BamHI (partial)-Hpal fragment of pDAP4307 was ligated into Hpal-BamHI- digested pBVl to give pDAP4511. In this construction, the dap A gene is also just ahead of the NPTII selec¬ table marker, but in the opposite orientation, such that it is transcribed divergently with respect to the NPTII. pDPZ4474 (ATCC No. 67721) and pDAP4511 were each transformed into Agrobacterium tumefaciens LBA4404 (pAL4404) using the method described in VanVliet et al., Plasmid, _1, 446-455 (1978). Transformants were selected on plates containing kanamycin (50 μg/ml) and streaked out for single colony purity.

G. Transformation of Nicotiana tabacum SRI and regen¬ eration of plants

Cultures of Agrobacterium tumefaciens LBA4404 (pAL4404) carrying pDPZ4474, pDAP4511, or pBVl were used in co-cultivation transformations of tobacco leaf

disks. One set of disks was treated with sterile water as a negative control.

Leaf disks of Nicotiana tabacum SRI plants were transformed essentially as described in Horsch et al., Science, 227, 1229-1231 (1985) with the following exceptions: Agrobacterium cultures were grown in 523 medium (Kado et al. , Physiol. Plant Pathol., 2 , 47-59 (1972)), supplemented with rifampicin (20 μg/ml), streptomycin (100 μg/ml), and kanamycin (50 μg/ml) at 28°C for 48 hours, washed and resuspended in sterile water prior to use with leaf disks; Black Mexican Sweet suspension cultures were used as feeder cultures on nurse plates; all solid media contained 2.5 gm/1 Gelrite (Scott Laboratories, Omaha, NE) in place of agar.

After two days on nurse plates, the leaf disks were placed on shooting medium (same as nurse medium, but containing 500 μg/ml carbeπicillin and kanamycin at 0, 100 or 200 μg/ml and lacking feeder cultures or filter paper). Plates were sealed with parafilm and incubated at 26°C with a 12-hour photoperiod. Shoots were excised when they reached 1 cm and were trans¬ ferred to rooting medium (no hormones) containing car- benicillin and kanamycin at the same concentration on which the shoot originated. When roots appeared, the plantlets were placed in vermiculite and then trans¬ planted to soil.

H. Screening for expression of dap A in regenerated tobacco plants

Plant tissue was initially screened for expression of the introduced dap A gene by evaluating resistance to the lysine analog S(2-aminoethyl)cysteine (AEC). AEC is a potent inhibitor of dihydrodipicolinic acid synthase of plants, but the bacterial DHDPS is

much less sensitive to it. Leaf disks were prepared from surface-sterilized leaves as described in the transformation protocol. The disks were placed, (four disks per plate, two plates per treatment) on shooting medium supplemented with 1 mM L-arginine and AEC at 0, 0.1, 0.2 or 0.4 mM. Plates containing shooting medium supplemented with 75 μg/ml kanamycin were included for each plant tested to check the selectable marker. After one week at 26°C in the light the individual leaf disks were weighed and the average fresh weight and standard deviation were determined for each treat¬ ment.

Leaf disks from all plants remained green and expanded in the absence of AEC. Fresh weight per disk increased from about 9 mg to about 60 mg in 1 week.

Growth was strongly inhibited in control plants in the presence of AEC. Leaf disks plated on medium contain¬ ing at least 0.2 mM AEC were brown and shriveled. In contrast, leaf disks from putative dap A+ plants were readily identified by their ability to remain green and growing in the presence of AEC. Many AEC-tolerant plants were identified using this screening procedure. The characterization of one such plant is described below in detail.

I. Characterization of dap A+ tobacco plant 327 1. AEC tolerance

The response of leaf disks from plant 327 to the presence of AEC was compared with that of leaf disks from water control plant 101.

Results are presented in Table 1.

Table 1 - Leaf Disk AEC Screen.*

Fresh Weight (Wt.-% of control)

AEC (mM) 327 101

0 100 100 0.1 88 32

0.2 81 15

0.4 71 10

* Fresh weights of leaf disks (4 discs per plate, 2 plates per AEC concentration) were measured after one week. Average fresh weights were determined for each AEC treatment and are expressed as the percent of the average fresh weight on plates with no AEC.

2. Lysine-tolerant DHDP synthase activity in leaf extracts

Young leaves of 327 and seed-grown NT-SRI plants were extracted as follows. One to two grams of deribbed leaves were ground at 4°C in 2 ml extraction buffer (0.1 M potassium phos¬ phate (pH 7.5 at 4°C), 2 M EDTA, 1 mM β- mercaptoethanol, 10 mM sodium pyruvate) in the presence of 0.14 gm polyvinylpyrrolidone per gram of tissue. The extract was filtered through Miracloth (Calbiochem) and then cen¬ trifuged for 10 minutes at 8000 x g (4°C). The supernatant was slowly brought to 40% saturation by adding finely ground solid ammonium sulfate at 4°C with gentle stirring. The precipitate was removed by centrifugation.

The supernatant was brought to 66% ammonium sulfate saturation and the 40-66% precipitate was collected by centrifugation as before. The pellet was resuspended in column buffer (50 mM Tris-HCl (pH 7.5 at 4°C), 1 mM EDTA, 10

mM sodium pyruvate, 10% glycerol) and passed through a Sephadex G-25 column (Sigma) to desalt.

Protein concentrations were determined by the dye-binding method of Bradford (REF) using bovine serum albumin fraction V powder in dis¬ tilled water as the standard.

Dihydrodipicolinic acid synthase activity was monitored using the o-aminobenzaldehyde (ABA) assay (Yugari et al., 3. Biol. Chern., 240, 4710-4716 (1965)). L-lysine hydrochloride stock solutions were made in distilled water, filter-sterilized, and added at the indicated concentrations at the beginning of the enzyme reaction. At set time-points, 200 μl aliquots were removed and the reaction quenched in 800 μl of stop buffer (0.21 M citric acid, 0.53 M sodium phosphate (pH 5 at 25°C), with 0.24 mg o-ABA added just prior to use from a freshly- made 10 mg/ l absolute ethanol stock). The pink color was allowed to develop for one hour at 25°C and then the optical density was measured at 520 nm.

DHDP synthase activity in the leaves of 327 was approximately 30-fold higher than that in leaves of seed-grown NT-SRl. More signifi- cantly, the enzyme activity in 327 leaves was completely resistant to added L-lysine at 100 μM and only 14% inhibited in the presence of 500 μM L-lysine. In contrast, 100 μM L-lysine inhibited the DHDP synthase activity in NT-SRl leaves by 87% and no activity could be detected at 500 μM L ^' -lysine.

3. Southern/Northern hybridization analysis

Genomic DNA was isolated from leaves of trans¬ formed plants and 101 according to the method of Shure et al., Cell, 55, 225-233 (1983). Ten micrograms of DNA from each plant was digested separately with BamHI and with BstEII and electrophoresed through 0.7% agarose. Digestion with BamHI produces a diagnostic 1540 bp internal fragment, whereas BstEII cuts once in the dap A coding sequence. The gel was blotted to a nylon membrane and hybridized (Southern, J^ Mol. Biol., 98, 503-517 (1975)), using the 1540 bp BamHI fragment of pDAP4307 labelled with ³² P as a probe. Auto- radiography of the blot clearly showed the characteristic 1540 bp band in the BamHI digest track and two bands in the BstEII digest track. These results, coupled with reconstruction markers for copy number, are consistent with the dap A gene being present in transformed plants as a single copy. No hybridization was evident in lanes containing DNA from plant 101.

Northern blot analysis was also performed and the results indicated that the dap A gene was being actively transcribed. Using both total RNA and oligo-dT-selected polyA RNA isolated from the leaves of transformed plants, the same probe described above showed hybridiza¬ tion to a single species of RNA of about 1100 nucleotides. No hybridization to RNA from 101 was observed.

4. Free lysine levels in leaves

Trichloroacetic acid (TCA) extracts of leaves from 327 and 101 were prepared to assay free amino acid levels. Young leaves (1-2 g ) were diced ^' into an ice cold mortar, covered with liquid nitrogen, and ground to a powder. The ground leaves were lyophilized overnight and dry weights determined. Each lyophilized sample was returned to a cold mortar and re- ground in the presence of 2 ml 10% TCA. The mortar was rinsed with an additional 2 ml of TCA. The TCA extracts were combined and incu¬ bated on ice for 30 minutes with occasional vortexing. The extraction mixture was centri- fuged for 20 minutes (4°C) and the supernatant was removed and placed on ice. The pellet was re-extracted with 2 ml 10% TCA as before and the supernatants were pooled. Two milliliters of the pooled extract were extracted three times with 5 ml ether. The ether extract was stored at -70°C until analyzed.

Free amino acid levels were determined by reverse phase HPLC using the o-phthaldialde- hyde derivatization method of Jones et al., 3.

Chromatog., 266, 471-482 (1983). Three samples of NT-SRl leaves averaged 75 μg free lysine per gram lyophilized tissue. Two samples of 327 leaves gave values of 15,450 and 14,630 μg free lysine per gram lyophilized tissue, respectively, or about a 200-fold increase in free lysine.

5. ^" Heritability of the dap A gene Tobacco plant 327 flowered and was allowed to self-pollinate and produce seed. Mature, dry

seed was collected, surface-sterilized with 10% bleach, and rinsed well with sterile water. Seeds were plated (50 seeds per plate, two plates per treatment) on gridded ger¬ mination plates (1/4 MS salts, 2.5 gm/1 Gelrite) containing AEC at 0, 1, 10, 30, 100 and 300 μM AEC. Seeds from self-pollinated 101 were treated identically as controls. All seeds germinated in 3-4 days. After 10 days, the number of green seedlings on each plate was counted. After two weeks, the seedlings were gently pulled from the Gelrite and root lengths were measured. Results are shown in Table 2 and clearly demonstrate that the AEC tolerance trait is inherited by the progeny of 327.

Table 2 - AEC Resistance in Seedlings

Avg. Root Leπgth (% of control)

AEC (μM) 327 101

0 100 100

1 111 46

3 87 33

10 51 27

30 46 18

100 21 0

Root lengths measured at two weeks post-plating. Average root lengths were determined for each AEC treatment and results are expressed as percent of average root length on medium lacking AEC.

The invention has been described with- refer¬ ence to various specific and preferred embodiments and techniques. However, it should be understood that many variations and modifications may be made while remain¬ ing within the spirit and scope of the invention.

Iπtβrnational Application No : PCT/

Fo— PCT'RC 13 . (Januar ^y 1981)