Nancy L. Paiva*, John D. Hipskind, and John D. Cooper

Plant Biology Division, P.O. Box 2180,The Samuel Roberts Noble Foundation, Ardmore, OK 73401

*Corresponding author (nlpaiva@noble.org)

Secondary metabolites can confer both beneficial and detrimental properties to alfalfa. Some secondary metabolites may deter the invasion by pathogens, attract pollinators and nodulating bacteria, or increase the nutritional value of alfalfa, while others may be seen as antiquality factors. The availability of easily transformed and regenerated alfalfa lines combined with advances in molecular techniques and increased knowledge of biochemical pathways have created the possibility of selectively increasing desirable compounds, decreasing detrimental compounds, or introducing totally foreign compounds into alfalfa. This review summarizes some recent progress in increasing fungal resistance in alfalfa by the introduction of a foreign phytoalexin (antimicrobial secondary metabolite) called resveratrol glucoside (RGluc) via genetic engineering. Transgenic alfalfa shoots accumulated up to 20 ug/g fresh weight in the shoots, and showed decreased infection by Phoma medicaginis under laboratory conditions. Pycnidia formation was completely inhibited on some infected leaves and greatly reduced on others, compared to non-transformed or empty vector-transformed control plants; suppression of sporulation may have a great impact upon spread of the disease in the field.

This genetic modification may improve alfalfa yield and quality in regions where Phoma is prevalent. Field studies are currently underway to determine if plants produce similar levels of RGluc in the field as compared to GH, and to compare forage yields of wild-type and transgenic plants. Progeny both from selfing the primary transgenics (in a RegenSY background) and from crosses with an individual plant line showing greater field performance and the ability to spontaneously self-pollinate are also being analyzed for inheritance of RGluc accumulation and for possible effects of increasing copy number on RGluc levels.

Resveratrol is not only considered as safe for consumption by animals, being present in certain existing forages and fruits, resveratrol is thought to be beneficial to human health, improving cardiovascular health and acting to prevent tumor formation, possibly through its antioxidant and other activities. The RGluc-accumulating alfalfa is currently being compared to parental alfalfa in animal feeding studies to test for increased tumor prevention activity.

Some problems encountered with the current approach are briefly mentioned, and other potential areas and new techniques for future applications are briefly described.

Secondary metabolites play many diverse roles in plants. In alfalfa, they can both confer benefits to the plant, or decreasing the palatability of the forage (Hanson, 1972; Hanson et al., 1988). Phytoalexins such as medicarpin, sativan and vestitol are thought to confer higher disease resistance to fungal pathogens (Higgins et al., 1972; Paiva et al., 1994), various isoflavonoids may effect bacterial and nematode pests (Baldridge et al., 1998), anthocyanin flower pigments help attract pollinating bees, and phenolic compounds such as luteolin and 2’-methoxychalcone (Maxwell et al., 1989) trigger nodulation by Rhizobium meliloti. Lignin and other polyphenolics can decrease digestibility (Jung et al., 1997), saponins may contribute to bloat or confer bitterness to alfalfa, and phytoestrogenic isoflavonoids such as coumestrol and formononetin can decrease fertility of grazing animals. While traditional plant breeding has often directly or indirectly selected for populations altered in the levels of secondary metabolites, current plant molecular biology techniques offer highly specific routes to manipulate compounds existing in alfalfa, or to introduce new compounds.

We have recently (Hipskind and Paiva, 1999) altered the secondary metabolite profile of alfalfa (Medicago sativa) by introducing a single gene from peanut (Arachis hypogaea) encoding resveratrol synthase (Tropf et al., 1994 ). No accumulation of stilbenes has been reported in wild-type alfalfa. Resveratrol and other stilbenes occur in a diverse range of plant species, including peanut, pine, rhubarb, grapevine, and various medicinal plants, where they appear to serve as preformed and induced antimicrobial compounds. One of the best known sources of resveratrol is wine (Soleas et al., 1997), where it is present as both free resveratrol and a glucose conjugate (Fig.1). Resveratrol accumulates in grape berries before ripening, and is thought to help protect the fruits from fungal pathogens. In peanut, the synthesis of resveratrol and other highly antifungal stilbene derivatives is induced in fungal-inoculated leaves. We chose to introduce resveratrol synthesis into alfalfa primarily to try to improve its resistance to certain fungal pathogens. While these studies may provide valuable resistant germplasm for cultivar enhancement, we also see it as a model system for examining the effects of such engineering on alfalfa, and determining where future metabolite engineering problems may lie.

INTRODUCTION OF RESVERATROL ACCUMULATION IN ALFALFA

While wild-type alfalfa does not accumulate stilbenes, it does contain the necessary precursors for their biosynthesis. Resveratrol (3,5,4’-trihydroxystilbene) is a common stilbene, and is synthesized from three molecules of malonyl CoA and one molecule of p-coumaroyl CoA by the action of resveratrol synthase (Figure 1). The precursors malonyl CoA and p-coumaroyl CoA are present in all plants as precursors of fatty acids, lignin, flavonoids, isoflavonoids, anthocyanins, and many other compounds, but the levels of the precursors within specific cells may vary with the development and environment of the plant. We obtained a peanut cDNA clone (Tropf et al., 1994), constructed an appropriate binary vector, and introduced the foreign gene into alfalfa via Agrobacterium tumefaciens-mediated transformation. The vector used in our initial study used a cauliflower mosaic virus (CaMV) 35S promoter with a duplicated enhancer region (reported to provide higher transcription rates) to drive expression of the resveratrol synthase gene, a tobacco etch virus (TEV) 5’ nontranslated region (NTR) (thought to increase translation efficiency of the resulting messenger RNA), and a CaMV 35S 3’ region (thought to assist in polyadenylation of the mRNA) (Fig. 2 A). (See Hipskind and Paiva, manuscript submitted to MPMI, for detailed information). The promoter, 5’- and 3’-regions are all derived from pRTL2 (Restrepo et al., 1990), and these elements together with the RS gene were transferred to the T-DNA region of the binary vector pGA482 (An, 1986).

Multiple independent primary tranformant lines were regenerated from a single cv. RegenSY parent plant line following etablished protocols (Bingham, 1991; Oommen et al., 1994). While previous reports (Hain et al., 1990 and 1993; Fischer et al., 1997; Thomzik et al., 1997) using similar technology in other plant species indicated that such transformants accumulated resveratrol (as determined by ELISA), our phytochemical analysis of the alfalfa transformants utilizing HPLC detected no free resveratrol. However, we did detect substantial amounts of resveratrol glucoside (RGluc; Fig.1) in a number of transformed lines (Hipskind and Paiva, 1999). The levels accumulated ranged from approximately 20 ug RGluc (measured as resveratrol equivalents)/g fresh weight down to undetectable levels (Table 1), even though the vector and parent plant line used for transformation were identical. Many of the higher accumulating transgenics contained only one copy of the transgene construct, as determined by Southern analysis, while the lower accumulating lines contained one, two or three copies. This indicates that so-called "position effects", or the site of integration of the transgene, played a major role in determining expression levels, while total transgene copy number was less important.

Primary transformants were propagated vegetatively by cuttings. RGluc levels in the shoots varied from harvest to harvest (See Hipskind and Paiva, 1999), and it was determined that the RGluc levels were higher in younger than older leaves, but the RGluc levels increased from the top to the bottom of the stem, particularly in the region in which lignification occurs(Inoue et al., 1998). Preliminary experiments had shown that the alfalfa leaf and stem pathogen Phoma medicaginis was inhibited approximately 60% by 0.5 mM resveratrol in agar plate bioassays (performed as described in Blount et al., 1993) (Table 2), and later experiments indicated that RGluc and free resveratrol were equally inhibitory (on an equivalent molar basis) against this pathogen in this bioassay system.

Spray inoculation of plants with Phoma medicaginis spore preprations provided highly variable symptoms in our hands, even on the wild-type parent lines. We developed a leaf needle inoculation assay, wherein a fine hypodermic needle was first dipped into one of a series of diluted spore suspensions, then used to pierce a young leaf on an excised shoot. The shoot was then inserted into a water-agar medium inside a sterile plastic box, the box was sealed, and symptoms were allowed to develop for 8-10 days. The necrotic lesion (dark brown region immediately surrounding the inoculation site) area was substantially reduced in the high RGluc-accumulating lines. Up to an 80% reduction in lesion area was observed with an inoculum of 10,000 colony forming units (CFU)/ml and approximately 40% reduction at 10 million CFU/ml.

Even more striking was a decrease in mycelial growth outside of the necrotic region and in pycnidia formation observed on the high RGluc-accumulating lines. Many inoculated transgenic leaves produced no pycnidia, the spore producing structures of this pathogen, while their formation was readily observed in wild-type plants under our conditions. We believe this effect could have a major contribution to the containment of this disease in the field. This also indicates that while agar plate bioassay data may suggest which compounds are active against a particular pathogen, such inhibitory effects may be enhanced in planta, possibly due to synergistic interactions with endogenous plant defense mechanisms.

In this study, there were at least two major factors contributing to the successful inhibition of the fungus. First, since alfalfa does not produce resveratrol or RGluc, and Phoma medicaginis is a fairly specialized pathogen, Phoma has had little chance of previously encountering RGluc and therefore little chance of evolving a highly effective defense against this compound. For example, many pathogens can specifically degrade their natural host’s phytoalexins, but are sensitive to phytoalexins from species they do not infect. Second, use of the CaMV35S promoter confers constitutive expression of the RS gene and therefore constitutive accumulation of RGluc prior to fungal attack. In alfalfa, the synthesis of the major known natural antifungal phytoalexins is induced in leaves only after the pathogen is recognized as having begun its invasion. Having a phytoalexin at high levels before the pathogen arrives may inhibit its progress until the natural defenses can be activated and prevent the pathogen from ever getting established.

In Table 2, several other alfalfa pathogens from an in-house collection showed significant inhibition by resveratrol. In theory, the current lines of RGluc-accumulating alfalfa may also have increased resistance to these and other pathogens which attack stems or leaves. However, the levels of RGluc which accumulated in roots were minute (<0.2 ug/g.fresh weight) even in those lines which accumulated the highest levels in leaves and stems. Therefore, pathogens which enter through roots such as Fusarium and Phytophthora may still be able to destroy the root system of transgenic plants. In our original study, we analyzed the mRNA levels of the transgene (RS) and consistantly found very high levels in leaves and stems but relatively low levels in alfalfa roots. This was surprising in that CaMV35S promoters are known to have some tissue specificity but are considered to have fairly strong root expression (Benfey and Chua, 1990). Also, we have previously made alfalfa transgenic plants harboring various versions of the CaMV35S promoter driving the reporter gene GUS (b-glucuronidase) and other messages; although not a direct measurement of promoter activity, the GUS transformants had shown good histological staining of both leaves and roots (Oommen et al., 1994).

We have recently analyzed the GUS mRNA levels in one of these CaMV35S:GUS transformants, harboring pKYGUS (Figure 2B; from Dr. C. Schardl, U. Kentucky). Northern analysis indicated that levels of GUS mRNA in roots (R) were very similar to those in leaves (L) and stems (S) (Figure 3A) in two independent RNA extractions (15 mg/lane in first set, 10 mg/lane in second set). For comparison, northern blots analyzing three independent CaMV35S:RS transgenic lines are shown, clearly illustrating the relatively low root RS mRNA levels (Figure 3B; modified from Hipskind and Paiva, 1999). The blots were re-probed with a soybean 18S rRNA clone probe, to prove that the lanes in each set contain approximately equal amounts of total RNA. These results suggest that either the two different promoter derivatives drive different levels of root expression, or something in the RS construct is destabilizing the message in the roots. In addition to the differences in the coding regions expressed, the CaMV35S:GUS and the CaMV35S:RS binary vectors contain many other differences; the portions of the CaMV35S promoter are different, one vector contains the TEV leader while the other does not, the 3’-regions in the binary differ (CaMV35S versus Rubisco small subunit, rbcS). In addition the BamHI site used in subcloning the RS cDNA clone is approximately 900 bp after the stop codon of the RS coding region, providing it with an unusually long 3’-untranslated region. We are currently analyzing existing transgenic alfalfa plants with a variety of transgenes and 3’-nontranslated regions driven by CaMV35S promoter derivatives to further address this question, and to further examine the elements which contribute to high message accumulation in alfalfa.

 
Figure 3

We are also transforming alfalfa with new constructs, in which the expression of RS is driven by a combination of regulatory elements from octopine and mannopine synthase genes (a so-called super-promoter; Ni et al., 1995). This promoter is reported to be as much as 100 times as active as the CaMV35S promoter in tobacco and other species, but has not yet been evaluated in alfalfa. It may not only help address the problem of the lack of RGluc accumulation in roots, but in theory might lead to higher accumulation of RGluc in stems and leaves.

A major concern in manipulating future targets may be the availability of tissue-specific, defense-specific, or sufficiently strong promoter elements for use in alfalfa. We have previously shown that the alfalfa isoflavone reductase promoter conferred constitutive root and nodule expression and was pathogen and elicitor-inducible in alfalfa. The same construct was not regulated in the same fashion when transferred to tobacco; it was expressed in additional leaf and stem tissues in tobacco, and was no longer pathogen or elicitor inducible (Oommen et al., 1994). One could easily imagine that promoters which appear ideal from studies in other plants will prove unsatisfactory in alfalfa. Also, one might wish to express or suppress a pathway in only one part of the plant, which requires developmentally regulated promoters. In our previous work, no detrimental effects on alfalfa development were observed despite the high accumulation of RGluc. In contrast, in a recent study with a similar construct in tobacco, the plants were rendered male-sterile and the flowers lost their anthocyanin pigments due to competition of the RS with CHS for flavonoid precursors (Fischer et al., 1997); promoters devoid of flower and pollen expression would avoid these problems.

Another possible cause for low accumulation of RGluc in alfalfa roots, other than low RS mRNA levels, is more competition from CHS for precursors. In alfalfa roots the levels of CHS message were extremely high, where CHS is involved in the synthesis of several root and nodule flavonoids and isoflavonoids. In other genetic engineering projects such as introducing Bt protein for increased insect resistance or other foreign protein to improve protein quality, the successful accumulation of the protein product is dependent upon mRNA levels (a function of promoter strength and RNA stability) and translation of the message to protein. In secondary metabolite manipulations such as for resveratrol accumulation, in addition to the above factors, we have also to consider enzyme stability, competition with endogenous pathways for precursors, and successful interactions with upstream enzymes. On the latter point, there is evidence that enzymes in a pathway may set up complexes or "channels" for substrates via protein-protein, and foreign enzymes may or may not be able to interact correctly to compete for the substrates.

PRELIMINARY FIELD TRIALS

In summer 1998 we transplanted cuttings of 10 of the independent primary transgenic lines to a field plot in southern Oklahoma, along with cuttings of the RegenSY parent line. Our first goal was to compare the relative levels of RGluc in the shoots, both the absolute levels in replicate plants grown in the greenhouse versus plants grown in the field, and among the chosen lines relative to each other. We are still analyzing the samples and data, but it is very clear that the highest accumulating lines in the greenhouse are also the highest accumulating lines in the field. It is more difficult to make a fair comparison of the absolute levels. In our initial studies, we found that young leaves from non-bloomimg shoots always contain higher levels of RGluc in the greenhouse, while older leaves contained less. However, the unusually high field temperatures during the first season caused all plants derived from the RegenSY cultivar to immediately produce buds on the regrowth, and the foliage from the field looks nothing like that obtained from the greenhouse. At several harvests, the field levels appear lower than the greenhouse levels, but it is not clear if this is due to the higher light intensities or higher temperatures in the field, or simply due to different developmental stages. We have developed faster, more accurate RGluc assays, and are hoping to be able to sample non-blooming shoots at a future date.

We are also comparing forage yields between 10 independent transformants and nontransformed line. Currently, we see no obvious detrimental effect of resveratrol accumulation on forage yield. However, we do see that while all of the lines grew similarly in the greenhouse, visible differences in growth were observed in the field. We believe these negative effects are primarily due to tissue culture artifacts, as has been reported in other labs.

PROGENY ANALYSIS:

The transgenics lines were hand-tripped and produced seed at approximately the same rate as the nontransformed parent line and empty vector-control lines. All seed types had high germination rates (95-100%). In theory, selfing a tetraploid such as alfalfa harboring a single transgene insert should produce a population with 3 plants harboring the RS gene to one plant without it. One third of the transgenic progeny should contain two copies of the transgene at the same locus, and since this is a high expressing locus, these plants might produce higher levels of RGluc than the single copy parents or siblings.

Figures 4 and 5 show the relative RGluc levels observed in the progeny from the self-pollination of two high expressing lines, arranged according to increasing expression levels. In both cases the ratio of RGluc-accumulating lines to non-accumulating lines exceed 3:1; for line 1-D1, the ratio was 39:9, while for line 1E2.1, the ratio was 38:9. This indicates that there is no bias against transfer of the transgene, for example, due to toxicity of the product to transgenic pollen. The levels of RGluc observed in the progeny form a gradually increasing distribution, as opposed to an obvious grouping of lower and higher expressers. This may be partly due to the fact that the parental RegenSY line is already highly inbred, which leads to obvious vigor differences in the selfed progeny; large variations in vigor superimposed upon the two transgene copy levels may explain the observed distributions. We plan to analyze selected higher and lower accumulating progeny for their copy numbers by molecular techniques.

Selected primary transgenic lines have also been crossed to a single plant line derived from the cultivar Cimarron. The Cimarron parent showed much greater forage production and greater vigor than RegenSY in the greenhouse and in the field. This Cimarron line also has a tendency to spontaneously self-pollinate, reducing the amount of hand labor required to produce seed for future analyses. Using the Cimarron as the female parent, seedlings produced from successful crosses are easily selected by qualitatively screening for RGluc accumulation using as little as one trifoliate. After further analyses, the highest accumulators were further visually evaluated for vigor and similarity in appearance to the Cimarron parent under greenhouse conditions; several progeny have inherited both the RS transgene and the tendency to spontaneously self-pollinate. Six of these progeny lines and the respective transgenic and nontransgenic parent lines are currently being evaluated in the field for RGluc levels and increased forage yield.

FUTURE STUDIES/ADDITIONAL APPLICATIONS:

The above studies were initiated to attempt to increase the disease resistance of alfalfa through secondary metabolite manipulations. However, there are many alternative reasons for producing additional secondary metabolites in plants, such as using these compounds as pharmaceuticals, chemical intermediates, or nutrients. Currently, the field of "nutriceuticals", or beneficial non-nutritive food components, has attracted great interest. Resveratrol, the molecule introduced in this example, is not only considered as safe for consumption by animals, it is thought to be beneficial to human health, improving cardiovascular health and acting to prevent tumor formation, possibly through its antioxidant and other activities (Soleas et al., 1997; Jang et al., 1997). Other compounds, such as the isoflavonoids daidzein and genistein, have also attracted much attention. Due to its high quality protein and high value as an animal feed, alfalfa serves as a novel vehicle for testing the value of genetically increasing the levels of potential nutriceuticals in food crops. The RGluc-accumulating alfalfa described above is currently being compared to parental alfalfa in mouse feeding studies to test for increased tumor prevention activity. These studies will indirectly look for any potential detrimental effects of our genetic-engineering strategy on animals. Resveratrol and its glucoside are present in high levels in certain wines and other grape products, plus several medicinal plants used for centuries, and could therefore fall into the category of "generally regarded as safe", or GRAS, compounds for human consumption. It is also present in peanut leaf hay and certain forage grasses, suggesting that it will not harm livestock.

In some cases, one must use caution in either chosing the targets for manipulation or in the levels of the alterations. Many studies indicate that decreasing lignin content in alfalfa, particularly in the stems, will increase digestibility of the forage. However, lignin is required for normal plant structure, water transport in vascular systems, and to some extent as a barrier against pathogens, so drastically reduced levels would be detrimental. The isoflavonoid coumestrol is considered an antiquality factor by some researchers, due to its estrogenic effects on grazing animals; consuming too much phytoestrogen can decrease the number of offspring produced. However, elimination of coumestrol may aggravate certain diseases, in that it is thought to inhibit certain pathogenic nematodes and bacteria. The phytoestrogens including formononetin were reportedly reduced in a clover species by selecting for the enzyme which converts daidzein to formononetin, but if the synthesis of formononetin is blocked entirely in alfalfa, plants will no longer make medicarpin and other related antifungal phytoalexins, possibly making them more susceptible to increased fungal disease. Saponins are considered a potential antiquality factor in alfalfa, possibly contributing to bloat-related foaming of the rumen, but have been demonstrated to act as antimicrobial defense compounds in oats. Even increasing the levels of beneficial endogenous compounds must be approached with caution: high levels of defense compounds have been shown to be potentially damaging to the producing plant, as well as animal consumers. As mentioned earlier, high levels of resveratrol caused male sterility in tobacco. A famous example of secondary metabolite overdose is that of breeders selecting for high levels of psoralens in celery to ward off pests, but exposure to the celery sap caused blisters among the celery handlers. The effects of high doses of some compounds on cattle and horses are as yet unknown.

In many cases, biochemistry and initial gene cloning are the rate-limiting steps. We have identified other potentially more effective antimicrobial compounds that may be useful to introduce into alfalfa (Dixon et al., 1995), but are still in pursuit of the necessary genes from other species. Other groups have indicated that introduction of condensed tannins (polymeric flavonoid derivatives) into alfalfa leaves would decrease the bloat potential of alfalfa (Koupai-Abyazani et al., 1993), but cloning of the necessary genes has not yet been reported.

New molecular techniques for manipulating metabolite levels are appearing constantly. Some require both a firm understanding of the biochemical pathways involved, plus the isolation of the necessary genes. Over-expression of pathway genes, either foreign or native to alfalfa, are now straight-forward, thanks to years of development of transformation and regeneration protocols and genotypes. Techniques such as antisense RNA and co-suppression or sense-suppression are routinely used in the plant molecular biology research journals to eliminate or reduce expression of specific gene products, but reports of their successful application in alfalfa are limited. These techniques require very high promoter activities to generate sufficiently high levels of RNA, and the promoters used to date have primarily been developed from work in other species, and may be sub-optimal for alfalfa. A new technique termed "chimeraplasty" has recently been adapted from animal systems to plants; the technique uses synthetic oligonucleotide analogs to very specifically alter an endogenous plant gene, without introducing any foreign DNA (Gura, 1999; Zhu et al.,1999; Beetham et al., 1999).

One recent approach which assumes no knowledge of the pathways or genes is that of activation tagging (Walden et al., 1994). A strong promoter element is placed facing outward from the T-DNA region of a binary vector, and hundreds of independent transgenic lines or pools are generated. The T-DNA will insert into the genome at somewhat random intervals. If the promoter lands sufficiently close to a coding region, the encoded gene will be over-expressed, providing a dominant mutation, possibly with an observable phenotype. Tagging a single regulatory gene may increase the activities of several genes in a pathway. While requiring a huge initial investment in labor to generate sufficient transgenic lines, such an approach is theoretically applicable to tetraploid alfalfa, and the population could be used to screen for alterations in a number of secondary metabolites, as well as numerous other phenotypes.

Traditional mutation and selection or screening schemes could also be applied to alfalfa at the diploid level, such as the CADL lines developed at Wisconsin, particularly in looking for recessive traits such as loss of a pathway. The current lines suffer from inbreeding depression and lack of transformation/regeneration capacity, but improved lines are possibly under development. The model legume Medicago truncatula, a self-fertile diploid annual medic, is rising in popularity and in the number of resources being dedicated to its study. Functional genomics programs for M. truncatula have begun in both the US and Europe, and should provide a wealth of sequences and expression data for use in isolating genes to use in manipulations, as well as mutant populations. Based on isoflavonoid pathway analysis done in our lab, certain metabolite profiles and gene sequences appear to be highly identical between alfalfa and M. truncatula, but other pathways have not been investigated.

We thank Cuc K. Ly, J. Scott McNeill, and Darla F. Boydston for assistance in preparing the figures, and Allyson D. Wilkins for assistance in preparing the text. The work was funded by the Samuel Roberts Noble Foundation.

Previous Page