Ribozymes Targeted to Stearoyl–ACP 9 Desaturase mRNA Produce Heritable Increases of Stearic Acid in Transgenic Maize Leaves

Correspondence to: Michael Zwick, mgzwick{at}rpi.com (E-mail), 303-449-6995 (fax).

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Ribozymes are RNAs that can be designed to catalyze the specific cleavage or ligation of target RNAs. We have explored the possibility of using ribozymes in maize to downregulate the expression of the stearoyl–acyl carrier protein (9) desaturase gene. Based on site accessibility and catalytic activity, several ribozyme constructs were designed and transformed into regenerable maize lines. One of these constructs, a multimer hammerhead ribozyme linked to a selectable marker gene, was shown to increase leaf stearate in two of 13 maize lines. There were concomitant decreases in 9 desaturase mRNA and protein. The plants with the altered stearate phenotype were shown to express ribozyme RNA. The ribozyme-mediated trait was heritable, as evidenced by stearate increases in the leaves of the R₁ plants derived from a high-stearate line. The increase in stearate correlated with the presence of the ribozyme gene. A catalytically inactive version of this ribozyme did not produce any significant effect in transgenic maize. This is evidence that ribozymes can be used to modulate the expression of endogenous genes in maize.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

In the last decade, considerable effort has been focused on downregulating genes as a means to modify plant phenotypes. Much of the early work emphasized the use of antisense (Smith et al. 1988 ; van der Krol et al. 1988 ) and sense (Napoli et al. 1990 ; van der Krol et al. 1990 ) suppression. Recently, a third downregulation technology, ribozymes, has received attention from the plant science community. Ribozymes are RNA molecules with the ability to act as sequence-specific endoribonucleases (Zaug et al. 1986 ). Self-cleaving RNA molecules were originally identified in the group I intron of the Tetrahymena preribosomal RNA (Kruger et al. 1982 ). A number of other catalytic RNAs have since been discovered. These include group II introns, the RNA subunit of RNase P, hammerhead, hairpin, hepatitis delta virus, and Neurospora VS RNA self-cleaving RNAs (Symons 1994 ). In nature, self-splicing RNA molecules are often involved in the maturation of some RNA transcripts and the processing of small pathogenic RNAs associated with certain plant and animal viruses.

The hammerhead is one of the smallest and most thoroughly studied of the catalytic RNAs. Hammerhead refers to a secondary structure common to the conserved catalytic domain found in a large number of infectious plant pathogenic RNAs (Forster and Symons 1987 ). These self-cleaving RNAs are thought to function in the processing of intermediates formed during the rolling-circle replication cycle. The hammerhead domain can be divided into three RNA helices and 13 conserved nucleotides that form a defined tertiary structure (Pley et al. 1994 ). Mutagenesis experiments have shown the flexibility of the sequences involved in the base pairing of the helices and have defined the conserved core residues (Ruffner et al. 1990 ). Engineered ribozymes have been shown to cleave substrate RNAs in trans (Uhlenbeck 1987 ) and have been refined further such that most of the conserved residues were placed into the catalytic portion of the hammerhead (Haseloff and Gerlach 1988 ). The requirements for trans-cleavage by hammerhead ribozymes include a UH (where H is an A, C, or U residue) recognition site in the target RNA and the ability to base pair with the target (Koizumi et al. 1988 ; Ruffner et al. 1990 ). The base-pairing region, or arms, provides specificity to direct the catalytic domain of the ribozyme to the target site of the substrate RNA. A single-base mismatch at or near the site of cleavage can eliminate or severely reduce catalytic activity (Werner and Uhlenbeck 1995 ).

Relative to the previously employed downregulation strategies, a potential advantage of ribozymes is their catalytic mechanism, theoretically requiring fewer molecules to be effective. Ribozymes are also highly target specific, with the ability to distinguish a one- or two-base difference in target sequence (Werner and Uhlenbeck 1995 ). Thus, ribozymes can be designed to inactivate one member of a multigene family (Bennett and Cullimore 1992 ) or targeted to conserved regions of related mRNAs. Expression of introduced genes in stably transformed plants often has been problematic and is related in part to the phenomenon of homology-dependent gene silencing (Finnegan and McElroy 1994 ; Matzke and Matzke 1995 ). Because ribozymes typically have less duplicated sequence than do antisense or sense constructs, they may be less prone to transgene inactivation.

Although there have been a large number of reports demonstrating the potential applications of ribozymes in mammalian systems, relatively few have documented application of ribozymes in plant systems. The amount of sequence that is complementary to the target may affect whether the ribozyme acts catalytically or by an "antisense-like" mechanism (de Feyter et al. 1996 ). Ribozymes have been shown to reduce expression of reporter genes in plant protoplasts (Steinecke et al. 1992 ; Perriman et al. 1993 , Perriman et al. 1995 ). Effects in transgenic plants also have been documented. Wegener et al. 1994 demonstrated that the effective reduction of neomycin phosphotransferase nptII in transient assays (Steinecke et al. 1992 ) also could be achieved in transgenic tobacco plants. Recently, two endogenous genes encoding UDP–glucose pyrophosphorylase in potatoes and a lignin-forming peroxidase in tobacco have been targeted (Borovkov et al. 1996 ; McIntyre et al. 1996 ). In both studies, decreased enzymatic activity of the target was observed in transgenic plants containing ribozyme or antisense constructs. Neither study included catalytically inactive ribozymes; therefore, in the ribozyme plants, it was unclear whether the effect was due to a ribozyme-based mechanism. One approach to demonstrating ribozyme catalytic activity is detection of an intact cleaved fragment of target RNA. Cleavage products have not been documented in transgenic plants but have been found in transient assays by using protoplasts (Steinecke et al. 1992 ; Perriman et al. 1995 ).

One area for application of ribozyme technology is the alteration of plant lipid biosynthesis for food and industrial uses. In plants, the first step in C-18 fatty acid desaturation is catalyzed by stearoyl–acyl carrier protein (stearoyl–ACP) desaturase, commonly known as 9 desaturase (Figure 1). This enzyme introduces a cis double bond at position 9/10 of the C-18 chain, converting stearoyl–ACP to oleoyl–ACP (stearic acid to oleic acid). Expression and regulation of 9 desaturase in plants have been studied extensively (Fawcett et al. 1994 ; Slocombe et al. 1994 ). cDNA sequences have been cloned from a number of plants, including safflower (Thompson et al. 1991 ), castor (Shanklin and Somerville 1991 ), rapeseed (Knutzon et al. 1992 ), and canola (Slocombe et al. 1992 ). Downregulation of 9 desaturase is expected to increase stearic acid content, which is an effect that would produce oil high in saturates. Oils high in saturates can be used for the production of margarine without additional hydrogenation, which results in formation of trans–fatty acids.

View larger version (27K): [in this window] [in a new window]   Figure 1. The Plant Fatty Acid Biosynthetic Pathway. The first step in C-18 fatty acid desaturation is performed by the enzyme 9 desaturase. Downregulation of 9 desaturase is expected to increase stearic acid (18:0) content. The black star indicates 9 desaturase.

Knutzon et al. 1992 described the antisense-mediated downregulation of 9 desaturase in rapeseed and canola. In rapeseed, desaturase activity was eliminated completely, resulting in stearate contents between 15 and 25%. The elimination of 9 desaturase activity reduced oil content in transgenic seed by 45% and resulted in a low germination rate. In canola, antisense downregulation of the 9 desaturase resulted in stearate contents >30% with normal oil contents and germination. In both rapeseed and canola, elevated stearate was associated with elevated C18:3. Using traditional plant breeding approaches, researchers have increased stearate levels in safflower (Ladd and Knowles 1970 ) and soybean (Hammond and Fehr 1984 ; Graef et al. 1985 ) without deleterious effects to the plant. Downregulation of the 9 desaturase has not been described in maize. This target was chosen for development of ribozyme downregulation technology because of the readily assayable and potentially valuable commercial phenotype that could result from successful downregulation.

The following study describes the use of ribozymes to control the expression of 9 desaturase in maize. We describe in detail the selection and optimization of ribozymes targeted to the 9 desaturase mRNA and describe experiments demonstrating ribozyme expression and control of 9 desaturase levels in transgenic maize plants. Furthermore, the majority of the data focus on one ribozyme construct shown to be most efficacious in transgenic maize plants. The inactive version of this ribozyme had no effect on stearate or the 9 desaturase mRNA or protein, suggesting that the primary mechanism of action for the ribozyme is catalytic. In this study, the ribozyme effect has been shown to be a heritable trait, indicating the utility of this technology in successive generations of transgenic plants.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Isolation and Characterization of the Maize 9 Desaturase cDNA
The sequence of the insert of cDNA clone pDAB424 and the predicted translation are shown in Figure 2. The 1621-nucleotide cDNA insert contained a 394–amino acid open reading frame (ORF), which upon translation possibly could encode a 44.7-kD polypeptide. The remaining 145-nucleotide 5' and 294-nucleotide 3' regions are predicted to be untranslated. The size of the predicted polypeptide (~45 kD) corresponds well to the size of the 9 desaturase precursors previously characterized from castor bean, cucumber, and rice (Shanklin and Somerville 1991 ; Akagi et al. 1995 ; Lindqvist et al. 1996 ). Comparison of the predicted ORF with the amino acid sequence of castor bean 9 desaturase (data not shown) indicated that extensive similarity exists between the two ORFs, with an 85% identity over the portion of the castor bean sequence representing the mature polypetide. Throughout the maize 9 desaturase sequence, the differences with castor bean are mainly the result of conservative replacements.

View larger version (74K): [in this window] [in a new window]   Figure 2. Nucleotide and Deduced Amino Acid Sequences of the Maize 9 Desaturase cDNA. The sequence of the coding strand of the cDNA insert of clone pDAB424 is shown together with the predicted translation of the only major ORF. The start of translation was chosen as the first methionine in the sequence, which was found as a good match to the Kozak consensus sequence for optimal eukaryotic translation initiation (AACAATGGC, where the boldface type indicates the codon for methionine). The putative N-terminal alanine of the mature polypeptide is underlined. The locations of the three ribozyme sequences complementary to the coding strand are shown by the dashed lines. The asterisks indicate the cleavage sites for each ribozyme.

This high degree of identity allowed the tentative identification of the ORF encoded by cDNA clone pDAB424 as the maize 9 desaturase. Comparison of the two sequences also allowed us to detect the putative chloroplast uptake processing site in the maize 9 desaturase sequence. N termini of mature imported chloroplast proteins often start with Ala–Ser, with cleavage occurring after the Met residue in the tripeptide Met–Ala–Ser. This frequently occurs 30 to 60 amino acids from the N terminus of the predicted precursor protein (De Boer and Weisbeek 1991 ). This tripeptide is present in the castor bean sequence, and a similar sequence of Val–Ala–Ser occurs in a similar position in the maize sequence. The putative N terminus of the maize 9 desaturase was therefore assigned as Ala. The resulting 363–amino acid mature polypetide has a calculated molecular mass of ~41.3 kD. Definitive assignment of the function of the encoded protein was made by overexpression in Escherichia coli of the putative mature ORF. Induced cells produced a soluble protein with high 9 desaturase activity (data not shown).

Differential Selection of Ribozymes Targeted to 9 Desaturase mRNA
RNase H cleavage assays were performed using full-length 9 desaturase cDNA transcripts to identify potential ribozyme-accessible sites on the 9 desaturase mRNA (Christoffersen et al. 1994 ). RNase H cleaves the RNA strand of an RNA:DNA hybrid (Tomizawa 1993 ) and can be used for directed DNA oligonucleotide cleavage of a target RNA molecule. Forty-nine DNA oligonucleotides, each 21 nucleotides in length, were designed to cover 108 potential hammerhead sites within the coding region of the 9 desaturase mRNA. The 9 desaturase T7 RNA transcript was prefolded in magnesium chloride to approximate the secondary structure of the RNA found in vivo. The results of this screening are shown in Figure 3. Hammerhead ribozymes were designed to the sites covered by the oligonucleotides that were most accessible in the RNase H assays. These ribozymes were then subjected to computer analysis for folding prediction (Zuker 1989 ), and those ribozymes that had significant secondary structure were rejected. Twenty sites were chosen for further study with synthetic ribozymes.

View larger version (32K): [in this window] [in a new window]   Figure 3. RNase H Accessibility Assay for 9 Desaturase RNA. DNA oligonucleotides (21-mer) were used to probe full-length T7 transcripts of 9 desaturase RNA for sites accessible to cleavage by RNase H. The 49 oligonucleotides cover 108 hammerhead sites within the coding region in 9 desaturase that were not predicted to be in secondary structures. Error bars represent the standard deviation from the average of three experiments. Numbering is based on the central position of the oligonucleotide.

Ribozymes were chemically synthesized to evaluate catalytic activity against full-length substrate RNA (Wincott et al. 1995 ). The ribozymes all contained a 3-bp stem II and 10-nucleotide arms spanning the catalytic domain. The results of the cleavage assays are shown in Figure 4. The 5' end of the transcript appears to be most amenable to cleavage, which is consistent with the RNase H data.

View larger version (30K): [in this window] [in a new window]   Figure 4. Cleavage of 9 Desaturase RNA by Synthetic Ribozymes at 26°C. Full-length substrate RNA was cleaved under ribozyme excess conditions with synthetic hammerhead ribozymes designed to the most accessible sites determined by RNase H assays. These data (±SD) were derived from the 120-min time points of three experiments. Quantitation was performed using a Molecular Dynamics PhosphorImager.

Because a limited number of ribozyme constructs could be evaluated in transgenic maize, several constructs containing multiple hammerhead ribozymes were designed. Based on cleavage activity in the synthetic ribozyme monomer screen, two multimers (Rz252 and Rz453) were engineered that placed multiple hammerhead ribozymes into a single transcript with the contiguous complementary sequence found between the ribozymes. In addition, one ribozyme monomer (Rz259) was evaluated further. A diagram of the ribozymes is shown in Figure 5A to C, and their locations within the 9 desaturase cDNA are shown in Figure 2. The monomer and multimer ribozymes were placed into several different expression cassettes to determine their catalytic activity within the context of the flanking sequences, which would be present in a transformed plant cell. One transcriptional unit fused the ribozyme monomer or multimers to the 3' end of the phosphinothricin acetyl transferase bar ORF.

View larger version (20K): [in this window] [in a new window]   Figure 5. 9 Desaturase Ribozymes and Target Sequences. (A) The 259-monomer ribozyme. (B) Multimer ribozyme to sites 252, 271, 313, and 326. (C) Multimer ribozyme to sites 453, 464, 475, and 484. Monomer and multimer hammerhead ribozymes were designed based on the synthetic ribozyme screen. The multimers contain the contiguous complementary sequence found between the cleavage sites.

A feature common to maize expression cassettes is inclusion of an intron in the 5' untranslated leader. A second transcriptional context tested mimicked the respective ribozymes flanked by the sequence of a spliced 5' leader. The transcription units were evaluated, and their in vitro activities are depicted in Figure 6. Among the ribozyme transcription units tested, the highest level of in vitro cleavage activity was produced by the 453 multimer (RPA118) in the context of the 5' spliced intron. The 259-monomer ribozyme (RPA114) also displayed its highest cleavage activity in the spliced intron context. In contrast, the 252-multimer ribozyme produced its highest level of in vitro cleavage activity when placed at the 3' end of the bar ORF (RPA85). As a result of this screening procedure, these three ribozyme transcription units were chosen for transformation into regenerable maize cultures. The two ribozyme genes that were not linked directly to the selectable marker were cloned into a plasmid containing the Basta resistance gene. These genes were expressed from a doubly enhanced cauliflower mosaic virus 35S promoter with alcohol dehydrogenase adh1 intron I and terminated with the nopaline synthase polyadenylation (nosA) signal. The bar–ribozyme fusion gene was regulated by the doubly enhanced 35S promoter, followed by the maize streak virus leader and adh1 intron I, and terminated with the nosA signal. Catalytically inactive versions of these three ribozymes also were synthesized, with mutations at conserved positions G-5 and A-14 (Hertel et al. 1992 ). The inactive ribozyme constructs are referred to as RPA119, RPA115, and RPA113 and correspond to the the active ribozymes in the order in which they are described above. Expression of all constructs was evaluated by transient assays in Black Mexican Sweet suspension cultures to ensure cellular expression of ribozyme RNA (data not shown).

View larger version (18K): [in this window] [in a new window]   Figure 6. Cleavage of 9 Desaturase RNA by Transcription Unit–Embedded Ribozymes. The 9 desaturase ribozymes were placed in several different contexts within the 35S transcription units. T7 transcripts were derived and tested for cleavage of full-length 9 desaturase RNA. When placed at the 3' end of the bar ORF, the 259-monomer and 453-multimer ribozymes cleaved <5% of the substrate RNA at 60 min in this assay. Error bars represent the standard error of triplicate assays. BAR, 3'end ORF of the bar gene; MM, multimer ribozyme; Mono, monomer ribozyme; SI, spliced intron vector.

Generation and Analysis of 9 Desaturase Transgenic Plants
Embryogenic type II callus was transformed by particle bombardment with the six ribozyme constructs described above and selected on the herbicide Basta. A single plasmid encoding both the bar and the respective ribozyme gene was used for transformation. A summary of the transformation results and description of the transformation events are shown in Table 1.

Before plant regeneration, we screened putatively transformed calli for evidence of 9 desaturase ribozyme genes by polymerase chain reaction (PCR) of genomic callus DNA. Primers were chosen that specifically amplified the 35S–adh1 intron I–ribozyme–nosA termination region of the gene insert. The results suggested a high frequency of intact gene insertions for isolates transformed with the nonfused ribozyme genes. PCR results for the fused ribozyme gene, RPA85, were more complex than those observed for the nonfused ribozymes and suggested multiple gene insertions or DNA rearrangements. Those samples with strong, predicted PCR products were selected for regeneration into plants (data not shown). In general, a total of 10 positive and two negative transgenic events per ribozyme construct were chosen for plant regeneration. On average, 15 plants were produced and analyzed for each transgenic line. Final confirmation that each regenerated line contained an intact copy of its respective ribozyme gene was made through DNA gel blot analysis of leaf tissue.

DNA gel blot analysis was used to characterize 264 primary regenerate (R₀) plants. These represented 62 individual lines transformed with one of the six different ribozyme constructs. Hybridization experiments with radiolabeled probe DNA, which was specific for each ribozyme construct, identified 40 lines containing the appropriate length integration products of 2.1 kb for the RPA85/RPA113 transformants, 1.2 kb for the RPA114/RPA115 transformants, and 1.3 kb for the RPA118/RPA119 transformants. The complexity of the integration event for each of the R₀ lines was also determined. A summary of the results from this analysis for the different ribozyme transformation series is found in Table 1.

Fatty Acid Analysis of Leaves from R₀ Ribozyme Transgenic Plants
The leaves of 428 plants from 35 lines transformed with active 9 desaturase ribozymes (RPA85, RPA114, and RPA118) and the leaves of 406 plants from 31 lines transformed with inactive 9 desaturase ribozymes (RPA113, RPA115, and RPA119) were assayed for fatty acid content. The levels of stearate observed in these leaves are presented in Table 2. Seven percent of the plants transformed with the active ribozymes had stearate levels >3%. Only 3% of the plants transformed with inactive ribozymes had stearate levels >3%, which was similar to that observed with the control plants. Two percent of the active ribozyme plants also had leaf stearate levels >5%. None of the inactive ribozyme or control plants had leaf stearate levels >5%.

The increased stearate observed in the active ribozyme plants was observed mainly in two lines, both of which were transformed with RPA85. Six of the 15 plants assayed from line RPA85-15 contained stearate levels that were between 6 and 9%, approximately fourfold greater than the average of the controls (Figure 7A). The average stearate content of the leaves from all of the plants of RPA85-15 was 3.83% (SE of 2.53). However, the average for the plants from this line with increased stearate was 6.88% (SE of 0.65), and the average for the plants from this line with normal stearate was 1.79% (SE of 0.14). When leaf analysis was repeated for the RPA85-15 plants at a later developmental stage, the stearate levels in leaves from plants previously shown to have normal stearate levels remained normal, and leaves from plants with high stearate were again found to have high levels (data not shown). A similar but less dramatic increase in stearate was observed for active ribozyme line RPA85-06. In this line, leaves with elevated stearate had a stearate content that averaged approximately twofold higher than that of the controls (data not shown). One line transformed with an inactive 9 desaturase ribozyme, RPA113, had three plants with leaf stearate levels slightly >3% (Figure 7A). The average stearate content of leaves from all of the plants of this line, RPA113-06, was 2.26% (SE of 0.65). The average stearate content of leaves from 15 control plants was 1.70% (SE of 0.6; Figure 7A).

View larger version (31K): [in this window] [in a new window]   Figure 7. Effect of Ribozyme Fusion on 9 Desaturase mRNA and Protein in Transgenic R0 Maize Leaves. Analysis of active ribozyme transgenic line RPA85-15 (bars and lanes 4 to 13), nontransformed (NT), and inactive ribozyme transgenic line RPA113-06 (bars and lanes 14 to 18). (A) Percentage of stearate content in leaves. (B) Ribozyme-mediated reductions in 9 desaturase mRNA levels in leaves. (C) Relative levels of 9 desaturase protein in leaves. Asterisks in (A) indicate plants that expressed detectable ribozyme mRNA. In (B) and (C), signals were quantified by using a scanning densitometer, and relative transcript and protein levels were calculated. The mean number (±SE) was determined for the nontransformed maize leaves.

Multimer Ribozyme Is Expressed in Active Ribozyme Transgenic Plants
An intact copy of the fused ribozyme multimer gene, RPA85, was detected by DNA gel blot analysis in R₀ plants of the high-stearate lines, RPA85-06 and RPA85-15. Within each line, plants were screened by reverse transcription–PCR for the presence of ribozyme RNA. Primers specific for amplification of the RPA85 ribozyme were used in all reactions. RPA85 ribozyme RNA was detected in each plant of the RPA85-15 line shown to contain high levels of stearic acid in the leaves (Figure 8, lanes 4, 7, 8, 10, 11, and 12). Several RPA85-15 plants with normal levels of leaf stearate also were tested for ribozyme expression. In three plants, no detectable level of ribozyme RNA was observed. RPA85 ribozyme RNA also was detected in the RPA85-06 plants shown to produce high stearate levels in their leaves (data not shown).

View larger version (109K): [in this window] [in a new window]   Figure 8. Reverse Transcription–PCR Demonstrates RPA85 Ribozyme Expression in RPA85-15 High-Stearate Plants and RPA113-06 Inactive Ribozyme Plants. Twenty microliters of the reverse transcription–PCR sample was run on a 2% 3:1 Nusieve agarose gel. Lane 1 contains the 123-bp Bethesda Research Laboratories marker; lanes 2 and 20, water controls without template DNA; lane 3, nontransformed leaf RNA; lanes 4, 7, 8, 10, 11, and 12, RPA85-15 high-stearate plants; lanes 5, 6, 9, and 13, RPA85-15 normal stearate plants; lanes 14 to 18, RPA113-06 inactive ribozyme plants; and lane 19, RPA85 plasmid.

9 Desaturase mRNA Is Reduced in Leaves of High-Stearate Transgenic Plants
Evidence of ribozymes acting on the 9 desaturase target would be demonstrated by a reduction in the level of 9 desaturase mRNA. RNA gel blot analysis was performed comparing the 9 desaturase mRNA levels of 10 plants from line RPA85-15 with nontransformed control plants. The blots were hybridized sequentially with radioactive probes specific for 9 desaturase RNA and a constitutively expressed actin RNA. The level of 9 desaturase mRNA was normalized to the level of actin transcript within each sample. This measurement distinguished variation in 9 desaturase mRNA levels as a result of loading errors from true ribozyme-mediated RNA reductions. Using densitometer readings, we calculated a ratio for each sample. Ranges in 9 desaturase/actin ratios from 0.35 to 0.53, with an average of 0.43, were calculated for the RPA85-15 high-stearate transgenic plants, whereas the average 9 desaturase/actin ratio for the nontransformed plants was 1.7. Comparing the average 9 desaturase/actin ratio between nontransformed controls and RPA85-15 high-stearate plants, we demonstrated a 3.9-fold reduction in RPA85-15 9 desaturase mRNA. An apparent threefold reduction in 9 desaturase mRNA level was observed for RPA85-15 high-stearate transgenic plants when 9 desaturase/actin ratios were compared between RPA85-15 high-stearate (Figure 7B, plants 4, 7, 8, 10, 11, and 12) and normal-stearate RPA85-15 plants (Figure 7B, plants 5, 6, 9, and 13). A similar reduction in 9 desaturase mRNA level was measured for the high-stearate plants of the RPA85-06 line (data not shown).

Normal Levels of 9 Desaturase mRNA Are Detected in Leaves of Inactive Ribozyme Transgenic Plants
Plants transformed with inactive versions of 9 ribozymes were produced and analyzed for the high-stearate phenotype. In these experiments, inactive controls were used to discriminate between effects resulting from ribozyme (catalytic) activity and those due to hybridization (antisense) effects. Inactive ribozymes cannot cleave their target RNA; however, they still specifically bind to the target sequences. Downregulation effects that are the result of ribozyme activity should be present only in active ribozyme plants. Effects that are the result of an antisense interaction would be observed in both active and inactive plants. A slightly elevated stearate level was measured in the leaf of one RPA113-06 plant (Figure 7A, inactive plant 14). Although the stearate level fell within the range of controls, RNA gel blot analysis was performed on a group of plants from this transgenic line. RPA113 is the inactive version of the RPA85-fused ribozyme multimer gene (Table 1).

Ribozyme expression in the RPA113-06 plants was evaluated by reverse transcription–PCR. The plant with a slightly elevated level of leaf stearate did not express detectable levels of ribozyme RNA (Figure 8, lane 14). Three of five plants tested did express RPA113 ribozyme RNA; however, no alterations in stearate levels, 9 desaturase mRNA, or protein levels were observed (Figure 7A and Figure 7B). This result contrasts with that observed for plants expressing active RPA85 ribozyme RNA. In those plants, increases in leaf stearate correlated with ribozyme expression and reductions in 9 desaturase mRNA and protein levels. These data support the hypothesis that the high-stearate phenotype displayed in the active RPA85 transgenic plants is associated with the presence of ribozyme activity.

Immunological Detection of 9 Desaturase in Maize Leaves
Further evidence of downregulation by ribozyme activity would be reductions in the 9 desaturase protein level in R₀ leaves of the RPA85-15 high-stearate plants. The 9 desaturase levels were monitored in nontransformed (Hi-II) plants, transformed active ribozyme line RPA85-15, and transformed inactive ribozyme line RPA113-17. The level of protein in leaves was assessed by extracting and enriching for 9 desaturase protein, separation by SDS-PAGE, and identification of protein by immunoblotting. A single protein with an apparent molecular mass of 38 kD was detected by antiserum raised against maize 9 desaturase in leaf extracts (Figure 7C). Figure 7C shows an analysis of the 9 desaturase protein from leaves of 10 plants transformed with the active ribozyme RPA85-15. Leaves from six of these plants had an apparent 50% reduction of 9 desaturase (Figure 7C, lanes 4, 7, 8, 10, 11, and 12), whereas other plants produced leaves with normal levels of 9 desaturase (Figure 7C, lanes 5, 6, 9, and 13) when compared with nontransformed control plants (Hi-II). Interestingly, the four plants that had normal levels of 9 desaturase and contained the active ribozyme gene were not shown to express ribozyme RNA (Figure 8). This may have been caused by the number of gene insertions (complex event) found for this particular event.

Inheritance of the High-Stearate Phenotype
The leaf tissues of R₁ plants derived from crosses with RPA85-15 high-stearate plants were subjected to FAME (Browse et al. 1986 ) and DNA and immunoblot analyses. The DNA gel blot analysis and FAME results from five crosses are summarized in Table 3. The percentage of plants with high stearate levels ranged from 20 to 50%. An intact copy of the gene coding for the fused 252-multimer ribozyme was present in a larger percentage of the plants (70 to 88%). Therefore, some that did not have the high-stearate phenotype did have a copy of the ribozyme gene. However, all of the plants with high leaf stearate levels that were analyzed by DNA gel blot analysis contained a copy of the ribozyme gene.

Figure 9A and Figure 9B compare the stearate levels and the relative 9 desaturase levels in leaves of R₁ plants from the RPA85-15 crosses. A 40 to 50% reduction of 9 desaturase protein was observed in plants that had increased levels of leaf stearate. This reduction was comparable with that observed in R₀ high-stearate plants. The decrease in 9 desaturase protein in R₁ plants correlates with an increase in leaf stearate and the presence of an intact ribozyme gene.

View larger version (26K): [in this window] [in a new window]   Figure 9. Stearic Acid Levels and Relative Levels of 9 Desaturase in Leaves of R1 Plants from Crosses with RPA85-15. Analysis of leaves from R1 plants from two crosses: (1) OQ414 x RPA85-15.06 or RPA85-15.11 (1 to 6) and (2) RPA85-15.06 and RPA85-15.07, or RPA85.10 selfed lines (7 to 14). (A) Percentage of stearic acid content. (B) Relative protein levels of 9 desaturase. Proteins were separated by SDS–PAGE (12% polyacrylamide gel) and transferred to ECL nitrocellulose membranes. The blots were incubated with antiserum raised against maize 9 desaturase. Signals were quantified by a scanning densitometer. Protein levels were determined relative to purified and quantified overexpressed 9 desaturase from E. coli.

A higher than expected percentage of plants containing the riboyzme transgene was noted in crosses made with inbred line OQ414. For these determinations, a limited sample size, eight and 10 plants, was analyzed, perhaps biasing the percentage of plants shown to contain transgenes. Also possible is integration of the transgenes at two loci resulting in non-Mendelian segregation. This atypical segregation in R₁ maize transgenic plants frequently has been noted in plants created for other studies (data not shown).

Fatty Acid Analysis of R₁ Seed Embryos
Fatty acid composition was determined for each zygotic embryo excised from the seeds of RPA85-15 plants. All of the embryos tested had normal stearic acid levels ranging between 1 and 2%. Mature seed embryos from lines transformed with active ribozymes also were assayed for fatty acid composition. A total of 582 seed embryos from 63 plants regenerated from eight separate lines (including RPA85-15) was tested, and all had normal stearic acid levels.

Immunological Detection of 9 Desaturase in R₁ Seed
We assessed the level of 9 desaturase synthesis in zygotic embryos of maize seeds by extracting total protein, separation by SDS-PAGE, and identification of proteins by immunoblotting. As would be predicted based on the results of fatty acid composition, none of the embryos tested showed an apparent reduction of 9 desaturase protein. These included the six plants from line RPA85-15, which had a 50% protein reduction in the leaf tissue (data not shown). Interestingly, based on total protein, the relative amount of 9 desaturase found in the zygotic embryos was approximately eightfold to 10-fold greater than the relative amount of 9 desaturase found in the leaves. The lack of variation in the accumulation of the 9 desaturase protein may have been attributed to either a rapid turnover of the enzyme in the leaf tissue as compared with the zygotic embryo or possibly an increase in the expression of the 9 desaturase gene within the seed.

To examine the expression of the 9 desaturase gene in maize plants, seeds (20 to 25 days after pollination) and leaf tissue were analyzed and compared for the presence of 9 desaturase transcripts. The results showed that 9 desaturase mRNA accumulated in seeds to a higher level than in the leaf material (data not shown). The fact that 9 desaturase mRNA is present at higher levels in the seed compared with the leaf suggested that 9 desaturase protein can be synthesized in the seeds at a higher level compared with the leaves and is not a result of instability of the protein in the leaf tissue.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Ribozymes represent an alternative to antisense and cosuppression strategies for downregulation of gene expression in plants. There are a few reports demonstrating ribozyme activity in plant cells, and only two in which the activity is against endogenous targets (Borovkov et al. 1996 ; McIntyre et al. 1996 ). Often, in these studies, ribozymes are embedded singly or as multimers in larger antisense sequences. Under these conditions, the unique contribution of the ribozyme to downregulation is difficult to demonstrate. Accessibility of the targeted cleavage site or sites has not been demonstrated consistently. The relative efficiency of cleavage at targeted sites is not known, and the catalytic efficiency of the ribozyme in its ultimate transcriptional context has not been determined. For these reasons, a stringent, sequential, and thorough evaluation of potential cleavage sites and ribozymes would provide the basis for successful demonstration of the desired phenotype.

Two hundred and fifty hammerhead ribozyme cleavage sites were identified in the maize 9 desaturase mRNA sequence. Accessibility of the sites in the mRNA was determined using RNase H assays. Based on these assays, 20 readily accessible sites were identified. Ribozymes targeted to these sites were synthesized with 10 complementary bases on either side of the catalytic domain (10/10 arms). Cleavage activity of these synthetic ribozymes in vitro was measured at 120 min on the full-length substrate. The highest rates of cleavage were observed in the 5' end of the mRNA, exceeding 50% cleavage at 120 min. Accessibility of the sites for cleavage as measured by the RNase H assay did not explain the majority of the variation observed in the long substrate cleavage assay. The correlation between the results of these two assays was ~0.40. Therefore, it may be critical to measure the accessibility of the sites as well as the in vitro cleavage of the ribozymes to enhance the probability of successful downregulation of the target. Based on the results of the cleavage assay, three lead ribozymes were designed, with two being multimers and one being a monomer. The 259 monomer, 453 multimer, and 252 multimer were tested in several transcriptional contexts.

The most catalytically efficient version of each of the three lead ribozymes was advanced for transformation into regenerable cultures. The transcriptional context of the ribozyme had a dramatic impact on its in vitro cleavage efficiency, as has been observed by others (Thompson et al. 1995 ). Included in the set is the 252 multimer linked 3' to the ORF of the bar selectable marker gene (RPA85). Fusion of the ribozyme gene to the ORF of a functional gene has been shown to be an effective strategy for expressing ribozymes in mammalian systems (Cameron and Jennings 1989 ; Borovkov et al. 1996 ). This particular construct has several advantages. First, selection and recovery of transgenic events simultaneously select for ribozyme expression. Second, the gene fusion may enhance ribozyme stability by protecting these typically short RNAs from nuclease attack. Translatable mRNAs have been shown to be more stable in plants (van Hoof and Green 1996 ). Finally, effectively extending the 3' untranslated region of the bar gene by the addition of the ribozyme sequence may increase the half-life of the bar mRNA, thereby enhancing the apparent expression level of the bar gene and Basta selection efficiency. The apparent transformation efficiency with catalytically active and inactive versions of this construct was >75% higher than was the average of all other constructs in this study (Table 1).

The most appropriate controls for these active ribozymes are transgenic plants containing catalytically inactive versions of the identical ribozyme rather than antisense sequences. Production of transgenics expressing either active or inactive ribozymes allowed us to determine whether ribozymes can effectively downregulate expression of an endogenous gene in the absence of a substantial antisense effect. Downregulation observed in transgenics with catalytically inactive ribozymes embedded in short antisense sequences is most likely attributable to antisense effects. Downregulation differences between transgenic plants expressing catalytically active versus inactive versions of the ribozymes must be attributable to catalytic activity, even in the absence of detectable cleavage products.

More than 330 independent Basta-resistant calli were obtained from selection. The presence of the ribozyme transgene, which was demonstrated by PCR amplification from callus DNA, was the criterion used for advancing transgenic calli to regeneration. Between nine and 13 transgenic events or lines were regenerated per construct, at least five of which contained the gene of interest, and at least one of which was a negative control. By using DNA gel blot analysis of leaf tissue, we showed that 23 active and 17 inactive ribozyme lines carry intact copies of the genes of interest.

More than 850 plants were regenerated and grown to reproductive maturity in the greenhouse. In regenerated plants, there were multiple means to characterize the expression and phenotypic impact of the introduced ribozyme constructs. The presence of the gene was demonstrated using gel blot analysis of genomic DNA extracted from leaf tissue. Expression of the ribozyme was confirmed using reverse transcription–PCR. The 9 desaturase transcript levels relative to internal controls were determined using RNA blot analysis. The 9 desaturase protein concentration was quantitated using immunoblot analysis. Finally, gas chromatography–fatty acid methyl ester (FAME) analysis (Browse et al. 1986 ) was used to determine alterations in fatty acid composition. Each of these techniques was equally applicable to leaf and seed tissue.

Gas chromatography–FAME analysis is sensitive, and differences in stearic acid content of <1% can be shown to be significant at the 0.05 level. Gas chromatography–FAME analysis of leaf tissue of >400 R₀ plants revealed only two lines (RPA85-06 and RPA85-15) with increased stearic acid. Both lines were transformed with the RPA85 construct (252 multimer fused 3' to the bar ORF). These two lines (RPA85-06 and RPA85-15) showed two- to fourfold increases in stearic acid. As we and others have observed with antisense downregulation, not all R₀ plants within a transformation event displayed the altered phenotype (Tabler 1993 ; Bourque 1995 ; T. Skokut and O. Folkerts, unpublished data). Fifteen plants each from lines RPA85-06 and RPA85-15 were analyzed, and six plants per line displayed the high-leaf-stearate phenotype. The RPA85-15 integration pattern is complex, with >10 DNA fragments hybridizing on DNA gel blots. Integration of RPA85-15 appears to be divided into two groups: one appears to be stable through transmission to progeny, and one is unstable. This complexity also may contribute to the variability seen among the R₀ plants of RPA85-15. Although the integration pattern of line RPA85-06 is only moderately complex, with four DNA fragments hybridizing on DNA blots, a similar range in variability was observed among R₀ plants. For either line, the observed plant-to-plant variability could be attributed to gene silencing by multiple copies of the ribozyme transgene. In a concurrent study using antisense to downregulate maize 9 desaturase activity, a comparable number of downregulated events were identified (two of 16 transgenic lines). Common to this study was the level of variability observed among plants within a downregulated line (data not shown).

Further analysis of leaf tissue of the high-stearate plants within RPA85-15 revealed detectable levels of ribozyme RNA, nearly a fourfold reduction in 9 desaturase mRNA, and a 50% reduction in 9 desaturase protein. An identical but less dramatic alteration in phenotype, 9 desaturase mRNA, and protein levels was observed for the six high-stearate plants within the RPA85-06 line (data not shown). All high-stearate plants within line RPA85-15 expressed ribozyme RNA. R₀ plants within line RPA85-15 that failed to display the high-stearate phenotype did not express detectable levels of ribozyme RNA, with the exception of one plant. It is unclear why no increase in leaf stearate was detected in this plant. Several possibilities may explain these results. Nonquantitative reverse transcription–PCR was used to confirm ribozyme expression; therefore, the absolute level of active ribozyme produced by each plant is unknown. A mutation inactivating the ribozyme could have been introduced during transformation and regeneration of the plant. Finally, precedence for a lack of correlation between antisense or ribozyme RNA expression and phenotype has been described by de Feyter et al. 1996 . To our knowledge, there are no reports addressing the correlation between ribozyme expression level and degree of endogenous target reduction in plants.

Similar results were not observed with inactive controls. In those lines, there was no alteration in phenotype, reduction in 9 desaturase mRNA, or reduction in 9 desaturase protein associated with detectable levels of ribozyme RNA. A marginal increase in leaf stearate was detected in one plant transformed with RPA113 (the inactive version of RPA85); however, no detectable ribozyme expression was observed for this plant.

The results for the active (RPA85) and inactive (RPA113) ribozyme transgenic plants suggest that the active riboyzme transgene is responsible for the high-stearate phenotype in the active ribozyme plants. A small number of independent transformation events were analyzed for each construct (13 for RPA85 and nine for RPA113). However, for those plants in the RPA85-06 and RPA85-15 lines that had high leaf stearate levels, there was a detectable ribozyme and a perfect correlation to reduced 9 desaturase mRNA and reduced concentration of 9 desaturase protein in the leaf. Inactive ribozyme RNA was detected in three of five RPA113-06 plants but was not associated with any detectable changes in phenotype, 9 desaturase mRNA, or protein levels. The lack of inhibitory effect by the inactive ribozyme suggests that the RPA85 ribozyme acts by a catalytic mechanism rather than an antisense interaction. There have been other reports in which ribozymes embedded into relatively long antisense molecules show a significant antisense effect (Perriman et al. 1993 ; de Feyter et al. 1996 ). Here, we intentionally have minimized the extent of complementarity to increase reliance on the catalytic nature of the ribozymes. Efforts were made to demonstrate in vivo cleavage products in RNA extracted from the leaves of the high-stearate plants. The experiments could not define clearly whether the putative cleavage products were generated in vivo or in vitro during the amplification procedures (data not shown).

Despite the alteration in leaf fatty acid profiles among R₀ progeny of RPA85-15, there was no modification of fatty acid profiles for seed embryos. Fatty acid biosynthetic genes are expressed constitutively; however, they appear to be coordinately upregulated in seed tissues in which oil is deposited. With the significant upregulation of the 9 desaturase gene that occurs in embryo tissue beginning as early as 10 days after pollination, the 9 desaturase protein and mRNA can be 10 times more abundant in the embryo 20 days after pollination than in leaf tissue (A. Owens Merlo and S. Young, unpublished data). To demonstrate an alteration of phenotype in the seed, an increase in ribozyme activity comparable with the increase of 9 desaturase mRNA would be required. This observation combined with the fact the constitutive promoter used in this study is not highly expressed in the embryo likely explains the lack of modification of the seed oil composition. Another possible explanation is that turnover rates of 9 desaturase mRNA and protein may have an impact on demonstration of phenotypes. In one report, turnover rates of target mRNA have been identified as critical to ribozyme activity (Bertrand et al. 1992 ).

R₁ seed of all experimental lines was produced either by crosses between R₀ progeny and proprietary inbreds (testcross) or self- or sibling pollinations among R₀ plants within transformation events. Both types of progeny were produced from RPA85-15. The results from DNA gel blot, immunoblot, and gas chromatography–FAME analyses of these plants confirm the results obtained with the R₀ plants. The high-stearate phenotype (5 to 8%) was observed consistently in leaves of plants that by DNA gel blot analysis were positive for the gene of interest. Plants displaying the high-stearate phenotype also displayed the previously observed 40 to 50% reduction in 9 desaturase protein concentrations. By expressing the fused 252-multimer ribozyme, the observed downregulation of 9 desaturase and the resulting high-stearate phenotype were shown to be stable and heritable.

In this study, we have demonstrated that ribozymes can be used to downregulate endogenous gene expression in maize. A stringent, sequential, and critical evaluation of target cleavage sites and ribozymes was performed to identify the most efficient ribozymes for evaluation in transgenic maize. Analysis of the resulting transgenic plants produced results supporting the conclusion that the observed downregulation was associated with the catalytic activity of the ribozymes. Inactive ribozyme controls also were used to provide further evidence for this conclusion. In contrast to Arabidopsis and tobacco, rapid generation of multiple transgenic maize events is not possible. In this study, too few events could be produced and evaluated to prove statistically that the observed downregulation of 9 desaturase in the leaves of RPA85 plants was due to catalytic cleavage. However, correlative data indicate that the expression of the active multimer fusion was responsible for the effect.

We believe several factors impacted the downregulation observed. Stabilization of the ribozyme by placing it 3' to an actively transcribed coding region resulted in the most dramatic levels of downregulation. Additional benefits are realized when the coding region is the selectable marker gene. We also believe that a thorough knowledge of the expression pattern of the target is important to successful ribozyme-mediated downregulation. Finally, to ensure a modification of seed fatty acid composition using ribozyme downregulation of 9 desaturase, or another fatty acid biosynthetic gene, a strong embryo-specific promoter is likely to be required to drive expression of the ribozyme.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Reagents
Restriction and modification enzymes were obtained from Gibco BRL and Boehringer Mannheim. Plasmid minipreparation kits came from Promega, and large-scale plasmid preparation kits came from Qiagen (Chatsworth, CA). Labeled nucleotides came from Du Pont–New England Nuclear (Beverly, MA) and Amersham.

Isolation of Stearoyl–Acyl Carrier Protein 9 Desaturase cDNA
Maize seed embryos of genotype CS608 were harvested at 20 days after pollination, and total RNA was isolated from 2 g of embryos that was ground to a fine powder in liquid nitrogen by the method of Murray et al. 1990 . Poly(A)⁺ RNA was purified using oligo(dT)–cellulose chromatography (type III; Collaborative Research, Chicago, IL) according to Sambrook et al. 1989 . Five micrograms of poly(A)⁺ RNA was converted to cDNA by using a UNI-ZAP cDNA synthesis kit and cloned in ZAPII, according to the protocols provided by the supplier (Stratagene, La Jolla, CA). In vivo rescue conversion of the library to plasmid form was performed by a modification and scale up of the in vivo rescue procedure for individual cDNA clones. A maize stearoyl–acyl carrier protein (ACP) probe was isolated from the plasmid library by amplification with the following degenerate primers: OF225 (5'-GARGARAAYMGNCAYGG-3') and OF226 (5'-TCRTGN-CKYT TYTCRTC-3'), based on the conserved amino acids in the castor stearoyl–ACP 9 desaturase implicated in the binding of the di-iron oxo group described by Shanklin and Somerville 1991 . Screening of 400,000 phage of the ZAP embryo cDNA library with the radiolabeled probe was done according to Ausubel et al. 1989 and resulted in identification of ~280 positive isolates. Sixteen positives were subjected to two rounds of plaque purification, in vivo rescue, and restriction analysis of the encoded inserts. The sequence of the longest clone (pDAB424) was determined using an Applied Biosystems Inc. (Foster City, CA) PRISM sequencing kit and an ABI370 (Applied Biosystems Inc.) sequencer.

Selection of Accessible Sites
There are 250 hammerhead sites in the maize 9 desaturase mRNA sequence. The secondary structure of the maize 9 desaturase mRNA was assessed by computer analysis using the MFOLD algorithms developed by M. Zuker (Zuker 1989 ). Regions of the mRNA that did not form secondary folding structures with RNA–RNA stems of more than eight nucleotides and that contained potential hammerhead ribozyme cleavage sites were identified.

RNase H Accessibility Assays
One hundred and eight hammerhead sites were tested for oligonucleotide accessibility by RNase H assays. Forty-nine DNA oligonucleotides, each 21 nucleotides long, were tested. Many of these cover more than one hammerhead site. The numerical designation of the oligonucleotide refers to the central nucleotide corresponding to the position covered in the 9 desaturase cDNA. RNA was screened for accessible cleavage sites by the method reported by Jarvis et al. 1996 . Briefly, 21-mer DNA oligonucleotides spanning ribozyme cleavage sites were synthesized (Macromolecular Resources, Fort Collins, CO). Target RNA used in this study was full length and contained cleavage sites for all of the hammerhead ribozymes targeted against 9 desaturase RNA. A template containing a T7 RNA polymerase promoter upstream of the 9 desaturase target sequence was polymerase chain reaction (PCR) amplified from a cDNA clone. Target RNA was transcribed from this template by using T7 RNA polymerase as described for the Maxiscript kit (Ambion, Inc., Austin, TX). The transcript was internally labeled during transcription by including -³²P–dCTP as one of the four ribonucleoside triphosphates. After transcription at 37°C for 2 hr, the mixture was treated with DNase I to digest away the DNA template. The resulting transcription mixture was resolved on a denaturing polyacrylamide gel. A band corresponding to full-length RNA was isolated from a gel slice, the RNA was precipitated with isopropanol, and the pellet was stored at 4°C.

Four to 10 nanograms of labeled transcript was mixed with 10 µM oligonucleotide in 1 x reaction buffer (20 mM Tris-HCl, pH 7.9, 100 mM KCl, 10 mM MgCl₂, 0.1 mM EDTA, and 0.1 mM DT T) and 0.8 units of RNase H (Gibco BRL) to a final volume of 10 µL and incubated for 10 min at 37°C. Reactions were terminated by addition of an equal volume of stop buffer (95% formamide, 20 mM EDTA, 0.5% SDS, 0.1% xylene cyanol, and 0.1% bromophenol blue), heated to 90°C, and separated on 4% denaturing polyacrylamide gels. Each reaction was performed in duplicate. The percentage of the substrate cleaved was quantified using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

Long Substrate Cleavage Assays
Hammerhead ribozymes were designed to the sites covered by the oligonucleotides that cleaved best in the RNase H assays. These ribozymes were then subjected to analysis by computer folding programs (Zuker 1989 ). The ribozymes that had significant secondary structure were rejected.

Ribozyme RNAs were chemically synthesized. The general procedures for RNA synthesis have been described by Scaringe et al. 1990 and Wincott et al. 1995 . The arm length chosen for the initial screening was 10/10. Ribozymes were purified by denaturing gel electrophoresis, using standard methods. The target 9 desaturase transcript was made as described above.

Ribozyme cleavage reactions were performed under ribozyme excess conditions (Herschlag and Cech 1990 ). Briefly, 1 mM ribozyme and <10 nM internally labeled target RNA were denatured separately by heating to 65°C for 2 min in the presence of 50 mM Tris-HCl, pH 7.5, and 10 mM MgCl₂. The RNAs were renatured by cooling to the cleavage reaction temperature (26°C) for 10 to 20 min. Cleavage reactions were initiated by mixing the ribozyme and target RNA at the reaction temperature. Aliquots were taken at regular time intervals and quenched by adding an equal volume of stop buffer (90% formamide, 20 mM EDTA, 0.1% xylene cyanol, and 0.1% bromophenol blue). The samples were heat denatured and resolved on a 4% sequencing gel. Quantitation was performed using a Molecular Dynamics PhosphorImager and imaging system.

Construction of Ribozyme Transcription Units
Ribozymes were expressed under the control of the doubly enhanced cauliflower mosaic virus 35S promoter. There were two basic vectors into which ribozymes were placed. The first vector, pDAB353, contained the doubly enhanced 35S promoter, alcohol dehydrogenase1 (adh1) intron I, the ß-glucuronidase (gus) reporter gene, and the nopaline synthase polyadenylation signal (nosA) in a pUC19 vector background. Ribozyme-encoding oligonucleotides were inserted by digesting the vector with BamHI and SstI to remove the gus reporter and ligated to ribozyme-encoding oligonucleotides with compatible ends. All of the ribozymes were first cloned into pDAB353 to facilitate their cloning into other expression plasmids. To construct a version of pDAB353, which mimicked a spliced intron transcript, the adh1 intron I was removed by XbaI-BamHI digestion and replaced with the following sequence: 5'-TCTAGAGGATCAAGTGCAAAGCTG-CGGACGGATCC-3'.

The parent plasmid pDAB367 contains the phosphinothricin acetyl transferase (bar) gene. The bar gene, which encodes for resistance to Basta (Hoechst Aktiengesellschaft, Germany), is regulated by the doubly enhanced 35S promoter, followed by the maize streak virus leader, adh1 intron I, and the nosA signal on a pUC19-based plasmid backbone. All of the ribozyme transcription units (except the bar fusion) were cloned into pDAB367 to facilitate stable transformation into Black Mexican Sweet and embryogenic callus. They were inserted into a HindIII site 5' to the doubly enhanced 35S promoter of the bar gene. Plasmid pDAB367 was digested with NotI and filled in with the Klenow fragment of DNA polymerase I to make a blunt acceptor site. This was then recut with HindIII. The ribozyme-containing plasmids (pDAB353 versions) were cut with EcoRI, filled in with the Klenow fragment, and recut with HindIII. The insert containing the entire ribozyme transcription unit was gel purified and ligated into the pDAB367 vector. The constructs were checked by digestion with SgfI-HindIII and XbaI-SstI and sequenced through the ribozyme for confirmation.

Ribozymes linked to the selectable marker gene bar were made as follows. Plasmid pDAB367 was partially digested with BglII, and the singly cut plasmid was gel purified. This product was then digested to completion with EcoRI, and the uppermost band was gel purified. This removed the nosA 3' untranslated region just past the end of the bar open reading frame (ORF). A BamHI-EcoRI digest of any of the ribozymes cloned into pDAB353 removed both the ribozyme and the nosA signal, which were then cloned directly into the bar plasmid vector.

Ribozymes Targeting 9 Desaturase
Monomers
Monomer ribozyme transcription units were constructed as follows. cDNA oligonucleotides were treated individually with T4 polynucleotide kinase in 1 x kinase buffer (Gibco BRL). They were combined for annealing using the following temperature scheme: 90°C for 2 min, 65°C for 2 min, 37°C for 5 min, and 25°C for 5 min; then they were placed on ice for at least 5 min. All ribozyme genes were designed with BamHI-compatible sites at the 5' end and SstI sites at the 3' end. The sequence of the 259 active ribozyme gene with a 3-bp stem II is 5'-GGATCCCCT TGGTGGACTGATGAGGCGAAA-GCCGAAACGGCGGACGGAGCTC-3'. This was cloned into pDAB353 as a BamHI-SstI fragment. It subsequently was introduced into the plant transformation vector pDAB367 as a HindIII fragment. This ribozyme was designated RPA114. Inactive hammerhead ribozyme genes were synthesized by substituting a T residue for the G at position 5 and a T for the A at position 14 (numbering from Hertel et al. 1992 ) such that the catalytic core reads 5'-CT TATGAGGCGAAAGCCGAT-3'. The inactive version of RPA114 was named RPA115.

Multimers
The multimer ribozymes were made by annealing complementary oligonucleotides, filling in with the Klenow fragment, restricting the DNA, and cloning it into the appropriate expression vector. The sequence of the oligonucleotide encoding ribozyme units of the 252– active multimer 3-bp stem II is as follows: 5'-GGATCCGGT-GGCAT TGCTGATGAGGCGAAAGCCGAAATGTGTAACCTGCTGATG-AGGCGAAAGCCGAAACATGTACCTCCCT TGGAGGAGCAAATGGCT-TCT TAT TCTCCTGATGAGGCGAAAGCCGAAACCT TGGTGGAGACG-GCGCTGATGAGGCGAAAGCCGAAACGTCATGGAGAGCTC-3'. This ribozyme gene, when fused to the 3' end of the bar ORF, was designated RPA85. The inactive complement of this riboyzme gene was named RPA113.

The sequence of the 453–active multimer 3-bp stem II is 5'-GGATCCGT TCTCTCTGATGAGGCGAAAGCCGAAAGCTCCTCTGAT-GAGGCGAAAGCCGAAAACT TCATCATCTGATGAGGCGAAAGCCGA-AAAATCCT TCACTGATGAGGCGAAAGCCGAAATGCTGGAGCTC-3'. This ribozyme gene was cloned into pDAB353 and recloned into pDAB367 as a separate transcription unit and designated RPA118. The inactive version of this ribozyme gene was named RPA119.

Cleavage of Transcription Unit–Based Ribozymes
Transcription unit–embedded ribozyme genes were synthesized from PCR-generated DNA templates by using bacteriophage T7 RNA polymerase (Milligan and Uhlenbeck 1989 ). Cleavage assays with T7 transcripts made from these multimer-containing transcription units were performed as described above for the synthetic ribozymes.

Transformation of Embryogenic (Regenerable) Cultures
A total of 140 µg of plasmid DNA was precipitated onto the surface of 60 mg of sterile, spherical gold particles, either 1.0 µm in diameter (Bio-Rad) or 1.5 to 3.0 µm in diameter (Aldrich Chemical Co.) per ribozyme construct (RPA85, RPA113, RPA114, RPA115, RPA118, and RPA119; see Table 1) before helium blasting (Pareddy et al. 1997 ). Precipitation was accomplished by adding 74 µL of 2.5 M calcium chloride and 30 µL of 0.1 M spermidine (free base) to 140 µg of plasmid DNA diluted with sterile H₂O to a volume of 300 µL. The mixture was vortexed for 30 sec, and then the gold–DNA was allowed to settle out of the solution. The resulting clear supernatant was removed, and the DNA-coated gold particles were resuspended in 1 mL of absolute ethanol. This suspension then was diluted 1:4 to obtain 15 mg of DNA–gold per mL of ethanol.

Approximately 600 mg of embryogenic type II callus tissue was spread onto 4SM "osmotic medium" (N6 salts and vitamins [ Chu et al. 1975 ], 1 mg/L 2,4-D, 0.2 M sorbitol, 0.2 M mannitol, and 7 g/L Gelrite [Schweizerhall, Inc., South Plainfield, NJ], pH 5.8) in 60 x 15-mm culture dishes. After a 4- to 16-hr pretreatment, the tissue was transferred to 60 x 20-mm culture dishes containing blasting medium (4SM osmotic medium solidified with 20 g/L tissue culture agar rather than 7 g/L Gelrite). The tissue was covered with a 104 -µm stainless steel screen cage and placed under a vacuum in the main chamber of