Because of the limited lysine content in corn grain, synthetic lysine supplements are added to corn meal-based rations for animal feed. The development of biotechnology, combined with the understanding of plant lysine metabolism, provides an alternative solution for increasing corn lysine content through genetic engineering. Here, we report that by suppressing lysine catabolism, transgenic maize kernels accumulated a significant amount of lysine. This was achieved by RNA interference (RNAi) through the endosperm-specific expression of an inverted-repeat (IR) sequence targeting the maize bifunctional lysine degradation enzyme, lysine-ketoglutarate reductase/saccharopine dehydrogenase (ZLKR/SDH). Although plant-short interfering RNA (siRNA) were reported to lack tissue specificity due to systemic spreading, we confirmed that the suppression of ZLKR/SDH in developing transgenic kernels was restricted to endosperm tissue. Furthermore, results from our cloning and sequencing of siRNA suggested the absence of transitive RNAi. These results support the practical use of RNAi for plant genetic engineering to specifically target gene suppression in desired tissues without eliciting systemic spreading and the transitive nature of plant RNAi silencing.
In plants, lysine is derived from aspartate in a number of enzymatic steps, with dihydrodipicolinate synthase (DHDPS) catalysing the first committed step for lysine production and lysine-ketoglutarate reductase/saccharopine dehydrogenase (LKR/SDH), a bifunctional enzyme, responsible for lysine catabolism (Galili, 2002). Lysine regulates its own accumulation by feedback inhibiting DHDPS and forward activating LKR/SDH (Galili, 2004). Since lysine is an important quality trait in crop plants, these two key steps in lysine metabolism have been the subject of studies for modifying seed lysine content. Seed-specific expression of a bacterial lysine feedback-insensitive DHDPS resulted in up to a 100-fold increase in free lysine accumulation in seeds of transgenic Arabidopsis, canola and soybean (Falco et al., 1995; Zhu and Galili, 2002). In contrast, significant accumulation of lysine in maize seed could only be achieved through deregulating lysine anabolism in the embryo (Mazur et al., 1999; Huang et al., 2005). This could be due to the presence of considerable amount of LKR/SDH activity in developing maize endosperm (Kemper et al., 1999) and it is likely that lysine catabolism rather than lysine biosynthesis modulates the lysine levels in developing maize endosperm. In this study we investigated whether the suppression of LKR/SDH alone in maize endosperm by RNA interference (RNAi) could lead to the accumulation of free lysine in mature kernels. Previously, the reduction of AtLKR/SDH either by T-DNA insertional mutagenesis or RNAi was shown to cause the rise of free lysine levels in Arabidopsis seeds (Zhu et al., 2001; Zhu and Galili, 2004).
RNAi results in gene-specific down-regulation triggered by double-stranded RNA (dsRNA) (Fire et al., 1998). While in Caenorhabditis elegans, Drosophila melanogaster and other animal organisms, RNAi may be achieved by the injection of dsRNA (Fire et al., 1998; Kennerdell and Carthew, 1998), expression of dsRNA using transgenes with inverted repeats (IR) of target sequences through Agrobacterium-mediated transient or stable transformation is usually how RNAi is deployed in plants (Schob et al., 1997; Chuang and Meyerowitz, 2000). In plants, the transgenic dsRNA are thought to be processed by RNase III-type enzymes called Dicer-like (DCL) enzymes and produce small short-interfering RNA (siRNA) (Xie et al., 2004; Bordersen and Voinnet, 2006). The siRNA then serve as the sequence specific guides to cleave homologous mRNA and direct DNA or histone modification at homologous loci and cause gene silencing at the transcriptional or post-transcriptional level (Bordersen and Voinnet, 2006). Although gene knockout by antisense suppression (Oeller et al., 1991) and sense co-suppression (Napoli et al., 1990) has long been a useful tool to study gene function in plants, gene suppression by dsRNA is far more efficacious (Chuang and Meyerowitz, 2000). Since all of these approaches utilize similar RNA silencing pathways that require the dsRNA intermediate, this is likely due to prompt dsRNA formation by transcribing directly from the IR transgene. In contrast, it has been proposed that the formation of dsRNA in antisense suppression and sense co-suppression approaches depends on either the transcription of an unintentional IR T-DNA insertion structure or the conversion of a single-stranded transgenic RNA by an endogenous RNA-dependent RNA polymerase (RdRP) (Tang et al., 2003; Sanders and Hiatt, 2005).
The study of plant RNA silencing pathways also reveals phenomenon such as transitive RNAi and systemic spreading that are not common in animal systems other than C. elegans (Klahre et al., 2002; Vaistij et al., 2002). Transitive RNAi is caused by secondary siRNA promoted by primary siRNA originating from the dsRNA with the action of RdRP (Vaistij et al., 2002). These secondary siRNA sequences may include any unpaired region of the transgenic IR transcript (Brummell et al., 2003) and the target mRNA sequences that lie upstream or downstream of the initial IR targeted region (Van Houdt et al., 2003). Therefore, the presence of transitive RNAi may lead to inactivation of mRNA species that are not targeted by the initial trigger sequence derived from the IR. Non-cell-autonomous suppression is another plant RNAi characteristic that has to be considered when applying RNAi to genetic engineering. Plant siRNA can travel intercellularly through plasmodesmata or utilize the vascular system for long-distance translocation (Yoo et al., 2004; Kim, 2005). This could impair the use of RNAi in cases when tissue-specific suppression is desired.
In this study, an IR sequence corresponding to part of the SDH coding region of maize LKR/SDH (ZLKR/SDH) cDNA (Kemper et al., 1999) was inserted into a transgene cassette driven by a maize endosperm-specific promoter. We investigated not only the efficacy of accumulating free lysine in mature kernels, but also whether the suppression of ZLKR/SDH was specific to the developing endosperm. The transitive nature of RNAi was also examined by analysing the expression of several endogenous genes that were potential targets of secondary siRNA generated from the nascent IR transcript and by sequencing siRNA cloned from transgenic endosperm tissue. Notably, this study represents the first transgenic RNAi study done on a major crop plant species such as maize and can provide important information for applying RNAi in plant biotechnology.
Transgenic F1 seeds showed reduced LKR/SDH protein accumulation
The T-DNA region of the binary vector used in the study, pMON79464, is shown in Figure 1(a). The maize endosperm-specific promoter, P-b32 (Hartings et al., 1990), was used to drive an IR sequence. The IR region, IR-SDH, corresponded to part of the SDH coding region of ZLKR/SDH cDNA. More than 50 independent R0 transgenic plants were generated. Only those containing a single copy of the epsps-cp4 marker gene as determined by an InvaderTM-based zygosity assay (Third Wave Technology, Madison, WI, USA) were pollinated with wild-type pollen to produce 21 F1 transgenic lines. Among them, 15 lines were found to have reduced maize LKR/SDH protein (ZLKR/SDH) in segregating F1 kernels as shown for representative lines S95775, S96063, S96761, S96773, S96774 and S96776. The remaining 6 lines had normal level of ZLKR/SDH as shown for S95538 and S96679 (Figure 1b).
Transgenic RNAi targeting ZLKR/SDH behaved as a dominant high lysine trait in F3 kernels
All F1 lines displaying the ZLKR/SDH reduction phenotype were planted and self-pollinated to obtain F2 seeds. Two randomly selected F2 lines, S95769 and S96100, were advanced to the F3 generation by planting 50 F2 seeds for each line. Since F2 plants were derived from self-pollinated hemizygous F1, they segregated homozygous, hemizygous and null and produced F3 ears that seeds contain 100%, 75% and 0% transgenic seeds, respectively. As shown in Figure 2, homozygous ears had up to a 20-fold increase in free lysine content compared to the null control. The average lysine increase in hemizygous ears was 77.7% (63.4%, 92.0%; α = 0.05) relative to homozygous ears. As expected, the transgenic RNAi targeting ZLKR/SDH appeared to behave as a dominant high lysine trait.
Endosperm-specific reduction of ZLKR/SDH in developing F4 transgenic kernels
Tissue from homozygous F3 plants and F4 developing kernels of lines S95769 and S96100, and wild-type were collected for ZLKR/SDH Western blot analysis. The ZLKR/SDH in transgenic lines was almost undetectable in the endosperm as early as 18 days after pollination (DAP) (Figure 3a). In the embryo, a faint ZLKR/SDH band was observed and appeared to be comparable to wild type. This was consistent with the results shown in F1 seeds where whole kernel tissue was used for analysis. Residual ZLKR/SDH protein observed in kernels that showed ZLKR/SDH suppression may be from the embryo (Figure 1b). Reduction of ZLKR/SDH was confirmed by assessing LKR activity. In comparison to wild type, a significant reduction in LKR activity was found only in the developing endosperm of transgenic kernels (Figure 3b). In Figure 3(c), Western blot data showed the ZLKR/SDH suppression lines and wild type had identical ZLKR/SDH expression patterns in other vegetative and reproductive tissues. ZLKR/SDH was present in shoots and immature ears, but absent in leaf and pollen/tassel tissues.
The endosperm-specific reduction of ZLKR/SDH in transgenic lines was also supported by Northern blot analysis. ZLKR/SDH mRNA is known to be relatively abundant in developing endosperm (Kemper et al., 1999). The ZLKR/SDH mRNA in transgenic endosperm was significantly reduced throughout the kernel developmental stages examined (Figure 4a). In the embryo, a longer exposure of the blot revealed that there was a similar ZLKR/SDH expression pattern between transgenic and wild-type embryos. The extra bands observed below the predicted ZLKR/SDH mRNA band in the blot are likely derived from the transgene. When the blot was probed with a DNA fragment corresponding to the LKR coding region which was not used in construction of the transgenic cassette, these bands were not observed (data not shown).
To verify that the transgenic suppression of ZLKR/SDH was mediated by RNA silencing, small RNA Northern blot analysis was performed. The typical 21- and 24-nt siRNA involved in RNA silencing were detected only in the developing endosperm of transgenic lines (Figure 4b). Even after prolonged exposure, the siRNA were not evident in the transgenic embryo. Although low transgene expression was detected in the embryo (Figure 4a), it apparently was not enough to induce the production of siRNA.
The expression of endogenous b32, Adh1 and Glb1 was unchanged in transgenic lines
The maize endogenous genetic elements b32 promoter, Adh1 intron, and Glb1 terminator were used along with the IR sequence targeting ZLKR/SDH to construct the transgenic cassette. It is known that secondary siRNA can be produced from unpaired RNA sequences linked to dsRNA and effect RNA silencing of their corresponding targets (Brummell et al., 2003). In addition to the Adh1 intron sequence, the transgenic transcript contained approximately 50 and 200 nt of 5′ and 3′ untranslated sequences from b32 and Glb1, respectively. As shown in Figure 5, the expression patterns of b32, Adh1 and Glb1 in developing kernels were similar between transgenic and wild-type plants. Based on the band intensity and the relative exposure time, b32 had higher expression in the endosperm and Glb1 had higher expression in the embryo as expected. The low level of b32 mRNA detected in the embryo was consistent with previous Northern blot analysis that showed a low level of B32-driven transgenic transcript in the embryo after 24 h of exposure (Figure 3a).
Cloning of small RNA from the developing endosperm of transgenic kernels
The presence of secondary siRNA can also be determined by the cloning of siRNA. Small RNA isolated from the 25 or 32 DAP developing endosperm of transgenic line S95769 for was used to construct a small RNA library for large-scale sequencing. Among approximately 200 000 sequences analysed, the most abundant ones were miRNA as expected and only 4568 sequences (2.29%) were found to be transgenic related siRNA (Figure 6a). The expected transgenic transcript formed a dsRNA and an unpaired loop region that spanned 2001–2459 and 1487–2000 bp of ZLKR/SDH cDNA, respectively. All but 14 transgenic siRNA matched the sequence of the dsRNA region and 12 matched the loop region. The last two transgenic related siRNA were mapped to the T-Glb1 sequence. More importantly, no secondary siRNA were generated from the endogenous ZLKR/SDH mRNA target. These results were consistent with the RNA gel blot analysis and suggested that transitive RNA silencing did not occur in the maize endosperm. As expected, these cloned transgenic siRNA were mostly 21 and 24 nt in length and consisted of both sense and antisense strands (Figure 6b).
In this study, we report the production of high lysine corn by suppression of the lysine catabolic enzyme ZLKR/SDH in the endosperm. This was achieved by RNA silencing through the endosperm-specific expression of an IR sequence that targeted part of the SDH coding region of ZLKR/SDH mRNA. Although details of RNA silencing pathways have been proposed in model plant systems, relatively little is known about crop plants. Using high lysine corn lines generated by RNAi, we carried out a detailed molecular characterization. These results not only provide some insight into how RNAi works in crop plants, but also provide an assessment of the potential use of RNAi for commercial purposes.
Lysine catabolism, not biosynthesis, in developing endosperm impacts lysine accumulation in corn grain
Previously it was found that in maize, the expression of a feedback-insensitive DHDPS in embryo tissue, but not in endosperm tissue resulted in free lysine accumulation in mature seeds (Mazur et al., 1999; Huang et al., 2005). It is generally thought that DHDPS is the rate limiting step in lysine metabolism. However, in the maize endosperm, the inability to accumulate high levels of lysine seems to be due to high rates of lysine catabolism (Kemper et al., 1999). The ability of transgenic maize kernels to accumulate free lysine by RNAi suppression of ZLKR/SDH alone supports this hypothesis. Also, maize DHDPS has been shown to lose over 90% of its activity at 200 µm of lysine in vitro (~30 p.p.m.) (Shaver et al., 1996) which approximates the lysine level in wild-type maize seed and is 1/20 the level of lysine in transgenic RNAi seed (Figure 2). This implies that the flux of lysine into the endosperm from the phloem probably occurred during endosperm development since the lysine biosynthesis may be inhibited by the high level of lysine. However, we cannot rule out the possibility that lysine biosynthesis and catabolism are partitioned in different compartments of the maize endosperm. We are currently investigating the possibility of further enhancing maize seed lysine content by coupling deregulated lysine biosynthesis with ZLKR/SDH reduction.
Maize endosperm-specific RNAi
Although RNAi induced by expression of dsRNA is effective and efficient in plants, it has been used mostly as a tool to study gene function in model plant systems. Only a few studies have examined the possibility of using RNAi to modify the expression of endogenous genes for agricultural applications in crop plants. For example, transgenic dsRNA were used to target seed storage proteins to change the grain quality of corn (Huang et al., 2006) and rice (Kusaba et al., 2003), or a regulatory protein to increase beneficial metabolites in tomato (Davuluri et al., 2005). In the first two instances, the tissue specificity of RNAi was not a concern since the target proteins were seed specific and therefore the suppression only occurred in seed regardless of whether the transgenic siRNA were systemically spreading or not. Specificity of RNAi becomes more important when the application involves modifying an endogenous pathway that is active in multiple tissues in a tissue-specific manner. Constitutive RNAi suppression may cause side-effects and render the technology undesirable. Such is the case in tomato where Davuluri et al. reported fruit-specific RNAi-mediated suppression (Davuluri et al., 2005). The authors observed that gene silencing signals were not passed beyond the seed stage. In each generation, the transgenic seeds gave rise to wild-type plants up until fruit development when the silencing was triggered again. It has been hypothesized that perhaps there is no transmission of silencing signals in and out of developing seeds (Kusaba, 2004). It could be the absence of plasmodesmata between the seed and its surrounding tissues, or the existence of a surveillance system that excludes silencing signals from passing into seed, as described for the shoot apex (Foster et al., 2002).
Similarly, our results indicated that the transgenic suppression of ZLKR/SDH was restricted to endosperm and suppression was not observed in embryo, seedlings, and tissues later in plant development. With the exception of the endosperm, ZLKR/SDH expression in transgenic plants was indistinguishable from wild-type plants. This lack of RNA silencing spreading between endosperm and embryo suggests the presence/absence of a silencing signal barrier/channel between these two tissues.
Absence of secondary siRNA production in maize endosperm
Transitive RNAi caused by secondary siRNA has to be considered to avoid adverse effects of unintended gene suppression. Secondary siRNA are promoted by primary siRNA and can be generated in two ways. In plants, a transgenic dsRNA is processed by DCLs to produce primary siRNA. The primary siRNA can then initiate ‘cis’ secondary siRNA production from the single-stranded region of the original transgenic transcript or ‘trans’ secondary siRNA production from their endogenous target mRNA, including sequences upstream and downstream of the initial trigger site. Using a transgene cassette containing a polygalacturonase (PG) cDNA fused with a heterologous IR sequence, Brummell and colleagues (2003) successfully suppressed endogenous PG and delayed the ripening of transgenic tomato. Such gene silencing by ‘cis’ secondary siRNA, however, was not detected in our transgenic lines. Through Northern bolt analysis and the cloning of small RNA, we concluded that the endogenous genes, whose genetic elements were used in the construction of the IR transgene cassette, were not subjected to RNA silencing. Moreover, we did not recover any ‘trans’ secondary siRNA from cloned small RNA. Interestingly, these transgene-derived siRNA, effective in silencing ZLKR/SDH, were found to represent only 2.29% of the endogenous small RNA population. Although plant small RNA pathways are known to be involved in defence and development, it is unlikely that transgenic siRNA at this level would impact these endogenous processes. Together, these findings provide important data regarding the commercial regulatory aspect of RNAi technology.
Application of RNAi for crop improvement
Although loss-of-function mutants can be recovered for the genetic improvement of crop plants through conventional breeding, they are usually recessive and lack tissue specificity which may limit their commercial utilization. Genetic engineering using RNAi offers a more useful technology to down-regulate endogenous genes in a dominant and tissue specific manner. As exemplified in the present study, a maize high lysine trait was engineered through endosperm suppression of ZLKR/SDH by RNAi. This would not be possible by isolating ZLKR/SDH loss-of-function mutants since high concentrations of lysine in vegetative tissues are known to cause agronomic abnormalities (Negrutiu et al., 1984; Frankard et al., 1992). We are now assessing the trait stability over successive generations and performance under different environmental conditions.
Although much has been learned about plant RNAi in recent years, its machinery and biological functions are still not fully understood. Nevertheless, there are examples of commercial cultivars produced by RNAi. Identified in the 1970s, LGC-1 (Low Glutelin Content-1) is low protein rice used by patients on a restricted protein diet. The cloning of this dominant mutant locus revealed a gene structure harbouring an IR of the glutelin gene and glutelin siRNA were found in the mutant (Kusaba et al., 2003). In soybean, pigmented seed coats are common in all wild accessions, but cultivated varieties have been selected for a yellow seed coat. This is due to an endogenous duplicated IR allele in cultivated varieties that drives silencing of CHS (chalcone synthase) genes, thereby inhibiting pigmentation of the seed coats (Tuteja et al., 2004). Another example is the Flavr Savr tomato, the first commercialized genetically modified food developed by Calgene (Davis, CA, USA; now part of Monsanto Company, St. Louis, MO, USA) in the early 1990s. Initially, antisense constructs against polygalacturonase (PG) cDNA were used to transform tomato to delay fruit ripening. However, molecular characterization of all the lines with the commercial efficacy had T-DNA inserts containing IR elements (Sanders and Hiatt, 2005). Recently, small RNA Northern blot analysis has confirmed the presence of siRNA corresponding to the PG cDNA sequence in these transgenic tomato lines (Allen and Krieger, pers. commun.).
Vector construction and plant transformation
The binary vector, pMON79464, contained the maize endosperm b32 promoter (Hartings et al., 1990), P-b32, maize Adh1 intron (Dennis et al., 1984), I-Adh1, an IR sequence corresponding to part of the SDH region of the maize LKR/SDH (ZLKR/SDH) cDNA (Kemper et al., 1999), IR-SDH, and maize Glb1 3′UTR (Belanger and Kriz, 1991), T-Glb1 (Figure 1a). The IR sequence, IR-SDH, consisted of a fusion of two segments corresponding to bases 2459-1487 and 2001-2459 of ZLKR/SDH cDNA, GenBank accession number AF003551 (start codon = 1). The selectable marker gene (epsps-cp4) and the rest of the genetic elements in the binary vector were similar to those previously described (Huang et al., 2004). The binary plasmid was electroporated into Agrobacterium tumefaciens ABI strain and introduced into maize embryos by Agrobacterium-mediated transformation (Armstrong and Rout, 2001).
LKR/SDH Western blot analysis
The Western blot procedures used were as previously described (Huang et al., 2005). The rabbit anti-ZLKR/SDH antibody was produced by immunizing rabbits against Escherichia coli expressing recombinant ZLKR/SDH (Cocalico Biologicals, Reamstown, PA, USA).
Plant RNA isolation and Northern blot analysis
Total RNA was isolated from dissected endosperm and embryo collected at desired developmental time points using Tri Reagent (Molecular Research Center, Cincinnati, OH, USA), and further separated to high molecular weight (HMW) RNA (> 200 nt) and low molecular weight (LMW) RNA (< 200 nt) by mirVana miRNA Isolation Kit (Ambion, Austin, TX, USA) following the manufacturers’ instructions. Fifteen µg of HMW or LMW RNA of each sample was loaded on a 1% (w/v) agarose/formaldehyde gel or a 15% TBE-Urea Criterion gel (Bio-Rad, Hercules, CA, USA), respectively, and transferred to Nytran SuperCharge (Schleicher and Schuell, Keene, NH, USA) or Zeta-Probe (Bio-Rad) nylon membranes, respectively. Prehybridization and hybridization were performed in the hybridization buffer PerfectHybTM Plus (Sigma, St. Louis, MO, USA). The washing and detection of the membranes were conducted following the manufacturer's procedures, except for the LMW RNA blots where a lower hybridization (38 °C) and washing (43 °C) temperatures was used. The PCR fragments corresponding to the cDNAs of various genes examined were radioactively labelled and used for probes.
Small RNA cloning and sequencing
Small RNA library construction was done as previously described with minor modifications (Lau et al., 2001; Llave et al., 2002). LMW RNA was separated on a 15% TBE-Urea Criterion gel (Bio-Rad) and small RNA eluted with Probe Elution Buffer (Ambion) according to the manufacturer's instructions. The library was sequenced commercially by 454 Life Sciences (Branford, CT, USA).
LKR activity assay
The protein extraction and LKR activity assay were performed as previously described (Huang et al., 2005).
Free lysine analysis by LC/MS/MS
Ground samples (50 mg) were extracted with extraction buffer (50% methanol, 50% water and 1% formic acid) with a 30 : 1 w/v ratio and vortexed for 20 min, and then kept at 4 °C overnight to complete the extraction. After centrifugation to pellet debris, the supernatant was transferred and necessary dilutions were made for LC/MS/MS analysis. A C18 column, 4.6 mm × 50 mm (Alltech Associates, Deerfield, IL, USA) and an API 3000 Triple quadrupole mass spectrometer (Applied Biosystems, Foster City, CA, USA) was used to separate and detect the lysine content in the sample extracts.
The authors wish to thank St. Louis corn transformation team for plant transformation, Monsanto trait development teams in Mystic and St. Louis for the greenhouse and field care, and Monsanto crop analytic team in St. Louis for lysine analysis. We also thank Ed Allen for his valuable comments on the manuscript.