Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Lysine (Lys) is the first limiting essential amino acid in rice, a stable food for half of the world population. Efforts, including genetic engineering, have not achieved a desirable level of Lys in rice. Here, we genetically engineered rice to increase Lys levels by expressing bacterial lysine feedback-insensitive aspartate kinase (AK) and dihydrodipicolinate synthase (DHPS) to enhance Lys biosynthesis; through RNA interference of rice lysine ketoglutaric acid reductase/saccharopine dehydropine dehydrogenase (LKR/SDH) to down-regulate its catabolism; and by combined expression of AK and DHPS and interference of LKR/SDH to achieve both metabolic effects. In these transgenic plants, free Lys levels increased up to ~12-fold in leaves and ~60-fold in seeds, substantially greater than the 2.5-fold increase in transgenic rice seeds reported by the only previous related study. To better understand the metabolic regulation of Lys accumulation in rice, metabolomic methods were employed to analyse the changes in metabolites of the Lys biosynthesis and catabolism pathways in leaves and seeds at different stages. Free Lys accumulation was mainly regulated by its biosynthesis in leaves and to a greater extent by catabolism in seeds. The transgenic plants did not show observable changes in plant growth and seed germination nor large changes in levels of asparagine (Asn) and glutamine (Gln) in leaves, which are the major amino acids transported into seeds. Although Lys was highly accumulated in leaves of certain transgenic lines, a corresponding higher Lys accumulation was not observed in seeds, suggesting that free Lys transport from leaves into seeds did not occur.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Lysine (Lys) is the first limiting essential amino acid in cereal grains (Galili et al., 1994). Among cereals, rice is the main source of calories and protein intake for billions of people worldwide. Thus, enhancing the content of Lys in rice would significantly contribute to the nutritional well-being of the world population. Lys in seeds is synthesized by a branch of the aspartate (Asp) family pathway, which is also responsible for the synthesis of two other essential amino acids, methionine (Met) and threonine (Thr) (Figure 1) (Galili, 1995, 2002; Jander and Joshi, 2010). Lys biosynthesis is strongly regulated by a feedback inhibition loop (Galili, 1995). The activity of aspartate kinase (AK), the first enzyme in the pathway, is the feedback inhibited by both Lys and Thr. In addition, Lys inhibits the activity of dihydrodipicolinate synthase (DHPS), the first enzyme specifically committed to Lys biosynthesis (Galili, 2002). Lysine ketoglutaric acid reductase/saccharopine dehydropine dehydrogenase (LKR/SDH) is a bifunctional enzyme that plays a key role in the Lys degradation pathway (Galili et al., 2001). Lys levels in plants are finely controlled by a super-regulated catabolic pathway as well.
Transgenic expression of bacterial Lys feedback-insensitive DHPS has been shown to drastically boost the content of free Lys in tobacco (Shaul and Galili, 1992a, 1993), canola (Falco et al., 1995), soybean (Falco et al., 1995), and Arabidopsis (Ben-Tzvi Tzchori et al., 1996; Zhu and Galili, 2003). When a similar method was used in maize, the bacterial DHPS induced Lys overproduction only when expressed in the embryo, but not in the endosperm (Frizzi et al., 2008; Huang et al., 2005; Mazur et al., 1999). For other cereals, such as barley (Brinch-Pedersen et al., 1996) and rice (Lee et al., 2001), only slight increases in the content of free Lys were detected in the mature seeds when a constitutive promoter was used. It was suggested that different regulatory mechanisms of Lys accumulation exist in these cereals. In an attempt to overcome Lys catabolism, reducing the activity of LKR/SDH by RNA interference (RNAi) technology was shown to dramatically increase the content of Lys in maize seeds (Frizzi et al., 2008; Houmard et al., 2007). However, similar research on Lys enhancement in rice is very limited, and the only related report showed that overexpressing the maize lysine feedback-insensitive DHPS in rice led only to a slight (2.5-fold) increase in Lys content in mature seeds but with a poor seed germination rate (Lee et al., 2001). In view of the importance of rice as a stable food for half of the world population, we have generated a transgenic rice line with a remarkable increase in free Lys content through overexpression of Lys feedback-insensitive AK and DHPS to boost the rate of Lys synthesis while simultaneously down-regulating the expression of LKR/SDH by RNAi to decrease Lys degradation in seeds. To evaluate the significance of this approach on Lys accumulation by regulating Lys biosynthesis and catabolism, two other transgenic lines were made, one using a construct constitutively expressing both AK and DHPS and the other with a construct down-regulating the expression of LKR by RNAi.
Metabolites, as the final products of gene activities, are good indicators of biochemical changes (Sweetlove and Fernie, 2005). Metabolic profiling has been used to phenotype a diverse array of genetically or environmentally manipulated plants (Aharoni et al., 2002; Gray and Heath, 2007; Hideyuki et al., 2005; Hirai et al., 2004; Roessner et al., 2000; Roessner-Tunali et al., 2003; Schauer et al., 2005). In this study, we applied metabolomic analysis not only to verify and investigate the induced variations in Lys metabolism by genetic engineering but also to gain further insights into the regulatory mechanism(s) of Lys accumulation in rice. However, a technical challenge in this study was detecting trace amounts of intermediates in the Lys metabolic pathway. Capitalizing on the recent development of Fourier transform ion cyclotron resonance/mass spectrometry (FTICR/MS), a powerful analytical tool allowing the acquisition of high-resolution mass spectra with high mass accuracy (Breitling et al., 2006; Brown et al., 2005; Dunn et al., 2005), we were able to compare differences in intermediate compounds involved in the Lys biosynthesis and catabolism pathways in the leaves and seeds of wild-type (WT) and transgenic rice plants.
The content of free Lys, through manipulating its biosynthesis and catabolism in rice, could be increased up to ~60-fold in seeds and ~12-fold in leaves. The free Lys-enriched rice seeds were shown to have a normal germination rate. Metabolomic analysis revealed the presence of Lys catabolism in rice leaves, but it played a more prominent regulatory role in seeds. In the transgenic leaves, the levels of glutamine (Gln) and asparagine (Asn), which are typically the major amino acids transported into seeds, were not substantially affected. Although Lys was highly accumulated in leaves of certain transgenic lines, a corresponding higher accumulation in seeds was not observed, suggesting that transport of free Lys from leaves into seeds probably did not occur.
Genetic engineering to alter metabolism of Lys to elevate its accumulation in rice
To co-ordinately increase Lys synthesis and reduce its catabolism as well as to investigate the regulation of Lys accumulation in rice, we generated transgenic lines from the japonica rice cultivar Wuxiangjing 9 (WXJ9), using three different constructs (Figure 2) to increase Lys synthesis, reduce its catabolism or both. These constructs included the following: (i) 35S, providing constitutive expression of the bacterial Lys feedback-insensitive AK and DHPS genes, both driven by the CaMV 35S promoter, (ii) Ri, inhibiting the expression of the LKR/SDH gene by LKR-RNAi construct under the rice endosperm-specific glutelin Gt1 promoter, and (iii) 35R, expressing the combined transgenes of constructs (i) and (ii).
Over 30 independent transgenic lines were obtained for each of the three constructs, and more than three stable transgenic lines were randomly chosen as representative lines for each construct. In both greenhouse and experimental fields, we observed that the transgenic rice lines, in comparison with the WT, exhibited normal plant growth and similar seed germination rates. PCR and Southern blot analyses showed that the transgenic cassette(s) were stably integrated into the genome of the chosen transgenic lines (Figure S1). Northern blot analysis revealed that the bacterial AK and DHPS genes were highly expressed in the developing seeds of the transgenic plants (Figure 3a), while the expression of the endogenous LKR/SDH genes was inhibited or down-regulated dramatically as expected (Figure 3a). The altered expression levels of the corresponding enzymes in Lys metabolism were further confirmed by Western blot analysis (Figure 3b).
Our results showed that through metabolic engineering, the free Lys content in transgenic rice could be successfully increased. Free Lys levels in mature seeds were enhanced by ~1.1-fold, ~10-fold, and ~60-fold in the 35S, Ri, and 35R transgenic lines, respectively (Figure 4c). The free Lys level also increased in leaves of the 35S (6- to 10-fold) and 35R (5- to 12-fold) transgenic lines (Figure 5b).
Detection of metabolic compounds in the Asp family pathway in rice seeds and leaves
To evaluate the regulatory significance of Lys biosynthesis and catabolism in Lys accumulation, we investigated the levels of compounds involved in these pathways in rice seeds and leaves by metabolomic analysis. Homozygous transgenic lines (T5 generation) and their WT counterpart were grown to maturity. The developing seeds were collected at the early grain-filling stage (5 days after flowering, DAF), milky stage (15 DAF), and mature stage. The leaf samples were collected at five developmental stages, including the tillering stage, panicle initiation stage, 5 DAF, 15 DAF, and mature seed stages. We were able to detect 11 compounds involved in Lys synthesis and catabolism in the Asp family pathway in both seed and leaf samples using liquid chromatography-Fourier transform mass spectrometry (LC-FTMS) (Table 1), including Asn, Asp, aspartate semialdehyde, 2,3-dihydrodipicolinate, 2,3,4,5-tetrahydrodipicolinate, 2,6-diaminoheptanedioate, Lys, saccharopine, 2-aminoadipate 6-semialdehyde, 2-aminoadipic acid, and 2-oxoadipate. The relative levels of these compounds in seeds at three developmental stages are provided in Figure S2 and Table S1, and in leaves at five plant developmental stages in Table S2. Thr, Met, homoserine, and Ile are synthesized by the other branch of the Asp family pathway, and their relative levels were also detected in seeds (Table S3). Other amino acids, such as Glu and Gln closely related to the Asp family pathway, were also measured in leaves (Table S2) and seeds (Table S4).
Table 1. Monoisotopic mass of metabolites involved in the Lys biosynthesis and catabolism pathways
Theoretical monoisotopic mass
Theoretical monoisotopic mass of negative mode
Theoretical monoisotopic mass of positive mode
For each construct, three independent transgenic lines were used for metabolomic analysis, and the profiling data of all the above compounds are listed in Tables S1–S4. Based on these results, one representative transgenic line for each construct was selected for subsequent analysis.
Profiles of metabolites during rice seed development
Metabolites involved in Lys biosynthetic and catabolic pathways
At 5 DAF, the levels of most of the investigated compounds in the caryopsis of transgenic rice were not substantially different from those of the WT. However, the levels of free Lys accumulated in developing seeds at this stage in 35S and Ri lines were somewhat lower than those in the WT and ~1.5-fold higher than those in the seeds of the 35R line (Figure 4a). At 15 DAF, large amounts of compounds were accumulated in the caryopsis, and the level of Lys in the caryopsis of 35R remained the highest, up to ~6-fold over that of the WT (Figure 4b). At the mature stage, free Lys levels in mature seeds of 35S, Ri and 35R were ~1.1-, ~10-, and ~60-fold higher, respectively, than those of the WT plants (Figure 4c). Thus, the Lys level of the 35S line, which overexpressed the bacterial Lys feedback-insensitive AK and DHPS without blocking Lys catabolism, remained about the same as that of the WT. Through the RNAi method of preventing Lys degradation in the Ri line, the Lys level increased by ~10-fold over that of the WT, suggesting that Lys catabolism is a major regulatory step in the accumulation of Lys in seeds. Furthermore, the free Lys level in seeds of 35R increased ~60-fold over that of the WT, rendering Lys the most abundant free amino acid in mature seeds, again supporting the notion that Lys catabolism plays a key role in free Lys accumulation in rice seeds.
Besides Lys, ten other compounds involved in the Lys metabolic pathway were altered in transgenic rice seeds in comparison with those in the WT. The accumulation of these compounds was at the highest level in the 35R line when compared with the other two transformants and the WT, especially at the mature stage (Figure 6; Table S1). Thus, in seeds of the 35R line, a robust alteration of the entire Lys biosynthesis and catabolism pathways must have occurred as a result of the effective RNAi-mediated prevention of Lys degradation and the enhancement of Lys synthesis by constitutive expression of AK and DHPS. Although the levels of all 11 examined compounds increased, their levels varied in mature seeds of the 35R line. The fold changes in 2,6-diaminoheptanedioate (Figure 6, c3), Lys (Figure 4c) and saccharopine (Figure 6, d3) at the mature stage were considerably higher than those of other compounds, which may have been due to these three compounds being either the direct precursors or product of the LKR enzyme.
The level of aspartate semialdehyde increased only in the 35S and 35R lines but not in the Ri line at the mature stage (Figure 6, a3). This result may be attributed to the constitutive overexpression of AK by the 35S promoter, causing AK to continuously convert aspartate into aspartate semialdehyde. The level of 2,3-dihydrodipicolinate, a specific intermediate in the Lys synthesis pathway, was relatively lower in 35S seeds than in the WT at the mature stage, while it was higher in the Ri line and observed at the highest level in the 35R line (Figure 6, b3). By not using RNAi to prevent Lys degradation in the 35S line, the pool of available 2,3-dihydrodipicolinate was drawn to form Lys. The higher level of 2,3-dihydrodipicolinate in the Ri line was likely due to the RNAi effect, while its highest level observed in the 35R line was due to the synergistic increase in Lys synthesis and decrease in Lys catabolism. As for other intermediates in the Lys biosynthesis pathway, the level of 2,6-diaminoheptanedioate in the 35S line, for example, was similar to that of the WT at the mature stage, while it was higher in the other two transgenic lines (Figure 6, c3). For metabolites in the Lys catabolism pathway, other studies have reported an association of the expression of bacterial Lys feedback-insensitive DHPS with enhanced levels of Lys catabolic products (Falco et al., 1995; Mazur et al., 1999). Our results also showed that the levels of saccharopine in the 35S line were higher than those of the WT at all three stages (Figure 6d), while levels of 2-aminoadipate 6-semialdehyde (Figure 6e) and 2-aminoadipic acid (Figure S2) were higher at 5 DAF and 15 DAF. This result may have been due to the overexpression of Lys feedback-insensitive AK and/or DHPS in the Lys synthesis pathway, as confirmed by enhanced LKR/SDH transcription in developing seeds of the 35S line compared with the WT (Figure 3), thereby triggering the regulatory mechanism of the Lys catabolism pathway to deregulate the level of Lys. The enhanced transcription level of LKR/SDH in the developing seeds of 35S line when compared with that in wild type (Figure 3) renders support to this notion. Although the LKR/SDH activity was reduced in the Ri and 35R lines, it could still convert some Lys to saccharopine, leading to higher levels than those found in the WT, especially at 15 DAF (Figure 6, d2) and the mature stage (Figure 6, d3).
Amino acids in the other branch of the Asp family pathway
As Thr, Met, homoserine, and Ile are synthesized by the other branch of the Asp family pathway that competes with the Lys branch for common substrates (Figure 1) (Galili, 1995; Jander and Joshi, 2010), we compared the changes of these amino acids in the three types of transgenic lines with those in the WT (Table S3). Because homoserine and Thr share identical theoretical mass and retention time, they cannot be separated by LC-FTMS. We thus used data from gas chromatography-mass spectrometry (GC-MS) analysis of homoserine, Thr, Ile, and Met for comparison. At the mature stage, Thr (Figure 7, b3) in the 35S and 35R lines increased ~3-fold and ~1.7-fold over the WT, respectively. The level of Met also showed increases of ~1.6-fold and ~1.8-fold over the WT in the 35S and 35R lines, respectively (Figure 7, c3). Since AK is the first enzyme in the Asp family pathway, overexpressing a Lys feedback-insensitive AK may have boosted the entire metabolic pathway, including the biosynthesis of Lys, Thr, and Met. It has been reported that expression of the bacterial AK led to ~14-fold overproduction of free Thr in tobacco seeds (Shaul and Galili, 1992b), supporting our metabolomic observation on the role of AK in the synthesis of Thr in rice seeds.
Aspartate semialdehyde, the branch point intermediate of Lys, Thr, and Met synthesis, is a common precursor for both DHPS and homoserine dehydrogenase (HSD). At the mature stage, levels of aspartate semialdehyde in seeds of the 35S and 35R lines were similar, ~1.4-fold higher than those in the WT (Figure 6, a3). Homoserine is the direct product of aspartate semialdehyde in Thr synthesis, and its level in the 35S line was higher than that in the 35R line (Figure 7, a3). The strong down-regulatory effect of LKR-RNAi on Lys catabolism led to the metabolic flow in the 35R line mainly towards Lys biosynthesis, resulting in high accumulation of 2,3-dihydrodipicolinate (Figure 6, b3) and less of homoserine (Figure 7, a3). The increased aspartate semialdehyde in the 35S line was mainly converted to homoserine due to its Lys catabolic regulatory mechanism. As a result, the accumulation of homoserine and that of its subsequent products such as Thr in the 35S line were higher than those in the 35R line.
Amino acids Glu, Asn, Asp and Gln
Lys is synthesized from Asp via one of the branches of the Asp family pathway (Galili, 1995). Asp can be produced in plant seeds either from Glu by Asp aminotransferase or from Asn by Asn synthetase (AS) (Lam et al., 1995), while Glu itself is synthesized largely from Gln as well as a product of the Lys catabolism pathway (Galili et al., 2001). We therefore also studied how changes in Lys metabolism conferred by targeted genetic intervention might affect the levels of Glu, Asp, Gln, and Asn (Figure 8). Although Lys catabolism was reduced in the 35R line, levels of its products, such as saccharopine and 2-aminoadipate 6-semialdehyde, were higher than those in the WT, suggesting that the Glu released from Lys catabolism in the 35R line should also have been higher than that of the WT. However, the accumulation of Glu in seeds of the three transgenic lines remained at levels similar to that of the WT at the mature stage (Figure 8b), suggesting that Glu may serve as a substrate in other metabolic pathways. One such possible pathway could be in the production of Asp via Asp aminotransferase (Lam et al., 1995), while another is in the synthesis of Gln (Lam et al., 1995), since levels of Asp (Figure 8a) and Gln (Figure 8d) in the 35R line were higher than those in the WT.
Temporal accumulation of Lys
Although its level varied in rice seeds at different stages, Lys accumulated at high levels from 15 DAF to the mature stage in both the WT and transgenic lines (Figure 4). When comparing the ratio of Lys between the 35S transgenic line and the WT at individual developmental stages, it was ~0.7-fold at both 5 DAF and 15 DAF and only increased slightly to ~1.1-fold at the mature stage. These results showed that the rate of Lys accumulation during seed development in the 35S line was not notably influenced by the expression of the transgenes. However, the ratio of Lys between the Ri or 35R line and the WT increased quickly during seed development, especially from 15 DAF onward, reaching their highest levels at the mature stage. These results revealed that Lys accumulated much more rapidly in Ri or 35R transgenic rice seeds than in the 35S transgenic or WT seeds, suggesting that Lys accumulation was mainly influenced by the LKR-RNAi transgenic effect and that Lys catabolism was more active at the later stage of seed development. Metabolites specific to Lys biosynthesis and catabolism in the Ri and 35R lines also accumulated quickly from 15 DAF onward, reaching their highest levels at the mature stage (Table S1), further confirming the notion that Lys catabolism plays a key role in its accumulation in rice seeds.
Profiles of metabolites in transgenic rice leaves
Metabolites involved in Lys biosynthetic and catabolic pathways
We also analysed and compared compounds involved in the Lys metabolic pathway in leaves of transgenic and WT rice during the five developmental stages in order to evaluate the transgenic effects on Lys accumulation in leaves (Table S2). Throughout the five developmental stages in leaves of the 35S and 35R lines, free Lys levels were higher than those in the WT, amounting to 6- to 10-fold and 5- to 12-fold higher, respectively (Figure 5b). Lys levels in these two transgenic lines were similar to each other at all stages, indicating that Lys enhancement in the transgenic leaves was due to the overexpression of the AK and DHPS under the control of the constitutive 35S promoter. Without the application of RNAi to deter Lys degradation, the free Lys level in leaves of the 35S line through overexpressing the Lys feedback-insensitive AK and DHPS by a constitutive promoter was still 6- to 10-fold higher than that of the WT. In contrast, the free Lys level was slightly increased by ~1.1-fold in the 35S line. These results imply that the accumulation of Lys in leaves is mainly regulated by its biosynthesis in leaves but by catabolism in seeds.
We were able to detect saccharopine, the direct product of Lys in the Lys catabolic pathway, in WT leaves (Figure 5c). As the level of saccharopine was quite low in leaves for its detection by other methods such as GC-MS, our results, although indirectly, suggested that Lys catabolic enzymes such as LKR/SDH were expressed and functioned, albeit weakly, in rice leaves.
Regarding other compounds, the levels of 2,6-diaminoheptanedioate (Figure 5a) and saccharopine (Figure 5c), the direct precursor and catabolic product of Lys, respectively, were dramatically higher in the 35S and 35R lines than in the WT, and the fold changes between transgenic and WT lines were consistent throughout leaf development. The increase in the level of saccharopine in the 35S and 35R lines supported the notion that LKR/SDH can be up-regulated by a high level of free Lys (Stepansky et al., 2005). Other compounds involved in the pathway did not show substantial changes in comparison with the WT plants (Table S2).
Amino acids Asn and Gln
The major amino acids transported into seeds are Asn and/or Gln (Lam et al., 1995; Sieciechowicz et al., 1988). Our results revealed that Asn and Gln did not show substantial differences between the leaves of transgenic and WT rice (Supplemental Table S4).
Generation of rice with enhanced free Lys content
Elevation of free Lys content is one approach to improving the nutritional quality of rice. Most previous efforts to improve Lys production in plants by metabolic engineering have utilized Lys feedback-insensitive AK and/or DHPS enzymes expressed by a constitutive or seed-specific promoter (Brinch-Pedersen et al., 1996; Falco et al., 1995; Mazur et al., 1999; Shaul and Galili, 1992a, 1993). Although constitutive expression of Lys feedback-insensitive enzymes may result in very high levels of Lys, it is often associated with abnormal plant growth and reduced seed germination (Ben-Tzvi Tzchori et al., 1996; Frankard et al., 1992; Shaul and Galili, 1992a, 1993). Targeting expression of these enzymes using a seed-specific promoter, however, previously has not resulted in a nutritionally desirable level of Lys in seeds (Karchi et al., 1994; Lee et al., 2001). For cereal grains, desirable Lys enhancement in seeds also has not been achieved by bioengineering of enzymes involved in the Lys biosynthesis pathway even when using a constitutive promoter (Brinch-Pedersen et al., 1996; Lee et al., 2001; Mazur et al., 1999). However, the approach of reducing LKR/SDH activity by RNAi technology to down-regulate Lys catabolism has been demonstrated to highly increase Lys content in maize seeds (Frizzi et al., 2008; Houmard et al., 2007). For rice, the only related study available reported a 2.5-fold low level increase in the content of free Lys in mature seeds through overexpressing the maize lysine feedback-insensitive DHPS (Lee et al., 2001). In the present study, through metabolic manipulations, we showed that by expressing the Escherichia coli feedback-insensitive AK and DHPS under a constitutive promoter (35S), the free Lys level could be increased by 6- to10-fold in leaves. Additionally, through inhibiting the expression of LKR/SDH by RNAi under the seed-specific promoter (Ri), the free Lys level in mature seeds could be enhanced ~10-fold. Finally, by combining 35S and Ri together (35R), the free Lys could be increased up to ~12-fold in leaves and ~60-fold in seeds of transgenic rice with normal plant growth and seed germination. Levels of two other essential amino acids in the 35S and 35R lines, Thr and Met, increased as well due to the overexpression of the Lys feedback-insensitive AK. Our results clearly demonstrated that free Lys can be genetically engineered to accumulate highly in the seeds and leaves of rice through manipulating its biosynthesis and catabolism.
Application of LC-FTMS to detect metabolites involved in the Lys metabolic pathway
Using GC-MS in early attempts, we could only detect the free amino acids in the rice seeds. In the wild-type rice leaves, even the free Lys could not be measured by GC-MS, which limit the study on the regulatory mechanism of Lys accumulation in the leaves. FTMS with its high sensitivity, coupled with LC, has allowed investigation of highly complex matrices in plants, such as complexes of oligosaccharides in Arabidopsis (Penn et al., 1997) as well as thousands of metabolites in strawberry (Aharoni et al., 2002). Applying this technology, we could analyse 11 intermediate compounds involved in the Lys metabolic pathway in both rice seeds and leaves (Table 1). Detection of these intermediates not only revealed the existence of Lys catabolism in leaves but also different regulatory mechanisms of Lys accumulation between rice leaves and seeds. Furthermore, other relevant compounds in the Lys metabolic pathway were also detected in our metabolomic study, providing an opportunity to investigate interactions and relationships between the Lys metabolic pathway and other networks. The platform established for the analysis of the regulation of Lys metabolism in rice by the present study may also be extended to other cereals and metabolic pathways.
Lys catabolism plays a major role in free Lys accumulation in rice seeds
Very few studies have centred on the relative contribution of Lys synthesis and catabolism in regulating the free Lys level in seeds, especially in rice seeds. For the model plant Arabidopsis, transgenic expression of the bacterial DHPS or its knockout mutant led to ~12- or 5-fold higher levels of free Lys in seeds, respectively, than the WT plants, and the combination of these two traits caused a synergistic ~80-fold increase of free Lys in seeds (Zhu and Galili, 2003). These results, however, did not reveal the relative contribution of Lys synthesis and catabolism to Lys accumulation. Sole expression of the bacterial DHPS has resulted in several- to hundredfold increases in free Lys in the seeds of Arabidopsis (Ben-Tzvi Tzchori et al., 1996; Zhu and Galili, 2004), much higher than that using RNAi to down-regulate LKR/SDH activity (Zhu and Galili, 2004). DHPS thus appears to be the major limiting factor in Lys accumulation in Arabidopsis seeds, while Lys catabolism plays an increasing regulatory role when Lys is over-accumulated. For cereal grains, in our current rice study, we used three different transgenic interventions to compare the relative contribution of Lys synthesis and catabolism in free Lys accumulation in seeds. While the free Lys level in the 35S line harbouring the AK and DHPS genes increased only by ~1.1-fold, the Ri line, in which LKR-RNAi was applied to deter Lys degradation, showed a much greater increase of ~10-fold over the WT. These results suggest that Lys catabolism is a major limiting factor in Lys accumulation in rice seeds. In comparing the 35S and 35R lines, the level of Lys in the 35R line was substantially elevated (~60-fold higher than that in the WT). This result again confirms that catabolism plays a major role in regulating the free Lys accumulation in rice seeds. Significant accumulation of Lys in maize seeds has also been reported through deregulating its catabolism (Frizzi et al., 2008). All these findings support that Lys catabolism is dominant in regulating Lys accumulation in cereal seeds.
Our results also showed that the level of LKR/SDH was dramatically enhanced in developing seeds of the 35S line when compared with that in the WT (Figure 3), indicating that overexpression of Lys feedback-insensitive AK and/or DHPS in the Lys synthesis pathway had triggered Lys catabolism to down-regulate the level of Lys. Even with a constitutive promoter, the effect of transgene(s) AK and/or DHPS was counterbalanced by the activity of Lys catabolism in sustaining a steady level of Lys in mature rice seeds.
Recent studies on Arabidopsis reported that overexpressing the bacterial DHPS in the LKR/SDH knockout mutant to increase the Lys level in seeds may retard seed germination due to a major negative effect on the levels of a number of tricarboxylic acid (TCA) cycle metabolites associated with cellular energy (Angelovici et al., 2009; Galili, 2011). It was suggested that the low levels of metabolites in the TCA cycle are due to the DHPS enzyme competitively utilizing pyruvate, serving also as a major entry precursor into the TCA cycle. However, the primary metabolites in the TCA cycle including pyruvate in our transgenic rice, especially in the 35R line (Table S6), did not show remarkable differences from those of the WT. We also did not find observable phenotypic changes or reduced seed germination in transgenic rice plants in comparison with the WT. Only ~1.0- to 1.9-fold differences of 2,3-dihydrodipicolinate, as the direct product of DHPS, were found between the 35R line and the WT during seed development (Figure 6, b3), suggesting that the DHPS enzyme did not notably compete with pyruvate involved in the TCA cycle in rice. This setting explains why metabolites in the TCA cycle did not change substantially in our transgenic rice, in turn supporting the idea that the regulation of Lys metabolism in cereal rice may be different in certain aspects in comparison with Arabidopsis.
Free Lys accumulation in rice leaves is mainly regulated by its biosynthesis
While Lys catabolism has previously been suggested to play an important role in Lys accumulation in plants, particularly in seeds (Ufaz and Galili, 2008), analysis of Lys accumulation in plant leaves is lacking. In our study, by overexpressing the Lys feedback-insensitive AK and DHPS driven by constitutive promoter, free Lys accumulation was increased up to 6- to 10-fold in the leaves, but only ~1.1-fold in the seeds of 35S transgenic rice in comparison with the WT. These and other results from this study suggest that in rice, Lys biosynthesis may play a more active role in regulating the accumulation of Lys in leaves, while Lys catabolism is the major regulatory factor in seeds.
LKR/SDH activity has been reported in plants but only in developing seeds (Arruda and Da Silva, 1983; Gaziola et al., 1997; Karchi et al., 1994). LKR/SDH transcripts were detected in several organs in Arabidopsis but most abundantly in the ovary and embryo (Tang et al., 1997). In maize, LKR/SDH mRNA is abundant in the endosperm, while it is completely absent in the embryo and scarcely detectable in roots, leaves and coleoptiles (Kemper et al., 1999). LKR/SDH mRNA was reportedly induced in rapeseed leaves upon osmotic stress (Deleu et al., 1999) and in Arabidopsis seedlings in response to hormonal, metabolic and stress signals and Lys (Stepansky et al., 2005). In these studies, LKR/SDH was not detectable at the transcriptional level (Deleu et al., 1999; Stepansky et al., 2005) and was present at a very low amount in WT leaves at the protein level (Stepansky et al., 2005). In the present study, through advanced metabolomic analysis, we were able to detect low levels of saccharopine, the direct product of Lys from its catabolism pathway, in the leaves of WT rice, providing indirect evidence of the presence of LKR/SDH and suggesting weak expression of the LKR/SDH gene in WT rice leaves. Our analysis also revealed that the level of LKR/SDH was obviously higher in the 35S and 35R transgenic lines, further suggesting that LKR/SDH could be up-regulated by the presence of high levels of free Lys. The up-regulated Lys catabolism in the 35S and 35R transgenic lines could counterbalance its biosynthesis to maintain a steady level of Lys in leaves that would not exert negative effects on the vegetative growth of rice. This finding contrasts with a previous study reporting that the constitutive expression of Lys-insensitive DHPS results in a very high level of Lys in all tissues of tobacco and Arabidopsis (Ben-Tzvi Tzchori et al., 1996; Shaul and Galili, 1992a, 1993), in turn leading to abnormal plant growth. Our study in rice first sheds light on the possibility that fine differences exist in the regulation of Lys metabolism in leaves between different species and thus warrants further study.
An elite rice cultivar, Oryza sativa ssp. Japonica cv.Wuxiangjing 9 (WXJ9), from China was used and planted in the greenhouse at the Chinese University of Hong Kong or the paddy fields at Yangzhou University (Yangzhou, Jiangsu Province, China).
Transgene constructs and rice transformation
Three key genes involved in plant Lys metabolism were used to construct transgene cassettes for enhancing Lys production in rice grains, including the mutated lysC gene coding for Lys feedback-insensitive AK from E. coli strain TOC R21 (Shaul and Galili, 1992b) and the dapA gene coding for DHPS from WT E. coli (Richaud et al., 1986), both kindly provided by Professor Gad Galili, the Weizmann Institute of Science, Israel, and the LKR/SDH gene from rice. The pea rbcS-3 chloroplast transit peptide (TP) was added to both lysC and dapA genes (Fluhr et al., 1986), and the partial coding sequences of LKR/SDH used for the RNAi construct were cloned from total RNA of developing rice seeds by reverse transcription PCR. Primers used for PCR are listed in Table S5. All three chimeric constructs were cloned into the binary vector pSB130 with double T-DNA regions (Figure 2).
Rice calli from immature or mature embryos were used as explants for Agrobacterium-mediated transformation according to our previous published procedure (Liu et al., 1998). Stably transformed plants were regenerated after screening for hygromycin resistance, and T0 transgenic plants were transferred and grown in soil for PCR and Southern blot analyses. Selected homozygous transgenic lines in T2 or later generations and their WT were propagated for composition and metabolomic analyses.
RNA and protein expression analyses
Total RNA was isolated from developing rice seeds (12 DAF) by a cold-phenol method (Liu et al., 2003). The isolated RNA was purified by treatment with DNase I and the RNeasy Plant Mini Kit (Qiagen, GmbH, Hilden, Germany). For Northern blot analysis, total RNA was electrophoresed in a 1% w/v formaldehyde/agarose gel and blotted onto a nylon membrane by capillary action. The hybridization and detection were performed according to the methods described in the DIG Nucleic Acid Detection Kit (Roche, Basel, Switzerland).
Total seed proteins were extracted at 4 °C from developing seeds by grinding in extraction buffer [125 mm Tris–HCl, 1% SDS, protease inhibitor cocktail (BD BaculoGold™, BD Biosciences, San Jose, CA)] with pH 6.8 for AK and LKR/SDH, or pH 7.5 for DHPS. After separation by SDS-PAGE, proteins were detected through immunoblotting using antibodies specific for E. coli AK and DHPS or rice LKR/SDH.
Experimental design for metabolomic analysis
Both the WT and transgenic rice used for metabolomic analysis were grown with permission for small-scale field trials from the Ministry of Agriculture China in the experimental fields in the summer of 2008 in Yangzhou, Jiangsu province, China. All transgenic plants were homozygous for the target transgene(s) in the T5 generation. Randomized plot design was used, and three plots in the fields were selected. For each plot, rice lines were planted randomly, and samples from five to ten individual plants were obtained for each line. Samples of five developmental stages were collected for analysis, including the tillering stage, panicle initiation stage, 5 DAF, 15 DAF, and mature stage. The uppermost, fully expanded leaves at each stage and the developing or mature seeds from the same plant at the later three stages were collected at 4:00 pm in the day. After harvesting, samples were immediately put into liquid nitrogen and freeze-dried for further use.
Sample preparation for metabolomic analysis
Methods for the applications of GC-MS (Fiehn, 2006; Frenzel et al., 2002; Lisec et al., 2006) and LC-FTMS (Aharoni et al., 2002; Hideyuki et al., 2005) were based on previous reports with necessary modifications as described below. Samples (45 mg leaves or 100 mg dehulled seeds) were grounded to powder in liquid nitrogen and extracted with 400 μL of extraction solution containing chloroform/methanol:H2O (1 :2.5 : 1, v/v/v) (pre-cooled at −20 °C and degassed). Ribitol (10 μL of 1 μg/μL stock in ddH2O) was added as an internal standard in GC-MS, and ampicillin (2.5 μL of 1 μg/μL stock in ddH2O) was added in LC-FTMS. The mixture was extracted for 30 min at 4 °C and centrifuged at 14 000 g for 10 min. The supernatant was dried in a speed vacuum system. For GC-MS, the dried residue was redissolved and derivatized for 120 min at 37 °C with 40 μL of 20 mg/mL hydroxylammonium chloride in pyridine, followed by a 30-min treatment at 37 °C with 70 μL of the silylating agent N-methyltrimethylsilyltrifluoroacetamide. The supernatant was transferred to a glass vial for GC-MS. For LC-FTMS, the residue was redissolved in 5% acetonitrile, and the supernatant was filtered through a 0.2-μm polyvinylidene difluoride (PVDF) filter.
Metabolic profiling by GC-MS
Metabolic profiling was carried out using the Agilent 5975 GC/MSD ChemStation with a HP-5s capillary column. Each sample (2 μL) was injected at 280 °C in splitless mode with helium carrier gas flow set to 1 mL/min. The temperature was isothermal for 2 min at 80 °C, followed by a 10 °C per min ramp to 325 °C and a hold at this temperature for 6 min. The transfer line temperature and ion source were 250 °C. The mass spectrum range was from m/z 40 to 800. Identification and quantification of peaks were processed using the GC/MSD ChemStation (Agilent, Santa Clara, CA) and Automated Mass Spectral Deconvolution and Identification System (AMDIS) software. Peak identities were confirmed by comparing with commercially available electron impact mass spectrum libraries NIST05 and mass spectrum of commercially available standards, which were obtained under identical trimethylsilyl deriviatization and GC-MS analysis.
Metabolic profiling by LC-FTMS
High-performance liquid chromatography (HPLC) separation was performed with the UltiMate 3000 LC system (Dionex, Bannockburn, IL) equipped with a Waters RP-18 Atlantis T3 column (2.1 × 100 mm, 3 μm) at a flow rate of 100 μL/min. MS detection was performed on a Bruker Daltonics APEX III FTMS equipped with a 7.0-Tesla actively shielded superconducting magnet with positive and negative modes. For the positive mode, solvents used were 0.1% formic acid in 98% H2O (A) and 0.1% formic acid in 98% acetonitrile (B). For the negative mode, solvents used were 0.5 mm NH4OH in 98% H2O (A) and 0.5 mm NH4OH in 98% acetonitrile (B). All solvents used were LC-MS grade. Both positive and negative modes used a 15-min linear gradient from 5% to 90% (B). Samples were internally calibrated for mass accuracy (<1 ppm) over the mass range of 100–500 amu using a mixture of standards. Identification of metabolites was confirmed by comparing the measured monoisotopic mass with the theoretical monoisotopic mass (<5 ppm) and the retention time with those of authentic standards. For metabolites without standards, the mass deviation should be <1 ppm. Peak identification and quantification of selective mass were accomplished using the software DataAnalysis™ and QuantAnalysis™ (Bruker, GmbH, Bremen, Germany). The 11 metabolites involved in Lys biosynthesis and catabolism pathways as well as their theoretical monoisotopic mass of positive mode and negative mode are listed in Table 1. The standards of Lys, saccharopine, 2-aminoadpic acid, 2-oxoadipate, Asn, Asp, Glu, Gln, Arg, Phe, Tyr, and Trp used in the LC-FTMS analysis were purchased from Sigma.
Three biological and three technical replicates were performed unless otherwise specified. Metabolite contents were normalized to an internal standard and expressed as a ratio of the plant dry weight (mg). All data are reported as means ± standard deviation/standard error (mean ± SD/SE). Statistical analysis to compare each transgenic line with the WT was performed using the Student's t-test followed by Pearson's correlation analysis. Results with a corresponding probability value of P <0.05, P <0.01 and P <0.001 were considered to be statistically significant, very significant and highly significant, respectively.
We gratefully thank Professor Gad Galili (The Weizmann Institute of Science, Israel) for providing the mutated lysC and dapA genes and Miss LEE Pui Kuen Jessie (CUHK) for her excellent technical assistance in LC-FTMS. The work described in this paper was supported by grants from the Bill and Melinda Gates Foundation (Grand Challenges in Global Health initiative, USA), the UGC of the HKSAR (Project No. AoE/B-07/99), the State Key Laboratory of Agrobiotechnology (CUHK), the Ministry of Science and Technology (Project No. 2012AA10A302-7 and 2011ZX08001-006), the Lee Hysan Foundation, the K.L. Lo Foundation, and the Jiangsu Natural Science Foundation of China (Project No. BK2012010).