Identification of OsGGR2, a second geranylgeranyl reductase involved in α-tocopherol synthesis in rice

Tocopherol (Toc) and tocotrienol (T3) are abundant in rice bran. Geranylgeranyl reductase (GGR) is an essential enzyme for Toc production that catalyzes the reduction of geranylgeranyl pyrophosphate and geranylgeranyl-chlorophyll. However, we found that a rice mutant line with inactivated Os02g0744900 (OsGGR1/LYL1/OsChl P) gene produces Toc, suggesting that rice plants may carry another enzyme with GGR activity. Using an RNA-mediated interference technique, we demonstrated that the Os01g0265000 (“OsGGR2”) gene product has GGR activity. This result supports the existence of two GGR genes (OsGGR1 and OsGGR2) in rice, in contrast to Arabidopsis thaliana (thale cress) and cyanobacterium Synechocystis that each have only one GGR gene. We also produced rice callus with inactivated OsGGR1 and OsGGR2 that produced T3 but not Toc. Such rice callus could be used as a resource for production of pure T3 for nutraceutical applications.


Identification of OsGGR2, a second geranylgeranyl reductase involved in α-tocopherol synthesis in rice
Eiichi Kimura 1 , Takumi Abe 2 , Kazumasa Murata 3 , Toshiyuki Kimura 4 , Yurika Otoki 2 , Taiji Yoshida 5 , Teruo Miyazawa 6 & Kiyotaka Nakagawa 2 Tocopherol (Toc) and tocotrienol (T3) are abundant in rice bran. Geranylgeranyl reductase (GGR) is an essential enzyme for Toc production that catalyzes the reduction of geranylgeranyl pyrophosphate and geranylgeranyl-chlorophyll. However, we found that a rice mutant line with inactivated Os02g0744900 (OsGGR1/LYL1/OsChl P) gene produces Toc, suggesting that rice plants may carry another enzyme with GGR activity. Using an RNA-mediated interference technique, we demonstrated that the Os01g0265000 ("OsGGR2") gene product has GGR activity. This result supports the existence of two GGR genes (OsGGR1 and OsGGR2) in rice, in contrast to Arabidopsis thaliana (thale cress) and cyanobacterium Synechocystis that each have only one GGR gene. We also produced rice callus with inactivated OsGGR1 and OsGGR2 that produced T3 but not Toc. Such rice callus could be used as a resource for production of pure T3 for nutraceutical applications.
Geranylgeranyl reductase (GGR) plays crucial roles in both vitamin E and chlorophyll biosynthesis. GGR is thought to have two functions: reduction of geranylgeranyl pyrophosphate (GGPP) to phytyl pyrophosphate (PPP), and reduction of geranylgeranyl-chlorophyll to chlorophyll 16 . Arabidopsis thaliana encodes only one GGR gene at the gene locus At1g74470. The homologous gene to Arabidopsis GGR in rice (Oryza sativa) is Os02g0744900 (OsGGR1/LYL1/OsChl P), which has a nucleic acid sequence that is 65% identical to that of At1g74470.
T3 is characteristically abundant in rice bran and is known to have greater antioxidant activity 17 , triglyceride-lowering effects 18 , and anti-angiogenesis activity 19 compared to Toc. As such, we aim to develop new varieties of T3-rich rice to produce high-purity T3 without Toc. During this effort, we previously described rice cultivars that are rich in T3 20 , and used quantitative trait loci (QTL) analysis to identify five loci on rice chromosomes that contribute to T3 production 21 .
T3 is biosynthesized from GGPP and homogentisic acid (HGA), whereas Toc is biosynthesized from PPP and HGA. PPP is generated from GGPP via the chlorophyll degradation pathway or direct reduction of GGPP by GGR catalytic activity (Fig. 1). We predicted that if GGR is inactivated, GGPP levels would increase, accompanied by an increase in T3 and absence of Toc synthesis. During the analysis of rice with an inactive GGR mutation, we found evidence for the existence of a second gene involved in GGR synthesis in addition to the Arabidopsis GGR orthologue OsGGR1.
In this study, we analyzed the GGR gene in rice and showed that rice has two GGR genes (OsGGR1 and Os01g0265000 ("OsGGR2")). We also showed that when both genes are inactivated in rice callus, Toc biosynthesis is eventually inhibited.
Quantitative analysis of Toc and T3 content in leaves and callus of OsGGR1 mutant rice. We analyzed the foliar vitamin E content of the three genotypes of the Tos17 mutant and WT. Although the Toc content in the OsGGR1 −/− mutant was significantly decreased compared with the other Tos17 mutants (OsGGR1 +/+ and OsGGR1 +/-) as well as WT, we confirmed the presence of substantial amounts of Toc in the OsGGR1 −/− mutant (Fig. 3A). T3 was not detected in the leaf samples. We also analyzed callus generated from the OsGGR1 Tos17 mutants and WT plants. Similar to the rice leaves, Toc in callus was present in the Tos17 mutants and WT (Fig. 3B). T3 was detected in the callus samples, although the amount of T3 in the Tos17 mutants was lower than that of WT.

Expression analysis of OsGGR2 mRNA. A Basic Local Alignment Search Tool (BLAST) search anal-
ysis revealed that rice has another gene that is similar to OsGGR1. We designated this OsGGR1 homologue Os01g0265000 as "OsGGR2". We next analyzed OsGGR2 mRNA expression because this gene is not registered in the full-length cDNA library database KOME (Knowledge-based Oryza Molecular Biological Encyclopedia) of NARO 23 (database now unavailable). We performed reverse transcriptase (RT)-PCR using the predicted sequences of the 5′ and 3′ non-coding regions as primers (Table S1). OsGGR2 was indeed expressed at the mRNA level in the callus and leaves of seedlings (Fig. 4C). The 1,374 bp nucleotide sequence of the cloned OsGGR2 mRNA coding region is GC-rich (76%) and lacks introns (Fig. 4A). Relative to OsGGR1, the OsGGR2 sequence has 64% and 53% similarity at the nucleic acid and amino acid level, respectively. Amino acid sequence alignment of OsGGR1 and OsGGR2 is presented in Fig. 4B. Further OsGGR2 expression pattern analysis in the grain filling stage showed that OsGGR2 is expressed in bran, the flag leaf, the third leaf from the flag leaf, and the flag leaf sheath (Fig. 4D).  OsGGR1 −/− /OsGGR2 RNAi double mutant. We then analyzed the vitamin E content of callus formed by the double mutant to assess OsGGR2 involvement in vitamin E biosynthesis (Fig. 5A). Results for RT-PCR analysis of OsGGR2 gene expression by a representative double mutant callus (clone No. 9) and average vitamin E content of WT callus, OsGGR1 −/− callus, and OsGGR1 −/− /OsGGR2 RNAi double mutant callus are shown in Fig. 5B and C. The Toc content of the double mutant callus was drastically reduced compared with WT and the OsGGR1 −/− callus. These results indicate that the OsGGR2 gene product has GGR activity and synthesizes Toc in rice plant cells.

Discussion
The vitamin E synthesis pathway (Fig. 1) has been elucidated mainly by studies using Arabidopsis thaliana and the Synechocystis mutant [4][5][6][7][8][9][10][11][12][13]15 . GGR was first identified in Arabidopsis thaliana 16 as an essential enzyme in the biosynthesis of Toc and chlorophyll 24,25 . GGR reduces GGPP to PPP, and also reduce geranylgeranyl-chlorophyll to chlorophyll. In addition to the direct reduction of GGPP to PPP by GGR, hydrolytic cleavage of the chlorophyll phytyl side chain produces phytol, which is then phosphorylated to form PPP. Toc is biosynthesized from PPP and HGA by the catalytic action of VTE2-1, 2, whereas T3 is biosynthesized from GGPP and HGA by the catalytic action of HGGT.
In this study, we prepared OsGGR1 Tos17 mutant rice samples. Since GGR is also necessary for chlorophyll production (Fig. 1), the phenotype of OsGGR1 −/− rice is incomplete albino (Fig. 2C), indicating that OsGGR1 is inactivated in OsGGR1 −/− genotype rice. Moreover, OsGGR1 −/− Tos17 mutant rice plants are sterile. Incidentally, OsGGR1/LYL1/OsChl P mutants isolated via ethylmethanesulfonate (EMS) mutagenesis or 60 Co irradiation are fertile 26,27 . Mutations in the fertile mutants would thus be expected to have moderate effects. Only one GGR gene is present in Arabidopsis thaliana, Nicotiana tabacum, and Synechocystis, which is consistent with the observation that a cyanobacterium mutant carrying inactivated GGR cannot grow photoautotrophically or produce Toc 28 . In this study, the OsGGR1 −/− Tos17 mutant was also unable to grow under photoautotrophic conditions, but the mutant did produce substantial amounts of Toc (Fig. 3A). Likewise, callus on OsGGR1 −/− plants contained Toc (Fig. 3B). Considering these findings (Fig. 3) and the biosynthesis pathway of vitamin E (Fig. 1), this result suggested that rice plants may carry another enzyme that has GGR activity. On the other hand, T3 was present in callus, but not in leaves (Fig. 3). This outcome is likely due to a lack of HGGT expression in rice leaves 29 .
We confirmed the existence of an OsGGR1 homologue in the rice genome using a BLAST search and designated this gene as OsGGR2 (Os01g0265000). We evaluated OsGGR2 expression by RT-PCR because OsGGR2 is not registered in the full-length KOME cDNA clone database (temporarily unavailable). The OsGGR2 gene is expressed in several rice organs, including leaf and bran. The existence of both OsGGR1 and OsGGR2 and their preservation throughout evolutionary history suggests that the functions of these two genes are not redundant and cannot substitute for one another. One possibility for the presence of two rather than one gene is that OsGGR1 and OsGGR2 may be distributed among different cell compartments and work individually by our speculation.
To determine OsGGR2 function in rice cells, we generated OsGGR1 and OsGGR2 double mutant callus tissue by suppressing OsGGR2 gene expression in OsGGR1 −/− genotype rice callus using RNAi. Toc content was drastically decreased in the double mutant callus compared with the OsGGR1 −/− single mutant, but did not reach zero (Fig. 5A,C). This residual production is likely because RNAi cannot completely abolish target transcripts. Moreover, gene expression inhibition efficiency is influenced by the rice genomic locus into which transfer-DNA (T-DNA) is integrated. These results further support the finding that rice has two active forms of GGR, OsGGR1 and OsGGR2.
According to the RiceXpro database 30 , OsGGR2 expression is relatively stronger during the early embryo stage, suggesting that this gene might play an important role in early plant development. Meanwhile, OsGGR1 is expressed strongly in leaves, which is consistent with the pale phenotype seen for the OsGGR1 Tos17 mutant (Fig. 2C). As described above and as shown in Fig. 1, there are two pathways of PPP synthesis, although we did not investigate the extent to which OsGGR2 can contribute to Toc biosynthesis in the two PPP production pathways  13 reported on VTE6, which exhibits phytyl-phosphate kinase activity when PPP is produced in the chlorophyll degradation pathway. According to this report, in Arabidopsis thaliana, PPP production in Toc synthesis occurs mainly through the chlorophyll degradation pathway and not by direct reduction of GGPP. We are currently examining our gene silencing mutant of OsGGR2 to further examine the functional differences between OsGGR1 and OsGGR2 in the two Toc biosynthesis pathways. In addition, enzymatic activity of OsGGR2, detailed comparison of gene expression level of OsGGR1 and OsGGR2, and relationship between message level of OsGGR2 and amount of Toc and T3 will be clarified in our future work.
Unlike Toc, T3 has potent anticancer activity by inhibiting angiogenesis 31 . T3 has thus attracted attention as a preventative and curative agent for diseases, as more than 50 diseases are associated with abnormal angiogenesis, including cancer, age-related macular degeneration and rheumatic diseases. The anticancer activity of T3 may be reduced by Toc through inhibition of its uptake 32 . Owing to their similar molecular structures, separating and purifying T3 from rice bran containing Toc is expensive and cumbersome. The production of Toc-free T3 produced from rice callus (Fig. 5A) would bypass these difficulties, and might be useful to generate pharmaceuticals aimed at suppressing angiogenesis. By generating callus with inactivation of both OsGGR1 and OsGGR2 activity, we showed that rice plant materials contained T3 but not Toc. This approach may provide a new pathway for the purification of T3 without Toc.  (Table S1).

Quantitative analysis of vitamin E content.
Vitamin E was extracted from rice samples with 2-propanol, and the extract was subjected to liquid chromatography with tandem mass spectrometry (LC-MS/ MS) as described previously 33 . Separation was performed at 40 °C using a silica column (ZORBAX Rx-SIL, 4.6 × 250 mm; Agilent, Palo Alto, CA, USA). A mixture of hexane/1,4-dioxane/2-propanol (100:4:0.5) was used as the mobile phase at a flow rate of 1.0 mL/min. Toc and T3 were detected in atmospheric pressure chemical ionization mode (APCI). MS/MS parameters were optimized with Toc and T3 standards in APCI mode (positive). Toc and T3 were detected using multiple reaction monitoring as follows: α-Toc, m/z 431. RT-PCR analysis of rice GGR expression. Total RNA was extracted with an RNeasy Plant Mini Kit ® (Qiagen, Hilden, Germany), followed by genomic DNA digestion with DNase I (TaKaRa, Shiga, Japan) at 37 °C for 30 min. The resulting total RNA was again purified with an RNeasy Plant Mini Kit to remove any remaining genomic DNA and DNase I. cDNA was synthesized from the total RNA using a QuantiTect ® reverse transcription kit (Qiagen) and subjected to PCR performed with appropriate primers (Table S1). We selected OsEF1-α as a positive control gene and used a DNase I-treated RNA template as a negative control. Transformation of rice callus. Transformation of rice callus was performed with Agrobacterium strain EHA 101 containing the gene silencing plasmid pANDA35HK. A partial sequence containing the 5′-noncoding 170 bp and 5′-coding 251 bp region of the OsGGR2 gene was inserted into pANDA35HK, which was a generous gift from the late Dr. Shimamoto and Dr. Miki (former affiliation: Nara Institute of Science and Technology) as reported previously 34 . Agrobacterium-mediated transformation of rice callus was performed according to the method described by Toki et al. 35 .
Statistical analysis. The data, expressed as mean ± SD, were subjected to the Kruskal-Wallis H-test followed by the Student-Newman-Keuls test. Statistical calculation was carried out using ystat 2000, an Excel statistical program file (IgakuTosho Shuppan, Tokyo, Japan). Differences with P < 0.05 were considered significant.