Two commercial varieties of Korean rice, O. sativa L. cv. Dongjin and Ilmi, were used to analyze endogenous gene expression, amplify the genes, isolate protoplasts, and perform a stable transformation. The Crop Biotech Institute in Kyung Hee University (Yongin, South Korea) generously provided a T-DNA-insertional rice mutant. Mature seeds were dehulled, surface-sterilized by 70% ethanol for 2 min and 2% sodium hypochlorite for 40 min, washed five times with distilled water, germinated for a week on Murashige and Skoog agar medium (Duchefa, Haarlem, Netherlands) in a plant growth chamber, transplanted into soil, and cultivated in a greenhouse under the conditions of 14-h light/10-h dark cycle at 28°C or in the paddy field until maturity during the summer. Mature seeds were harvested 60 days after flowering, and their endosperm color was visually compared after dehusking (TR-200 Electromotion rice husker, Kett, Tokyo, Japan) and polishing (Pearlest Polisher, Kett).

Rice samples of diverse organs obtained at different developmental stages and 2-week-old seedlings treated with six phytohormones, including 100-μM abscisic acid (ABA), 100-μM gibberellic acid (GA3), 100-μM auxin (IAA), 100-μM kinetin (KT), 100-μM methyl jasmonic acid (MeJA), and 200-μM salicylic acid (SA), were used for total RNA extraction using mainly RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany) and PureLink^® Plant RNA Reagent (Invitrogen, Waltham, MA, United States) for seed case. The real-time quantitative reverse transcriptase PCR (qRT-PCR) was performed using iQ™ SYBR^® Green Supermix (Bio-Rad) and CFX Connect™ Real-Time System (Bio-Rad), with PCR conditions and data calculated relative to the rice ubiquitin 5 gene (Os01g22490) to normalize RNA amounts, as previously described (You et al., 2020), and each qRT-PCR was performed in three technical repeats. Semiquantitative RT-PCR was performed under the following conditions: a cycle at 95°C for 3 min, 30 cycles at 95°C for 15 s, 55°C for 15 s, 72°C for 1 min/kb, followed by 3 min at 72°C with KOD FX DNA polymerase (Toyobo, Osaka, Japan).

We extracted the rice genomic DNAs using a Genomic DNA Preparation Kit (QIAGEN). The genotyping PCRs were performed using an EmeraldAmp^® PCR Master Mix (TaKaRa Bio, Shiga, Japan). To examine the transgene homozygosity, TaqMan PCR was performed to detect Bar as a transgene with an internal reference in the rice genome, i.e., the α-tubulin gene (Os11g14220), as previously described (Ha et al., 2019). Supplementary Table 10 contains all sequence information of primers.

The coding region of OsIspH1 (Os03g0731900) was amplified using gene-specific primer pair from the total RNA of 10-day-old seedlings of rice (O. sativa L. cv. Ilmi) and cloned into the pDONR221 vector using the Gateway^® BP^® Clonase II Enzyme Mix (Invitrogen), yielding pDONR221-OsIspH1. To determine the subcellular localization, OsIspH1 was fused to the N-terminus of superfolder green fluorescent protein (sGFP) and cloned into the pB2GW7 binary vector using Gateway^® LR Clonase^® II Enzyme Mix (Invitrogen). Protoplast isolation, DNA transfection, and confocal microscopy analysis were conducted, as previously described (You et al., 2016). The fluorescence was detected and imaged using a confocal laser scanning microscope (Carl Zeiss LSM 880, Jena, Germany).

Using the pDONR221-OsIspH1 as a template, two non-synonymous substitution mutations (H145P and K407R) in OsIspH1 were created to have one or both of each through site-directed mutagenesis with sequence-substituted primers (Supplementary Table 10). Original and mutated genes were cloned into pMW118 (Nippon Gene, Tokyo, Japan) and transformed into the E. coli DLYT1 strain, which was disrupted in LytB/EcIspH encoding E. coli HDR gene (Kim et al., 2008). An E. coli DLTY1 strain transformed with pTTQ18-EcIspH/LytB was used as a positive control for complementation analysis. The complementation tests were performed on the Luria-Bertani (LB) medium containing 50 μg/ml kanamycin, 100 μg/ml spectinomycin, and 0.01% (w/v) (±)-mevalonolactone (MVA) (Sigma-Aldrich, St. Louis, MO, United States).

After amplification of a double mutated gene using the pMW118-OsIspH1^H145P/K407R as a template, a constitutive expression cassette was prepared by consecutive cloning into the pDONR221 vector and the p600-PGD1 vector containing a rice phosphogluconate dehydrogenase 1 promoter using the Gateway procedures (Park et al., 2010), resulting in pIspH1^MUT. It was further introduced into pGlb:stPAC vector, in which endosperm specifically produces β-carotene (Jeong et al., 2017), resulting in pIspH1^MUT-stPAC. Two final binary vectors were transformed via triparental mating of Agrobacterium LBA4404 harboring pSB1 plasmids and cocultivated with embryogenic calli differentiated from matured rice seeds (O. sativa L. cv. Ilmi), as previously described (You et al., 2020). The putative transgenic plants were generated on a selection medium containing phosphinothricin (4 mg/L) and cefotaxime (500 mg/L) using shooting and rooting procedures, according to a previously described method (Ko et al., 2018).

In planta complementation assay was performed using conventional interbreeding during the flowering season at the T3 generation between heterozygous plants of osisph1 as the female parent and homozygous pIspH1^MUT-overexpressed plants as the male parent. Filial (F) 1 progenies were self-pollinated in the paddy field for three more generations to obtain homozygous seeds for transgene traits of both parents.

Rice leaves and dehusked seeds were ground to a fine powder using a pestle in liquid nitrogen. Total chlorophylls were extracted from rice leaf powders with 100% methanol at 70°C for 30 min using ThermoMixer Comfort (Eppendorf AG, Hamburg, Germany) at 500 rpm and centrifuged for 10 min at 4°C and 3,000 rpm. A spectrophotometer was used to measure the absorbance of the supernatant at 666 and 653 nm (Mecasys Co., Daejeon, South Korea). The chlorophyll content was calculated using the formula stated by Wellburn (1994).

Carotenoids were extracted from 100 mg powders of each rice leaf and seed, following the previous report with minor modifications (Kim et al., 2010). Notably, 0.1% ascorbic acid-containing ethanol was added to 100 mg powder to be incubated at 85°C for 5 min, after vortex-mixing for 20 s. After 10 min of saponification with potassium hydroxide (80%, w/v) in an 85°C water bath, samples were immediately placed on ice, and we added cold deionized water with β-apo-8′-carotenal (25 μg/ml) as an internal standard. Hexane-extracted carotenoid layers were separated two times by centrifugation at 1,200 × g. Aliquots were dried under a nitrogen stream and redissolved in 50:50 (v/v) dichloromethane/methanol solution. High-performance liquid chromatography (HPLC) was performed on a C30 YMC column (4.6 mm × 250 mm, 3 μm; Waters Corporation, Milford, MA, United States) using gradient elution at 1 ml/min with solvent A [methanol/water (92.8, v/v) with 10 mM ammonium acetate] and B (100% methyl tert-butyl ether) under the following conditions: 0 min, 83% A/17% B; 23 min, 70% A/30% B; 29 min, 59% A/41% B; 35 min, 30% A/70% B; 40 min, 30% A/70% B; 44 min, 83% A/17% B; 55 min, 83% A/17% B. Chromatograms were generated at 450 nm. The calibration curves were set using 5 concentrations of carotenoid standards, except for lycopene, from 0.3 to 5.0 μg/ml based on the peak area ratios obtained with the internal standard. Linearity was tested by least-squares regression analysis of the corrected peak area ratios against increasing weight ratios. The calibration curves of lutein, zeaxanthin, α-carotene, and β-carotene were linear (r² = 0.9988–0.9998). Lycopene undergoes degradation via isomerization and oxidation. Thus, the quantitative calculation of lycopene was based on the peak area of 5.0 μg/ml of lycopene standard (Kim et al., 2017).

To investigate the evolutionary relationship of the rice IspH family, the predicted amino acid sequences of 79 IspHs were first collected from GreenPhylDB¹, confirmed whether they contained a full-length open reading frame (ORF) in the National Center for Biotechnology Information Web BLAST², and finally selected for the phylogenetic tree analysis using the maximum-likelihood method (Figure 1). As a result, the tree classified them into six clades, namely, bacteria, mosses, ferns, gymnosperms, monocotyledons, and dicotyledons, as they showed a sequential relationship based on general evolutionary processes. All bacteria have a single IspH gene, whereas plants possess one to three copies. Considering the closer relationship between cyanobacteria and plants, it was proposed that plants obtain IspH genes from cyanobacteria via endosymbiosis, which might have evolved independently with non-photosynthetic bacteria from the common prokaryotic ancestor, as previously reported (Hsieh and Hsieh, 2015). Particularly, family genes belonging to the same species were mostly grouped into the same clade or closely related, indicating that they might be multiplicated during evolutionary processes.

In the O. sativa genome, two IspH genes with 10 introns, OsIspH1 (Os03t0731900-01) and OsIspH2 (Os03t0732000-00), designated in this study are annotated in Rice Annotation Project Database (RAP-DB). They have 68% amino acid sequence homology and are adjacent genes about 2.8 Mbp apart in the same chromosome 3. In monocotyledon plants, IspH genes are analyzed to exist as two copies in date palm (Phoenix dactylifera), banana (Musa acuminata), and a domesticated-rice (O. sativa; AA genotype) genomes and as just a single copy in other wild rice species (Oryza brachyantha, FF genotype; Oryza meyeriana, GG genotype), sorghum, corn, barley, wheat, and so on (Figure 1). As shown in the phylogenetic tree, IspH genes exist mostly as two or three copies in moss, fern, and gymnosperm genomes, one or two copies in monocotyledon genomes of angiosperm plants, and mostly a single copy in bacteria, dicotyledon genomes of angiosperm plants, indicating that the copy number of IspH genes has evolved to increase from a single copy in bacteria to two or three copies in gymnosperms but has evolved to decrease back to a single copy in angiosperms, during the genomic evolution. Of two rice IspH genes, an OsIspH1 was evolutionarily grouped with a single-copy IspH gene of other monocotyledon plants in a phylogenetic tree, but OsIspH2 was distant from them (Figure 1), supposing that OsIspH1 might have a conservative IspH function in angiosperms, rather than OsIspH2 having it.

To define the structures of two rice IspH genes, we extracted genomic DNA sequence information from the RAP-DB³ and constructed their genome structures, which are equally composed of ten exons and nine introns (Figure 2A). The 1st and 3rd introns of OsIspH2 have unusually longer sequences, 1,061 bp and 13.5 kbp, respectively, compared with 652 and 409 bp of OsIspH1 when predicted only with the genomic DNA sequence because the cDNA of OsIspH2 has never been reported. It meant that the functionality of OsIspH2 was largely unknown.

To compare the expression patterns of two rice IspH genes, we examined them in various tissues at different developmental stages with the gene-specific primers located in a 3′-untranslated region (Figure 2B and Supplementary Table 3). The spatial expression pattern revealed that OsIspH1 transcripts were 20–40 times higher in the leaves than roots at all developmental stages, indicating a preference for leaves. They were also expressed in florets and seeds. OsIspH2 was not expressed in any leaf tissue and florets but was found in low levels in roots and seeds, indicating that OsIspH1 might be mainly functional in all rice tissues. Furthermore, we examined the expression patterns of two rice IspH genes against the treatments of six phytohormones, namely, ABA, CK, GA, IAA, MeJA, and SA (Figure 2C and Supplementary Table 4). Transcript levels of OsIspH1 were slightly suppressed by ABA but were induced by other five hormones in early time posttreatment, proposing different responses. Those of OsIspH2 were barely detectable before and after treatment.

To simultaneously identify the cDNAs of two rice IspH genes, total RNAs were prepared from roots at the seedling stage using the spatial expression pattern in Figure 2B, and RT-PCRs were performed using diverse primer combinations (Figure 2D). OsIspH1 exhibited a single 1,410-bp-long band (including 1,380 bp encoding 459 amino acids) as predicted in amplification, including the full-length ORFs; however, OsIspH2 unexpectedly showed a longer single 2,705-bp band than the predictable 1,644 bp (including 1,449 bp encoding 482 amino acids). Through further RT-PCRs of OsIspH2 with different combinations of exon-specific primers and sequencing of corresponding amplicons, we confirmed the only splicing error in the 1st intron, generating the longer 1st exon, shorter 1st intron, and quite longer 2nd exon than expected size, as indicated by the dotted line in Figure 2A. Interestingly, the extended 1st exon region contained the early stop codon, indicating that OsIspH2 could be a non-functional gene by causing a disabled small ORF of 324 bp encoding 107 amino acids.

Accordingly, OsIspH1 turned out to be the only functional HDR gene in rice. Its localization was examined using a GFP-fused protein transformation in rice protoplasts prepared from green seedlings and confirmed in stroma inside chloroplasts, with a similar pattern to stroma marker, rice ribulose-1,5-bisphosphate carboxylase/oxygenase small subunit transit peptide, but different from cytosol marker (Figure 2E).

To evaluate the role of OsIspH1 in rice plants, we obtained a rice mutant with a T-DNA insertion in the region of 5′-untranslated region of OsIspH1 (PFG_1B-11039)⁴, as depicted in Figure 2A. We confirmed that the osisph1 knockout mutant exhibits albinism in homozygous plants based on genotyping and qRT-PCR (Figure 2F), and the albino phenotypes were observed at a 3:1 segregation ratio to identify a single-copy insertion of T-DNA, supposing that the albino phenotype might be derived from the knockout of OsIspH1 gene functions. Similarly, another knockout mutant of an OsIspH1 gene showed also the lethal albinism and the defectiveness of chloroplast biogenesis, which was constructed by a CRISPR-Cas9 gene-editing system (Liu et al., 2020), supposing that the albinism phenotype of osisph1 might also be mediated by the defective chloroplast biogenesis. Taken together, OsIspH1 is suggested to be crucial in photosynthetic pigments, including each or both chlorophylls and carotenoids, and that OsIspH2 could not compensate for the loss of function of OsIspH1.

To confirm the conservation in key residues of OsIspH1, we aligned amino acid sequences among 12 monocot plants, including rice (Figure 3). Previously known key residues, an N-terminal conserved domain (NCD), four Cys-residues consisting of Fe-S clusters at active sites, and eight substrate-binding sites are highly conserved among these IspHs, indicating that OsIspH1 could function as an HDR. Interestingly, a potential key residue, such as protein kinase C (PKC) phosphorylation site in the C-terminal region, was newly predicted to be conserved among IspHs in the corresponding position of OsIspH1, Ser-Tyr-Lys^405––407, through a Motif Scan analysis⁵, and Lys⁴⁰⁷ was also predicted as one of the ubiquitination sites through computational prediction of ubiquitination site⁶.

Referring to the well-studied key residue of His¹⁵² in Arabidopsis (Hsieh et al., 2014), we mutated OsIspH1 to have non-synonymous substitutions in the corresponding His¹⁴⁵ residue and newly predicted Lys⁴⁰⁷ residue into Pro¹⁴⁵ and Arg⁴⁰⁷, respectively. Two single-mutated OsIspH1^H145P and OsIspH1^K407R genes and a double-mutated gene of OsIspH1^H145P/K407R were transformed into an E. coli IspH1/LytB-lesion mutant strain, a DLYT1 (Kim et al., 2008). As a result, OsIspH1, OsIspH1^H145P, and OsIspH1^K407R successfully complemented the lethal phenotype of DLYT1 when EcIspH1/LytB was transformed as a positive control (Figure 4A). It confirmed that individual single mutation does not affect HDR activity. However, a double-substitution mutant, OsIspH1^H145P/K407R (also called OsIspH1^MUT hereafter), barely restored DLYT1 growth, indicating the complementary relationship of two residues for HDR enzyme activity. Additionally, to inquire whether substitution mutations of amino acids cause structural change in OsIspH1, we simulated the three-dimensional modeling based on PDB ID: 3DNF from A. aeolicus IspH using SWISS-MODEL⁷, as depicted in Figure 4B. According to a previous report, the first mutation from a positively charged His to a non-polar Pro is crucial for the “induced fit” effect to bind the substrate HMBPP (Hsieh et al., 2014). In this study, the predicted modeling showed the possibility that 1st key His¹⁴⁵ residue might be faced to the Lys⁴⁰⁷ residue at nearly position (Figure 4B), supporting the possibility that the complementary relationship of two residues might be important for HDR enzyme activity in rice plants, although the mechanism is largely unknown.

Based on the albino phenotype of the loss-of-function mutants, osisph1, as depicted in Figure 2F, we constitutively overexpressed an OsIspH1^MUT to elucidate how an OsIspH1 affects chlorophyll and carotenoid biosynthesis in rice leaves and seeds. To learn more about its effects on carotenoid metabolism in seeds involved in the Golden Rice trait, the OsIspH1^MUT cassette was further combined with a stPAC cassette, which produces β-carotene in rice endosperm (Jeong et al., 2017), as a single T-DNA vector (Figure 5A). Using three-independent homozygous T3 plants, which were selected by considering the lower copy number of one and two (Figure 5B), the expression patterns of a double-mutated transgene and an endogenous gene encoding rice HDR/IspH1 were examined in leaves (Figure 5C) and seeds (Figure 5D) of T4 rice plants. The overexpression of transgene, OsIspH1^MUT, in both organs and seed-specific expression of stPAC was confirmed relative to non-transgenic (NT) rice plants. Interestingly, the ectopic expression of an OsIspH1^MUT distinctly affected the expression patterns of endogenous OsIspH1 in both organs. In leaves of the IspH1^MUT transgenic plants, the expression of endogenous OsIspH1 was significantly suppressed by an average of about 50% (P < 0.001 in IspH1^MUT and P < 0.05 in IspH1^MUT-stPAC), and the combined transcript levels of OsIspH1^MUT and OsIspH1 were analyzed to be also significantly lower than the normal transcript level of endogenous OsIspH1 (P < 0.01 in IspH1^MUT and P < 0.05 in IspH1^MUT-stPAC), even under the ectopic expression of OsIspH1^MUT transgene (Figure 5C and Supplementary Table 5). In contrast to the suppression patterns in leaves, the combined transcript levels of OsIspH1^MUT and OsIspH1 were analyzed to be at least 4 times higher than the transcript level of endogenous OsIspH1 (P < 0.001 in IspH1^MUT and P < 0.05 in IspH1^MUT-stPAC), and the transcript levels of endogenous OsIspH1 were not suppressed in seeds of all transgenic plant (Figure 5D and Supplementary Table 6), indicating that OsIspH1^MUT was expressed four times more than the intrinsic OsIspH1 gene. Finally, the distinct expression patterns between leaves and seeds were statistically significantly conservative in six independent transgenic plants of IspH1^MUT and IspH1^MUT-stPAC, and taken collectively, it was supposed that the expression of OsIspH1 might be differentially regulated between leaves and seeds.

In the next step, we analyzed two terpene-derived photosynthetic pigments, including chlorophylls and carotenoids, using HPLC. In leaves, although there were differences between the independent transgenic plants, the total contents of chlorophylls (an average of 84.6% level, Supplementary Table 7) and carotenoids (an average of 84.3% level, Supplementary Table 8) were decreased in two transgenic plants with the significant differences compared to NT plants (Figure 6A). In addition to the functional complementation of OsIspH1^MUT as HDR/IspH in E. coli (Figure 4A), in planta functional complementation was examined through cross-interbreeding between two transgenic rice plants, namely, osisph1 and IspH1^MUT. In rice plants, the albinism of osisph1 was not restored by OsIspH1^MUT overexpression at all, of which homozygosity was ascertained by genotyping (Figure 6B), indicating that OsIspH1^MUT was a protein defective in HDR activity in rice plants. Collectively, it was supposed that the overexpression of OsIspH1^MUT might cause the relative decrease in active HDR proteins through the competition of OsIspH1^MUT with the endogenous OsIspHs and consequently decrease chlorophyll and carotenoid contents in the leaves. In the seeds, the carotenoid content of IspH1^MUT (Figure 7A) and IspH1^MUT-stPAC (Figure 7B) was statistically significantly reduced on the average of 61.4 and 54.2%, respectively, compared to NT, indicating a more severe pattern in seeds than leaves (Figure 7 and Supplementary Table 9). The results were considered to be closely related to the higher expression level of OsIspH1^MUT than the endogenous OsIspH1 gene in the seeds.

Taken together, OsIspH1 is supposed to have crucial roles such as the rate-limiting regulation on carotenoid biosynthesis in seeds and photosynthetic pigments in leaves.

The MEP pathway consists of seven enzymatic steps that start with the catalysis of DXS and end with that of IspH. While the five intermediate enzymes are encoded by a single gene, DXS and IspH have been reported as one or multiple genes in diverse plants, proposing the potent roles as flux-controlling steps for terpenoid production. Plant DXSs have been classified into three clades, with different influences on the downstream pathways: DXS1s for chlorophyll biosynthesis with a housekeeping role, DXS2s for the production of specialized terpene metabolites, including artemisinins, ginkgolides, resins, and carotenoids, in different plants, and DXS3 for considerably understudied but still tentative roles in a few plants, as recently summarized (Cordoba et al., 2011; Zhang et al., 2018; de Luna-Valdez et al., 2021; Tian et al., 2021). Meanwhile, plant IspH has been reported as a single-gene family in most plant species but as a multiple-gene family of 2–3 copies in some species. In G. biloba and Pinus taeda of gymnosperm plants, GbIspH1 and PtIspH1 are for housekeeping roles, and GbIspH2 and PtIspH2 are for individually specialized ginkgolide and resin roles (Kim et al., 2008, 2009). In angiosperm plants, two C. melo genes showed different spatial expression: CmIspH1 for the most abundant in all organs and CmIspH2 in specialized contexts, such as de-etiolating seedlings and orange-fleshed fruit during ripening (Saladie et al., 2014). Thereby, it could be proposed that IspH1s play housekeeping roles, but IspH2s can contribute to plant species-specific terpene production. In the case of O. sativa, two IspH genes are predicted in the genome and have been previously presumed the different function: OsIspH1 as a predominantly light-induced gene but OsIspH2 as an upregulated gene in the dxr mutants without light response, proposing a compensating route awaiting validation (Jung et al., 2008). However, Liu et al. (2020) presented that the genome-edited OsIspH1 defective mutant showed the albino lethal phenotype, and similarly in this study, the T-DNA mutant of OsIspH1, osisph1, showed the albino lethal phenotype (Figure 6), supposing that OsIspH2 might not have any compensate function for it. Interestingly in this study, the sequence analysis of OsIspH2 mRNA presented that the extended 1st exon region had an early stop codon to consequentially generate the truncated-short polypeptide fragments, which was spliced at the quite different region from the predicted-splicing position provided by the RAP-DB (Figures 2A,D). Collectively, an OsIspH2 gene is supposed to be non-functional, indicating that an OsIspH1 is a unique functional HDR gene in rice.

The X-ray crystal structure of IspH protein has been first reported in the hyperthermophilic eubacterium A. aeolicus (Rekittke et al., 2008), its critical residues have been elucidated mainly in bacterial systems, such as A. aeolicus, E. coli, and cyanobacteria (Rekittke et al., 2008; Grawert et al., 2009; Hsieh and Hsieh, 2015). Meanwhile, the plant-type IspH structure has also been reported in Arabidopsis and the same key residues to bacterial ones by complementation assay in E. coli isph mutant (Hsieh et al., 2014). By Hsieh et al. (2014), two bacterial critical residues showed the difference in the corresponding ones of Arabidopsis, His¹⁵² and His²⁴¹, not individually but cooperatively critical for HDR activity with more significance of His¹⁵² than His²⁴¹. In this study, we found another conserved motif, referred to as a predicted PKC phosphorylation site in Figure 3, and to determine whether it could be a novel crucial residue of OsIspH1, we mutated each or together at both positions of His¹⁴⁵ and Lys⁴⁰⁷ to generate OsIspH1^H145P and OsIspH1^K407R, and OsIspH1^H145P/K407R, and the former is the corresponding site to His¹⁵² of AtIspH, and the latter is a newly predicted site for PKC phosphorylation or ubiquitination in this study. The complementary experiments in the isph mutants of bacteria and rice plants showed that the simultaneous mutation of His¹⁴⁵ and Lys⁴⁰⁷, an OsIspH1^H145P/K407R, largely reduced the HDR activity of OsIspH1 not only in the bacteria expression system (Figure 4A) but also in the rice plant system (Figure 6B), but every single mutation did not have any effects on the HDR activity of it in bacteria system. These results supposed that the complementary relationship of two residues, namely, His¹⁴⁵ and Lys⁴⁰⁷, might be crucial to the HDR activity of OsIspH1.

The regulation of plant genes involved in MEP and MVA pathways occurs at multiple levels of transcription, posttranscription, translation, and posttranslation (Cordoba et al., 2009; Hemmerlin, 2013; Vranova et al., 2013). Particularly, plant IspHs contribute to the MEP pathway primarily through transcriptional regulation with tissue specificity and different responses against environmental conditions, such as light, circadian, mechanical wounding, and fungal elicitors in diverse plants (Kim et al., 2008, 2009; Cordoba et al., 2009; Ma et al., 2017). To scrutinize the intrinsic rice IspH regulation, we constitutively overexpressed the malfunctional OsIspH1^MUT (Figure 5) in rice plants, which was identified in two hdr mutants, namely, E. coli DLYT1 strains (Figure 4) and rice osisph1 plants (Figure 6B). Interestingly, the effects of OsIspH1^MUT-ectopic expression were observed to be quite differential between rice leaves and seeds. In leaves, not only the transcript level of an endogenous OsIspH1 was downregulated, but also the combined transcript level of endogenous OsIspH1 and exogenous OsIspH1^MUT remained lower than that of NT (Figure 5C). In contrast, the expression of an endogenous OsIspH1 was not downregulated in the seeds and remained higher than that of NT (Figure 5D), suggesting that the expression of an intrinsic OsIspH1 gene was differentially regulated in leaves and seeds. Similarly, organ-specific differential roles of OsDXS2 and OsDXR on carotenoid accumulation between leaves and seeds have also been reported (You et al., 2020). Collectively, these results suggest that the DXS, DXR, and IspH steps belonging to the MEP pathway might be regulated to differentially affected carotenoid accumulation between leaves and seeds.

Despite such differential regulation, the ectopic expression of OsIspH1^MUT simultaneously caused the decrease of chlorophyll and carotenoid content in leaves and seeds (Figures 6, 7), and the reduction patterns were proportional to the overexpression patterns of OsIspH1^MUT, indicating the relative decrease in active HDR enzyme function by competition between endogenous OsIspH1 and OsIspH1^MUT. In other words, the intrinsic OsIspH1 is supposed to play an important role in the biosynthesis of chlorophyll and carotenoid levels in rice leaves and seeds and to be a promising target to biofortify the functional terpene metabolites in rice plants.