Cloning and Expression Analysis of MEP Pathway Enzyme-encoding Genes in Osmanthus fragrans

The 2-C-methyl-d-erythritol 4-phosphate (MEP) pathway is responsible for the biosynthesis of many crucial secondary metabolites, such as carotenoids, monoterpenes, plastoquinone, and tocopherols. In this study, we isolated and identified 10 MEP pathway genes in the important aromatic plant sweet osmanthus (Osmanthus fragrans). Multiple sequence alignments revealed that 10 MEP pathway genes shared high identities with other reported proteins. The genes showed distinctive expression profiles in various tissues, or at different flower stages and diel time points. The qRT-PCR results demonstrated that these genes were highly expressed in inflorescences, which suggested a tissue-specific transcript pattern. Our results also showed that OfDXS1, OfDXS2, and OfHDR1 had a clear diurnal oscillation pattern. The isolation and expression analysis provides a strong foundation for further research on the MEP pathway involved in gene function and molecular evolution, and improves our understanding of the molecular mechanism underlying this pathway in plants.


Introduction
Osmanthus fragrans, also known as sweet osmanthus, sweet olive, and tea olive, is a traditional aromatic flowering tree that is native to China and has been cultivated for over 2500 years. It is considered to be one of top 10 Chinese traditional flowers and is also cultivated as an urban ornamental tree [1]. Owing to its pleasant aroma and evergreen properties, it is now widely distributed in Asian countries, such as China, Japan, Thailand, and India [2]. Today, 166 registered cultivars of O. fragrans have been classified into five groups based on morphological characteristics and growth habit. These are the Luteus group, the Albus group, the Aurantiacus group, the Asiaticus group, and the Colour group [3,4]. Generally, the cultivars in the Luteus group have golden-yellow flowers that only appear in the fall, whereas the Asiaticus group cultivars bloom all year round and have creamy-yellow flowers [5]. The fresh flowers are very abundant in aromatic compounds, including terpenoids, fatty acid derivatives, and phenylpropanoids/benzenoids [6,7]. Although the relative contents of the volatiles vary among different cultivars and developmental stages, the main aromatic components are the terpenoids, including monoterpenes ocimene and linalool [8][9][10]. Most of these substances are the primary components of perfumes and essential oils [11]. Because of the importance of these terpenoid

Plant Materials
Two cultivars of O. fragrans "Boye Jingui" and O. fragrans "Rixiang Gui" were grown in the campus of Nanjing Forestry University in Jiangsu, China. Florets of cymose inflorescences (FCI) with the same anthesis were harvested at bud-eye stage (S1), primary blooming stage (S2), full blooming stage (S3), and flower fading stage (S4) in September 2014. For tissue-specific gene expression studies, roots, stems and leaves, as well as florets of cymose inflorescences at full blooming stage were collected in September 2014. Materials used for diel analysis were collected every two hours (from 2:00 am to 24:00 pm) at full blooming stage (S3) on 11 October 2015. All these samples were immediately frozen in liquid nitrogen and stored at −80 °C for further use.

Plant Materials
Two cultivars of O. fragrans "Boye Jingui" and O. fragrans "Rixiang Gui" were grown in the campus of Nanjing Forestry University in Jiangsu, China. Florets of cymose inflorescences (FCI) with the same anthesis were harvested at bud-eye stage (S1), primary blooming stage (S2), full blooming stage (S3), and flower fading stage (S4) in September 2014 ( Figure 2). For tissue-specific gene expression studies, roots, stems and leaves, as well as florets of cymose inflorescences at full blooming stage were collected in September 2014. Materials used for diel analysis were collected every two hours (from 2:00 a.m. to 24:00 p.m.) at full blooming stage (S3) on 11 October 2015. All these samples were immediately frozen in liquid nitrogen and stored at −80 • C for further use.

Total RNA Extraction and Gene Cloning
The total RNA was obtained from the florets of cymose inflorescences of O. fragrans using RNAprep pure Kit (Tiangen Biotech, Beijing, China). The obtained RNA ratio of A260/280 was quantified by NanoDrop 2000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA). The RNA integrity was evaluated by 1.5% agarose gel electrophoresis. Then the first strand cDNA reaction with 1 μg total RNA was performed using RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific). According to the MEP pathway unigene sequences from the transcriptomic data of O. fragrans, the specific primers were designed to clone OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfMDS, OfHDS, OfHDR1, OfHDR2, and OfIDI (Table S1). By using LA Taq (Takara Biotechnology, Dalian, China), purified DNA fragments of polymerase chain reaction (PCR) were ligated into pEASY ® -T1 cloning vector (Transgen Biotech, Beijing, China) and transformed into E. coli chemically competent cells. Positive recombinant clones were identified and sequenced using the universal M13 primers.

Cloning of Full Length Genes by RACE
Rapid amplification of cDNA ends (RACE) was used to obtain the 3′ ends and 5′ ends of target genes according to the manufacture's procedure (Takara Biotechnology). The specific primers for 3′ RACE and 5′ RACE were designed using Oligo 6.0 software based on the obtained partial sequences. The primer sequences and PCR conditions were listed (Table S2). By sequential nested PCR, these unknown regions were amplified and sequenced. Then full-length genes were assembled together by the Lasergene 7.0 software (Dnastar, Madison, WI, USA). The coding regions were confirmed by PCR detection from start codon to stop codon.

Gene Expression Analysis
Following the MIQE guidelines, the primers of target genes for quantitative real-time PCR (qRT-PCR) were selected using primer premier 5.0 software (Premier biosoft, Palo Alto, CA, USA) (Table  S3), and the absence of hairpin structure and primer dimer were predicted by Oligo 6.0 software (Molecular biology insights, Colorado Springs, CO, USA) [40]. Total RNA preparation was performed as described previously according to the manufacturerʹs instructions. Then first strand cDNA was synthesized from 1 μg total RNA and diluted five-fold for gene expression experiment. The qRT-PCR experiment was carried out by using an ABI StepOnePlus Systems (Applied Biosystems, Carlsbad, CA, USA) and SYBR Premix Ex Taq (Takara Biotechnology). The PCR conditions were as follows: 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s, and 58 °C for 30 s.

Total RNA Extraction and Gene Cloning
The total RNA was obtained from the florets of cymose inflorescences of O. fragrans using RNAprep pure Kit (Tiangen Biotech, Beijing, China). The obtained RNA ratio of A260/280 was quantified by NanoDrop 2000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA). The RNA integrity was evaluated by 1.5% agarose gel electrophoresis. Then the first strand cDNA reaction with 1 µg total RNA was performed using RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific). According to the MEP pathway unigene sequences from the transcriptomic data of O. fragrans, the specific primers were designed to clone OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfMDS, OfHDS, OfHDR1, OfHDR2, and OfIDI (Table S1). By using LA Taq (Takara Biotechnology, Dalian, China), purified DNA fragments of polymerase chain reaction (PCR) were ligated into pEASY ® -T1 cloning vector (Transgen Biotech, Beijing, China) and transformed into E. coli chemically competent cells. Positive recombinant clones were identified and sequenced using the universal M13 primers.

Cloning of Full Length Genes by RACE
Rapid amplification of cDNA ends (RACE) was used to obtain the 3 ends and 5 ends of target genes according to the manufacture's procedure (Takara Biotechnology). The specific primers for 3 RACE and 5 RACE were designed using Oligo 6.0 software based on the obtained partial sequences. The primer sequences and PCR conditions were listed (Table S2). By sequential nested PCR, these unknown regions were amplified and sequenced. Then full-length genes were assembled together by the Lasergene 7.0 software (Dnastar, Madison, WI, USA). The coding regions were confirmed by PCR detection from start codon to stop codon.

Gene Expression Analysis
Following the MIQE guidelines, the primers of target genes for quantitative real-time PCR (qRT-PCR) were selected using primer premier 5.0 software (Premier biosoft, Palo Alto, CA, USA) (Table S3), and the absence of hairpin structure and primer dimer were predicted by Oligo 6.0 software (Molecular biology insights, Colorado Springs, CO, USA) [40]. Total RNA preparation was performed as described previously according to the manufacturer's instructions. Then first strand cDNA was synthesized from 1 µg total RNA and diluted five-fold for gene expression experiment. The qRT-PCR experiment was carried out by using an ABI StepOnePlus Systems (Applied Biosystems, Carlsbad, CA, USA) and SYBR Premix Ex Taq (Takara Biotechnology). The PCR conditions were as follows: 95 • C for 30 s, followed by 40 cycles of 95 • C for 5 s, and 58 • C for 30 s. The qRT-PCR for each sample was repeated three times. Previous validated genes OfRAN, OfRPB2, and OfACT were used as internal normalizations for different organs, different flowering stages and diel variations, respectively [12]. Each primer pair was validated the specificity by melt curve analysis, and the gene expression levels were calculated by the 2 −∆∆CT method. The qRT-PCR results were analyzed by using ABI StepOne software (Applied Biosystems).

Cloning and Sequence Analysis of MEP Pathway Genes
To clone OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfMDS, OfHDS, OfHDR1, OfHDR2, and OfIDI, degenerate primers were designed to obtain 2232 bp, 1085 bp, 1453 bp, 596 bp, 1306 bp, 628 bp, 2366 bp, 359 bp, 604 bp, and 669 bp sized amplicons respectively. Based on the partial gene sequences, the 3 region and 5 region were amplified and sequenced by RACE. The size of full-length cDNA, open reading frame (ORF), amino acids, molecular weight, and isoelectric point (pI) were listed (Table 1). We submitted the 10 sweet osmanthus full-length cDNAs of MEP genes to NCBI GenBank with the accession number KX400841-KX400850. The amino acid sequences of 10 MEP genes were aligned with other plants to reveal identities and conserved domains in the NCBI database. Multiple alignments showed that the identities ranged from 73% to 92%, in which OfDXS1, OfDXR, OfHDS, and OfIDI shared the higher identities between 88% and 92%, OfMCT, OfCMK, and OfMDS shared the lower identities between 73% and 82%, and OfDXS2 had the intermediate identity between 85% and 87%. Both OfDXS1 and OfDXS2 contained three conserved domains, a thiamine diphosphate-dependent domain at the N-terminus, a pyrimidine binding domain at the medial position, and a transketolase domain at the C-terminus. OfDXR also contained three reductoisomerase domains, which was located at the position 80-208 aa, 222-305 aa, and 337-459 aa. OfMCT had an IspD domain at the C-terminus, which is also known as ygbP domain. OfIDI showed a NUDIX hydrolase domain at the position 53-205 aa. The amino acid sequences among MEP proteins were composed of multiple conserved residues, which were crucial to form distinct dimensional structures and specific biological functions (Table S4).

Expression Analysis of MEP Genes in Different Organs
To investigate the tissue-specific expressions of the MEP genes, the transcript levels of two O. fragrans cultivars, "Boye" and "Rixiang", were detected in different organs including roots, stems, leaves, and inflorescences by using qRT-PCR. The transcript levels of OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfMDS, OfHDS, OfHDR1, OfHDR2, and OfIDI were measured ( Figure 3).  The tissue-specific results suggested that OfDXS1, OfDXS2, OfDXR, OfCMK, and OfHDR2 expressions were significantly abundant in the inflorescences, compared with other organs. In cultivar "Boye", the OfDXS1 and OfCMK transcript levels in the inflorescences were almost 8-fold and 22-fold higher than that in the roots respectively, while in cultivar "Rixiang" they were only 2.8fold and 11-fold, respectively. In both cultivar "Boye" and "Rixiang", OfDXR showed over 15-fold transcript levels in the inflorescences than in the roots. As for OfDXS2, the inflorescences were found to contain the highest transcript level. In cultivar "Boye", the OfDXS2 transcript level in the The tissue-specific results suggested that OfDXS1, OfDXS2, OfDXR, OfCMK, and OfHDR2 expressions were significantly abundant in the inflorescences, compared with other organs. In cultivar "Boye", the OfDXS1 and OfCMK transcript levels in the inflorescences were almost 8-fold and 22-fold higher than that in the roots respectively, while in cultivar "Rixiang" they were only 2.8-fold and 11-fold, respectively. In both cultivar "Boye" and "Rixiang", OfDXR showed over 15-fold transcript levels in the inflorescences than in the roots. As for OfDXS2, the inflorescences were found to contain the highest transcript level. In cultivar "Boye", the OfDXS2 transcript level in the inflorescences was virtually 427-fold higher than that in the roots, whereas in cultivar "Rixiang" it was merely 224-fold. For cultivar "Boye" and "Rixiang", OfMCT, OfMDS, OfHDS, and OfHDR1 showed higher transcript levels in the leaves and the inflorescences than in the roots and the stems. In addition, the OfMCT transcript levels were paralleled in the leaves and the inflorescences. In cultivar "Boye", OfHDS and OfHDR1 showed higher transcript levels in the inflorescences than in the rest of the organs. Whereas in cultivar "Rixiang", OfMDS and OfHDR1 showed higher transcript levels in the leaves. Furthermore, OfIDI showed slightly different transcript profiles among the four organs, which were consistent in the two cultivars. The OfIDI transcript level in the petals was 1.5-fold higher than that in the roots, and almost 3-fold higher than that in the stems or the leaves.

Expression Analysis of MEP Genes Over Flower Development
To determine the expression patterns during flower development, qRT-PCR were conducted to detect the transcript levels of MEP genes at four flowering stages, including bud-eye stage (S1), primary blooming stage (S2), full blooming stage (S3), and flower fading stage (S4) (Figure 4).
The experimental results showed the MEP genes were all detected at four flowering stages, however their transcript patterns varied from each other. For OfDXS1, the transcripts in "Boye" and "Rixiang" both displayed downregulated trends coincidently during the first three stages. The OfDXS2 transcript levels showed entirely opposite trends in the two cultivars: in "Boye" high transcript level was maintained in the first three stages, whereas in "Rixiang", it first increased steadily from S1 to S3 stage, and then declined sharply at S4 stage. The OfDXR transcript levels declined constantly in the two cultivars from S1 to S4 stage. For OfMCT and OfCMK, their transcript levels in "Boye" showed a regularly downregulated trend at the four stages. However, their transcripts in "Rixiang" remained at a high level from S1 to S3 stage, and then decreased considerably from S3 to S4 stage. For OfMDS and OfHDS, their transcript levels in "Boye" did not show a significant change at the four stages. However, their transcript levels in "Rixiang" showed a slight rise from S1 to S2 stage, and then declined from the S3 to S4 stage. For OfHDR1 and OfHDR2, their transcript levels in "Rixiang" showed a similar profile with that of OfMDS and OfHDS, while their transcript levels in "Boye" showed a reverse trend from S1 to S2 stage. For OfIDI, the transcripts in "Boye" and "Rixiang" maintained almost identical levels during the first three stages, but ascended to 1.25-fold and 2.15-fold from S3 to S4 stage, respectively. inflorescences was virtually 427-fold higher than that in the roots, whereas in cultivar "Rixiang" it was merely 224-fold. For cultivar "Boye" and "Rixiang", OfMCT, OfMDS, OfHDS, and OfHDR1 showed higher transcript levels in the leaves and the inflorescences than in the roots and the stems. In addition, the OfMCT transcript levels were paralleled in the leaves and the inflorescences. In cultivar "Boye", OfHDS and OfHDR1 showed higher transcript levels in the inflorescences than in the rest of the organs. Whereas in cultivar "Rixiang", OfMDS and OfHDR1 showed higher transcript levels in the leaves. Furthermore, OfIDI showed slightly different transcript profiles among the four organs, which were consistent in the two cultivars. The OfIDI transcript level in the petals was 1.5fold higher than that in the roots, and almost 3-fold higher than that in the stems or the leaves.

Expression Analysis of MEP Genes Over Flower Development
To determine the expression patterns during flower development, qRT-PCR were conducted to detect the transcript levels of MEP genes at four flowering stages, including bud-eye stage (S1), primary blooming stage (S2), full blooming stage (S3), and flower fading stage (S4).
The experimental results showed the MEP genes were all detected at four flowering stages, however their transcript patterns varied from each other. For OfDXS1, the transcripts in "Boye" and "Rixiang" both displayed downregulated trends coincidently during the first three stages. The OfDXS2 transcript levels showed entirely opposite trends in the two cultivars: in "Boye" high transcript level was maintained in the first three stages, whereas in "Rixiang", it first increased steadily from S1 to S3 stage, and then declined sharply at S4 stage. The OfDXR transcript levels declined constantly in the two cultivars from S1 to S4 stage. For OfMCT and OfCMK, their transcript levels in "Boye" showed a regularly downregulated trend at the four stages. However, their transcripts in "Rixiang" remained at a high level from S1 to S3 stage, and then decreased considerably from S3 to S4 stage. For OfMDS and OfHDS, their transcript levels in "Boye" did not show a significant change at the four stages. However, their transcript levels in "Rixiang" showed a slight rise from S1 to S2 stage, and then declined from the S3 to S4 stage. For OfHDR1 and OfHDR2, their transcript levels in "Rixiang" showed a similar profile with that of OfMDS and OfHDS, while their transcript levels in "Boye" showed a reverse trend from S1 to S2 stage. For OfIDI, the transcripts in "Boye" and "Rixiang" maintained almost identical levels during the first three stages, but ascended to 1.25-fold and 2.15-fold from S3 to S4 stage, respectively (Figure 4). (e) OfCMK (f) OfMDS

Expression analysis of MEP Genes during Diel Oscillations
To further study the expression patterns during diel oscillations from sweet osmanthus flowers over time, we chose 12 sampling time points with two-hour intervals in full blooming stage for daily analysis. Using qRT-PCR, the transcript levels of the 10 MEP genes were detected in the two cultivars.
Each of the MEP genes showed a particular oscillating pattern during the daytime and night cycles. In cultivar "Boye" and "Rixiang", the OfDXS1 transcript levels both exhibited a clear peak in the morning. The OfDXS2 transcript levels showed a typical diurnal oscillation both in cultivars "Boye" and "Rixiang", which escalated in the morning, reached the peak at midday, then deescalated in the afternoon. For OfDXR, the transcript level in "Boye" showed a slight peak at 06:00 h, whereas peak transcript of "Rixiang" appeared to a later time point (10:00 h). The OfMCT transcript in "Boye" achieved the highest level at 06:00 h, while in "Rixiang" slight oscillations occurred during the whole day. Compared with other time points, the OfCMK transcript in "Boye" maintained higher level from 12:00 to 22:00 h. Yet in "Rixiang", the OfCMK transcript level decreased after 14:00 h. For OfMDS, the transcript level in "Boye" oscillated steadily during the whole day, whereas the transcript level in "Rixiang" revealed a significant peak in the morning with 1.73-fold higher than the corresponding predawn level. In "Boye", the OfHDS transcript peaked to 1.5-fold higher levels at 04:00, 14:00, and 20:00 h, while significant oscillation was undetected in "Rixiang". For OfHDR1, the transcript levels in "Boye" and "Rixiang" firstly crested at 08:00 h, thereafter reaching another peak at 14:00 and 16:00 h, respectively. While the OfHDR2 transcript levels showed a slight peak between 06:00 and 08:00 h in the two cultivars. As for OfIDI, the transcript levels showed a gradual decline in the daytime.

Expression analysis of MEP Genes during Diel Oscillations
To further study the expression patterns during diel oscillations from sweet osmanthus flowers over time, we chose 12 sampling time points with two-hour intervals in full blooming stage for daily analysis. Using qRT-PCR, the transcript levels of the 10 MEP genes were detected in the two cultivars ( Figure 5).
Each of the MEP genes showed a particular oscillating pattern during the daytime and night cycles. In cultivar "Boye" and "Rixiang", the OfDXS1 transcript levels both exhibited a clear peak in the morning. The OfDXS2 transcript levels showed a typical diurnal oscillation both in cultivars "Boye" and "Rixiang", which escalated in the morning, reached the peak at midday, then de-escalated in the afternoon. For OfDXR, the transcript level in "Boye" showed a slight peak at 06:00 h, whereas peak transcript of "Rixiang" appeared to a later time point (10:00 h). The OfMCT transcript in "Boye" achieved the highest level at 06:00 h, while in "Rixiang" slight oscillations occurred during the whole day. Compared with other time points, the OfCMK transcript in "Boye" maintained higher level from 12:00 to 22:00 h. Yet in "Rixiang", the OfCMK transcript level decreased after 14:00 h. For OfMDS, the transcript level in "Boye" oscillated steadily during the whole day, whereas the transcript level in "Rixiang" revealed a significant peak in the morning with 1.73-fold higher than the corresponding predawn level. In "Boye", the OfHDS transcript peaked to 1.5-fold higher levels at 04:00, 14:00, and 20:00 h, while significant oscillation was undetected in "Rixiang". For OfHDR1, the transcript levels in "Boye" and "Rixiang" firstly crested at 08:00 h, thereafter reaching another peak at 14:00 and 16:00 h, respectively. While the OfHDR2 transcript levels showed a slight peak between 06:00 and 08:00 h in the two cultivars. As for OfIDI, the transcript levels showed a gradual decline in the daytime.

Discussion
The MEP pathway genes have been isolated and identified in a number of plant species, including Arabidopsis [26,27], peppermint [30,31], tomato [32,33], and rice [24]. However, this pathway has not yet been studied in sweet osmanthus. In this study, gene cloning allows the analysis

Discussion
The MEP pathway genes have been isolated and identified in a number of plant species, including Arabidopsis [26,27], peppermint [30,31], tomato [32,33], and rice [24]. However, this pathway has not yet been studied in sweet osmanthus. In this study, gene cloning allows the analysis of the MEP pathway genes sequences in sweet osmanthus and the results will facilitate further research on gene function and molecular evolution.

The MEP Pathway Genes of O. fragrans Are Highly Related to Those from Other Plants
The MEP pathway contains eight enzymatic steps and previous research has shown that terpenoids biosynthesis is regulated by a series of structural and functional genes. [24]. DXS, the first committed enzyme in the MEP pathway, influencing the accumulation of downstream isoprenoids, is encoded by a small multigene family in higher plants. In this study, we successfully isolated two OfDXS genes from sweet osmanthus. In the second enzymatic step of the MEP pathway, DXR also has a rate-limiting effect on the accumulation of MEP-derived isoprenoids [41,42]. Furthermore, the biosynthesis of MEP limits the production of downstream isoprenoids in Arabidopsis [43]. In the third step of the MEP pathway, AtMCT contain a plastid targeting sequence in Arabidopsis [44]. The CMK genes, which contain putative ATP binding sites and plastid target sequences, were also cloned from tomato and peppermint [45]. The GbMDS from ginkgo biloba is well conserved in the protein family and highly similar (over 70% identity) to other plants [46]. Although less is known about HDS and HDR than other genes in the MEP pathway, it has been shown that HDR is encoded by muticopy genes in plants, and we obtained two OfHDR genes from sweet osmanthus [35]. The last step in the MEP pathway is an isomerization reaction, which is catalyzed by IDI, and is also a rate-limiting step during isoprenoid synthesis [47].
In this study, through the efforts of transcriptome mining and RACE, we successfully obtained 10 full-length MEP pathway cDNAs from sweet osmanthus. These included OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfMDS, OfHDS, OfHDR1, OfHDR2, and OfIDI. Their deduced protein sequences were all highly similar to those of other plants. Interestingly, the sequence alignments of OfDXS1, OfDXS2, OfDXR, OfMCT, OfCMK, OfHDS, OfHDR1, and OfHDR2 showed 92%, 87%, 91%, 80%, 82%, 92%, 86%, and 88% identity with reported corresponding proteins from Sesamum indicum respectively, which suggested that there was an evolutionary conserved relationship between sweet osmanthus and S. indicum. However, OfMDS shared a high similarity with Salvia miltiorrhiza. By bioinformatics analysis, we found that the MEP genes were conserved over their protein sequences, but their detailed evolutionary relationships need further investigation.

Expression Patterns of the MEP Pathway Genes Suggest that Enhanced Biosynthesis of Substrate Contributes to the Production of Monoterpenes in O. fragrans Flowers
The biosynthesis of terpenoids can be regulated at, at least, two levels: the level of terpene synthases and the level of the substrates of TPSs. While a previous study has shown the importance of the regulation of TPS genes [16], the present study indicates the importance of the regulation of the substrate biosynthesis.
The transcript results showed that the MEP genes were all highly expressed in the inflorescences compared with other organs. This suggested that there was a tissue-specific expression profile among these genes, which led to the biosynthesis of specific downstream isoprenoid-derived products [24]. Similarly, it has been reported that terpene synthase (TPS) genes involved in volatile terpenoid synthesis have been cloned and shown to be flower-specific in Clarkia breweri [48], Antirrhinum majus [49], and A. thaliana [50]. OfDXS2, which is involved in the first step of the MEP pathway, showed a clear flower-specific transcript profile and its transcript level was several hundred-fold higher than in the roots. It has been suggested that the floral and vegetative tissues are the main scent sources in many plants [51]. Therefore, the enormous expression might lead to the specific accumulation of monoterpenes and sesquiterpenes in flowers [52].
Previous studies have shown that the pigment and essential oil compositions vary in O. fragrans floral developmental process [9,53]. Furthermore, previous research has suggested that the aromatic compounds and relative contents differ among O. fragrans cultivar groups, including cultivars "Boye" and "Rixiang" [54]. In this study, the expression profiles of these MEP pathway genes were investigated at different developmental stages by qRT-PCR. Notably, OfDXS1 and OfDXS2 remained continuous high expression during the anthesis, as well as OfMDS, OfHDS, and OfHDR1. These results were consistent with the expression level of OfDXS in "Yanhong Gui", where there was a substantial accumulation of αand β-carotene [1]. Moreover, carotenoids are considered to be the crucial substrates for the OfCCD1 enzyme, which produces αand β-ionone aroma compounds in flowers of O. fragrans [8]. In contrast, the OfDXR expression level was down-regulated dramatically from S1 to S4 stage. Interestingly, the OfMCT and OfCMK expression patterns in "Boye" and "Rixiang" were not the same. These variations in expression might be caused by cultivar differences.

Expression Patterns of MEP Pathway Genes during Diel Oscillations
It has been observed that the MEP pathway genes fluctuated rhythmically over a daily light/dark cycle [55]. In A. thaliana, the expression of MEP pathway genes was reported to be controlled by light [56]. However, in snapdragon flowers, MEP gene expressions follows a diurnal rhythm, which is regulated by an endogenous circadian clock [57]. Recent study has shown that in many plants, scent emission can be regulated either by circadian clock or by light, mostly by the gene expression levels [58].
Previous research has suggested that the flower volatile emissions of sweet osmanthus follow a diurnal pattern, and sweet osmanthus flowers release the highest total amount of volatiles at 10:00 h in the daytime [2]. In this study, we monitored the transcript levels of the MEP pathway genes every 2 h for 24 h at the O. fragrans full blooming stage. The OfDXS2 expression level, not OfDXR, showed a diurnal oscillation profile that increased under light conditions with the highest accumulation occurring between 12:00 and 14:00 h. Similarly, the transcript levels of OfDXS1 and OfHDR1 showed considerable intraday variations with a clear peak in the morning. Apart from OfDXS1, OfDXS2, and OfHDR1, the rest of the MEP pathway genes lacked obvious diurnal oscillation patterns. Although previous mathematical modelling data suggested that flux through the MEP pathway is due to the photosynthesis-dependent supply of metabolic substrates, additional experimental work is needed to clarify the contribution of enzymatic substrate biosynthesis to diurnal patterns of volatile emission [59].

Conclusions
In this study, the genes in the MEP pathway in O. fragrans were cloned, compared with those of other plants, and analyzed for their expression patterns. The sequence alignment analysis revealed that the MEP pathway genes of O. fragrans had high sequence identities with other reported proteins, which suggests that an evolutionarily conserved relationship exists. The qRT-PCR results showed that many MEP pathway genes had a higher-level of expression in the inflorescences, supporting that the enhanced production of the prenyl diphosphates-the substrates of terpene synthases-contributes to the biosynthesis of monoterperne floral volatiles. In addition, the expressions of several genes, such as OfDXS1, OfDXS2, and OfHDR1, exhibited a diurnal oscillation pattern. Our results lay an important foundation for future research on functional and molecular evolutionary analysis of the terpene pathway genes involved in the production of terpene floral volatiles in O. fragrans.
Author Contributions: Chen Xu, Huogeng Li and Lianggui Wang conceived and designed the experiments; Chen Xu and Hongna Mu performed the experiments; Chen Xu, Xiulian Yang and Chunsun Gu analyzed the data; Huogeng Li and Lianggui Wang contributed reagents/materials/analysis tools; Chen Xu and Yuanzheng Yue wrote the paper.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript: