Introduction

Geranylgeranyl diphosphate (GGPP) is an important metabolic intermediate in plants that serves as a precursor for the biosynthesis of tocopherols, gibberellins, carotenoids, chlorophylls, and other diterpenoids (Okada et al. 2000; Lange and Ghassemian 2003). GGPP is usually synthesized via the head-to-tail condensation of three isoprenyl diphosphate (IPP) moieties to the allyl head of dimethylallyl diphosphate (DMAPP), as catalyzed by GGPP synthase (GGPS) (Scolnik and Bartley 1994). Some GGPSs can alternatively use geranyl diphosphate (GPP) or farnesyl diphosphate (FPP) as a substrate to produce GGPP (Hefner et al. 1998). The first GGPS genes were identified in the bacteria Pantoea stewartii, Pantoea ananatis (previously known as Erwinia herbicola and E. uredovora, respectively), and Rhodobacter capsulatus, and were designated as crtE genes (Armstrong et al. 1990a, b; Misawa et al. 1990; Mergaert et al. 1993). Subsequently, GGPS homologs have been identified in various photosynthetic and non-photosynthetic carotenoid-accumulating organisms, including bacteria, fungi, red yeasts, and higher plants.

The GGPS gene family is the largest and most diverse group of isoprenyl synthase genes (van Schie et al. 2013). In higher plants, it usually comprises genes involved in the biosynthesis of GGPP and other isoprenoids. There are 12 GGPS homologs in the genome of Arabidopsis thaliana, but only five (GGPS1, GGPS3, GGPS4, GGPS7, and GGPS11) have been functionally characterized as producers of GGPP (Okada et al. 2000). Although carotenoid biosynthesis exclusively occurs in the plastid in higher plants, the subcellular localization of GGPSs has revealed that they are also present in other compartments, such as mitochondria (Okada et al. 2000; Beck et al. 2013). It has been proposed that plants use different GGPSs to supply GGPP for different metabolic branches beyond this intermediate (Okada et al. 2000; van Schie et al. 2013).

Although the carotenoid metabolic pathway has been well studied in higher plants (Ruiz-sola and Rodríguez-Concepción 2012), it remains poorly understood in red algae. To the best of our knowledge, only two red algal enzymes involved in carotenoid metabolism, a lycopene cyclase of Cyanidioschyzon merolae (Cunningham et al. 2007) and a β-carotene hydroxylase (PuCHY1) of Pyropia umbilicalis (previously as Porphyra umbilicalis) (Yang et al. 2014), have been functionally characterized. Here, we cloned the gene encoding GGPS in P. umbilicalis (PuGGPS) and determined that it functions in the biosynthesis of GGPP for carotenoid biosynthesis.

Materials and methods

Leafy thalli of Pyropia umbilicalis strain P.umb.1 (deposited as UTEX #LB-2591 CCAP 1379/4), which is the same strain previously used for transcriptome sequencing (Chan et al. 2012), were obtained from Dr. Susan Brawley (University of Maine). P. umbilicalis thalli were cultivated in Provasoli’s enriched seawater (Starr and Zeikus 1993) under previously described conditions (Yang et al. 2013). The materials were rinsed briefly with distilled water, surface dried with paper towels, ground into a fine powder with a mortar and pestle in liquid nitrogen and either stored at −80 °C or used promptly. Leafy thalli used for the quantification of gene expression were treated with 5 μM norflurazon (Sigma, USA) for 2 and 4 h, and untreated samples were used as controls. All materials were cultivated under the same conditions.

Transcriptome analysis and homolog identification

A. thaliana is one of the eukaryotic organisms in which the carotenoid metabolic pathway has been well studied (Ruiz-sola and Rodríguez-Concepción 2012). The open reading frames (ORFs) of 12 genes encoding GGPSs of A. thaliana were used as queries to blast against the transcriptome database (http://dbdata.rutgers.edu/nori) of P. umbilicalis (Chan et al. 2012). Two GGPS genes identified from the cyanobacterium Synechococcus sp. PCC 6301 and the unicellular red alga Cyanidioschyzon merolae were also incorporated as queries. We used the tblastx algorithm with an e-value cutoff of 1 × 10−10 to screen for putative homologs, as previously reported (Yang et al. 2014). The queries are listed in Table 1.

Table 1 Queries of the geranylgeranyl diphosphate synthase (GGPS) genes from Arabidopsis thaliana and other organisms used to blast against the Pyropia umbilicalis transcriptome database
Full size table

Cloning of PuGGPS

To obtain the coding sequence of a transcript (P_umbilicalis_esContig5139) that resembles the known GGPS genes, total RNA from P. umbilicalis was isolated and reverse transcribed as previously described (Yang et al. 2013). Based on the sequence of the putative homologous transcript, a pair of primers, GGPS-HF and GGPS-ER, was designed to amplify the coding region from our complementary DNA (cDNA) pool. Genomic DNA was also extracted from P. umbilicalis as previously described (Yang et al. 2013). All primers used in this study are listed in Table 2. All amplicons were subcloned into a pMD19-T vector (Takara, Japan) and sequenced. The full-length cDNA sequence was deposited in GenBank under the accession number KP863500, and the gene was named PuGGPS. High-fidelity PrimeStar DNA polymerase (Takara) was used throughout this study unless otherwise indicated, and dimethylsulfoxide (DMSO) was added to the polymerase chain reaction (PCR) system at a final concentration of 5 % (v/v) to overcome the high GC contents of the templates (Sun et al. 2010).

Table 2 Primers used in the present study
Full size table

Heterologous expression and functional characterization of PuGGPS

The ORF of PuGGPS was amplified using the primers GGPS-SacI-HF and GGPS-HindIII-ER to incorporate restriction sites. The amplicon was purified, digested with Sac I-Hind III, cloned into pET-32b (Merck Millipore, Germany), and sequenced for confirmation. The construct was designated as pET-PuGGPS. To test the enzymatic activity of PuGGPS in vivo, pET-PuGGPS was co-transformed into E. coli BL21(DE3) (Merck Millipore) with pAC-94N, which harbors all the genes needed for the biosynthesis of β-carotene in bacteria (crtB, crtI and crtY of Pantoea stewartii) except for GGPS (crtE) (Cunningham and Gantt 2007). The vector pET-NtGGPS1, which was constructed in our lab and harbors an active GGPS gene of Nicotiana tabacum (GGPS1, KC484701.1) (Orlova et al. 2009), was co-transformed with pAC-94N into E. coli BL21(DE3) as a positive control. An empty pET-32b vector was co-transformed with pAC-94N as a negative control. Transformant colonies accumulating β-carotene were selected from Luria broth (LB) plates and then inoculated in LB medium as previously reported (Yang et al. 2014). Isopropyl β-D-1-thiogalactopyranoside (IPTG) (0.5 mM) was used to induce the expression of PuGGPS for 3 h at 250 rpm and 28 °C. A 200-μL aliquot of the culture was added to 20 mL of fresh medium containing 0.5 mM IPTG and inoculated under the same conditions. After inoculation, cells in the 20 mL culture were collected by centrifugation at 5000×g for 5 min. The pelleted cells were sequentially mixed with 400 μL of 80 % acetone, 250 μL of ethyl acetate, and 250 μL of distilled water for 15 s, with high-speed vortexing after each step. After centrifugation at 10,000×g for 5 min at 4 °C, the organic phase was transferred into a new tube and dried under a stream of nitrogen gas. The dried extract was re-dissolved in 100 μL of ethyl acetate and either analyzed immediately or stored at −80 °C. A Waters 2695 separation module and 2998 photodiode array detector (PDA) (Waters, USA) were used for high-performance liquid chromatography (HPLC) analysis of carotenoids on a Spherisorb ODS2 column (5 μm, 4.6 × 250 mm) (Waters) using a 45-min gradient of ethyl acetate (0–100 %) in acetonitrile-water-triethylamine (9:1:0.01, v/v) at a flow rate of 1 mL min−1 and monitored at 440 nm (Norris et al. 1995). At least five replicates were performed for each sample.

Homology modeling

The protein sequence of PuGGPS was imported into Accelrys Discovery Studio Client (Accelrys, USA) with its transit peptide sequence removed. The truncated sequence was aligned with proteins with available crystal structures in the Protein Data Bank (PDB) using the blast search algorithm. The crystal structures of five proteins with high sequence identities, including chain A of GGPS of Sinapis alba (PDB ID: 2J1P), GGPS of Lactobacillus brevis (3M9U), GPS of Mentha piperita (3KRA), and FPSs of Marinomonas sp. MED 121 (4F62) and Staphylococcus aureus (1RTR), were used as templates for homology modeling (Levit et al. 2012).

Gene expression profile

To discern whether PuGGPS is involved in carotenoid biosynthesis in P. umbilicalis, the expression levels of transcripts encoding phytoene desaturase (PuPDS) (GenBank accession number KP863501) and PuGGPS were relatively quantified after leafy thalli were treated with the carotenoid biosynthesis inhibitor norflurazon for 2 and 4 h. RNA was isolated from norflurazon-treated and control materials using RNAiso Plus and reverse transcribed using a PrimeScript II First Strand cDNA Synthesis Kit (Takara) as described previously (Yang et al. 2013). Quantitative PCR was performed with a StepOne Plus Real-Time PCR System (Applied Biosystems, USA) using a SYBR Premix Ex Taq (Tli RNaseH Plus) Kit (Takara). A P. umbilicalis homolog (PuACT2) of the Pyropia yezoensis Actin2 gene was used as a reference (Kitade et al. 2008). The reaction system and thermal profile were described previously (Yang et al. 2014). The primer pairs used for the amplification of fragments of PuPDS, PuGGPS, and PuACT2 are listed in Table 1 (each with a -QF or -QR suffix after the name of the gene to indicate the forward or reverse orientation, respectively). For each treatment, PCR was performed in triplicate for each triplicate sample. The expression levels were calculated according to the 2−ΔCT method (Sun et al. 2010).

Phylogenetic analysis

We searched GenBank for GGPS homologs of different types according to previous reports (Wang and Ohnuma 2000; Vandermotern et al. 2009; van Schie et al. 2013). Sequences were aligned using the ClustalX program (Chenna et al. 2003). A phylogenetic tree was constructed using the neighbor-joining (NJ) method in MEGA 5 with 1000 bootstrap replicates (Tamura et al. 2011).

Results

We used the ORFs of the 12 A. thaliana GGPS genes as queries to blast against the transcriptome of P. umbilicalis, and all queries returned only a single transcript species: P_umbilicalis_esContig5139. When we used the GGPS genes from Cyanidioschyzon merolae and Synechococcus sp. PCC 6301 as queries, the same transcript species was identified. We also used these 14 queries to blast against the transcriptome of Porphyra purpurea, another Bangiales red algal species, and similarly found only one transcript species (data not shown).

Sequence analysis of the GGPS homolog transcript revealed a translation initiation codon (ATG) at the 259th nucleotide. Using a primer pair designed according to the transcript sequence, both cDNA and genomic DNA sequences corresponding to this putative gene were amplified, cloned, and sequenced. ORF of this putative gene was found to be 1038 bp, encoding a protein of 345 amino acid residues (data not shown), of which the relative molecular weight was approximately 35 kDa. No introns were found when comparing the cDNA and genomic DNA sequences of PuGGPS. The sequences sharing the highest sequence identities with PuGGPS were GGPSs from Thalassiosira pseudonana (65 %), Synechococcus sp. JA-3-3Ab (58 %), and Cyanidioschyzon merolae (57 %). A transit peptide that targets this protein to the plastid was predicted at the N-terminus (M1-D65) by the online server TargetP (Emanuelsson et al. 2000).

Because previous studies have shown that GGPS homologs might not produce GGPP alone and that they are involved in different metabolic pathways, we assessed the function of this PuGGPS by constructing a pET-PuGGPS expression vector, transforming it into E. coli BL21(DE3) and inducing the expression of a fusion protein with IPTG. The expressed PuGGPS with a His tag had a molecular weight of 55.3 kDa (data not shown), indicating that this construct could be used for a color complementation assay. This construct was then co-transformed into E. coli BL21(DE3) cells with pAC-94N, which harbors the full set of genes needed to synthesize β-carotene when GGPP is supplied. When we cultivated the transformed bacteria in LB medium and induced gene expression with IPTG for 3 h, the E. coli cells co-transformed with pET-PuGGPS and pAC-94N changed from white to orange yellow, indicating β-carotene accumulation and demonstrating that the expressed PuGGPS was able to produce GGPP for the subsequent metabolic processes driven by the enzymes expressed by pAC-94N. Bacterial cells transformed with both pET-NtGGPS1, which expresses a functionally confirmed tobacco GGPS, and pAC-94N also accumulated β-carotene, as expected. Bacterial cells harboring a pET-32b empty vector and pAC-94N did not turn orange yellow (Fig. 1).

Fig. 1
figure 1

Color complementation results for PuGGPS in Escherichia coli. BL21(DE3) cells harboring different combinations of plasmids were cultivated in LB medium and induced with IPTG for 3 h before the cells were pelleted to show the accumulation of β-carotene. −, pET-32b + pAC-94N as the negative control; PuGGPS, pET-PuGGPS + pAC-94N; +, pET-NtGGPS1 + pAC-94N as the positive control

Full size image

The pigments in the cultivated E. coli cells were extracted and separated by HPLC using previously reported methods (Yang et al. 2014). Our results showed there was no carotenoid accumulation in the negative control harboring pET-32b and pAC-94N (Fig. 2a). However, cells harboring pAC-94N with either pET-PuGGPS or pET-NtGGPS1 accumulated β-carotene together with a low amount of lycopene (Fig. 2b, c). This result indicates that PuGGPS is capable of synthesizing GGPP in vivo for the biosynthesis of carotenoids, such as NtGGPS1.

Fig. 2
figure 2

HPLC analysis of pigment accumulation in E. coli cells harboring different combinations of plasmids after IPTG induction. a pET-32b + pAC-94N as the negative control, b pET-PuGGPS + pAC-94N, and c pET-NtGGPS1 + pAC-94N as the positive control. The peaks are lycopene (1) and β-carotene (2)

Full size image

The herbicide norflurazon is an inhibitor of phytoene desaturase (PDS), which is a key enzyme in the biosynthesis of carotenoids (Chamovitz et al. 1993). To determine whether PuGGPS is functionally involved in carotenoid biosynthesis in planta, we quantified the expression of the PuGGPS and PuPDS genes following the perturbation of carotenoid metabolism by this inhibitor. Our results showed that after P. umbilicalis leafy thalli were treated with norflurazon for 2 h, the expression of PuPDS was slightly increased and that of PuGGPS was increased by approximately twofold compared with the control levels. Following treatment with norflurazon for 4 h, the expression of PuPDS was increased by twofold, whereas that of PuGGPS remained at the same level observed at 2 h (Fig. 3).

Fig. 3
figure 3

Quantitative real-time PCR determination of the transcript abundances of PuGGPS and PuPDS in leafy thalli that were treated with norflurazon for 2 and 4 h. Untreated thalli were used as control. Relative expression was calculated as the ratio between the expression levels of the gene studied and the reference gene PuAct2. The data represent the mean ± SE (n = 6)

Full size image

In higher plants, GGPSs are encoded by multiple members of a gene family. However, in this primitive red alga, we could not identify other transcript homologs sharing reasonable sequence similarity to suggest a possible GGPS function. For a better understanding of PuGGPS, we blasted its sequence against the non-redundant protein database using the blastp algorithm in NCBI. Most of the hits were GGPSs from different organisms. We aligned the sequences of GGPS11, GGPS4, and GGPS6 from A. thaliana, which are known to be localized to different organelles, and that of the GGPS from Synechococcus sp. PCC 6301 with the PuGGPS sequence. The multiple alignment showed that PuGGPS shares seven typical conserved motifs with other GGPSs from both cyanobacteria and higher plants, including the first aspartate-rich motif (FARM) and the second aspartate-rich motif (SARM), which are signature sequences for GGPSs (Fig. 4).

Fig. 4
figure 4

Alignment of the sequences of GGPS homologs from Pyropia umbilicalis (PuGGPS), Arabidopsis thaliana (AtGGPS), and Synechococcus sp. PCC 6301 (SeGGPS). Conserved motifs are underlined, and the first and second aspartate-rich motifs (FARM and SARM, respectively) are labeled

Full size image

GGPSs are categorized into three different types (I, II, and III) based on the sequences of the chain-length determination (CLD) regions, including the FARM and several of its upstream amino acid residues (Hemmi et al. 2003). PuGGPS is a typeType II GGPS, of which the FARM is DXXXXD (X is any amino acid), instead of DDXXD in the FARM of the other two types (Hemmi et al. 2003). A phylogenetic tree based on GGPS protein sequences of different types (Wang and Ohnuma 2000; Vandermotern et al. 2009; van Schie et al. 2013) indicated that PuGPPS was closely clustered with GGPSs from the cyanobacterium Synechocystis sp. PCC 6803, the unicellular red alga Cyanidioschyzon merolae, and the diatom Phaeodactylum tricornutum (Fig. 5). GGPSs from the green algae Chlamydomonas reinhardtii and Volvox carteri were more closely clustered with those from higher plants (Fig. 5).

Fig. 5
figure 5

Phylogenetic analysis of PuGGPS and its homologs from Arabidopsis thaliana (At), Oryza sativa (Os), Nicotiana tabacum (Nt), Physcomitrella patens (Pp), Selaginella moellendorffii (Sm), Chlamydomonas reinhardtii (Cr), Volvox carteri (Vc), Synechocystis sp. PCC 6803 (Sy), Cyanidioschyzon merolae (Cm), Phaeodactylum tricornutum (Pt), Pantoea dispersa (Pd), Homo sapiens (Hs), and Sulfolobus acidocaldarius (Sa). A bootstrap (1000 replicates) neighbor-joining unrooted phylogenetic tree was generated using MEGA 5. The bootstrap values are labeled beside the branches, and the scale bar indicates 10 % sequence divergence

Full size image

To ascertain whether the tertiary structure of PuGGPS resembles that of GGPSs for which crystal structures have been elucidated, the tertiary structure of PuGGPS was constructed using homology modeling (Levit et al. 2012). The results showed that the constructed three-dimensional model of PuGGPS resembled the crystal structures of GGPSs from Sinapis alba (PDB ID: 2J1P) and Lactobacillus brevis (3M9U) (Fig. 6a). There were 11 α-helices with no β-sheets in the constructed model, which is a structure typical of prenyl diphosphate synthase called the “terpenoid synthase fold.” Several of the α-helices form the active cavity, and the product GGPP can precisely dock into this cavity (Fig. 6b). The FARM and SARM signatures are located on the two sides of the cleft to facilitate the chelation of Mg2+, which is required for the activity of GGPS. The reliability of the constructed model was tested using a Ramachandran plot, and we found most of the intersection angles between the amino acids to be reasonable, with a few exceptions. However, abnormal intersection angles were noted between amino acids in the loops or at the corners, which may be caused by the attraction of amino acids during normal folding (Fig. 6c, d).

Fig. 6
figure 6

Homology modeling of PuGGPS. a 3D structures of PuGGPS (red) and homologs from different organisms, showing their similarities. b Docking of GGPP (yellow) to PuGGPS and the location of Mg2+ (green). c Ramachandran analysis of the 3D structure of PuGGPS. The red spots show abnormal intersection angles, and the green spots show normal angles. d Locations of the amino acids with abnormal intersection angle (indicated with the stick ball)

Full size image

Discussion

GGPP, synthesized by GGPS, is a precursor for the biosynthesis of many biologically important metabolites, including carotenoids and chlorophylls. The gene that we cloned and identified from the transcriptome of P. umbilicalis encodes a protein of 345 amino acids with an estimated molecular weight of 35 kDa, which is consistent with previously reported GGPSs (Engprasert et al. 2004; Thabet et al. 2012). This protein has high sequence similarity with GGPSs from cyanobacteria, unicellular red algae, and diatoms, supporting an early evolution of carotenoid metabolism from cyanobacteria to heterokonts (Yang et al. 2014). Furthermore, PuGGPS shares seven conserved motifs, including the FARM and SARM signatures (Fig. 4), and has a three-dimensional structure that is very similar to those of other GGPSs (Fig. 6a). Phylogenetic analysis also showed that it belongs to the type II GGPSs of eubacteria and plants (Fig. 5) (Wang and Ohnuma 2000; Vandermotern et al. 2009). These results indicate that PuGGPS is a member of the GGPS protein family. This function was further supported by a color complementation assay, HPLC analysis, and three-dimensional modeling (Figs. 1, 2, and 6b).

The application of norflurazon perturbs metabolic flux because of its competition with a cofactor (plastoquinone) required by PDS (Norris et al. 1995; Breitenbach et al. 2001). In our studies, the treatment of P. umbilicalis leafy thalli with norflurazon resulted in the prompt induction of PuGGPS expression (Fig. 3), suggesting that PuGGPS is likely to be involved in the biosynthesis of carotenoids and that it accumulates in response to the decreased catalytic activity of PuPDS. However, the transcript abundance of PuGGPS did not increase with prolonged norflurazon treatment (Fig. 3). One possible reason for this finding is that the norflurazon supplied in the medium was not sufficient to compete with plastoquinone for PuPDS. The decreased catalytic activity of PuPDS was then compensated for by the induced expression of PuPDS and the accumulation of the enzyme; thus, it was unnecessary to express PuGGPS at a higher level. This agrees with a previous report in higher plants that the promoter activity of PDS is activated by end-product regulation (Corona et al. 1996). Additionally, because the transcriptome of P. umbilicalis has been sequenced under different conditions and is thought to include a near-complete repertoire of genes (Chan et al. 2012), PuGGPS is most likely the only enzyme that supplies GGPP for the biosynthesis of not only carotenoids but also chlorophylls and other diterpenoid derivatives; thus, its expression is not regulated solely by the carotenoid metabolic pathway.

Previous studies have shown that carotenoid metabolism in red algae is not as complicated as that in higher plants (Schubert et al. 2006; Cunningham et al. 2007). Epoxycarotenoids, such as antheraxanthin and violaxanthin, have not been identified in either P. umbilicalis or other related species (Gantt et al. 2010; Schubert et al. 2006; Yang et al. 2014). Therefore, carotenoids in P. umbilicalis are unlikely to be utilized in a xanthophyll cycle to confer photoprotection (Yang et al. 2014). Both cyanobacteria and red algae lack transmembrane light harvesting complex II (LHCII), and it has been reported that carotenoid-binding protein is involved in energy quenching in cyanobacteria (Kirilovsky 2007; Rakhimberdieva et al. 2007). The functions of carotenoids in primitive red algal species remain unclear. The detailed characterization of the entire metabolic pathway under different environmental conditions should shed light on the regulatory mechanisms governing different metabolic branches beyond GGPP and also on the early evolution of carotenoid metabolism.