S. pararoseus NGR was isolated from strawberry fruit in the green house of Shenyang Agricultural University (41°49′N, 123°34′E). The strain number is recorded in the China General Microbiological Culture Collection Center as CGMCC 2.5280. S. pararoseus NGR was incubated in YPD medium (10 g/L yeast extract, 20 g/L peptone, 20 g/L dextrose, pH 6.0) in shaking flasks at 28°C. Escherichia coli DH5α and BL21 (DE3) strains for the cloning and expression of recombinant plasmids were purchased from TIANGEN Bio. Co., Ltd. (Beijing, China). E. coli DH5α and BL21 (DE3) strains were incubated at shaking flasks at 37°C in LB medium (10 g/L peptone, 5 g/L yeast extract, 10 g/L NaCl, pH 7.0).

Fresh cells of S. pararoseus NGR were collected by centrifugation after 3 days of culture in liquid YPD medium (17,000 × g, 1 min, 4°C), and genomic DNA was prepared by the Yeast Genomic DNA Extraction Kit (Solarbio, Beijing, China). Then, total RNA was extracted using the Spin Column Fungal Total RNA Purification Kit (Sangon, Shanghai, China). The total cDNA of S. pararoseus NGR was prepared by the PrimeScript II Ist Stand cDNA Synthesis Kit (Takara, Dalian, China), and was stored at −20°C for subsequent experiments.

According to the complete GGPPS gene sequence of S. pararoseus NGR resulting from transcriptome sequencing, a pair of specific primers (forward primer crtE-F and reverse crtE-R, listed in Table 1) were designed to clone the crtE with restriction enzymes sites for KpnI and BamHI. The 2 × High Fidelity PCR Master Mix (Sangon, Shanghai, China) was used to amplify crtE from genomic DNA or cDNA of S. pararoseus NGR, and the reaction procedure was as following: 1 cycle of 95°C for 3 min; 35 cycles of 95°C for 15 s, 55°C for 15 s, and 72°C for 1 min; at last, 72°C for 5 min. The amplified fragment was ligated into pMD18-T vector (Takara, Dalian, China) and transformed into E. coli DH5α, then extracted the plasmid pMD18T-crtE by TIANprep Mini Plasmid Kit (Tiangen Biotech Co., Ltd., Beijing, China) and sequenced before the further experiments.

Sequence alignment of crtE cDNA with its genomic DNA was performed using DNAMAN. The basic physicochemical properties of CrtE were predicted in the ProtParam online tool¹ with default parameters. Protein sequence alignment was performed according to the ClustalW algorithm with default parameters. Protein secondary structure prediction using the SPOPMA online tool. The three-dimensional structure of CrtE was predicted on the SWISS-MODEL online tool,² and the resulting CrtE model was visualized using the software Pymol (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.).

Plasmids pMD18T-crtE and pET32a were digested with restriction enzymes KpnI and BamHI (Takara, Dalian, China). The target gene fragment and pET32a linear fragment were ligated with T4 DNA ligase (Takara, Dalian, China) to construct the recombinant plasmid pET32a-crtE.

pET32a-crtE was transformed into BL21 (DE3) competent cells and pET32a empty vector was used as control. The BL21 (DE3) strain containing plasmids pET32a-crtE and pET32a, respectively, were cultured for 12 h, seeded in LB medium (containing 50 μg/mL ampicillin) at 1:20 and incubated at 37°C with shaking flasks until OD₆₀₀ = 0.6, then added with isopropyl β-D-thiogalactoside (IPTG) at final concentration of 1 mM for 4–5 h induction. The cells were centrifuged at 17,000 × g for 1 min, and the supernatant was discarded. The precipitates were washed twice with phosphate buffered saline (PBS) buffer. The precipitates were suspended with non-denatured cracking solution (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH8.0) and added PMSF (Yuanye Bio-Technology, Shanghai, China) with a final concentration of 1 mM. The suspending cells were broken on the ultrasonic cell crusher to release internal solutes then centrifuged at 21,000 × g for 30 min. After centrifugation, the supernatant and precipitation were used as samples for Sodium Dodecyl Sulfate-PolyAcrylamide Gel Electrophoresis (SDS-PAGE) to analyze the expression of CrtE and total protein.

To obtain purified CrtE proteins, the supernatant after cell crushing was subjected to HisTrap HP column (Cytiva, Marlborough, USA), and washed with five column volumes of wash solution (50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole, pH 8.0), and eluted with five column volumes of elution solution (50 mM NaH₂PO₄, 300 mM NaCl, 500 mM imidazole, pH 8.0). The resulting purified CrtE was concentrated by Amicon Ultra-0.5 Centrifugal Filter (10 kDa) (Millipore, Billerica, MA, USA) and exchanged with 100 mM HEPES buffer (containing 5 mM MgCl₂, pH7.5). Protein concentrations were determined by the Bradford Protein Assay Kit (Solarbio, Beijing, China).

Addition of 10 ng CrtE protein, 50 μM allylic substrate (DMAPP, GPP, or FPP) and 50 μM IPP to the enzymatic activity reaction system and made up to 200 μL with 100 mM HEPES buffer. DMAPP, GPP, FPP, and IPP were purchased from Sigma-Aldrich (Sigma-Aldrich, St Louis, USA). The reaction was incubated at 28°C for 2 h. Then added 200 μL Tris-HCl (pH 9.5) buffer containing 2 U SAP (Shrimp Alkaline Phosphatase) (Takara, Dalian, China) and 2 U TIPP (Thermostable Inorganic Pyrophosphatase) (NEB, USA), and reacted for 12 h at 30°C for dephosphorylation of the product. The reactions were stopped on ice, and added 400 μL hexane for extraction, repeated twice. Then the hexane layer was collected and evaporated to 100 μL under a N₂ stream, stored at −80°C for thin layer chromatography (TLC) analysis.

The above samples collected in hexane were analyzed by TLC in TLC Silica gel 60 F₂₅₄ (Merck, Germany), and GOH, FOH, and GGOH mixtures were used as standards, developed with toluene/acetonitrile/ethylacetate/acetic acid (35:5:15:0.15, v/v/v/v). The standards GOH, FOH, and GGOH were purchased from Sigma-Aldrich (Sigma-Aldrich, St Louis, USA). Color development was performed with iodine vapor and the product distribution was observed by TLC scanner KH-3100 (KEZHE, Shanghai, China).

The following protocol was used for in vivo functional analysis of the putative crtE gene. Plasmid pAC-LYC carries three carotenogenic genes crtE, crtB, and crtI from Erwinia herbicola and can synthesize lycopene in E. coli. As shown in Figure 1, a non-functional S. pararoseus strain NGR large subunit ribosomal RNA sequence (26S) (GenBank accession number ON197886) was inserted into the SacI site of pAC-LYC to construct plasmid pAC-LYC (ΔE). Since E. coli cannot synthesize GGPP by itself (Umeno et al., 2005), the plasmid pAC-LYC (ΔE) cannot synthesize lycopene in E. coli and therefore behaves as white. The plasmids mentioned above are shown in Table 2. Plasmid pET32a-crtE was transformed into the E. coli BL21 (DE3) carrying the plasmid pAC-LYC (ΔE), pET32a empty vector co-expressed with pAC-LYC (ΔE) in BL21 (DE3) as a control strain. Recombinant strains were incubated in LB medium (containing 50 μg/mL ampicillin and 25 μg/mL chloramphenicol) at 37°C until OD₆₀₀ = 0.6. IPTG with a final concentration of 1 mM was added for induction and cultured in dark for 48 h at 37°C. The cells were collected by centrifugation (11,500 × g, 10 min) to observe the color of recombinant E. coli as a preliminary identification. Lastly, qualitative and quantitative analysis of carotenoids were conducted using high performance liquid chromatography (HPLC) method.

Added 10 mL of dimethyl sulfoxide to the E. coli cells collected from 1 L of culture medium, mixed thoroughly, and treated in a water bath at 65°C for 30 min, shaking several times during the period. Then added 20 mL acetone and stranded 20 min, the pigment supernatant was collected by centrifugation (16,000 × g, 1 min). Sucked the supernatant and mixed it with 15 mL hexane, let it stood for a moment, and took the hexane layer containing pigment. The hexane layer was concentrated under nitrogen before HPLC detection. Carotenoids were detected using the Agilent 1290 Infinity II LC system (Agilent Technologies, Palo Alto, CA, USA) equipped with a reverse phase C₁₈ column (5 μm, 150 × 4.6 mm) (Thermo Fisher Scientific, Waltham, MA, USA) with an isocratic solvent system consisting of acetonitrile/methanol/isopropanol (85:10:5, v/v/v) at a flow rate of 1 mL/min, an injection volume of 20 μL, a column temperature of 32°C and a wavelength of 450 nm. Lycopene standard was purchased from Meilun Biotech Co., Ltd. (Dalian, China).

In order to clone the putative geranylgeranyl diphosphate synthase gene (crtE) from S. pararoseus NGR, specific primers, crtE-F and crtE-R, were designed based on the transcriptome data. We obtained a full-length cDNA and genomic DNA clone of the crtE gene, 1,134 and 1,722 bp, respectively. Comparison of the two sequences showed that the crtE genomic DNA contains eight introns and all splicing sites following the GT-AG rule (Figure 2A). The corresponding full length cDNA clone contains a complete open reading frame (ORF), encodes a polypeptide with 377 amino acids that was predicted in the ProtParam Tool to have a protein molecular weight is 42.59 kDa and the isoelectric point is 5.66. The cDNA sequences and deduced amino acids of crtE reported in this study are available in the GenBank databases under the accession number: KY652916.1.

To further explore the molecular features of the crtE gene expression product, MEGA7 software was utilized to compare the CrtE protein with the GGPPS of other species. The comparison of these amino acid sequences was shown in Figure 2B, with five highly conserved domains (from I to V), a result similar to that previously reported (Chen et al., 1994). Two of these aspartic acid-rich motifs are located in domains II and V, respectively, where near the N-terminal of the amino acid sequence is known as FARM (first aspartic acid-rich motif) and near the C-terminal is known as SARM (second aspartic acid-rich motif) (Zhang et al., 2021). In the previous crystal structures of prenyltransferase showed that the allylic substrate is bound in the middle of FARM and SARM via magnesium ions (Aaron and Christianson, 2010). In addition, the region which contains FARM and its the upstream five amino acids was defined as CLD (chain length determination region). This region can broadly distinguish the types of prenyltransferase. The CLD region of CrtE is identical to that of Rhodotorula toruloides GGPPS and differs from the human GGPPS CLD region by only the first amino acid upstream of FARM. The FARM region of type I GGPPS is DDXXD and its upstream fifth position is an aromatic amino acid with a large side chain and its upstream fourth position is a non-aromatic amino acid; the FARM region of type II GGPPS is DDXXXXD and its upstream fifth and fourth positions are non-aromatic amino acids; the FARM region of type III GGPPS is DDXXXXD and its upstream fifth and fourth positions are non-aromatic amino acids. Therefore CrtE is consistent with the CLD characteristics of type III GGPPS.

Used the SPOPMA online tool for the CrtE protein secondary structure prediction, as shown in Figure 3A, 55.97% of the α-helices, 10.34% of the extended chains, 5.84% of the β-turns, and 27.85% of the random coils. The main secondary structure of the CrtE protein is the α-helix, with the adjacent α-helices linked by other states.

Then, we used the online software SWISS-MODEL to predict the 3D-model of CrtE protein, and the predicted CrtE model has 46.9% sequence identity with the template protein (PDB ID code 6C56) with high reliability. To further explore the enzymatic activity characteristics of CrtE, we compared the CrtE model with the Saccharomyces cerevisiae GGPPS model (PDB ID code 2E8T) in the protein visualization software Pymol and embedded the ligands FPP (PDB ID code 2E90), IPP (PDB ID code 2E8U), and Mg²⁺ in CrtE. As shown in Figure 3B, CrtE is a dimeric structure where each subunit is composed of multiple α-helices as well as other secondary structures that link the α-helices in tandem. Eight of these α-helices are spatially arranged vertically, crossed below, and capped above by three short α-helices to form an activity cavity in the core. In addition, a terminal α-helix straddles between the two subunits of dimer and serves to maintain dimerization. We enlarged the activity cavity and the pyrophosphate group of the substrate FPP head was immobilized on the space between FARM and SARM by two Mg²⁺, while IPP was immobilized on the other position.

In order to express CrtE, plasmid pET32a and pET32a-crtE was transformed into E. coli BL21 (DE3) and expression was induced with IPTG. Following collection of cells for ultrasonic fragmentation, soluble proteins and insoluble precipitates were separated by centrifugation and subjected to SDS-PAGE assay. The total protein and CrtE were visualized by Coomassie Brilliant Blue R250 stain. It could be appeared through (Figure 4A) that the recombinant protein was expressed in the supernatant and inclusion bodies, and more in the supernatant, with a band of about 60 kDa. The molecular weight of the recombinant protein including one Trx⋅tag, one His⋅tag, one S⋅tag, and CrtE was predicted to be 59.06 kDa in ProtParam Tool, consistent with the results obtained by SDS-PAGE.

Recombinant protein pET32a-CrtE carries a His⋅tag and was purified using HisTrap HP column. As shown in Figure 4B the ninth lane corresponding to elution 4 was used to analyze the CrtE enzymatic activity. The recombinant pET32a-CrtE was used to study GGPPS enzymatic properties without cleavage of the His⋅tag, because of the latter has a negligible effect on enzymatic activity (Barbar et al., 2013).

The enzymatic activation reaction of CrtE protein occurred in HEPES buffer. The synthesis of the products was observed using three different allylic substrates and IPP as substrates for the reaction. As shown in Figure 5, both three allylic substrates can be used as substrates to synthesize the unique product GGPP. Among them, DMAPP produced the least GGPP as a substrate, which showing a very small spot, utilization of GPP was the second most abundant, and largest number of GGPP was produced using FPP. Furthermore, we can see that a portion of the substrate was remained in the system at the end of the reaction. Because we wanted to see if there was product generation other than GGPP by adding an excess of substrate to the reaction system, the results indicated that a specific product, GGPP, was produced regardless of which allylic substrate was utilized.

We successfully constructed plasmid pAC-LYC (ΔE) for the identification of the putative GGPPS gene of S. pararoseus (Figure 1). In order to test the functional activity of plasmid pAC-LYC (ΔE), we cloned the crtE gene (GenBank accession number M87280) from Erwinia herbicola and ligated it to the expression vector pET32a, which was named pET32a-crtE-ori. The colonies of E. coli BL21 (DE3) carrying plasmid pAC-LYC (ΔE) showed white. When transformed with plasmid pET32a-crtE-ori, the colonies changed to red, while the colonies remained white after transforming the empty vector pET32a (as shown in Figure 6). And HPLC analysis showed that plasmid pAC-LYC (ΔE) and pET32a-crtE-ori co-expressing strains accumulated lycopene (as shown in Figure 6). To sum up, these results suggest that plasmid pAC-LYC (ΔE) can exert its functional activity in E. coli BL21 (DE3).

The above complementary experiments are used to identified the functional of the crtE gene of S. pararoseus. The color of strains changed from white to red after transformation of plasmid pET32a-crtE into E. coli BL21 (DE3) carrying plasmid pAC-LYC (ΔE). HPLC analysis showed that the accumulating lycopene of the recombinant strain carrying pAC-LYC (ΔE) transformed with pET32a-crtE plasmid (as shown in Figure 6). The above results indicate that the crtE gene encodes a GGPPS functional protein in E. coli and its metabolites synthesize lycopene under the action of phytoene synthase (CrtB) and phytoene desaturase (CrtI). Furthermore, we found that E. coli harboring the plasmid pAC-LYC (ΔE) and S. pararoseus crtE produces fewer pigments than that of harboring pAC-LYC (ΔE) and E. herbicola crtE. A similar phenomenon has been observed in the process of functional characterization of the gene encoding GGPPS from Dunaliella bardawil (Liang et al., 2015). The expression of DbGGPS in E. coli led to less accumulation of the lycopene than that of the gene crtE from Erwinia uredovora. This could be due to the fact that CrtE of E. herbicola has a higher expression level than CrtE of S. pararoseus in E. coli or has a stronger enzymatic activity. In addition, certain characteristics of the protein itself or the culture conditions of E. coli may be the decisive factors of enzymatic activity.