Membrane-bound type prenyltransferases for aromatic substrates play crucial roles in the biosynthesis of various natural compounds. Lithospermum erythrorhizon p-hydroxybenzoate: geranyltransferase (LePGT1), which contains multiple transmembrane α-helices, is involved in the biosynthesis of a red naphthoquinone pigment, shikonin. Taking LePGT1 as a model membrane-bound aromatic substrate prenyltransferase, we utilized a baculovirus-Sf9 expression system to generate a high yield LePGT1 polypeptide, reaching ~ 1000-fold higher expression level compared with a yeast expression system. Efficient solubilization procedures and biochemical purification methods were developed to extract LePGT1 from the membrane fraction of Sf9 cells. As a result, 80 μg of LePGT1 was purified from 150 mL culture to almost homogeneity as judged by SDS/PAGE. Using purified LePGT1, enzymatic characterization, e.g. substrate specificity, divalent cation requirement and kinetic analysis, was done. In addition, inhibition experiments revealed that aromatic compounds having two phenolic hydroxyl groups effectively inhibited LePGT1 enzyme activity, suggesting a novel recognition mechanism for aromatic substrates. As the first example of solubilization and purification of this membrane-bound protein family, the methods established in this study will provide valuable information for the precise biochemical characterization of aromatic prenyltransferases as well as for crystallographic analysis of this novel enzyme family.
Aromatic substrate prenyltransferases (AS-PTs) catalyze the substitution reaction of an aromatic proton with a prenyl group provided by prenyl diphosphate to synthesize prenylated aromatics, which play various important biological roles in all organisms ranging from bacteria to humans. For example, AS-PTs mediate the key reaction step in the biosynthesis of naturally occurring quinones, such as ubiquinone (UQ) , plastoquinone [2, 3] and also vitamin E [4, 5], in which their basic skeletons are formed via prenylation of aromatic intermediates. In addition, prenylation reactions largely contribute to the diversification of the chemical structures and biological activities of aromatic natural products [6-11].
Due to the beneficial properties of prenylated compounds for human health in general and the crucial role of prenyl moieties for the biological activities of prenylated compounds in particular [6-11], many intensive studies have been carried out to characterize AS-PTs. For instance, whereas some soluble-type AS-PT genes involved in antibiotic biosynthesis were cloned from Streptomyces spp. and the crystal structures of representative enzymes, CloQ and NphB (Orf2), were solved [12, 13], there are no plant homologues of these soluble-type AS-PTs. For a membrane-bound AS-PT, in contrast, no example of purification to homogeneity has been reported even in bacteria, and thus no structural analysis of membrane-bound AS-PTs has been done yet, whereas heterologous expression and purification of a membrane-intrinsic prenyltransferase for non-aromatic substrates was reported from Sulfolobus solfataricus . Despite their biological importance, there has been a major hurdle for the study of membrane-bound AS-PTs because of difficulties in achieving high yield overexpression and in handling these hydrophobic proteins biochemically.
Para-hydroxybenzoic acid (PHB) prenyltransferases (PPTs) are a representative subfamily of membrane-bound AS-PTs, which have typically nine transmembrane α-helices in the 290 (ubiA) to 407 (AtPPT1) amino acid sequence [15, 16]. Genes coding for PPTs have been identified in many organisms from prokaryotes to eukaryotes, and their gene products are mostly responsible for UQ biosynthesis [15-20]. In the PPT family for UQ biosynthesis, all members characterized thus far show broad specificity for prenyl diphosphate substrates of different chain lengths. In contrast, Lithospermum erythrorhizon PHB geranyltransferase (LePGT1) , which is involved in the biosynthesis of a red naphthoquinone, shikonin, a representative secondary metabolite of this medicinal plant, shows strict specificity for geranyl diphosphate (GPP) as the prenyl donor. LePGT1 is also known as a key regulatory enzyme of shikonin biosynthesis  and not relevant to UQ formation .
This study employed LePGT1 as a model enzyme of the AS-PT family because this protein can be expressed in the non-plant host organism as its native form due to the lack of an N-terminal mitochondrial targeting signal, which is common for all eukaryotic members of this enzyme family. The high catalytic activity of LePGT1 is also advantageous [21, 22]. In addition, recombinant LePGT1 can be functionally expressed in budding yeast, and a mutational study was done to reveal critical amino acids directly involved in the enzymatic function . In this study, we overexpressed LePGT1 in Sf9 cells, from which the recombinant protein was solubilized and purified to homogeneity for the first time, and then characterized its enzymatic properties.
Heterologous expression of LePGT1 in insect cells
To express LePGT1 polypeptide fused with a histidine-tag (His-tag) at its C-terminus, we constructed a plasmid, pDEST8-LePGT1-His. A recombinant bacmid containing LePGT1-His was prepared by use of the Bac-to-Bac system (Invitrogen) for insect cell expression. Monolayer cultures of Sf9 cells were then infected with the virus and the expression of LePGT1 with His-tag was monitored by immunoblotting using a Penta-His antibody. As shown in Fig. 1A,B, a clear signal was observed in immunoblotting at the expected molecular weight of 29 kDa, whereas no signal was observed in the negative control (control virus infected cells).
The enzymatic activity of recombinant LePGT1 expressed in Sf9 cells was confirmed by the measurement of PHB geranyltransferase activity using the microsome fraction of transformed Sf9 cells. We detected high PHB geranyltransferase activity (3.99 ± 0.57 μmol·h−1·mg−1 protein), whereas no geranyltransferase activity was detected in the microsomes of negative control Sf9 cells (Fig. 1C,D).
Solubilization and purification of LePGT1 expressed in insect cells
To purify recombinant LePGT1, we prepared a suspension culture of Sf9 cells in a 300 mL flask (75 mL culture medium) from a static culture. Total activity present in 150 mL LePGT1-expressing Sf9 culture was estimated to be 233 μmol·h−1, corresponding to ~ 1000-fold the activity obtained by the yeast expression system previously reported for this membrane protein . First, we tested three cell homogenization methods – (a) 40 mL Dounce tissue grinder (glass homogenizer, Wheaton); (b) Teflon homogenizer (5 mL Teflon vessel, Sansyo); and (c) Sonicator (Sonifier 250, Branson) – to evaluate the recovery rate of LePGT1 protein and its enzymatic activity in the crude membrane fraction from cell homogenate. Immunoblotting and PHB geranyltransferase activity in three fractions, i.e. 2500 g supernatant, 20 000 g supernatant and 20 000 g precipitate, obtained from three different homogenization methods revealed that sonication was the most suitable method to enrich the functional LePGT1 polypeptide in the crude membrane fraction as judged by both the total recovery of LePGT1 polypeptide and the high specific activity (Fig. S1).
To optimize the solubilization conditions in order to preserve the enzymatic activity, we screened several detergents (n-dodecyl-β-d-maltoside; n-octyl-β-d-glucoside; 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate; Triton X100; sodium deoxycholate, DOC). All detergents used at the indicated concentrations with microsomal fraction expressing LePGT1 showed only mild effects on enzyme activity in vitro (Fig. S2A). Among the detergents, DOC showed the highest recovery of LePGT1 activity from microsome as the soluble form (Fig. S2B). The concentration of DOC and the treatment time to solubilize functional LePGT1 were then optimized to 6 mm DOC and a treatment time of 60 min (Fig. S2C). It should be noted, however, that a short treatment time of 5 min was actually sufficient to solubilize the functional LePGT1, while a long treatment did not improve solubilization efficiency. Rather, a gradual decrease in enzyme activity was observed during treatment (Fig. S2D). In fact, LePGT1 solubilized with DOC exhibited low stability during storage after His-tag-mediated affinity purification, i.e. most enzyme activity was lost within 3 days after solubilization (Fig. S3). Thus, DOC was replaced by 0.2 mm digitonin shortly after the solubilization step for the subsequent affinity purification because solubilized LePGT1 exhibited much higher stability in digitonin-containing buffer (Fig. S3). By this procedure, LePGT1 was recovered from the Ni2+-resin with 500 mm imidazole revealing a single band in SDS/PAGE (Fig. 2). The yield of purified LePGT1 was 80 μg from 34.3 mg crude protein of Sf9 cells derived from 150 mL culture volume. The final purification factor was 41.3-fold, and the overall yield was 9.10% (Table 1).
Table 1. Affinity purification of LePGT1-His from 1.5 × 107 Sf9 cells
Total protein (mg)
Total activity (μmol·h−1)
Specific activity (μmol·h−1·mg−1)
Characterization of purified LePGT1
We conducted enzymatic characterization of LePGT1 with purified recombinant enzyme. To assess the substrate specificity, we used a series of prenyl diphosphates of different chain lengths, i.e. dimethylallyl diphosphate (DMAPP, C5), GPP (C10), farnesyl diphosphate (FPP, C15) and geranylgeranyl diphosphate (GGPP, C20), for the prenyltransferase assay. [14C]-PHB was used as the common prenyl acceptor. The reaction product was separated by normal-phase TLC, and the radioactive signals were detected with a BAS1800 image analyzer (Figs 3A and S4A). Purified LePGT1 exhibited strict substrate specificity for GPP, which is in agreement with data obtained from the crude microsomal fraction of transgenic yeast and that of native enzyme in L. erythrorhizon cultured cells . We also investigated various aromatic compounds as prenyl acceptor substrate in addition to PHB. Among the tested aromatic compounds, LePGT1 revealed high specificity for PHB to give m-geranyl-p-hydroxybenzoic acid (GBA) (100%). The other aromatic compounds gave no apparent peak or an extremely small unidentified peak in HPLC analysis within the retention time range of the geranylated compounds. The quantities of these putative reaction products were estimated as GBA equivalents: o-hydroxybenzoic acid, 5.8%; m-hydroxybenzoic acid, 2.8%; and benzoic acid, no detectable peak. These data suggest that purified LePGT1 has narrow substrate specificity for aromatic substrates.
Eukaryotic PPTs exhibit an absolute requirement for divalent metal ions for their enzyme activity . Therefore, we examined the preference of various divalent cations using PHB and GPP as substrates. As shown in Fig. 3B, Mg2+ was the most preferred cation for the activity of LePGT1 among the divalent cations examined. The other divalent cations gave apparently lower or no activity of PHB geranyltransferase: Co2+, 34.8%; Fe2+, 15.0%; Mn2+, 8.54%; and Ca2+, Cu2+ and Zn2+, 0%; where the value of Mg2+ was set to 100%. Enzyme assays at various pH values indicated that the optimal pH was 7.0–7.5 (Fig. S4B) and the optimal temperature was about 45°C (Fig. S4C).
To obtain further information on the recognition mechanism of aromatic substrates, a competition analysis was carried out using various phenolic compounds in the LePGT1 activity assay (Fig. 4). Each of 10 aromatic compounds shown in Fig. S5 was added to the reaction mixture at the same concentration of PHB, and the influence of the added aromatic compound on PHB geranyltransferase activity was evaluated by HPLC. Among them, only a limited number of compounds, such as hydroquinone, homogentisic acid and catechole showed strong inhibition (2.0%, 7.1% and 11% activities of control reaction, respectively), whereas chlorogenic acid and caffeic acid showed only a weak influence on the activity, i.e. ~ 70%–80% of control. Compounds showing strong competition/inhibition are simple phenols having phenolic OH residues, while carboxylic acid does not seem to influence the enzyme activity. Phenylpropanoid gave no apparent inhibition.
Substrate saturation kinetics of the purified LePGT1 for PHB and GPP showed a typical Michaelis–Menten-type kinetics for this transferase (Fig. S6). Kinetic parameters were estimated from double-reciprocal plots of the initial velocity versus the substrate concentration: Km values for PHB and GPP were 35.8 μm and 430 μm, respectively, and Vmax was 450 μmol·h−1·mg−1 protein (Table 2).
Table 2. Kinetic parameters of LePGT1 of various preparations (Km μm, Vmax μmol·h−1·mg−1). ND, not determined
Our previous report, using a crude microsomal fraction of yeast transformant of LePGT1, described that PHB prenyltransferase family members had highly conserved aspartate-rich sequences in two hydrophilic loops, and that the substitution of aspartate with alanin, especially D87 and D212, caused a dramatic decrease in enzyme activity . To demonstrate the importance of these essential aspartates with homogeneously purified enzyme, we prepared two mutant enzymes whose conserved aspartates were replaced with alanin (D87A and D212A, respectively) using the Sf9-baculovirus system. As a result, only D87A mutant could be successfully expressed, whereas the D212A mutant showed almost no expression of the recombinant protein. The enzyme assay revealed that the purified D87A had completely lost its PHB geranyltransferase activity, consistent with data from experiments using crude microsomes of LePGT1-expressing yeast.
In this study we have established a method for high-level functional expression, solubilization and effective purification of LePGT1. Using purified LePGT1, we have characterized the substrate specificity and kinetic parameters of this enzyme. In our previous report, three-dimensional molecular modeling of LePGT1 suggested that the tight binding pocket formed by the N-terminal aspartate-rich region in the catalytic center caused the strict preference for GPP as the prenyl substrate . This model was supported by biochemical data obtained from chimeric enzyme expressed in a yeast expression system . In the present study, purified LePGT1 expressed in Sf9 cells was used for enzymatic characterization to show again the strict substrate specificity for GPP, indicating that the solubilized and purified LePGT1 retained the tight binding pocket for GPP. In addition, other enzymatic properties, such as divalent cation requirement, pH optimum and temperature optimum were also consistent with the native LePGT previously characterized with crude microsomal fraction. These data strongly support that the purification procedures, including solubilization, did not alter the critical enzymatic properties of LePGT1.
Two conserved hydrophilic regions in the PPT subfamily (i.e. regions I and III ) play a central role for recognition and catalysis of two substrates coordinately [23, 25]. In particular, the conserved aspartates in regions I and III are the most important amino acids for the catalytic function of LePGT1, e.g. D87 in region I recognizes the diphosphate moiety of GPP via two Mg2+ atoms through chelating bonds . In agreement with this, D87A mutant completely lost enzyme activity, in both yeast recombinant enzyme and the homogeneously purified enzyme from Sf9. The importance of the C-terminus for substrate affinity has also been suggested in the present study. The purified LePGT1 with His-tag fused at the C-terminus showed an apparently higher Km value for GPP than no-tagged LePGT1 expressed in yeast (Table 2) . The native LePGT partially purified from L. erythrorhizon cultured cells  also showed lower Km than C-terminus His-tagged recombinant LePGT1 (Table 2), suggesting that the C-terminal region of LePGT1 influences the affinity for GPP, whereas GPP recognition via Mg2+ would instead be directly mediated by the N-terminal D-rich region as predicted by analogy with other prenyltransferases for chain elongation.
Our purified enzyme showed strong substrate specificity for PHB. Although some other aromatic compounds may also be recognized as prenyl acceptors, their relative activities were extremely low compared with PHB, in a similar manner as was found for the partially purified native LePGT from L. erythrorhizon , e.g. 2.8% for m-hydroxybenzoic acid in our study and 2.6% in partially purified native LePGT. The consistency in aromatic substrate preference between the purified recombinant LePGT1 and the native enzyme also shows that our purification procedures are appropriate for obtaining pure LePGT1 protein retaining native enzymatic properties.
The influence of various aromatic compounds on LePGT1 activity was screened in the presence of the native substrate, PHB. As revealed, simple compounds having two free phenolic hydroxyl groups, such as hydroquinone, catechole and homogentisic acid, substantially inhibited the enzyme activity of LePGT1 (Figs 4 and S5). The IC50 of the strongest inhibitor, hydroquinone, was calculated to be 3.8 μm (Fig. S7). Such a strong inhibition was not observed when one phenolic hydroxyl group was replaced with a carboxyl group. Moreover, most phenylpropanoids, such as ferulic acid and p-coumaric acid, gave almost no inhibition, probably due to the bulky propionic acid portion. These findings suggested that, in addition to PHB, two free phenolic hydroxyl groups of simple aromatic compounds can be recognized at the aromatic substrate binding site of LePGT1. Three-dimensional modeling of LePGT1 depicted that the PHB molecule is situated between two α-helices of the conserved hydrophilic loops , where small aromatic compounds with two free phenolic hydroxyl groups may be inserted into the catalytic center and spatially interfere with the recognition of PHB, although the prenylation reaction does not occur.
In nature, there are many prenylated aromatic compounds  whose biosynthesis involves membrane-bound AS-PTs. As a result of recent intensive studies on plant AS-PTs, some new AS-PT genes have been identified [26-28]. Among these members and LePGT1, many common structural features are seen, e.g. similar putative topology, D-rich motif, and the position of the motif in the hydrophilic loops. Also biochemically these new members showed a similar requirement for a divalent cation for their enzyme activity and the same narrow substrate specificity. Our results reported here will provide valuable information for further investigations of various membrane-bound prenyltransferases at a precise biochemical level, and also make it possible to carry out crystallization and three-dimensional structural analyses of this membrane-bound enzyme family in the near future.
Materials and methods
DMAPP, GPP, FPP and GGPP were synthesized as described previously . [14C]-PHB was purchased from Sigma (St Louis, MO, USA). DOC and digitonin were purchased from Wako Pure Chemicals (Osaka, Japan). Other chemical reagents including substrate and inhibitors were provided by Wako Pure Chemicals, Nakarai Tesque (Kyoto, Japan) and Sigma-Aldrich (St Louis, MO, USA).
LePGT1 cDNA fused to histidine-tag (His-tag) at the C-terminus was subcloned into the pENTR3C vector (Invitrogen, Carlsbad, CA, USA) via EcoRI and XhoI sites to yield pENTR-LePGT1-His vector. This entry vector was used for LR recombination with pDEST8 to construct pDEST8-LePGT1-His. MAX Efficiency DH10Bac Competent Cells (Invitrogen) were used for production of a recombinant bacmid used in the Bac-to-Bac Baculovirus Expression System (Invitrogen) to produce LePGT1 in Sf9 cells (Spodoptera frugiperda) according to the manufacturer's protocol.
Expression of LePGT1 in Sf9 cells
Sf9 cells were grown at 27°C as a monolayer culture in serum-free Sf900 medium (Gibco, Carlsbad, CA, USA) or as a suspension culture in Grace's insect medium with 10% fetal bovine serum plus 0.1% Pluronic F-68 (Invitrogen). The monolayer culture was utilized to increase virus titer, and also for small-scale functional expression of recombinant LePGT1. Suspension cultures were maintained in a rotary shaker (BioShaker BR-13FP, Taitec, Saitama, Japan) with a rotation speed of 150 rpm at 27°C. Sf9 cells in suspension culture were infected with the baculovirus at a multiplicity of infection of 0.5 according to the manufacturer's instructions; 72 h after infection, cells were harvested and washed with ice-cold phosphate-buffered saline. Cells were then stored at −80°C until use.
Preparation of microsomal membrane
All steps were performed at 0–4°C. Frozen Sf9 cells from suspension culture (75 mL culture, 75 × 106 cells) were thawed and resuspended in 5 mL of sonication buffer (final concentration 50 mm Tris/HCl pH 7.5, 1 mm EDTA, 10% (v/v) glycerol). The cell suspension was sonicated for 10 s × three times (duty cycle 50%, output control 3) with a sonicator (Sonifier 250, Branson, Danbury, CT, USA) and then centrifuged at 2500 g for 5 min at 4°C to remove cell debris and nuclei. The supernatant was further centrifuged at 100 000 g for 30 min to pellet the microsomal fraction. The microsome pellet was resuspended in 2 mL of ice-cold buffer A (final concentration 50 mm Tris/HCl pH 7.5, 300 mm NaCl, 10% (v/v) glycerol) containing 20 mm imidazole with a Dounce homogenizer (Wheaton, Millville, NJ, USA). At the first selection step of the homogenization methods shown in Fig. S1, centrifugation at 20 000 g for 30 min was applied to yield crude membrane fraction.
Solubilization and purification of LePGT1 from Sf9 microsome membrane using His-tag
To solubilize recombinant LePGT1, the microsomal membrane was resuspended with buffer A containing 10 mm DOC and 20 mm imidazole and kept on ice with gentle mixing for 5 min. The insoluble fraction was removed by centrifugation (100 000 g, 30 min). Solubilized proteins were applied to Ni2+-nitrilotriacetic acid-agarose (Ni2+-NTA-agarose) (Qiagen, Venlo, Netherlands), pre-equilibrated with buffer A containing 10 mm DOC, and the mixture was rotated for 5 min. The resin was then washed four times with a 2.5-fold bed volume of buffer A containing 0.2 mm digitonin and 20 mm imidazole. The solubilized LePGT1 was eluted with a 2-fold bed volume of buffer A containing 0.2 mm digitonin and 500 mm imidazole. Protein concentrations were determined by the Bradford method using a protein assay kit (Bio-Rad, Hercules, CA, USA). Bovine serum albumin was used as standard.
SDS/PAGE and immunoblotting analysis
SDS/PAGE and immunoblotting were carried out according to Terasaka et al. , with slight modifications. In brief, proteins were denatured in denaturation buffer [10 mm Tris/HCl pH 8.0, 40 mm dithiothreitol, 1 mm EDTA, 10% (w/v) sucrose, 10 mg·mL−1 pyronine Y and 2% (w/v) SDS] for 15 min at 50°C, subjected to SDS/PAGE (10% or 12% gel) and then transferred to an Immobilon polyvinylidene difluoride membrane (Millipore, Billerica, MA, USA). For blocking, the membrane was treated with Blocking One (Nacalai Tesque) overnight at room temperature. Then, the membrane was incubated with Penta-His antibody (Penta-His HRP Conjugate, Qiagen) and subsequently the signal was visualized by chemiluminescence (Western Lighting Chemiluminescence Reagent Plus, PerkinElmer Life Sciences, Waltham, MA, USA).
Measurement of LePGT1 enzymatic activity
PHB prenyltransferase activity measurements using cold or radioisotope-labeled substrate were performed according to Yazaki et al. , with slight modifications. In non-radioisotope experiments, the reaction product GBA was quantitatively measured by HPLC (Shimadzu LC-10A system, Kyoto, Japan) using the peak area at 254 nm absorbance with testosterone propionate as an internal standard. The reaction product was identified by direct comparison with the standard sample. HPLC conditions were as follows: column, LiChrosphere 100RP-18 (Merck, Rahway, NJ, USA) 4 × 250 mm; column oven, 40°C; solvent system, methanol/H2O/acetic acid (80 : 20 : 0.3); flow rate, 1 mL·min−1; detection, 230–320 nm with an SPD6A photodiode array detector (Shimadzu). With a radioactive substrate of [14C]-PHB, the reaction products were analyzed by silica gel TLC (Kiesel gel, Merck, 20 cm × 20 cm) with a solvent system of benzene/ethyl acetate (4 : 1). TLC plates were exposed to an imaging plate (Fujifilm) at room temperature for 7 days and then analyzed with a BAS1800 image analyzer (Fujifilm, Tokyo, Japan).
We thank Dr Masaharu Mizutani of Kobe University for providing the Sf9 cells, Dr Masayoshi Maeshima of Nagoya University for valuable suggestions on the purification technique and the Life Research Support Center in Akita Prefectural University for analysis of DNA sequencing. This work was supported in part by a Grant-in-Aid for Scientific Research (No. 24310156 to K.Y.) of the Ministry of Education, Culture, Sports, Science and Technology, a grant from the Research for the Future Program ‘Molecular mechanisms on regulation of morphogenesis and metabolism leading to increased plant productivity’ (No. 00L01605 to K.Y.) of the Ministry of Education, Culture, Sports, Science and Technology of Japan, and a Research Fellowship from the Japan Society for the Promotion of Science for Young Scientists (No. 17.2011 to K.O.).