Kinetics of Phosphomevalonate Kinase from Saccharomyces cerevisiae

The mevalonate-based isoprenoid biosynthetic pathway is responsible for producing cholesterol in humans and is used commercially to produce drugs, chemicals, and fuels. Heterologous expression of this pathway in Escherichia coli has enabled high-level production of the antimalarial drug artemisinin and the proposed biofuel bisabolane. Understanding the kinetics of the enzymes in the biosynthetic pathway is critical to optimize the pathway for high flux. We have characterized the kinetic parameters of phosphomevalonate kinase (PMK, EC 2.7.4.2) from Saccharomyces cerevisiae, a previously unstudied enzyme. An E. coli codon-optimized version of the S. cerevisiae gene was cloned into pET-52b+, then the C-terminal 6X His-tagged protein was expressed in E. coli BL21(DE3) and purified on a Ni2+ column. The KM of the ATP binding site was determined to be 98.3 µM at 30°C, the optimal growth temperature for S. cerevisiae, and 74.3 µM at 37°C, the optimal growth temperature for E. coli. The KM of the mevalonate-5-phosphate binding site was determined to be 885 µM at 30°C and 880 µM at 37°C. The Vmax was determined to be 4.51 µmol/min/mg enzyme at 30°C and 5.33 µmol/min/mg enzyme at 37°C. PMK is Mg2+ dependent, with maximal activity achieved at concentrations of 10 mM or greater. Maximum activity was observed at pH = 7.2. PMK was not found to be substrate inhibited, nor feedback inhibited by FPP at concentrations up to 10 µM FPP.


Introduction
The mevalonate pathway is an important conduit for the production of crucial metabolites with a wide array of functions, including terpenoids [1,2], hormones and steroids [3]. The heterologous expression of this pathway in Escherichia coli has enabled high-level production of the antimalarial drug artemisinin [4][5][6], but the chemical structures of these metabolites also make them interesting targets for solving some of the most crucial problems in the energy market [7,8]. With only slight modifications to mevalonate pathway intermediates and products, either in vivo or through traditional chemical engineering processes post cell culture extraction, these molecules can be transformed into biofuels that, depending on our ability to scale-up, could offset or replace traditional liquid fuels [9]. This would allow us to replace petroleum-based, CO 2 producing fuels with fuels that are carbon neutral. Although industrial-scale corn-based ethanol production is already a reality in the energy market, ethanol is a less than desirable biofuel because not only does it divert crops from the food supply, it is not compatible with our current distribution infrastructure or vehicle fleet [10].
Whether these fuel alternatives are five-carbon alcohols derived from the mevalonate pathway intermediates isopentenyl pyrophosphate and dimethylallyl pyrophosphate [11], or downstream, terpene-based molecules like bisabolene [8], further improvement of titers may be realized through a more robust understanding of the enzymes in the mevalonate pathway and the ways in which those enzymes are regulated by metabolic intermediates. In particular, proteomics data has previously shown that the fourth and fifth enzymes in the pathway-mevalonate kinase (MK) and phosphomevalonate kinase (PMK), respectively-are expressed at relatively low levels and may be targets for increasing overall isoprenoid production [12,13]. Previous work has also shown that substrate inhibition and feedback inhibition of MK may be responsible for limiting flux through the pathway [14]. Because MK-a phosphotransferase that acts on mevalonate and ATP to yield mevalonate-5-phosphate-and PMK-a phosphotransferase that acts on mevalonate-5-phosphate and ATP to yield mevalonate-5-diphosphate-both require ATP to function and downstream prenyl phosphates might act as general ATP binding site inhibitors, PMK was identified as another potential source of pathway regulation.
PMKs from other sources have been studied revealing implications for pathway engineering. For example, PMK from E. faecalis is Mn 2+ dependent rather than Mg 2+ dependent [15]. Pig-derived PMK is substrate inhibited by ATP under high ATP, low mevalonate phosphate concentrations [16]. If S. cerevisiae PMK is similarly dependent or inhibited it would make an ideal target for protein engineering. Furthermore, S. cerevisiae prefers to grow at 30uC, but much of our production takes place in E. coli, which necessitates understanding how PMK activity is affected by a change in growth temperature from 30uC to 37uC. Herein we report cloning a codon-optimized sequence of S. cerevisiae PMK into an expression vector, the expression and purification of PMK in E. coli, and the kinetic characterization of the purified enzyme.

Results and Discussion
Although phosphomevalonate kinase (PMK) from S. cerevisiae has previously been studied in partially purified lysates [17], and even utilized to study the kinetics of another enzyme [18], this is the first time PMK from S. cerevisiae has been kinetically characterized in isolation. In a study of the partially purified enzyme it was reported that pH did not affect PMK activity, but we found that PMK does have an optimal activity at pH = 7.2, and its activity drops off below pH = 6.5 and above pH = 8.0 ( Figure 1). Although at first glance there is an apparent ''shoulder'' in the pH profile, careful consideration of the profile shows that the shoulder is within error and therefore cannot be considered to conclusively exist. Although we did not test a wide array of storage conditions, solutions with high PMK concentrations were found to be stable long term only at pH = 8.0 with 800 mM NaCl. As found previously S. cerevisiae PMK shows a cation dependence on Mg 2+ , with 10 mM corresponding to maximal activity ( Figure 2).
Kinetic constants were determined by nonlinear regression analysis using the solver function in Microsoft Excel. The K M for ATP, K M ATP , was determined to be 98.3 mM and 74.3 mM at 30uC and 37uC, respectively. The K M for mevalonate-5phosphate, K M mev-p , was determined to be 885 mM and 880 mM at 30uC and 37uC, respectively ( Figure 3). V max was determined to be 4.51 mmol/min/mg enzyme and 5.33 mmol/min/mg enzyme at 30uC and 37uC, respectively ( Figure 4). In contrast, the K M ATP , K M mev-p , and V max for the Enterococcus faecalis PMK, which is Mn dependent, were reported to be 170 mM, 190 mM, and 3.9 mmol/ min/mg enzyme [15]. The values for the Streptococcus pneumonia PMK were reported to be 74 mM, 4.2 mM, and 5.5 mmol/min/mg enzyme [19]. The values for pig liver PMK have been reported to be 43 mM, 12 mM, and 51 mmol/min/mg enzyme [16]. For the recombinant human PMK, the values were reported to be 107 mM, 34 mM, and 46 mmol/min/mg enzyme [20]. The high K M mev-p for the S. cerevisiae PMK makes it less ideal than enzymes with a low K M , as it would only reach its maximal rate at a high concentration of mevalonate-5-phosphate. Because of the Mn dependence of the E. faecalis PMK, it may not function fully if expressed in E. coli or other organisms. In contrast, the S. pneumonia, pig, and human PMKs have reasonable values for K M ATP and K M mev-p , making them better choices for a heterologous pathway. In terms of maximum rates, the mammalian enzymes are high than the microbial enzymes.
Because the S. cerevisiae PMK has been used heterologously in E. coli for production of isoprenoids [4][5][6][7][8]11], the temperature effect on PMK activity is important, particularly at E. coli's optimal growth temperature of 37uC. Despite expectations that PMK activity might diminish with increasing the temperature from the preferred 30uC growth temperature of S. cerevisiae to the 37uC preferred by E. coli, PMK activity was shown to slightly increase with the increase in temperature. This increased activity bodes well for the production of isoprenoid products, including advanced biofuels, via the mevalonate pathway if the low protein expression levels currently observed can be increased [12,13]_ENREF_9. It should be noted that although we were able to achieve very high yields of PMK using pET-52b+ for the purpose of isolating and purifying the enzyme, increasing PMK expression in production strains by using high copy plasmids would be counterproductive to increasing overall biofuels production as doing so would divert an unnecessary amount of resources into the production of protein to the detriment of fuel titers.
One regulatory mechanism for controlling PMK activity we can rule out is feedback inhibition, as the presence of farnesyl . Of the publications reporting the kinetics of PMKs from various organisms, none have reported inhibition by prenyl phosphates. Furthermore, unlike S. cerevisiae mevalonate kinase [18], PMK did not demonstrate substrate inhibition. The lack of feedback and substrate inhibition in the S. cerevisiae PMK is an attractive feature for increasing production of a desired isoprenoid. Nevertheless, S. pneumonia PMK, which has a high V max and low K M s, is a much better enzyme and should be incorporated into future production strains. An additional advantage of the S. pneumonia PMK is that its crystal structure of the has been solved   [21] and the kinetic mechanism of its catalysis has been described in detail [19].
With the addition of PMK from this study, the S. cerevisiaederived mevalonate pathway enzymes that have been kinetically characterized include hydroxymethylglutaryl synthase [22], hydroxymethylglutaryl reductase [14], mevalonate kinase [18], phosphomevalonate decarboxylase [23], and farnesyl pyrophosphate synthase [24], leaving acetyl-CoA C-acetyltransferase and isopentenyl diphosphate isomerase uncharacterized. Although isopentenyl diphosphate isomerase has been isolated and studied [25], the difficulty associated with detecting the isomerization of a single bond is likely why the kinetic constants have yet to be determined. In combination with traditional genetic engineering techniques, such as varying promoter strength, and newly developed technologies for varying expression, such as RBS calculators [26], studying the kinetics of these remaining enzymes should allow isoprenoid production from engineered microbes to be optimized more rationally.

Codon Optimization of PMK
The original S. cerevisiae PMK sequence (accession number NM_001182727), which was downloaded from the BioCyc.org database, was codon optimized by DNA2.0 (Menlo Park, CA) for expression in E. coli. Codon optimization replaced codons rare for E. coli with more frequently used codons. The sequences of the original and codon-optimized versions of the genes are presented in Figure S1.

Expression Plasmid Construction
A chemically-competent strain of E. coli DH10B was transformed with pET-52b+ (Novagen, Germany) and then used to prepare the plasmid according to the instructions and materials in a Qiagen (Valencia, California) Spin Miniprep Kit. The codon-optimized PMK sequence was PCR amplified with primers that added a BsaI restriction site with an NcoI overhang on the 59 end of the sequence and a SacI restriction site on the 39 end of the sequence, then digested with the appropriate restriction enzymes (all enzymes from New England BioLabs, Ipswich, Massachusetts), and cloned into pET-52b+ to make expression plasmid pET-52b+ _coPMK-His. Confirmation of expression plasmid construction was accomplished by sequencing the cloning region using T7 primers (sequencing and primers from Quintara Biosciences, Albany, California).

PMK-His Expression and Purification
Ideal conditions for PMK expression were screened on NuPAGE 10% Bis-Tris SDS-PAGE gels and the supplies indicated in the accompanying protocol (Invitrogen, Grand Island, New York) from 5-mL cultures that spanned a range of media types, growth temperatures, inducer concentrations, and growth times. Protein expression was ultimately accomplished by growing a 2-L culture in Terrific Broth (Invitrogen) to OD 600 = 0.6 at 37uC, inducing with 100 mM IPTG (Sigma Aldrich, St. Louis, Missouri), then growing at 18uC for approximately two days (until stationary phase was reached). Cells were pelleted in 250-mL portions, flash frozen in liquid nitrogen after medium removal, and then stored at 280uC prior to further processing. On ice, cells from one 250-mL portion were suspended in 25 mL of a lysis buffer (10 mM Imidizole, 300 mM NaCl, 50 mM NaH 2 PO 4 , pH = 8.0; Sigma Aldrich), sonicated for 10 minutes in a water bath to break up residual clumps, then homogenized with two passes through an EmulsiFlexH-C3 (Avestin, Canada). Cell debris was removed by centrifugation at 12,000 X g for 30 minutes. Cleared lysate was bound to 2-mL of Ni-NTA resin (Qiagen) at 4uC by rocking gently for 30 minutes. The resin was then bedded in a column, washed with 20 column volumes (CV) of buffer containing 20 mM imidizole, then the protein was eluted with 10 CV of buffer containing 500 mM imidizole. Buffer exchange into 20 mM Tris, 50 mM NaCl, pH = 7.0 was accomplished on an AKTA (GE Healthcare Life Sciences, Pittsburgh, Pennsylvania) using a GE Healthcare HiPrep 26/10 Desalting Column (17-5087-01). Protein was then concentrated using VivaSpin 20 3,000-MWCO filters (Sartorius, Bohemia, New York). Protein concentration was determined using a Nanodrop (Thermo Scientific, West Palm Beach, Florida). The protein was then diluted so that glycerol (Sigma) was 50% v/v and stored at 220uC.

Activity Assay
All chemicals and supporting enzymes were purchased from Sigma-Aldrich. Reaction progress was monitored spectrophotometrically at 339 nm for NADH consumption on a 96-well plate in a Spectramax M2 (Molecular Devices, Sunnyvale, California). 100-mL enzymatic assay mixtures contained 200 mM Tris (pH = 7.2), 100 mM KCl, 10 mM MgCl 2 , 0.81 mM NADH, 1.5 mM phosphoenolpyruvate, 0.682U pyruvate kinase, 0.990 U lactate dehydrogenase, 0.1 mg PMK, 0.1-8.0 mM ATP, and 0.2-10.0 mM mevalonate-5-phosphate. Stock concentrations of NADH and pH neutralized ATP were confirmed through their extinction coefficients ( ATP e 259 nm = 15.4 mM 21 cm 21 , NADH e 339 nm = 6.22 mM 21 cm 21 ). All conditions were repeated twelve times for statistical analysis, from which K M (mM) and reaction velocities (mM mev-PP formed*minute 21 * mg PMK 21 ) were calculated. When studying pH effect and divalent cation dependence, ATP and mevalonate-5-phosphate were held constant and data were normalized to the maximum observed reaction velocities. To ensure PMK was the rate-limiting enzyme, when necessary the following standard controls and results were verified: doubling the PMK added doubled the observed rate, doubling the supporting enzymes added did not affect the observed rate, and doubling the phosphoenolpyruvate concentration did not affect the observed rate.