Stable carbon isotope discrimination by human 3-hydroxy-3-methylglutaryl-coenzyme A reductase.

The aim of this study was to investigate the possible existence and magnitude of stable carbon isotope discrimination by human 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR). The catalytic portion of HMGR was expressed and purified. The reaction product mevalonate was lactonized and extracted from the reaction mixture by a solid-phase extraction protocol. Stable carbon isotope ratios of mevalonolactone (MVL) were analyzed by gas chromatography-combustion-isotope ratio mass spectrometry. An average fractionation factor (12)k/(13)k of 1.0031 +/- 0.0004 for all carbon atoms contained in MVL was estimated by the method of internal competition. The value was calculated by nonlinear curve fitting, where the ratio (13)C/(12)C of MVL was plotted versus the fraction of reaction.

Abstract The aim of this study was to investigate the possible existence and magnitude of stable carbon isotope discrimination by human 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR). The catalytic portion of HMGR was expressed and purified. The reaction product mevalonate was lactonized and extracted from the reaction mixture by a solid-phase extraction protocol. Stable carbon isotope ratios of mevalonolactone (MVL) were analyzed by gas chromatography-combustion-isotope ratio mass spectrometry. An average fractionation factor 12 k/ 13 k of 1.0031 6 0.0004 for all carbon atoms contained in MVL was estimated by the method of internal competition. The value was calculated by nonlinear curve fitting, where the ratio 13 C/ 12 C of MVL was plotted versus the fraction of reaction.-Lüdke, S. M., U. Flenker, W. Schänzer, and D. Schomburg. Stable carbon isotope discrimination by human 3-hydroxy-3-methylglutaryl-coenzyme A reductase. J. Lipid Res. 2008. 49: 2620-2626.
Supplementary key words mevalonate pathway • steroids • kinetic isotope effect • isotope ratio mass spectrometry • curve fitting • doping control Kinetic isotope effects (KIEs) present in enzyme-catalyzed reactions may evidence the corresponding reaction mechanism (1) and account for the natural variation of the stable isotope signatures of biological compounds (2,3). These parameters are of rapidly increasing interest in the geological, biological, and forensic sciences, to name but some.
The interpretation of such data will be facilitated by a sound understanding of the underlying processes. Although a perfect analysis of isotope fluxes in metabolic networks is virtually impossible (3), and purely empirical "isotope fingerprinting" is applied routinely and successfully, reliable predictors can be found and quantified from time to time. For example, lipids are typically strongly depleted in 13 C versus other classes of biological compounds (4). This phenomenon usually is attributed to KIEs during the decarboxylation of pyruvate (5).
Considering this phenomenon, the stable carbon isotope ratios of human steroid hormones exhibit somewhat unusual values. Steroids isotopically are only slightly lighter than carbohydrates and proteins (6). Hence, steroid biosynthesis is unlikely to be limited by pyruvate decarboxylation, and other processes must control the stable isotope ratios of steroids.
We hypothesized that owing to its key position in steroid biosynthesis, 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR) might play a pivotal role here. The hypothesis is supported by the observation that steroid hormones reveal stronger 13 C depletion when contraceptives are administered (6). This probably indicates the emergence of rate limitation resulting from feedbackinduced downregulation of the enzyme. HMGR is among the most highly regulated enzymes in nature (7). The different levels of regulation include transcription, translation, enzyme degradation, and phosphorylation. HMGR (EC 1.1.1.34) catalyzes the four-electron reduction of HMG-CoA to mevalonate and CoASH: HMG-CoA 1 2 NADPH 1 2 H1 → mevalonate 1 CoA-SH 1 2 NADP1. The human enzyme assumes a rate-limiting position in the mevalonate pathway. Numerous biomolecules originate from this process, including haem, ubiquinone, and cholesterol. Besides its diverse functions in animals, the latter forms the backbone of steroid hormones.
Chemical reactions involving bond making or breaking almost always reveal normal isotope effects, which means slower reaction of the isotopically heavier isotopologues (1). Therefore, the reaction of HMGR is likely to discriminate substrate molecules with carbonyl-13 C vicinal to the thioester bond between the HMG and the CoA moiety (Fig. 1). The aim of this study was to investigate the pos-sible existence and magnitude of isotope discrimination by HMGR in order to provide better understanding of the factors that control the 13 C/ 12 C ratios of sterols, steroids, and other products of the mevalonate pathway.

Vector design
A plasmid containing the gene for human HMGR was purchased from the American Type Culture Collection. The catalytic subunit (residues 420-888) (8) was amplified by PCR introducing an AflIII and BamHI restriction site at the ends (restriction enzymes from New England Biolabs). The fragment was inserted into the pNHis vector (9) between the NcoI and BamHI restriction sites using standard ligation procedures with T4 DNA ligase (Roche). The pNHis vector introduces an N-terminal polyhistidine tag that permits affinity purification. The ligation mixture was transformed into Escherichia coli DH5a. Vectors containing the insert were identified by agarose gel electrophoresis of minipreparations of plasmid DNA from single colonies. A suitable vector was amplified in DH5a cells (Qiagen plasmid midi kit). To avoid fragmentary expression starting from an internal translation initiation site (10), M485 was mutated to I485 by side-directed mutagenesis (reagents from NEB). According to Istvan et al. (8), the enzymatic activity of the M485I mutant protein is unchanged compared with the wild-type protein.

Protein expression and purification
For expression of the fusion protein, the vector was transformed into E. coli BL(DE3). Cells were grown in Luria Bertani medium (11) at 37°C until they reached an optical density at 600 nm of 0.8, then induced with 1 mM isopropyl-b-D-thiogalactoside and grown at 25°C for 4 h. Purification was carried out using an Ni-NTA Agarose affinity column (Qiagen) according to a vendor-provided protocol. Protein concentrations were determined by the method of Bradford (12). The two main fractions were combined, concentrated by centrifugal filtration (Millipore am-icon ultra 30,000), and re-buffered in potassium phosphate buffer (13) (0.2 M KCl, 0.16 M potassium phosphate, 0.004 M EDTA, and 0.01 M DTT, pH 6.8). Enzyme purity was checked by SDS-PAGE. For storage, the solution was spiked with 10% (v/v) glycerol, and aliquots of 200 ml were kept at 220°C.

Enzymatic reaction
The reaction system consisted of 160 mM potassium phosphate (pH 6.8), 200 mM KCl, 420 myM NADPH (Sigma), and 311 myM (R,S)-HMG-CoA (Fluka) in a final volume of 800 ml. The reaction was initiated by adding the enzyme to the complete assay mixture in a volume that was sufficient to complete the conversion within approximately 14 min. The HMG-CoA-dependent oxidation of NADPH was monitored at 340 nm in a spectrophotometer (Phar-macia Biotech Ultrospec 2000) equipped with a cell holder adjusted at 37°C. Up to three samples of 250 ml per run were taken and transferred into ice-cooled glass vials. The vials had been equipped with a solution of pravastatin (Sigma) to inhibit further reaction. For samples taken at fractions of reaction smaller than 0.15, volumes of 500 ml were pipetted.

Sample preparation
For lactonization of mevalonic acid, the samples volume was doubled by addition of 0.1 M HCl, and the reaction was left unattended at room temperature for about 1 h. Solid-phase extraction was conducted as described by Saini et al. (14), with slight modifications. Briefly, the samples were individually transferred to a solid-phase extraction cartridge (Separtis Isolute ENV1 100 mg/3 ml) that had been washed and preconditioned with 3 ml methanol followed by 1 ml water. The sample was allowed to enter the column bed by gravity flow, and the column was washed with 500 ml HCl followed by 500 ml water. The cartridges were dried under gentle vacuum suction. The column bed was conditioned with 500 ml methanol, and another volume of 500 ml eluted the analytes. The methanol extract solution was dried in a vacuum dessicator over phosphorus pentoxide. For gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS) analysis, analytes were taken up in individually determined volumes of tert-butyl methyl ether, in order to minimize concentration varieties.

Stable isotope analysis
Stable carbon isotope analysis of mevalonolactone (MVL) was performed by GC-C-IRMS. The hardware encompassed an Agilent GC 6890 (Agilent Technologies) coupled to a Thermo Delta-plus XP gas isotope ratio mass spectrometer. The system was largely modified as described by Flenker, Güntner, and Schänzer (15).
Briefly, the solvent removal, which is essential to GC-C-IRMS, was performed before passage of the GC column. A KAS-4 Gerstel cooled injection system supported by an auxiliary vacuum was employed for this purpose.
Solvent removal was achieved at a temperature of 5°C within an interval of 1 min. The vent flow was adjusted to 50 ml/min, and the auxiliary vacuum was set to 20.1 bar. The sample transfer to the GC column was performed at 200°C within another minute. The material was re-condensed on the column at 35°C. Subsequently, the injection port was purged at a temperature 300°C, where the gas flow was set to 100 ml/min. The temperature rates between the different steps amounted to 10 K/s. The GC column was a Macherey and Nagel Optima-1701. The dimensions were 30 m length, 320 mm inner diameter, and 0.5 mm film thickness. The carrier gas flow was set to 1.8 ml/min, where constant flow mode was used.
The initial GC temperature (35°C) was maintained for 2 min. The temperature then was raised to 135°C at 40 K/min. Subsequently, the temperature was changed to 250°C at a rate of 10 K/min. Fig. 1. Enzyme reaction of 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR). Isotope effects are expected at C-1, C-2, and C-3, the group of atoms adjoining the cleaved C-S bond.
Typically, two reference gas pulses were employed to calculate the 13 C/ 12 C ratio of MVL. For this purpose, one uncontaminated pulse before and one after the MVL signal were selected. Figure 2 shows a typical analysis and illustrates the procedure. The reference gas was calibrated versus an n-alkane mixture kindly provided by the University of Indiana (16).

Data evaluation
The spectrophotometric absorbance was plotted against time. During sampling, the cover of the spectrophotometer had to be opened. Data measured within these intervals were removed from the plot. A cubic smoothing spline (17) was fitted to the re-mainder of the data. It was assumed that the observed decrease of NADPH concentration depended on two different factors, the consumption of NADPH during the enzyme reaction and the enzyme-independent decay of NADPH. The latter depends on pH and temperature (18).
In order to isolate the enzymatic effect, the curve was corrected by addition of the absolute difference of two decay lines. The latter were individually calculated according to the NADPH concentration before and after complete conversion of (S)-HMG-CoA, respectively. The fraction of reaction then was calculated as the ratio of the difference between initial absorbance and absorbance at the time of sampling divided by the total difference of absorbances (Fig. 3).
The fractionation factor (a) and the isotopic ratio of the product molecule after complete conversion (R ∞ ) were estimated by nonlinear least-squares curve fitting. The software was R in its latest version where the nls-library was employed (19).
The model formula was derived from the rate laws of the stable carbon isotopes (see Appendix): R t is the isotopic ratio of the sample at time t and f is the corresponding fraction of reaction. The KIE of the enzyme reaction is then given by KIE 5 (a 2 1) 5 ( 12 k/ 13 k) 2 1; (Eq: 2) where 12 k and 13 k indicate the rate constants for 12 C and 13 C, respectively.
RESULTS AND DISCUSSION Figure 2 shows a typical GC-C-IRMS analysis of MVL following our protocol for enzymatic production, sample preparation, and 13 C/ 12 C analysis. Fig. 2. Gas chromatography-combustion-isotope ratio mass spectrometry analysis of mevalonolactone (MVL) isolated from the HMGR reaction mixture. R1-R5, reference gas pulse; UNK, unkown contaminant. R3 and R5 were used to calculate the 13 C/ 12 C of MVL in this run. In this case, the reference gas pulses R3 and R5 were employed to calculate the 13 C/ 12 C of MVL. The contamination of R4 appeared irregularly and probably results from the solid-phase extraction. The MVL signal itself is clearly resolved and certainly yields valid results. Because the HMGR reaction was stopped after varying intervals, the resulting MVL concentrations were rather different. The required amounts of solvent, however, were estimated in advance, and thus, approximately identical peak heights could be obtained. In fact, these varied by less than a factor of 3, which falls well within the linear dynamic range of IRMS analyses. Figure 4 shows the observed relation between f and R t . The fitted curve according to equation 1 is plotted. Analysis of the residuals suggested presence of some outliers. In fact, two outliers were confirmed by Huber statistics (20). The curve was refitted after removal of these points. The corresponding estimates for a, R ∞ , and the calculated KIE are listed in Table 1.
The standard deviation of the residuals is close to the known repeatability precision of GC-C-IRMS analyses. Moreover, the residuals are independent from f and exhibit random scatter. It can therefore be assumed that the relation between R and f is described adequately.
Both estimates for a and R ∞ and their corresponding confidence intervals could be confirmed by application of bootstrap methods (21).
KIEs are commonly determined (22)(23)(24) by the method of internal competition (1). The calculation employs the equation proposed by Tong and Yankwich (25) or its simplified version by OʼLeary (26), both describing a in terms of f, R t , and R ∞ . This method requires several measurements of the isotopic composition R t of the starting material or the reaction product over the range from as little as 1% reaction to as much as 50% reaction. Second, the value of R ∞ , which is the isotopic composition of the product at total conversion, must be known.
Because of the steep dependence of the product isotopic composition at the end of the reaction (see Fig. 1), R ∞ is afflicted with large experimental errors and can be regarded as the major source of bias. However, the curve-fitting method used in this study does not require any exact value for R ∞ . In fact, R ∞ and a are estimates obtained from the fitting process.
The model equation, which was used to calculate the KIE, is based upon the assumption of a pseudo first-order irreversible enzyme reaction. Human HMGR utilizes two equivalents of NADPH as a cosubstrate. In order to meet the first-order demand, limitation by a substrate other than HMG-CoA was avoided by excess NADPH.
The calculated KIE has to be treated as the average isotope effect on all carbon atoms contained in the product molecule. Assuming that the isotope effect would be largely restricted to C-1 and only normal isotope effects contribute to the observed value, the maximum expected depletion of 13 C at C-1 can be assessed, at least roughly. Thereto the average isotope effect is multiplied with the number of carbon atoms present in the product molecule, in this case six for MVL. The resulting value of 18.6‰ is not unusual in enzymatic reactions in which carbon bonds are cleaved. For example, bacterial decarboxylases have been reported to exhibit KIEs between 14.7‰ and 18.2‰ at C-1 (22,26).
However, Melzer and Schmidt (5) investigated the pyruvate dehydrogenase reaction from two different organisms and calculated position-specific isotope effects. In fact, 13 C primary isotope effects from 9‰ to 25‰ were observed and secondary isotope effects still caused depletions around 3‰.
From the above, it is certainly inadequate to attribute the entire isotope effect of HMGR to C-1. The effect would rather show some kind of distribution between the atoms C-1, C-2, and possibly even C-3 (Fig. 1). Nonetheless, a calculated propagation of 13 C depletion from mevalonate to cholesterol would remain unaffected. Cholesterol incorporates all three of the carbon atoms concerned from six molecules of mevalonate. The depletion would therefore propagate with 18.6 3 6/27 5 4.1‰ to the C 27 -body of cholesterol (Fig. 5). Likewise, pregnanes incorporate five groups of C-1 to C-3 and would show a 13 C depletion of about 18.6 3 5/21 5 4.4‰.
In contrast, androstanes incorporate C-1 and C-2 from mevalonate at equal numbers, but lack one atom C-3 (Fig. 5). The expected 13 C depletion of this compound  depends on the actual effect attributed to C-3. Considering it to fall between one-third and zero of the overall KIE, androstanes would exhibit depletions between (18.6 3 5 2 18.6/3)/19 5 4.6‰ and 18.6 3 5/19 5 4.9‰, respectively. All of these numbers represent estimates that rely on sole limitation of steroid biosynthesis by HMGR.
The results are of practical relevance for doping control, where stable isotope ratios are used to identify the primary sources of steroids. The method compares the 13 C/ 12 C ratios of androgens or corresponding metabolites to those of steroids from independent pathways. For example, testosterone (T, C 19 ) may be compared with pregnanediol (C 21 ). The latter thus serves as "endogenous reference compound." Synthetic androgenic steriods betray their origin in that their 13 C/ 12 C values significantly differ from those of the endogenous reference compounds. Up to now, with very few exceptions, only 13 C depletions have been observed for synthetic steroids.
Consequently, during the application of synthetic T, the 13 C/ 12 C ratios of T itself and its metabolites will be reduced compared with the purely biosynthetic pregnanes. Depending on the considered target and reference compounds, a difference of at least 3‰ on the y-scale 1 must be exceeded in order to tell the presence of synthetic steroids (28).
The described doping control methodology presumes that physiological factors do not significantly increase the corresponding differences in stable isotope composition. As demonstrated above, full limitation by HMGR could result in a maximum difference in 13 C depletion of 4.9 2 4.4 5 0.5‰ when comparing C 19 -and C 21 -steroid hormones. Therefore 13 C discrimination by HMGR would only be able to slightly quench the 13 C enrichment attributed to doping. It can be stated that the influence of the HMGR isotope effect will not significantly exceed the analytical error generally present in GC-C-IRMS. It is un-likely that corresponding results will be corrupted by the described phenomena. Nonetheless, it should be kept in mind that different steroids, although originating from an identical pathway, do not necessarily exhibit identical stable isotope signatures. Because the atoms of mevalonate will have a nonstatistical distribution of 13 C, the measured average depletion depends on the atoms incorporated into or removed from the analyzed molecule.
Apart from practical issues, our results allow one to draw interesting conclusions concerning the nature of steroid biosynthesis. Crude lipids were found to be depleted in 13 C by 7‰ relative to biomass (4,27). The rate limitation of acetate formation catalyzed by pyruvate dehydrogenase is likely to be responsible for this phenomenon (5). However, among lipids, FFAs exhibit considerable 13 C/ 12 C depletions of around 8‰, whereas steroids are only sightly depleted by ?1‰, compared with biomass (6). The lack of 13 C depletion rules out rate limitation of steroid formation by the decarboxylation of pyruvate.
It is usually assumed that HMGR features the rate-limiting step in sterol synthesis. In the present study of the HMGR reaction, a depletion propagating with 4-5‰ to sterols has been calculated. Considering that sterols appear much less depleted than can be predicted from this result, it may be concluded that under physiological conditions, most of the HMG-CoA emerging in the relevant compartment is converted to mevalonate. This would result in the disappearance of significant HMGR-induced isotope fractionation, for the most part.
In general, the influences of in vivo isotope effects are modulated by various factors such as the branching of carbon fluxes to alternative metabolic pathways or compartmentalization. To interpret patterns of isotopic depletion, the structure and characteristics of the related network of reactions has to be considered (3). Thus, the results presented contribute to the interpretation of isotopic patterns in the mevalonate pathway.
APPENDIX R t AS A FUNCTION OF f, a, AND R ∞ We presume a system of pseudo first-order irreversible reactions of the form A → B with the rate constant k. 1 The y 13 C value is defined as: d 13 C(‰) 5 ( @ 13 C#/@ 12 C# sample @ 13 C#/@ 12 C# standard 2 1) Á 1000 The standard is carbon dioxide from PeeDee Belemnite isotopic standard limestone (29).

Fig. 5
Cholesterol incorporates carbon atoms C-1 to C-3 from six mevalonate molecules. The denoted positions 1 to 3 refer to the mevalonate molecule. C 19 -steriode hormones lack the side chain and concurrently one of the C-3 atoms.