- Email: [email protected]
Isolation and identification of two isomeric forms of malonyl-coenzyme A in commercial malonyl-coenzyme A
ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 328 (2004) 203–209 www.elsevier.com/locate/yabio
Isolation and identiWcation of two isomeric forms of malonyl-coenzyme A in commercial malonyl-coenzyme A Paul E. Minkler,a Vernon E. Anderson,b Nakul C. Maiti,b Janos Kerner,c and Charles L. Hoppela,d,¤ a
Medical Research Service, Louis Stokes Department of Veterans AVairs Medical Center, Cleveland, OH 44106, USA b Department of Biochemistry, Case Western Reserve University, Cleveland, OH 44106, USA c Department of Nutrition, Case Western Reserve University, Cleveland, OH 44106, USA d Department of Pharmacology and Medicine, Case Western Reserve University, Cleveland, OH 44106, USA Received 9 December 2003
Abstract Two isomers of malonyl-coenzyme A (malonyl-CoA) were detected in a commercial preparation of malonyl-CoA. These compounds were separated by preparative high-performance liquid chromatography (HPLC) and characterized by HPLC/ultraviolet (UV)/mass spectrometry. Both compounds had a UV absorbance maximum at 259–260 nm. Both compounds underwent negative electrospray ionization to produce a [M-H]¡quasi-molecular ion at m/z 852 and both compounds underwent collision-induced dissociation to produce a characteristic fragment at m/z 808, all consistent with the structure of malonyl-CoA. Nuclear magnetic resonance spectrometry showed that the two chromatographically distinguishable malonyl-CoAs are structural isomers: the major component is the naturally occurring malonyl-CoA and the contaminant is 30-dephospho-20-phospho-coenzyme A. 2004 Elsevier Inc. All rights reserved. Keywords: Malonyl-coenzyme A; HPLC; Electrospray ionization; Mass spectrometry; NMR
Malonyl-coenzyme A (malonyl-CoA)1 is the prime regulatory molecule in the control of cellular long-chain fatty acid oxidation due to its inhibition of mitochondrial carnitine palmitoyltransferase I [1–4]. The quantiWcation of malonyl-CoA in biological systems has contributed to a better understanding of its role in regulation of mitochondrial fatty acid oxidation [5]. Bench-top electrospray ionization/ion-trap mass spectrometers (capable of high-sensitivity full-scan MS and tandem MS) employed as detection devices following HPLC have proven to be a substantial technical
¤
Corresponding author. Fax: +216-229-8509. E-mail address: [email protected] (C.L. Hoppel). 1 Abbreviations used: malonyl-CoA, malonyl-coenzyme A; 20-AMP, adenosine 20-monophosphate; 30-AMP, adenosine 30-monophosphate; 50-AMP, adenosine 50-monophosphate; 20,50-ADP, adenosine 20,50-diphosphate sodium salt; 30,50-ADP, adenosine 30,50-diphosphate sodium salt; 30-dephoCoA, 30-dephospho-coenzyme A; 20-phospho-malonylCoA, 30-dephospho-20-phospho-coenzyme A. 0003-2697/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.01.015
innovation [6]. We applied this technology to the detection of malonyl-CoA but, in the course of developing an HPLC separation compatible with mass spectrometric detection, we observed what appeared to be a contaminant in a commercial preparation of malonyl-CoA. Further reWnement of the chromatographic conditions led to the clear chromatographic separation of the contaminant from malonyl-CoA. In this communication, we provide evidence that establishes the contaminant as an isomer of malonyl-CoA. To accomplish this, the two compounds were separated by preparative HPLC and characterized by HPLC/ UV/MS. While the two forms are chromatographically diVerentiated, they are indistinguishable by both UV spectrophotometry and mass spectrometric detection. NMR spectroscopy demonstrated conclusively that the two malonyl-CoA isomeric forms present in the commercial preparation are the naturally occurring malonyl-CoA and the contaminant 30-dephospho-20-phospho-coenzyme A (20-phospho-malonyl-CoA).
204
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
Materials and methods Chemicals and solvents Malonyl-CoA (lots 052K7029, 111K7026, and 122K 7019), adenosine 20-monophosphate (20-AMP), adenosine 30-monophosphate (30-AMP), adenosine 50-monophosphate (50-AMP), adenosine 20,50-diphosphate sodium salt (20,50-ADP), adenosine 30,50-diphosphate sodium salt (30,50ADP), 30-dephospho-coenzyme A (30-dephoCoA), and ammonium formate were purchased from Sigma–Aldrich (Milwaukee, WI). Methanol (HPLC-grade) was purchased from Fisher ScientiWc (Cleveland, OH). HPLCgrade water was prepared using a Milli-Q reagent water system (Millipore, Bedford, MA). The D2O used in NMR experiments was purchased from Cambridge Isotope Laboratories (Andover, MA). The ion-trap mass spectrometer’s electrospray source used nitrogen for both the sheath and the auxiliary gases provided from liquid nitrogen dewar tanks available through Praxair Distribution (North Royalton, OH). Helium (5.0 ultra-high purity), which was employed as both the damping and the collision gas, was purchased from Praxair. Preparative HPLC Preparative-scale puriWcation of malonyl-CoA was performed using a BioCAD SPRINT perfusion chromatography system (PerSeptive Biosystems, Cambridge, MA) conWgured to perform preparative-scale reversedphase HPLC. The column used was an Alltima C18LL 5, 250 £ 22-mm purchased from Alltech Associates, (DeerWeld, IL). The eluents used were A (95% 0.1 M ammonium formate/5% methanol) and B (50% 0.1 M ammonium formate/50% methanol). The instrument was programmed to create a binary gradient of 100% A to 80% A/20% B over 40 min. A 2-ml injection loop was installed in the manual injector. The pump Xow rate was 10 ml/min and detection was by UV at 254 nm. Fractions (0.5 min) were collected with a Model FC-205 fraction collector (Gilson, Middleton, WI).
preparative HPLC. The Xow rate was 0.5 ml/min. The HPLC diode array detector was set to collect a single wavelength (259 nm with a reference wavelength of 400 nm) and full-scan UV spectra from 190 to 400 nm. The mass spectrometer was an LCQ Deca ion-trap instrument purchased from ThermoFinningan (San Jose, CA). The instrument was operated using the electrospray ionization source in the negative ion mode with the following settings: sheath gas, 90 units; auxiliary gas, 30 units; spray voltage, 2 kV; capillary temperature, 350 °C; capillary voltage, ¡22 V. The mass spectrometer was programmed to generate data from two sequential scan events: (1) full-scan MS, from m/z 700 to 1000, and (2) full-scan MS/MS on 852.2 from m/z 230 to 1000 with a collision energy setting of 26% and an isolation width of m/z 3.0. 1
H NMR
NMR samples were taken to dryness under vacuum in a SpeedVac (Savant), redissolved in 0.5 ml of D 2O to remove exchangeable protons, evaporated, and redissolved in D2O. NMR spectra were obtained with a 600-MHz Varian Inova (Varian, Lake Forest, CA) spectrometer at ambient temperature.
Results and discussion Isolation of malonyl-CoA and 20-phospho-malonyl-CoA Fig. 1 shows the preparative-scale HPLC chromatogram of malonyl-CoA. This result was generated by dissolving 8.2 mg of malonyl-CoA lithium salt (Sigma lot 052K7029) in 100 l of eluent A and injecting this into the preparative HPLC. Two large peaks were observed
Analytical HPLC with UV spectrophotometric and mass spectrometric detection The analytical HPLC/UV/MS system consisted of an HP1100 series quaternary pump with on-line degasser, autosampler, column heater, and diode array detector purchased from Agilent (Wilmington, DE). The column used was a Hypersil BDS C18 3 100 £ 4-mm analytical column cartridge with an integrated precolumn (4 £ 4 mm) packed with 5 Hypersil BDS C18, also purchased from Agilent. The column heater was operated at 40 °C. The HPLC pump was programmed to create a binary gradient of 100% A to 50% A/50% B over 20 min, where eluents A and B were identical to those used for
Fig. 1. Chromatogram resulting from an injection of 8.2 mg of malonyl-CoA (Sigma lot 052K7029) into the preparative HPLC chromatograph. The conditions are described under Materials and methods. The two chromatographic peaks labeled Peak 1 (malonyl-CoA) and Peak 2 (20-phospho-malonyl-CoA) were collected and the material contained isolated.
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
(labeled Peak 1 and Peak 2 in Fig. 1). Peak 1 was collected between 23.5 and 27 min and Peak 2 was collected between 27.5 and 29.5 min. Both Peak 1 and Peak 2 fractions were lyophilized and the residues reconstituted in water to give Wnal concentrations of 2.7 mM for Peak 1 and 0.37 mM for Peak 2 (the concentrations were determined spectrophotometrically using 15.6 cm¡1 mM¡1 as the absorbtivity coeYcient for coenzyme A at 258 nm). UV spectrophotometric and mass spectrometric characterization of malonyl-CoA and 20-phospho-malonylCoA Peak 1 and Peak 2 solutions were then injected (20 l) into the analytical HPLC. The chromatograms for Peak 1 (light tracings) and Peak 2 (dark tracings) are displayed in Fig. 2. In this Wgure, the two chromatographic injection results are superimposed and each of three simultaneous chromatographic data sets are displayed: UV at 259 nm (top), full-scan MS from m/z 700 to 1000
205
(middle), and full-scan MS/MS on 852.2 from m/z 230 to 1000 (bottom). One chromatographic species is dominant in each chromatogram, and the spectral data associated with each of these peaks is displayed to the left (Peak 1) and right (Peak 2) of the dominant peak. The UV spectra absorption maximas for both Peak 1 and Peak 2 at 259–260 nm are consistent with the absorption maximum of coenzyme A. The electrospray ionization mass spectra are consistent with a malonyl-CoA [M-H]¡ quasi-molecular ion at m/z 852.2 (C24H37N7O19P3S D 852.107784 u). The collision-induced dissociation product ion (MS/MS) spectra with prominent signals at m/z 834.2 and 808.2 are consistent with the loss of H2O (C24H35N7O18P3S D 834.097219 u) or the loss of CO2 (C23H37N7O17P3S D 808.117955 u) and are mechanistically reasonable malonyl-CoA products. Our conclusions after examination of the Peak 1 and Peak 2 spectra (UV, MS, and MS/MS) are that (1) they are consistent with malonyl-CoA and (2) the two chromatographic peaks are indistinguishable by either UV spectroscopy or mass spectrometry.
Fig. 2. Chromatograms resulting from injections of the isolated Peak 1 material (malonyl-CoA) and Peak 2 material (20-phospho-malonyl-CoA) from Fig. 1 using the analytical HPLC/UV/MS system. The conditions are described under Materials and methods. The chromatograms from the injection of Peak 1 material are displayed with the lighter tracings and the chromatograms from the injection of Peak 2 material are displayed with the darker tracings. The three stacked chromatograms resulted from simultaneous collection of three data sets: UV at 259 nm (top), full-scan MS from 700 to 1000 m/z (middle), and full-scan MS/MS on 852.2 from 230 to 1000 m/z (bottom). Spectral data associated with each of these peaks is displayed to the left (Peak 1) and right (Peak 2) of the dominant chromatographic peak.
206
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
IdentiWcation of malonyl-CoA and 20-phospho-malonylCoA by NMR The 1H NMR spectra of Peak 1 and Peak 2 shown in Fig. 3 are both consistent with the presence of coenzyme A thiolesters with peaks attributed to adenine, pantetheine, and ribose all present [7]. The thiolmethylene resonance near 2.9 ppm indicates the presence of a thiolester [8]. The absence of any additional peak arising from the acyl moiety is consistent with the presence of a malonyl thiolester as the malonyl C2 protons readily exchange with the solvent D2O [9].
There are signiWcant diVerences between the 1H NMR spectra of Peak 1 and Peak 2, indicating that the two compounds separated by HPLC are diVerent and do not interconvert either on the NMR time scale or in the time that elapsed between isolation and acquisition of the 1H NMR spectra (i.e., days). These diVerences center on the resonances for the ribose protons and speciWcally for the 20 and 30 protons. In 1H NMR spectra of coenzyme A and common coenzyme A thiolesters, these protons have resonances at 4.65–4.75 ppm and have been unequivocally assigned both by 2D COSY spectra and by the 3bond coupling of the 30 1H to the 31P [7]. These features
Fig. 3. 1H NMR spectra. (Bottom) Peak 1 (malonyl-CoA), (Top) Peak 2 (20-phospho-malonyl-CoA). The conditions are described under Materials and methods. Numerical labels indicate identities of protons.
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
207
Fig. 4. 1H NMR spectra. (Bottom) 30,50-ADP. (Top) 20,50-ADP. The conditions are described under Materials and methods. Numerical labels indicate identities of protons. The “*” labeled signal is the HOD peak.
are shared by the 1H NMR spectrum of 30,50-ADP shown in Fig. 4. Phosphorylation of a ribose hydroxyl group has two eVects on the 1H NMR spectrum. It results in a deshielding of »0.3 ppm, as reXected by the comparisons in Table 1, and an additional splitting due to the 3JHP of »7 Hz. These two features are observed in the diVerence between the 1H NMR spectra of Peak 2 and Peak 1 shown in Fig. 3. Transferring the phosphate from the 30hydroxyl to the 20-hydroxyl results in a »0.2-ppm deshielding of the 20 1H and a corresponding increase in shielding of the 30 1H so that they appear at 4.90 and 4.24 ppm, respectively. This isomerization of the phosphate position is mirrored in the diVerences between the standard 20,50-ADP and the 30,50-ADP spectra shown in
Fig. 4, conWrming the assignment. Therefore, the two chromatographically distinguishable species of malonylCoA present in the commercial preparation (identical by both UV and MS) are unequivocally shown by NMR to be the naturally occurring malonyl-CoA (Peak 1) and the contaminant 20-phospho-malonyl-CoA (Peak 2). Speculation on the origin of the 20-phospho-malonyl-CoA contamination The Wrst lot of malonyl-CoA that we obtained (lot 052K7029), whose content launched this investigation, was determined to have 12–13% 20-phospho-malonyl-CoA by UV detection using either preparative or
208
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
Table 1 NMR data for standards vs malonyl-CoA (Peak 1) and 20-phospho-malonyl-CoA (Peak 2) Compound
Chemical shifts (ppm)
Malonyl-CoA (Peak 1) 30,50-ADP 30-AMP 30-Dephospho-CoA 20-Phospho-malonyl-CoA (Peak 2) 20,50-ADP 20-AMP
10H
20H
30H
40 H
50H
6.014 6.025 5.955 6.049 6.097 6.098 6.114
4.701 4.669 4.662 4.655/4.646 4.902 4.921 5.042
4.665 4.657 4.564 4.269 4.240 4.247 4.164
4.428 4.392 4.280 4.414 4.478 4.465 4.427
4.073 3.980 3.746 4.105 4.087, 4.057 3.946 3.749
Table 2 Percent of 20-phospho-malonyl-CoA found in commercial malonyl-CoA Malonyl-CoA lot No.
HPLC detection method
Amount injected
20-Phospho-malonyl-CoAa (%)
052K7029 052K7029 052K7029 111K7026 111K7026 122K7019 122K7019
UV from preparative HPLC UV from analytical HPLC MS/MS from analytical HPLC UV from analytical HPLC MS/MS from analytical HPLC UV from analytical HPLC MS/MS from analytical HPLC
8.2 mg; N D 1 0.2 nmol; N D 1 0.2 nmol; N D 1 6 nmol; N D 10 0.2 nmol; N D 10 6 nmol; N D 10 0.2 nmol; N D 10
12.8 12.5 9.1 6.0 § 0.3b 3.1 § 0.3b 4.7 § 0.3b 2.4 § 0.3b
a b
Determined by dividing the MS/MS or UV peak areas of HPLC Peak 2 by the sum of the peak areas of Peak 1 + Peak 2 times 100%. Mean § standard deviation.
analytical HPLC (Table 2). By analytical HPLC with MS/MS detection the value was 9%. We have subsequently purchased two additional lots of malonyl-CoA from the same vendor (lots 111K7026 and 122K7019) and examined them for 20-phospho-malonyl-CoA content and found lower, although still detectable, amounts of 20-phospho-malonyl-CoA (Table 2). Since the migration of the ribose 30-phosphate to the 20-phosphate does not occur at a signiWcant rate under physiological conditions, the appearance of the 20-phospho-malonyl-CoA contaminant presumably arises during synthesis. Acid catalyzes the phosphate migration, and acid quenches used in coenzyme A thiolester syntheses may promote isomerization [10]. A second mechanism for isomerization could come from activating the phosphate group [11]. The activation of the phosphate may result directly from reaction with the activated acyl group or by reaction with residual activating reagent if the activated acyl group is not isolated prior to addition to the coenzyme A. In the presence of activating reagent, the 30-phosphate is activated and displaced by the 20-hydroxyl group to form the 20,30 cyclic phosphodiester, which hydrolyzes to give both 20- and 30-phosphate esters. SigniWcance of contaminant The signiWcance of a contaminant in a reagent is a function of the combined eVects of the amount of contamination and the perturbation that the contaminant has on the system being used. For us, a primary issue with the presence of 20-phospho-malonyl-CoA in the commercial malonyl-CoA is the accuracy of preparing
standardized solutions. We standardize solutions of malonyl-CoA by determining the UV contribution of coenzyme A at 258 nm and then calculating the malonyl-CoA concentration using 15.6 cm¡1 mM¡1 as the absorbtivity coeYcient for coenzyme A at 258 nm. This procedure is accurate to within a few percent. However, in the commercial material, the UV response of the coenzyme A moiety is from not only the biologically relevant malonyl-CoA but also the contaminant 20-phospho-malonyl-CoA. In the instance of lot 052K7029, at 9–13% contamination, we felt that it was necessary to purify the malonyl-CoA. Using the described chromatographic system the contamination is not detrimental because the chromatographic system easily resolves these two species (Fig. 2). However, if the two malonyl-CoA isomers are not resolved or poorly resolved, depending on the percent of contamination, the result could negatively inXuence the accuracy of an HPLC analysis procedure. As a substrate in enzymatic assays, the contribution of 20phospho-malonyl-CoA is not known. However, as an inhibitor of carnitine palmitoyltransferase-I, we have found that the eVect of the 20-phospho-malonyl-CoA depends on the isoform of that enzyme (unpublished). In summary, we have characterized a contaminant in commercial malonyl-CoA and, with NMR spectroscopy, unequivocally identiWed it as 20-phospho-malonyl-CoA. We have also described a preparative HPLC procedure for puriWcation of milligram amounts of the naturally occurring 30-phospho-malonyl-CoA from the contaminant 20-phospho-malonyl-CoA. This procedure can be used to prepare puriWed malonyl-CoA for analytical methods that require highly pure material when the
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
contaminant content of the commercial malonyl-CoA is beyond an appropriate threshold.
Acknowledgments We thank Stephen T. Ingalls for reviewing the manuscript and oVering helpful suggestions. This work is supported in part by NIH P01 AG15885, NSF MCB0110610, and VA Medical Research Service. Part of the data in this paper was presented in a late-breaking poster at the Experimental Biology 2003 meeting in San Diego, CA, April 2003.
References [1] S.E. Mills, D.W. Foster, J.D. McGarry, Interaction of malonylCoA and related compounds with mitochondria from diVerent rat tissues. Relationship between ligand binding and inhibition of carnitine palmitoyl transferase I, Biochem. J. 214 (1983) 83–91. [2] J.D. McGarry, G.P. Mannaerts, D.W. Foster, A possible role for malonyl-CoA in the regulation of hepatic fatty acid oxidation and ketogenesis, J. Clin. Invest. 60 (1977) 265–270. [3] N.B. Ruderman, A.K. Saha, D. Vavvas, L.A. Witters, Malonyl-CoA, fuel sensing, and insulin resistance, Am. J. Physiol. 276 (1999) E1–E16.
209
[4] V.A. Zammit, The malonyl-CoA-long-chain acyl-CoA axis in the maintenance of mammalian cell function, Biochem. J. 343 (Pt. 3) (1999) 505–515. [5] J.D. McGarry, M.J. Stark, D.W. Foster, Hepatic malonyl-CoA levels of fed, fasted and diabetic rats as measured using a simple radioisotopic assay, J. Biol. Chem. 253 (1978) 8291–8293. [6] K.R. Jonscher, J.R. Yates III, Review: the quadrupole ion trap mass spectrometer—a small solution to a big challenge, Anal. Biochem. 244 (1997) 1–15. [7] R.L. Dordine, P. Paneth, V.E. Anderson, 13C NMR and 1H-1H NOEs of coenzyme A: conformation of the pantoic acid moiety, Bioorg. Chem. 23 (2) (1995) 169–181. [8] J.J. Mieyal, L.T. Webster Jr., U.A. Siddiqui, Benzoyl and hydroxybenzoyl esters of coenzyme A. PuriWcation and nuclear magnetic resonance characterization; conformation in solution, J. Biol. Chem. 249 (8) (1974) 2633–2640. [9] B. Sedgwick, J.W. Cornforth, S.J. French, R.T. Gray, E. Kelstrup, P. Willadsen, The biosynthesis of long-chain fatty acids. Incorporation of radioactivity from stereospeciWcally tritiated malonyl thiol esters, and the stereochemistry of the acetyl-CoA carboxylase reaction, Eur. J. Biochem. 75 (2) (1977) 481–495. [10] M. Oivanen, R. Schnell, W. PXeiderer, H. Lonnberg, Interconversion and hydrolysis of monomethyl and monoisopropyl esters of adenosine 20- and 30-monophosphates: kinetics and mechanisms, J. Org. Chem. 56 (11) (1991) 3623–3628. [11] M. Komiyama, Y. Takeshige, Regioselective P-O(3 0) cleavage of 2 0,3 0-cyclic monophosphates of ribonucleosides catalyzed by - and -cyclodextrins, J. Org. Chem. 54 (20) (1989) 4936– 4939.
Isolation and identiWcation of two isomeric forms of malonyl-coenzyme A in commercial malonyl-coenzyme A Paul E. Minkler,a Vernon E. Anderson,b Nakul C. Maiti,b Janos Kerner,c and Charles L. Hoppela,d,¤ a
Medical Research Service, Louis Stokes Department of Veterans AVairs Medical Center, Cleveland, OH 44106, USA b Department of Biochemistry, Case Western Reserve University, Cleveland, OH 44106, USA c Department of Nutrition, Case Western Reserve University, Cleveland, OH 44106, USA d Department of Pharmacology and Medicine, Case Western Reserve University, Cleveland, OH 44106, USA Received 9 December 2003
Abstract Two isomers of malonyl-coenzyme A (malonyl-CoA) were detected in a commercial preparation of malonyl-CoA. These compounds were separated by preparative high-performance liquid chromatography (HPLC) and characterized by HPLC/ultraviolet (UV)/mass spectrometry. Both compounds had a UV absorbance maximum at 259–260 nm. Both compounds underwent negative electrospray ionization to produce a [M-H]¡quasi-molecular ion at m/z 852 and both compounds underwent collision-induced dissociation to produce a characteristic fragment at m/z 808, all consistent with the structure of malonyl-CoA. Nuclear magnetic resonance spectrometry showed that the two chromatographically distinguishable malonyl-CoAs are structural isomers: the major component is the naturally occurring malonyl-CoA and the contaminant is 30-dephospho-20-phospho-coenzyme A. 2004 Elsevier Inc. All rights reserved. Keywords: Malonyl-coenzyme A; HPLC; Electrospray ionization; Mass spectrometry; NMR
Malonyl-coenzyme A (malonyl-CoA)1 is the prime regulatory molecule in the control of cellular long-chain fatty acid oxidation due to its inhibition of mitochondrial carnitine palmitoyltransferase I [1–4]. The quantiWcation of malonyl-CoA in biological systems has contributed to a better understanding of its role in regulation of mitochondrial fatty acid oxidation [5]. Bench-top electrospray ionization/ion-trap mass spectrometers (capable of high-sensitivity full-scan MS and tandem MS) employed as detection devices following HPLC have proven to be a substantial technical
¤
Corresponding author. Fax: +216-229-8509. E-mail address: [email protected] (C.L. Hoppel). 1 Abbreviations used: malonyl-CoA, malonyl-coenzyme A; 20-AMP, adenosine 20-monophosphate; 30-AMP, adenosine 30-monophosphate; 50-AMP, adenosine 50-monophosphate; 20,50-ADP, adenosine 20,50-diphosphate sodium salt; 30,50-ADP, adenosine 30,50-diphosphate sodium salt; 30-dephoCoA, 30-dephospho-coenzyme A; 20-phospho-malonylCoA, 30-dephospho-20-phospho-coenzyme A. 0003-2697/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.01.015
innovation [6]. We applied this technology to the detection of malonyl-CoA but, in the course of developing an HPLC separation compatible with mass spectrometric detection, we observed what appeared to be a contaminant in a commercial preparation of malonyl-CoA. Further reWnement of the chromatographic conditions led to the clear chromatographic separation of the contaminant from malonyl-CoA. In this communication, we provide evidence that establishes the contaminant as an isomer of malonyl-CoA. To accomplish this, the two compounds were separated by preparative HPLC and characterized by HPLC/ UV/MS. While the two forms are chromatographically diVerentiated, they are indistinguishable by both UV spectrophotometry and mass spectrometric detection. NMR spectroscopy demonstrated conclusively that the two malonyl-CoA isomeric forms present in the commercial preparation are the naturally occurring malonyl-CoA and the contaminant 30-dephospho-20-phospho-coenzyme A (20-phospho-malonyl-CoA).
204
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
Materials and methods Chemicals and solvents Malonyl-CoA (lots 052K7029, 111K7026, and 122K 7019), adenosine 20-monophosphate (20-AMP), adenosine 30-monophosphate (30-AMP), adenosine 50-monophosphate (50-AMP), adenosine 20,50-diphosphate sodium salt (20,50-ADP), adenosine 30,50-diphosphate sodium salt (30,50ADP), 30-dephospho-coenzyme A (30-dephoCoA), and ammonium formate were purchased from Sigma–Aldrich (Milwaukee, WI). Methanol (HPLC-grade) was purchased from Fisher ScientiWc (Cleveland, OH). HPLCgrade water was prepared using a Milli-Q reagent water system (Millipore, Bedford, MA). The D2O used in NMR experiments was purchased from Cambridge Isotope Laboratories (Andover, MA). The ion-trap mass spectrometer’s electrospray source used nitrogen for both the sheath and the auxiliary gases provided from liquid nitrogen dewar tanks available through Praxair Distribution (North Royalton, OH). Helium (5.0 ultra-high purity), which was employed as both the damping and the collision gas, was purchased from Praxair. Preparative HPLC Preparative-scale puriWcation of malonyl-CoA was performed using a BioCAD SPRINT perfusion chromatography system (PerSeptive Biosystems, Cambridge, MA) conWgured to perform preparative-scale reversedphase HPLC. The column used was an Alltima C18LL 5, 250 £ 22-mm purchased from Alltech Associates, (DeerWeld, IL). The eluents used were A (95% 0.1 M ammonium formate/5% methanol) and B (50% 0.1 M ammonium formate/50% methanol). The instrument was programmed to create a binary gradient of 100% A to 80% A/20% B over 40 min. A 2-ml injection loop was installed in the manual injector. The pump Xow rate was 10 ml/min and detection was by UV at 254 nm. Fractions (0.5 min) were collected with a Model FC-205 fraction collector (Gilson, Middleton, WI).
preparative HPLC. The Xow rate was 0.5 ml/min. The HPLC diode array detector was set to collect a single wavelength (259 nm with a reference wavelength of 400 nm) and full-scan UV spectra from 190 to 400 nm. The mass spectrometer was an LCQ Deca ion-trap instrument purchased from ThermoFinningan (San Jose, CA). The instrument was operated using the electrospray ionization source in the negative ion mode with the following settings: sheath gas, 90 units; auxiliary gas, 30 units; spray voltage, 2 kV; capillary temperature, 350 °C; capillary voltage, ¡22 V. The mass spectrometer was programmed to generate data from two sequential scan events: (1) full-scan MS, from m/z 700 to 1000, and (2) full-scan MS/MS on 852.2 from m/z 230 to 1000 with a collision energy setting of 26% and an isolation width of m/z 3.0. 1
H NMR
NMR samples were taken to dryness under vacuum in a SpeedVac (Savant), redissolved in 0.5 ml of D 2O to remove exchangeable protons, evaporated, and redissolved in D2O. NMR spectra were obtained with a 600-MHz Varian Inova (Varian, Lake Forest, CA) spectrometer at ambient temperature.
Results and discussion Isolation of malonyl-CoA and 20-phospho-malonyl-CoA Fig. 1 shows the preparative-scale HPLC chromatogram of malonyl-CoA. This result was generated by dissolving 8.2 mg of malonyl-CoA lithium salt (Sigma lot 052K7029) in 100 l of eluent A and injecting this into the preparative HPLC. Two large peaks were observed
Analytical HPLC with UV spectrophotometric and mass spectrometric detection The analytical HPLC/UV/MS system consisted of an HP1100 series quaternary pump with on-line degasser, autosampler, column heater, and diode array detector purchased from Agilent (Wilmington, DE). The column used was a Hypersil BDS C18 3 100 £ 4-mm analytical column cartridge with an integrated precolumn (4 £ 4 mm) packed with 5 Hypersil BDS C18, also purchased from Agilent. The column heater was operated at 40 °C. The HPLC pump was programmed to create a binary gradient of 100% A to 50% A/50% B over 20 min, where eluents A and B were identical to those used for
Fig. 1. Chromatogram resulting from an injection of 8.2 mg of malonyl-CoA (Sigma lot 052K7029) into the preparative HPLC chromatograph. The conditions are described under Materials and methods. The two chromatographic peaks labeled Peak 1 (malonyl-CoA) and Peak 2 (20-phospho-malonyl-CoA) were collected and the material contained isolated.
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
(labeled Peak 1 and Peak 2 in Fig. 1). Peak 1 was collected between 23.5 and 27 min and Peak 2 was collected between 27.5 and 29.5 min. Both Peak 1 and Peak 2 fractions were lyophilized and the residues reconstituted in water to give Wnal concentrations of 2.7 mM for Peak 1 and 0.37 mM for Peak 2 (the concentrations were determined spectrophotometrically using 15.6 cm¡1 mM¡1 as the absorbtivity coeYcient for coenzyme A at 258 nm). UV spectrophotometric and mass spectrometric characterization of malonyl-CoA and 20-phospho-malonylCoA Peak 1 and Peak 2 solutions were then injected (20 l) into the analytical HPLC. The chromatograms for Peak 1 (light tracings) and Peak 2 (dark tracings) are displayed in Fig. 2. In this Wgure, the two chromatographic injection results are superimposed and each of three simultaneous chromatographic data sets are displayed: UV at 259 nm (top), full-scan MS from m/z 700 to 1000
205
(middle), and full-scan MS/MS on 852.2 from m/z 230 to 1000 (bottom). One chromatographic species is dominant in each chromatogram, and the spectral data associated with each of these peaks is displayed to the left (Peak 1) and right (Peak 2) of the dominant peak. The UV spectra absorption maximas for both Peak 1 and Peak 2 at 259–260 nm are consistent with the absorption maximum of coenzyme A. The electrospray ionization mass spectra are consistent with a malonyl-CoA [M-H]¡ quasi-molecular ion at m/z 852.2 (C24H37N7O19P3S D 852.107784 u). The collision-induced dissociation product ion (MS/MS) spectra with prominent signals at m/z 834.2 and 808.2 are consistent with the loss of H2O (C24H35N7O18P3S D 834.097219 u) or the loss of CO2 (C23H37N7O17P3S D 808.117955 u) and are mechanistically reasonable malonyl-CoA products. Our conclusions after examination of the Peak 1 and Peak 2 spectra (UV, MS, and MS/MS) are that (1) they are consistent with malonyl-CoA and (2) the two chromatographic peaks are indistinguishable by either UV spectroscopy or mass spectrometry.
Fig. 2. Chromatograms resulting from injections of the isolated Peak 1 material (malonyl-CoA) and Peak 2 material (20-phospho-malonyl-CoA) from Fig. 1 using the analytical HPLC/UV/MS system. The conditions are described under Materials and methods. The chromatograms from the injection of Peak 1 material are displayed with the lighter tracings and the chromatograms from the injection of Peak 2 material are displayed with the darker tracings. The three stacked chromatograms resulted from simultaneous collection of three data sets: UV at 259 nm (top), full-scan MS from 700 to 1000 m/z (middle), and full-scan MS/MS on 852.2 from 230 to 1000 m/z (bottom). Spectral data associated with each of these peaks is displayed to the left (Peak 1) and right (Peak 2) of the dominant chromatographic peak.
206
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
IdentiWcation of malonyl-CoA and 20-phospho-malonylCoA by NMR The 1H NMR spectra of Peak 1 and Peak 2 shown in Fig. 3 are both consistent with the presence of coenzyme A thiolesters with peaks attributed to adenine, pantetheine, and ribose all present [7]. The thiolmethylene resonance near 2.9 ppm indicates the presence of a thiolester [8]. The absence of any additional peak arising from the acyl moiety is consistent with the presence of a malonyl thiolester as the malonyl C2 protons readily exchange with the solvent D2O [9].
There are signiWcant diVerences between the 1H NMR spectra of Peak 1 and Peak 2, indicating that the two compounds separated by HPLC are diVerent and do not interconvert either on the NMR time scale or in the time that elapsed between isolation and acquisition of the 1H NMR spectra (i.e., days). These diVerences center on the resonances for the ribose protons and speciWcally for the 20 and 30 protons. In 1H NMR spectra of coenzyme A and common coenzyme A thiolesters, these protons have resonances at 4.65–4.75 ppm and have been unequivocally assigned both by 2D COSY spectra and by the 3bond coupling of the 30 1H to the 31P [7]. These features
Fig. 3. 1H NMR spectra. (Bottom) Peak 1 (malonyl-CoA), (Top) Peak 2 (20-phospho-malonyl-CoA). The conditions are described under Materials and methods. Numerical labels indicate identities of protons.
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
207
Fig. 4. 1H NMR spectra. (Bottom) 30,50-ADP. (Top) 20,50-ADP. The conditions are described under Materials and methods. Numerical labels indicate identities of protons. The “*” labeled signal is the HOD peak.
are shared by the 1H NMR spectrum of 30,50-ADP shown in Fig. 4. Phosphorylation of a ribose hydroxyl group has two eVects on the 1H NMR spectrum. It results in a deshielding of »0.3 ppm, as reXected by the comparisons in Table 1, and an additional splitting due to the 3JHP of »7 Hz. These two features are observed in the diVerence between the 1H NMR spectra of Peak 2 and Peak 1 shown in Fig. 3. Transferring the phosphate from the 30hydroxyl to the 20-hydroxyl results in a »0.2-ppm deshielding of the 20 1H and a corresponding increase in shielding of the 30 1H so that they appear at 4.90 and 4.24 ppm, respectively. This isomerization of the phosphate position is mirrored in the diVerences between the standard 20,50-ADP and the 30,50-ADP spectra shown in
Fig. 4, conWrming the assignment. Therefore, the two chromatographically distinguishable species of malonylCoA present in the commercial preparation (identical by both UV and MS) are unequivocally shown by NMR to be the naturally occurring malonyl-CoA (Peak 1) and the contaminant 20-phospho-malonyl-CoA (Peak 2). Speculation on the origin of the 20-phospho-malonyl-CoA contamination The Wrst lot of malonyl-CoA that we obtained (lot 052K7029), whose content launched this investigation, was determined to have 12–13% 20-phospho-malonyl-CoA by UV detection using either preparative or
208
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
Table 1 NMR data for standards vs malonyl-CoA (Peak 1) and 20-phospho-malonyl-CoA (Peak 2) Compound
Chemical shifts (ppm)
Malonyl-CoA (Peak 1) 30,50-ADP 30-AMP 30-Dephospho-CoA 20-Phospho-malonyl-CoA (Peak 2) 20,50-ADP 20-AMP
10H
20H
30H
40 H
50H
6.014 6.025 5.955 6.049 6.097 6.098 6.114
4.701 4.669 4.662 4.655/4.646 4.902 4.921 5.042
4.665 4.657 4.564 4.269 4.240 4.247 4.164
4.428 4.392 4.280 4.414 4.478 4.465 4.427
4.073 3.980 3.746 4.105 4.087, 4.057 3.946 3.749
Table 2 Percent of 20-phospho-malonyl-CoA found in commercial malonyl-CoA Malonyl-CoA lot No.
HPLC detection method
Amount injected
20-Phospho-malonyl-CoAa (%)
052K7029 052K7029 052K7029 111K7026 111K7026 122K7019 122K7019
UV from preparative HPLC UV from analytical HPLC MS/MS from analytical HPLC UV from analytical HPLC MS/MS from analytical HPLC UV from analytical HPLC MS/MS from analytical HPLC
8.2 mg; N D 1 0.2 nmol; N D 1 0.2 nmol; N D 1 6 nmol; N D 10 0.2 nmol; N D 10 6 nmol; N D 10 0.2 nmol; N D 10
12.8 12.5 9.1 6.0 § 0.3b 3.1 § 0.3b 4.7 § 0.3b 2.4 § 0.3b
a b
Determined by dividing the MS/MS or UV peak areas of HPLC Peak 2 by the sum of the peak areas of Peak 1 + Peak 2 times 100%. Mean § standard deviation.
analytical HPLC (Table 2). By analytical HPLC with MS/MS detection the value was 9%. We have subsequently purchased two additional lots of malonyl-CoA from the same vendor (lots 111K7026 and 122K7019) and examined them for 20-phospho-malonyl-CoA content and found lower, although still detectable, amounts of 20-phospho-malonyl-CoA (Table 2). Since the migration of the ribose 30-phosphate to the 20-phosphate does not occur at a signiWcant rate under physiological conditions, the appearance of the 20-phospho-malonyl-CoA contaminant presumably arises during synthesis. Acid catalyzes the phosphate migration, and acid quenches used in coenzyme A thiolester syntheses may promote isomerization [10]. A second mechanism for isomerization could come from activating the phosphate group [11]. The activation of the phosphate may result directly from reaction with the activated acyl group or by reaction with residual activating reagent if the activated acyl group is not isolated prior to addition to the coenzyme A. In the presence of activating reagent, the 30-phosphate is activated and displaced by the 20-hydroxyl group to form the 20,30 cyclic phosphodiester, which hydrolyzes to give both 20- and 30-phosphate esters. SigniWcance of contaminant The signiWcance of a contaminant in a reagent is a function of the combined eVects of the amount of contamination and the perturbation that the contaminant has on the system being used. For us, a primary issue with the presence of 20-phospho-malonyl-CoA in the commercial malonyl-CoA is the accuracy of preparing
standardized solutions. We standardize solutions of malonyl-CoA by determining the UV contribution of coenzyme A at 258 nm and then calculating the malonyl-CoA concentration using 15.6 cm¡1 mM¡1 as the absorbtivity coeYcient for coenzyme A at 258 nm. This procedure is accurate to within a few percent. However, in the commercial material, the UV response of the coenzyme A moiety is from not only the biologically relevant malonyl-CoA but also the contaminant 20-phospho-malonyl-CoA. In the instance of lot 052K7029, at 9–13% contamination, we felt that it was necessary to purify the malonyl-CoA. Using the described chromatographic system the contamination is not detrimental because the chromatographic system easily resolves these two species (Fig. 2). However, if the two malonyl-CoA isomers are not resolved or poorly resolved, depending on the percent of contamination, the result could negatively inXuence the accuracy of an HPLC analysis procedure. As a substrate in enzymatic assays, the contribution of 20phospho-malonyl-CoA is not known. However, as an inhibitor of carnitine palmitoyltransferase-I, we have found that the eVect of the 20-phospho-malonyl-CoA depends on the isoform of that enzyme (unpublished). In summary, we have characterized a contaminant in commercial malonyl-CoA and, with NMR spectroscopy, unequivocally identiWed it as 20-phospho-malonyl-CoA. We have also described a preparative HPLC procedure for puriWcation of milligram amounts of the naturally occurring 30-phospho-malonyl-CoA from the contaminant 20-phospho-malonyl-CoA. This procedure can be used to prepare puriWed malonyl-CoA for analytical methods that require highly pure material when the
P.E. Minkler et al. / Analytical Biochemistry 328 (2004) 203–209
contaminant content of the commercial malonyl-CoA is beyond an appropriate threshold.
Acknowledgments We thank Stephen T. Ingalls for reviewing the manuscript and oVering helpful suggestions. This work is supported in part by NIH P01 AG15885, NSF MCB0110610, and VA Medical Research Service. Part of the data in this paper was presented in a late-breaking poster at the Experimental Biology 2003 meeting in San Diego, CA, April 2003.
References [1] S.E. Mills, D.W. Foster, J.D. McGarry, Interaction of malonylCoA and related compounds with mitochondria from diVerent rat tissues. Relationship between ligand binding and inhibition of carnitine palmitoyl transferase I, Biochem. J. 214 (1983) 83–91. [2] J.D. McGarry, G.P. Mannaerts, D.W. Foster, A possible role for malonyl-CoA in the regulation of hepatic fatty acid oxidation and ketogenesis, J. Clin. Invest. 60 (1977) 265–270. [3] N.B. Ruderman, A.K. Saha, D. Vavvas, L.A. Witters, Malonyl-CoA, fuel sensing, and insulin resistance, Am. J. Physiol. 276 (1999) E1–E16.
209
[4] V.A. Zammit, The malonyl-CoA-long-chain acyl-CoA axis in the maintenance of mammalian cell function, Biochem. J. 343 (Pt. 3) (1999) 505–515. [5] J.D. McGarry, M.J. Stark, D.W. Foster, Hepatic malonyl-CoA levels of fed, fasted and diabetic rats as measured using a simple radioisotopic assay, J. Biol. Chem. 253 (1978) 8291–8293. [6] K.R. Jonscher, J.R. Yates III, Review: the quadrupole ion trap mass spectrometer—a small solution to a big challenge, Anal. Biochem. 244 (1997) 1–15. [7] R.L. Dordine, P. Paneth, V.E. Anderson, 13C NMR and 1H-1H NOEs of coenzyme A: conformation of the pantoic acid moiety, Bioorg. Chem. 23 (2) (1995) 169–181. [8] J.J. Mieyal, L.T. Webster Jr., U.A. Siddiqui, Benzoyl and hydroxybenzoyl esters of coenzyme A. PuriWcation and nuclear magnetic resonance characterization; conformation in solution, J. Biol. Chem. 249 (8) (1974) 2633–2640. [9] B. Sedgwick, J.W. Cornforth, S.J. French, R.T. Gray, E. Kelstrup, P. Willadsen, The biosynthesis of long-chain fatty acids. Incorporation of radioactivity from stereospeciWcally tritiated malonyl thiol esters, and the stereochemistry of the acetyl-CoA carboxylase reaction, Eur. J. Biochem. 75 (2) (1977) 481–495. [10] M. Oivanen, R. Schnell, W. PXeiderer, H. Lonnberg, Interconversion and hydrolysis of monomethyl and monoisopropyl esters of adenosine 20- and 30-monophosphates: kinetics and mechanisms, J. Org. Chem. 56 (11) (1991) 3623–3628. [11] M. Komiyama, Y. Takeshige, Regioselective P-O(3 0) cleavage of 2 0,3 0-cyclic monophosphates of ribonucleosides catalyzed by - and -cyclodextrins, J. Org. Chem. 54 (20) (1989) 4936– 4939.