Anal Bioanal Chem (2002) 374 : 835–840 DOI 10.1007/s00216-002-1554-x

O R I G I N A L PA P E R

Joseph J. Dalluge · Steven Gort · Russell Hobson · Olga Selifonova · Frank Amore · Ravi Gokarn

Separation and identification of organic acid-coenzyme A thioesters using liquid chromatography/electrospray ionization-mass spectrometry

Received: 18 June 2002 / Revised: 12 August 2002 / Accepted: 12 August 2002 / Published online: 8 October 2002 © Springer-Verlag 2002

Abstract A method has been developed for the direct determination of coenzyme A (CoA) and organic acid-CoA thioesters in mixtures using directly combined liquid chromatography/electrospray ionization-mass spectrometry (LC/ESI-MS). Mixtures of CoA and organic acid-CoA thioesters were analyzed by LC/ESI-MS with detection of protonated molecular ions and characteristic fragment ions for each compound. The identities of the CoAthioesters were established based on LC retention times and simultaneously recorded mass spectra. Monitoring of the CoA specific fragment ion at m/z 428 throughout the chromatogram provides a unique fingerprint for CoA content in the samples that corroborates the identification of organic acid-CoA thioesters in the mixtures. Furthermore, fragment ions arising from the ester linkage portion of the molecule allow unambiguous identification of the CoA esters in the samples. A second LC elution system was developed that allows the simultaneous separation and identification of 2-hydroxypropionyl-CoA (lactyl-CoA) and 3-hydroxypropionyl CoA (3HP-CoA), which have the same mass and identical MS fragmentation behavior. The utility of LC/ESI-MS employing this elution system is demonstrated by the determination of 3HP-CoA and lactyl-CoA (converted to CoA-thioesters from their corresponding free acids using CoA-transferase) in fermentation broths from Escherichia coli strains engineered for the production of 3-hydroxypropionic acid (3HP). External calibration employing a purified 3HP-CoA standard allowed indirect quantification of 3HP content in the broth with a precision of 1% (RSD). The feasibility of extending the method described above to perform LC/selected reaction monitoring-tandem mass spectrometry for direct determination of organic acid-CoA thioesters in cells was also demonstrated.

J.J. Dalluge (✉) · S. Gort · R. Hobson · O. Selifonova · F. Amore · R. Gokarn Cargill Central Research, Minneapolis, MN 55440–5702, USA e-mail: [email protected]

Keywords Liquid chromatography/mass spectrometry · Liquid chromatography/tandem mass spectrometry · Metabolite · Metabolomics · Metabolic engineering

Introduction Coenzyme A (CoA) and CoA-thioesters are essential substrates involved in amino acid, lipid, and carbohydrate metabolisms. It is estimated that approximately 4% of all known enzymes require CoA or CoA-esters as substrates [1]. Because these compounds enable enzymes to catalyze stereoselective carbon-carbon bond formation (e.g. fatty acid, polyketide biosynthesis), they are potentially useful for synthesizing complex organic molecules. Furthermore, biochemical manipulation and strategic engineering of the metabolic pathways in which CoA-thioesters play a role could provide a unique platform for large-scale production of industrially useful organic acids. Development of a rapid, sensitive, and selective means of separating and identifying CoA and CoA-esters is therefore essential to support ongoing research in these areas. Liquid chromatography (LC) with ultraviolet (UV) absorbance detection has been the analytical method of choice for determination of CoA and CoA-esters in complex mixtures [2, 3, 4]. UV detection, however, cannot provide molecular mass or structural information to verify identification of individual organic acid-CoA thioesters, and is confounded by a lack of selectivity for identification of these analytes in complex mixtures. Furthermore, a lack of pure standards for some of these analytes makes mass spectral information critical for the verification of observed UV peaks. Combined LC-continuous flow fast atom bombardment (FAB)-mass spectrometry (MS) has been reported and overcomes the limitations of LC/UV while providing “soft” ionization of the very labile CoA esters that cannot be measured adequately using gas chromatography, thermospray MS, or electron ionization MS [5]. LC/continuous flow FAB-MS, although effective, depends on complex in-house designed instrumentation that is expensive, unreliable, and inaccessible to the non-ex-

836

pert. Finally, reported methods for the determination of CoA-thioesters, including LC/continuous flow FAB-MS, fail to demonstrate the separation of structural isomers such as 3-hydroxypropionyl CoA (3HP-CoA) and 2-hydroxypropionyl-CoA (lactyl-CoA). To overcome the limitations of previous methods, a method based on directly combined LC/electrospray ionization (ESI)-MS has been developed as a simple and robust method for the efficient determination of these compounds, including the simultaneous determination of 3HP-CoA and lactyl-CoA derived from fermentation broths of bacteria engineered for production of 3-hydroxypropionic acid (3HP). Finally, development of new analytical methods for direct determination of intracellular metabolites such as CoA-thioesters [6] will have a significant impact on the fields of microbiology, biomedical research, and metabolic engineering, by enabling improved process and strain development, rapid and accurate metabolic flux analysis, and monitoring of cellular response to changes in environment. Toward this end, the method described herein was extended to perform LC/selected reaction monitoringtandem MS (LC/SRM-MS-MS) to demonstrate the feasibility of direct determination of organic acid-CoA thioesters in cells.

Materials and methods Chemicals. CoA, acetyl-CoA, and propionyl-CoA were purchased from Sigma (St. Louis, Mo., USA), trifluoroacetic acid (TFA), ammonium acetate, and acetic acid were purchased from J.T. Baker (Phillipsburg, N.J., USA), triethylamine was purchased from Fisher (Fair Lawn, N.J., USA), and acetonitrile (ACN) was purchased from EM Science (Gibbstown, N.J., USA). HPLC-grade water (18 mΩ), prepared using a Millipore Simplicity purification system (Millipore, Bedford, Mass.), was used to prepare all solutions. Synthesis of organic acid-CoA thioesters. Acrylyl-CoA, lactyl-CoA and 3HP-CoA were synthesized enzymatically according to the method of Selmer et al. [7]. The reaction was carried out in 50 mM potassium phosphate buffer (pH 7.0) with 1 mM acetyl-CoA and 100 mM of sodium acrylate, lithium lactate, or sodium 3-hydroxypropionate. Purified propionyl-CoA transferase from Megasphaera elsdenii was added to a final concentration of 0.05 mg/mL. The reaction was incubated at room temperature for 15 min. TFA was added to a final concentration of 0.1% (v/v). The solution was purified using a Sep-Pak cartridge (Waters, Milford, Mass.). The cartridge was conditioned with methanol and washed twice with 0.1% TFA. The sample was then applied to the cartridge and the cartridge was washed twice more with 0.1% TFA. The sample was eluted with 40% ACN, 0.1% TFA. The ACN was removed from the sample by vacuum centrifugation. Preparation of CoA/organic acid-CoA thioester standards. Standard stock solutions of CoA and the organic acid-CoA thioesters (1 mg/mL), and a solution containing 0.1 µg/µL of each of the six components in water were prepared and used for methods development. The 3HP-CoA standard was purified further employing a Waters Fractionlynx-controlled preparative LC/ESI-MS with automated fraction collection based on the mass of 3HP-CoA. Isolated 3HP-CoA (≥98% purity) was dried, weighed, and reconstituted in water to a stock concentration of 0.9 mg/mL. Growth of 3-hydroxypropionate fermenting strains. A saturated culture of ALS484 (lambdaDE3) with the plasmid encoding genes

for 3HP production was diluted 1:100 and 50 µL was used to inoculate 100 mL of media in a sealed serum bottle. The medium used in the fermentation was Luria broth (LB) supplemented with 50 µg/mL carbenecillin, 100 mM MOPS, 0.5 g/L cysteine, 2 mM KH2PO4, 10 mM Na2HPO4, 15 mM NaCl, 17 µM CaCl2, 10 µM MgCl2, 5 µM MnCl2, 0.4 µM CoCl2, 30 µM (NH4)2SO4, 1 µM FeSO4 and either 0.4% or 0.8% glucose (v/v). The sealed serum bottle was incubated overnight at 37 °C. After the overnight incubation, the absorbance of the sample at 600 nm (OD600) was measured. The culture was diluted in fresh modified LB and glucose was added to a final concentration of 0.4% (v/v) in one sample, and 0.8% (v/v) in another. The diluted culture was incubated at 37 °C until the OD600 was between 0.5 and 0.6. At that time the cultures were induced by adding isopropyl β-D-thiogalacto-pyranoside to a final concentration of 100 µM. Aliquots of the culture were taken at various time points. The samples were centrifuged and the supernatant was filtered through a 0.45 µm Aerodisc 13 nylon syringe filter (Pall, Ann Arbor, Mich.). The filtrate was stored either on ice or at 4 °C. Conversion of organic acids to CoA thioesters. The filtrate above was used at a concentration of 50% in 100 mM potassium phosphate buffer, pH 7.0, 1 mM acetyl-CoA and 0.05 mg/mL purified transferase. The reaction was allowed to proceed for 20 min at room temperature. The reaction was stopped by adding 1 volume of 10% TFA. The sample was purified using Sep Pak Vac columns (Waters). The column was conditioned with methanol and washed twice with 0.1% TFA. The sample was then applied to the column and the column was washed twice with 0.1% TFA. The sample was eluted with 40% ACN, 0.1% TFA. The ACN was removed from the sample by vacuum centrifugation. LC/ESI-MS analysis. Analyses of the standard CoA/CoA-thioester mixtures and the CoA-thioester mixtures derived from fermentation broths were carried out using a Waters/Micromass ZQ LC/MS instrument consisting of a Waters 2690 liquid chromatograph with a Waters 996 photo-diode array absorbance monitor placed in series between the chromatograph and the single quadrupole mass spectrometer. LC separations were made using a 4.6×150 mm YMC ODS-AQ (3 µm particles, 120 D pores) reversed-phase chromatography column at room temperature. Two gradient elution systems were developed using different mobile phases for the separation of the CoA esters. These two systems are summarized in Table 1. Elution system 1 was developed to provide the most rapid and efficient separation of the five-component CoA/CoAthioester mixture (CoA, acetyl-CoA, lactyl-CoA, acrylyl-CoA, propionyl-CoA), whereas elution system 2 was developed to provide baseline separation of the structurally isomeric esters lactyl-

Table 1 Gradient elution systems for the separation of organic acid-coenzyme A thioesters. ACN Acetonitrile System

Buffer A

Buffer B

Gradient Time (min)

%B

1

25 mM ammonium acetate; 0.5% acetic acid

ACN, 0.5% acetic acid

0 40 42 47 50

10 40 100 100 10

2

25 mM ammonium acetate; 10 mM TEA; 0.5% acetic acid

ACN, 0.5% acetic acid

0 10 45 50 53 54

10 10 60 100 100 10

837 CoA and 3HP-CoA, in addition to separation of the remaining esters listed above. In all cases, the flow rate was 0.250 mL/min. All parameters of the electrospray MS system were optimized and selected based on generation of protonated molecular ions ([M+H]+) of the analytes of interest, and production of characteristic fragment ions. The following instrumental parameters were used for ESI-MS detection of CoA and organic acid-CoA thioesters in the positive ion mode: Capillary, 4.0 V; cone, 56 V; extractor, 1 V; RF lens, 0 V; source temperature, 100 °C; desolvation temperature, 300 °C; desolvation gas, 500 L/h; cone gas, 40 L/h; low mass resolution, 13.0; high mass resolution, 14.5; ion energy, 0.5; multiplier, 650. Uncertainties for reported m/z and molecular masses are ±0.01%. Quantification of 3HP-CoA in an Escherichia coli fermentation broth. Four calibration solutions were prepared for 3HP-CoA to achieve four different concentrations in the solutions. These standards were analyzed by LC/ESI-MS, and the peak areas corresponding to 3HP-CoA were calculated automatically using the data analysis software. The resulting data were subjected to a linear least squares analysis to generate an external calibration curve for quantification of 3HP-CoA in a fermentation broth. Organic acids (including 3HP) in the fermentation broth supernatant of E. coli engineered for 3HP production were converted to CoA-thioesters as described above (near 100% conversion), and determined by LC/ESI-MS. 3HP-CoA was measured in triplicate and quantified using the external calibration. LC/SRM-MS-MS for identification of lactyl-CoA in cell-free extracts. SRM experiments for determination of lactyl-CoA in cellfree extracts were performed on a Micromass Ultima triple quadrupole mass spectrometer. LC separations were performed using gradient elution system 1 (Table 1) with the mass spectrometer set to detect the m/z=840 to m/z=333 transition specific to lactyl-CoA.

Results and discussion Examination of the general utility of LC/ESI-MS and LC/MS-MS for sensitive detection of organic acid-CoA Fig. 1A–F Liquid chromatography/electrospray ionizationmass spectrometry (LC/ESIMS) of coenzyme A (CoA) thioesters. A Total ion chromatogram illustrating the separation of CoA and four CoAorganic acid thioesters. 1 CoA, 2 2-hydroxypropionyl-CoA (lactyl-CoA), 3 acetyl-CoA, 4 acrylyl-CoA, 5 propionylCoA. B Mass spectrum of CoA. C Mass spectrum of lactyl-CoA. D Mass spectrum of acetyl-CoA. E Mass spectrum of acrylyl-CoA. F Mass spectrum of propionyl-CoA. Peaks labeled with asterisks were shown from their corresponding mass spectra not to be CoA-thioesters

thioesters in fermentation samples and cell-free extracts, and development of a method for the separation and identification of 3HP-CoA and lactyl-CoA in a complex matrix were the specific aims of this study. Determination of CoA and organic acid-CoA thioesters in a standard mixture using LC/ESI-MS The efficacy of the separation and detection of CoA thioesters using LC/ESI-MS was tested on a standard mixture of four organic acid-CoA thioesters and CoA. The results of this experiment are illustrated in Fig. 1. Figure 1A shows a total ion chromatogram for this analysis that demonstrates separation of the five-component mixture within 20 min utilizing LC elution system 1 (Table 1). The identities of the five components were established from LC retention times and from continuously recorded mass spectra. Characteristic mass spectra for each component in the mixture are illustrated in Fig. 1B–F. The mass spectra of these analytes include protonated molecular ions ([M+H]+), as well as fragment ions characteristic of CoA and CoA-thioesters. The structure of CoA is illustrated in Fig. 2. Regions of the molecule subject to cleavage within the ESI source, and m/z values for the resulting ions, are indicated. In the case of CoA, the [M+H]+ ion =768, and cleavage of CoA within the phosphate backbone region of the molecule (Fig. 2) results in two distinct ions at m/z=428 and m/z=261 (Fig. 1B). In the case of an organic acid-CoA thioester such as lactyl-CoA (Fig. 1C), the [M+H]+ is shifted to m/z=840, due to the attached lactyl moiety (72 Da). The characteristic fragment ion arising from the region of CoA where the lactyl group is

838

Fig. 2 Structure of CoA. Regions of the molecule subject to insource fragmentation, and the m/z values of the characteristic ions are indicated

attached also shifts by 72 from m/z=261 to m/z=333. The CoA-specific fragment ion at m/z=428, however, remains unchanged, as this fragment arises from the nonvariable region of the CoA structure. This fragmentation pattern is consistent for all CoA thioesters measured in this study, and allows unambiguous identification of these analytes in any mixture. Determination of 3HP-CoA and lactyl-CoA in a mixture containing both analytes

Fig. 3A, B LC/ESI-MS separation of 3-hydroxypropionyl CoA (3HP-CoA) (10.4 min) and lactyl-CoA (11.0 min). A Lactyl-CoA, selected ion chromatogram of m/z=840 and mass spectrum recorded under peak 2 (insert). B 3HP-CoA plus lactyl-CoA, selected ion chromatogram of m/z=840 and mass spectrum recorded under peak 1 (insert). Peak identification: 1 3HP-CoA, 2 lactyl-CoA. The peak labeled with an asterisk was shown from its corresponding mass spectrum not to be a CoA-thioester

The issue of separating 3HP-CoA from lactyl-CoA is essential from the standpoint of monitoring the production of 3HP (via its CoA-thioester) in bacteria where lactic acid (and subsequently lactyl-CoA) represents a major interferent. Because these structural isomers have identical masses and mass spectral fragmentation behavior, the separation and identification of these analytes in a mixture depends on their chromatographic separation. While elution system 1 (Fig. 1) provided excellent separation of the CoA thioesters tested, it was unable to resolve 3HP-CoA and lactyl-CoA. The chromatographic separation of 3HP-CoA and lactyl-CoA employing a phosphate buffer containing tetrabutylammonium hydrogen sulfate (TBAHS) has been reported, although no chromatograms illustrating the separation were presented [4]. Because both phosphate and TBAHS are nonvolatile, they cannot be employed for directly combined LC/MS which is needed to provide the added dimension of selectivity for identification of these compounds in complex matrixes. For this reason, an alternative LC elution system was developed to achieve this goal. In general, ammonium acetate represents a useful MS-compatible replacement for phosphate buffers in LC separations. And although it was not discussed in the prior report [4], TBAHS was probably employed as an ion-pairing reagent, and could be substituted by a more volatile amine such as triethylamine. Ammonium acetate and triethylamine were therefore used in the

839

development of elution system 2 (Table 1). The utility of this method for separation of 3HP-CoA (1) and lactyl-CoA (2) was tested on a mixture of these two compounds and the results are illustrated in Fig. 3. Figure 3A shows a selected ion chromatogram for m/z=840 in the analysis of lactyl-CoA (2) only. The mass spectrum recorded under peak 2 is displayed as an insert, and allows identification of this peak as the lactyl-CoA thioester (compare with Fig. 1C). Figure 3B illustrates the selected ion chromatogram for m/z=840 in the analysis of the 3HP-CoA/ lactyl-CoA mixture. The mass spectrum recorded under peak 1 of Fig. 3B is displayed as an insert. Upon comparison of Fig. 3A and Fig. 3B, the retention times of each component, and in consideration of the mass spectra corresponding to each peak, assignment of peak 1 as 3HP-CoA and peak 2 as lactyl-CoA was made. Finally, although elution system 2 has been demonstrated for the simultaneous determination of CoA and the five CoA-thioesters described herein, including the lactylCoA/3HP-CoA pair (data not shown), this system requires longer analysis times for determination of

Fig. 4A, B Production of 3-hydroxypropionic acid (3HP) in Escherichia coli. A LC/ESI-MS determination of organic acid-CoA thioesters derived from E. coli containing the plasmid. B Control: LC/ESI-MS determination of organic acid-CoA esters in E. coli without the insert. Peak identification: 1 3HP-CoA (simultaneously recorded mass spectrum illustrated as an insert to A. 2 Lactyl-CoA (simultaneously recorded mass spectrum illustrated as an insert to B. Peaks labeled with asterisks were shown from corresponding mass spectra not to be CoA-thioesters

acrylyl-CoA and propionyl-CoA, which elute in the range 25–27 min. LC/ESI-MS for monitoring production of 3HP in E. coli Initial efforts to develop a LC system for the separation of 3HP and lactic acid were unsuccessful. These efforts were subsequently shifted toward developing the LC/ESI-MS system for the separation of the corresponding CoA thioesters of these acids. In this way, production of the free acids in bacteria could be monitored by conversion to their corresponding CoA esters using a CoA transferase enzyme. An example of the utility of this method for monitoring production of 3HP in E. coli is given in Fig. 4. Figure 4A illustrates the selected ion chromatogram for m/z=840 in the analysis of a CoA transferase-treated fermentation broth aliquot collected from a culture of E. coli containing a plasmid with the insert that encodes propionyl-CoA transferase (1), lactyl-CoA dehydratase (2), and 3HP-CoA dehydratase (3) for production of 3HP from lactic acid. The mass spectrum corresponding to peak 1 is illustrated as an insert. As a control, a CoA transferasetreated fermentation aliquot from a culture of the same E. coli strain that did not contain the insert encoding enzymes 1, 2, and 3 was analyzed by LC/ESI-MS. This result is shown in Fig. 4B. Comparison of Fig. 4A with Fig. 4B allows the unambiguous assignment of peak 1 as 3HP-CoA and peak 2 as lactyl-CoA, indicating successful production of 3HP in E. coli. An external standard curve for 3HP-CoA measurement is illustrated in Fig. 5. The concentrations of standards used were chosen to span the range of 3HP concentrations expected in the broth. This curve was used to quantify 3HP-CoA derived from the fermentation broth of E. coli containing the insert for 3HP production. 3HP-CoA was measured in triplicate resulting in a measured 3HP concentration of 45 µM (±1%) in the

Fig. 5 External calibration curve for quantification of 3HP-CoA derived from fermentation of E. coli containing the plasmid insert for 3HP production

840

Fig. 6 Selected reaction monitoring of lactyl-CoA (1.5 µM, RT= 10.1 min) in a cell-free extract. The transition m/z=840 (M+H]+ of lactyl CoA) to m/z=333 (protonated lactyl-pantetheine fragment of lactyl-CoA) was monitored throughout the chromatogram

broth. The limit of detection for 3HP determination using this method is approximately 10 µM. Determination of lactyl-CoA in a cell-free extract using LC/MS-MS LC/MS-MS was used to demonstrate the feasibility of direct determination of lactyl-CoA in bacterial cells. For this purpose, lactyl-CoA was spiked into a cell-free extract of an organism that is unable to produce this metabolite. The first quadrupole mass analyzer of the triple quadrupole mass spectrometer was set to transmit only the protonated molecular ion of lactyl-CoA ([M+H]+m/z= 840). Argon was introduced into the second quadrupole (collision cell) to induce fragmentation of the parent species. The third quadrupole was set to detect a fragment ion specific for lactyl-CoA (m/z=333, the protonated lactylpantetheine moiety, see Fig. 2). This type of MS-MS experiment is termed selected reaction monitoring (SRM), and is the method of choice for maximizing selectivity and sensitivity in the determination of analytes from complex mixtures. The SRM chromatogram illustrating the identification of lactyl-CoA (1.5 µM) in a cell-free extract is shown in Fig. 6. This figure demonstrates the feasibility

of rapid and selective detection of intracellular organic acid-CoA metabolites with detection limits well below typical levels of such metabolites (typical range, 2–350 µM) [8] found in normal cells. The decreased retention of lactyl-CoA during this analysis compared to previous analyses (Fig. 1, Fig. 3, Fig. 4) may be attributed to the analysis being performed with a newer column on a different chromatographic system. MS and MS-MS analysis of the peak verify its identity as lactyl-CoA. In conclusion, the LC/ESI-MS and MS-MS methods described in this report should find widespread use for direct monitoring of organic acid-CoA thioester metabolites in cells for metabolic flux analysis, metabolomics, and strain and process development, as well as in indirect monitoring of organic acid production in bacteria or other hosts. Further, the sensitivity of the technique, together with appropriate standards, allows indirect quantification of organic acid production in these organisms, and should allow direct quantification of intracellular levels of CoA metabolites. Acknowledgements The authors acknowledge Dr. Wolfgang Buckel and Dr. Thorsten Selmer for their assistance in the purification of propionyl-CoA transferase.

References 1. Lee C-H, Chen AF (1982) Immobilized coenzymes and derivatives. In: Everse J, Anderson B, You K (eds) The pyridine nucleotide coenzymes. Academic, New York, p189 2. King MT, Reiss PD (1985) Anal Biochem 146:173–179 3. Patel SS, Walt DR (1987) J Biol Chem 262:7132–7134 4. Valentin HE, Steinbuchel A (1994) Appl Microbiol Biotechnol 40:699–709 5. Norwood DL, Bus CA, Millington DS (1990) J Chromatogr A 527:289–301 6. Buchholz A, Takors R, Wandrey C (2001) Anal Biochem 295: 129–137 7. Selmer T, Willanzheimer A, Hetzel M (2002) Eur J Biochem 269:1–9 8. Suzanne Jackowski (1996) Biosynthesis of pantothenic acid and coenzyme A. In: Neidhardt FC (ed.) Escherichia coli and Salmonella: cellular and molecular biology, vol. 1. ASM, Washington D.C., pp. 687–694