Anal. Chem. 2005, 77, 6737-6740
Discovery of Enzymatic Activity Using Stable Isotope Metabolite Labeling and Liquid Chromatography-Mass Spectrometry Joseph J. Dalluge,*,† Hans Liao,‡ Ravi Gokarn,‡ and Holly Jessen‡
Cargill Scientific Resources Center and Biotechnology Development Center, Cargill Incorporated, P.O. 5702, Minneapolis, Minnesota 55440-5702
Stable isotope labeling of an intracellular chemical precursor or metabolite allows direct detection of downstream metabolites of that precursor, arising from novel enzymatic activity of interest, using metabolite profiling liquid chromatography-mass spectrometry. This approach allows the discrimination of downstream metabolites produced from novel enzymatic activity from the unlabeled form of the metabolite arising from native metabolic processes within the cell. Even for the case in which a given product of novel enzymatic activity is a transient, the novel enzymatic activity that produced it can be demonstrated to exist indirectly by identification of product-specific downstream metabolites. Therefore, direct or indirect discovery of novel enzymatic machinery engineered within a cell can be accomplished without a requirement for direct product purification or identification. The application of this approach to confirm the presence of a novel metabolic activity, alanine 2,3-aminomutase, obtained by mutagenesis and selection are discussed. The advantages of metabolite profiling approaches to metabolic engineering in terms of accelerating enzyme discovery and development of intellectual property will also be highlighted. Metabolic engineering is a rapidly developing technology with significant potential to impact the long-term sustainability of economic systems via the production of biobased industrial products.1-3 Current interest in pathway engineering toward the commercial production of chemicals, materials, and biomolecules necessitates the development of analytical strategies that expedite both the discovery and characterization of novel metabolic activities, as well as establishment of intellectual property surrounding these activities. This article describes a rapid and sensitive method for identifying novel enzymatic activity using stable isotope metabolite * Corresponding author. E-mail:
[email protected]. Phone: 952-742-3038. Fax: 952-742-3010. † Cargill Scientific Resources Center. ‡ Biotechnology Development Center. (1) Bailey, J. E. Science 1991, 252, 1668-1675. (2) Nakamura, C. E.; Whited, G. M. Curr. Opin. Biotechnol. 2003, 14, 454459. (3) Chotani, G.; Dodge, T.; Hsu, A.; Kumar, M.; LaDuca, R.; Trimbur, D.; Weyler, W.; Sanford, K. Biochim. Biophys. Acta 2000, 1543, 434-455. 10.1021/ac051109y CCC: $30.25 Published on Web 09/21/2005
© 2005 American Chemical Society
labeling and liquid chromatography-mass spectrometry. By labeling an intracellular chemical precursor or metabolite with 13C-stable isotope labels, downstream metabolites of that precursor, arising from novel enzymatic activity of interest, may be detected directly in cell extracts. The application of this approach to confirm the presence of alanine 2,3-aminomutase, a novel enzymatic activity obtained by mutagenesis and selection, is presented here. EXPERIMENTAL SECTION Reagents. Coenzyme A was purchased from Sigma Chemical Co. (St. Louis, MO). Trifluoroacetic acid, ammonium acetate, and acetic acid were purchased from J. T. Baker (Phillipsburg, NJ). Acetonitrile was purchased from EM Science (Gibbstown, NJ). HPLC-grade water was purchased from Fisher Scientific (Fair Lawn, NJ). Construction of Escherichia coli ∆panD::CAT. E. coli ∆panD::CAT was constructed using the gene inactivation method of Datsenko and Wanner4 using E. coli strains BW25113/pKD46 and BW 25141/pKD3 from the E. coli Genetic Stock Center. The CAT gene of pKD3 was amplified using primers 5′-TATCAATTCGTTACAGGCGATACATGGCACGCTTCGGCGCGTGTAGGCTGGAGCTGCTTC and 5′-GATGTCGCGGCTGGTGAGTAACCAGCCGCAGGGATAACAACATATGAATATCCTCCTTAG, where the underlined sequence corresponds to the regions in the E. coli chromosome immediately upstream and downstream of the panD locus, respectively, and the nonunderlined regions are homologous to regions in pKD3 that permit amplification of a fragment containing the CAT gene. The PCR product was transformed into BW25113/pKD46 expressing the λ recombination functions and transformants were recovered on LB plates containing 25 µg/mL chloramphenicol and 5 µM β-alanine. Chloramphenicol-resistant transformants were tested for retention of chloramphenicol resistance, loss of ampicillin resistance (indicating curing of pKD46), and requirement for β-alanine for growth on M9-glucose minimal medium. Confirmation of correct insertion of the CAT gene into the panD locus was carried out by colony PCR of the resultant ∆panD::CAT strain using primers that flank the insertion locus (5′-TTACCGAGCAGCGTTCAGAG; and 5′-CACCTGGCGGTGACAACCAT). While the wild(4) Datsenko, K. A.; Wanner, B. L. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 66406645.
Analytical Chemistry, Vol. 77, No. 20, October 15, 2005 6737
type panD locus is expected to yield a PCR product of 713 base pairs, the ∆panD::CAT construct yielded a 1215-base pair product. A derivative in which the inserted CAT gene was removed by the activity of the FLP recombinase encoded by plasmid pCP20 was constructed as described.4 A strain carrying a second deletion in the panC gene (encoding pantothenate synthetase) was constructed in a similar manner and the mutation transduced into the ∆panD strain. The ∆panD/∆panC strain is unable to form pantothenate from β-alanine and requires supplementation with pantothenate for growth. Cloning and Mutagenesis of the Bacillus subtilis Lysine 2,3-Aminomutase (KAM Gene). The B. subtilis KAM gene encoding lysine 2,3-aminomutase was cloned from ATCC 6051 using primers based on the sequence described5 (GCGCGAGGAGGAGTTCATATGAAAAACAAATGGTATAAAC and CGGGCACCGCTTCGAGGCGGCCGCACCATTCGCATG), where the underlined nucleotides are the NdeI and NotI sites used for cloning the PCR product into pPRONde, a derivative of pPROLar.A122 (Clontech Laboratories, Inc., Palo Alto, CA) in which an NdeI site was constructed at the initiator ATG codon by oligonucleotidedirected mutagenesis using the QuikChange Site-Directed Mutagenesis kit (Stratagene, Carlsbad, CA). In vitro mutagenesis was carried out using the methods of Xu et al.6 yielding libraries of >50 000 clones with an average mutation rate of 1.3 altered nucleotides/kb, or of Cadwell and Joyce,7 yielding an average mutation rate of 0-4.7 altered nucleotides/kb. Mutagenized plasmid was transformed into electrocompetent E. coli ∆panD, and transformants were plated on M9 minimal medium supplemented with 0.4% glucose and 25 µg/mL kanamycin to select for complementation of the ∆panD phenotype. Transformants growing on the minimal medium plates arose at a frequency of ∼1 × 10-4 relative to the number to total transformants growing on LB plus 25 µg/mL kanamycin, and plasmid DNA from these colonies was retransformed into the ∆panD host to confirm that the ability to grow in the absence of added β-alanine was conferred by a function carried by the plasmid. Plasmid DNA was prepared from retransformed colonies and the KAM gene sequenced to determine any changes relative to the wildtype B. subtilis KAM gene sequences. All putative alanine 2,3-aminomutase clones carried a B. subtilis KAM gene sequence with the substitutions L103M, M136V, and D339H (where the first amino acid is the wild-type sequence, the number is the amino acid position, and the second amino acid is the sequence observed in the alanine 2,3-aminomutase sequence). Biosynthesis of [13C]Coenzyme A from [3-13C]r-Alanine. Cells of E. coli ∆panD transformed with pPRO-Nde vector, or pPRO plasmid carrying wildtype or variant KAM, were grown overnight at 37 °C in M9 minimal medium with 25 µg/mL kanamycin, 10 µM Fe2(NH4)2(SO4)2, 1 mM L-alanine (unlabeled), and 10 µM β-alanine. The cultures were diluted 100-fold in minimal medium with 25 µg/mL kanamycin, 10 µM Fe2(NH4)2(SO4)2, and 11 mM [3-13C]alanine (99%, Cambridge Isotope Laboratories, Andover, MA) but no unlabeled L-alanine or β-alanine. Following growth at 30 °C for ∼20 h, the cells were recovered by centrifuga(5) Chen, D.; Ruzicka, F. J.; Frey, P. A. Biochem. J. 2000, 348, 539-549. (6) Xu, H.; Pettersen, E. I.; Petersen, S. B.; El-Gewely, M. R. BioTechniques 1999, 27, 1102-1108. (7) Cadwell, R. C.; and Joyce, G. F. PCR Methods Appl. 1992, 2, 28-33.
6738
Analytical Chemistry, Vol. 77, No. 20, October 15, 2005
Scheme 1. Engineered Pathway (Boldface) for 3-Hydroxypropionic Acid Production
tion and extracted8 in the presence of 10 mM dithiothreitol to convert thioesters of coenzyme A to the free sulfhydryl form. Directly Combined Liquid Chromatography-Electrospray Ionization-Tandem Mass Spectrometry. Determination of coenzyme A in standards and cell extracts was carried out as previously described9 using a Waters/Micromass liquid chromatography-tandem mass spectrometry (LC/MS/MS) system consisting of a Waters 2795 liquid chromatograph directly coupled to a Micromass Quattro Ultima triple quadrupole mass spectrometer. LC separations were made using a 4.6 × 150 mm YMC ODS-AQ (3-µm particles, 120-Å pores) reversed-phase chromatography column at room temperature. The LC mobile phase consisted of (A) 25 mM ammonium acetate containing 0.5% acetic acid and (B) acetonitrile containing 0.5% acetic acid. The elution conditions were as follows: isocratic at 10% B, 0-10 min, linear gradient from 10% B to 60% B, 10-12 min, isocratic at 60% B, 1213 min, linear from 60% B to 95% B, 13-14 min, isocratic at 95% B, 14-17 min, linear from 95% B to 10% B, and 17-18 min, with a 7-min reequilibration period between runs. The flow rate was 0.25 mL/min. All parameters of the ESI-MS/MS system were optimized and selected based on optimal in-source generation of the protonated molecular ions of coenzyme A, as well as production of coenzyme A-specific fragment ions. The following instrumental parameters were used for LC/MS analysis of coenzyme A in the positive ion mode: capillary, 3.5 kV; cone, 20 V; Hex 1, 15 V; aperture, 1 V; Hex 2, 0 V; source temperature, 100 °C; desolvation temperature, 350 °C; desolvation gas, 500 L/h; cone gas, 40 L/h; low mass resolution (Q1), 15.0; high mass resolution (Q1), 15.0; ion energy, 0.2; entrance, 50 V; collision energy, 2; exit, 50 V; low mass resolution (Q2), 15; high mass resolution (Q2), 15; ion energy (Q2), 0.5; multiplier, 650. For identification of the daughter ions of coenzyme A, MS/MS analysis was performed using the following instrumental parameters: capillary: 3.5 kV; cone: 20 V; Hex 1: 15 V; aperture: 1 V; Hex 2: 0 V; source temperature, 100 °C; desolvation temperature, 350 °C; desolvation gas, 500 L/h; cone gas, 40 L/h; low mass resolution (Q1), 12.0; high mass resolution (Q1), 12.0; ion energy, (8) Jackowski, S.; Rock, C. O. J. Bacteriol. 1981, 148, 926-932. (9) Dalluge, J. J.; Gort, S.; Hobson, R.; Selifonova, O.; Amore, F.; Gokarn, R. Anal. Bioanal. Chem. 2002, 374, 835-840.
0.2; entrance, -5 V; collision energy, 20; exit, 1 V; low mass resolution (Q2), 15; high mass resolution (Q2), 15; ion energy (Q2), 0.5; multiplier, 650. RESULTS AND DISCUSSION A proposed engineered metabolic pathway for the production of 3-hydroxypropionic acid (3-HP) from glucose is summarized in Scheme 1. In addition, native metabolic pathways for key intermediates are indicated, as these paths have impacted the discovery of alanine 2,3-aminomutase activity and subsequent optimization of the engineered route to 3-HP. Three of the four enzymatic activities (alanine aminotransferase, aminobutyrate aminotransferase, 3-hydroxybutyrate dehydrogenase) required for production of 3-HP were known to occur naturally, were cloned into the model production host, and their activities demonstrated by enzymatic assay and by measuring the enzymatic production of downstream products from upstream reactants fed to the engineered host. For example, the concerted activities of aminobutyrate aminotransferase and 3-hydroxybutyrate dehydrogenase in this pathway was demonstrated by feeding β-alanine to the host strain and monitoring production of the stable product 3-HP by LC/MS analysis of cell-free extracts. By contrast, alanine 2,3-aminomutase was unknown in nature previous to this report, although a similar activity, lysine 2,3-aminomutase has been reported.5 The general strategy for the engineering and discovery of alanine aminomutase was 3-fold: (a) random mutagenesis of lysine 2,3-aminomutase to shift substrate preference to R-alanine; (b) selection for mutants with putative alanine 2,3-aminomutase activity; and (c) verification of alanine 2,3-aminomutase activity by measuring cellular production of β-alanine from R-alanine using LC/MS. Parts a and b were accomplished as described in the Experimental Section, resulting in an enzyme variant with the substitutions L103M, M136V, and D339H (where the first amino acid is the wild-type lysine 2,3-aminomutase residue, the number is the amino acid position, and the second amino acid is the sequence observed in the variant with alanine 2,3-aminomutase activity). It is impossible to speculate on the effect these mutations would have on the structure of this enzyme as a detailed structural determination of the aminomutase enzyme has not been published at this time. The variant with alanine 2,3-aminomutase activity was identified by selection in a host carrying a deletion of the panD gene (∆panD, Scheme 1) and thus requiring supplementation with, or an alternative source of, β-alanine.10 Thus, ∆panD strains containing alanine 2,3-aminomutase would produce β-alanine and test positive for growth, whereas those without this novel activity would not grow without supplemental β-alanine. This mutagenesis and selection approach, while valuable for initial identification of promising mutants, cannot provide definitive evidence of the desired activity, and a more rigorous assay for detection of metabolites that are the product of this novel activity was necessary. As activity levels for the engineered aminomutase were so low as to prevent detection of β-alanine using in vitro enzymatic assays, we proposed the use of liquid chromatography-mass spectrometry to detect β-alanine production in ∆panD strains containing the putative alanine 2,3-aminomutase. To maximize accumulation (10) Kennedy, J.; Kealey, J. T. Anal. Biochem. 2004, 327, 91-94.
Figure 1. Identification of alanine 2,3-aminomutase activity in an engineered microbe. (A) Structure of coenzyme A and characteristic fragment ions observed in the positive ion electrospray mass spectrum of this metabolite. (B) Positive ion electrospray mass spectrum of native 12C-coenzyme A in a cell extract. (C) Positive ion electrospray mass spectrum of coenzyme A detected in ∆panD cells fed 13C-Ralanine and containing the putative alanine 2,3-aminomutase activity. Inset: MS/MS analysis of 13C-coenzyme A.
of β-alanine, the test strains, in addition to lacking panD, were ∆panC (Scheme 1) and did not contain the engineered activities for the production of 3-HP from β-alanine. This approach, however, was not successful for the direct identification of β-alanine in cell extracts. The lack of detectable β-alanine was possibly due to its metabolism by an unidentified branch metabolic pathway, such that it could not accumulate to a level greater than the detection limit of the analytical method (LOD ) 1 µM). As an alternative to the direct detection of β-alanine, the host strain was engineered in such a way as to allow the indirect determination of β-alanine production by measuring the natural downstream metabolite of this amino acid, coenzyme A11 (Scheme 1). To ensure that the coenzyme A measured was a result of the novel enzymatic activity in question and not the result of other native metabolic pathways, the following strategy, based on stable isotope metabolite labeling and LC/MS, was employed: (a) test strains were fed 13C-R-alanine; (b) test strains were ∆panD and contained the putative alanine (11) Jackowski, S. In Escherichia coli and Salmonella: Cellular and Molecular Biology; Niedhardt, F. C.; et al. Eds.; ASM Press: Washington DC, 1996; pp 687-694.
Analytical Chemistry, Vol. 77, No. 20, October 15, 2005
6739
Table 1. 13C-CoA/12C-CoA Ratios in Microbes
13C-r-Alanine-Fed
sample
13C-CoA/12C-CoA
std (theoretical ) 0.257) 12C-CoA std A (∆panD) B (∆panD) C (∆panD) D (∆panD) E (∆panD) F (∆panD/∆panC) G (∆panD/∆panC) H (∆panD/∆panC)
0.251
12C-CoA
0.254 3.57 3.03 5.98 9.42 4.01 0.232 0.278 0.264
2,3-aminomutase gene, such that production of 13C-β-alanine could only arise from the chemical conversion of 13C-R-alanine by the novel enzymatic activity of interest (Scheme 1); (c) test strains did not contain the enzymes to convert β-alanine to 3-HP (Scheme 1) but were panC+, so that any 13C-β-alanine produced would be metabolized to 13C-coenzyme A (Figure 1A). Detection of a significant enrichment of 13C-coenzyme A in cells fed 13C-R-alanine would provide proof of the novel alanine 2,3-aminomutase activity in the engineered microbe. Figure 1B illustrates a mass spectrum of natural coenzyme A in a cell extract, displaying a [M + H]+ ion at m/z 768, and two characteristic fragment ions9 at m/z 428 and 261, corresponding to the adenine nucleotide and pantetheine moieties, respectively (Figure 1A). By comparison, in ∆panD cells engineered with the putative alanine 2,3-aminomutase, and fed 13C-R-alanine, the predominant
6740
Analytical Chemistry, Vol. 77, No. 20, October 15, 2005
species detected is 13C-coenzyme A at m/z 769, suggesting that 13C-R-alanine is a precursor and providing support for the existence of alanine 2,3-aminomutase activity in these cells. The fact that the predominant daughter of m/z 769 is found at m/z 262 (Figure 1C), corresponding to the β-alanine/pantothenic acidcontaining region of coenzyme A provides additional support for this discovery. Ratios of detected 13C-coenzyme A/12C-coenzyme A demonstrating the enrichment of isotopically labeled CoA in 13C-R-alanine fed ∆panD cells are tabulated in Table 1. As a control to ensure that 13C-CoA was being produced via the conversion of 13C-R-alanine to 13C-β-alanine, and was not arising via a alternative metabolic pathway from 13C-R-alanine, the same experiment was run with a ∆panD/∆panC strain, and as expected, the 13C-CoA/ 12C-CoA ratio returns to a naturally occurring value (Table 1). CONCLUSIONS LC/MS with stable isotope metabolite labeling has been demonstrated to be a powerful tool for identifying a novel enzymatic activity (alanine 2,3-aminomutase) in metabolically engineered cells, allowing observation of novel activity even for the case in which the product is a transient metabolic intermediate. Finally, this technology allows the direct or indirect identification of novel enzymes without the need for direct characterization or purification, accelerating discovery and establishment of intellectual property. Received for review June 22, 2005. Accepted August 4, 2005. AC051109Y