All plasmids and strains used in this work are listed in Table 1, and plasmid maps are provided in Additional file 1: Figure S1. We first tested conversion of TP to 3MX using a strain of E. coli that contained plasmid pAD1 [23]. Resting cells (OD600 = 2.5) converted approximately 0.3 mM TP to 3MX over 1 h, after which the reaction essentially stopped (Fig. 2). In order to increase activity, plasmids dAA, dDD, and dDA were added to the strain carrying pAD1, resulting in three new strains. These new strains allowed us to test the effect of different levels of ndmA and ndmD copy numbers on 3MX production (see Additional file 1: Table S2 for approximate gene copy numbers of each strain). Addition of ndmA only (strain pAD1dAA) had very little effect on activity (Fig. 2). Increasing the copy number of both genes (strain pAD1dDA) greatly increased the activity over strain pAD1dAA, with nearly complete conversion in 3 h. However, increasing the ndmD gene copy number only (strain pAD1dDD) resulted in complete conversion of TP within 2 h (Fig. 2). Strain pAD1dDD, which contained the lowest ndmA copy number, exhibited a slightly higher activity than did strain pAD1dDA, suggesting that plasmid pAD1 provided a sufficient ndmA gene dosage. These results also indicated that the reaction was limited by the amount of soluble NdmD produced inside the cells, since the activity increased with increasing ndmD copy number.
Read moreTable 1 Plasmids and strains used in this study Full size table Degradation of TP by metabolically engineered E. coli resting cells. Shaded triangle strain BL21(DE3) (negative control); Open circle strain pAD1; Shaded circle strain pAD1dAA; Open triangle strain dDA; Shaded square strain pAD1dDA; Open square strain pAD1dDD. Cells (OD600 = 2.5) were incubated with 1 mM TP in 50 mM KPi buffer at 30 °C with 400 rpm shaking, and metabolites were quantified via HPLC Full size image In the case of plasmid pAD1, the ndmD gene is separated from the T7 promoter by approximately 1.1 kb of sequence containing the ndmA ribosomal binding site and gene, followed by a short synthetic ribosomal binding site of unknown strength just before the ndmD gene (Additional file 1: Figure S1). SDS-PAGE of strain pAD1 (Additional file 1: Figure S2) showed a strong band of soluble NdmA, but very little NdmD (soluble or insoluble). In contrast, strain pAD1dDD had very strong soluble and insoluble NdmD bands. Based on activity and electrophoretic analysis, plasmid pAD1 clearly did not produce sufficient soluble intracellular NdmD. This was confirmed using resting cells (OD600 = 2.5) of an E. coli strain containing only plasmid dDA, which consumed 0.8 mM TP over 300 min (Fig. 2). Plasmid dDA is based on the pACYCDuet-1 backbone, giving a plasmid (and gene) copy number approximately fourfold lower than that of pAD1. In spite of the lower overall gene dosage, activity was much higher in strain dDA than in strains pAD1 and pAD1dAA. Methods to increase expression of ndmD from plasmid pAD1 only could involve using a known strong ribosomal binding site and/or a second T7 promoter between ndmA and ndmD. In order to increase intracellular levels of NdmD, a plasmid containing the ndmD gene placed immediately downstream of the T7 promoter and ribosomal binding site in pET28a(+) [43] was used. Compatible plasmids containing one or two copies of ndmA (plasmids dA and dAA, respectively) were then added to a strain of E. coli harboring pET28-His-ndmD. This resulted in strains with a low (pDdA) or medium (pDdAA) ndmA gene dosage, based on estimated copy number and number of genes in each plasmid. The activity and protein expression levels of these two strains were then compared with strain pAD1dDD, which had the highest ndmA dosage of the three (Additional file 1: Table S2). Strains pDdA, pDdAA, and pAD1dDD grew to a similar OD600 in 100 mL Luria–Bertani broth (LB) (Additional file 1: Table S3) when gene expression was induced as described in the “Methods” section. SDS-PAGE revealed that soluble (active) protein expression is about the same for NdmA and NdmD among the three strains (Additional file 1: Figure S2). Each wet cell paste was used to test the conversion of TP to 3MX by resuspending in KPi buffer to a final cell concentration of 30 mg/mL and initial TP concentration of 4 mM. After 90 min of the reaction time, TP was reduced 56, 51, and 43 % by suspensions of pDdA, pDdAA, and pAD1dDD, respectively. Approximately 84, 82, and 81 % of the consumed TP was converted to 3MX in strains pDdA, pDdAA, and pAD1dDD, respectively, with the remaining TP forming 1MX (Additional file 1: Table S3). Based on these results, strain pDdA was chosen for further studies due to the highest yield of 3MX from TP. Clearly, the additional gene dosage of ndmA (pDdAA) did not improve 3MX yield, relative to single gene dose (pDdA). Therefore, the activity of the cells was proven to be independent of the ndmA gene dosage and highly dependent on the ndmD gene dosage and expression in each E. coli strain. The effect of culture medium on cell growth and activity was also measured by growing strain pDdA in Luria–Bertani Lennox (LB) and super broth (SB) media. SB produced approximately 50 % more cells than did LB (Additional file 1: Table S4). Cells were resuspended to 15 mg/mL, and the initial TP concentration in activity assays was lowered to 1 mM in order to achieve complete conversion, which would facilitate downstream purification and product recovery. TP was completely consumed in SB-grown cells within 90 min (Fig. 3). After 2 h, nearly all of the TP was consumed in both reactions (Additional file 1: Table S4). 3MX yield from TP was 82–83 %, with an additional 12–13 % being 1MX. Because the cells are capable of performing both N
1- and N
3-demethylations on both TP and also 1- and 3MX, some small amount of xanthine was also formed from the monomethylxanthine products. These results demonstrate that the media composition had no significant effect on product ratio. Given the complete conversion of TP achieved in shorter time and 50 % more biocatalyst harvested from SB, this medium was chosen for the production of 3MX to supply clients. Production of methylxanthines from TP by strain pDdA grown in SB. 1 mM TP (open circle) was converted to 0.81 ± 0.002 mM 3MX (shaded circle) and 0.13 ± 0.002 1MX (open square) within 90 min by 15 mg/mL of resting cells. Concentrations reported are means with standard deviations of triplicate results Full size image Although yield of 3MX is high, minor production of 1MX decreases the overall yield of 3MX. The slight N
3-demethylation of TP by NdmA to form 1MX is surprising and in contrast with our previous findings that NdmA is highly specific for the N
1 methyl group of caffeine and TP [43]. We therefore tested the activity of strain pDdA on caffeine and observed a slight (<2 %) N
3-demethylation activity to form paraxanthine (1,7-dimethylxanthine, data not shown). The enzyme in the previously reported work was expressed in E. coli BL21(DE3) with a C-terminal hexahistidine (His6) tag for facile purification and assayed in vitro, and produced only 3MX from TP. 1MX was shown to be produced from TP by the highly-specific N
3-demethylase NdmB-His. The present study utilizes NdmA expressed in the same microbial chassis without the His6 tag, and the reaction is carried out in vivo. It is unclear whether performing the reaction in vivo, elimination of the His6 tag from NdmA, enzyme expression level, and/or enzyme solubility [46] are involved in the change in site specificity. In our in vitro studies, the minimum amount of enzyme was used in order to determine the kinetics [43], and the paraxanthine and 1MX products may have been below the detection limit. However, the reduction in enzyme expression level (comparing strains pAD1 and dDA vs. strain pDdA) in this work did not result in a lower ration of products. Clearly, an in vitro approach would not be economical, as it would require addition of external NADH. It should be noted, however, that addition of a His6 tag has been implicated in changing substrate specificity of the thioesterase I in E. coli due to a slight change in active site geometry [47]. The reason for the discrepancy between NdmA and NdmA-His6 is currently under investigation. The original strain of P. putida CBB5 produced approximately twice as much 3MX as 1MX [38], however, the 1MX production in this strain, besides slight specificity of NdmA at N
3-position, can mostly be attributed to NdmB [43]. Future work to reduce the N
3-demethylation activity of NdmA in vivo when expressed in E. coli should create a more efficient process for production of 3MX, while simultaneously simplifying downstream recovery of 3MX. The reaction conditions for conversion of TP to 3MX were optimized by evaluating different concentrations of cells (5, 10, 15, 30, and 60 mg wet cells/mL) and initial substrate concentration (1, 2, and 4 mM TP). It is clear from the data presented in Fig. 4 that a reaction containing 1 mM TP and 15 mg/mL resting cells provides linear conversion of TP to 3MX. At these reaction conditions, the product concentration and reaction volume suited the prep high pressure liquid chromatography (HPLC) column for complete product recovery. Goto Guangxing to know more. Effect of cell and substrate concentrations on 3MX production by E. coli pDdA. Resting cell assays were performed using 5 (open triangle), 10 (open square), 15 (open triangle), 30 (open diamond), and 60 (open circle) mg/mL wet cells. TP concentrations were 1 mM (a), 2 mM (b), and 4 mM (c). Concentrations of TP (left), 3MX (middle), and 1MX (right) are shown as means with standard deviations of triplicate reactions Full size image Production of 3MX was scaled up by growing the pDdA strain in SB media in four 2.8 L Fernbach flasks, resulting in a total yield of 20 g wet cells. The cell yield was sufficient to perform a 1.3 L reaction with an initial TP concentration of 1 mM at 15 mg/mL resting cell suspension. Initial analysis by HPLC showed complete consumption of TP after 2 h of reaction time with formation of 0.81 and 0.13 mM 3MX and 1MX, respectively. The products were separated by preparative chromatography (Additional file 1: Figure S3). Resolution of 3MX (retention time of 116 min) and 1MX (retention time of 136 min) was sufficient to collect each of the two products separately. The two products were crystallized through evaporation and freeze-drying, resulting in 106 mg 3MX and a minor amount of 1MX. Because the very small amount of 1MX produced could not be collected from the walls of the freeze dryer tray, 1MX was not further characterized. We are attempting to produce 1MX from TP via a metabolically engineered E. coli host containing ndmB and ndmD. The NdmB enzyme has been shown to be highly specific for N
3-demethylation [43], and a purified NdmB-His6 produced only 1MX in vitro. The theoretical amount of 3MX produced in the reaction was 175 mg (~81 % mole to mole conversion from TP); however 36 % of the post-reaction mixture was used to optimize the preparative chromatographic separation. Therefore, a total of 111 mg 3MX (64 % of the post-reaction mixture) was loaded onto the column for purification and recovery. The resulting 106 mg pure 3MX indicates very little loss during separation with a purification yield of 95.5 % after optimization of separation in the prep column. Improving the selectivity of NdmA so that it only produces 3MX from TP would further increase the yield. The reaction conditions described here could produce 135 mg/L 3MX. To our knowledge, this is the first report describing the non-transient microbial production of 3MX. Until now, all microbial production of 3MX has been as an intermediate in the caffeine and TP catabolic pathways [38, 48]. Therefore, there are no values in the literature with which to compare this yield. However, there was adequate amount for further analytical work and supply of samples to our clients. Because the ndm genes have only recently been discovered [43, 46], previous attempts to produce methylxanthines through a biocatalytic route have focused primarily on metabolism and enzymology studies for conversion of caffeine to theobromine. Research has shown that addition of certain divalent metal ions, such as Co2+, Ni2+, Cu2+, and Zn2+ have a strong inhibitory effect on degradation of theobromine accumulated from caffeine in whole cells of P. putida [49, 50]. However, there are no known specific inhibitors to stop the reaction at the desired, high-value methylxanthines such as paraxanthine and 1-, 3-, and 7-methylxanthine. Also, this approach would not be optimal for methylxanthine production, as the wild type P. putida strains (CBB5 and others) have lower growth rates and cannot produce the same amount of enzyme (hence, catalytic activity) as can E. coli expressing the recombinant ndm genes. Jin et al. [51] cloned genes from the caffeine biosynthetic pathway of coffee and tea into Saccharomyces cerevisiae. The resulting strain produced a very low level (0.38 mg/L) of caffeine when fed exogenous xanthosine. Without addition of xanthosine, no caffeine was detected. Besides the low production level, use of plant genes restricts the possible products to 7-methylxanthine, theobromine, and caffeine, which are the metabolites of the caffeine biosynthetic pathway. Caffeine is mostly produced during the decaffeination of coffee beans [52, 53]. Theobromine and TP are mostly produced synthetically [54, 55], although extraction from plants is possible [56]. Thus, further strain optimization and protein engineering will be required before use of plant-based genes can be used in a microbial system to produce high value methylxanthines. The purity of both 3MX and 1MX was analyzed by analytical HPLC using appropriate authentic standards. The retention time of the biologically produced products (Additional file 1: Figure S3) and authentic standards were identical. The High Resolution LC-MS spectrum of biologically produced and standard 3MX matched very well (Fig. 5) and corresponded to the published spectrum [57]. LC/MS was recorded on ESI positive mode; distinct M + 1 ion peak at 167.0569 m/z was observed both in the standard (Fig. 5a) and the biologically produced 3MX (Fig. 5b). LC–MS spectra of 3MX samples. a Spectrum of 3MX standard purchased from Sigma–Aldrich. b 3MX produced in this work. Inset to b: Purified, crystallized 3MX produced in this work Full size image The 1H NMR spectrum of biologically produced and standard 3-methyl xanthine also matched very well (Additional file 1: Figure S4). 1H NMR was recorded on a Bruker 500 MHz spectrophotomer using DMSO-d6 as solvent. Standard 3-methylxantine showed presence of peaks at δ 13.48 (s, 1H) and 11.07 (s, 1H) corresponding to –NH proton, and peaks at δ 8.01 and 3.3 corresponding to –C = H (s, 1H) and –CH3 group (s, 3H). The biologically produced 3MX also showed peaks at δ 13.48 (s, 1H) and 11.07 (s, 1H) corresponding to –NH proton, and peaks at δ 8.0 and 3.3 corresponding to –C = H (s, 1H) and –CH3 group (s, 3H). If you want to learn more, please visit our website theophylline synthetic raw material manufacturer.Comparison of growth media
Larger scale reaction, preparative chromatography, and purification of 3MX
Analytical characterization of biologically produced 3MX