Metabolic engineering of Escherichia coli for production of mixed isoprenoid alcohols and their derivatives

Engineering a synthetic pathway for a mixture of isoprenoid-based C₅ and C₁₅ alcohols

In order to convert IPP/DMAPP to isopentenol and FPP to farnesol in E. coli, it is necessary to overexpress endogenous or exogenous phosphatases for dephosphorylation of these prenyl diphosphates. The endogenous phosphatase NudB effectively hydrolyzes IPP and DMAPP to isopentenol [13, 41–43]. As Nudix hydrolases generally show broad substrate specificities [38], we hypothesized that NudB could also have catalytic activity toward FPP. Thus, nudB was cloned with ispA encoding FPP synthase into a strong expression vector, pTrc99A, resulting in plasmid pT-MBF1.0 (Fig. 1b). Plasmid pSNA, expressing an exogenous mevalonate (MVA) pathway, was used to effectively supply prenyl diphosphate precursors for synthesis of mixture isoprenoid alcohols (Fig. 1b) [44]. Both of the plasmids were transformed into E. coli DH5α, resulting in strain NA-MBF1.0 (Additional file 1: Table S1). After culture of the strain for 48 h, the culture broth was analyzed by gas chromatography (GC), which showed seven peaks that were subsequently identified through gas chromatography–mass spectrometry (GC–MS) (Fig. 2; Additional file 1: Fig. S1). Three of the seven peaks were identified to be isoprenol, prenol (C₅ alcohols), and farnesol (C₁₅ alcohol) as expected (Fig. 2). The other four unexpected peaks corresponded to isoprenyl acetate, farnesyl acetate, Z,E-farnesal, and E,E-farnesal (Fig. 2). Thus, the strain NA-MBF1.0 successfully produced 221 mg/L of C₅ alcohols and their derivatives, and 626 mg/L of a C₁₅ alcohol and its derivatives, which was a total of 847 mg/L isoprenoid biofuel mixture (Fig. 3a). In addition to C₅ products, 65.1% of the total isoprenoid mixture produced in the strain NA-MBF1.0 was farnesol, a fourfold increase over the reference strain (135.5 mg/L) overexpressing IspA only [37]. Previously, the overexpression of NudB was used for the production of C₅ products, isopentenol, because it was considered to have its activity only to IPP/DMAPP [13, 41–43]. However, our result indicates that NudB is able to dephosphorylate FPP to farnesol (C₁₅ products), as well as IPP/DMAPP to isopentenol. NudB was thus presumed to be a potential phosphatase, hydrolyzing a wide range of prenyl diphosphates, and is likely appropriate for production of a mixture of isoprenoid alcohols as biofuel. The new promiscuous activity of NudB toward FPP was verified by in vitro study (Additional file 2: Fig. S6).

Engineering a synthetic pathway for a mixture of isoprenoid-based C₅ and C₁₀ alcohols

Identification of enzymes leading to the conversion of alcohols to their derivatives

Isoprenyl acetate and farnesyl acetate are esters of isoprenol and farnesol, respectively. The formation of esters has not been reported in the production of isopentenol and farnesol [13, 37, 41, 46–49]. Chloramphenicol acetyltransferase (CAT) has been reported to possess nonspecific esterification activity toward some isoprenoid alcohols such as perillyl alcohol [50], geraniol [26], and retinol [51]. Thus, the formation of isoprenyl acetate and farnesyl acetate in strain NA-MBF1.0 was suspected due to the chloramphenicol antibiotic marker (Cm^R) gene (cat) in plasmid pSNA. To validate this theory, plasmid pSNAK with a kanamycin antibiotic marker (Km^R) gene was used instead of pSNA. The strain NAK-MBF1.0 harboring pT-MBF1.0 and pSNAK produced isoprenoid alcohols (isoprenol, prenol, and farnesol), farnesal, and farnesyl acetate, but no formation of isoprenyl acetate was observed (Fig. 3a). However, the production of farnesyl acetate was still observed in strain NAK-MBF1.0 (Fig. 3a). This indicates that CAT contributes to the esterification of isoprenol to its acetyl ester, but is not responsible for the formation of farnesyl acetate. Hence, some unknown endogenous acetyltransferase in E. coli seems to be involved in the esterification of farnesol into farnesyl acetate (Fig. 1a). The cell growth of NAK-MBF1.0 was slightly lower than that of NA-MBF1.0 (Additional file 1: Fig. S2).

Farnesal is an oxidized product of farnesol. The oxidation of farnesol to farnesal and even to farnesoic acid by alcohol dehydrogenase has been reported in Aedes aegypti [52]. Oxidation of geraniol to geranial and neral by alcohol dehydrogenases has been reported in E. coli [26]. The formation of farnesal in strains NA-MBF1.0 and NAK-MBF1.0 was thus suspected to be a result of oxidation of farnesol by some endogenous alcohol dehydrogenases/aldehyde reductases. To address this, YjgB, YahK, YddN, and AdhE from E. coli were overexpressed using the vector pTrc99A. YjgB and YahK are NADPH-dependent aldehyde reductases, exhibiting broad substrate specificity toward aldehydes and alcohols [53]. YddN is a NADH-dependent acetaldehyde reductase with a catalytic activity toward medium-chain alcohols [54]. AdhE is an aldehyde-alcohol dehydrogenase, catalyzing the conversion of primary alcohols to aldehydes [55]. Authentic standard farnesol was fed to the culture of E. coli strains DH5α-Trc, DH5α-YjgB, DH5α-YahK, DH5α-YddN, and DH5α-AdhE harboring plasmids pTrc99A, pT-YjgB, pT-YahK, pT-YddN, and pT-AdhE, respectively (Additional file 1: Table S2). The strain DH5α-YjgB significantly transformed farnesol to farnesal (Additional file 1: Table S2). To confirm the dehydrogenase activity of YjgB toward farnesol, farnesol was fed to a culture of mutant strain MGΔYjgB with a deletion of yjgB; no farnesal peak was observed in GC analysis of the culture (data not shown).

Similarly, formation of isoprenyl, prenyl, and geranyl acetates in strain NA-MBF2.0 was also suspected due to esterification of isoprenol, prenol, and geraniol, respectively, by the CAT of plasmid pSNA. There was no formation of the isoprenoid acetate esters in the strain NAK-MBF2.0 harboring pT-MBF2.0 and pSNAK (Fig. 3b). The citral produced from strains NA-MBF2.0 and NAK-MBF2.0 is the dehydrogenated product of geraniol formed by YjgB [26]. Cell growth of strain NAK-MBF2.0 was slightly reduced in comparison to that of strain NA-MBF2.0 (Additional file 1: Fig. S2).

Effects of ispA and nudB operon gene order along with overexpression of aphA on total titer of the isoprenoid mixture

Since we successfully constructed strains for synthesis of C₅–C₁₀ and C₅–C₁₅ isoprenoid-based biofuel mixtures, we next sought to improve the production of the C₅–C₁₅ isoprenoid-based biofuels mixture. A balanced expression of production pathway genes is critical to achieve high titers of desirable products. There is a fundamental relationship between expression level of genes and their order in an operon. In general, expression of a gene increases as its position in an operon moves closer to a promoter [56]. Positional modulation of the constituent genes in a synthetic operon has significantly enhanced productions of zeaxanthin, taxadiene, and protoilludene in E. coli [57–59]. The order of the ispA and nudB genes in pT-MBF1.0 was reversed to construct pT-MBF1.1 (Fig. 1b). Cell growth of strain NA-MBF1.1 harboring pT-MBF1.1 and pSNA was significantly enhanced in comparison with the growth of strain NA-MBF1.0 (Additional file 1: Fig. S3). The total titer of the isoprenoid mixture from strain NA-MBF1.1 was 1652 mg/L, a twofold increase over the production of strain NA-MBF1.0 (847 mg/L) (Fig. 4). The significant enhancement of the total production was mainly due to the increased production of C₅ products (prenol, isoprenol, and isoprenyl acetate) from 221 to 979 mg/L, whereas the production of C₁₅ products (farnesol, farnesal, and farnesyl acetate) was not significantly different between the strains (Fig. 4). These results show that insufficient expression of NudB from pT-MBF1.0 caused the lower production of C₅ products in strain NA-MBF1.0 because of the limited conversion of IPP and DMAPP to C₅ products, which would also cause an intracellular accumulation of IPP and DMAPP. The lower cell growth of strain NA-MBF1.0 was presumed to be due to the accumulation of IPP and DMAPP, which are toxic and can inhibit normal cell growth [60].

NudB was thought to convert IPP and DMAPP directly to isoprenol and prenol, respectively [13]. However, a recent study has shown that NudB hydrolyzes IPP and DMAPP into their monophosphate forms, IP and DMAP, respectively, in E. coli, and the subsequent hydrolysis of IP and DMAP to isopentenol is performed by other phosphatases such as AphA, Agp, and YqaB [42]. The aphA gene was thus cloned into plasmid pT-MBF1.1 after ispA, resulting in pT-MBF1.2 (Fig. 1b). The strain NA-MBF1.2 harboring pT-MBF1.2 and pSNA produced 2027 mg/L of total isoprenoid mixture (1105 mg/L of C₅ products and 922 mg/L of C₁₅ products), which was an increase of 1.2-fold over the production of strain NA-MBF1.1 (Fig. 4). The overexpression of AphA enhanced the production of not only the C₅ products but also the C₁₅ products. This suggests that FPP is sequentially dephosphorylated to farnesol by the cooperation of NudB and AphA, similar to the dephosphorylation of IPP/DMAPP to isopentenol. A significant decrease in farnesol production has been reported in a previous study of an AphA mutant strain [37], indicating the involvement of AphA in farnesol production. The cell growth of strain NA-MBF1.2 was slightly reduced in comparison with that of strain NA-MBF1.1 (Additional file 1: Fig. S3). To date, NudB has been only known to have its activity on IPP/DMAPP and used for production of C₅ isoprenoid alcohols (Table 1). However, NudB has a promiscuous activity on isoprenyl diphosphates (IPP/DMAPP, GPP, and FPP), and is successfully applied here for production of mixed isoprenoid alcohols (Table 1). Although George et al. reported 2230 mg/L of a single C₅ alcohol (isoprenol), their production of mixed C₅ alcohols (isoprenol, prenol, and isopentanol) was only 1150 mg/L [41]. Here, we produced the comparable amount of mixed C₅ alcohols (1105 mg/L) in addition to C₁₅ alcohol (922 mg/L), a total mixed isoprenoid alcohol of 2027 mg/L.

Augmented expression of the idi gene changes the compositional profile of the isoprenoid mixture

The prenol fractions of the isoprenoid mixtures obtained from strains NA-MBF1.1 and NA-MBF1.2 were very small compared to those of isoprenol. This is probably due to a lower intracellular level of DMAPP than IPP or a higher specificity of NudB toward IPP than DMAPP. IPP isomerase (IDI) catalyzes the reversible isomerization of IPP to DMAPP and adjusts its ratio to 3:7 [61]. In order to increase the conversion of IPP to DMAPP, the idi gene was cloned downstream of the ispA gene of pT-MBF1.1 and pT-MBF1.2, resulting in pT-MBF1.1a and pT-MBF1.2a, respectively (Fig. 1b). The effect of the additional overexpression of IDI on the compositional profile of the isoprenoid mixture was investigated in strains NA-MBF1.1a and NA-MBF1.2a harboring pT-MBF1.1a/pSNA and pT-MBF1.2a/pSNA, respectively (Fig. 4; Additional file 1: Fig. S4). The additional overexpression of IDI drastically changed the compositional profile with a significant decrease and increase in C₅ and C₁₅ products, respectively, but a small increase of prenol. IPP-derived products (isoprenol and isoprenyl acetate) of the strains NA-MBF1.1a and NA-MBF1.2a were 303 and 345 mg/L, respectively, which were distinctively lower than 965 and 1086 mg/L of the corresponding strains NA-MBF1.1 and NA-MBF1.2, (Fig. 4, Additional file 1: Fig. S4). The concentrations of DMAPP-derived prenol in strains NA-MBF1.1a and NA-MBF1.2a were 53 and 67 mg/L, respectively, which were increased more than threefold compared to those in strains NA-MBF1.1 and NA-MBF1.2; however, the absolute increases were only 39 and 47 mg/L, respectively (Additional file 1: Fig. S4). However, the concentrations of FPP-derived products (farnesol, farnesal, and farnesyl acetate) of the strains NA-MBF1.1a and NA-MBF1.2a significantly increased to 1174 and 1419 mg/L, respectively, in comparison with 706 and 922 mg/L in strains NA-MBF1.1 and NA-MBF1.2, respectively (Fig. 4; Table 1; Additional file 1: Fig. S4). The increase in FPP-derived products could be ascribed to an increase in the intracellular levels of DMAPP through the expression of IDI. The FPP synthase IspA possesses a higher Km with DMAPP (50 µM) than with IPP (4 µM) [62], and therefore a higher intracellular DMAPP level through the additional overexpression of IDI could be a benefit to divert the flux of isoprenyl diphosphates to FPP. Moreover, the DMAPP would initiate additional FPP synthesis as a reaction-priming molecule (Fig. 5). Although the additional overexpression of IDI increased the prenol fraction in the isoprenoid mixture, there was still a significant difference between the titers of isopentenol (IPP-derived product) and prenol (DMAPP-derived product) (Additional file 1: Fig. S4), which could indicate a preference of NudB for IPP over DMAPP. The cell growths of strains NA-MBF1.1a and NA-MBF1.2a were slightly lower than those of strains NA-MBF1.1 and NA-MBF1.2 (Additional file 1: Fig. S3). The specific total production of the isoprenoid mixtures from strains NA-MBF1.1a and NA-MBF1.2a was enhanced over the corresponding strains NA-MBF1.1 and NA-MBF1.2 (Fig. 6). FPP-derived products were a major fraction of the isoprenoid mixtures obtained from strains NA-MBF1.1a and NA-MBF1.2a. This suggests that NudB is an effective phosphatase even with FPP, and the compositional profile of the isoprenoid mixture can be modulated by changes in metabolic flux toward various isoprenyl diphosphates (Fig. 5). It presents an important issue in field of metabolic engineering. Product profile of an enzyme is often restricted by its surroundings such as an intracellular level of metabolites (intermediates or precursors to the enzymatic reaction), and thus a specificity of enzymes is designated in a given specific condition. This study successfully shows a modulation of product profile by implementations of the promiscuous activity of NudB and a change of intracellular synthesis of the precursor isoprenyl diphosphates (IPP/DMAPP, GPP, and FPP). The maximum titer of farnesol production reported until now is 526 mg/L [49], while we produced 1419 mg/L of farnesol in addition to 412 mg/L of isopentenol. This is the highest titer of farnesol production reported.

Strains, media, and culture conditions

The E. coli DH5α strain was used as a host for cloning and production of desired compounds. For plasmids construction, cells were grown overnight at 37 °C in LB (Luria–Bertani) medium (10 g tryptone, 5 g yeast extract, and 10 g sodium chloride per liter), while cultures for seed preparation and isoprenoid production were grown in 2YT medium (16 g tryptone, 10 g yeast extract, and 5 g sodium chloride per liter). Antibiotics were added to the culture medium at a final concentration of 100 mg/L of ampicillin and 50 mg/L of kanamycin or chloramphenicol as required. To make the seed culture, colonies were inoculated in 5 mL of 2YT medium with appropriate antibiotics and grown overnight at 30 °C with shaking at 250 rpm. The production culture was made by inoculating the overnight seed culture into 4 mL of 2YT medium to an optical density at 600 nm (OD_600 nm) of 0.1 of with 2% (v/v) of glycerol as the main carbon source. The culture was carried out in two-phases by overlaying 1 mL of oleyl alcohol over 4 mL of the culture in order to prevent volatile loss of the products [41, 63]. The two-phase production culture was initially induced with 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and was incubated for 48 h with 250 rpm shaking in a rotary shaker at 30 °C. Cellular growth of the culture was measured in terms of OD_600 nm using a spectrophotometer. Glass tubes (2.5 cm × 15 cm) with plastic caps were used for aerobic culture for all experiments conducted in this study. Genomic DNA of E. coli MG1655 was used for PCR amplification of the genes cloned in this study. All strains used in this study are listed in Additional file 1: Table S1.

Plasmid construction

PCR primers and plasmids used in this study are listed in Additional file 1: Table S1. PCR was performed using Pfu DNA polymerase (SolGent, Daejeon, Korea) according to the manufacturer’s instructions for amplification of DNA fragments. The ispA [Genbank: NC_000913.3: 440202–441101] fragment was PCR amplified from the genomic DNA of E. coli MG1655 with primers IspA-F and IspA-R, digested with restriction enzymes SacI and KpnI, and ligated between the corresponding restriction sites of pTrc99A to create pT-SBL. For the construction of pT-MBF1.0, nudB [Genbank: NC_000913.3: 1948180–1948632] was amplified from the genomic DNA with primers NudB-F1 and NudB-R1, digested with BamHI and HindIII, and subsequently inserted between the corresponding restriction sites of pT-SBL. The ispA mutant ispA* was PCR amplified with primers IspA*-F and IspA*-R using the plasmid pT-GPSGES [26] as a template. The PCR product was digested with KpnI and BamHI and ligated into the corresponding restriction sites of pT-SBL to create pT-MBF2.0. Another fragment of nudB was PCR amplified from pT-MBF1.0 with primers NudB-F2 and NudB-R2, digested with NcoI and SacI, and ligated upstream of ispA in pT-SBL to create pT-MBF1.1. The idi [Genbank: NC_000913.3: 3033065–3033613] fragment was PCR amplified from the genomic DNA with primers Idi-F and Idi-R. The resulting fragment was digested with BamHI and SalI and inserted into the corresponding restriction sites of pT-MBF1.1 to construct pT-MBF1.1a. The DNA fragment of aphA [Genbank: NC_000913.3: 4269414–4270127] was PCR amplified from the genomic DNA with primers of AphA-F and AphA-R, digested with SalI and HindIII, and inserted between the corresponding restriction sites of pT-MBF1.1 and pT-MBF1.1a to create pT-MBF1.2 and pT-MBF1.2a, respectively. The genes overexpressed for transformation of farnesol to farnesal, yjgB [Genbank: NC_000913.3: 4495190–4496209], yahK [Genbank: NC_000913.3: 342884–343933], and yddN [Genbank: NC_000913.3: 1552828–1553838], were previously cloned in the plasmids pT-YjgB, pT-YahK, and pT-YddN, respectively [26]. For construction of pT-AdhE, the DNA fragment of adhE [Genbank: NC_000913.3: 1295446–1298121] was PCR amplified from the genomic DNA with primers adhE-F and adhE-R, digested with EcoRI and BamHI, and inserted between the corresponding restriction sites of pTrc99A. To convert acetyl-CoA to IPP and DMAPP via the mevalonate (MVA) pathway, plasmid pSNA encoding the MVA pathway was previously constructed by Yoon et al. [44]. The plasmid pSNA was constructed by cloning of the top portion of the MVA pathway (mvaE and mvaS from Enterococcus faecalis) and its bottom portion (mvaK1, mvaK2, and mvaD from Streptococcus pneumonia, and idi from E. coli) into the pSTV28 vector under control of a lac promoter. The cat gene encoding chloramphenicol acetyltransferase (CAT) in plasmids pSTV28 and pSNA had been replaced with a kanamycin resistance gene, and the resultant plasmids were named pSTV28 K and pSNAK, respectively [26].

Farnesol feeding of the culture

To investigate the conversion of farnesol to farnesal, 1 mL of decane containing 1 g/L of farnesol was put over 4 mL of 2YT medium, with 2% (v/v) glycerol as a carbon source, which was inoculated with the recombinant strains DH5α-Trc, DH5α-YjgB, DH5α-YahK, DH5α-YddN, or DH5α-AdhE, or the mutant strain MGΔYjgB. The culture of each strain was initially induced with 0.2 mM IPTG and incubated at 30 °C with 250 rpm shaking in a rotary shaker for 48 h. The decane phase was collected at 12-h intervals and subsequently subjected to GC analysis to observe the conversion of farnesol to farnesal.

Analysis of products by GC–MS and GC-FID

The entire culture broth including oleyl alcohol phase was extracted with ethyl acetate after 48 h of culture [41]. The ethyl acetate fraction was obtained by high-speed centrifugation and subjected to analysis by GC–MS and GC-FID. Formation of isoprenoid mixtures, composed of isoprenol, isoprenyl acetate, prenol, prenyl acetate, geraniol, citral, geranyl acetates, farnesol, farnesal, and farnesyl acetate, was identified by GC–MS (GC–MS-QP2010, SHIMADZU, Japan). All products were quantitatively analyzed using GC (Agilent Technologies 7890A) equipped with a flame ionization detector (FID). Samples were injected into a DB-WAX123-7032 column (30 m in length, 0.320 mm in internal diameter, and 0.25 μm in film thickness), and nitrogen was used as a carrier gas. The GC oven temperature started at 40 °C and was raised with a gradient of 5 °C/min until 80 °C and held for 1 min. The temperature was then raised at a rate of 60 °C/min until 235 °C where it was held for 5 min. The detector temperature was maintained at 260 °C. The standard curves for isoprenol, prenol, isoprenyl acetate, geraniol, citral, geranyl acetates, farnesol, farnesal, and farnesyl acetate were constructed for the quantitative calculation of their production (Additional file 1: Fig. S5).