Systematic engineering pinpoints a versatile strategy for the expression of functional cytochrome P450 enzymes in Escherichia coli cell factories
Microbial Cell Factories volume 22, Article number: 219 (2023)
Production of plant secondary metabolites in engineered microorganisms provides a scalable and sustainable alternative to their sourcing from nature or through chemical synthesis. However, the biosynthesis of many valuable plant-derived products relies on cytochromes P450 – enzymes notoriously difficult to express in microbes. To improve their expression in Escherichia coli, an arsenal of engineering strategies was developed, often paired with an extensive screening of enzyme variants. Here, attempting to identify a broadly applicable strategy, we systematically evaluated six common cytochrome P450 N-terminal modifications and their effect on in vivo activity of enzymes from the CYP79 and CYP83 families. We found that transmembrane domain truncation was the only modification with a significantly positive effect for all seven tested enzymes, increasing their product titres by 2- to 170-fold. Furthermore, when comparing the changes in the protein titre and product generation, we show that higher protein expression does not directly translate to higher in vivo activity, thus making the protein titre an unreliable screening target in the context of cell factories. We propose the transmembrane domain truncation as a first-line approach that enables the expression of wide range of highly active P450 enzymes in E. coli and circumvents the time-consuming screening process. Our results challenge the notion that the engineering strategy must be tailored for each individual cytochrome P450 enzyme and have the potential to simplify and accelerate the future design of E. coli cell factories.
Biorxiv, CC-BY-NC-ND 4.0 International license. https://doi.org/10.1101/2022.12.13.520134v1.abstract.
Cytochromes P450 (CYPs or P450s) are amongst the most versatile enzymes in nature and draw the attention of metabolic engineers with their ability to catalyse unique reactions on a myriad of complex substrates. In planta, P450s unlock the diversity of secondary metabolism  and play a key role in the biosynthesis of natural products used as medicines, cosmetics, colourants, and flavours [2, 3]. Reconstituting these biosynthetic pathways in microbial cell factories, such as Escherichia coli or Saccharomyces cerevisiae, allows the production of high-value natural products without the problems associated with the native hosts: seasonal nature of the production, low amounts of the product per plant, and tedious extraction from plant material. Well-known examples include the semi-synthetic production of the antimalarial drug artemisinin ; biosynthesis of taxadiene, an intermediate of the anti-cancer medicine Taxol ; and de novo biosynthesis of analgesic opioids [6, 7]. Still, functional expression of heterologous P450s remains a challenge that researchers must address to reach competitive production levels of plant pathways in microbes.
All plant P450s require a redox partner for electron transfer and most localise on the membrane of the endoplasmic reticulum [8, 9], making E. coli – without eukaryotic organelles and no native P450 enzymes – a particularly challenging host. The N-terminal hydrophobic amino acids, which anchor P450s in membranes and act as a membrane localisation signal, cause the proteins to aggregate in inclusion bodies if the proteins cannot assume their native fold at a pace matching the speed of translation [10, 11]. Larson et al. pioneered the strategy of N-terminal truncation, removing the hydrophobic amino acids of rabbit CYP2E1 and showing that the enzyme maintains its original activity . Barnes et al. optimised the 5’ codons of bovine CYP17A, increasing the translation initiation speed and enabling the protein to express in E. coli . N-terminal truncation and insertion of the MALLLAVF peptide, the so-called Barnes sequence, were later used to establish the expression of many other eukaryotic P450s in bacteria [14,15,16,17,18]. Because these modifications can increase the ratio of P450s localised in the cytosol [19, 20], alternative techniques were developed to maintain P450 membrane localisation: insertion of leader sequences from bacterial membrane proteins  and exchange of the N-terminal domain with transmembrane sequences from E. coli membrane proteins [22, 23] or well-expressed eukaryotic P450 enzymes .
Although researchers can now choose from an arsenal of engineering strategies, no clear favourite exists and establishing novel plant pathways requires extensive optimisation and screening of the P450 enzyme expression and functionality [22, 23]. Considering that most of the available research has focused on increasing the P450 protein titre, which does not always translate to higher in vivo activity , it is difficult to use previous results to guide the P450 engineering strategy in the context of cell factories. Furthermore, plant biosynthetic pathways often rely on multiple P450-mediated steps , while most studies in the field aim to overexpress only a single P450 enzyme at a time .
Here, we suggest a versatile strategy to express highly functional P450s in E. coli cell factories by systematically comparing six common N-terminal modifications. We evaluate the changes of in vivo activity of CYP79A2 together with two members of the CYP83 family and then validate the best strategy on four additional P450 enzymes. CYP79 and CYP83 catalyse two consecutive steps at the entrance to the glucosinolate pathway and accept standard amino acids as substrates (Fig. 1) . This makes them a convenient model system for screening of product generation that avoids the need to feed complex substrates typical for plant secondary metabolism and allows the evaluation of concurrent engineering of multiple P450s.
N-terminal sequence modifications affect CYP79A2 functionality and expression levels
We chose the conversion of phenylalanine to phenylacetaldoxime (Phe-Ox) catalysed by CYP79A2 for the initial assessment of N-terminal modifications because CYP79A2 constitutes a bottleneck in its biosynthetic pathways [27, 28], and it could be improved by N-terminal truncation  and the introduction of Barnes sequence . We cloned the full-length, codon-optimized CYP79A2 from Arabidopsis thaliana and six engineered variants with different N-terminal modifications into operons with A. thaliana reductase 1 (ATR1, Fig. 2D). All tested modifications were previously shown to increase the functional expression of P450s and fall into four general categories: (1) N-terminal truncation by removing either the transmembrane domain (ΔTM) or both the transmembrane domain and the adjacent hydrophilic region (ΔTM+) [12, 29], (2) 5’-codon optimisation by inserting short expression enhancing peptides in front of the P450 protein sequence (Barnes and 28aa), (3) transmembrane domain exchange with E. coli native membrane anchor from SohB probable protease (SohB) , and (4) introduction of E. coli leader sequence from outer membrane protein A (OmpA)  (Fig. 2A).
We assayed CYP79A2 variants by measuring the generation of Phe-Ox after 72 h fermentation in T7 autoinducing media (Fig. 2B). Although CYP79A2 was functional without any sequence modifications, producing 0.52 mM Phe-Ox, many engineered variants reached significantly higher titres. Those with the transmembrane domain removed or replaced were the top producers and [ΔTM]CYP79A2, where we truncated the transmembrane domain, but left the hydrophilic region intact, reached the highest Phe-Ox titre at 0.89 mM. The introduction of the Barnes sequence was beneficial only when the MALLLAVF peptide was inserted in front of the full-length enzyme, rather than substituting the first eight amino acids (Fig. S2). Using OmpA leader sequence or 28aa tag impaired CYP79A2 function, reducing Phe-Ox titre by approximately 30%. Because ATR2 and Δ44ATR2 are often used in literature, we compared the different reductases and their effect on [ΔTM]CYP79A2 activity. In our expression system, ATR1 was superior to both and led to 2-fold and 3-fold higher Phe-Ox titres than ATR2 and Δ44ATR2, respectively (Fig. S3) and was therefore used in all following experiments. Ensuring the rich medium did not skew our results, we have compared the strains expressing [ΔTM]CYP79A2 and native CYP79A2 in supplemented M9 minimal media and obtained comparable improvement as in TB medium (Fig. S4).
To relate the metabolite data with the P450 protein titre, we performed targeted proteomics measuring CYP79A2 protein expression levels on six of the replicates used above (Fig. 2C). We observed a disparity between the effect the modifications had on CYP79A2 protein expression compared to the increase in product generation. [ΔTM]CYP79A2 showed the highest protein levels and highest Phe-Ox production, but the relative protein titre increased by 28-fold, while Phe-Ox titre improved by only 1.8-fold. The introduction of 28aa tag increase protein levels 7-fold compared to native enzyme, while decreasing Phe-Ox production by 40%. These results corroborate previous reports that protein titre is not indicative of in vivo activity of P450 enzymes . In the following sections, we continue with [ΔTM]CYP79A2 as the best-performing variant using product generation as the metric to evaluate our engineering strategies.
Concurrent engineering of a dual P450 module
Plant biosynthetic pathways for valuable secondary metabolites often proceed through multiple P450-mediated steps . To study the simultaneous engineering of two P450 enzymes, we paired native and truncated CYP79A2 with variants of CYP83B1 or CYP83A1, which both convert Phe-Ox into phenylacetonitrile oxide (Fig. 1). We expressed CYP79A2 with either one of the CYP83s and ATR1 from a single operon under the control of T7 promoter (Fig. 3A). When assaying the dual P450 module, we measured the generation of S-phenylacetohydroxymoyl-L-glutathione (Phe-GSH, Fig. 1), a more suitable product to quantify than the reactive and unstable nitrile oxide . Although the conjugation of nitrile oxides with glutathione can occur spontaneously, we co-expressed glutathione-S-transferase F11 (GSTF11) alongside the P450s and ATR1 to improve the formation of Phe-GSH (Fig. 3A) .
We observed that strains with native CYP83s failed to convert the increased supply of Phe-Ox by [ΔTM]CYP79A2 into nitrile oxide, producing less Phe-GSH than with native CYP79A2 and accumulating Phe-Ox instead (Fig. 3B and D). This was not specific to the truncated CYP79A2, as expressing the SohB variant led to the same result (Fig. S5). To achieve nearly complete conversion, we had to engineer the N-termini of both CYP79A2 and CYP83s. Although the highest Phe-GSH titres were similar between the CYP83s, interestingly, the effect of the N-terminal modifications varied between the two enzymes. For CYP83B1, multiple variants showed a significant increase in product titre with ΔTM, SohB and OmpA variants generating approximately 1.5 mM Phe-GSH, more than double compared to the native control (Fig. 3C). For CYP83A1, only ΔTM variant had a significantly positive effect on product titre and generated 1.47 mM Phe-GSH, on par with CYP83B1 (Fig. 3E). Our results suggest ΔTM N-terminal truncation is superior to the other approaches as it was the only modification that significantly increased the product generation by all three enzymes CYP79A2, CYP83B1 and CYP83A1.
Transmembrane domain truncation as a first-line approach for P450 functional expression in E. coli
To validate transmembrane domain truncation as a strategy that could be applied to a broad range of P450s, we chose four additional CYP79 enzymes with diverse N-terminal sequences, distinct substrate preferences and different plant origins. CYP79A1 from Sorghum bicolor converts tyrosine into its corresponding oxime , CYP79B2 from A. thaliana accepts tryptophan as the sole substrate , CYP79D2 from Manihot esculenta acts on both valine and isoleucine , and CYP79F6 from Barbarea vulgaris accepts only non-canonical, carbon chain-elongated amino acids (here homophenylalanine) . We cloned the four CYP79s, CYP83 and ATR1 into operons where both CYP79 and CYP83 were either native full-length or with the transmembrane domain truncated.
We assayed the enzymes by measuring the generation of putative glutathione conjugates derived from their respective amino acid substrates by tandem mass spectrometry (Q-TOF LC-MS/MS), comparing the relative production in strains expressing the native and truncated P450s (Fig. 4). We identified the glutathione conjugates by extracting molecular features with matching m/z ratios and a characteristic fragmentation pattern in the mass spectrum [36, 37], which we confirmed with Phe-GSH standard (Fig. S6). The results further corroborated the broad applicability of ΔTM as the transmembrane domain truncation led to higher product accumulation for all four new CYP79s. Although CYP79A1 and CYP79B2 were functional without any modification, truncating the enzymes yielded almost double of their respective products despite being limited by insufficient supply of tyrosine and tryptophan in TB media (Fig. S7). The strain with [ΔTM]CYP79F6 and [ΔTM]CYP83A1 produced almost 7-fold more homo-phenylalanine-derived glutathione conjugate than the strain with native P450s and in the case of CYP79D2, which showed only trace activity without truncation, the improvement was 150-fold and 170-fold for valine and isoleucine products, respectively. Taken together, we show that transmembrane domain truncation improves functional expression and in vivo product accumulation for all seven studied P450 enzymes, being especially effective for enzymes with low native activity.
We produced a broad and systematic screening of six common P450 N-terminal sequence modifications evaluating their effect on the in vivo enzymatic activity. Using two consecutive steps catalysed by CYP79A2 and either CYP83A1 or CYP83B1 as a model system, we found that truncation of the transmembrane domain outperformed the other N-terminal modifications as it significantly improved product generation across all three P450 enzymes. We sought to validate this finding by truncating four additional CYP79 enzymes and showed that ΔTM variants again led to increased product titres in all four cases. Amongst the six modifications, only the transmembrane domain truncation was beneficial for all studied combinations of enzymes from CYP79 and CYP83 families, improving the product titres 2- to 170-fold. This strongly suggests that the transmembrane domain truncation serves as a versatile strategy to improve the functional expression of plant P450s in E. coli. Moreover, in the context of microbial cell factories, we demonstrated that in pathways with multiple P450s concentrating on the rate-limiting step is insufficient and all P450s must be engineered concurrently.
Classically, P450 engineering studies measure protein titre as a metric of success [25, 26]. However, we observed a disparity between the effect the modifications had on the protein levels and the in vivo activity of CYP79A2 enzyme. ΔTM, ΔTM + and SohB variants improved protein expression far greater than product titre whereas [28aa]CYP79A2 increased protein levels 7-fold while reducing Phe-Ox production by 30% compared to the native enzyme. Zhou et al. reviewed several studies that report a similar discrepancy  and Christensen et al. showed that CYP79A1 fused to a membrane anchor of E. coli signal peptidase 1 was amongst the variants with the highest protein levels but only marginally improved in vivo activity . Changes at the N-terminus can increase the P450 fraction in the cytosol [9, 19, 20, 26], limiting their access to the P450 reductase, and may interfere with the protein-protein interaction between the reductase and the P450s on the membrane . It is possible that the modifications impair the electron transfer between the reductase and the P450 variants, where protein amount does not translate to in vivo activity. Furthermore, at the very high expression levels of [ΔTM]CYP79A2, the availability of substrates, heme or NADPH may replace protein titre as the limiting factor. Together, these results indicate that screening P450 protein expression levels is insufficient to guide the engineering strategy in the context of cell factories. While screening P450 in vivo activity is more exact, it becomes cumbersome at scale requiring access to diverse and complex substrates as well as established quantification methods for the generated products.
We set out to identify an engineering strategy that could be applied as a first-line approach without screening a library of variants, as, in our opinion, no clear favourite has emerged in literature. Available studies use different growth conditions and combine multiple modifications to tailor individual P450 enzymes for the highest possible protein titre, making it difficult to abstract a broadly applicable engineering strategy. We studied the modifications individually and found that only ΔTM caused a significant increase in product generation across all tested enzymes, and while other modifications could lead to similar improvements, they failed to have a consistently positive effect. Barnes and SohB variants were significantly better than native CYP79A2 and CYP83B1 but caused no statistically significant change in CYP83A1. OmpA was among the top variants for CYP83B1 but led to a dramatic decrease in Phe-Ox production by CYP79A2. The final culture density was not a general determinant of high in vivo activity, but specifically for OmpA and Barnes variants with low product titres, slower growth may contribute to their poor performance (Table S1 and S3). These results show that the transmembrane domain is a major factor that impedes the functional expression of P450s in E. coli, and its truncation, in front of the hydrophilic region, is sufficient to alleviate the problem in a wide range of P450 enzymes.
Balancing the expression of individual enzymes or modules is a standard approach in pathway engineering. We noticed a similar requirement for P450 engineering in pathways containing multiple P450 enzymes. [ΔTM]CYP79A2, which catalyses the rate-limiting reaction at the entry point of the glucosinolate pathway [27, 28], produced almost double Phe-Ox over the native enzyme, but this benefit was lost upon adding the subsequent unmodified CYP83s, which were unable to convert the extra Phe-Ox into nitrile oxide. Only when engineering both CYP79A2 and CYP83s, we successfully translated the doubled Phe-Ox titre into doubled Phe-GSH. As P450s draw from shared cellular resources such as heme incorporation, electron supply through NADPH, and electron transfer from NADPH to P450 by cytochrome P450 reductase, the 28-fold increase in expression we observed for [ΔTM]CYP79A2 might cause CYP79A2 to monopolize the access to these resources and, in turn, impair CYP83 activity.
Although optimization and validation at a larger scale is necessary, we expect our findings to translate well from a 24-well plate screening to a bioreactor setting. Bioreactors provide a more controlled fermentation environment, which can support much higher bacterial biomass and thus higher product titre. Especially the control over dissolved oxygen and aeration rate will have a great impact on the in vivo activity of P450s, which require oxygen for catalysis. Furthermore, due to cost, minimal media is common for fermentation at scale and in the industry. We have performed a control experiment in supplemented M9 minimal media showing that the benefit of N-terminal truncation is independent of rich media and high product titres are maintained (Fig. S4). Taken together, we identify the N-terminal truncation as the best and the most versatile modification to achieve high in vivo activity across a wide range of plant P450 enzymes in E. coli. Our results emphasise the discrepancy between P450 protein titre and in vivo activity, suggesting that screening of product generation is necessary to guide P450 engineering. To avoid the extensive screening of various N-terminal modifications, we propose ΔTM as a first-line approach to yield highly functional P450 enzymes and aid future design and establishment of P450-containing biosynthetic pathways in E. coli cell factories.
Materials and methods
Oligonucleotides, Plasmids, Bacterial strains, Transformation protocol and LC-MS/MS procedures for small molecules and peptides. Detailed description can be found in Supplementary Material.
Construction of operons for P450 expression
Purified PCR products were assembled using an adapted USER protocol from Geu-Flores et al.  and Cavaleiro et al.  in a 10 µL reaction as follows: 1 µL USER enzyme mix, 1 µL 10x CutSmart buffer, 1 unit of DpnI, 20 ng PCR product of the plasmid backbone and 40 ng PCR product of all fragments to assemble. The reaction was incubated for 1 h at 37 °C, then at gradually decreasing temperatures around the melting temperature (Tm) of USER overhangs (31 − 26 °C) with 5 min at each temperature step and finished with 30 min at 10 °C. The heterologous genes were assembled into synthetic operons with a maximum of three genes per operon. All operons were flanked by consensus T7 promoters with T7 terminators, and all genes shared the same ribosome binding site originating from the T7 bacteriophage major capsid protein.
To define transmembrane domains of P450 enzymes, we used Uniprot annotations in combination with a transmembrane domain prediction tool TMHMM v2.0 (now deprecated in favour of DeepTHMHH)  (Table 1). To identify an appropriate truncation site for ΔTM+, the hydrophilic region was annotated as a short amino acid stretch abundant in arginine and lysine residues together with other hydrophilic amino acids. The P450 N-terminal sequence modifications include: (1) truncation of the membrane domain in front of the hydrophilic region (ΔTM), (2) truncation of the membrane domain after the hydrophilic region (ΔTM+), (3) addition of Barnes peptide MALLLAVF in front of the full-length P450 sequence (Barnes), (4) substitution of the first 8 amino acids with Barnes peptide MALLLAVF, (5) addition of 28 amino acid synthetic peptide in front of the full-length P450 sequence (28aa), (6) substitution of the native transmembrane domain with the transmembrane domain of SohB probable protease from E. coli (SohB), and (7) addition of E. coli outer membrane protein A (OmpA) signal peptide in front of the full-length P450 sequence (OmpA). 28aa, SohB and OmpA sequences were ordered from Twist Bioscience as short DNA fragments, amplified with uracil-containing primers, and used to create modified P450s on a separate plasmid backbone before they were cloned into the expression vector. Barnes sequence was short enough to be inserted into the primer overhangs that were used to create the modifications (3) and (4), and truncated P450s were prepared with primers annealing at within the gene sequences. DNA sequences of all used modifications are shown in Table S4.
Overnight cultures were started by picking single colonies into 1.2 mL LB media with appropriate antibiotics in a 24-well plate. Then, the overnight cultures were diluted 1000-fold into 1.2 mL autoinducing medium in round bottom 24-well plates and the fermentation was carried out in a rotary shaker at 18 °C, 200 RPM for 72 h. Autoinducing medium, adapted from Studier et al. , was used during all fermentations unless stated otherwise. The original recipe was adjusted to match Terrific broth (TB) in the content of yeast extract and tryptone and in the buffer composition. The TB-based autoinducing medium contained 24 g/L yeast extract, 12 g/L tryptone, 100 mM KH2PO4-K2HPO4 buffer pH 7.0, 0.05% glucose, 0.5% glycerol, 0.2% α-lactose, 2 mM MgSO4, and trace metal mix (50 µM FeCl3, 20 µM CaCl2, 10 µM MnCl2, 10 µM ZnSO4, 2 µM CoCl2, 2 µM CuCl2, 2 µM NiCl2, 2 µM Na2MoO4, 2 µM Na2SeO3, 2 µM H3BO3). To improve heme biosynthesis necessary for functional P450 enzymes, 0.5 mM 5-aminolevulinic acid was supplied to the media. Corresponding antibiotics were also added at the following concentrations: carbenicillin (50 µg/mL), spectinomycin (50 µg/mL), kanamycin (50 µg/mL) and chloramphenicol (34 µg/mL). 1 mM L-homophenylalanine (TCI Chemicals, Tokyo, Japan) was supplemented to the strain expressing BvCYP79F6. Glutathione is one of the most abundant metabolites in E. coli  and as such was not supplied in the media above the levels found in yeast extract.
M9 minimal media recipe was adjusted to allow for autoinduction of protein expression, and supplemented with 5-aminolevulinic acid, thiamine and trace metals. The final composition was 12.8 g/L Na2HPO4.7H2O, 3 g/L KH2PO4, 1 g/L NH4Cl, 0.5 g/L NaCl, 2 mM MgSO4, 1mM thiamine, 0.1 mM CaCl2, 0.5 mM 5-aminolevulinic acid, 0.05% glucose, 0.5% glycerol, 0.2% α-lactose and trace metal mix (50 µM FeCl3, 20 µM CaCl2, 10 µM MnCl2, 10 µM ZnSO4, 2 µM CoCl2, 2 µM CuCl2, 2 µM NiCl2, 2 µM Na2MoO4, 2 µM Na2SeO3, 2 µM H3BO3).
Metabolomics and targeted proteomics analyses
Media and cell samples were collected separately after the cells were spun down from 1 mL cultures in deep 96-well plates at 3700 x g for 5 min. The media samples were stored at − 20 °C and the cell pellet at − 80 °C until analysis. For metabolite analysis an aliquot of the media was diluted 25-fold in water and filtered through 0.22 μm filters. The glutathione conjugates derived from Tyr, Trp, Leu, Val and HPhe were analysed with Q-TOF LC-MS/MS using their characteristic fragmentation pattern as described previously [36, 37] (Fig. S6). Phe-Ox and Phe-GSH were analysed by triple quadrupole LC-MS/MS after diluting the media further 10-fold with 13 C-,15 N-labeled amino acid mix (250-fold total dilution, 10 µg/ml, Isotec, Miamisburg, US). The titres were quantified with 13 C-, 15 N-Phe as internal standard using response factors calculated from dilution series of the analytes in spent E. coli media to account for the matrix effect (Table S6). Outliers were defined as data lower than Q1 – 2 * IQR and higher than Q3 + 2 * IQR, and were removed from the dataset, where IQR is interquartile range and Q1 and Q3 are lower and upper quartiles, respectively. Student’s upper-tailed t test was performed to identify engineered variants with significant increase in product generation compared to the native enzymes. Statistical analysis and visualisation were done using R (version 4.2.2, ) in RStudio 2022.07.2.576  relying on tidyverse  and ggplot2  packages. For targeted proteomics, the cells were disrupted with BugBuster Protein Extraction Reagent (Merck, Darmstadt, Germany) and the sample preparation method was adapted from Batth et al.  as previously described .
The data supporting the conclusions of the study are included in the supplementary material.
O’Connor SE. Engineering of secondary metabolism. Annu Rev Genet. 2015;49:71–94. https://doi.org/10.1146/annurev-genet-120213-092053.
Yun B-W, Yan Z, Amir R, Hong S, Jin Y-W, Lee E-K, et al. Plant natural products: history, limitations and the potential of cambial meristematic cells. Biotechnol Genet Eng Rev. 2012;28:47–60. https://doi.org/10.5661/bger-28-47.
Hussain MS, Fareed S, Ansari S, Rahman MA, Ahmad IZ, Saeed M. Current approaches toward production of secondary plant metabolites. J Pharm Bioallied Sci. 2012;4:10–20. https://doi.org/10.4103/0975-7406.92725.
Paddon CJ, Keasling JD. Semi-synthetic artemisinin: a model for the use of synthetic biology in pharmaceutical development. Nat Rev Microbiol. 2014;12:355–67. https://doi.org/10.1038/nrmicro3240.
Ajikumar PK, Xiao W-H, Tyo KEJ, Wang Y, Simeon F, Leonard E, et al. Isoprenoid pathway optimization for Taxol Precursor Overproduction in Escherichia coli. Science. 2010;330:70–4. https://doi.org/10.1126%2Fscience.1191652.
Galanie S, Thodey K, Trenchard IJ, Interrante MF, Smolke CD. Complete biosynthesis of opioids in yeast. Science. 2015;349:1095–100. https://doi.org/10.1126%2Fscience.aac9373.
Nakagawa A, Matsumura E, Koyanagi T, Katayama T, Kawano N, Yoshimatsu K, et al. Total biosynthesis of opiates by stepwise fermentation using engineered Escherichia coli. Nat Commun. 2016;7:1–8. https://doi.org/10.1038/ncomms10390.
Hu B, Zhao X, Wang E, Zhou J, Li J, Chen J, et al. Efficient heterologous expression of cytochrome P450 enzymes in microorganisms for the biosynthesis of natural products. Crit Rev Biotechnol. 2022;0:1–15. https://doi.org/10.1080/07388551.2022.2029344.
Hausjell J, Halbwirth H, Spadiut O. Recombinant production of eukaryotic cytochrome P450s in microbial cell factories. Biosci Rep. 2018;38. https://doi.org/10.1042/bsr20171290.
Gillam EMJ, Guo ZY, Martin MV, Jenkins CM, Guengerich FP. Expression of cytochrome P450 2D6 in Escherichia coli, purification, and spectral and catalytic characterization. Arch Biochem Biophys. 1995;319:540–50. https://doi.org/10.1006/abbi.1995.1329.
Biggs BW, Lim CG, Sagliani K, Shankar S, Stephanopoulos G, Mey MD, et al. Overcoming heterologous protein interdependency to optimize P450-mediated taxol precursor synthesis in Escherichia coli. Proc Natl Acad Sci. 2016;113:3209–14. https://doi.org/10.1073/pnas.1515826113.
Larson JR, Coon MJ, Porter TD. Alcohol-inducible cytochrome P-450IIE1 lacking the hydrophobic NH2-terminal segment retains catalytic activity and is membrane-bound when expressed in Escherichia coli. J Biol Chem. 1991;266:7321–4. https://doi.org/10.1016/S0021-9258(20)89447-2.
Barnes HJ, Arlotto MP, Waterman MR. Expression and enzymatic activity of recombinant cytochrome P450 17 alpha-hydroxylase in Escherichia coli. Proc Natl Acad Sci. 1991;88:5597–601. https://doi.org/10.1073%2Fpnas.88.13.5597.
Haudenschild C, Schalk M, Karp F, Croteau R. Functional expression of Regiospecific Cytochrome P450 Limonene Hydroxylases from Mint (Mentha spp.) in Escherichia coli and Saccharomyces cerevisiae. Arch Biochem Biophys. 2000;379:127–36. https://doi.org/10.1006/abbi.2000.1864.
Sandhu P, Guo ZY, Baba T, Martin MV, Tukey RH, Guengerich FP. Expression of Modified Human cytochrome P450 1A2 in Escherichia coli: stabilization, purification, spectral characterization, and Catalytic Activities of the enzyme. Arch Biochem Biophys. 1994;309:168–77. https://doi.org/10.1006/abbi.1994.1099.
Gillam EMJ, Wunsch RM, Ueng Y-F, Shimada T, Reilly PEB, Kamataki T, et al. Expression of cytochrome P450 3A7 in Escherichia coli: Effects of 5′ modification and Catalytic characterization of recombinant enzyme expressed in Bicistronic Format with NADPH-Cytochrome P450 reductase. Arch Biochem Biophys. 1997;346:81–90. https://doi.org/10.1006/abbi.1997.0286.
Park H-G, Lim Y-R, Han S, Kim D. Expression and characterization of truncated recombinant human cytochrome P450 2J2. Toxicol Res. 2014;30:33–8. https://doi.org/10.5487/tr.2014.30.1.033.
Cheesman MJ, Baer BR, Zheng Y-M, Gillam EMJ, Rettie AE. Rabbit CYP4B1 engineered for high-level expression in Escherichia coli: ligand stabilization and processing of the N-terminus and heme prosthetic group. Arch Biochem Biophys. 2003;416:17–24. https://doi.org/10.1016/s0003-9861(03)00278-9.
Pernecky SJ, Larson JR, Philpot RM, Coon MJ. Expression of truncated forms of liver microsomal P450 cytochromes 2B4 and 2E1 in Escherichia coli: influence of NH2-terminal region on localization in cytosol and membranes. Proc Natl Acad Sci. 1993;90:2651–5. https://doi.org/10.1073/pnas.90.7.2651.
Cheesman MJ, Traylor MJ, Hilton ME, Richards KE, Taylor MC, Daborn PJ, et al. Soluble and membrane-bound Drosophila melanogaster CYP6G1 expressed in Escherichia coli: purification, activity, and binding properties toward multiple pesticides. Insect Biochem Mol Biol. 2013;43:455–65. https://doi.org/10.1016/j.ibmb.2013.02.003.
Pritchard MP, Ossetian R, Li DN, Henderson CJ, Burchell B, Wolf CR, et al. A General Strategy for the expression of recombinant human cytochrome P450s in Escherichia coli. Using bacterial Signal peptides: expression of CYP3A4, CYP2A6, and CYP2E1. Arch Biochem Biophys. 1997;345:342–54. https://doi.org/10.1006/abbi.1997.0265.
Christensen U, Vazquez-Albacete D, Søgaard KM, Hobel T, Nielsen MT, Harrison SJ, et al. De-bugging and maximizing plant cytochrome P450 production in Escherichia coli with C-terminal GFP fusions. Appl Microbiol Biotechnol. 2017;101:4103–13. https://doi.org/10.1007/s00253-016-8076-5.
Vazquez-Albacete D, Cavaleiro AM, Christensen U, Seppälä S, Møller BL, Nørholm MHH. An expression tag toolbox for microbial production of membrane bound plant cytochromes P450. Biotechnol Bioeng. 2017;114:751–60. https://doi.org/10.1002/bit.26203.
Ichinose H, Hatakeyama M, Yamauchi Y. Sequence modifications and heterologous expression of eukaryotic cytochromes P450 in Escherichia coli. J Biosci Bioeng. 2015;120:268–74. https://doi.org/10.1016/j.jbiosc.2015.01.019.
Zhou A, Zhou K, Li Y. Rational design strategies for functional reconstitution of plant cytochrome P450s in microbial systems. Curr Opin Plant Biol. 2021;60:102005. https://doi.org/10.1016/j.pbi.2021.102005.
Zelasko S, Palaria A, Das A. Optimizations to achieve high-level expression of cytochrome P450 proteins using Escherichia coli expression systems. Protein Expr Purif. 2013;92:77–87. https://doi.org/10.1016/j.pep.2013.07.017.
Halkier BA, Gershenzon J. Biology and Biochemistry of Glucosinolates. Annu Rev Plant Biol. 2006;57:303–33. https://doi.org/10.1146/annurev.arplant.57.032905.105228.
Petersen A, Crocoll C, Halkier BA. De novo production of benzyl glucosinolate in Escherichia coli. Metab Eng. 2019;54:24–34. https://doi.org/10.1016/j.ymben.2019.02.004.
Miki Y, Asano Y. Biosynthetic pathway for the cyanide-free production of phenylacetonitrile in Escherichia coli by utilizing plant cytochrome P450 79A2 and bacterial aldoxime dehydratase. Appl Environ Microbiol. 2014;80:6828–36. https://doi.org/10.1128/aem.01623-14.
Hansen CH, Du L, Naur P, Olsen CE, Axelsen KB, Hick AJ, et al. CYP83B1 is the Oxime-metabolizing enzyme in the glucosinolate pathway in Arabidopsis. J Biol Chem. 2001;276:24790–6. https://doi.org/10.1074/jbc.m102637200.
Mikkelsen MD, Buron LD, Salomonsen B, Olsen CE, Hansen BG, Mortensen UH, et al. Microbial production of indolylglucosinolate through engineering of a multi-gene pathway in a versatile yeast expression platform. Metab Eng. 2012;14:104–11. https://doi.org/10.1016/j.ymben.2012.01.006.
Halkier BA, Nielsen HL, Koch B, Moller BL. Purification and characterization of recombinant cytochrome P450TYR expressed at high levels in Escherichia coli. Arch Biochem Biophys. 1995;322:369–77. https://doi.org/10.1006/abbi.1995.1477.
Mikkelsen MD, Hansen CH, Wittstock U, Halkier BA. Cytochrome P450 CYP79B2 from Arabidopsis catalyzes the Conversion of Tryptophan to Indole-3-acetaldoxime, a Precursor of Indole Glucosinolates and Indole-3-acetic acid. J Biol Chem. 2000;275:33712–7. https://doi.org/10.1074/jbc.m001667200.
Andersen MD, Busk PK, Svendsen I, Møller BL. Cytochromes P-450 from cassava (Manihot esculenta Crantz) catalyzing the first steps in the biosynthesis of the cyanogenic glucosides linamarin and lotaustralin. Cloning, functional expression in Pichia pastoris, and substrate specificity of the isolated recombinant enzymes. J Biol Chem. 2000;275:1966–75. https://doi.org/10.1074/jbc.275.3.1966.
Wang C, Crocoll C, Agerbirk N, Halkier BA. Engineering and optimization of the 2-phenylethylglucosinolate production in Nicotiana benthamiana by combining biosynthetic genes from Barbarea vulgaris and Arabidopsis thaliana. Plant J. 2021;106:978–92. https://doi.org/10.1111/tpj.15212.
Geu-Flores F, Møldrup ME, Böttcher C, Olsen CE, Scheel D, Halkier BA. Cytosolic γ-Glutamyl peptidases process glutathione conjugates in the biosynthesis of glucosinolates and camalexin in Arabidopsis. Plant Cell. 2011;23:2456–69. https://doi.org/10.1105/tpc.111.083998.
Xie C, Zhong D, Chen X. A fragmentation-based method for the differentiation of glutathione conjugates by high-resolution mass spectrometry with electrospray ionization. Anal Chim Acta. 2013;788:89–98. https://doi.org/10.1016/j.aca.2013.06.022.
Mustafa G, Nandekar PP, Camp TJ, Bruce NJ, Gregory MC, Sligar SG, et al. Influence of transmembrane Helix mutations on cytochrome P450-Membrane interactions and function. Biophys J. 2019;116:419–32. https://doi.org/10.1016/j.bpj.2018.12.014.
Geu-Flores F, Nour-Eldin HH, Nielsen MT, Halkier BA. USER fusion: a rapid and efficient method for simultaneous fusion and cloning of multiple PCR products. Nucleic Acids Res. 2007;35:e55. https://doi.org/10.1093/nar/gkm106.
Cavaleiro AM, Kim SH, Seppälä S, Nielsen MT, Nørholm MHH. Accurate DNA Assembly and Genome Engineering with Optimized Uracil Excision Cloning. ACS Synth Biol. 2015;4:1042–6. https://doi.org/10.1021/acssynbio.5b00113.
Hallgren J, Tsirigos KD, Pedersen MD, Armenteros JJA, Marcatili P, Nielsen H, et al. DeepTMHMM predicts alpha and beta transmembrane proteins using deep neural networks. bioRxiv. 2022;20220408487609. https://doi.org/10.1101/2022.04.08.487609.
Studier FW. Protein production by auto-induction in high-density shaking cultures. Protein Expr Purif. 2005;41:207–34. https://doi.org/10.1016/j.pep.2005.01.016.
Bennett BD, Kimball EH, Gao M, Osterhout R, Van Dien SJ, Rabinowitz JD. Absolute metabolite concentrations and implied enzyme active site occupancy in Escherichia coli. Nat Chem Biol. 2009;5:593–9. https://doi.org/10.1038/nchembio.186.
R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. 2021. Available from: https://www.R-project.org/.
RStudio Team. RStudio: Integrated Development for R, RStudio. PBC, Boston. 2020. Available from: http://www.rstudio.com/.
Wickham H, Averick M, Bryan J, Chang W, McGowan LD, François R, et al. Welcome to the Tidyverse. J Open Source Softw. 2019;4:1686. https://doi.org/10.21105/joss.01686.
Wickham H. ggplot2: elegant graphics for data analysis. New York, NY: Springer; 2016. https://doi.org/10.1007/978-0-387-98141-3.
Batth TS, Keasling JD, Petzold CJ. Targeted proteomics for metabolic pathway optimization. Methods Mol Biol. 2012;944:237–49. https://doi.org/10.1007/978-1-62703-122-6_17.
We are thankful to William Thomas Wajn for helpful discussion of the study.
Open access funding provided by Royal Library, Copenhagen University Library. This work was supported by Danish National Research Foundation grant DNRF99 awarded to B.A.H. and Novo Nordisk Foundation grant NNF20OC0065061 awarded to B.A.H.
Open access funding provided by Royal Library, Copenhagen University Library
The authors declare there are no competing interests.
Ethics approval and consent to participate
Consent for publication
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
About this article
Cite this article
Poborsky, M., Crocoll, C., Motawie, M.S. et al. Systematic engineering pinpoints a versatile strategy for the expression of functional cytochrome P450 enzymes in Escherichia coli cell factories. Microb Cell Fact 22, 219 (2023). https://doi.org/10.1186/s12934-023-02219-7