Abstract
Background
Terpenes are one of the most diverse and abundant classes of natural biomolecules, collectively enabling a variety of therapeutic, energy, and cosmetic applications. Recent genomics investigations have predicted a large untapped reservoir of bacterial terpene synthases residing in the genomes of uncultivated organisms living in the soil, indicating a vast array of putative terpenoids waiting to be discovered.
Results
We aimed to develop a high-throughput functional metagenomic screening system for identifying novel terpene synthases from bacterial metagenomes by relieving the toxicity of terpene biosynthesis precursors to the Escherichia coli host. The precursor toxicity was achieved using an inducible operon encoding the prenyl pyrophosphate synthetic pathway and supplementation of the mevalonate precursor. Host strain and screening procedures were finely optimized to minimize false positives arising from spontaneous mutations, which avoid the precursor toxicity. Our functional metagenomic screening of human fecal metagenomes yielded a novel β-farnesene synthase, which does not show amino acid sequence similarity to known β-farnesene synthases. Engineered S. cerevisiae expressing the screened β-farnesene synthase produced 120 mg/L β-farnesene from glucose (2.86 mg/g glucose) with a productivity of 0.721 g/L∙h.
Conclusions
A unique functional metagenomic screening procedure was established for screening terpene synthases from metagenomic libraries. This research proves the potential of functional metagenomics as a sequence-independent avenue for isolating targeted enzymes from uncultivated organisms in various environmental habitats.
Similar content being viewed by others
Background
Terpenes are the largest and structurally most diverse group of natural organic products [1]. Their functions in living organisms are essential and as diverse as their structures, ranging from common metabolites to highly specialized functional molecules [1, 2]. Terpenes and their derivatives are commercially valuable resources for flavors, cosmetics, medicines, and biofuels [3]. Terpenes have commonly been extracted from natural sources for commercial production, but the extraction processes are often limited by low concentrations in these settings [2]. Alternative chemical syntheses of terpenes are expensive and low-yielding due to the structural complexity of terpene molecules [2, 4]. Therefore, microbial engineering for overproducing terpenes from inexpensive sugars has become a scalable, cost-effective, and environmentally friendly alternative to the conventional approaches [2].
Escherichia coli and Saccharomyces cerevisiae are the two host systems that have been most actively engineered as platforms for the biosynthesis of terpenes and their derivatives [5]. Both strains have native metabolic pathways generating prenyl pyrophosphate precursors for the terpene biosynthesis, namely, the 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway in E. coli and the mevalonate (MVA) pathway in S. cerevisiae [6]. Nevertheless, optimization of metabolic fluxes through the pathways is necessary to overproduce terpenes in those host systems [2]. In the case of E. coli, overexpression of key enzymes of the native MEP pathway or the heterogenous MVA pathway has successfully enhanced the production of terpenes and terpene derivatives [7,8,9,10]. However, excessive prenyl pyrophosphate pools resulting from the overexpression of key enzymes in the MEP and MVA pathways severely inhibit the growth of host E. coli; this precursor toxicity can be alleviated by channeling the excessive prenyl pyrophosphates into terpenes through an exogenous terpene synthase [9].
Identification of novel terpene synthases (TSs) can immensely expand the range of terpene products and their applications, together with rapid advances in platform microbial engineering [11,12,13,14]. However, compared to the high number of structurally characterized terpenes, the identification of terpene biosynthetic elements remains limited [15]. Thanks to rapidly growing genomic and metagenomic sequencing data, high-throughput prediction approaches based on amino acid sequence similarity have emerged to efficiently screen potential TS genes [16]. However, these approaches have been limited in the prediction of bacterial TSs because of their low levels of sequence similarity to plant and fungal TSs and their relatively low mutual similarities [17]. Previous studies mining TSs from bacterial genome databases using profile hidden Markov models (HMMs) demonstrated the promise of bacterial communities as bioresources of novel TSs [17,18,19].
In a complementary approach, Withers et al. showed the potential of function-based genome mining to screen biosynthetic elements of terpene and terpene derivatives from bacterial genomes. The system employed the toxicity of excessive prenyl pyrophosphates in E. coli as a selection pressure. Two isopentenol biosynthetic enzymes, which mitigate the growth inhibition by consuming the toxic precursors, were discovered from Bacillus subtilis genomic DNA libraries through this system [20]. Similarly, Furubayashi et al. developed another functional TS screening method based on the colorimetric changes of E. coli by active TSs that compete with carotenoid pathways for prenyl pyrophosphate precursors [21].
Here we present a functional metagenomics approach for screening TSs by finely optimizing the precursor toxicity mechanism [9, 20] and expanding the range of novel TS sources from single microorganisms to bacterial metagenomes. Our approach achieved a unique and effective platform for isolating TS genes from bioresources independently of bacterial cultures or genome databases (Fig. 1). We demonstrated the potential of functional metagenomic screening by isolating a novel β-farnesene synthase from a human fecal metagenomic library that does not show amino acid sequence similarity to known β-farnesene synthases.
Results and discussion
Optimization of the selection pressure for the precursor toxicity-based screening
We hypothesized that the toxicity of prenyl pyrophosphate pools, namely, precursors of terpene biosynthesis, could be harnessed to functionally search through the vastly uncharacterized genomic content of environmental microbial communities for enzymes catalyzing terpene biosynthesis. To test this hypothesis, we devised a precursor toxicity-based functional metagenomics approach (Fig. 1). We introduced the synthetic MBIS operon consisting of S. cerevisiae ERG12, ERG8, MVD1, and E. coli idi and ispA [9] under the control of an isopropylthio-β-galactoside (IPTG)-inducible promoter PlacUV5, to ensure the overproduction of prenyl pyrophosphates in E. coli. We used pA5c carrying chloramphenicol resistance and the p15A origin of replication as a backbone vector of this synthetic operon, considering its compatibility with pZE21-based metagenomic library plasmids containing the kanamycin resistance marker and pBR322 origin of replication. The resulting plasmid, pA5c–MBIS, enabled E. coli to biosynthesize prenyl pyrophosphates from supplemented mevalonate with IPTG induction (Fig. 2a). Two control plasmids were constructed to determine the optimal concentration of mevalonate as a selection pressure for isolating genes encoding TS. pZE21–AgBis carrying the bisabolene synthase gene from Abies grandis (AgBis), one of the best-characterized sesquiterpene synthases [10, 22], was constructed as a positive control. We substituted the green fluorescent protein (GFP) gene for AgBis to prepare a negative control plasmid pZE21–GFP. We tracked the growths of the two E. coli DH10B strains carrying pA5c–MBIS and one of the control plasmids on varied mevalonate concentrations in the medium from 0 to 20 mM, with or without induction of the MBIS operon. Mevalonate supplementation alone did not inhibit the growth of both control E. coli strains without induction (Fig. 2b). With induction, mevalonate hindered the growth of the negative control strain; mevalonate concentration was positively correlated with the time required for the negative control E. coli to reach the maximum specific growth rate (Fig. 2b and Additional file 1: Fig. S1). On the other hand, the growth inhibition by mevalonate was substantially alleviated in the positive control E. coli (Fig. 2b and Additional file 1: Fig. S1), corroborating the conversion of the excess prenyl pyrophosphate pools to bisabolene that is not toxic to E. coli [10]. 8 mM of mevalonate efficiently inhibited the growth of negative control without damage to the fitness of positive control (Fig. 2b, c).
Minimizing spontaneous mutations
Spontaneous mutations under lethal stresses may allow E. coli to evade the selection pressure of excessive prenyl pyrophosphates, obviating the need for TS activities from the library plasmid [23]. To minimize the spontaneous mutation rate during the precursor toxicity-based screening, we adopted a new E. coli host strain, LowMut, a reduced-genome strain lacking most genes irrelevant for laboratory applications, including active IS elements and error-prone DNA polymerases [24,25,26]. We compared the stabilities of the two host strains, DH10B and LowMut, on the screening plate after transformation with pA5c–MBIS and one of positive (pZE21–AgBis) and negative (pZE21–GFP) control plasmids. To avoid potential carry-over transmission of free mobile DNA elements, such as transposases, integrases, and recombinases [27] from the original cloning host into LowMut, plasmid transformation into LowMut, re-extraction, and re-transformation were repeated 3 times.
Positive controls of both host strains exhibited comparable colony numbers on the 8 mM mevalonate plate and the non-selective plate with no mevalonate (Fig. 3a). On the other hand, the negative control of LowMut exhibited a significantly lower colony number on the 8 mM mevalonate plate than that of DH10B (Fig. 3b), indicating that LowMut is a more suitable host than DH10B regarding genetic stability under the selection pressure of excessive prenyl pyrophosphates. Higher mevalonate concentrations than 8 mM diminished the number of colonies of both positive and negative control strains, regardless of the host system (Fig. 3a, b), similar to the trend observed in the liquid cultivations for optimizing the mevalonate concentration (Fig. 2b). We collected 4 colonies of the negative controls from each screening medium agar plate to check for potential mutations in the pA5c–MBIS plasmid responsible for generating prenyl pyrophosphates. Indeed, 14 among the 24 colonies contained transposon insertions that directly interrupted open reading frames (ORFs) of the MBIS operon in the pA5c–MBIS. All insertions were in the first (ERG12) or second (ERG8) ORFs, while the promoter PlacUV5 and following ORFs of the MBIS operon (Mvd1, idi, ispA) were not perturbed in any tested colonies (Fig. 3b). The other 10 colonies did not show any mutation on the MBIS pathway, suggesting that spontaneous mutations in the genomic DNA might rearrange the metabolism of E. coli to mitigate the toxicity of excessive prenyl pyrophosphates [23].
Validation of the optimized precursor toxicity-based screening procedure
To validate the capability of the optimized toxicity-based screening to isolate TSs from libraries containing both TS ORFs and non-TS ORFs, the screening procedure was simulated with well-characterized TSs and GFP. We adopted limonene synthase from Mentha spicata (MsLim) as a model monoterpene synthase in addition to the sesquiterpene synthase AgBis. To construct MsLim-expressing positive control, pZE21–MsLim was transformed into the LowMut pA5c–MBIS after 3 transformation–extraction passages through LowMut, as described above. Cells of the two LowMut-based positive controls expressing one of the model terpene synthases were mixed with approximately 105-fold more cells of LowMut-based negative control expressing GFP (Additional file 1: Fig. S2a, b). The toxicity-based screening on 1 mM IPTG and 8 mM mevalonate successfully recovered the positive control strains from the artificial library mixture (Fig. 3c and Additional file 1: Fig. S2). Still, a few colonies grown on screening plates were negative controls (Additional file 1: Fig. S2b, c), reaffirming that the LowMut host strain can overcome the designated selection pressure via other avenues than terpene synthase activity. The high recovery rate and decent selectivity validated in this simulation demonstrate the reliability of precursor toxicity-based screening as an avenue for isolating varied terpene synthases consuming farnesyl pyrophosphate (FPP) and geranyl pyrophosphate (GPP).
Screening of potential TSs from metagenomic libraries
Human fecal and soil metagenomic library plasmids were transformed into the LowMut pA5c–MBIS strain (see “Minimizing spontaneous mutations” section for detailed description). The resulting library cells were incubated on solid media screening plates. For the first screening, growth patterns of the library transformants were compared with the positive (LowMut pA5c–MBIS pZE21–AgBis) and negative (LowMut pA5c–MBIS pZE21–GFP) control strains, and the colonies showing similar growth to the positive control were selected as transformants expressing a potential TS. We double-checked the first screening outcomes by retransforming the screened library plasmids into fresh LowMut pA5c–MBIS cells and spotting each transformant on the screening plate. The library plasmids whose new transformants did not show uniform growth were excluded as false positives; their original colonies from the first screening might overcome the precursor toxicity through spontaneous mutagenesis (Fig. 3d). It has been previously shown that mixed colony sizes may represent high mutator phenotypes allowing a fast response to the toxicity even during the second screening [23]. 27 library plasmids were screened through the precursor toxicity-based functional metagenomic screening (Additional file 1: Fig. S3). The insert regions of screened library plasmids were sequenced, putative ORFs were predicted from assembled contigs, and their amino acid sequences were further analyzed to predict their potential functions.
Despite these screening steps, we identified false positives from the selected libraries which apparently evaded the selection pressure without bona fide TS activity. 6 library plasmids included ORFs predicted to encode multidrug and toxic compound extrusion (MATE), auxin efflux carrier (AEC), or ATP-binding cassette (ABC) transporters (Additional file 1: Fig. S3) which may transport excessive intracellular prenyl pyrophosphates out from the E. coli cells, thereby alleviating intracellular toxicity. We also found 2 library plasmids expressing a transcriptional repressor and a putative transposase (Additional file 1: Fig. S3) which might induce fast spontaneous mutations inactivating a part of the MBIS operon or rearranging host metabolism. In addition, some of the selected library plasmids carried ORFs that are likely too short to be potential TSs (< 450 bp) [28] or did not have any ORFs (Additional file 1: Fig. S3), probably owing to the spontaneous mutagenesis (Fig. 3a, b). Finally, we screened 10 unique TS candidates from 15 screened library plasmids (Additional file 1: Fig. S3).
Characterization of TS candidates
Amino acid sequences of all ORFs from the selected library plasmids were novel (see Additional file 1: Supplementary texts), except TS36R1, which has an identical amino acid sequence to a histidine phosphatase family protein identified from Faecalibacterium prausnitzii (NCBI Reference Sequence: WP_158395513.1). The other 9 novel proteins of potential TS activity were cloned into the pET28b( +) backbone for the functional validation of the selected candidates. However, the cloning of most selected ORFs was unsuccessful regardless of the location of the hexa histidine tag; their cloning yield was extremely low, and all the resulting plasmids either destroyed the start codon or frameshifted the insert. TS10F1 was the only protein whose ORF was successfully cloned into pET28b( +) among the 9 candidates.
The contig containing TS10F1 was discovered from multiple library plasmids of a human fecal metagenome (Fig. 4a). TS10F1 exhibited 75% identity to potential histidine phosphatase family protein of Prevotella pectinovora (NCBI Reference Sequence: WP_044075392.1), and the predicted catalytic core regions of the two proteins were conserved (Fig. 4a and Additional file 1: Fig. S4). We tested in vivo functionality of TS10F1 by cultivating recombinant BL21(DE3) carrying both pA5c–MBIS and pET28–TS10F1 (BM_TS10F1) in a defined medium with a dodecane overlay to capture potential terpene products. Positive control BM_AgBis (pA5c–MBIS and pET28–AgBis) and negative control BM_Empty (pA5c–MBIS and pET28) were cultivated in the same culture condition to define BM_TS10F1-specific terpene-like molecules; IPTG was not added to the BM_Empty culture due to the precursor toxicity by pA5c–MBIS. BM_TS10F1 generated more β-farnesene and its derivatives, such as farnesol and farnesol acetate, than control strains (Fig. 4b, d).
The β-farnesene production in detectable levels by the positive and negative control strains connotes the evasion of the toxic accumulation of FPP by the plasticity of terpenoid biosynthesis in E. coli [9, 29]. In addition, innate E. coli phosphatases catalyzing farnesol formation from FPP, such as AphA, PgpA, and PgpB [30], might cause the basal farnesol levels of the two control strains. Furthermore, chloramphenicol acetyltransferase, the chloramphenicol resistance marker in the pA5c–MBIS plasmid, catalyzes nonspecific acetylation toward terpene alcohols, such as farnesol acetate formation [31]. We hypothesized that TS10F1 increased β-farnesene production through its β-farnesene synthase activity, and the native promiscuous E. coli metabolic activities and above-mentioned enzymes caused additional farnesol and farnesol acetate accumulation by hydrating and esterifying the excess β-farnesene [32]. To test this hypothesis, we performed an enzyme assay of the purified TS10F1 (Fig. 4e). Indeed, TS10F1 generated β-farnesene from FPP, not from GPP, regardless of cofactors (Fig. 4f). There was no spontaneous conversion of FPP into β-farnesene (Additional file 1: Fig. S5), suggesting that TS10F1 possesses the β-farnesene synthase activity removing diphosphate from FPP (EC 4.2.3.47).
Farnesene overproduction by recombinant yeast expressing TS10F1
We selected S. cerevisiae as a new host to test the potential of TS10F1 in microbial β-farnesene production on a sugar-based carbon source without noticeable innate β-farnesene biosynthesis. The TS10F1 ORF was codon-optimized and introduced into an engineered S. cerevisiae SK1Ze that limits downstream metabolism of the MVA pathway to accumulate FPP and shunting upper glycolysis toward the pentose phosphate pathway to efficiently regenerate NADPH, the key cofactor of the MVA pathway (Fig. 4a) [33]. TS10F1 was episomally overexpressed under a strong constitutive promoter (pRS426_PCCW12–TS10F1) together with acetyl-CoA C-acetyl transferase (ERG10) and truncated 3-hydroxy-3-methylglutaryl-co-enzyme-A reductase (tHMG1), the key enzymes of the MVA pathway (Fig. 5a). The resulting strain SHE_TS10F1 and negative control SHE_Empty carrying an empty pRS426_PCCW12 plasmid (Additional file 1: Table S1) exhibited similar growth patterns on glucose (Fig. 5b − d). SHE_TS10F1 produced β-farnesene efficiently in the glucose consumption phase (95 h, 3.89 ± 0.19 mg/g glucose), while SHE_Empty did not produce β-farnesene at detectable levels (Fig. 5e). SHE_TS10F1 exhibited relatively inefficient β-farnesene production during ethanol utilization, and the level of produced β-farnesene decreased until the ethanol depletion timepoint, 215 h (Fig. 5e). The detected highest β-farnesene titer was 119.74 ± 2.85 mg/L at 165 h. Interestingly, we discovered farnesoic acid from the 215 h sample analysis (Fig. 5e), implicating that innate promiscuous metabolic activities of S. cerevisiae, which were repressed during the glucose phase [34] or ethanol-dependently activated [35], converted β-farnesene into other metabolites including farnesoic acid in the SHE_TS10F1 culture.
Future perspectives and challenges
This study demonstrated the potential of precursor toxicity-based functional metagenomic screening as a platform for bioprospecting novel TSs from uncultivated microbiota. We envision that this approach can be exploited further to identify additional novel TSs from other bioresources than the human fecal and soil samples used in this study. Furthermore, the selection pressure of functional metagenomic screening can be redesigned to discover other valuable metabolic functions residing in the metagenomes of uncultivated organisms, such as bioconversion of toxic compounds and decomposition of recalcitrant wastes [36, 37].
To enhance the functional metagenomic screening efficiency for TSs, improved control of false positives during the screening step would be required. Although the mutation rate of the LowMut strain is close to zero [24, 25], it could not eliminate spontaneous mutations resulting in toxicity mitigation without a TS activity (Fig. 2e). Additional safeguards, such as overexpression of dnaE [38], may reinforce the regulation of spontaneous mutagenesis and reduce the occurrence of false positives. Another area warranting further optimization is the low cloning efficiency of screened TS candidate ORFs which limited the number of characterizable candidates. We found indels destroying the expression cassette of TS candidates from few yielded constructs of the pET28b( +) cloning of TS candidates. These aberrant constructs with indels probably arose due to the metabolic consequence of the TS candidates [39]; they might be efficient enough to severely inhibit the E. coli growth even at the basal leaky expression level of the T7 system by depleting the native prenyl pyrophosphate pool without the overproduction of prenyl pyrophosphates via the MBIS pathway [40, 41]. Further studies with hosts and culture conditions enabling fine modulation of the prenyl pyrophosphate level, or the use of cell-free transcription–translation systems [42, 43], may be required to understand the potential toxicity of the screened candidates and to improve their cloning efficiency.
Conclusions
The current study exploits functional metagenomics for screening TSs from metagenomic libraries based on the toxicity of prenyl pyrophosphates, the universal precursors of terpenes. The selection pressure of the functional metagenomic screening was optimized by controlling prenyl pyrophosphates biosynthesis via the MBIS pathway and mevalonate supplementation. An ORF, TS10F1, encoding a novel β-farnesene synthase from a human fecal metagenomic library was discovered through functional metagenomic screening and subsequent functional validation. Recombinant S. cerevisiae expressing the TS10F1 successfully produced β-farnesene from glucose. Our results highlight the potential of functional metagenomics as a unique approach for screening targeted metabolic pathways from uncultivated organisms from diverse environmental habitats.
Methods
Routine culture and cloning conditions
E. coli MegaX DH10B (Thermo Fisher Scientific, Waltham, MA) was used for routine cloning and selection pressure optimization. E. coli LowMut (Scarab Genomics, Madison, WI) was used as a host strain for the functional metagenomic screening. E. coli BL21(DE3) (Thermo Fisher Scientific) was used for in vivo enzyme tests and protein production. S. cerevisiae SK1Ze [33] was used for the terpene production from glucose. Details of strains used in this study are described in Additional file 1: Table S1.
E. coli strains were cultured in Luria Bertani medium (LB, 5 g/L yeast extract, 10 g/L tryptone, and 5 g/L NaCl). 34 μg/mL of chloramphenicol, 50 μg/mL of kanamycin, or 100 μg/mL of ampicillin were supplemented if required. Unless otherwise mentioned, E. coli strains were grown at 37 °C and 250 rpm. Yeast peptone medium with glucose (YPD, 10 g/L yeast extract, 20 g/L peptone, 20 g/L of glucose) was used for routine yeast cultures. Yeast transformants carrying auxotrophic plasmids were grown in a synthetic complete medium containing yeast nitrogen base, complete supplement mixture without histidine, leucine, and uracil (MP Biomedicals, Santa Ana, CA), and 40 g/L of glucose (SCD-3). The SCD-3 medium was adjusted at pH 5.5 using 50 mM potassium hydrogen phthalate buffer. Yeast strains were grown at 30 °C and 250 rpm unless otherwise mentioned.
Sanger sequencing was performed by Genewiz (Azenta Life Sciences, Chelmsford, MA), PCR was performed using Q5 HighFidelity DNA polymerase (New England Biolabs, Ipswich, MA), and codon-optimization and DNA fragment synthesis were accomplished through Integrated DNA Technologies (Coralville, IA). Plasmids and primers used in this study are described in Additional file 1: Tables S1 and S2.
Optimization of the precursor toxicity
The MBIS operon [9] was expressed under the control of an IPTG-inducible promoter PlacUV5 in the pA5c backbone (pA5c–MBIS) to switch on growth inhibition by generating excessive prenyl pyrophosphates from mevalonate. The MBIS operon and pA5c backbone DNA fragments were prepared from pBbB5k–MBIS and pA5c–RFP by BamHI and EcoRI digestion. 1 M mevalonate solution was prepared by mixing 2 M mevalonate and 2 M KOH in a 1:1.02 ratio (v/v) and incubating at 37 °C for 30 min. Optimal mevalonate concentration for the screening process was determined through E. coli cultures with 200 μL of LB medium with kanamycin, chloramphenicol (LBKC), 0.5 mM of IPTG, and mevalonate with varied concentrations in clear bottom 96-well plates. E. coli growth was tracked every 5 min for 18 h at 37 °C using a Synergy H1 microplate reader (BioTek, Winooski, VT). The functional metagenomic screening was performed on solid media screening plates, composed of LBKC agar supplemented with 1 mM IPTG and 8 mM mevalonate, with plating of ~ 106 colony forming units of functional metagenomic library cells or control strains.
Simulation of precursor toxicity-based screening
Colony forming units (CFUs) of cell solutions of two positive controls (LowMut pA5c–MBIS pZE21–MsLim and LowMut pA5c–MBIS pZE21–AgBis) and one negative control (LowMut pA5c–MBIS pZE21–GFP) were measured via spotting assay on LBKC plates and frozen at − 80 °C with 20% glycerol. Two positive control strain cells were mixed with approximately 105-fold more cells of negative control strain cells. The mixed cells were washed in sterilized PBS twice before spreading on screening medium plates, namely, LBKC with 1 mM IPTG and 8 mM mevalonate. To measure the actual CFUs of each strain in the mixture, each strain was processed in an identical way but with sterilized PBS instead of cell solutions of other strains and spread on LBKC without IPTG and mevalonate. Colonies on screening medium plates were identified via colony PCR amplifying the insert region of pZE21 (Additional file 1: Table S2 and Fig. S2a, d). The recovery rate of each strain was calculated using the CFUs on screening medium plates and CFUs on LBKC.
Metagenomic library
Metagenomic library plasmids used in this study were previously prepared by cloning sheared metagenomic DNA fragments (2–5 kb) of human fecal specimens collected from El Salvador (40101, 40203, and 40301) or soil samples from Michigan (S18) [44, 45]. The backbone plasmid pZE21 constitutively expresses ORFs under the control of PLtetO-1 in both DH10B and LowMut host strains, both of which lack the tetracycline repressor [46].
Characterization of TS candidates
Metagenomic insert region sequences of the selected library plasmids were identified via Sanger sequencing. The sequencing was performed initially with primers 5253 and 5254 and continued with new primers designed based on the initial sequencing results. Amino acid sequences of the ORFs were analyzed using protein BLAST (https://blast.ncbi.nlm.nih.gov), ConSurf [47], and PyMOL 2.5.2 [48]. Selected ORFs from the library plasmids were PCR amplified and introduced into pET28b( +) (MilliporeSigma, St. Louis, MO) with N-terminal or C-terminal hexa histidine-tag for in vivo functionality tests and protein production.
To investigate putative products of the screened TS in vivo, BL21(DE3) transformed with pA5c–MBIS and the TS expressing plasmid was recovered in 5 mL LBKC at 37 °C overnight. Recovered cells were harvested, washed in sterilized PBS, and inoculated into 5 mL EZ rich defined medium (Teknova, Hollister, CA) with kanamycin and chloramphenicol in a glass tube at initial OD600nm 0.1. 12 mM mevalonate and 0.1 mM IPTG were supplemented at OD600nm 0.5. After adding 0.5 mL filtered dodecane, the cultures were continued at 20 °C. Dodecane layer was collected at 10-, 24-, 34-, and 48-h timepoints.
To perform in vitro enzyme assays, target TS protein was overproduced in BL21(DE3) without pA5c–MBIS. The production culture was performed in 200 mL LB with kanamycin (LBK) in a 1 L baffled flask at initial OD600nm 0.1, and IPTG was added to a final concentration of 0.1 mM at OD600nm 0.5. Then the culture was continued at 20 °C for 18 h. Target TS protein was extracted and purified using HisTALON™ Gravity Columns Purification Kit (Takara Bio, San Jose, CA). Purified protein samples were desalted using Amicon® Ultra-0.5 10 K (MilliporeSigma) and concentrated to 1 μg/μL in 2X reaction buffer (80 mM MgCl2, 8 mM MnCl2, 8 mM DTT, 50 mM HEPES, pH 7.2). The enzyme reaction was initiated in a glass vial by combining 100 μL of the concentrated enzyme solution with the mixture of 100 μL of the mixture of a potential substrate (GPP or FPP, 0.1 mM) and a cofactor (NADH, NADPH, 0.1 mM, or water) in water. To rule out the spontaneous conversion of substrates and define the baseline chromatogram for the gas chromatography–mass spectrometry (GC–MS) analysis of reaction products (see “Analysis of microbial metabolites and enzyme reaction products” section), we also prepared negative control reaction mixtures by adding 2X reaction buffer instead of the enzyme solution. After brief vortexing, 200 µL filtered dodecane was added to capture enzyme reaction products, and the dodecane layer was sampled at 0 and 3 h timepoints.
Terpene production in the engineered yeast
A selected TS ORF was codon-optimized, amplified, and introduced into pRS426_PCCW12 [33] for constitutive expression in yeast. tHMG1 and ERG10 coding sequences were amplified from S. cerevisiae S288C genomic DNA and introduced into pRS423_PTDH3 and pRS425_PTEF1 (EUROSCARF), respectively. The lithium acetate/single strand carrier DNA/polyethylene glycol method [49] was used for yeast transformations.
Yeast strains were recovered from frozen glycerol stocks on SCD-3 plates in a 30 °C incubator. A single colony of each strain was inoculated in 5 mL of SCD-3 broth and grown at 30 °C and 250 rpm overnight as a seed culture. For precultures, seed cultures were transferred into fresh 5 mL liquid SCD-3 at initial OD600nm 0.5 and incubated for additional 6 h. Main cultures were performed with 50 mL of SCD-3 in a 250 mL baffled flask at initial OD600nm 0.1. After inoculation, 5 mL of filtered dodecane was added to capture β-farnesene, and flasks were incubated at 30 °C, 300 rpm. Culture broth (bottom) was collected for tracking yeast cell growth, glucose, and soluble extracellular metabolites. The dodecane layer (top) was analyzed to quantify the targeted terpene and its hydrophobic derivatives. Intracellular hydrophobic metabolites were extracted from the cell pellets following the procedure previously described [50].
Analysis of microbial metabolites and enzyme reaction products
Terpenes and terpene derivatives collected in dodecane were analyzed through GC–MS composed of Agilent 7820A GC equipped with an HP-5 ms column and 5977E MSD (Agilent Technology, Wilmington, DE). The split ratio was 1:1, and filaments were heated at 250˚C for 2 min. The initial temperature of the column oven was 100˚C for 3 min, followed by a 10˚C/min ramp to 280˚C for 3 min. Dodecane samples from in vitro enzyme assays and yeast cultures were analyzed using 7890 GC and 5977B MSD (Agilent Technology) in identical conditions. Terpenes were identified using the National Institute of Standards and Technology (NIST) database. The identification results were confirmed and quantified using corresponding standard chemicals.
Microbial cell density in the culture broth was monitored by absorbance at 600 nm (A600) using DiluPhotometer (Implen, Westlake Village, CA). Glucose and extracellular metabolites in culture broth were quantified by HP 1050 HPLC system equipped with 1100 Refractive Index Detector (Agilent Technology), Rezex ROA Organic Acid H + (8%) column (Phenomenex, Torrance, CA), and 0.005 N H2SO4 as a mobile phase (0.6 mL/min and 50 °C).
Availability of data and materials
All assembled sequences have been deposited to Genbank and are available via the BioProject PRJNA835076 (https://www.ncbi.nlm.nih.gov/bioproject/835076).
Abbreviations
- MEP pathway:
-
2-C-methyl-D-erythritol 4-phosphate pathway
- MVA pathway:
-
Mevalonate pathway
- TS:
-
Terpene synthase
- AgBis:
-
Abies grandis Bisabolene synthase
- GFP:
-
Green fluorescent protein
- MsLim:
-
Mentha spicata Limonene synthase
- IPTG:
-
Isopropylthio-β-galactoside
- ORF:
-
Open reading frame
- MATE:
-
Multidrug and toxic compound extrusion
- AEC:
-
Auxin efflux carrier
- ABC:
-
ATP-binding cassette
- FPP:
-
Farnesyl pyrophosphate
- GPP:
-
Geranyl pyrophosphate
References
Gershenzon J, Dudareva N. The function of terpene natural products in the natural world. Nat Chem Biol. 2007;3:408–14.
Chang MCY, Keasling JD. Production of isoprenoid pharmaceuticals by engineered microbes. Nat Chem Biol. 2006;2:674–81.
Tippmann S, Chen Y, Siewers V, Nielsen J. From flavors and pharmaceuticals to advanced biofuels: production of isoprenoids in Saccharomyces cerevisiae. Biotechnol J. 2013;8:1435–44.
Helfrich EJN, Lin G-M, Voigt CA, Clardy J. Bacterial terpene biosynthesis: challenges and opportunities for pathway engineering. Beilstein J Org Chem. 2019;15:2889–906.
Wang C, Liwei M, Park J-B, Jeong S-H, Wei G, Wang Y, et al. Microbial platform for terpenoid production: Escherichia coli and yeast. Front Microbiol. 2018;9:2460.
Kirby J, Keasling JD. Biosynthesis of plant isoprenoids: perspectives for microbial engineering. Annu Rev Plant Biol. 2009;60:335–55.
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.
Anthony JR, Anthony LC, Nowroozi F, Kwon G, Newman JD, Keasling JD. Optimization of the mevalonate-based isoprenoid biosynthetic pathway in Escherichia coli for production of the anti-malarial drug precursor amorpha-4,11-diene. Metab Eng. 2009;11:13–9.
Martin VJJ, Pitera DJ, Withers ST, Newman JD, Keasling JD. Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nat Biotechnol. 2003;21:796–802.
Peralta-Yahya PP, Ouellet M, Chan R, Mukhopadhyay A, Keasling JD, Lee TS. Identification and microbial production of a terpene-based advanced biofuel. Nat Commun. 2011;2:483.
Keasling J, Garcia Martin H, Lee TS, Mukhopadhyay A, Singer SW, Sundstrom E. Microbial production of advanced biofuels. Nat Rev Microbiol. 2021;19:701–15.
Moser S, Pichler H. Identifying and engineering the ideal microbial terpenoid production host. Appl Microbiol Biotechnol. 2019;103:5501–16.
Rinaldi MA, Ferraz CA, Scrutton NS. Alternative metabolic pathways and strategies to high-titre terpenoid production in Escherichia coli. Nat Prod Rep. 2022;39:90–118.
Zhang C, Hong K. Production of terpenoids by synthetic biology approaches. Front Bioeng Biotechnol. 2020;8:347.
Bian G, Deng Z, Liu T. Strategies for terpenoid overproduction and new terpenoid discovery. Curr Opin Biotechnol. 2017;48:234–41.
Milshteyn A, Schneider JS, Brady SF. Mining the metabiome: identifying novel natural products from microbial communities. Chem Biol. 2014;21:1211–23.
Yamada Y, Kuzuyama T, Komatsu M, Shin-Ya K, Omura S, Cane DE, et al. Terpene synthases are widely distributed in bacteria. Proc Natl Acad Sci USA. 2015;112:857–62.
Komatsu M, Tsuda M, Omura S, Oikawa H, Ikeda H. Identification and functional analysis of genes controlling biosynthesis of 2-methylisoborneol. Proc Natl Acad Sci U S A. 2008;105:7422–7.
Reddy GK, Leferink NGH, Umemura M, Ahmed ST, Breitling R, Scrutton NS, et al. Exploring novel bacterial terpene synthases. PLoS ONE. 2020;15: e0232220.
Withers ST, Gottlieb SS, Lieu B, Newman JD, Keasling JD. Identification of isopentenol biosynthetic genes from bacillus subtilis by a screening method based on isoprenoid precursor toxicity. Appl Environ Microbiol. 2007;73:6277–83.
Furubayashi M, Ikezumi M, Kajiwara J, Iwasaki M, Fujii A, Li L, et al. A high-throughput colorimetric screening assay for terpene synthase activity based on substrate consumption. PLoS ONE. 2014;9: e93317.
Bohlmann J, Crock J, Jetter R, Croteau R. Terpenoid-based defenses in conifers: cDNA cloning, characterization, and functional expression of wound-inducible (E)-alpha-bisabolene synthase from grand fir (Abies grandis). Proc Natl Acad Sci U S A. 1998;95:6756–61.
Swings T, Van den Bergh B, Wuyts S, Oeyen E, Voordeckers K, Verstrepen KJ, et al. Adaptive tuning of mutation rates allows fast response to lethal stress in Escherichia coli. Elife. 2017;6: e22939.
Csörgo B, Fehér T, Tímár E, Blattner FR, Pósfai G. Low-mutation-rate, reduced-genome Escherichia coli: an improved host for faithful maintenance of engineered genetic constructs. Microb Cell Fact. 2012;11:11.
Pósfai G, Plunkett G, Fehér T, Frisch D, Keil GM, Umenhoffer K, et al. Emergent properties of reduced-genome Escherichia coli. Science. 2006;312:1044–6.
Umenhoffer K, Fehér T, Balikó G, Ayaydin F, Pósfai J, Blattner FR, et al. Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular and synthetic biology applications. Microb Cell Fact. 2010;9:38.
Frost LS, Leplae R, Summers AO, Toussaint A. Mobile genetic elements: the agents of open source evolution. Nat Rev Microbiol. 2005;3:722–32.
Jia Y, Xia D, Louzada ES. Molecular cloning and expression analysis of a putative terpene synthase gene from citrus. J Am Soc Hortic Sci. 2005;130:454–8.
Pérez-Gil J, Rodríguez-Concepción M. Metabolic plasticity for isoprenoid biosynthesis in bacteria. Biochem J. 2013;452:19–25.
Wang C, Yoon S-H, Shah AA, Chung Y-R, Kim J-Y, Choi E-S, et al. Farnesol production from Escherichia coli by harnessing the exogenous mevalonate pathway. Biotechnol Bioeng. 2010;107:421–9.
Zhou J, Wang C, Yoon S-H, Jang H-J, Choi E-S, Kim S-W. Engineering Escherichia coli for selective geraniol production with minimized endogenous dehydrogenation. J Biotechnol. 2014;169:42–50.
Guzmán GI, Utrilla J, Nurk S, Brunk E, Monk JM, Ebrahim A, et al. Model-driven discovery of underground metabolic functions in Escherichia coli. Proc Natl Acad Sci U S A. 2015;112:929–34.
Kwak S, Yun EJ, Lane S, Oh EJ, Kim KH, Jin Y-S. Redirection of the glycolytic flux enhances isoprenoid production in Saccharomyces cerevisiae. Biotechnol J. 2019;15: e1900173.
Kayikci Ö, Nielsen J. Glucose repression in Saccharomyces cerevisiae. FEMS Yeast Res. 2015. https://doi.org/10.1093/femsyr/fov068.
Lewis JA, Broman AT, Will J, Gasch AP. Genetic architecture of ethanol-responsive transcriptome variation in Saccharomyces cerevisiae strains. Genetics. 2014;198:369–82.
Igiri BE, Okoduwa SIR, Idoko GO, Akabuogu EP, Adeyi AO, Ejiogu IK. Toxicity and bioremediation of heavy metals contaminated ecosystem from tannery wastewater: a review. J Toxicol. 2018;2018:2568038.
Patel AK, Singhania RR, Albarico FPJB, Pandey A, Chen C-W, Dong C-D. Organic wastes bioremediation and its changing prospects. Sci Total Environ. 2022;824: 153889.
Fijalkowska IJ, Dunn RL, Schaaper RM. Mutants of Escherichia coli with increased fidelity of DNA replication. Genetics. 1993;134:1023–30.
Rood JI, Sneddon MK, Morrison JF. Instability in tyrR strains of plasmids carrying the tyrosine operon: isolation and characterization of plasmid derivatives with insertions or deletions. J Bacteriol. 1980;144:552–9.
Briand L, Marcion G, Kriznik A, Heydel JM, Artur Y, Garrido C, et al. A self-inducible heterologous protein expression system in Escherichia coli. Sci Rep. 2016;6:33037.
Campbell TL, Brown ED. Characterization of the depletion of 2-C-methyl-D-erythritol-2,4-cyclodiphosphate synthase in Escherichia coli and Bacillus subtilis. J Bacteriol. 2002;184:5609–18.
Bowie JU, Sherkhanov S, Korman TP, Valliere MA, Opgenorth PH, Liu H. Synthetic biochemistry: the bio-inspired cell-free approach to commodity chemical production. Trends Biotechnol. 2020;38:766–78.
Silverman AD, Karim AS, Jewett MC. Cell-free gene expression: an expanded repertoire of applications. Nat Rev Genet. 2020;21:151–70.
Forsberg KJ, Patel S, Gibson MK, Lauber CL, Knight R, Fierer N, et al. Bacterial phylogeny structures soil resistomes across habitats. Nature. 2014;509:612–6.
Pehrsson EC, Tsukayama P, Patel S, Mejía-Bautista M, Sosa-Soto G, Navarrete KM, et al. Interconnected microbiomes and resistomes in low-income human habitats. Nature. 2016;533:212–6.
Lutz R, Bujard H. Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic Acids Res. 1997;25:1203–10.
Ashkenazy H, Abadi S, Martz E, Chay O, Mayrose I, Pupko T, et al. ConSurf 2016: an improved methodology to estimate and visualize evolutionary conservation in macromolecules. Nucleic Acids Res. 2016;44:W344-350.
Schrödinger LLC. The PyMOL molecular graphics system, Version 2.5. 2021.
Gietz RD, Schiestl RH. High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nat Protoc. 2007;2:31–4.
Rodriguez S, Kirby J, Denby CM, Keasling JD. Production and quantification of sesquiterpenes in Saccharomyces cerevisiae, including extraction, detection and quantification of terpene products and key related metabolites. Nat Protoc. 2014;9:1980–96.
Acknowledgements
We thank members of the Dantas lab for their help with protocol optimization and for thoughtful discussions on the results and analyses presented herein. We appreciate the gifts pA5c–RFP (FZ105), pBbB5k–MBIS (FZ239), pW1a–AgBis (FZ260), and MsLim template from the laboratory of Dr. Fuzhong Zhang at the Department of Energy, Environmental & Chemical Engineering, Washington University in St. Louis. We thank the Edison Family Center for Genome Sciences & Systems Biology at Washington University School of Medicine in St. Louis staff (Eric Martin, Brian Koebbe, Jessica Hoisington-López, and MariaLynn Crosby) for technical support in high-throughput sequencing and computing.
Funding
Research reported in this publication was supported in part by awards to G.D. through the International Center for Advanced Renewable Energy and Sustainability at Washington University, and the NIH Director’s New Innovator Award (DP2-DK-098089). N.C. received support from the NIDDK Pediatric Gastroenterology Research Training Program of the NIH under award number T32-DK-077653 (Phillip I. Tarr, Principal Investigator). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
Author information
Authors and Affiliations
Contributions
SK, NC, AY, and GD conceived the study design, experiments, and analyses. SK, NC, AY, and NA optimized the functional metagenomic screening system. SK, NC, NA, JN, and JC performed the screening and characterization of terpene synthase candidates. SK, JN, and JC performed the construction, cultivation, and product analyses of engineered yeast strains. SK, NC and GD drafted the manuscript and figures. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Ethics approval and consent to participate
No applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Additional file 1.
Additional data.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Kwak, S., Crook, N., Yoneda, A. et al. Functional mining of novel terpene synthases from metagenomes. Biotechnol Biofuels 15, 104 (2022). https://doi.org/10.1186/s13068-022-02189-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13068-022-02189-9