Introduction

Increasing concerns of climate change and energy security have necessitated microbial biosynthesis to produce biofuels from renewable carbon source as a sustainable and stable alternative to the fossil fuel-based approaches [1, 2]. Biofuels research to date has predominantly focused on conversion of sugars (hexoses and pentoses), the primary components of deconstructed lignocellulosic biomass [3]. However, lignin is another major component of lignocellulosic biomass and its catabolism has also been extensively studied recently [4,5,6]. Development of a new microbial chassis that enables full utilization of the lignocellulosic biomass-derived carbon sources is critical to achieve economically viable biofuel production [7].

Pseudomonas putida KT2440 has recently emerged as a promising microbial host due to its capability of utilizing a broad range of carbon sources and its high tolerance to xenobiotics [8]. As P. putida is usually isolated from soils [9], the natural living environment conveys to P. putida versatile metabolism to degrade different types of substrates as carbon sources and it is adapted to tolerate various physicochemical stresses. Particularly, P. putida can utilize lignin-derived intermediates and aromatics, such as p-coumarate, benzoate, toluene as sole carbon sources, and thus has great potential to be developed as a new microbial workhorse to convert renewable carbon sources during bio-based production. P. putida KT2440 has been generally recognized as safe (GRAS) and is widely used for metabolic engineering studies as its full genome sequence is available [10]. It can share some genetic parts (plasmid backbone, promoter, RBS, etc.) with Escherichia coli, which could facilitate the genetic modification in P. putida. However, P. putida also showed different sugar metabolism from the model hosts that use classic glycolysis pathway, such as E. coli, Saccharomyces cerevisiae. P. putida oxidizes glucose to gluconate and 2-ketogluconate in the periplasm, followed by the phosphorylation to 6-phophogluconate (6PG) toward Entner–Doudoroff (ED) pathway [11] (Fig. 1). Due to the lack of phosphofructokinase (PFK) that catalyzes the rate-limiting phosphorylation of fructose-6-phosphate (F6P) to fructose-1,6-diphosphate (FBP) in glycolysis, P. putida does not catabolize glucose through the typical glycolysis but by the ED pathway [11].

Fig. 1
figure 1

Central metabolism and isoprenol production pathways in P. putida. Isoprenol production pathways, including the MEP pathway, the original MVA pathway, and the IPP-bypass MVA pathway are presented and the key engineering efforts to overexpress or knockout genes are presented in red. G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; FBP, fructose-1,6-diphosphate; G3P, glyceraldehyde-3-phosphate; 3PG, 3-phosphoglycerate; PEP, phosphoenolpyruvate; 2 KG, 2-ketoglutarate; MVA, mevalonate; MVAP, mevalonate phosphate; MVAPP, mevalonate diphosphate; IPP, isopentenyl diphosphate; DMAPP, dimethylallyl diphosphate; IP, isopentenyl phosphate; PHA, polyhydroxyalkanoate; MK, mevalonate kinase; PMK, phosphomevalonate kinase; PMD, phosphomevalonate decarboxylase; NudB, dihydroneopterin triphosphate diphosphatase. MvaES, HMGS and HMGR genes from Enterococcus faecalis; phaABC, PHA synthase

Full size image

Microbial production of isoprenoids has been considered a critical route for developing biofuels [12]. The biosynthesis of isoprenoids starts with two key isoprene units, isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP), naturally synthesized by two isoprenoid pathways: the mevalonate (MVA) pathway and non-mevalonate (methylerythritol 4-phosphate, MEP) pathway [13], respectively. IPP and DMAPP are precursors of isopentenols (C5) [14], and they can also be condensed to geranyl diphosphate (GPP) and farnesyl diphosphate (FPP) to produce monoterpenes (C10) [15, 16] and sesquiterpenes (C15) [17, 18], respectively. All C5, C10, and C15 isoprenoids are important candidates for fuel, especially jet fuel replacements [19]. Typical isoprenoid fuel molecules include isoprenol (C5) [20], limonene (C10) [15], 1,8-cineole (C10) [16], bisabolene (C15) [17], epi-isozizaene (C15) [18], etc. Among them, isoprenol (3-methyl-3-buten-1-ol) has received more attention due to its increasing applications as a valuable drop-in fuel molecule and important precursor of commodity chemicals. For example, US Navy has recently developed a high-performance jet fuel, 1,4-dimethylcyclooctane (DMCO), which can be produced from isoprenol [21]. Isoprenol is the dephosphorylated molecule of isopentenyl phosphate (IP) [22]. Unlike monoterpenes and sesquiterpenes, isoprenol synthesis does not require IPP which is toxic to cell growth [23]. An IPP-bypass pathway was developed to overcome the IPP toxicity and showed advantages in isoprenol (C5) production in both E. coli [24] and yeast [20].

Given that P. putida has emerged as a new workhorse strain, it has attracted interests for engineering of isoprenoid production [25,26,27]. For example, there have been a few literatures for isoprenoids production in P. putida, and mostly, the endogenous MEP pathway was engineered for the isoprenoids production and frequently focused on the oxidation of terpenes using P450 enzymes as P. putida is known to be tolerant to oxidative stress [25]. The heterologous MVA pathway was also expressed in P. putida, but the performance was not as good as what was shown in E. coli when a similar engineering strategy was attempted, and only a low productivity and titers were achieved [26].

In this study, we engineered the heterologous MVA pathway in P. putida KT2440 to produce isoprenoids, including isoprenol (C5) and epi-isozizaene (C15). We compared the differences among the MEP, MVA, and IPP-bypass MVA pathways during isoprenol production (Fig. 1). Since isoprenol can be utilized as a carbon source by P. putida KT2440 [28], we investigated strategies to prevent isoprenol self-consumption. Metabolomics was performed to reveal the metabolic difference during isoprenol degradation. We also showed the engineered P. putida can produce isoprenol using p-coumarate as the sole carbon source. Our results showed that P. putida is a promising microbial chassis for isoprenoids production with the improved capability of carbon utilization from lignocellulosic biomass for biofuel production.

Materials and methods

Strains and plasmid construction

All strains and plasmids used in this study are listed in Table 1. Strains and plasmids along with their associated information have been deposited in the public version of the JBEI Registry (https://public-registry.jbei.org; entries JPUB_019914 to JPUB_019988) and are available from the authors upon request. P. putida KT2440 was used for isoprenoid production, and E. coli DH5α was used for the general cloning.

Table 1 Strains and plasmids used in this study
Full size table

Transformation of P. putida was performed by electroporation using a Bio-Rad (Bio-Rad Laboratories, Hercules, CA) MicroPulser preprogrammed EC3 setting (0.2 cm cuvettes with 50 μL cells, ~ 5 ms pulse, 3.0 kV) [29]. LB medium and LB agar medium were used for cell outgrowth and colony selection at 30 °C, respectively. Kanamycin (50 µg/mL) or gentamicin (30 µg/mL) was used as the selective antibiotics when needed. Gene knockout of P. putida was performed based on the homologous recombination followed by a suicide gene (sacB) counter-selection using modified pK18-mobSacB plasmids [30]. The genotypes of gene-knockout mutants were confirmed by colony PCR using specific primers, followed by DNA sequencing (GENEWIZ, South San Francisco, CA, USA).

Isoprenol production in P. putida

An overview figure of typical process of isoprenol production and analysis is presented in the Supplementary information.

P. putida KT2440 strains bearing isoprenol pathway plasmids (Table 1) were used for isoprenol production. Starter cultures of all production strains were prepared by growing single colonies in LB medium containing 50 µg/mL kanamycin at 30 °C with 200-rpm shaking overnight. The starter cultures were diluted in 5 mL EZ-rich defined medium (Teknova, CA, USA) or M9 minimal medium [29], containing 10 g/L or 20 g/L glucose (1% or 2%, w/v), 25 µg/mL kanamycin in 50-mL culture tubes, and 0.5 mM IPTG was added to induce protein expression with OD600 at 0.4–0.6. When strains were cultivated in a 24-well microtiter plate, 2 mL medium was used and the plate was sealed with a gas-permeable film (Sigma-Aldrich, St. Louis, MO). When strains were cultivated in a 250-mL shake flask, 50-mL medium was used. l-glutamate was supplemented into the minimal medium at the indicated concentration when needed. For isoprenol production using p-coumarate as the carbon source, 10 g/L or 20 g/L (1% or 2%, w/v) p-coumarate was used to replace glucose in the EZ-rich defined medium. The P. putida cultures were incubated in rotary shakers (200 rpm) at 30 °C for 48 h.

Evaluation of isoprenol consumption

P. putida strains (Table 1) were used to investigate isoprenol consumption. Starter cultures were prepared by inoculating glycerol stocks in LB medium at 30 °C with 200-rpm shaking overnight. The starter cultures were diluted with OD600 at 0.01 in 5 mL M9 minimal medium or EZ-rich defined medium (Teknova, CA, USA) containing 10 g/L glucose (1%, w/v) or no glucose (0%, w/v), added with 1 g/L isoprenol in 50-mL culture tubes. Amino acids (Additional file 1: Table S1) were added individually into the M9 minimal medium at desirable concentrations when needed. The P. putida cultures were incubated in rotary shakers (200 rpm) at 30 °C for 48 h. Blank media without strain inoculation were used in parallel to evaluate isoprenol evaporation loss.

Quantification of isoprenol

The measurement and quantification of isoprenol were conducted by collecting 250 µL of cell culture and combining it with 250 µL of ethyl acetate containing 1-butanol (30 mg/L) as an internal standard. The mixture of ethyl acetate and cell culture was vigorously shaken for 15 min and subsequently centrifuged at 21,130 g for 3 min to separate the ethyl acetate phase from the aqueous phase. The ethyl acetate layer was collected and 1 µL was analyzed by gas chromatography-flame ionization detection (GC-FID, Thermo Focus GC) equipped with DB-WAX column (15 m, 0.32 mm inner diameter, 0.25 µm film thickness, Agilent, USA). The GC oven was programmed as follows: 40–100 °C at 15 °C/min, 100–230 °C at 40 °C/min, held at 230 °C for 2 min. The inlet temperature was 200 °C.

Production and quantification of epi-isozizaene

P. putida KT2440 bearing the pathway plasmid (Table 1) was used for epi-isozizaene production. Starter cultures of all production strains were prepared by growing single colonies in LB medium containing 50 µg/mL kanamycin at 30 °C with 200-rpm shaking overnight. The starter cultures were diluted in a 5 mL EZ-rich defined medium (Teknova, CA, USA) containing 10 g/L glucose (1%, w/v), 25 µg/mL kanamycin in 50-mL culture tubes. 0.5 mM IPTG was added to induce protein expression with OD600 at 0.4–0.6, and 0.5 mL nonane (10%, v/v) was added as a solvent overlay. The P. putida cultures were incubated in rotary shakers (200 rpm) at 30 °C for 72 h.

For epi-isozizaene measurement, the solvent overlay was sampled and centrifuged at 21,130 g for 3 min. The overlay layer was collected and diluted with ethyl acetate containing 5 mg/L guaiazulene as the internal standard. 1 µL was analyzed by Agilent GC–MS equipped with HP-5 column (Agilent, USA). The GC oven was programmed from 40 °C (held for 3 min) to 295 °C at 15 °C/min. The concentration of epi-isozizaene was estimated using the TIC areas with alternative standard (−)-trans-caryophyllene as described in a previous study [31].

Quantification of metabolites

The concentrations of glucose and organic acids from the culture were measured with an Agilent 1100 Series HPLC system, equipped with an Agilent 1200 Series refractive index detector (RID) (Agilent Technologies, CA) and Aminex HPX-87H ion-exclusion column as described in a previous study [32]. The quantification of glucose and organic acids was calibrated with authentic standards.

For metabolomics analysis, 1.5 mL cell culture was collected at 24 and 48 h and centrifuged at 13,000 g for 1 min at room temperature. The cell pellet was quenched with 250 µL methanol, vortexed, and stored at − 20 °C. For sample preparation, 250 µL water was added to the methanol lysate and mix thoroughly. Centrifuge the methanol/water lysate at 13,000 g for 10 min at 4 °C. The supernatant was filtered by a Millipore Amicon Ultra 3 kDa cut-off filter (Billerica, MA) at 13,000 g at − 2 °C for 30–60 min until most of the sample has been filtered. The intracellular metabolite concentrations were quantified by liquid chromatography and mass spectrometry (LC–MS) methods as previously described by Baidoo et al. (with reference to note 6) [33].

Results

Engineering P. putida for isoprenol production

P. putida natively possesses the MEP pathway for isoprenoids biosynthesis. To produce isoprenol in P. putida, we first attempted to use the endogenous MEP pathway and overexpressed the E. coli dihydroneopterin triphosphate diphosphatase (NudB) that has a promiscuous activity to catalyze the conversion of IPP to IP which is hydrolyzed to isoprenol by endogenous phosphatases [22]. In this case, P. putida KT2440 was transformed with a high-copy plasmid pBbB1k-NudB (Table 1) using a modified broad host range replication origin BBR1 [34] and a Trc promoter which works both in E. coli and P. putida. The engineered P. putida strain (JPUB_019966, Table 1, Fig. 2A) could produce a low level of isoprenol at 2 mg/L after 48 h from 1% glucose (Fig. 2C).

Fig. 2
figure 2

Engineering heterologous pathway for isoprenol production in P. putida. A Configuration of P. putida strains with the engineered isoprenol pathway plasmids. B–D Production results by the engineered P. putida strains from 1% glucose. B Cell growth. The initial OD600 of 0 h was set at 0.05. C Isoprenol production. Isoprenol levels of different strains were not detectable at 0 h; D Glucose consumption. Error bars indicate one standard deviation of triplicates. NudB, dihydroneopterin triphosphate diphosphatase (E. coli); AtoB, acetoacetyl-CoA synthase (E. coli); HMGS, HMG-CoA synthase; HMGR, HMG-CoA reductase; MvaS, HMG-CoA synthase; MvaE, HMG-CoA reductase; MK, mevalonate kinase; PMK, phosphomevalonate kinase; PMD, phosphomevalonate decarboxylase. The footnote of enzymes indicates their sources: Sc, S. cerevisiae; Sa, Staphylococcus aureus; Ef, Enterococcus faecalis. BBR1, broad host range replication origin; KanR, kanamycin-resistant antibiotic marker

Full size image

We then engineered a heterologous MVA pathway, which has shown high isoprenol production in E. coli [14]. To construct the MVA pathway, two operons were used to express the MVA pathway genes onto the plasmid backbone of pBbB5k. The expression of the top portion of the MVA pathway (AtoB, HMGS, HMGR) was driven by a LacUV5 promoter, and the expression of the bottom portion enzymes (MK, PMK, PMD) as well as NudB were driven by a Trc promoter. The resulting engineered P. putida strain (JPUB_019967, Table 1, Fig. 2A) produced up to 12 mg/L isoprenol at 24 h and the titer decreased at a later time point (Fig. 2C).

Finally, we engineered the IPP-bypass MVA pathway to compare the isoprenol production by using the promiscuous activity of PMD [22] in P. putida. Three different MVA pathway top-portion operons (MevT, MTSA, and MvaES) were studied, which the HMGS and HMGR genes are from Saccharomyces cerevisiae, Staphylococcus aureus, and Enterococcus faecalis, respectively (JPUB_019968 to JPUB_019971, Table 1, Fig. 2A，Additional file 1: Figure S9). Results showed that the highest isoprenol production (up to 74 mg/L after 24 h) was observed in the strain with the MvaES top portion operon (Fig. 2C). Compared with the strain using the original MVA pathway, strains with the IPP-bypass MVA pathways (except the one using the MTSA top portion operon) showed a 4- to 6-fold increase of isoprenol production. This suggests the IPP-bypass MVA pathway can be used in P. putida for isoprenol production and it shows higher efficiency than the original MVA pathway or the endogenous MEP pathway.

Optimization of isoprenol production in P. putida

Given that E. coli has shown much higher isoprenol production [24] than what we achieved in P. putida, we compared the metabolic difference between P. putida and E. coli to identify limiting steps and target them to optimize isoprenol production in P. putida. We used the published 13C-metabolic flux data of P. putida [11] and E. coli [35] for the comparison (Additional file 1: Figure S1). Interestingly, we found P. putida derived threefold more carbon flux from acetyl-CoA to TCA cycle compared with E. coli, which indicated less acetyl-CoA availability in P. putida for isoprenol production. Another difference is P. putida possesses both phosphoenolpyruvate carboxylase (ppc) and pyruvate carboxylase (pyc) that can direct 2.8-fold more flux from glycolysis to the TCA cycle, whereas the pyc gene does not exist in E. coli. In addition, P. putida can naturally synthesize polyhydroxyalkanoate (PHA) from acetyl-CoA as a carbon sink [36]. Therefore, we attempted to knockout ppc, pycAB, and phaABC (PHA synthase) genes to improve acetyl-CoA pool and isoprenol production in P. putida. As the Additional file 1: Figure S2 shows, the knockout of ppc and pycAB genes did not improve isoprenol production compared with the wild-type strain. The double knockout of ppc and pycAB genes produced an even lower amount of isoprenol. However, the deletion of phaABC genes increased 24% of isoprenol production compared with the wild-type strain during the screening in a microtiter plate, which suggests they are promising targets.

On the other hand, we noticed that the production results in the previous section showed decreased isoprenol levels and depleted glucose after 24 h (Fig. 2C). This indicated 1% glucose concentration might be insufficient to support a 48-h production process. Thus, we increased the initial glucose concentration to 2%, and the IPP-bypass MVA pathway with the MevT operon improved the isoprenol production level from 9 mg/L to 47 mg/L after 48 h (Fig. 3). When using the ΔphaABC strain, isoprenol production was further increased to 86 mg/L from 2% glucose at 48 h (Fig. 3). By applying this new condition (ΔphaABC strain + 2% glucose) to the best producing pathway (IPP-bypass_MvaES), the engineered P. putida strain (JPUB_019974, Table 1) reached 101 mg/L isoprenol production after 48 h (Fig. 3), which was a ~ 2-fold improvement of isoprenol production from the starting conditions (53 mg/L from the wild-type strain and 1% glucose).

Fig. 3
figure 3

Optimizing isoprenol production. Isoprenol production from 2% glucose with different background strains, top-portion MVA pathway, and MK-PMD genes. Error bars indicate one standard deviation of triplicates. MK, mevalonate kinase; PMD, phosphomevalonate decarboxylase. Sc, S. cerevisiae; Mm, Methanosarcina mazei; HKQ, a mutant of PMDSc containing three mutations (R74H, R147K, and M212Q)

Full size image

Given that MK and PMD are key steps to converting MVA to isoprenol, we also tested different combinations of the MK-PMD gene cassettes to optimize isoprenol production. Based on previous results in E. coli [24], we selected two efficient enzyme versions, MKMm (MK from Methanosarcina mazei) and PMDHKQ (a mutant of PMDSc containing three mutations [37]) to construct four combinations of the MK-PMD cassette. Results showed that the strain with MKMm-PMDHKQ (JPUB_019977, Table 1) produced the highest isoprenol at 104 mg/L after 48 h from 2% glucose in a culture tube (Fig. 3). More production details were studied by culturing this best producer in a shake flask. As the Additional file 1: Figure S3 shows, glucose was not fully depleted and ~ 2.5 g/L of residual glucose was detected in the culture after 48 h, which suggested the initial glucose concentration at 2% was sufficient in supporting a 48-h production. No significant amounts of organic acids were detected as fermentative by-products except the small amount of acetate (0.8 g/L) and succinate (0.2 g/L) observed only at 24 h. While the isoprenol titer was lower in the shake flask (80 mg/L), it might be attributed to potentially faster isoprenol evaporation in the flask than in the culture tube. Unlike the top-portion MVA pathway, changing MK-PMD genes did not significantly improve isoprenol production. Collectively, we engineered the IPP-bypass MVA pathway in P. putida KT2440 for isoprenol production and achieved the highest production titer from glucose at up to 104 mg/L (c.f. the maximum theoretical yield from glucose is 0.319 g/g glucose [38]).

Investigation of isoprenol consumption in P. putida

While the above isoprenol production was performed in the EZ-rich defined medium, it is also important to perform the production in the minimal medium, which is more frequently used for bioreactor fermentation and metabolic flux analysis [24]. Using the highest isoprenol producer (JPUB_019977, Table 1) from the EZ-rich defined medium, we tested isoprenol production in M9 minimal medium but observed low levels of isoprenol (~ 1 mg/L) after 48 h from 2% glucose (Additional file 1: Figure S4). Since P. putida has shown the capability of utilizing isoprenol as a carbon source [28], this urged us to investigate the difference between the two media that were used for isoprenol production. We first compared isoprenol consumption in the M9 minimal medium and EZ-rich medium. It was observed that the addition of glucose could help to preserve isoprenol in the medium, and the consumption was significantly slower in the EZ-rich medium when glucose is present (5 mg/L/hour) than in the M9 minimal medium (11 mg/L/hour) in 48 h (Fig. 4A).

Fig. 4
figure 4

Investigation of isoprenol consumption in P. putida. A Isoprenol consumption by P. putida ΔphaABC strain (JPUB_019964) in M9 minimal medium and EZ-rich medium containing 1% glucose or no glucose, respectively. B Isoprenol consumption by P. putida ΔphaABC strain in M9 minimal medium supplemented with individual amino acid. The working concentrations of 8 amino acids were shown in Additional file 1: Table S1. Control, no amino acid supplemented; Mix, the mixture of all 8 amino acids with the same individual concentration. Error bars indicate one standard deviation of triplicates

Full size image

To find out which other component of the EZ-rich medium contributed to slowing down the isoprenol consumption, we compared the recipes of two media and identified 8 amino acids that are present at a higher concentration in the EZ-rich medium formulation (Additional file 1: Table S1). By supplementing these 8 amino acids individually into the M9 minimal medium at the same concentration used in the EZ-rich medium, surprisingly, we found that the addition of L-glutamate (L-Glu) or L-glutamine (L-Gln) preserved isoprenol to a similar level that was observed in the EZ-rich medium (Fig. 4B). We chose L-Glu as the supplement to investigate isoprenol production in the minimal medium and observed that the addition of 6 mM L-Glu resulted the highest isoprenol production level to 15 mg/L after 48 h (Additional file 1: Figure S4), which is nearly a 15-fold increase compared with the previous level without any supplements (~ 1 mg/L).

Based on the findings of the L-Glu supplementation experiment, we continued to investigate the mechanism that L-Glu involves in isoprenol preservation in P. putida. We compared the intracellular metabolites between the conditions with and without the L-Glu supplement. When isoprenol is presented in the medium without the L-Glu supplement, it showed a significant difference in metabolites of central carbon and energy metabolism after 24 h (Fig. 5). Although isoprenol could provide an additional carbon source, the difference in pyruvate, succinate, malate, ATP, NADH, and NAD+ levels indicated an insufficient energy supply and imbalanced redox compared with the control group. In contrast, supplementing L-Glu restored those metabolites to comparable levels to the control group (Fig. 5). Since L-Glu is considered as a favored carbon source for P. putida [39], the L-Glu-mediated prevention of isoprenol self-consumption could be attributed to carbon catabolite repression (CCR). To verify this, we deleted the CCR regulator gene (crc) from the P. putida chromosome and studied the isoprenol consumption with the Δcrc strain (JPUB_019978, Table 1). Results showed the prevention of isoprenol from self-consumption by supplementing L-Glu was significantly reduced when crc is deleted (222 mg/L, Additional file 1: Figure S5B), compared with the strain without crc deletion (523 mg/L, Fig. 4B). This suggests L-Glu assisted isoprenol preservation in P. putida may be attributed to CCR, in which L-Glu is a preferred carbon source, rather than isoprenol, in supporting rapid cell growth [40]. However, overexpressing crc with the isoprenol pathway did not increase isoprenol production, but even lowered the production titer (6 mg/L, Additional file 1: Figure S6).

Fig. 5
figure 5

Metabolite analysis of l-glutamate supplement during isoprenol consumption in P. putida (JPUB_019964). Metabolites under four conditions were investigated after 24 h. Control, no isoprenol and no l-glutamate added; Isoprenol only, 1 g/L isoprenol added; l-Glu only, 6 mM l-glutamate added; Isoprenol + l-Glu, 1 g/L isoprenol and 6 mM l-glutamate added. Error bars indicate one standard deviation of triplicates

Full size image

Isoprenol production using p-coumarate as a carbon source

P-Coumarate is a prominent compound used as a representative lignin-derived aromatics and there are efforts to increase p-coumarate content in lignocellulosic biomass [41]. We attempted to use p-coumarate as the carbon source to investigate isoprenol production in the engineered P. putida strain. Results showed that the engineered P. putida strain (JPUB_019977, Table 1) can produce up to 25 mg/L isoprenol from 2% p-coumarate (c.f. the maximum theoretical yield from p-coumarate is 0.273 g/g p-coumarate) after 48 h (Fig. 6), which is 24% of the isoprenol titer achieved from 2% glucose. We observed that the cell growth at 2% p-coumarate was 27% lower than the 1% p-coumarate condition after 48 h. More residual p-coumarate was detected in the medium for the 2% condition (Fig. 6), suggesting a higher concentration of p-coumarate may inhibit cell growth. Though the isoprenol titer was lower from p-coumarate than from glucose, it showed the possibility of utilizing aromatics as well as sugars as the carbon source in biofuel production. This demonstrates that P. putida is a promising host for the comprehensive conversion of carbons from lignocellulosic biomass for bio-based production.

Fig. 6
figure 6

Isoprenol production from p-coumarate in the engineered P. putida. A p-Coumarate catabolic pathway; B Isoprenol production by engineered P. putida (JPUB_019977). EZ-rich medium was used as the base medium by changing the carbon source to 1% or 2% p-coumarate. The blank EZ-rich medium without any carbon sources was used as a control. Error bars indicate one standard deviation of triplicates. The initial OD600 of 0 h was set at 0.05. Isoprenol levels were not detectable at 0 h

Full size image

Engineering P. putida for other larger terpenes production

To expand the isoprenoid production profile in P. putida via the MVA pathway, we engineered the MVA pathway for monoterpenes and sesquiterpenes. We chose two monoterpenes (limonene, C10: 49 mg/L, 1,8-cineole, C10: 6 mg/L; Additional file 1: Figure S8) and one sesquiterpene (epi-isozizaene, C15) as targets for production. To demonstrate the capability of producing the larger terpene molecule (C15), we performed a similar strategy in engineering epi-isozizaene production. Herein, the MEP pathway was used as a control by overexpressing the epi-isozizaene synthase. As shown in Fig. 7, the MVA pathway (2 mg/L) showed a higher level of epi-isozizaene than the MEP pathway (1 mg/L). By applying the ΔphaABC strain for the MVA pathway, a higher production for epi-isozizaene (5 mg/L) was observed from 1% glucose, which is consistent with the results obtained from isoprenol production. However, compared with isoprenol, the production level of epi-isozizaene was much lower, and more efforts are needed to optimize the pathway for sesquiterpenes.

Fig. 7
figure 7

Production of sesquiterpene in the engineered P. putida. A Epi-isozizaene synthesis pathway from glucose. B–D Production results by the engineered P. putida strains from 1% glucose. B Cell growth; C Epi-isozizaene production; D Glucose consumption. Error bars indicate one standard deviation of triplicates. FPP, farnesyl diphosphate; FPPS, farnesyl diphosphate synthase; EizS, epi-isozizaene synthase from Streptomyces coelicolor. Error bars indicate one standard deviation of triplicates

Full size image

Discussion

In this study, we engineered the heterologous MVA pathway in P. putida KT2440 to produce isoprenoids, including isoprenol (C5) and epi-isozizaene (C15). Unlike the E. coli system, the use of a heterologous MVA pathway showed very limited improvement of isoprenoid production (Figs. 2, 7). These results are consistent with a previous report [26] and the reason might be the result of the distinct central metabolism in P. putida and its different flux distribution with acetyl-CoA (Additional file 1: Figure S1). For isoprenol, we also engineered the IPP-bypass MVA pathway, and it showed advantages compared with using the MEP and the original MVA pathway. The highest isoprenol titer from engineered P. putida was 104 mg/L from 2% glucose.

While the use of the IPP-bypass MVA pathway made a substantial improvement during isoprenol production, this is still much lower than the batch culture titer (~ 2,500 mg/L) reported in E. coli [24]. Compared with the E. coli system, the low isoprenol titer might be attributed to two reasons. First, the isoprenol degradation pathway in P. putida competes with the synthesis pathway, leading to a reduced accumulation of isoprenol. In contrast, E. coli does not show the capability of consuming isoprenol as a carbon source. Due to isoprenol consumption being very significant in P. putida (up to 714 mg/L isoprenol was consumed in 24 h, Fig. 4A), this could be one of the main reasons that the engineered P. putida cannot show a comparable isoprenol titer to the similarly engineered E. coli strains. We tried deleting a gene (PP_2675) reported to be associated with P. putida’s growth on isoprenol [28], but this deletion did not improve the isoprenol titer (Additional file 1: Figure S7). As many genes have been identified as being involved in isoprenol catabolism [28], additional gene deletion may be required to achieve reduction in isoprenol catabolism and degradation without compromising isoprenol production. Second, the balancing of isoprenol-pathway enzymes was harder to achieve in P. putida than in E. coli as the number and the variety of plasmids are limited in P. putida. In the E. coli system, two plasmids could be used for isoprenol production to achieve a well-balanced pathway proteins expression for the best isoprenol production. For example, the 1st plasmid was selected to be a medium-to-low copy plasmid to drive the top MVA portion genes (atoB, HMGS, HMGR) to reduce accumulation of the final metabolite (mevalonate) and prevent the substrate inhibition against the next enzyme (MK) in the pathway. The 2nd plasmid contained a high copy origin and drove MK and PMD genes expression under a strong promoter as higher level expression of these enzymes is required to drive the pathway toward the product [24]. Therefore, to optimize the balancing of pathway enzymes in P. putida, a systematic comparison of pathway constructs such as using a 1-plasmid vs 2-plasmid system, varying plasmid copy numbers, and changing the strength of the promoter or RBS in driving different pathway genes will be required as previously shown in the E. coli isoprenoid studies [42].

As P. putida consumes isoprenol, we investigated the possibilities of preventing the consumption of isoprenol by supplementing specific medium components. Interestingly, we found supplementing l-Glu in the culture medium showed a significant preservation of isoprenol. Using metabolomics, we revealed the difference of intracellular metabolites and attempted to explain the possible scenarios during isoprenol degradation. The metabolites analysis showed an insufficient energy availability and an imbalanced redox status during isoprenol degradation. This may be associated with the alcohol degradation mechanism as P. putida utilizes pyrroloquinoline quinone (PQQ)-dependent alcohol dehydrogenases for alcohol degradation [43], which may change the balance of cellular redox when processing isoprenol degradation. In addition, since l-Glu is a precursor of PQQ biosynthesis [44], supplementing l-Glu could increase substrate availability toward PQQ biosynthesis, which might contribute to the rebalancing of redox status as well as restoring the cellular metabolism. On the other hand, a few studies reported the development of isoprenol utilization pathways for isoprenoid synthesis, such as isopentenol utilization pathway (IUP) [45] and isoprenoid alcohol (IPA) pathway [46]. In these pathways, the alcohol kinase (e.g., yeast choline kinase, [45]) was identified and used for isoprenol phosphorylation, which indicated alcohol phosphorylation might be an alternative route besides alcohol dehydrogenation related to isoprenol degradation.

As P. putida is an emerging microbial host, there are still many challenges to engineer this host as a bioproduction workhorse. For example, even though some P. putida species can utilize xylose as a carbon source, the most widely studied P. putida microbial platform (KT2440) cannot naturally utilize xylose. Thus, engineering for the simultaneous utilization of glucose, xylose, and lignin-derived aromatic substrates may need additional efforts to achieve optimal carbon utilization without comprising the production yields [32, 47]. The versatile metabolism of P. putida which allows it to survive with broad substrates also brings issues of the self-degradation of biosynthetic products. These issues could be challenging to overcome since multiple genes and regulations may be involved in the degradation process [28]. Additionally, the polyploid property nature of P. putida may increase the instability of using a high-copy plasmid for gene expression [48], and consistent with this we observed significant variations among colonies when screening for productions. Even with these issues, the unique capability of P. putida to utilize lignin-derived intermediates and aromatics as carbon sources are clear advantages over the widely used microbial hosts such as E. coli and S. cerevisiae as a next-generation industrial microbial host for converting lignocellulosic biomass to biofuels and bioproducts. In this study, we demonstrated that the engineered P. putida strains can utilize p-coumarate, as the sole carbon source to produce isoprenol. It is foreseeable that P. putida can achieve an economically feasible production of isoprenol and other bio-based products from lignocellulosic biomass via systematic strain engineering combining the efforts of computation and analytics using the Design-Build-Test-Learn research cycle [49].

Conclusions

P. putida can naturally utilize broad carbon sources and is tolerant to xenobiotics, which shows great potential to be developed as an emerging industrial microbial workhorse especially in maximally converting carbon from lignocellulosic biomass to biofuels and bioproducts. In this study, we engineered the heterologous MVA pathway in P. putida KT2440 to produce isoprenoids, including isoprenol (C5) and epi-isozizaene (C15). IPP-bypass MVA pathway showed advantages during isoprenol production. Through comparing flux distribution and identifying gene-knockout target, we optimized the production strain to achieve an increase of isoprenol production to 104 mg/L in a batch flask experiment. Due to the isoprenol degradation in P. putida, we investigated the strategy to prevent self-consumption of isoprenol, and supplementation of l-Glu in the medium was found to show significant preservation for isoprenol. The engineered P. putida strain can also produce isoprenol using p-coumarate as the sole carbon source. Our results presented a good demonstration of developing P. putida as a new microbial chassis for biofuel production with improved carbon utilization from lignocellulosic biomass.