Genes were simplified as three elements: ES (upper MVA pathway), S1 (lower MVA pathway), and PG (Pinene synthetic pathway; Figure 1A) cassettes to drive heterologous protein expression for improved pinene biosynthesis (Figure 1B).

Three pinene pathways, PG1, PG2, and PG3, from different species, were constructed (Figure 1B). To verify which heterogeneous expression pathway is needed for optimum pinene production in strains DH411 and DH416. In this regard, the strain DH416 was transformed with plasmid pETL carrying PG1, PG2 and PG3 cassettes to generate three strains HSY001 (DH416/pETL-PG1), HSY002 (DH416/pETL-PG2) and HSY003 (DH416/pETL-PG3), respectively. On the other hand, three more strains HSY004 (DH411/pETL-PG1), HSY005 (DH411/pETL-PG2), HSY006 (DH411/pETL-PG3) were generated. Lycopene and pinene are classified as terpenoids and share similar precursor requirements (IPP and DMAPP). When the activity of the pinene pathway was increased, the intracellular precursor levels for lycopene biosynthesis were decreased. This occurs because both pinene and lycopene synthesis pathways compete for the same precursors. Strain (DH411/pETL-PG1) with the lowest lycopene level was selected for further experiments, indicating the corresponding pinene synthetic pathway (PG1) is more appropriate than PG2 and PG3 (Supplementary Figure S1). According to the result, it was difficult to transform DH416 into a pinene-producing strain, hence E. coli DH411 was selected as the starting strain. For instance, Yang et al. (2013) found that an accumulation of 5.44 mg/L pinene was achieved in E. coli via heterologous GPPS (A. grandis) and PS (P. taeda) expression (Yang et al., 2013). Similar findings have also been reported by Sarria et al. (2014), who demonstrated that the heterologous expression of PS (A. grandis) and GPPS (A. grandis) generated 28 mg/L of pinene in E. coli (Sarria et al., 2014). Afterwards, the pinene production from these six strains was analyzed (Figure 2A).

Based on the above-mentioned results, strain HSY004 (DH411/pETL-PG1) was identified as the most favorable integrated strain in comparison to other strains. It exhibited the maximal level of pinene production about 5.66 mg/L (Figure 2A). The PG1 cassette was determined to be the optimal choice for heterologous pinene production. To further enhance pinene production, integration of one PG1 cassette in a counterclockwise orientation in region 8 was performed, resulting in the creation of strain HSY007. DH411 strains lacked the expression cassette for pinene synthesis at the 8th site. After that, the GC cassette was knocked off in region 23 of strain HSY007 to construct a new strain HSY008. While comparing the pinene production of both HSY004 and HSY008, it was found that the integrated strain HSY008 generated a 3-fold increase (15.34 mg/L) in pinene production when compared to strain HSY004 (Figure 2B). More so, it was found that the knockout of the lycopene biosynthesis pathway (GC) resulted in enhanced pinene production in integrated strain HSY008.

To investigate whether further overexpression of either the upper MVA pathway or lower MVA pathway was necessary, strain HSY009 was generated. In this strain, one S1 cassette was integrated at the LpxM region (non-essential position). This integrated strain consisted of two copies of S1 cassette which is more than that of starting strain DH411. Then, the integrated strain HSY009 was transformed with the single copy plasmid pCC1FOS carrying ES and S1 cassettes, which in turn generated the HY09E (HSY009/pCC1E) and HY09S (HSY009/pCC1S) strains. Notably, the increase in either the MVA upper pathway or the MVA lower pathway to generate HY09E (ES: S1: PG 2: 2: 1), and HY09S (ES: S1: PG 1: 3: 1) strains did not enhance the pinene production, as shown in Figure 3. The pinene production of HSY009 (ES: S1: PG 1: 2: 1) was 34.35 mg/L, which is significantly highest among three newly constructed strains HY09E, and HY09S, respectively (Figure 3). Our findings align with the study of Hussain et al. (2021), who stated that a further increase in either the lycopene expression pathway or MVA lower pathway was not beneficial in terms of specific lycopene production after the generation of 11 models (Hussain et al., 2021). This may be due to the phenotype mimicking the loss-of-function mutation when there is a high gene expression level. Hence, the strain HSY009 was declared to be the best-performing strain. This integrated strain was comprised of one copy of the upper MVA pathway, two copies of the lower MVA pathway, and one copy of the pinene synthetic pathway. Therefore, understanding the copy number of heterogeneous expression pathways is crucial for designing an efficient expression pathway.

To improve the pinene production in strain HSY009, the expression cassettes PG were integrated into three non-essential regions of the genome. These regions were the 44th, 58th, and 23rd, as illustrated in Figure 4A. The region’s selection was due to its wide distribution throughout the chromosome (Wei et al., 2017).

By integrating the PG cassettes one by one into these regions, three new strains were generated: HSY010 (HSY009 Δ44th::T7 PG1), HSY011 (HSY009 Δ58th::T7 PG1), and HSY012 (HSY009 Δ23rd::T7 PG1). Moreover, it was found that the best pinene production of 69.08 mg/L belonged to strain HSY012, as depicted in Figure 4B.

These results indicated that the integrated strain HSY012 (ES: S1: PG 1: 2: 2) carrying one copy of the MVA upper pathway, two copies of the MVA lower pathway, and two copies of the pinene synthetic pathway, exhibited the highest pinene production followed by HSY011, HSY010, and HSY009 strains, respectively (Figure 4B). Furthermore, it was observed that the growth rate of the HSY012 strain was higher compared to strains HSY011, HSY010, and HSY009, as depicted in Figure 4C. This finding suggests that integrating the PG1 cassette at the 23rd region increased pinene production and growth rate. This result is in line with Wei et al. (2017), who compared the lycopene level by changing the direction (clockwise and counter-clockwise direction) and position (8th, 23rd, 58th, and 64th nonessential regions) of GC (Lycopene expression pathway) integration. It was found that the coupling integration positions with expression direction resulted in a 5- and 10-fold increase in lycopene production in E. coli (Wei et al., 2017).

To test whether the overexpression of the pinene expression pathway was required after the development of HSY012, the transformation of a single copy plasmid pCC1FOS carrying PG1 cassette was executed in strain HSY012. This transformation resulted in the generation of HSY013 (HSY012/pCC1P) strain. However, there was no significant increase in the pinene yield, as shown in Figure 4D. This result indicated that the overexpression of the pinene synthetic pathway was futile at this point, as the HSY012 strain (ES: S1: PG 1: 2: 2) accumulated more pinene up to 69.08 mg/L than that of HSY013 strain (ES: S1: PG 1: 2: 3) (Figure 4D). Therefore, HSY012 was identified as the best-performing strain among the tested strains. It consisted of one copy of the upper MVA pathway, two copies of the lower MVA pathway, and two copies of the pinene synthetic pathway. Table 1 summarizes the pinene level (mg/L) of five recombinant E. coli strains after integrating the pinene expression pathway at different integration sites. The colony formation and pinene production are illustrated in Figure 5.

The final strain, HSY012, was further utilized in shake fermentation experiments to produce pinene. The agitation speed, aeration rate, and pH were set at 1.5 vvm, 220 rpm, and 7.2, respectively. Optimization of culture medium and processing conditions can be utilized to increase the quantity and quality of biofuel. Fermentation was carried out using flask equipment set under optimized conditions. These conditions included the cultivation temperature, which needed to strike a balance between cell growth and product formation; the nitrogen source, which could impact the vitality and viability of bacteria; and the carbon source, which influenced the concentration of intracellular acetyl-CoA (Gao et al., 2023; Rajpurohit and Eiteman, 2024; Ayivi et al., 2022). To improve the pinene titers, the “one-factor at-a-time” method was exploited to optimize all the parameters (Figure 6). The highest pinene production (90.76 mg/L) was obtained when the HSY012 strain was grown on a fermentation medium containing 3% glucose and 2.5% yeast extract, with induction using 0.2% arabinose at a temperature of 32°C. Using glucose as a carbon source resulted in the highest pinene yield compared to maltose, fructose, sucrose, xylose, and lactose by HSY012, with maltose being the second-best carbon source.

The final strain, HSY012, consisting of a biosynthetic pathway of pinene, accumulated pinene up to 90.76 mg/L under flask fermentation conditions. Afterwards, a 5 L bio-fermentor was used for pinene production by strain HSY012 using the optimized treatment that gave the best results in shake flask experiments. In batch fermentation, glucose as a carbon source, along with yeast extract, was used successfully to obtain better results in terms of pinene production. The induction temperature was fixed at 32°C. Following induction in the bio-fermentor, there was a rapid increase in α-pinene production, peaking at approximately 280.78 mg/L between 18 and 24 h. The highest pinene concentration of 436.68 mg/L with a mean productivity of 14.55 mg/L/h was achieved after 30 h culture in a 5-L fermenter. The cell density of the engineered strain reached about 40, while the glucose concentration reduced to 5 g/L after 36 h of fermentation (Figure 7). Besides the unknown mechanism of chromosomal sites (8, 23, 44, 58, and LpxM) imparting toxicity stress tolerance to strains, optimizing cofactor metabolic pathways achieves further improvement in pinene production and cell growth. Furthermore, comprehensive analysis based on proteomics, transcriptomics, or metabolomics can help identify major restriction factors such as product synthesis, growth, and cofactor consumption, which can then be targeted for further optimization.

The details about plasmids and strains are provided in Table 2. The construction of plasmids was accomplished through E. coli DMT, DH5α, and DH1. For eliminating the methylated plasmid in vivo, E. coli DMT was used. All primers’ manufacturing was performed by the local renowned GENEWIZ company (listed in Supplementary Table S1). Restriction enzymes, PrimeStar Max DNA polymerase, Takara DNA ligation, Clonexpress Ultra cloning, Plasmid miniprep, and Gel extraction kit were used. All operations were performed as per instructions.

The engineered strains DH411, DH416, and HSY009 were constructed via the lambda-Red recombination method. Moreover, the remaining other strains were constructed via CRISPR/Cas9/lambda-Red techniques. Using E. coli DH411 as a host, the expression element PG1 was integrated counterclockwise into the 8th region to generate the strain HSY007/pCas9. Using plasmid pCP-8r as a template, PCA3rc and PCA5arc as primers, the target fragment containing homologous region 8 and expression element PG1 was amplified by PCR at about 4.1 kb (see Supplementary Table S2 for CRISPR plasmids). Recombination utilizes plasmids pCas9 and pRNA-8r. After that, E. coli HSY007 as a host, the 23 region was knocked out to construct the strain HSY008. Using plasmid pCP-23d as template, PCA3rc, and PCA5rc as primers, a target fragment containing the homologous region of region 23 was amplified by PCR at about 1.1 kb. Following that, recombination was carried out utilizing plasmids pCas9 and pRNA-23d. To eliminate the plasmid pCas9 in the host, plasmid pKD-15A was introduced into the host. Dehydrytetracycline was added to induce the expression and cleavage of pCas 9 and sgRNA, respectively. Furthermore, E. coli HSY008 was used as the host, and expression element S1 was integrated into the LpxM locus clockwise to generate the strain HSY009. In the next step, plasmids pBDC-58i and pCNA were recombined, followed by the second step of recombination using plasmids pSNA and pKILS. The first step of recombination employed primers ES-0 and ES-1; while the second step of recombination was verified by PCR using primers M-0 and M-1. Lastly, E. coli HSY009 was used as the host, and the expression element PG1 was integrated sequentially into the 44th, 58th, and 23rd locus to generate three strains HSY010, HSY011, and HSY012, respectively.

Three heterogeneous expression cassettes were utilized: ES cassette, S1 cassette, and PG cassette. Each cassette consists of specific genes obtained from different organisms. The ES cassette consists of two genes mvaE and mvaS isolated from Enterococcus faecalis. This cassette is responsible for catalyzing the conversion of acetyl-CoA into mevalonic acid. It is commonly referred to as the upper MVA pathway. The S1 cassette consists of genes mvaK2, mvd1, fni, and mvk isolated from Streptococcus pneumoniae. This cassette catalyzes the reaction to convert mevalonic acid to IPP and DMAPP. These compounds serve as the major precursors for the enhanced production of terpenoids. The S1 cassette is commonly named the lower MVA pathway. The third expression cassette, PG cassette, consists of two genes, GPPS and PS, which were isolated from Abies grandis. The PG cassette is responsible for the production of pinene, which is the target product of interest (see Supplementary Table S3).

In this study, two plasmids, pRNA-8r and pRNA-GC, were generated from pRNA (Reisch and Prather, 2015). The fragment F1 (3.5 kb) was amplified by PCR using primers 8F and BetaR and pRNA as a template. Similarly, the fragment F2 (3.5 kb), was amplified by PCR using primers pkdseq1 and 8R. These two PCR products, F1 and F2, were subsequently assembled to create the plasmid pRNA-8r. RNA8-1 and BetaR. PCR primers RNA8-1 and BetaR were used to confirm the presence of positive pRNA-8r colonies. The first fragment was amplified by PCR using primers 23 GF and BetaR and pRNA as a template, while the second fragment was amplified by PCR using primers pkdseq1 and 23G. These two PCR products were assembled to create the plasmid pRNA-GC. PCR identification was conducted using primers RNA-23G and BetaR.

Additionally, the donor plasmids pCP-8r and pCP-23d were employed in this study. The 2.2 kb backbone CP containing cat and p15A ori was amplified by PCR using primers PCA5 and PCA3 and pACYCDuet-1 as a template. The PG fragment containing element PG1 was amplified by PCR using primers TE5 and TE3 and pET3b-PG1 as templates. The E. coli DH1 genome was used as a template, the 0.6 kb 8 L fragment containing the left homology region was amplified by PCR using primers L8-5 and L8-3; while the 0.6 kb 8R fragment containing the right homology region was amplified by PCR using primers R8-5 and R8-3, respectively. Four PCR products, CP, PG, 8 L, and 8R were assembled to create the plasmid pCP-8r.

The left homologous region of the 23 region was amplified by PCR (about 500 bp) using primers L23-5 and L23d-3 and the E. coli DH1 genome as a template. The right homologous region (about 500 bp) of the 23 region was amplified by PCR using primers R23d-5 and R23-3 and E. coli DH1 genome as a template. Plasmid pCP-23d was constructed after recombination of cytoskeleton CP, left homologous fragment of 23 region, and right homologous segment of 23 region. The target fragments for homologous recombination were amplified by PCR using primers PCA3rc and PCA5rc and plasmids pCP-8r or pCP-23d as a template.

Two primers RP5 and FP3 were used to amplify the backbone of pCC1P (fragment CC1) and plasmid pCC1FOS was used as the template. The fragment PGCC, consisting of T7 PG flanked by 15–20 bp homologous region of fragment CC1 was amplified from plasmid pETP with the assistance of two primers BI5 and BI3, respectively. Afterwards, the cloning kit (TaKaRa Co.) was used to assemble two fragments, named CC1 and PGCC to generate plasmid pCC1P which was identified by PCR via two primers, namely FP-0 and RP-1, respectively.

The isolated GPPS gene fragment was digested from pUC57-GS with Xba I and BamH I, followed by insertion into Spe I/BamH I sites of plasmid pUC57-PS to create pUC57-GS-PS. The isolated PG fragment was digested from pUC57-GS-PS with Xba I and BamH I and then ligated into the corresponding sites of pET3b to create pETP.

The PG1, PG2, and PG3 fragments were digested from pUC57-PG1, pUC57-PG2, and pUC57-PG3 and inserted one by one into plasmid pETL to create pETL-PG1, pETL-PG2, pETL-PG3. The PG1 fragment was inserted in pET3b to create pET3b-PG1. The primers used to amplify the PG1 fragment from pUC57-PG1 were RT5 and RT3, primer FT5 and FT3 for the PG2 fragment, and primer HT5 and HT3 for the PG3 fragment, respectively.

For culture, the single colony of strain HSY012 was grown overnight at 32°C and 220 rpm in 100 mL of LB medium containing NaCl (10 g/L), tryptone (10 g/L), and yeast extract (5 g/L). Then, the inoculation was carried out in 50 mL of the medium at an optical density (OD₆₀₀) of 0.1. The fermentation medium consisted of 5 g/L of yeast extract, 5.0 g/L of NaH₂PO₄.2H₂O, 7 g/L of K₂HPO₄.3H₂O, 2.5 g/L of NaCl, 5 g/L of Tween 80, 10 g/L of glycerol, 0.5 g/L of MgSO₄ and 10 g/L of glucose (Hussain et al., 2021). The culture system was induced with 0.2% of L-arabinose when the OD₆₀₀ reached ∼0.6–0.8. For strain HSY012, fermentation was carried out in a 5 L fermenter with an initial working volume of 3 L. The system was induced with 0.2% of L-arabinose after 7 h. During fermentation, the optimum temperature and pH for culture incubation were 32°C and 7 (controlled by NaOH). Optimal pinene production was obtained at 220 rpm agitation and 1.5 vvm aeration rate. The fermentation medium was supplemented again with L-arabinose (0.2%) after 12 h (Xiong et al., 2015). Furthermore, a spectrophotometer evaluated the bacterial growth by measuring the OD₆₀₀. At the start of the experiment, 15% (6 mL) of dodecane, purchased from Skyrun Co., Ltd, was poured into a fermentation medium through a 5 mL pipette. It was applied as an overlay to trap the pinene. This chemical is kept at room temperature.

The effects of three factors, such as carbon source, nitrogen source, and temperature, on pinene production were studied. The fermentation medium was supplemented individually with five different nitrogen sources (peptone, beef powder, yeast extract, ammonium sulfate, and corn steep liquor) at a concentration of 2.5%, followed by the addition of different carbon sources (maltose, glucose, fructose, sucrose, xylose, and lactose), with the concentration of 3%. The effect of different temperatures (25°C, 30°C, 32°C, 35°C) on pinene production was also studied.

For the quantification of pinene, 1.5 mL of microcentrifuge tube carrying 500 μL of the dodecane layer was centrifuged (25,000 g, 1 min). Afterwards, the dilution of 50 μL of dodecane was carried out in 450 μL of ethyl acetate that spiked with the internal standard cyclohexanone. GC/MS used a standard curve of α-pinene (Sigma Aldrich) to analyze the samples. Agilent 7,890 GC system with 5,975 MSD was used. 19091S-433 Agilent chromatographic column thickness was 0.25 mm, and the length was 30 m. The inlet temperature was 300°C, flow was at 1 mL/min, shunt ratio was at 100: 1, and injection amount was 0.2 L. The oven temperature was initially set at 50°C for 30 s. Subsequently, there was a ramping phase with a rate of 40°C/min, increasing the oven temperature from 50°C to 70°C. Following this, there was another ramping phase at a rate of 25°C/min, raising the temperature from 70°C to 250°C. The final temperature of 250°C was held for 5 min. The reference retention time for cyclohexanone and pinene was 3.8 min and 4.7 min, respectively. All experiments were done with triplicates, and the error bar showed the standard deviation. The standard curve was plotted with the concentration of reference substance as the abscissa, and the ratio of peak heights of pinene to cyclohexanone was plotted on the ordinate (y-axis), as shown in Supplementary Figure S2.