Development of a semi-endogenous PRT-based expression system for Methylotuvimicrobium alcaliphilum and its application to enhance 3-hydroxybutyric acid production from methanol

Abstract

Background

Endogenous expression systems, such as those incorporating promoters, ribosome-binding sites (RBS), and terminators (i.e., PRT systems), have been developed for various bacterial host strains. However, intact native promoters are typically long sequences that include regulated regions, which hinder the production of recombinant proteins.

Results

In this study, a semi-endogenous PRT system based on 20-40 nucleotide motif sequences derived from the genome of the target host strains was developed using machine learning-based prediction tools. This system was subsequently fine-tuned and optimized based on transcription and translation rates as well as the Gibbs free energy of the terminator structure. This system was validated, and its effectiveness was demonstrated by using it to induce the expression of dTomato, Zeocin^R, and phaCAB cluster genes in E. coli. The semi-endogenous PRT was also used for enhanced production of 303 ± 6.39 mg/L 3-hydroxybutyric acid (3-HB) from methanol as the sole carbon source in a 50-mL nitrate mineral salt (NMS) flask after 234 h with overexpression of exogenous phaA and phaB genes in the Methylotuvimicrobium alcaliphilum 20ZX strain.

Conclusions

The semi-endogenous PRT system was efficiently designed and served as an endogenous expression system in both E. coli and methanotroph. This system will offer more choices for selecting an endogenous expression platform tailored to the target host strain.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13036-025-00592-y.

Background

Genetic tools in biotechnology have recently focused on areas such as clustered regularly interspaced short palindromic repeats (CRISPR)-Cas systems, synthetic biology platforms, high-throughput screening, omics technologies, machine learning (ML), data analytics, genome-scale metabolic models, multiplexed genome editing, and genome-wide association studies (GWAS). Investigations in these areas have focused on enhancing the precision, efficiency, and scalability of metabolic engineering applications ranging from biofuel production and pharmaceutical synthesis to bioremediation, and efforts have been made to develop and broadly apply a set of robust engineering toolkits. Many ML-based tools with high accuracy have been developed. ML algorithms can create ribosome-binding site (RBS) libraries for an RBS sequence-phenotype model that predicts a large number of producers [1]. The Extended Vision Mutant Priority (EMPV) framework, an ML-based model, can predict strong promoters with mutations [2]. Various deep-learning models have been used to predict regulators [3] and target DNA regulatory regions [4]. These tools have been employed to detect and predict promoters, RBSs, terminators, and regulators, as well as their efficiency in producing enzymes or valuable substances. These components are essential for constructing expression systems in host cells. Several studies have introduced or validated novel ML-based tools in recent years. Nonetheless, many of these studies primarily focused on model architecture, such as determining the optimal number of layers or blocks, tuning training parameters in epochs, batch size, and learning rate, and selecting appropriate optimizers or algorithms on the basis of available experimental datasets. Consequently, research that applies these ML tools directly to practical problems in genetic engineering is limited. For instance, DeepRegFinder, which employs convolutional or recurrent neural network architectures [3], and LegNet, which utilizes large-scale attention-based transformer models [4], are representative deep-learning frameworks designed for detecting regulatory elements and genomic regions. However, additional studies are required to adapt and apply these tools in real-world biological contexts to validate their usability, robustness, and generalizability.

Protein-translation systems essentially include a promoter, RBS, target gene, and terminator. The selection of the expression system is based on the findings of previous studies and the existing systems. However, conducting experiments on novel strains with unknown characteristics remains a challenge. The available systems may be incompatible or work inefficiently with novel strains [5, 6], thereby synthesizing small amounts of the target product. Testing and identifying new tools that can be applied to a novel strain is a time-consuming and expensive process. In this regard, using native expression systems that include endogenous promoters, RBS, and terminators can solve this problem. Using a native promoter, RBS, and a terminator stem from a novel target strain, the expression of the target product and enzyme can be effectively achieved. Many recent studies have attempted to use endogenous expression systems, such as overexpression and secretion of recombinant proteins in P. tricornutum using the endogenous promoter HASP1 [7], which has since been shown to be inducible under phosphate limitation [8], the endogenous promoter for expressing single-guide RNA [9], some native promoters were stronger than commonly used synthetic promoters for expressing heterologous proteins in Zymomonas mobilis [10], enhancement of protein expression using the endogenous promoter and codon optimization of the moss Physcomitrella patens [11], expression and production of β-galactosidase in Leuconostoc citreum EFEL2700 after screening strong endogenous promoters based on transcriptome analysis [12], characterization of two endogenous promoters of two genes, glyceraldehyde-3-phosphate dehydrogenase (GapC1) and glutamine synthetase (GS), in Phaeodactylum tricornutum [13], and identification of the strong endogenous promoter CRT (Pcrt) from transcriptome-based screening in the oleaginous microalga Ettlia sp. YC001 [14]. Endogenous promoters are easily accepted in the host because they stem from endogenous genes in the host genome. However, a limitation of these promoters is that they are regulated by both negative or positive effects on the host cell system, necessitating the development of new or modified endogenous systems to avoid endogenous regulation of host cells cultured under certain conditions.

Industrial strains such as E. coli, Pseudomonas putida, and Saccharomyces cerevisiae have been extensively studied using many engineering tools. These strains generally utilize glucose, sucrose, lactose, or glycerol to produce valuable products, such as 1,2-propanediol [15], protocatechuic acid [16], 2,3-butanediol (2,3-BDO) [17], 3-hydroxyvalerate [18], and 3-hydroxybutyrate [19]. In comparison with conventional strains, non-conventional strains have certain benefits, such as utilization of novel carbon sources. Methanotrophs utilize methane as their sole carbon and energy source to solve environmental problems by mitigating greenhouse gas emissions. Biogas is a sustainable and renewable carbon resource because it contains 50%-70% methane [20]. The key enzymes in methanotrophs, soluble methane monooxygenase (sMMO) and particulate MMO (pMMO), are involved in the first step in converting methane to methanol before it moves further into the metabolic pathway. Achieving high-level expression of foreign genes is a prerequisite to produce non-natural target products from methane. Although genetic toolboxes are limited, various target products have been synthesized in methanotrophs, such as ectoine, 2,3-BDO, 3-hydroxybutyric acid, and 1,2-propane diol [21–23] as a proof-of-concept.

This study describes the prediction, design, and selection of a semi-endogenous system consisting of a promoter, RBS, and terminator (i.e., a PRT system). Endogenous 40-nucleotide sequences of the promoter or terminator were predicted using an ML tool, whereas the endogenous 20-nucleotide RBS sequence was directly selected upstream of the ATG codon after annotating genes from the sequencing FASTA file. All final PRT sequences were created using motif-based sequences. These motifs represent all endogenous sequences belonging to the entire genome and were identified using the MEME tool. The promoter was optimized and modified, and the transcription and translation rates of the RBS were calculated. The terminators were optimized on the basis of the lowest energy of the loop structure. Ensemble candidate PRT sequences were constructed using a suitable vector and expressed in the target strains, including E. coli and methanotroph. These systems were evaluated for the expression of the dTomato and Zeocin^R genes and the production of the biodegradable polymer polyhydroxybutyrate (PHB) through the expression of phaCAB cluster genes in E. coli as the reference strain. The semi-endogenous PRT was also used to synthesize 3-hydroxybutyrate, a PHB monomer, in Methylotuvimicrobium alcaliphilum 20ZX. This system worked effectively in E. coli and methanotroph, offering benefits such as easy synthesis by insertion into oligo primers and easy acceptance by target cells because of self-originated endogenous sequences.

Methods

Bioinformatics analysis, predictions, and calculations

The genome sequences of E. coli K12 MG1655 (NC_000913) and Methylotuvimicrobium alcaliphilum 20Z (chromosome NC_016112.1) in FASTA and GBK formats were downloaded from NCBI. These files were used as input files to predict the promoter, RBS, and terminator. PGAP version 2023-10-03. build7061 was used to annotate the bacterial genome [24]. MEME version 5.1.1 was used to identify and search for motifs and sequences [25, 26]. PROMOTECH v1.0 was used to predict promoters from bacterial genome sequences [27]. iTerm-PseKNC, a sequence-based tool, was used to predict the terminator on the basis of a support vector machine, which was a standalone package [28]. De Novo DNA (https://www.denovodna.com/) [29] and RBSDesigner [30] were used to calculate the transcription and translation rates, respectively. Seqfold (https://github.com/Lattice-Automation/seqfold) was used to predict the lowest free energy structure of the nucleic acids. Forgi, a Python library, was used to analyze the secondary structure of the RNA [31]. The small in-house scripts for the sequencing process were written using Python version 3.8 and Jupyter Notebook version 6.4.1. Two Python library packages, Biopython version 1.78 and Pandas 1.5.3, were also used to write the scripts.

Chemicals and reagents

Sodium (S)-3-hydroxybutyrate was purchased from Santa Cruz Biotechnology (Dallas, TX, US) and used as the standard for high-performance liquid chromatography (HPLC) analysis. Xylose was purchased from Glentham Inc. (UK). Methane gas was supplied by UNIONGAS (Seoul, South Korea). Methanol for molecular biology, supplied by Merck, was used as the carbon source. Luria–Bertani (LB) powder and bacteriological grade agar were supplied by FORMEDIUM (UK). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) or TCI (Tokyo, Japan). DNA sequencing and oligonucleotide synthesis were performed by Macrogen (Seoul, South Korea). The reagents for polymerase chain reaction (PCR) analyses were obtained from BioFACT (Daejeon, South Korea). An Expin™ Gel SV Kit (GeneAll, Seoul, South Korea) was used for gel extraction and DNA purification. The Exprep Plasmid SV was supplied by GenAll (Seoul, South Korea) for plasmid isolation. A ClonExpress Ultra One-step Cloning Kit was purchased from Vazyme and was used for the Gibson assembly reaction.

Bacterial strains

E. coli DH5α and E. coli K12 MG1655 were used as reference strains for testing dTomato and Zeocin^R gene expression with the PRT system. These strains harbored pAWP89-PRT^E-dTomato, pAWP89-PRT^E-Zeocin, pAWP89-Spectinomycin^R-Ptac-dTomato, pAWP89-spectinomycin^R-PRT^E-dTomato, and pAWP89-spectinomycin^R-PRT^E-phaCAB for testing the dTomato signal, resistance to zeocin, fluorescence-activated cell sorting (FACS) experiments, and production of the biodegradable polymer PHB in E. coli. The engineered xylose-utilizing M. alcaliphilum 20ZX strain [22] was used as the platform strain to investigate the 3-hydroxybutyric acid (3-HB) content in pAWP89-P_syn^20Z-R_syn^20Z-phaA-R_syn^20Z-phaB-T_syn^20Z (pAWP89-PRT^20Z-phaAB).

Media and cultural conditions

E. coli DH5α and E. coli K12 MG1655 were used as the reference strains for growth on LB. LB + kanamycin (final concentration, 50 mg/L) and LB + zeocin (final concentration, 0.03% v/v) were used to culture the recombinant strain in 10-mL Falcon tubes with wildtype and recombinant strains, respectively. The recombinant E. coli strain was screened using LB with 1.5% (w/v) agar and antibiotics (50 mg/L kanamycin or 0.03% v/v zeocin). All cultivation steps were performed at 37 °C and 200 rpm. To produce PHB from xylose in E. coli, the recombinant pAWP89-spectinomycin^R-PRT^E-phaCAB-harboring E. coli was cultured in 10 mL of LB or 10 mL of M9 medium containing 2 mM Mg₂SO₄, 0.1 mM CaCl₂, 10 g/L xylose as the sole carbon source, trace elements, and vitamins, as previously described [22].

M. alcaliphilum 20ZX and M. alcaliphilum 20ZX-3HBPRT strains were cultured in nitrate mineral salt (NMS) medium or NMS agar with 1% (v/v) methanol as the sole carbon source at 30 °C and 230 rpm, as previously described [22].

Cloning, transformation, and expression

dTomato and Zeocin^R genes were obtained from a previous study. Semi-endo PRT was specifically designed for gene expression in E. coli. Entire Semi-endo PRT sequences were included into an oligonucleotide primer (Table S2), synthesized, amplified by PCR, and ligated to the vector using a ligation reaction (Cloning Kit). The recombinant vector was used to transform E. coli DH5α using the heat-shock method and E. coli K12 MG1655 by electroporation. E. coli K12 MG1655 competent cells were prepared by washing twice with glycerol 10%, after which 50 µL of competent cells was mixed with 5–10 µL of the pAWP89-P_syn-R_syn-dTomato/Zeocin^R-T_syn sample. EC2 (~ 2.5 KV) and 1-mm cuvettes were used for electroporation in a Bio-Rad machine.

The phaA (H16_RS07140) and phaB (H16_RS07145) genes were cloned from R. eutropha H16 into pAWP89 using a semi-endo PRT system. The semi-endo PRT sequences were inserted into oligo primers synthesized by Macrogen (Seoul, South Korea) and ligated with the target gene and pAWP89 vector using a Cloning Kit. The ligated plasmid was amplified by transfer into E. coli DH5α, after which plasmids from positive colonies were isolated and purified using the Exprep Plasmid SV Kit. The purified plasmid was also used to transform the M. alcaliphilum 20ZX strain by electroporation using a Bio-Rad machine (1-mm cuvette, 1300 V, 200 Ω, 25 µF). Colonies of the M. alcaliphilum 20ZX-3HBPRT recombinant strain were screened on an NMS agar plate, and PCR was performed using methanol as the carbon source. 3-HB experiments were performed using NMS with methanol, methane, and xylose as carbon sources.

Flow cytometry analysis

The fluorescence signal at the single-cell level was measured using Attune^® NxT (Thermo Fisher Scientific, US) with a yellow laser source (561 nm) for excitation of the dTomato protein. Cells were cultured in 10 mL of media from positive colonies, collected after 24 h, and washed once with phosphate-buffered saline (PBS), and 1 mL of PBS was added. The samples were diluted to optical density (OD) 0.5 in a 1.5-mL e-tube. For each sample, the data of a single-cell region (R1) were counted for 10,000 events under forward scatter (FSC) = 300 V and side scatter (SSC) = 300 V, with other values set at default (400 V). The thresholds were set at FSC = 1 × 1000 and SSC = 0.5 × 1000. Acquisition volume was set at 40 µL and 12.5 µL/min. Data was analyzed using Attune flow cytometry software. The mean fluorescence intensity (MFI) value of the region of interest (R2) was used as a quantitative measure of fluorescence intensity.

Visualization of fluorescence gene-harboring colonies

The transformed target dTomato and sfGFP-inserted plasmid colonies were visualized using ultraviolet (UV) excitation. A WiseUv (Daihan, Korea) WUV-M20 UV transilluminator was used for this purpose. Tomato and sfGFP-harboring colonies were excited under UV light at 312 nm.

Analysis and detection by HPLC

OD values were measured at a wavelength of 600 nm in a Beckman spectrophotometer using 1.5-mL cuvettes and a 1-cm path length. For these analyses, 1 mL of the supernatant in a 50-mL shake-flask cultured media with each wildtype and recombinant M. alcaliphilum 20ZX-3HBPRT strains was collected by centrifugation twice at 7000 ×g, 7 min at 4 °C and filtered by 0.2-µm hydrophilic PTFE-H polytetrafluoroethylene (PTFE-H) with 1 mL of KOVAX-SYRINGE. 3-HB was detected using HPLC (Agilent, USA) with the following parameters: refractive index detector (RID) at 35 °C, 5 mM sulfuric acid mobile phase, and Aminex HPX-87 H organic acid column (Bio-Rad, Hercules, USA) maintained at 35 °C with a flow rate of 0.6.

Analysis and detection by GC

For measuring PHB content, cells were collected at different time points and then dried by freezing for 4-12 h. Methanolysis was performed by incubation at approximately 100 °C for more than 8 h with 2 mL of chloroform and 2 mL of methanol (containing 15% v/v H₂SO₄), with 20 mg of benzoic acid serving as the internal control. The organic phase was collected after adding 2 mL of H₂O and vortexing to separate the two phases. The sample was filtered by a 0.2-µm filter (PTFE membrane, Agilent) and a 1-µL sample was injected into the gas chromatography (GC) for polyhydroxyalkanoate (PHA) analysis (GC, Agilent 8890).

Statistical analysis

In this study, the values were calculated as the mean of triplicate experiments, with error bars presenting the standard deviation. Student’s t-test was used to determine the significant difference with p values less than 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Results

In-silico prediction, design, and selection of the endogenous promoter, RBS, and terminator

The selection of an expression system for the host strain remains a major research challenge. Testing with the available tools and systems incurs enormous workloads. In this study, we developed a semi-endo PRT system, an expression system in the host strain that facilitated the expression of the desired product with a rapidly selecting promoter, RBS, and terminator. This system included an endogenous promoter (P), ribosome-binding site (RBS-R), and terminator (T), as predicted from the host genome sequence. These sequences were then fine-tuned to generate a semi-endogenous PRT (or semi-endo PRT). Semi-endogenous sequences possess half of the properties of the original endogenous sequence. In addition, selecting multiple or consensus motif-based sequences for the entire PRT genome will not completely match a certain sequence; however, maintaining the constitutive motif allows the host cell to recognize function-responsive sequences. The final semi-endo PRT set can be constructed into a suitable vector with the target gene to produce the desired product, such as an enzyme or a valuable metabolite.

The workflow for PRT development based on the host genome sequence is shown in detail in Fig. 1a. The challenge in selecting promoters and terminators is the location of genes in an operon. In this case, only one promoter and one terminator can be detected. In contrast, RBSs are normally attached to each gene. Identification of RBSs is easier than detection of promoters or terminators. The locations of promoters and terminators can be determined using deep learning. PROMOTECH and iTerm-PseKNC are two examples of artificial intelligence (AI)-based tools that can be used for prediction. Thousands of promoters, RBS, and terminators are present in a host genome, and only one or a few sequences are selected and used in expression systems. A motif is the best option for creating a unique sequence for an expression system, with benefits such as maintaining function while presenting all feature sequences in the genome. MEME provides a motif based on an assembly sequence set. The fine-tuning step was performed by optimizing the sequence and calculating the transcription and translation rates for the promoters and RBSs. The terminator sequence was selected, followed by the calculation of loop energy and prediction of the 2D structure using Seqfold and Forgi.

Fig. 1.

Workflow for selection of the semi-endo PRT in a specific strain. (a) The process of selecting the promoters, RBS, and terminators. (b) Semi-endogenous PRT in E. coli. (c) Semi-endogenous PRT in 20ZX. In detail, the 40-nucleotide sequence promoter was predicted by PROMOTECH, and the motif was identified by MEME after assembling the entire sequences using an in-house Python script. RBS sequences were assembled by selecting 20 nucleotides upstream of ATG of the coding sequence (CDS) in the genome. Terminators with 40-nucleotide sequences were selected and predicted by RhoTermPredict, and the motif-responding sequence was assembled by MEME and the in-house script. Finally, these sequences were constructed together to create the candidate sequences. These candidate endogenous PRT sequences were fine-tuned on the basis of the transcription and translation rates with De Novo DNA and RBSdesigner for promoter and RBS, respectively. The loop structure energy and 2D structure were predicted by Seqfold and Forgi for the terminator

This workflow was adapted to obtain a candidate semi-endo PRT sequence from E. coli (Fig. 1b) and M. alcaliphilum 20Z (Fig. 1c). The genome sequences (FASTA format) of E. coli K12 MG1655 (NC_000193.3) and M. alcaliphilum 20Z (NC_016112) downloaded from the NCBI database served as the starting point. The tac promoter (P_tac) and RBS containing 29 and 28 nucleotides, respectively, in the vector pAWP89 (supplementary data S1) were used as the references. In-house Python scripts were written to assemble the predicted promoter sequences. Fifty predicted endogenous promoter sequences with scores above the threshold of 0.9 in E. coli and 451 sequences with scores above the threshold of 0.8 in M. alcaliphilum 20Z were selected (supplementary data S2, S4). MEME predicted the motifs of these sequences, yielding a multilevel of 40 nucleotides of the motif-response sequence (Fig. 1b and c, and supplementary data S4-S7). For the RBS sequence, 20 nucleotides upstream of the ATG start codon were selected and assembled from all genes in the genome. In total, 4302 and 3920 RBS sequences were assembled from E. coli and M. alcaliphilum 20Z, respectively (supplementary data S8 and S9). MEME yielded an RBS motif-based sequence in both E. coli and M. alcaliphilum 20Z strains (Fig. 1b and c, and supplementary data S10 and S11).

On the basis of MEME, the results suggested some motif-RBS sequences in E. coli and M. alcaliphilum 20Z. The promoter and RBS sequences were attached, and the transcription and translation rates were calculated using De Novo DNA (Figs. 2a and 5a) and RBSDesigner (Table 1).

Fig. 2.

Validation of the semi-endo PRT^E system in E. coli with the dTomato and Zeocin genes. (a) Fine-tuning of the promoter on the basis of the transcription rate. (b) Selection of the terminator on the basis of the binding energy and structure. (c) Construction of the dTomato and Zeocin genes with semi-endo PRT^E in vector pAWP89. The results of dTomato and Zeocin gene expression are presented in d and e, respectively. Transcription rate (Tx rate) of the Ptac promoter with the dTomato gene. Transcription rate data were obtained using the Promoter Calculator option in De Novo DNA. After fine-tuning, the endogenous promoters became semi-endogenous promoters. i: Internal loop. e: external. h: hairpin loop. b: bulge. s: stacked pairs. m: multibranched loop. Gibb’s free energy was determined using Seqfold and visualized by Forgi

Fig. 5.

Enhancing the production of 3-hydroxybutyric acid from methanol with semi-endo PRT^20Z in the 20ZX strain. (a) Fine-tuned 20Z promoter based on the transcription rate. (b) 20Z Terminator selection based on the loop energy and structure. (c) Construction of phaA and phaB genes with semi-endo PRT20Z in vector pAWP89. (d) The maximum concentration of 3-HB synthesized by 20ZX and 20ZX-3HBPRT. (e) 3HB concentration, OD, and MeOH consumption following the course of 20ZX and recombinant strains 20ZX-3HBPRT. (f) 3-HB biosynthesis pathway from methanol. (g) HPLC results showed the detected 3-HB peak overlapping with the standard peaks at around 12.8 min. Transcription rate (Tx rate) of the Ptac promoter with the Zeocin^R gene. Transcription rate data were obtained using the Promoter Calculator option in De Novo DNA. After fine-tuning, the endogenous promoters become semi-endogenous promoters. i. internal loop. e. external. (h) hairpin loop. b. bulge. s. stacked pairs. m. multibranched loop. Gibb’s free energy was obtained using Seqfold and visualized by Forgi. (i) internal loop. e. external. h. hairpin loop. b. bulge. s. stacked pairs. m. multibranched loop. Gibb’s free energy was obtained using Seqfold and visualized by Forgi. The data represent the mean value ± SD (n = 3). H6P, hexulose 6-phosphate; X5P, xylulose 5-phosphate; Ru5P, ribulose 5-phosphate; R5P, ribose 5-phosphate; S7P, sedoheptulose 7-phosphate; G3P, glyceraldehyde 3-phosphate; E4P, erythrose 4-phosphate; F6P, fructose 6-phosphate; G6P, glucose 6-phosphate; 6PGL, 6-phosphogluconolactone; 6PGC, 6-phospho-gluconate; KDPG, 2-keto-3-deoxy-6-phosphogluconate; F1,6P, fructose 1,6 bisphosphate; 3HB: 3-hydroxybutyric acid

Table 1.

Calculation of translational efficiency by RBSDesigner

		E. coli				M. alcaliphilum 20ZX
	Control: Ptac_RBS_pAWP89	P_R_1*	P_R_2*	P_R_2_MD1*	P_R_2_MD2*	P_R_1*	P_R_2*	P_R_2_MD1*	P_R_2_MD2*
Translational Efficiency	1.86E-06	0.17	1.71E-04	1.91E-04	1.91E-04	0.16	6.96E-03	2.52E-03	3.27E-03
RBS Exposure Probability	2.55E-18	1.07E-04	2.01E-14	2.72E-14	2.72E-14	6.67E-05	2.62E-09	1.69E-10	3.41E-10
RBS-Ribosome Complex Probability	4.80E-15	0.19	6.26E-10	8.45E-10	8.45E-10	0.15	1.38E-05	8.88E-07	1.80E-06
SD Hybridization Energy	-4.1 kcal/mol	-4.8 kcal/mol	-7.1 kcal/mol	-7.1 kcal/mol	-7.1 kcal/mol	-4.8 kcal/mol	-5.2 kcal/mol	-5.2 kcal/mol	-5.2 kcal/mol
Spacer Length	6	10	7	7	7	10	8	8	8
Spacer Efficiency	0.86	0.73	1	1	1	0.73	1	1	1

* Sequences of the control P_tac-RBS of pAWP89, P_R_1, P_R_2, P_R_2_MD1, and P_R_2_MD2 in E. coli and M. alcaliphilum 20ZX are listed in Supplementary Table S4

For the terminator, transcription terminators were predicted using iTerm-PseKNC, as described in the Methods section. To date, two types of transcriptional terminators have been investigated. Intrinsic or rho-independent terminators require a stem loop to terminate the transcription process [32]. Conversely, rho-dependent terminators require a specific rho protein to locate and bind to the signal sequence [33]. Thus, rho-dependent termination requires additional expression of the exogenous rho-coded gene in the host cell. In an endogenous system, host cells contain rho-coded genes. Therefore, the predicted transcription terminator from iTerm-PseKNC was selected to construct a semi-endo PRT system. iTerm-PseKNC predicted 239/4302 and 347/3920 terminator sequences in E. coli and M. alcaliphilum 20Z, respectively, with scores above the threshold of 0.9 (supplementary data S12). For M. alcaliphilum 20Z, the terminator motifs were extended in all genes because of the low number of predicted sequences with threshold 0.9 (supplementary data S13). MEME determined three and four motifs in E. coli (Fig. 1b) and M. alcaliphilum 20Z (Fig. 1c), respectively, with the highest number of matched sequences. These motifs were used to calculate the loop structure energy and predict the 2D structure using Seqfold and Forgi (supplementary data S20). Finally, candidate endogenous PRT sequences were prepared for the fine-tuning step before constructing the desired genes.

Fine-tuning the semi-endo PRT

After predicting the endogenous PRT system, a fine-tuning step will be carried out to optimize the sequence before conducting experiments. The endogenous promoters were initially optimized by altering nucleotides in the − 35 and − 10 boxes, following the canonical sequences of the sigma factor rpoD motif-binding site [34, 35]. The results showed that the change of the step-by-step of 1 nucleotide and 2 or 3 nucleotides in E. coli and M. alcaliphilum 20ZX increased the transcription rate (Tx rate) in the forward strand (Figs. 2a and 5a), no significant effect was observed in modifying some nucleotides in the reverse strand (supplementary materials, Figure S1). P_tac promoter and RBS sequence of pAWP89 with Zeocin^R gene (Fig. 2a bottom) or dTomato gene (Fig. 5a bottom) were used as a control. The input sequences for running the Promoter Calculator option in De Novo DNA were listed in Table S4 with P_R_2_MD2 and the target genes. The final fine-tuned endogenous promoter, which has changed the specific nucleotides, was named the semi-endogenous promoter. Semi-endogenous promoters were also attached to the predicted endogenous RBS sequences, and the transcription rate was continuously calculated using RBSDesigner (Table 1; the green sequences in Table S4). Two RBS sequences were given after the prediction steps (Fig. 1a and b). The feasible one was selected based on a translation rate calculation using P_tac-RBS in pAWP89 as the control. The translation rate efficiency of 0.000191 and 0.00327 for P_R_2_MD2 in E. coli and the M. alcaliphilum 20ZX strain were closest to the value of the conventional system, 0.00000186, in pAWP89. These sequences, P_R_2_MD2, were selected to conduct further experiments.

The promoter and RBS were independently evaluated for their terminator secondary structure and loop energy. Figures 2b and 5b present the predicted loop structural energy and 2D structure. Four terminators from M. alcaliphilum 20ZX showed loop energies of 1.7, -0.1, -0.6, and − 1.4, while three terminators from E. coli showed values of -3.3, -2.9, and 0.1, respectively. These energies were obtained using Seqfold, a Python-based tool for predicting the minimum Gibb free energy structure of nucleic acid. The prediction temperature was maintained at the default temperature, 37 °C. According to the definition of Gibb free energy, lower (more negative) values indicate more thermodynamically favorable structures. Therefore, terminator sequences with lowest energies were selected for constructing the expression system.

Finally, P_R_2_MD2 and terminators with ΔG values of -1.4 and − 3.3 were assembled with the target genes for further experiments. To evaluate the efficiency of the semi-endo PRT, the constructions were expressed in reference strain, E. coli, and type I methanotroph M. alcaliphilum 20ZX.

Evaluation of the semi-endo PRT system in the E. coli strain

Validation of the system by evaluating the expression of the dTomato and Zeocin^R genes in E. coli DH5α

To evaluate the efficiency of the semi-endo PRT system, a specific semi-endo PRT^E was constructed in E. coli using the dTomato and Zeocin^R genes of pAWP89 in Fig. 2. The candidate sequences shown by the workflow in Fig. 1a for the E. coli host strain were fine-tuned, followed by the calculation of the transcription rate (Tx) in Fig. 2a, and terminators were selected on the basis of structure 2b. The semi-endo sequences of P, R, and T were inserted into oligo primers for easy construction and were synthesized by Macrogen (described in the Materials and Methods section). E. coli DH5α (NZ_CP080399.1) and E. coli K12 MG1655 (NC_000913.3) possessed similar genome sizes (4.49 Mb and 4.46 Mb, respectively), and had similar numbers of genes (4429 and 4639, respectively). The semi-endo PRT system designed and synthesized for E. coli K12 MG1655 can be used for both strains. After transformation by heat-shock, the E. coli DH5α strain showed pinkish colonies in Fig. 2d in comparison with the control and E. coli K12 MG1655 strains (supplementary material Figure S3). This proved that the specifically designed and synthesized semi-endo PRT^E system could function in E. coli. In addition, a short sequence gene, the 375-bp Zeocin^R gene, was constructed using a semi-endo PRT^E system and expressed in E. coli strains. Only pAWP89-P_syn-R_syn-Zeocin^R-T_syn harbored colonies that could be grown on LB medium plus the zeocin antibiotic, as shown in Fig. 2e in comparison with the control. The wildtype strain of E. coli could not resist kanamycin (Kan) or zeocin on the LB agar plate plus Kan, as shown by the transparent circles. These results confirmed that the semi-endo PRT^E system functioned well in E. coli.

Short sequences of the semi-endo PRT^E system were successfully expressed in E. coli. To compare the expression levels and efficiencies of the Ptac promoter and PRT^E system, FACS was used to measure the dTomato signal in a single cell and to check the polyhydroxybutyrate (PHB) content by expressing phaCAB, a cluster gene required for biodegradable polymer biosynthesis. The dTomato and Zeocin proteins do not affect host metabolism. In contrast, the phaCAB cluster gene directly converts the core metabolite acetyl-coenzyme A (CoA) into polyhydroxybutyrate. Therefore, expression of the phaCAB cluster gene can demonstrate the efficiency of the PRT system in host metabolism.

Detection of the dTomato signal of semi-endo PRT system in the E. coli strain by FACS and evaluation of polyhydroxybutyrate biosynthesis with this system

To quantify the dTomato signal, we used FACS to measure the signal in a single recombinant E. coli cell line. In comparison with E. coli wildtype (E. coli DH5α) and E. coli with pAWP89-empty (E. coli DH5α including vector pAWP89 without gene), the recombinant strain including E. coli DH5α and E. coli K12 MG1655 showed a notable signal in Fig. 3a. Counting in the single-cell region (R1) and region 2 (R2) was gated on the basis of the fluorescence-intensive regions of interest, E. coli wildtype in Fig. 3c, and positive samples in Fig. 3d (supplementary material Figure S5). In R2, the values of the fluorescence intensity (MFI) are provided for each sample. On the basis of the MFI values, dTomato expression in the recombinant strain pAWP89-PRT^E-dTomato E. coli DH5α was 4-fold lower than in the recombinant strains pAWP89-P_tac-dTomato E. coli DH5α and pAWP89-P_tac-dTomato E. coli K12 MG1655. Thus, the semi-endo PRT activity was weaker than that of the P_tac system under the same conditions. In addition, we observed the expression of sfGFP and dTomato protein using the PRT system on LB agar plates under normal and UV light excitation in Fig. 3e.

Fig. 3.

Quantification and detection of the dTomato signal by FACS. (a) dTomato signal in different strains, including E. coli WT dTomato (E. coli DH5α), E. coli-p89-empty (E. coli DH5α with pAWP89-empty), E. coli DH5α with pAWP89-tac-dTomato, E. coli K12 MG1655 with pAWP89-tac-dTomato, and E. coli DH5α with pAWP89-PRT-dTomato. (b) Measurement and comparison of the dTomato signal on the basis of mean fluorescence intensity (MFI) values. The R1 region consisted of 10,000 single cells, and the R2 region was gated on the basis of the dTomato signal (MFI) in the region of interest in E. coli DH5α. (c) Recombinant pAWP89-PRT-dTomato-harboring E. coli DH5α. (d, e). Visualization of sfGFP and dTomato genes in pAWP89-PRT-sfGFP and pAWP89-PRT-dTomato under normal light and UV light. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3

This system was adapted to produce biodegradable polyhydroxybutyrate (PHB) in E. coli. We compared the efficiencies of the Ptac and PRT systems by constructing them using the phaCAB cluster gene in Fig. 4a. The results showed that E. coli grew better in the PRT system with higher OD and dry cell weight (DCW) values (Fig. 4a, b). Notably, E. coli expressing the P_tac promoter produced 6-fold less PHB than the PRT system in both M9 with xylose and LB medium, in terms of productivity (Fig. 4c) and titer (Fig. 4d). The Ptac promoter is known to be a strong promoter and was constructed using the high-copy number plasmid pAWP89. Producing PHB by competing with and using the core metabolite acetyl-CoA may affect cell growth, which was proven by the low OD and DCW values as well as the PHB content. PRT has a weaker expression system, as shown in Fig. 3b, which may enhance its biosynthesis. The trend of using weak or moderate promoters has been introduced in recent research [36]. The use of weak or moderate promoters can avoid the classic paradox associated with strong promoters in synthetic biology, which cause reduced growth rate due to ribosomal or machinery competition [37]; this scenario imposes a metabolic burden and results in a reduced yield by combining high-copy-number plasmids [38].

Fig. 4.

Validation of the semi-endo PRT^E system in E. coli with biosynthesis of the biodegradable polymer PHB. (a) Optical density (OD) of two recombinant strains that underwent transformation with pAWP89-Ptac-phaCAB and pAWP89-PRT-phaCAB with xylose as the sole carbon source. (b) Dry cell weight (DCW) of two recombinant strains. (c, d). Productivity (% mg PHB/mg DCW) and titer (mg/L) of two recombinant strains. M9X: M9 medium plus xylose 10 g/L. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3

Considering the effectiveness of the PRT system, we attempted to adapt it to produce PHB in engineered M. alcaliphilum 20ZX. However, we found that the phaZ gene encoded for a depolymerase enzyme. Therefore, we attempted to boost and produce 3-hydroxybutyric acid, a monomer of PHB, in M. alcaliphilum 20ZX, a type I methanotroph strain.

Application of the semi-endo PRT system in M. alcaliphilum 20ZX for producing 3-HB from methanol

The M. alcaliphilum 20Z genome contains genes that produce acetoacetyl-CoA and 3-hydroxybutyryl-CoA, as shown in Fig. 5e. However, we could not find phaC genes that encode polyhydroxyalkanoate synthase in its genome, although the phaZ gene could function as a depolymerase. The FadB gene can convert 3-hydroxybutyryl-CoA into crotonyl-CoA, and then into acetoacetyl-CoA and acetyl-CoA by the FadB and FadA genes in fatty acid oxidation. However, we were unable to detect PHB in our experiments. The presence of phaZ in the M. alcaliphilum 20Z genome requires phaZ deletion for PHB production if we wish to engineer the 20ZX strain to produce PHB. Although the 20Z strain could produce small amounts of 3HB, it did not exist for a long time because of the reverse pathway involving FadA and FadB. This is illustrated in Fig. 5e. To boost and produce higher 3-HB content in M. alcaliphilum 20ZX, the engineered M. alcaliphilum 20Z, which utilizes xylose as a carbon source [22] was used as the platform strain. The semi-endo PRT system was specifically designed for M. alcaliphilum 20ZX and constructed with the phaA and phaB genes from R. eutropha H16 in pAWP89, which encodes the acetyl-CoA acetyltransferase and acetoacetyl-CoA reductase enzymes in Fig. 5c. Two enzymes can convert acetyl-CoA, a core metabolite, into acetoacetyl-CoA, and then 3-hydroxybutyryl-coA, endogenous, which cleaves the CoA group in the cell and produces 3-hydroxybutyric acid (3-HB). The promoter of the 20Z strain was a fine-tuned nucleotide based on the calculated transcription rate, and the terminator was selected, followed by loop energy and structural prediction in Fig. 5a, b. During metabolism, engineered M. alcaliphilum 20ZX consumed methanol through the RuMP, ED, and/or EMP pathways, boosting and producing 3-HB, a monomer of the biodegradable polyhydroxybutyrate (PHB) polymer, through the expression of the constructed pAWP89-P_syn-R_syn-phaA-R_syn-phaB-T_syn in Fig. 3c, d, e. A positive recombinant 20ZX-3HBPRT colony was cultured in 50 mL of NMS medium with MeOH 1% (v/v) as the sole carbon source, and 3-HB was detected by HPLC. The results showed that a 3HB-response peak can be observed to overlap with the standard peaks (Fig. 5f). Additionally, the recombinant strain produced a large amount of formic acid (2 g/L) (Fig. 5g, supplementary material Figure S6). To observe the duration of 3-HB existence, 20ZX and recombinant strains were investigated over a period of time. The experiment was performed for 234 h to determine the OD, methanol consumption, and 3-HB production (Fig. 5e). The highest 3-HB content was 303 ± 6.39 mg/L after 234 h in recombinant strain 20ZX-3HBPRT, and it was 2-fold higher in 20ZX strain, 123.37 ± 16.16 mg/L, in Fig. 5d. However, the growth rate of recombinant 20ZX-3HBPRT was lower than that of the 20ZX strain.

Screening of promoters in M. alcaliphilum for IAA production revealed that the P_tac promoter yielded the highest titer of IAA at 10.5 mg/L compared to P_hps, P_MxaF and P_S−layer [39]. However, only 1.29% PHB was produced with P_tac promoter using xylose and methane as carbon sources [40]. Other promoters were screened in M. alcaliphilum, such as P_sps or inducible promoter (P_m) [41], their effectiveness in biosynthetic application has not been clearly evaluated. Here, we present an additional option with promised strategy for the expression system in M. alcaliphilum.

Discussion

De Novo DNA and RBSDesigner were developed using databases of common reference strains and lacked methanotroph-specific designs. Consequently, their application in methanotrophs is only partially effective, creating a gap between theoretical predictions and experimental outcomes. All translation and transcription calculations in PRT were higher than those in the reference, but the experimental results showed lower values (Table 1; Figs. 2a and 5a, and 3b). A comprehensive calculation of the combination is better than separate calculations. However, overexpression of phaA and phaB achieved a 3HB yield of 303 ± 6.39 mg/L, which was more stable and higher than those of the control 20ZX strain without the phaAB gene. In addition, the semi-endo PRT system can express the constructed genes from the log phase to the stationary phase in the 20ZX strain (Fig. 5e). The recombinant strain may have required time to adapt to this system. Interestingly, the PRT system can be produced weak dTomato signals and more PHB contents with a higher growth rate (Figs. 3b and 4a, c and d), it may be consistent with the findings of previous studies on the classic paradox in synthetic biology [37, 38, 42]. The use of the promoter, RBS, and terminator stems from the genome of the target host strain could easily adapted in host strain. This approach may have solved problems such as reducing the toxicity of the recombinant proteins or overexpression burdens on the cells to prevent plasmid loss, which present challenges to generate stable strains [43–47] or problems with strong promoters [37, 38]. Promoter library screening has been performed in Methylomonas sp. DH-1, where the recombinant strain produced 18.12 mg/L cadaverine [48], and in M. alcaliphilum, which produced 10.5 mg/L IAA [39]. However, further development and optimization of expression systems are needed to achieve higher titers and yields. The semi-endo PRT system represents a promising strategy for improving production performance, as it resulted in a two-fold increase in 3-HB production. To improve production in methanotrophs, engineering challenges should be comprehensively addressed, such as stabilized plasmid stabilization and prevention of plasmid loss, which was discussed in Cupriavidus necator H16 [49].

Genetic tools have been recently developed using specific mutagenized methane monooxygenase promoters in methanotrophs [50], and the phenol-inducible promoter of the CRISPR-base editor (BE) has been adopted to produce 2,090 mg/L mevalonate from methane [51]. These strains utilize methane as the sole source of carbon. CRISPRi was established in Cupriavidus necator H16 [52], and a strain that can grow on H₂ and CO₂ as energy and carbon sources was shown to produce 75 wt% (w/w) of the biodegradable polymer PHA from glucose [53] and 69.2 wt% (w/w) of cellular content copolymer (3-hydroxybutyrate-12.5 mol% 3-hydroxy hexanoate) on a gas mixture (wt% refers to mass fraction, and mol% refers to molar fraction) [54]. However, C. necator H16 lacks specific soluble methane monooxygenase (p/sMMO) genes that utilize methane. Newly isolated methanotrophic strains with novel characteristics, such as moderately thermophilic methanotrophs [55], have been obtained. Methanotrophic strains possess p/sMMO, but they lack genes for PHA synthesis, such as phaC, a PHA synthase, in type I methanotrophs, or they have an entire gene cluster for PHA synthesis; however, synthesis occurs under conditions of nitrogen/phosphorus limitation and an excess carbon source with low efficiency in type II methanotrophs, strict regulation [56–62]. Because of the need for a toolkit, the use of methane-utilizing strains poses challenges in the genetic engineering of metabolism to achieve desired products with high titers and yields. Almost all reports describing high amounts of target products involved engineered strains. This has led to the demand for the development of genetic toolkits for metabolic engineering.

This study demonstrated that a constitutive semi-endogenous PRT system can be obtained from target host strains and used as endogenous expression systems. The genetics of the newly isolated strains can be revealed by the rapid development of whole-genome sequencing techniques. Semi-endo PRT systems can be rapidly synthesized using genome sequences from the host strain. The desired gene can be easily constructed using a semi-endo PRT-based expression system with a compatible plasmid.

Conclusions

In this study, semi-endo PRT systems with promoters, RBSs, and terminator motif sequences in the genomes of target host bacteria were constructed and used as endogenous expression systems. The semi-endo PRT-based expression systems were developed using ML-based tools, and the transcription rate, translation rate, and Gibbs free energy of the terminator structure with a compatible plasmid were determined. This system was successfully tested with dTomato, Zeocin^R, and phaCAB genes in E. coli, and it produced 303 ± 6.39 mg/L 3HB after 234 h by utilizing methanol as the sole carbon source in the 20ZX strain expressing the phaA and phaB genes. This system will provide additional options for selecting an endogenous expression system for a target host strain.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PRT

Promoter, ribosome-binding site, and terminator

RBS

Ribosome-binding site

NMS

Nitrate minimal salt

ML

Machine learning

FACS

Fluorescence-activated cell sorting

FSC

Forward scatter

SSC

Side scatter

GFP

Green fluorescent protein

HPLC

High-performance liquid chromatography

GC

Gas chromatography

OD

Optical density

RID

Refractive index detection

PHB

Polyhydroxybutyrate

PHA

Polyhydroxyalkanoate

3-HB

Hydroxybutyric acid

TCA

Tricarboxylic acid cycle

p/sMMO

Particulate/solute methane monooxygenase

20ZX

M. alcaliphilum 20Z utilizing xylose

20ZX-3HBPRT

Engineered strain 20ZX with 3HBPRT producing 3-HB using the PRT system

Author contributions

KNP developed and performed the experiments and wrote and drafted the manuscript. EYL coordinated the study and edited, reviewed, and finalized the manuscript.

Funding

This research was supported by the Bio and Medical Technology Development Program of the National Research Foundation (NRF), funded by the Korean Government (MSIT) (No. RS-2025-02214232). This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean Government (MSIT) (RS-2024-00466473).

Data availability

Scripts are available on Github: (https://github.com/khoi91hp/semi-endoPRT) Or Figshare: (https://doi.org/10.6084/m9.figshare.27219729).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.