Transient analysis and engineering of bacterial metabolic pathways using cell-penetrating peptide-peptide nucleic acid conjugates

Abstract

Background

Conventional genetic engineering approaches for bacterial metabolic pathway manipulation, although highly applicable, still face limitations including metabolic burden, irreversibility, and dependency on host cellular machinery. Cell-penetrating peptide-peptide nucleic acid conjugates (CPP-PNAs), known for their applicability as antibacterial tools and in the elucidation of protein function, offer a promising alternative to overcome such limitations. Since the application of CPP-PNA in metabolic engineering and pathway elucidation remains largely unexplored, we developed and validated a CPP-PNA platform using Synechocystis sp. PCC 6803 as a model system to demonstrate targeted metabolic pathway evaluation.

Results

High compatibility and dose-dependent permeation efficiency in strain PCC 6803 was first observed when the amphipathic CPP (KFF)₃K was employed, achieving clear cell growth inhibition at 10 µM and above. Specific targeting of D-lactate dehydrogenase (Ddh) using CPP-Syn6803ddh conjugates achieved near-complete protein translation knockdown within 24 h, as confirmed by Western blot analysis. Metabolomics analysis using LC-MS on predetermined metabolites revealed that CPP–PNA treatment produced metabolic effects comparable to stable genetic knockout strains, with both approaches showing a significant 2.5-fold increase in pyruvate accumulation compared to wild-type controls. Further elucidating the reason for pyruvate accumulation, we observed compensatory activation of the glyoxalase pathway at 48 h post-treatment, resulting in 3-fold increased D-lactate production presumably through methylglyoxal detoxification. Validating this observation, RT-qPCR analysis confirmed 2-3-fold upregulation of the glyII gene, encoding for the glyoxalase II (GlyII) enzyme, in both CPP-PNA treated and knockout strains, while double CPP-PNA inhibition experiments targeting both Ddh and glyoxalase pathways suppressed D-lactate accumulation.

Conclusions

This study establishes CPP-PNAs as efficient tools for rapid, and simple metabolic pathway investigation. The approach produces results comparable to conventional genetic knockouts while offering dose-dependent control and avoiding permanent genomic alterations. Our findings reveal unexpected metabolic complexity in Synechocystis sp. PCC 6803 D-lactate synthesis under light conditions and demonstrate the utility of CPP-PNA for uncovering compensatory pathway activation. This platform represents a valuable addition to bacterial genetic engineering, addressing some of the critical limitations faced by conventional approaches, while showing potential for further understanding the biochemistry of metabolite-producing bacteria.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12934-025-02909-4.

Introduction

Advancements in bacterial genetic engineering has made it possible to develop “super strains” or genetically modified bacteria that act as biological factories for sustainable production of countless bioproducts including pharmaceuticals, biofuels, and industrial chemicals from renewable carbon sources. Our ability to produce such strains has enabled breakthrough applications across clinical medicine, diagnostics, and biotechnology, while supporting the development of circular bio economies and contributing to global efforts toward reducing carbon emissions and promoting environmental sustainability [1–3]. However, central to developing these engineered bacterial strains, gene expression regulation and metabolic pathway manipulation approaches are key and crucial to enable optimized cellular outputs and precise control over microbial behavior.

Current approaches for genetic manipulation in bacteria include plasmid-based overexpression systems [4], homologous recombination for chromosomal integration [5], CRISPR/Cas9-mediated genome editing [6], and CRISPR interference (CRISPRi) [7]. These tools have enabled high-throughput pathway optimization, combinatorial gene control, and rewiring of regulatory circuits. Nevertheless, despite their utility, each approach carries inherent limitations that constrain broader application across diverse bacterial species and limit the ability to fully elucidate the dynamics of metabolic pathway regulation. Plasmid-based approaches, while easy to implement, often impose a significant metabolic burden on host cells, interfere with native regulatory networks, and suffer from instability in long-term cultures [8–10]. In contrast, chromosomal modifications offer greater stability but are labor-intensive, irreversible, and can introduce off-target effects or polar mutations [11–13]. CRISPR-based systems, on the other hand, while revolutionary, require co-expression of guide RNAs and effector proteins, increasing the genetic load and raising concerns over cytotoxicity and unintended editing outcomes [14, 15]. Crucially, most current tools are also highly dependent on the host’s transcriptional and translational machinery, resulting in the difficulty to establish simple, rapid and tunable evaluation systems.

Thus, there is a strong demand for strategies that can further address the limitations of conventional genetic manipulation tools. One such potential approach involves the use of peptide nucleic acids (PNAs)—synthetic DNA analogs characterized by a peptide-like backbone [16, 17]. PNAs exhibit exceptional binding affinity and sequence specificity and their resistance to enzymatic degradation makes them ideal for interfering with gene expression at the transcriptional or translational level. Unlike traditional genetic tools, PNA-based regulation is transient and does not rely on the host’s genetic machinery, offering highly tunable and programmable control over gene function. However, the effective intracellular delivery of PNA remains a key challenge [18]. In this context, cell-penetrating peptides (CPPs) have emerged as powerful facilitators of intracellular delivery. Initially explored for their roles in eukaryotic systems, CPPs—short, typically cationic peptides—have proven highly effective at transporting various biomolecules, including nucleic acids and proteins [19]. By interacting with negatively charged membrane components, CPPs enable direct translocation or endocytosis-independent entry into cells. Recent studies have harnessed CPPs to deliver PNAs specifically designed to silence critical bacterial genes, demonstrating their potential as precise antibacterial agents [20, 21].

CPP-PNAs, having a straightforward design, chemical stability and efficient cellular permeation applicable with diverse bacterial strains have emerged as an effective tool for targeted protein translation inhibition. These molecular constructs combine the cell-penetrating capabilities of amphipathic peptides with the sequence-specific binding properties of peptide nucleic acids, enabling rapid and selective modulation of protein translation without genetic manipulation (Figure S1). In antimicrobial research, CPP-PNAs have shown significant promise for combating antibiotic resistance, where they were first used to sensitize resistant E. coli to ampicillin by targeting β-lactamase genes [22]. Subsequent studies have successfully targeted essential genes including mecA in MRSA for oxacillin sensitivity restoration [23] and ftsZ for cell division disruption in Staphylococcus aureus [24]. Infectious disease research has extended these applications to tuberculosis, where anti-inhA PNAs target fatty acid synthesis pathways in Mycobacterium tuberculosis [25], and to intracellular pathogens including E. coli K1, Salmonella, and Shigella for enhanced antibiotic delivery [26]. In the field of microbiome engineering, selective bacterial growth inhibition in consortia for creating artificial microbiomes was demonstrated [27], while synthetic biology applications include CPP-mediated transformation of large plasmids in E. coli [28] and genetic engineering through plasmid curing and marker gene excision [19]. Agricultural biotechnology has benefited from CPP-PNA applications in plant pathology, particularly for fire blight control in Erwinia amylovora as an alternative to antibiotic treatments [29].

While these diverse applications demonstrate the versatility of CPP-PNAs across multiple bacterial systems, their potential as tools for metabolic pathway analysis in metabolite-producing bacteria remains largely unexplored. In this study, we present a platform for using CPP–PNAs as molecular tools to probe and understand bacterial metabolism in metabolite-producing bacteria. Metabolite-producing bacteria, including Streptomyces, Pseudomonas and various cyanobacteria, have long been targets for metabolic pathway analysis. However, despite numerous proposed engineering methods for these microbes, limitations persist, particularly regarding strain-specific transformation efficiency and genetic accessibility [30, 31]. As a result, we still depend on bacterial species such as Escherichia coli, in which genetic engineering approaches are well established. Nevertheless, using non-native strains to elucidate metabolic pathways presents several challenges. First, the molecular cloning of large gene clusters can be tedious and time-consuming [32], while expression of the genes upon successful transformation may cause cytotoxicity or prove incompatible with the new host [33, 34]. Additionally, critical aspects such as understanding metabolic fluxes and the relationships between different metabolic pathways cannot be adequately addressed [35]. These challenges are more pronounced in non-model organisms, where established transformation protocols may be inefficient or are yet to be established.

To demonstrate the potential of CPP-PNA technology for metabolic pathway investigation and its applicability with metabolite-producing bacteria, we selected the cyanobacterium Synechocystis sp. PCC 6803 as our model organism. This bacterium serves as an ideal proof-of-concept system because its metabolic background is well characterized and its well-established genetic methods allow us to validate CPP-PNA results against conventional approaches, particularly through the establishment of stable knockout strains [36]. By demonstrating similar metabolic trends between CPP-PNA and genetic knockouts in this model, we aim to establish confidence in the method’s accuracy and provide a foundation for subsequent application to other metabolite-producing bacteria or non-model strains where genetic validation remains challenging. We first evaluated the compatibility and delivery efficiency of CPP–PNA in the target strain, then demonstrated the potential of these conjugates to transiently regulate metabolic fluxes and provide new insights into pathway dynamics by specifically targeting the D-lactate dehydrogenase (Ddh) enzyme that regulates the D-lactate-pyruvate pathway. This work highlights the versatility of CPP–PNAs as rapid, non-genomic, programmable regulators for metabolic studies and engineering applications in bacteria.

Materials and methods

Bacterial strain and culture conditions

Synechocystis sp. PCC 6803 was obtained from the Pasteur Culture Collection (PCC), Institut Pasteur, Paris, France. Maintenance and cultivation of the wild-type and knockout strains for all experiments were performed under white light illumination (60 µmol m⁻² s⁻¹) provided by an LC-LED450W system (TAITEC Corp., Saitama, Japan) at 30 °C with shaking at 120 rpm using a BR-43 FL shaker (TAITEC Corp., Saitama, Japan). For establishing the working culture, cryopreserved cells were inoculated into a 100 mL conical flask containing 20 mL of BG11 liquid medium. Once cell growth was confirmed, the culture was continuously maintained in liquid medium throughout the experiment. The working culture was passaged every two weeks or when preparing experimental subcultures for CPP-PNA evaluation. Passaging was performed by transferring 200 µL of the previous culture into 20 mL of fresh medium. To prepare experimental subcultures, a one-week-old working culture was transferred to a 50 mL centrifuge tube, centrifuged at 3,000 × g for 10 min at 30 °C, and the supernatant was discarded. The bacterial pellet was then resuspended in a fresh medium and adjusted to an optical density at 730 nm (OD₇₃₀) of 0.2 before use. The Δddh and Δddh-ΔglyII strains, lacking the gene encoding for D-lactate dehydrogenase (Ddh), ddh (slr1556), or both the ddh and the gene encoding for glyoxalase II (GlyII), glyII (sll1019), respectively, were established using standard homologous recombination. Detailed information on knockout strain establishment is provided in Additional File 1.

CPP-PNA design and synthesis

The CPP-PNAs used in this study were designed and synthesized based on a previous report [37]. In brief, the CPP employed was KFFKFFKFFK ((KFF)₃K), an amphipathic 10-amino-acid peptide composed of lysine (K) and phenylalanine (F) [38]. The PNA sequences for the CPP-Syn6803acpP, CPP-Syn6803dnaB, CPP-Syn6803ftsZ, CPP-Syn6803ddh, and CPP-Syn6803glyII conjugates were specifically designed to target the start codons of mRNAs of genes encoding the acyl carrier protein (AcpP), acpP (ssl2084), replicative DNA helicase (DnaB), dnaB (slr0833), cell division FtsZ protein (FtsZ), ftsZ (sll1633), Ddh and GlyII, respectively. As a negative control, a non-specific CPP-PNA conjugate, CPP-ncPNA, designed not to bind to the start codon of any mRNA in strain PCC 6803, was also synthesized. Sequences of each of the conjugates are presented in the Additional File 1 (Table S1). The PNA sequences were conjugated to the C-terminus of the CPP (KFF)₃K via a 2-aminoethoxy-2-ethoxy acetic acid (AEEA) linker. All CPP-PNAs were synthesized using standard solid-phase peptide synthesis protocols, purified by reversed-phase high-performance liquid chromatography (RP-HPLC; Shimadzu, Kyoto, Japan), and verified by mass spectrometric analysis using a MALDI-TOF MS system (Autoflex Speed TOF, Bruker Japan K.K., Kanagawa, Japan). The synthesized CPP-PNAs were stored in water at a concentration of 1 mM as stock solutions at -20 °C for subsequent experiments.

CPP-PNA permeation validation and targeted metabolic inhibition

To validate the permeation of CPP-PNA in Synechocystis sp. PCC 6803, the CPP-Syn6803acpP, the CPP-Syn6803dnaB and the CPP-Syn6803ftsZ conjugates were used. All experiments hereafters were conducted in triplicate for reproducibility and result validation. Several 10 mL experimental subcultures were prepared at an initial OD₇₃₀ of 0.2 in 100 mL conical flasks containing BG11 medium and pre-incubated for 23 h to ensure culture stabilization prior to CPP-PNA treatment. At the 23-hour mark, the CPP-PNAs were added to achieve subcultures with final probe concentrations of 2, 4, 6, 8, and 10 µM, respectively. Additional subcultures with no conjugate addition or with addition of 10 µM of the non-specific CPP-ncPNA conjugate were also prepared as negative controls. Hour 0 was defined as the time point 24 h after pre-incubation initiation and one hour after CPP-PNA addition. Cell growth was subsequently monitored by measuring OD₇₃₀ at 0-, 6-, 24-, 48-, and 72-hour time points.

For the protein translation inhibition of metabolic pathways using CPP-PNA, we targeted the Ddh enzyme responsible for D-lactate-pyruvate conversion (EC 1.1.1.28) using the CPP-Syn6803ddh conjugate and evaluated its effects on pyruvate and its surrounding metabolites. To further validate the effects of D-lactate-pyruvate pathway inhibition, the CPP-Syn6803glyII conjugate was additionally employed. Both conjugates were used at a final concentration of 10 µM. Cells were harvested at 24- and 48-hour time points, and the designated metabolites were isolated and measured using LC-MS. The Δddh and Δddh-ΔglyII strains were used as controls to validate the effects of CPP-PNA inhibition.

Metabolite extraction and sample preparation

At each OD₇₃₀ measurement time point, 8 mL of culture was harvested, and cells were gently entrapped onto a 0.2 μm membrane (Millipore Sigma, MA, USA) placed in a filter holder under aspiration filtration. Cell-entrapped membranes were transferred to disposable petri dishes and were briefly rinsed with 10 mL of 20 mM ammonium bicarbonate (pre-chilled at 4 °C). Upon washing, the membranes were each placed in 5 mL centrifuge tubes, and the tubes were submerged and rapidly cooled using liquid nitrogen for 30 s to immediately halt cellular metabolism. Cold methanol-water mixture (1 mL, 80:20, v/v) was added to the samples, and further quenching was performed for 1 h at -80 °C. Chloroform (1.5 mL) was subsequently added to the samples, and bacteria adhering to the membrane were washed off by gentle pipetting. The samples were then transferred to 15 mL centrifuge tubes and agitated for 10 min at 4 °C using a petiSyzer (HiPep Laboratories, Kyoto, Japan). After adding 1.25 mL of deionized water, the samples underwent another round of shaking at 4 °C for 5 min. Following phase separation by centrifugation at 3,000 × g for 15 min at 4 °C, 1.6 mL of the aqueous (top) phase was collected and transferred to 2 mL centrifuge tubes. Methanol was removed from the aqueous phase by evaporation using a centrifugal concentrator (CC-105; TOMY, Osaka, Japan) for 45 min. To further concentrate the metabolites, samples were lyophilized at -80 °C for approximately 24 h. The lyophilized metabolites were resuspended in 50 µL of deionized water, thoroughly mixed, and prepared for mass spectrometric analysis.

LC-MS analysis and metabolite quantification

LC-MS analysis was performed using an LC-20AD solvent delivery system (Shimadzu, Kyoto, Japan) coupled to an X500R Q-TOF mass spectrometer (AB SCIEX, Framingham, MA, USA) operating in negative ESI mode. The interface parameters were set as follows: curtain gas, 45 psi; ion spray voltage, 5500 V; temperature, 700 °C; ion source gas 1, 70 psi; ion source gas 2, 60 psi; and collision gas, 7 psi. Chromatographic separation was achieved using an Intrada Organic Acid column (150 × 2 mm) maintained at 37 °C with a flow rate of 0.2 mL/min. The mobile phases consisted of solvent A (acetonitrile/water/formic acid = 10:90:0.1, v/v/v) and solvent B (acetonitrile/100 mM ammonium formate = 10:90, v/v), with the following gradient: 0–1 min, 0% B; 1–7 min, 0-100% B; and 7–10 min, 100% B. For quantification, standard curves were prepared using pure standards of D-lactate, pyruvate, phosphoenolpyruvate (PEP), acetyl-CoA, and malate at concentrations of 1, 5, 10, 25, 50, and 100 µg/mL. Acetyl-CoA, pyruvate, and malate were prepared as a mixture, while D-lactate and PEP were prepared separately.

Protein quantification and western blotting

10 mL of bacterial cultures were centrifuged at 3,000 × g for 10 min, and the pellets were washed twice with BG11 medium before resuspension in 1 × PBS. The cell suspensions were rapidly frozen in liquid nitrogen and stored at − 80 °C. For protein extraction, cells were disrupted using 0.5 mm silica beads (BioSpec Products Inc., Bartlesville, OK, USA) through 10 cycles of 30 s vortexing on ice. Following centrifugation (15,000 × g, 20 min, 4 °C), the supernatant was collected, and protein concentration was determined by the Bradford method using albumin standards (Thermo Fisher Scientific Inc., Waltham, MA, USA). Proteins (2.0 µg per sample) were denatured in SDS sample buffer at 95 °C for 10 min, separated by SDS-PAGE (SuperSep™ Ace 10–20%, FUJIFILM Wako Pure Chemical Corp., Osaka, Japan), and transferred onto nitrocellulose membranes. Membranes were blocked with PBS-MT (5% skim milk in PBS-T) and incubated with anti-Ddh primary antibody (Cosmo Bio Co., Ltd., Tokyo, Japan; 1:4,000), followed by HRP-conjugated secondary antibody (Promega Corp., Madison, WI, USA; 1:2,500). Signals were detected using Immobilon solution (EMD Millipore Corp., Billerica, MA, USA) and visualized with a LAS-4000 mini scanner (GE Healthcare Life Sciences, Little Chalfont, Buckinghamshire, UK). Band intensities were visually analyzed and quantified accordingly.

RNA extraction and quantitative PCR analysis

After centrifugation of three 10 mL subculture samples (total volume = 30 mL) at 3,000 × g for 15 min at 4 °C, the cell pellets were resuspended in 150 µL of 1 × PBS and then preserved in 2.0 mL of RNAlater (Thermo Fisher Scientific K. K., Tokyo, Japan). RNA extraction was performed using TRI Reagent (Cosmo Bio Co., Ltd., Tokyo, Japan) and chloroform according to the manufacturer’s protocol. After vortexing, samples were incubated for 10 min at room temperature, followed by centrifugation at 10,000 × g for 15 min. The aqueous phase was collected and mixed with chloroform, then centrifuged again at 10,000 × g for 5 min. The resulting upper aqueous phase, containing total RNA, was carefully transferred and mixed with an equal volume of 75% ethanol (v/v) in nuclease-free water. Total RNA was then isolated and purified using the RNeasy Protect Mini Kit (QIAGEN K. K., Tokyo, Japan). Genomic DNA was removed by on-column DNase digestion using the QIAGEN RNase-Free DNase Set. The DNase I solution (20 µL) was applied directly onto the RNeasy column and incubated at 30 °C for 30 min, following the manufacturer’s instructions. Purified total RNA (2 µg) was reverse transcribed into cDNA using PrimeScript™ IV 1st strand cDNA Synthesis Mix (Takara Bio Inc., Shiga, Japan). Quantitative PCR was performed using 5 ng of cDNA with PowerUp™ SYBR™ Green Master Mix (Thermo Fisher Scientific K. K., Tokyo, Japan) to determine the transcription levels of glyII and rnpB. The rnpB gene encodes the RNA subunit of RNase P and serves as a widely used internal reference gene in Synechocystis sp. PCC 6803 [39] (Table S2). Relative expression levels were calculated using the ΔΔCt method and normalized to the expression of rnpB.

Data and image analysis

All graphs were generated, and statistical analyses were performed using GraphPad Prism10 v10.4.0 (GraphPad Software Inc., San Diego, CA, USA). One-way ANOVA was used for statistical comparisons, and significance was determined at p < 0.05. Band intensities from Western blot images were normalized to the SDS-PAGE results (Figure S2) and quantified using ImageJ (https://imagej.net/ij/) and analyzed as described in Additional File 1.

Results

CPP-PNA demonstrates compatibility and specificity in Synechocystis sp. PCC 6803

Prior to employing CPP-PNAs for understanding and elucidating metabolic pathways in Synechocystis sp. PCC 6803, we first validated the compatibility of the CPP (KFF)₃K and evaluated the inhibitory effects of PNA within the bacterium. This validation was performed using the 3 CPP-PNAs, each designed to bind to the mRNA of the AcpP, DnaB and FtsZ proteins respectively and inhibit cell growth. Our results demonstrated that the conjugates induced dose-dependent growth inhibition from 24 h, with moderate suppression at 4 µM and substantial reduction at 10 µM (Fig. 1). Validation of targeted inhibition was supported by treating strain PCC 6803 with a non-specific CPP-PNA probe (CPP-ncPNA), which resulted in no growth inhibition and exhibited a growth curve similar to the wild-type strain. These results indicate that each of the CPP-PNAs was compatible with and efficiently permeates strain PCC 6803, while specific translation inhibition of target proteins can be achieved through proper PNA design. Based on these findings, we proceeded with using the CPP (KFF)₃K and selected 10 µM as the working concentration for subsequent metabolic pathway investigations.

Fig. 1.

Confirmation of CPP-PNA intracellular permeation. Growth curves were obtained for strains treated with CPP-Syn6803acpP, CPP-Syn6803dnaB and CPP-Syn6803ftsZ conjugates, each inhibiting the translation of the AcpP, DnaB and FtsZ proteins respectively. Cells were cultured in the presence of various concentrations of CPP-PNA (2, 4, 6, 8, and 10 µM), and OD₇₃₀ values were measured over time. WT (wild type) represents cells without CPP-PNA treatment, and NC (negative control) represents cells treated with 10 µM of non-specific CPP-ncPNA. Data represent the mean ± standard deviation (SD) of triplicates

Having validated the compatibility and inhibitory effects of CPP-PNA in Synechocystis sp. PCC 6803, we next confirmed the specificity of CPP-PNA towards the target enzyme of interest. In this work, we targeted the Ddh enzyme, which is responsible for D-lactate-pyruvate conversion. Upon treatment of strain PCC 6803 with the CPP-Syn6803ddh conjugate and monitoring its effects on Ddh protein levels, the time-course evaluation by Western blot analysis revealed progressive downregulation of the enzyme over 48 h, with near-complete knockdown achieved by 24 h (Fig. 2). The target Ddh was most prominent at the molecular weight of 36.9 kDa, validating that the CPP-Syn6803ddh probe was specific, and the 24-hour timepoint was optimal for subsequent experiments since it corresponds with the growth inhibition effects observed in Fig. 1. Other bands were observed above the target enzyme but since the intensities were low and did not show a dose-dependent effect, we concluded that they were possibly non-specific bound proteins to the Ddh antibody and therefore negligible.

Fig. 2.

Western blot analysis of Ddh protein expression following CPP-PNA treatment. Wild-type Synechocystis sp. PCC 6803 cells and cells inhibited with 10 µM of the CPP-Syn6803ddh conjugate were used. The numerical values below the observed bands represent the relative expression levels of Ddh normalized to the untreated control (set to 1.00, shown in red). Protein loading, unified at 2 µg, was normalized based on the SDS-PAGE analysis

Metabolomics analysis showed CPP–PNA efficacy in pathway analysis

To comprehensively assess the effects of Ddh inhibition due to CPP-PNA treatment and to compare the results with conventional genetic techniques, we performed quantification of pyruvate at the 24-hour timepoint within the wild-type Synechocystis sp. PCC 6803 as the negative control, a strain treated with the CPP-Syn6803ddh conjugate as the CPP-PNA treated sample and a Δddh knockout strain. For further comprehension, we extended this analysis to the surrounding metabolites directly related to pyruvate as well. Using LC/MS, metabolomics analysis interestingly showed a significant 2.5-fold increase in both the Δddh and CPP-PNA treated strains compared to the untreated wild-type strain (Fig. 3a), while D-lactate levels were consistently low across all three strains (Fig. 3b). Other metabolites directly linked to pyruvate, phosphoenolpyruvate (PEP), malate, and acetyl-CoA, on the other hand, showed minimal or no changes in concentration (Fig. 3c–e). These observations suggest that CPP-PNA produced results comparable to conventional genetic knockout approaches. It also showed that inhibition of the Ddh enzyme, whether through genetic knockout or CPP-PNA-mediated translational suppression, impacts pyruvate accumulation.

Fig. 3.

Quantification of pyruvate accumulation and its surrounding metabolites. Wild-type Synechocystis sp. PCC 6803 cells, cells inhibited with 10 µM of the CPP-Syn6803ddh conjugate, and a Δddh strain were used. The relative amounts of pyruvate (a) and its surrounding metabolites (b–e; D-lactate, PEP, malate, and acetyl-CoA) were quantified at 24 h. Each metabolite level was normalized to the WT at each time point, and measurements were performed using LC-MS. Data are presented as mean + SD (N = 3). Statistical significance was determined by ANOVA (* *p < 0.01)

Ddh Inhibition suggested compensatory glyoxalase pathway activation and D-lactate production in Synechocystis sp. PCC 6803

After observing increased intracellular pyruvate levels at 24 h post-Ddh inhibition, we hypothesized that this accumulation resulted from altered metabolic flux through other pathways connected to pyruvate. Here, we specifically investigated whether enhanced glycolytic activity might be responsible for pyruvate accumulation. To validate this assumption, we looked at the glyoxalase pathway since in Synechocystis sp. PCC 6803, methylglyoxal (MG), a toxic byproduct, is produced during increase glycolysis which is then converted to D-lactate for detoxification (Fig. 4a; [40]). Although D-lactate synthesis was low at the 24-hour timepoint, assuming that there may be a delayed response, we extended our analysis to 48 h post-treatment and observed a commendable 3-fold increase in D-lactate accumulation in both the CPP-Syn6803ddh treated and Δddh strains compared to the wild-type strain (Fig. 4b). To further confirm that this D-lactate accumulation resulted from upregulation of the glyoxalase pathway rather than alternative routes, we conducted two validation experiments.

Fig. 4.

D-lactate biosynthesis and intracellular accumulation 48 h post-CPP-PNA treatment. a The schematic illustrates the glyoxalase pathway related to glycolysis and D-lactate synthesis. Methylglyoxal (MG), a toxic byproduct of increased glycolysis, is detoxified to D-lactate by the glyoxalase II (GlyII) enzyme. b The bar graph shows the relative intracellular D-lactate levels in 10 µM of CPP-PNA-treated cells and the Δddh strain 48 h after CPP-PNA treatment, quantified by LC-MS and normalized to the untreated wild-type (WT) control. Data represent the mean + standard deviation (SD) of at least three independent replicates. Statistical significance was determined by ANOVA (* p < 0.05)

First, we examined expression levels of the glyII gene, which encodes for GlyII, the enzyme responsible for converting MG to D-lactate. RT-qPCR quantification of glyII transcript levels showed an approximately 2–3-fold increase in both the CPP-PNA-Syn6803ddh–treated strain and the Δddh strain (Fig. 5a). Second, we performed a double inhibition experiment targeting both the D-lactate-pyruvate and glyoxalase pathways simultaneously. When strain PCC 6803 was treated with CPP-PNAs targeting both the mRNA of Ddh and GlyII, we observed suppression of D-lactate to levels nearly equivalent to the wild-type strain at 48 h post-treatment (Fig. 5b). We validated these findings by comparing the results with a double knockout strain (Δddh-ΔglyII), which showed a similar metabolic profile. Collectively, these results provide evidence that D-lactate accumulation observed after Ddh inhibition, stems from upregulation of the glyoxalase pathway, likely in response to increased glycolytic flux contributing to pyruvate accumulation.

Fig. 5.

Validation of upregulation of the glyoxalase pathway upon CPP-PNA treatment. a Relative expression levels of the glyII in CPP-PNA-treated cells and the Δddh strain 48 h after CPP-PNA treatment, measured by RT-qPCR. Data represent the mean + standard deviation (SD) of at least three independent replicates. The difference between the two groups was not statistically significant (ns; unpaired two-tailed t-test, P = 0.53). b The bar graph shows the relative intracellular D-lactate levels of the wild-type Synechocystis sp. PCC 6803 strain (WT), cells treated with 10 µM CPP-Syn6803ddh and CPP-Syn6803glyII conjugates, and the Δddh-ΔglyII strain at 48 h post-CPP-PNA treatment, quantified by LC-MS and normalized to the untreated WT control. Data represent the mean + standard deviation (SD) of at least three independent replicates. Statistical significance was determined by ANOVA (* p < 0.05, ** p < 0.01)

Discussion

In this study, we explored the potential of CPP-PNA as tools for metabolic pathway analysis, extending their use beyond previous antibacterial and molecular biology applications. Our findings demonstrate that CPP-PNAs can be effectively applied to pathway elucidation in cyanobacteria, highlighting their potential for broader use in metabolite-producing bacteria.

Before applying CPP-PNA to study metabolic pathways in cyanobacteria or any other bacterial strain, it is crucial to first determine the compatibility of the CPP used and, secondly, assess the specificity of PNA to inhibit the target protein or enzyme. CPP compatibility is required since CPP permeation in bacteria tends to be strain specific [41]. Furthermore, high permeation efficiency is particularly important as it ensures robust inhibition of target proteins and enables the dose-dependent control of protein activity necessary for comprehensive enzyme characterization in metabolic pathways [29]. Prior to this work, our group has evaluated and optimized the permeability of CPPs to improve its applicability with diverse bacterial strains and have employed the found conditions with several bacterial strains including E. coli, 10 bacterial strains classified within the gut microbiome family and against a Paenibacillus isolate [37, 42, 43]. Similarly, we demonstrated in this work, that the CPP (KFF)₃K was compatible with Synechocystis sp. PCC 6803, as evidenced by similar dose-dependent growth inhibition following treatment with CPP-PNAs, compatible with the mRNAs of the AcpP, DnaB and FtsZ proteins, respectively (Fig. 1).

Having validated the delivery system, we next sought to demonstrate the regulatory potential of CPP-PNA in understanding metabolic pathways. Focusing on the metabolite for analysis, we selected pyruvate as our model because pyruvate metabolism is one of the most well-characterized metabolites present across all bacteria, with comprehensively understood surrounding pathways and readily quantifiable metabolites [44, 45]. Pyruvate concentration, as far as we know, is dependent on 4 metabolites, phosphoenol pyruvate (PEP), Malate, D-lactate and Acetyl-CoA, where their conversion to pyruvate and vice-versa are closely regulated by the respective conversion enzymes. Since CPP-PNAs target and inhibit the translation of proteins, for us to demonstrate the direct effects of CPP-PNA in the elucidation of metabolic pathways, we needed to target either one of the metabolite conversion enzymes. Furthermore, direct visualization of the enzyme due to CPP-PNA inhibition was also important. Thus, we needed to select an enzyme with commercially available antibodies. Among the metabolite conversion enzymes, only malate dehydrogenase, associated to malate-pyruvate conversion, and Ddh, associated to D-lactate-pyruvate conversion, had commercial antibodies. We selected Ddh as our primary target because it was the least well-characterized of the two enzymes.

Following successful Ddh inhibition as shown in the Western blot analysis (Fig. 2), we evaluated pyruvate concentration and its related metabolites with and without CPP-PNA treatment and compared these results with the Δddh control strain. Interestingly, we found elevated pyruvate concentrations in both the CPP-PNA treated strain and the Δddh control (Fig. 3a). Since we did not observe any significant changes in concentrations of other surrounding metabolites, including PEP, malate, Acetyl-CoA and expectedly D-lactate (Fig. 3b–e), we hypothesized that pyruvate accumulation may have resulted from increment in metabolic flux via one of the pathways connected to pyruvate.

Based on a previous report, we knew that D-lactate synthesis in Synechocystis sp. PCC 6803 can occur via another pathway, the glyoxalase pathway. This pathway is triggered when glycolysis increases in the bacterium, where the toxic methylglyoxal (MG) is produced as a byproduct but is neutralized by converting MG to D-lactate [40]. If the glyoxalase pathway were triggered due to increased glycolytic activity, we would expect to observe D-lactate production in both the CPP-PNA treated cells and the Δddh strain at the 24-hour time point. However, this was not observed, prompting us to consider whether metabolic pathway activation might also be dependent on cellular response. To test this possibility, we extended our CPP-observation of D-lactate to 48 h. This time lag hypothesis, although speculative, could be explained by reports suggesting that pathway activation may require a period of adaptation even in stable knockout strains such as the Δddh strain. Similarly, for the CPP-PNA treated cells, D-lactate accumulation may point toward a non-instantaneous engagement of the glyoxalase pathway, possibly triggered upon pyruvate elevation [46, 47]. Intriguingly, both strains showed an increase in D-lactate synthesis compared to the wild-type strain (Fig. 4). Following these observations, we performed RT-qPCR to confirm upregulation of the glyII expression (Fig. 5a) and conducted double CPP-PNA treatment experiments with the Δddh-ΔglyII strain to confirm downregulation of D-lactate synthesis (Fig. 5b).

From the 48-hour D-lactate synthesis result, possibly derived from the glyoxalase pathway, we therefore propose that inhibiting the D-lactate-pyruvate pathway may affect pyruvate accumulation. The absence of D-lactate production in the wild-type strain supports this interpretation. This result was surprising since the D-lactate-pyruvate pathway controlled by the Ddh enzyme is typically activated under dark fermentation conditions [48]; thus, inhibiting the Ddh enzyme would be expected to have no effect under light conditions. Since our experiments were conducted under light conditions, which we believe have not been evaluated before in strain PCC 6803, D-lactate synthesis in this strain may be more complicated than we initially presumed. While these findings reveal unexpected metabolic complexity, our main goal in this work was to demonstrate the applicability of CPP-PNA to understand and elucidate metabolic pathways. Therefore, additional targeted experiments are still required to justify and confirm our findings.

Shown in this study, we demonstrated that CPP-PNA has the potential to serve as a simple and rapid method for evaluating metabolic pathways within Synechocystis sp. PCC 6803. We also showed that dose-dependent evaluation can be performed without permanent genetic alterations. Despite these promising results, several limitations and challenges remain for broader CPP-PNA implementation. As demonstrated, this approach is only effective for gene knockdown experiments and does not address other gene regulation phenomena, such as gene upregulation or activation, that may also affect metabolic fluxes. For gene upregulation or activation in bacteria using stable knockout strains, chromosomal integration is typically required but this cannot be achieved due to the structural nature of PNAs having a peptide backbone. PNA’s limitations in gene upregulation applications might be addressed by developing CPP-DNA delivery systems that could enable homologous recombination [46], but this represents an area requiring future research and optimization.

PNA length and specificity have also been subjects of debate, as shorter PNA sequences are known to contribute to off-target effects. However, recent reports have shown that although longer PNAs exhibit higher binding affinity and reduced off-target effects, their in vivo efficacy was significantly lower than shorter PNAs [49, 50]. This reduction is primarily attributed to the cellular uptake limitations of CPPs harboring longer PNAs [51]. In bacteria, the optimal trade-off between PNA length and CPP permeation efficiency occurs with 10–12 nts [52]. Therefore, in this work, where efficient PNA delivery is required for dose-dependent metabolic pathway evaluation, 10 nts were employed. Additionally, in the context of translation inhibition, PNAs are most effective when bound to the start codon region of the target mRNA [53]. Consequently, to avoid possible off-target effects, all our conjugates in this work were checked to ensure that they bind exclusively to the start codon region of target mRNAs by mapping against the genome of strain 6803 (Table S1). Nevertheless, sequence similarities among PNAs remain possible when design is restricted to start codon regions. Thus, further evaluation of new CPPs to accommodate efficient delivery of longer PNAs could be necessary.

Even with close precaution in the design of our CPP-PNAs, we did however observe slight decolorization in all cultures treated with CPP-PNA, indicating the possibility of an off-target effect. Although it was difficult for us to fully clarify the reason of this off-target effect, we performed a background experiment to prove that the off-target effect did not affect the results of this work. We synthesized a CPP-PNA that binds in the middle of the Ddh mRNA (CPP-Syn6803ddh_Asp130) and upon treatment of strain 6803 with this conjugate at 48 h, although we did see decolorization of the culture (Figure S3), we did not see an accumulation of D-lactate (Figure S4). This indicated that D-lactate accumulation was due to the inhibition of the CPP-Syn6803ddh conjugate. Furthermore, to ensure that the decolorization did not affect the metabolites within our samples, all metabolites measured were individually normalized with the OD₇₃₀ values of the culture prior to comparison. As shown here, off-target effects non-related to the specificity of the PNA sequence may also occur and precautions need to be made prior to use CPP-PNA in metabolic pathway evaluation. In addition, with the current design of our CPP-PNAs, we are currently limited to the application of CPP-PNA to evaluate specific metabolic pathways rather than using it as a tool for global analysis on the effects of enzyme inhibition. Other limitations include the strain-specific nature of CPP permeation, which requires individual optimization for each target organism, the need for compatible commercial antibodies that limit the availability of validation methods, and the potential for CPP-PNA effects to complicate long-term metabolic studies.

In conclusion, based on the results attained in this work, in its current application, the approach can serve as a primary screening tool for identification of a target metabolic pathway prior to establishing stable knockout strains. We however believe further improvements and optimization to CPP-PNAs can overcome the limitations and challenges, allowing the platform to offer an advancement for bacterial genetic engineering, particularly in contexts where conventional genetic methods encounter substantial constraints, and provides new opportunities to investigate the biochemical mechanisms of metabolite-producing bacteria.

Author contributions

YK performed most of the experiments and analyses. SK assisted in establishing the knockout strains and extraction of metabolites. JY synthesized the CPP-PNA conjugates. GK and MK assisted in setting up the facilities for cyanobacteria cultivation and optimized the protocols for western blotting. RS and SI assisted in the optimization and setup of the LC-MS for metabolite measurements. KS, TM, RA and TM (Corresponding author; TM-CA) conceptualized and designed the study. YK and TM-CA wrote the manuscript. All authors have read and approved the final version of the manuscript.

Funding

This research was supported by the Grant-in-Aid for Scientific Research (A) (JSPS KAKENHI Grant number JP24H00519), Grant-in-Aid for Challenging Exploratory Research (JSPS KAKENHI Grant number JP21K19137) and by The Japan Science and Technology Agency (Fusion Oriented Research for disruptive Science and Technology program) all awarded to TM-CA.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.