Strains and plasmids used in this paper are listed in Table 1. Plasmid construction in this work follows the overlap extension PCR cloning procedure as described previously (Nelson and Fitch, 2011). E. coli strain Turbo (Tsingke Biotech) was used for competent cell preparation. The wspR, mrkA promoter, and mrkH sequences were synthesized by Tsingke Biotech, while the dgcT, dgcI, pdeH, and dgcF sequences were amplified from the genome of the E. coli MG1655 strain and the proC promoter and mScarlet I sequences were amplified from pEB2-mScarlet I-mVenus^NB-MrkH vector (Addgene No. 182291).

The mrkH gene was inserted into the medium-copy plasmid pBAD33, generating the plasmid pBAD33-MrkH. Subsequently, the pdeH or wspR D70E gene, along with an additional ribosome binding site, was inserted downstream of the mrkH gene, resulting in the plasmids pBAD33-MrkH-PdeH and pBAD33-MrkH-WspR D70E, respectively. Plasmid pET28b(+)-PmrkA-mScarlet I was constructed by replacing the T7 promoter of the high-copy plasmid pET28b(+)-MCS with mrkA promoter and mScarlet I-encoding gene. For the plasmids pBAD33-PdeH-proC-MrkH and pBAD33-WspR D70E-proC-MrkH, which express PdeH or WspR D70E under the control of the araBAD promoter and constitutively express MrkH via the proC promoter respectively, were constructed by first inserting the PdeH or WspR D70E gene downstream of the araBAD promoter, followed by the insertion of the proC promoter, an additional ribosome binding site, and mrkH gene downstream of the pdeH or wspR D70E gene.

Quick change mutagenesis method was employed for introducing point mutations of the wspR, dgcT, dgcI, and dgcF genes (Liu and Naismith, 2008). All the oligonucleotides, promoters, and genes used in this work are listed in Supplementary Tables S1, S2.

Overnight cultures of the strains harboring cdiGEBS and either PdeH or WspR D70E were diluted 1:100 into 4 mL of fresh LB medium supplemented with 100 μg mL⁻¹ kanamycin, 34 μg mL⁻¹ chloramphenicol, and 0.002% L-arabinose. Cells were then cultivated at 30°C for 6 h with a shaking speed of 220 rpm. Subsequently, 0.1 OD_{600 nm} cells were collected and resuspended in 40 µL phosphate-buffered saline (PBS). A 5 µL aliquot was placed on a microscope slide, coated with 1% agarose, and covered with a coverslip. Fluorescence images were captured using a Zeiss LSM 880 Confocal Laser Scanning Microscope, with the excitation wavelength set to 568 nm.

Overnight cultures of E. coli co-expressing cdiGEBS and PdeH, or cdiGEBS and each WspR mutant, or cdiGEBS and each putative DGC were inoculated at a 1:100 dilution-fold into 4 mL fresh LB medium supplemented with 100 μg mL⁻¹ kanamycin, 34 μg mL⁻¹ chloramphenicol, and 0.002% L-arabinose. After grown at 30°C for 6 h, 1 OD_{600 nm} cells were collected and resuspended in 220 µL PBS for fluorescence measurements by an automated microplate reader (SynergyHTX hybrid reader, BioTek) with the excitation and the emission set at 540 ± 35 nm and 600 ± 40 nm, respectively.

Overnight cultures of E. coli co-expressing cdiGEBS and various CSPs were inoculated at a 1:100 dilution-fold into 4 mL fresh LB medium supplemented with 100 μg mL⁻¹ kanamycin, 34 μg mL⁻¹ chloramphenicol, 100 μg mL⁻¹ streptomycin, 100 μg mL⁻¹ spectinomycin, and 0.002% L-arabinose. Cells were then cultured at 30°C, 220 rpm until the OD_{600 nm} reached 0.3, followed by adding 0.1 mM isopropyl-β-D-thiogalactoside (IPTG) to induce the expression of 6 × His-MBP-CSP. The 6 × His-MBP tag was fused with various CSP to prevent cellular degradation of CSP (Hee et al., 2020). Following 6 h of 6 × His-MBP-CSP expression, cells equivalent to 4 OD_{600 nm} units were harvested and resuspended in 2 mL of PBS buffer for fluorescence intensity and spectral measurements using a fluorescence spectrometer (FS5 Spectrofluorometer, Edinburgh Instruments), with the excitation wavelength set at 568 nm and an emission range of 575–650 nm.

Overnight cultures of E. coli harboring cdiGEBS were diluted 1:100 into fresh LB medium supplemented with 100 μg mL⁻¹ kanamycin, 34 μg mL⁻¹ chloramphenicol, and 0.002% L-arabinose. Then 200 μL cell cultures were transferred to each well of a 96-well plate containing different concentrations of tyrosol or ampicillin. Fluorescence intensity and cell density (OD_{600 nm}) were monitored by using a Synergy HTX Multi-Mode Microplate Reader with a 15-min detection period for 21 h at 30°C. For fluorescence intensity measurement, the excitation and emission wavelength were set at 540 ± 35 nm and 600 ± 40 nm, respectively.

Cells were cultivated and induced for protein expression as described in the “Fluorescence Measurement by a Microplate Reader” Section. 1 OD_{600 nm} cells were collected and resuspended in PBS to a final OD_{600 nm} of 5.0. Then cell samples were mixed with 5 × protein loading buffer, and subjected to 12% SDS-PAGE after boiling for 10 min. For Western blot analysis, proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Merck) using a wet transfer apparatus (Tanon). Then the PVDF membranes were blocked with 5% non-fat milk in TBST (Tris buffer saline containing 0.1% Tween-20), followed by incubation with the primary antibodies: anti-6 × His from mouse or anti-trigger factor from rabbit (1:4,000, GenScript, Cat. No. A01329) in TBST containing 5% non-fat milk. Afterward, the membranes were incubated with fluorescently labeled IRDye 800 CW secondary antibodies (1:10,000, LI-COR Biosciences), and fluorescence images were recorded using the Odyssey Sa Infrared Imaging System (LI-COR Biosciences).

The transcriptional regulator MrkH is a member of the PliZ family and exhibits a high affinity for c-di-GMP in the nanomolar range (Schumacher and Zeng, 2016; Wang F. et al., 2016). Upon c-di-GMP binding, MrkH undergoes a conformational change-induced activation (Figure 1A) (Schumacher and Zeng, 2016), allowing it to bind the mrkA promoter and highly upregulate the transcription of the downstream mrkABCDF operon (Wilksch et al., 2011). Based on this, we introduced the c-di-GMP sensing element MrkH under the regulation of the araBAD promoter and the fluorescent reporter element mScarlet I under the control of the mrkA promoter in E. coli to establish a genetically encoded fluorescent biosensor for c-di-GMP detection. Specifically, the mrkH gene was placed downstream of the araBAD promoter in the medium-copy number plasmid pBAD33 vector, while the mScarlet I gene was inserted downstream of the mrkA promoter (PmrkA) on the high-copy plasmid pET28b(+) backbone with the original T7 promoter removed (Figure 1B).

We observed a significant fluorescence increase when the strain containing the biosensor was induced with arabinose for MrkH expression, suggesting that the endogenous c-di-GMP in E. coli can substantially activate the transcriptional activity of MrkH and thus upregulate mScarlet I expression (Figure 1C). To further confirm that the observed increase in mScarlet I fluorescence was due to the transcriptional activation of MrkH upon binding c-di-GMP, we introduced the R113A mutation in MrkH to substantially diminish its c-di-GMP-binding ability (Vrabioiu and Berg, 2022). We found that the resultant strain expressing the MrkH R113A mutant exhibited a significantly decreased fluorescence intensity compared to the strain expressing the wild-type MrkH (Figure 1C), indicating that the fluorescence increase in the strain expressing wild-type MrkH was indeed mediated by the binding between MrkH and c-di-GMP.

To assess whether cdiGEBS can respond to varying intracellular levels of c-di-GMP, we co-expressed the biosensor with either a constitutively activated diguanylate cyclase WspR D70E or the phosphodiesterase PdeH to induce high or low c-di-GMP concentrations, respectively (Dippel et al., 2018). By optimizing the arabinose concentration, cultivation temperature, and induction time, we found that the biosensor could effectively distinguish between high and low c-di-GMP levels, exhibiting a remarkable fluorescence change readout of over 23-fold (Supplementary Figure S1 and Figure 1C). These findings were further supported by fluorescence imaging (Figure 1D), and Western blot analysis confirmed that the observed differences in fluorescence intensity were not due to variations in MrkH expression levels (Supplementary Figure S2). Taken together, these results indicate that cdiGEBS can respond to high and low c-di-GMP concentrations with highly dynamic fluorescence fold changes.

To further investigate the sensitivity of cdiGEBS in distinguishing varying cellular c-di-GMP levels, we maintained the mScarlet I reporter element and modified the sensing element as follows: 1) MrkH was constitutively expressed under the control of the proC promoter, and 2) WspR D70E or PdeH expression was induced by the araBAD promoter (Figure 2A). By titrating the resulting strains with different concentrations of L-arabinose, we were able to gradually adjust the levels of the c-di-GMP synthase WspR D70E or the degradation enzyme PdeH, leading to progressively increased or decreased cellular c-di-GMP concentrations, respectively.

We observed that the strain co-expressing MrkH and WspR D70E exhibited a gradual increase in mScarlet I fluorescence with increasing concentrations of L-arabinose (Figure 2B). In contrast, the strain co-expressing MrkH and PdeH showed a dose-dependent decrease in mScarlet I fluorescence intensity with L-arabinose (Figure 2C). Furthermore, by titrating the strain containing only the biosensor, without WspR D70E or PdeH, with different concentrations of L-arabinose, we excluded the possibility that these fluorescence changes were due to the effects of L-arabinose on the mScarlet I fluorophore or MrkH transcriptional activity (Supplementary Figure S3). Overall, these results demonstrate that cdiGEBS is sensitive and capable of distinguishing subtle variations in intracellular c-di-GMP levels, which result from finely tuned adjustments in the expression of c-di-GMP synthesis or degradation enzymes.

The capability of cdiGEBS to distinguish varying c-di-GMP levels promoted us to explore its application in assessing the activity of diguanylate cyclases (DGCs). There are 29 genes in Escherichia coli K-12 participating in c-di-GMP biosynthesis or degradation (Hengge et al., 2016; Povolotsky and Hengge, 2016). Previous studies have shown that overexpression of DgcT resulted in a significant increase of intracellular c-di-GMP, implying that DgcT processed diguanylate cyclase activity (Jonas et al., 2008). Additionally, the uncharacterized membrane-associated proteins DgcI and DgcF have also been annotated as putative DGCs in E. coli since strains lacking DgcF or DgcI (ΔdgcF or ΔdgcI) display increased swimming motility (Sanchez-Torres et al., 2011; Povolotsky and Hengge, 2016). DGCs generally function as homodimers, with their GG(D/E)EF domains being responsible for the production of c-di-GMP (Dahlstrom and O’Toole, 2017). To gain structural insights into these putative DGCs, we predicted the potential dimeric structures of these proteins using AlphaFold 3 and identified the presence of typical GG(D/E)EF domains in all of them (Figure 3A) (Abramson et al., 2024).

We then investigated the in vivo diguanylate cyclase activities of these putative DGCs using cdiGEBS, with the constitutively active DGC WspR D70E serving as the positive control. Strains co-expressing the biosensor and any of the putative DGCs showed a substantial increase in mScarlet I fluorescence intensity, comparable to the strains co-expressing the biosensor and WspR D70E (Figure 3B). This result indicates that these three putative DGCs can indeed catalyze the synthesis of c-di-GMP in vivo. Moreover, to investigate the effect of the GG(D/E)EF domain of DgcT, DgcF, and DgcI on their DGC activity, we substituted the third residue of the GG(D/E)EF domain with alanine. We found that the resulting strains expressing the protein mutants (DgcT E361A, DgcI D371A, and DgcF E224A) showed significantly decreased fluorescence intensities compared to the strains expressing the wild-type proteins (Figure 3B). This suggests that DgcT, DgcI, and DgcF all catalyze the synthesis of c-di-GMP in a GG(D/E)EF domain-dependent manner.

The sensitive performance of our biosensor in tracking DGC activities drove us to further investigate whether it could discriminate the subtle differences in the enzymatic activities of various DGC mutants. To examine this, we characterized the activities of the canonical DGC WspR and its mutants with defined DGC activities in vivo using cdiGEBS (Figures 3C, D) (Huangyutitham et al., 2013). The D70A substitution disrupts the phosphorylation site, resulting in the loss of cyclase activity (De et al., 2009). Similarly, the E253A mutation inactivates WspR by impairing the active site (Malone et al., 2007). In contrast, the D70E mutation mimics the phosphorylated state, leading to constitutive activation (Wang X. C. et al., 2016). The V72D mutant also exhibits constitutive diguanylate cyclase activity, as identified in a random mutagenesis study of WspR (Huangyutitham et al., 2013). Additionally, studies have demonstrated that the tetrameric form of WspR displays high activity levels, with the WspR L170D variant enhancing activity by stabilizing this tetrameric form (De et al., 2008).

We found that strains co-expressing the biosensor and the wild-type WspR showed a significant increase in fluorescence intensity compared with the strain containing only the biosensor. In contrast, strains co-expressing the biosensor and the inactive mutants WspR D70A or WspR E253A exhibited low fluorescence intensities similar to the strain harboring the biosensor only, implying that these mutations nearly completely abolished DGC activities (Figure 3D). Consistent with the literature, we observed that strains co-expressing the biosensor and activity-enhanced mutants D70E, V72D, and L170D all showed a trend of increased fluorescence intensity compared to the strain co-expressing the biosensor and the wild-type WspR, although the increase in fluorescence intensity of the strain expressing the mutant L170D was not statistically significant (Figure 3D) (Huangyutitham et al., 2013). Additionally, Western blot analysis suggested that the observed fluorescence differences were not due to variations in the expression levels of MrkH and WspR mutants (Supplementary Figure S4). Taken together, these results suggested that cdiGEBS was sensitive and could be used for functional analysis and activity evaluation of DGCs in vivo.

High c-di-GMP levels could stimulate biofilm formation, leading to a 10 to 1,000-fold increase in antibiotic resistance of bacteria in the sessile state than in the planktonic state (Dostert et al., 2019; Ciofu and Tolker-Nielsen, 2019; Dostert et al., 2021). Decreasing the cellular c-di-GMP levels is a promising strategy for combating biofilm-related antibiotic resistance and potential chronic infection. Therefore, researchers have developed a number of structurally diverse compounds to diminish c-di-GMP concentrations in bacteria, either by directly sequestering c-di-GMP or by enhancing PDE activity (Qvortrup et al., 2019; Hee et al., 2020). Encouraged by the sensitivity of cdiGEBS in monitoring c-di-GMP levels in vivo, we next sought to examine whether this sensor was capable of reflecting the effects of typical anti-biofilm chemicals on the cellular c-di-GMP level.

Hee et al. designed a serial of short c-di-GMP-sequestering peptides (CSP) which could directly bind c-di-GMP and thus reduce the concentration of free cellular c-di-GMP in Pseudomonas aeruginosa and consequently inhibit biofilm formation (Hee et al., 2020). We found that the strain co-expressing the biosensor and the peptide CSP2 which bound c-di-GMP tightest exhibited a substantial decrease in mScarlet I fluorescence intensity compared with the strain harboring the biosensor only (Figures 4A–C). In contrast, co-expressing the biosensor and the binding affinity attenuated peptides CSP3 or CSP2 R169A only led to a moderate fluorescence decrease of the resulting strains (Figures 4B, C).

Furthermore, Choi et al. demonstrated that tyrosol could decrease c-di-GMP levels by activating PDE activity (Choi and Kim, 2024). We observed that treating the strain containing cdiGEBS with increasing concentrations of tyrosol resulted in decreased mScarlet I fluorescence intensity compared with the control strain which was treated with 1% DMSO (Figure 4D). In contrast to biofilm dispersal chemicals, studies have shown that subinhibitory concentrations of ampicillin can facilitate c-di-GMP production (Boehm et al., 2009; Ho et al., 2013; Gao et al., 2022). We also found that cdiGEBS could be used for monitoring the increasing effect of ampicillin on the cellular c-di-GMP levels (Figure 4F). Additionally, neither over-expressing the CSPs nor tyrosol treatment substantially hinders bacterial growth (Supplementary Figure S5). For ampicillin treatment, although bacterial growth was slightly reduced at 1.0 μg mL⁻¹, the unaltered growth at 0.1–0.5 μg mL⁻¹ confirms that the observed increase in fluorescence is genuine and not a result of signal normalization (Supplementary Figure S5). Furthermore, neither tyrosol nor ampicillin treatment affected the fluorescence of the strain constitutively expressing mScarlet I under the control of the J23100 promoter, indicating that these chemicals do not interfere with the mScarlet I fluorophore (Supplementary Figure S6). Taken together, these results demonstrate that cdiGEBS is capable of reflecting the effects of typical chemicals on cellular c-di-GMP levels and even has the potential to be used for validating the binding affinities between various chemicals and c-di-GMP in vivo.

Recent studies have emphasized the strategy of developing anti-biofilm compounds by reducing cellular c-di-GMP levels for the treatment of biofilm-associated diseases (Martin et al., 2021; Hultqvist et al., 2024). The high sensitivity of our sensor suggests strong potential for its use as a high-throughput screening platform to identify novel agents that reduce intracellular c-di-GMP concentrations. Future efforts focusing on screening compounds that inhibit DGC activity, activate PDE activity, or sequester free c-di-GMP by combining tools of our sensor and fluorescence-activated cell sorting (FACS) could offer new insights for combating biofilm-associated diseases.