HomeScienceEngineering probiotics to inhibit Clostridioides difficile infection by dynamic regulation of intestinal...

Engineering probiotics to inhibit Clostridioides difficile infection by dynamic regulation of intestinal metabolism

[ad_1]

Hydrolase-mediated bile salt deconjugation inhibits C. difficile germination and growth

To validate our hypothesis that the modulation of bile salt profiles controls C. difficile germination, we assessed the effect of taurocholate and its deconjugated form, cholate, on the germination and growth of C. difficile (Supplementary Fig. 1A). While the physiological level of bile salts varies along the intestinal tract, we have taken 2 mM of bile salts as an average representative concentration to use for in vitro assays9,20. Figure 2A, B show that cholate significantly decreased the germination of the endospores by up to 81% and the number of viable vegetative cells of C. difficile strains by up to 82%. We then chose the bile salt hydrolase Cbh of Clostridium perfringens as an enzyme to be introduced into probiotics to deconjugate taurocholate into cholate. We used this enzyme because Cbh uses taurocholate as a preferred substrate and remains more functional at physiological pH than other bile salt hydrolases21. Figure 2C shows that recombinant Cbh (Supplementary Fig. 1B, C) converted 99.2% of the taurocholate into cholate. Cbh also deconjugated glycocholate, another conjugated primary bile salt present in the gut, with similar efficiency. Glycocholate is also reported to be a germinant of C. difficile, and the deconjugating activity of Cbh against both taurocholate and glycocholate achieved >99% reduction of conjugated bile salts (Fig. 2D). Figure 2E, F show that the group with recombinant Cbh exhibited a 96% lower endospore germination rate and 89% fewer viable vegetative cells than the control group with taurocholate and no Cbh. These results suggested that recombinantly expressed Cbh deconjugated taurocholate into cholate and that this enzymatic deconjugation significantly inhibited the germination of endospores and the growth of vegetative cells of C. difficile.

Fig. 2: The bile salt hydrolase Cbh inhibits C. difficile via deconjugation of taurocholate.
figure 2

A Germination of endospores from C. difficile strains ((i) CD630, (ii)VPI10463, (iii) BAA1870, and (iv) 9689), under different concentrations of bile salt (taurocholate or cholate). The experiments were set with the same initial endospore number and indicated the amount of germination as a percentage of the maximal germinated CFUs. n = 2 independent experiments. B Growth of vegetative cells from C. difficile with 2 mM bile salt (taurocholate or cholate) or without bile salt (control) in BHIS media. n = 3 independent experiments. C Quantification of bile salts following incubation with Cbh. The concentrations of taurocholate, glycocholate and cholate were determined through LC/MS. Unpaired two-sided Student’s t-tests were performed to compare the bile salt concentrations between the Cbh-treated and control groups (p < 0.0001). ****P < 0.0001. n = 3 independent experiments. D Quantification of bile salts (5 mM taurocholate and 5 mM glycocholate) following incubation with Cbh. The concentrations of taurocholate, glycocholate and cholate were determined through LC/MS. n = 3 independent experiments. E Germination of purified C. difficile endospores and F growth of vegetative C. difficile following treatment with Cbh for 1 or 12 h, respectively. The reactions were set up as follows: the control group contained blank buffer, the Cbh-only group contained purified Cbh-his6 without bile salt (final concentration of 2 mM), the taurocholate-only group contained taurocholate without enzymes and the Cbh + taurocholate group contained purified Cbh and taurocholate. Relative endospore germination was calculated by normalising to taurocholate-only group. Unpaired two-sided Student’s t-test was performed to compare the Cbh + taurocholate group and the taurocholate-only control group (p = 0.0085). *P < 0.05. n = 3 independent experiments. All data were presented as mean values with error bars representing SEMs of triplicates unless stated otherwise. Source data are provided as a Source Data file.

Engineering of dysbiosis-sensing circuits in probiotics

We chose E. coli Nissle 1917 as the probiotic host because it is an extensively studied probiotic with a long safety record in humans22,23. Furthermore, as a gram-negative bacterium, E. coli Nissle 1917 can be employed in combination with the current CDI therapy, which entails the use of antibiotics that target gram-positive bacteria. We used the previously reported auxotrophic E. coli Nissle 1917 strain as the base strain, which requires exogenous d-alanine for survival (denoted by EcN), because it allows antibiotic-free selection, enhances plasmid stability, and provides a means for biocontainment22,24.

Based on increasing evidence that microbiome dysbiosis preludes the onset of CDI7,8,12,13,19,25, we hypothesised that a genetic sensor that regulates the expression of Cbh in response to microbiome dysbiosis might increase the efficacy of our engineered probiotics against CDI. We chose sialic acid as a proxy signal for dysbiosis because the level of sialic acid is elevated in the intraluminal space upon antibiotic treatment, which has been postulated to be a result of a dysbiosis-mediated imbalance between sialic acid-catabolising and/or sialidase-expressing members of the microbiome19. Sialic acid also supports the pathogenesis of gastrointestinal infections, including CDI, possibly because pathogens can use sialic acid as a carbon source and decorate the cellular surface with sialic acid to evade host responses19,26,27,28.

We selected the promoter pNanA as a core sensor element because it has been postulated to be responsive to sialic acid29. We characterised pNanA with the transcriptional regulator NanR (Supplementary Fig. 2A). Supplementary Fig. 2B shows that pNanA was responsive to sialic acid when NanR was expressed, while pNanA alone resulted in a high background with a low response. Then, to improve the dynamic range of the pNanA-NanR sensor (termed ‘Sensor’), we modulated the expression level of NanR using a set of constitutive promoters and ribosome binding sites. Fig. 3A shows that the combination of J23113 (promoter) and rbs4 (RBS) led to the highest dynamic range in response to sialic acid. We also observed that the pNanA-NanR sensor dynamically regulated the expression of GFP (reporter protein), where the removal of sialic acid brought the expression back to the basal level (Supplementary Fig. 2C, D). Notably, the pNanA sequence contains a catabolite activator protein (CAP) binding site. In line with this feature, Fig. 3B shows that glucose significantly reduced the GFP expression in response to sialic acid. The level of glucose is significantly lower in the ileum, caecum and colon30, where C. difficile primarily colonises13. Therefore, we hypothesised that the pNanA-NanR sensor can provide another layer of expression control such that the expression of Cbh increases when the engineered probiotics reach the target sites: the ileum, caecum and colon.

Fig. 3: Development and characterisation of a sialic acid-responsive biosensor.
figure 3

A Optimisation of pNanA induction via modulation of the nanR co-expression level with different combinations of constitutive promoters (Pcon) and ribosome binding sites (rbs). GFP was expressed under the control of pNanA as a reporter gene in all constructs. Induction was performed with sialic acid. The relative GFP expression after 3 h of induction is shown. Signal-to-noise represents the ratio between the induced to the uninduced GFP expression. B Relative GFP fluorescence of EcN expressing the selected pCon-rbs combination (J23113-rbs4-nanR-pNanA-gfp) after 3 h of incubation with varying concentrations of glucose with or without 0.2% sialic acid. C Germination efficiency of purified C. difficile endospores in taurocholate and/or sialic acid treated with EcN expressing Cbh from a selected sialic acid sensor construct (J23113-nanR-pNanA-cbh; EcN-pNanA-Cbh). No-EcN reaction and circuit control (EcN-circuit ctrl) were used as controls. The circuit control expressed GFP in lieu of Cbh under the same construct design. The germinated endospore CFU values were normalised to the no-EcN control values. D Quantification of bile salts following incubation with EcN-Cbh from a selected sialic acid sensor construct (J23113-nanR-pNanA-cbh; EcN-pNanA-Cbh). No-Cbh reaction and purified Cbh reaction were used as controls. The concentrations of taurocholate and cholate were determined through HPLC. E Circuit design of the final sensor-amplifier construct (J23113-nanR-pNanA-cadC-pCadBA-gfp) consisting of sensor, amplifier, and actuator modules. F Relative GFP fluorescence of EcN expressing the sensor-only construct (J23113-nanR-pNanA-gfp) or the sensor-amplifier construct (J23113-nanR-pNanA-cadC-pCadBA-gfp). The relative GFP expression after 3 and 6 h of induction is shown. G Immunoblot of Cbh-his6 expressed from different constructs in EcN. The pBad construct (araC-pBad-cbh) was induced with l-arabinose for 16 h, and the sensor-only construct (J23113r4-nanR-pNanA-cbh) and sensor-amplifier construct (J23113r4-nanR-pNanA-cadC-pCadBA-cbh) were induced with sialic acid for 16 h. S represents the soluble fraction, and IS represents the insoluble fraction. The expected size of Cbh-his6 is 38 kDa. The band on the ladder corresponds to 35 kDa. All data were presented as mean values with error bars representing SEMs of triplicates (n = 3 independent experiments). Source data are provided as a Source Data file.

Engineering of sensor-amplifier-actuator circuits in probiotics

Next, we examined whether the expression of Cbh (termed ‘Actuator’), when induced by the sialic acid sensor, could inhibit the germination of C. difficile endospores. Figure 3C shows a reduction in endospore germination by up to 47%, which indicated a significantly lower efficacy than that of purified recombinant Cbh (Fig. 2D). We also observed low conversion of taurocholate into cholate (Fig. 3D). Therefore, we hypothesised that the low germination inhibition might be due to insufficient expression of Cbh. To test this hypothesis, we added an amplifier module to the sensor-actuator circuit. Specifically, the transcriptional activator gene cadC was placed under the control of the promoter pNanA so that CadC (termed ‘Amplifier’) could activate the promoter pCadBA to amplify Cbh expression. The amplifier module was evaluated for sialic acid-responsivity (Supplementary Fig. 2E). CadC regulates the cad operon and has been shown to be pH sensitive31, providing an additional layer of expression control32. A module with constitutive GFP expression was also evaluated for comparison (Supplementary Fig. 2F). The final sensor-amplifier-actuator circuit, where Cbh served as an actuator (Fig. 3E), resulted in significantly increased actuator (Cbh) expression of the circuit (Fig. 3F). The expression level of Cbh under sialic acid induction was comparable to that of purified recombinant Cbh (Fig. 3G), which led to significant germination and growth inhibition (Fig. 2D, E). These results suggest that the EcN harbouring the aforementioned sensor-amplifier-actuator circuit (denoted by EcN-Cbh), which comprises pNanA, NanR, CadC, pCadBA and Cbh, might significantly inhibit the germination and growth of C. difficile.

The engineered probiotics inhibit C. difficile germination and growth

We then examined the extent to which EcN-Cbh could deconjugate taurocholate into cholate and reduce the germination of C. difficile endospores. In response to sialic acid, EcN-Cbh fully converted taurocholate into cholate (Fig. 4A) and caused a 98% reduction in endospore germination (Fig. 4B). Notably, the absence of sialic acid led to less but still considerable deconjugation (45%) and germination reduction (90%), likely due to the basal expression of Cbh in EcN-Cbh. Basal expression of Cbh did not significantly alter the growth of the host strains (Supplementary Fig. 3A). In addition, no extracellular deconjugation activity of EcN-Cbh was observed (Supplementary Fig. 4), suggesting that the deconjugation action of EcN-Cbh remained intracellular, although Cbh is reportedly secreted in gram-positive C. perfringens33.

Fig. 4: The engineered probiotics inhibit C. difficile endospore germination and growth and reduce C. difficile toxin secretion.
figure 4

A Quantification of taurocholate and cholate after 3 h incubation of taurocholate and/or sialic acid with EcN or EcN expressing the sensor-amplifier-cbh construct (EcN-Cbh). Conversion efficiency (cholate/taurocholate) is shown. No-EcN reaction and circuit control (EcN-circuit ctrl) reaction were used as controls. The circuit control expressed GFP in lieu of Cbh under the same construct design. B Germination efficiency of purified C. difficile endospores in taurocholate (tau) and/or sialic acid (SA) treated with EcN-Cbh. The germinated endospore CFU values were normalised to a no-EcN control group. Unpaired two-sided Student’s t-test was performed to compare the EcN-Cbh and EcN-circuit ctrl groups. *P < 0.05. C CFU values from coculture assays between vegetative C. difficile cell cultures and wild-type EcN (EcN WT), EcN-circuit ctrl, or EcN-Cbh cultures preincubated for 1 h with taurocholate. n = 2 independent experiments. D Immunoblot of the C. difficile toxin TcdA from concentrated supernatants at different time points of germinating C. difficile culture. Reactions with or without taurocholate were treated with EcN-Cbh and then incubated with C. difficile endospores. No-EcN reaction and EcN-circuit ctrl reaction were used as controls. EcN was induced with taurocholate and/or sialic acid. The blot was probed with an anti-tcdA antibody. L represents the protein ladder. The expected size of TcdA is 308 kDa. E Relative cell viability of Caco-2 cells treated with supernatants collected from germinating C. difficile cultures under the conditions outlined for Fig. 4D (i) no-EcN control, (ii) EcN-circuit ctrl, (iii) EcN-Cbh with taurocholate, and (iv) EcN-Cbh without taurocholate. The supernatants were diluted to final concentrations of 1X (filled), 0.1X (grey), and 0.01X (empty). The relative cell viability was determined by MTT assay and normalised to that of the untreated Caco-2 control group. n = 2 independent experiments. The numerical relative cell viability data are shown in a colour-graded matrix. Bottom: the coloured bar corresponds to the relative Caco-2 viability. All data were presented as mean values with error bars representing SEMs of triplicates (n = 3 independent experiments), unless stated otherwise. Source data are provided as a Source Data file.

Next, we determined whether EcN-Cbh could inhibit the growth of the vegetative cells of C. difficile. Figure 4C shows that EcN-Cbh strongly reduced the number of viable vegetative cells. To further investigate the mechanism of this reduction, we evaluated the numbers of viable vegetative C. difficile cells after culture with EcN-Cbh, taurocholate and cholate. Supplementary Fig. 3B, C indicate that the growth inhibition was due to cholate. Figure 4C shows that 1-hour preincubation with taurocholate was sufficient to significantly inhibit the vegetative growth of C. difficile cultured with EcN-Cbh. Together, these results suggest that EcN-Cbh significantly inhibits the germination of endospores and vegetative cells of C. difficile.

The engineered probiotics reduce C. difficile toxicity

To investigate whether and the extent to which germination and growth inhibition by EcN-Cbh could lead to a reduction in the toxicity of C. difficile, we first assessed the amount of the C. difficile exotoxin TcdA when C. difficile was cocultured with EcN-Cbh. Figure 4D shows that EcN-Cbh significantly decreased the level of TcdA in the culture, which likely resulted from lowered germination and growth of C. difficile. Then, we evaluated the viability of human epithelial colorectal adenocarcinoma cells (Caco-2 cells) upon exposure to the supernatant of C. difficile. EcN-Cbh significantly improved the viability of Caco-2 cells exposed to C. difficile (Fig. 4E). These results together suggest that pre-treatment with EcN-Cbh reduced C. difficile toxicity.

The engineered probiotics inhibit CDI in a murine model

The aforementioned in vitro results prompted us to hypothesise that EcN-Cbh might inhibit CDI in vivo. To test this hypothesis, we evaluated whether and the extent to which EcN-Cbh could reduce mortality and morbidity in a murine model of CDI that was previously established34,35 (Fig. 5A). The model mice were given engineered probiotics prior to being exposed to a virulent strain of C. difficile, VPI10463, that has been reported to induce significant mortality and symptomatic displays34,36. The evaluation was conducted using a treatment group (EcN-Cbh) and five control groups: (i) a no-sensor control (EcN-Cbh-S) group, (ii) a no-amplifier control (EcN-Cbh-A) group, (iii) a no-actuator control (EcN-Cbh) group, (iv) a wild-type control (EcN-WT) group and (v) an infection control (CD) group (Fig. 5B). Figure 5C shows the extent of deconjugation activities of the probiotics. Supplementary Figs. 5, 6 and 8 show further characteristics of these probiotics, such as growth, expression, antimicrobial sensitivity and intestinal viability. We assessed the clinical symptoms, mortality and weight of the CDI model mice as previously described35,37,38. Challenge with C. difficile (107 CFU) was performed on day 0. The mortality, weight, and clinical symptoms of the mice were then monitored over the course of 9 days. The clinical symptoms were scored according to a previously established standard38.

Fig. 5: The engineered probiotics improve infection prognosis and outcomes of CDI in a murine model.
figure 5

A Animal experiment timeline. B Table of the groups used in this study and the constructs expressed in the probiotics. An additional group for infection control in which the animals were given sucrose instead of probiotic was also used. The growth and expression characteristics of the probiotics are presented in Supplementary Fig. 5. C Quantification of taurocholate and cholate after incubation of the probiotics with taurocholate and sialic acid. A control reaction without EcN (“Control”) was set up. ND not detected. The error bars represent the SEMs of results from two independent experiments. D Survival curves of the groups following C. difficile challenge. The log-rank Mantel–Cox test was performed to compare the treatment and control groups (p value against no-sensor control group <0.0001; no-amplifier control group = 0.0058; no-actuator control group = 0.0001; wild-type control group = 0.0002; infection control group = 0.0153). *P < 0.05; ***P < 0.001. E Relative mean weight of animals following C. difficile challenge. Animals were removed from the calculation on the day following their death. F The Clinical Sickness Score (CSS) of each animal was recorded as the highest score attained by the individual animal within 6 days postinfection. Daily CSSs are shown in Supplementary Fig. 7. The infection severity based on the score is indicated by dotted lines. Dunnett’s multiple comparisons test was performed to compare the treatment and control groups (p value against no-sensor control group = 0.0004; no-amplifier control group = 0.0015; no-actuator control group <0.0001; wild-type control group <0.0001; infection control group = 0.0010). **P < 0.01. The number of independent samples for each group used are: treatment group, n = 12; no-sensor control group, n = 7; no-amplifier control group, n = 10; no-actuator control group, n = 10; wild-type control group, n = 5; and infection control group, n = 15. Source data are provided as a Source Data file.

Figure 5 D shows that EcN-Cbh led to a 100% survival rate in the model mice, while all the controls resulted in significantly lower survival rates ranging from 60 to 14.3%. Figure 5E indicates that the infection model mice fed EcN-Cbh exhibited the least weight loss, especially between day 2 and day 4, when the model mice showed the most severe symptoms. Figure 5F shows that the group of model mice fed EcN-Cbh had the lowest clinical symptom score (CSS), 4.42. CSSs were used to indicate the severity of the infection and ranged from normal (0 to 2) to mild (3 to 5), moderate (6 to 8) or severe (9 to 12) based on a stool, behaviour and weight loss38. All the controls exhibited significantly higher CSSs than the EcN-Cbh-fed model mice, with mean scores ranging from 7.73 to 10.00. Finally, we performed 16 S rRNA metagenomic sequencing on a microbiome library extracted from the faeces of the animals in all the groups. Figure 6A shows that only EcN-Cbh led to a decrease in the abundance of C. difficile in the model mice. Figure 6B shows that among the groups that expressed Cbh, EcN-Cbh and the no-amplifier control, both of which harboured the sensor, improved the diversity of the microbiome of the model mice, as indicated by a significant increase in the Shannon diversity index. These results together suggest that EcN-Cbh significantly reduced CDI in the model mice, as evidenced by a 100% survival rate, improved clinical symptoms and a decrease in the abundance of C. difficile.

Fig. 6: The engineered probiotics improve infection prognosis and outcomes of CDI in a murine model.
figure 6

A Relative abundance of C. difficile from metagenomic sequencing of 16 S rRNA from a genomic library of microbiomes extracted from faecal samples of all treatment and control groups. Faecal samples from day 1 and day 3 postinfection were sequenced. Unpaired two-sided Student t-test was performed to compare relative abundance of C. difficile between day 1 and day 3 (p value for treatment group = 0.05, no-sensor control group = 0.2979; no-amplifier control group = 0.0412; no-actuator control group <0.0001; wild-type control group = 0.0002; infection control group = 0.0014). *P < 0.05; ***P < 0.001. Results of two independent samples for each group are used for each respective day, with three technical replicates. B Shannon index (alpha diversity indicator). Unpaired Student t-test was performed to compare the Shannon index between day 1 and day 3. *P < 0.05; **P < 0.01; ***P < 0.005. Results of two independent samples for each group are used for each respective day, with three technical replicates. C Histological Injury Scores (HISs) of colon tissues were assessed in a blind fashion. The infection severity based on the score is indicated by dotted lines. Mann–Whitney test was performed (outlier (*) was excluded for analysis). *P < 0.05 (p value for no-infection control and treatment = 0.036; no-infection and wild-type = 0.024; infection and treatment = 0.036). The numbers of independent samples for each group are no-infection control group, n = 3; for all other groups, n = 6. D Representative microscopic images of H&E-stained colon tissue from each group were assessed. Submucosal oedema (*) and neutrophil infiltration (arrow) were identified. The scale bar corresponds to 100 µm. Source data are provided as a Source Data file, and the black bars indicate the means and SEMs of the groups.

The engineered probiotics ameliorate histopathological injury in model mice

To examine the extent of the injury caused by CDI in the model mice, we conducted a histopathological assessment of colon tissues harvested from the model mice on days 3 and 4 following death or euthanasia. Figure 6C shows the histologic injury scores (HISs) of the harvested colon tissue. The HISs were used to indicate the severity of the infection and ranged from normal (0 to 1) to mild (2 to 3), moderate (4 to 6) or severe (7 to 9) based on epithelial tissue damage, mucosal oedema and neutrophil infiltration38 (Fig. 6D). Model mice treated with EcN-Cbh displayed lower HISs than the control mice. These results suggest that EcN-Cbh treatment markedly alleviates tissue damage due to CDI in model mice.

The engineered probiotics modulate bile salt composition

Next, to determine whether and how much the sialic acid level was elevated and to assess whether EcN-Cbh modulated bile salt profiles in vivo as hypothesised, we analysed the faeces of infection model mice as described in Fig. 7A. Figure 7B shows that the level of sialic acid was increased 6-fold upon antibiotic treatment from day -6 to -3, when EcN-Cbh was introduced, supporting our use of a sialic acid biosensor in the probiotics. Figure 7C shows that EcN-Cbh decreased the taurocholate level and increased the cholate level in the infection model mice from day -3 to day 0 before C. difficile was introduced. Figure 7D shows that the EcN-Cbh and the no-sensor control resulted in significantly reduced taurocholate and correspondingly elevated cholate levels. We also quantified other bile salts, glycocholate, chenodeoxycholate, lithocholate, and deoxycholate, in the faeces of the model mice upon administration of the engineered probiotics (Supplementary Fig. 9). Supplementary Fig. 9 shows that EcN-Cbh and the no-sensor control led to an increased level of deoxycholate, a derivative of cholate, while the Cbh-expressing probiotics (i.e. EcN-Cbh, the no-sensor control, and the no-amplifier control) resulted in an increased level of chenodeoxycholate and its derivative lithocholate. These results suggest that EcN-Cbh increases cholate levels and decreases taurocholate levels in model mice, as hypothesised.

Fig. 7: The engineered probiotics modify bile salt composition in the host gut.
figure 7

A Timeline of sample collection for metabolite examination in relation to the animal experiment timeline. B Quantification of sialic acid in faecal samples during antibiotic-containing water treatment. Paired Student’s t-test was performed to compare the day −6 and day -3 data (p = 0.0027). **P < 0.01. The black bars indicate the means and SEMs of the groups. n = 3 independent samples. C Quantification of (i) taurocholate and (ii) cholate in faecal samples prior to C. difficile challenge from day −6 to 0 in the EcN-Cbh and infection control groups (“No probiotic”). ND not detected. Unpaired Student’s t-test was performed between the EcN-Cbh and infection control groups for taurocholate on day −1 (p = 0.0507) and cholate on day −3 (p = 0.0056). **P < 0.01. The black bars indicate the means and SEMs of the groups. n = 3 independent samples. D Quantification of (i) taurocholate and (ii) cholate in faecal samples collected from day −1 to day 3 from each group. Mixed-model ANOVA was performed to compare groups (p value for taurocholate for no-sensor control group = 0.1362; no-amplifier control group = 0.1117; no-actuator control group = 0.0152; wild-type control group <0.0001; infection control group = 0.0051; for cholate for no-sensor control group = 0.8709; no-amplifier control group = 0.2829; no-actuator control group = 0.0002; wild-type control group = 0.0001; infection control group <0.0001). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. The black bars indicate the means and SEMs of the groups and days. LOD limit of detection. Source data are provided as a Source Data file.


[ad_2]

Source link

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments

%d bloggers like this: