Bioengineering of Escherichia coli Nissle 1917 for Production and Excretion of Spermidine, a Key Metabolite in Human Health

Microbiota-derived metabolites have biological importance for their host. Spermidine, a metabolite described for its protective effect in age-related diseases, is now studied for its role in the resolution of inflammation and gut homeostasis. Strategies to modulate its production in the gastrointestinal tract are of interest to increase host spermidine intakes. Here, we show that metabolic engineering can be used to increase spermidine production by the probiotic Escherichia coli Nissle 1917 (EcN), used in humans. First, we found that increasing the expression of genes involved in polyamine biosynthesis, namely the S-adenosylmethionine synthase speD and the spermidine synthase speE, resulted in an increase in spermidine produced and excreted by our engineered bacteria. The major drawback of this first attempt was the production of acetylated forms of spermidine. Next, we propose to solve this problem by increasing the expression of the spermidine exporter system MdtI/MdtJ. This second strategy had a major impact on the spermidine profile found in the culture supernatant. Our results demonstrate, for the first time, the feasibility of rationally engineering bacterial probiotic strains to increase their ability to deliver the microbiota-derived metabolite, spermidine. This work illustrates how metabolomic and synthetic biology can be used to design and improve engineered Live Biotherapeutic Products that have the potential to be used in personalized medicine.


Introduction
The human intestine is colonized by trillions of microorganisms which participate in the regulation of the host's physiology. There is increasing evidence of microbiota's involvement in the regulation of human physiology and in the development of human pathologies, such as diseases involving inflammatory processes. In the gastrointestinal tract, the microbiota are involved in a constant dialogue with the host immune system. Notably, metabolites derived from the microbiota are essential to this dialogue, as they are involved in immune regulation [1]. Their role in the maintenance of the intestinal barrier is now well described, showing the importance of preserving the microbiota and its functionality. An imbalance within this ecosystem, described as dysbiosis, leads to a loss of microbial diversity and is accompanied by a depletion of major metabolites that play a role in maintaining intestinal and immune homeostasis.
Manipulation of the intestinal microbiome is proposed as a potential strategy to prevent or treat a wide range of human diseases [2]. Current strategies rely primarily on the use of probiotics, prebiotics, fecal microbiota transplantation, symbiotic microbial consortia or microbiota-derived proteins and metabolites. Some probiotic strains are now examined Table 1. Primers used in this study. speE (spermidine synthase) and speD (S-adenosylmethionine synthase). mdtJ and mdtI genes code for the spermidine export proteins MdtIJ.

Name
Primer Sequence (5 -3 ) Each plasmid was sent for sequencing with the oSeq1 and oSeq2 primers for Sanger sequencing (Eurofin Genomics, France).

Sample Collection and Sample Preparation
During the culture, 400 µL was collected at the beginning and 24 h after the start of the culture. Each bacterial suspension was then centrifuged at 10,000× g for 10 min at 4 • C. Afterwards, the supernatant was collected and stored at −80 • C while waiting to follow the sample preparation protocol.
For sample preparation, we used a protein precipitation strategy previously described [20]. Briefly, each supernatant was treated with methanol (80% methanol, 20% Sample) and mixed using a vortex mixer for 20 s, left on ice at 4 • C for 30 min to allow protein precipitation, then centrifuged for 20 min at 13,000× g. After centrifugation, the supernatant was collected and subjected to desiccation before being stored at −80 • C for analysis.

Liquid Chromatography Coupled with High-Resolution Mass Spectrometry (LC-HRMS) Analysis
Untargeted metabolomic profiling of the culture supernatant was conducted using ultra-high-performance liquid chromatography (Vanquish Flex, Thermo Fisher Scientific, Waltham, MA, USA) coupled with a Q Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA).
For the acquisition presented in the result part 3.1, we used the following protocol: • The mobile phases consisted of A (100% water + 0.1% formic acid) and B (100% acetonitrile (ACN) + 0.1% formic acid). Elution started with an isocratic step of 3 min at 1% mobile phase B, followed by a linear gradient from 1% to 100% mobile phase B for the next 4 min. These proportions were kept constant for the next 3 min before returning to 1% B for 4 min.
For the acquisition presented in the result part 3.2, we used the following protocol: • The mobile phases consisted of A (100% water + 0.1% formic acid) and B (100% acetonitrile (ACN) + 0.1% formic acid). Elution started with an isocratic step of 2 min at 1% mobile phase B, followed by a linear gradient from 1% to 100% mobile phase B for the next 12 min. These proportions were kept constant for the next 4 min before returning to 1% B for 5 min.
The mass spectrometer was fitted with an electrospray source (ESI) operating in positive modes. It was operated with a capillary voltage at 5 kV in the positive ionization and a capillary temperature set at 280 • C. The temperature of the autosampler compartment was set at 4 • C, and the injection volume was 5 µL. Detection was carried out from m/z 60 to 900. During the acquisition, we used a quality-control (QC) pooled samples strategy to ensure that measurements and peak detection were stable. This QC sample was injected every 10 injections. Features with a coefficient of variation (CV) higher than 30% were removed from the analysis.

LC-HRMS Data Processing
Raw LC-HRMS data were converted to m/z extensible markup language (.mzXML) in centroid mode using MSConvert ProteoWizard (release version 3.0.20350-23fa10407). Preprocessing of the data was performed on MzMine 2.53 [21]. Normalization and statistical analyses were performed on MetaboAnalyst 5.0 [22]. Mass detection was realized keeping the noise level at 5.0 × 10 6 for MS level 1 and 1.0 × 10 6 for MS level 2. An ADAP chromatogram builder was employed, using a minimum group size of scans of 4, a group intensity threshold of 3.0 × 10 6 , and an m/z tolerance of 0.001. An ADAP wavelet deconvolution algorithm was used with the following standard settings: S/N threshold of 10, coefficient/area threshold of 20, peak duration range of 0.00-4 min and RT wavelet range of 0.01-0.10. MS2 scans were paired using an m/z tolerance range of 0.01 Da and a RT tolerance range of 0.2 min. Isotopes were grouped using the isotopic peak grouper algorithm (m/z tolerance of 5 ppm, RT tolerance of 0.2 min, maximum charge of 3). For the peak alignment, this step was performed using the Join aligner module (m/z tolerance of 5 ppm and weight of 75 and 25 were attributed for m/z and RT, respectively; RT tolerance of 0.8 min).

Metabolite Annotation
Metabolites identification of putrescine and spermidine was performed by injecting authentic chemical standards (Sigma, St Louis, MO, USA) under the same analytical conditions. In the absence of an available authentic chemical standard, metabolites of interest were only considered as putatively annotated based on accurately measured masses and interpretation of the MS/MS spectra when available, as described by Aros-Calt et al., 2015 [23]. First, feature annotation was performed by using fragmentation spectra in the spectral library of the Global Natural Product Social Molecular Networking (GNPS) platform [24]. The Library Search feature of GNPS allowed us to identify N8acetylspermidine (GNPS Library Spectrum CCMSLIB00006122363). In a second attempt to perform metabolite identification, SIRIUS and CSI: FingerID [25] were used. SIRIUS identifies metabolites based on their MS/MS spectra and by creating fragmentation trees that can explain the fragmentation spectrum. Then, CSI:FingerID uses the fragmentation tree to predict the compound's molecular fingerprint. Using this second approach, we obtained a similarity score of 100 for N8-acetylspermidine and 92.1 for diacetylspermidine (see Supplementary Figures S1 and S2).

Statistical Analysis
Results in figures are represented as mean per group ± standard deviation (SD). The statistical test used for each hypothesis is described in the figure legend. The individual data were obtained from eight independent cultures. All statistical analyses were performed on GraphPad Prism ver.9.0 for Windows (GraphPad Software, San Diego, CA, USA) and MetaboAnalyst 5.0. Results were considered significant as follows: * p < 0.05.

Overexpression of the Genes speD and speE Increases the Amount of Spermidine
In E. coli, the flux of PAs depends on import, export, synthesis and catabolism ( Figure 1A). Putrescine synthesis relies on ornithine or arginine decarboxylation performed, respectively, by ornithine decarboxylase (SpeC) or arginine decarboxylase (SpeA) and agmatinase (SpeB) [26,27]. Spermidine synthesis begins with the synthesis of decarboxylated S-adenosylmethionine, which depends on the conversion of methionine to S-adenosylmethionine (SAM) and its decarboxylation by the enzyme SAM decarboxylase (SpeD) [28]. Once putrescine is produced, it is used to produce spermidine with the spermidine synthase SpeE, where decarboxylated SAM serves as a substrate in this reaction [29].
To increase the synthesis of Spermidine in EcN, we first decided to increase the amount of the enzymes SpeD and SpeE that catalyze the last steps of the synthesis of spermidine. For that purpose, the genes speD and speE were assembled into the plasmid pUCBB-ntH6-eGFP, an optimized BioBrick™ vector, derived from the pUC backbone [18]. We constructed pCC25, a plasmid where speD and speE were placed under the control of the P lacP' promoter, and pCC26, a plasmid where speD and speE transcription was regulated by the P fnrs promoter and where an optimized RBS was used. P lacP' is a constitutive promoter derived from the lac promoter, and P fnrs is an anaerobic inducible promoter regulated by the fumarate and nitrate reductase (FNR) transcription factor [30]. The P fnrs promoter was chosen because it has already shown its interest in anaerobic environments, such as the gut [3]. After designing these plasmids, we cultured the EcN, EcN pCC25 and EcN pCC26 strains in an LB medium and monitored their growth. We did not see any difference in the growth curves between the three groups ( Figure 2A). We then used RT-qPCR to relatively quantify the gene expression of speD and speE, and we observed an overexpression of speE and speD genes (Supplementary Figure S3). Then, to look further for the effect of these overexpressions, we collected culture supernatants at times 0 and 24 h from 8 independent cultures and analyzed them by LC-MS/MS (see the Methods section and reference papers [12,31]). As shown in Figure 2B, EcN, transformed with the plasmid pCC25 and pCC26, have increased spermidine concentrations in their supernatants. In our experiment, the putrescine concentration increased in all conditions between 0 h and 24 h. However, the amount of putrescine present at the end of the cultures was lower if the strains were transformed either with pCC25 or pCC26 ( Figure 2C). The putrescine and spermidine concentration profiles between wild type EcN and engineered EcN suggests that EcN are releasing putrescine into the extracellular medium. The overproduction of SpeD/E tends to limit the amount of putrescine present at the end of the culture, indicating that there is a higher consumption of putrescine. However, the rise in spermidine was also accompanied by a significant increase in acetylated forms of spermidine, as shown by the study of the mono-acetylated and diacetylated forms ( Figure 2D,E). pCC25 and pCC26, have increased spermidine concentrations in their supernatants. In our experiment, the putrescine concentration increased in all conditions between 0 h and 24 h. However, the amount of putrescine present at the end of the cultures was lower if the strains were transformed either with pCC25 or pCC26 ( Figure 2C). The putrescine and spermidine concentration profiles between wild type EcN and engineered EcN suggests that EcN are releasing putrescine into the extracellular medium. The overproduction of SpeD/E tends to limit the amount of putrescine present at the end of the culture, indicating that there is a higher consumption of putrescine. However, the rise in spermidine was also accompanied by a significant increase in acetylated forms of spermidine, as shown by the study of the mono-acetylated and diacetylated forms ( Figure 2D,E). List of transporters and enzymes: MdtI/J, spermidine export proteins; PotABCD, ATP-binding spermidine-preferential uptake system; PotFGHI, ATP-binding putrescine-specific uptake system; PotE, putrescine/ornithine antiporter; SapABCDF, Putrescine export system permease; SpeA, biosynthetic arginine decarboxylase; SpeB, agmatinase; SpeC/SpeF, biosynthetic/degradative ornithine decarboxylase; SpeE, spermidine synthase; SpeG, spermidine acetyltransferase (B): General outline of constructions presented in this study. Created with BioRender.com. List of transporters and enzymes: MdtI/J, spermidine export proteins; PotABCD, ATP-binding spermidine-preferential uptake system; PotFGHI, ATP-binding putrescine-specific uptake system; PotE, putrescine/ornithine antiporter; SapABCDF, Putrescine export system permease; SpeA, biosynthetic arginine decarboxylase; SpeB, agmatinase; SpeC/SpeF, biosynthetic/degradative ornithine decarboxylase; SpeE, spermidine synthase; SpeG, spermidine acetyltransferase (B) General outline of constructions presented in this study. Created with BioRender.com.

Putrescine Supplementation of the Medium Increases Spermidine Production
With our previous experiment, we identified that putrescine is more consumed by our engineered EcN strains. We, therefore, asked whether the substrate, putrescine, was limiting for spermidine biosynthesis, and we then cultured EcN, EcN pCC25 and EcN pCC26 with LB media supplemented with 1 mM putrescine. This concentration is a typical concentration of putrescine in the intestinal lumen of healthy humans [32]. As shown in Figure 3B,C, we reconfirmed our observations, showing that EcNs transformed with plasmid pCC25 or pCC26 have increased concentrations of spermidine and decreased concentration of putrescine in their supernatants. Interestingly, putrescine supplementation results in the increase in spermidine compared with the unsupplemented medium in EcN pCC25 and EcN pCC26 ( Figure 3D). However, the addition of putrescine did not result in an increase in the production of spermidine by the wild type EcN. In conclusion, the metabolic engineering of the strain (EcN pCC25 and EcN pCC26) combined with the supplementation of LB medium maximizes the amount of spermidine produced.

Putrescine Supplementation of the Medium Increases Spermidine Production
With our previous experiment, we identified that putrescine is more consumed by our engineered EcN strains. We, therefore, asked whether the substrate, putrescine, was limiting for spermidine biosynthesis, and we then cultured EcN, EcN pCC25 and EcN pCC26 with LB media supplemented with 1 mM putrescine. This concentration is a typical concentration of putrescine in the intestinal lumen of healthy humans [32]. As shown in Figure 3B,C, we reconfirmed our observations, showing that EcNs transformed with plasmid pCC25 or pCC26 have increased concentrations of spermidine and decreased concentration of putrescine in their supernatants. Interestingly, putrescine supplementation results in the increase in spermidine compared with the unsupplemented medium in EcN pCC25 and EcN pCC26 ( Figure 3D). However, the addition of putrescine did not result in an increase in the production of spermidine by the wild type EcN. In conclusion, the metabolic engineering of the strain (EcN pCC25 and EcN pCC26) combined with the supplementation of LB medium maximizes the amount of spermidine produced.

Excretion and Production of Acetylated Forms of Spermidine Can Be Limited by Overexpressing the Spermidine Exporter MdtI/MdtJ
We demonstrated that increasing the concentration of SpeD and SpeE is an effective strategy to increase the amount of spermidine available in the extracellular compartment. However, as we have shown previously, this accompanies an increase in the production and excretion of acetylated forms of spermidine. This increase makes sense considering the biology of PAs in E. coli and the mechanisms by which this species regulates intracellular polyamine levels. As it was described before, spermidine accumulation is toxic for E. coli [33]. Two strategies are used by these bacteria to limit spermidine toxicity, namely, spermidine overexcretion using specific transporters [34] or spermidine acetylation using the spermidine N(1)-acetyltransferase SpeG. Nevertheless, since our goal is to maximize the amount of spermidine in the extracellular compartment, it may be interesting to promote spermidine excretion to limit the amount of acetylated spermidine. To address this issue, we have added a second biological circuit on pCC25 and pCC26 to increase the production of the two-component exporter system MdtI/MdtJ. This new circuit was composed of the constitutive promoter P J23102 , an optimized RBS and the coding sequence of MdtI and MdtJ. The two new plasmids were designed and assembled and named plasmid pCC31 (P fnrs -speD-speE-B0015-P J23102 -mdtJ-mdtI) and plasmid pCC35 (P LacP' -speD-speE-B0015-P J23102 -mdtJ-mdtI). Each plasmid was transformed into EcN, and we used the LB media supplemented with putrescine and the same methodology to assess whether the addition of the MdtIJ transporters improved spermidine production and excretion. Metabolic profiling of the different strains showed that the addition of the MdtI/MdtJ biological system induced a major change in the metabolite levels of the PAs pathway.
Indeed, spermidine is significantly increased in EcN pCC31 and EcN pCC35 compared with EcN (fold change: 3 and 5, respectively) or EcN transformed with pCC25 or pCC26 (fold change: 2.8 between EcN pCC35 and EcN pCC25) ( Figure 4D). Furthermore, the accumulation of acetylated forms of spermidine observed with plasmids pCC25 and pCC26 disappeared in strains transformed with plasmids pCC31 and pCC35 ( Figure 4E,F). These observations suggest that these engineering strategies profoundly impacted the metabolites fluxes belonging to the PAs family, in favor of spermidine production and excretion.

Excretion and Production of Acetylated Forms of Spermidine Can Be Limited by Overexpressing the Spermidine Exporter MdtI/MdtJ
We demonstrated that increasing the concentration of SpeD and SpeE is an effective strategy to increase the amount of spermidine available in the extracellular compartment. However, as we have shown previously, this accompanies an increase in the production and excretion of acetylated forms of spermidine. This increase makes sense considering the biology of PAs in E. coli and the mechanisms by which this species regulates intracellular polyamine levels. As it was described before, spermidine accumulation is toxic for E. coli [33]. Two strategies are used by these bacteria to limit spermidine toxicity, namely, spermidine overexcretion using specific transporters [34] or spermidine acetylation using the spermidine N(1)-acetyltransferase SpeG. Nevertheless, since our goal is to maximize the amount of spermidine in the extracellular compartment, it may be interesting to promote spermidine excretion to limit the amount of acetylated spermidine. To address this issue, we have added a second biological circuit on pCC25 and pCC26 to increase the production of the two-component exporter system MdtI/MdtJ. This new circuit was composed of the constitutive promoter PJ23102, an optimized RBS and the coding sequence of   midine is significantly increased in EcN pCC31 and EcN pCC35 compared with EcN (fold change: 3 and 5, respectively) or EcN transformed with pCC25 or pCC26 (fold change: 2.8 between EcN pCC35 and EcN pCC25) ( Figure 4D). Furthermore, the accumulation of acetylated forms of spermidine observed with plasmids pCC25 and pCC26 disappeared in strains transformed with plasmids pCC31 and pCC35 ( Figure 4E,F). These observations suggest that these engineering strategies profoundly impacted the metabolites fluxes belonging to the PAs family, in favor of spermidine production and excretion.  The individual data were obtained from independent cultures. All data shown represent the mean ± SD. Statistical significance was calculated using Kruskal-Wallis test and Benjamini, Krieger and Yekutieli Multiple comparisons. Statistical significance indicated as * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

EcN Engineering Seems to Have a Limited Impact on Its Overall Metabolism
We used an untargeted LC-MS/MS metabolomics approach to pinpoint the main differences in the metabolome of the engineered EcN and the wild type. In all, 178 features were detected and met our quality requirements presented in the method section, and volcano plot analysis was used to compare EcN pCC25 and EcN pCC35 with EcN ( Figure 5). Analysis of the EcN pCC25 and pCC35 strains showed that the majority of features were unchanged at 24 h if the strains' supernatants were respectively compared to the EcN strain ( Figure 5A,B). This first observation highlights the fact that this metabolic engineering did not affect the global metabolism of the EcN pCC25 and pCC35. However if we look in detail at the effect of the addition of pCC25 and pCC35 to the wild type strain, we can see that the abundance of five features was decreased (putrescine, methionine and three features related to N-Acetyl-isoleucine) and seven increased (spermidine, diacetylspermidine, acetyl-spermidine and four features unidentified) in EcN pCC25 supernatant compared to EcN ( Figure 5A), while in pCC35, the abundance of eight features (arginine, acetyl-putrescine, methionine, alanyl-valine and, four features were unidentified) was decreased and seven increased (spermidine, N-Acetyl-spermidine, and five features were unidentified) in its supernatant compared with EcN ( Figure 5B). In conclusion, the main metabolic changes observed by this unsupervised method demonstrate that the changes brought about by both types of engineering (pCC25 and pCC35) are related principally to the polyamines pathway.
were detected and met our quality requirements presented in the method section, and volcano plot analysis was used to compare EcN pCC25 and EcN pCC35 with EcN ( Figure  5). Analysis of the EcN pCC25 and pCC35 strains showed that the majority of features were unchanged at 24 h if the strains' supernatants were respectively compared to the EcN strain ( Figure 5A,B). This first observation highlights the fact that this metabolic engineering did not affect the global metabolism of the EcN pCC25 and pCC35. However if we look in detail at the effect of the addition of pCC25 and pCC35 to the wild type strain, we can see that the abundance of five features was decreased (putrescine, methionine and three features related to N-Acetyl-isoleucine) and seven increased (spermidine, diacetylspermidine, acetyl-spermidine and four features unidentified) in EcN pCC25 supernatant compared to EcN ( Figure 5A), while in pCC35, the abundance of eight features (arginine, acetyl-putrescine, methionine, alanyl-valine and, four features were unidentified) was decreased and seven increased (spermidine, N-Acetyl-spermidine, and five features were unidentified) in its supernatant compared with EcN ( Figure 5B). In conclusion, the main metabolic changes observed by this unsupervised method demonstrate that the changes brought about by both types of engineering (pCC25 and pCC35) are related principally to the polyamines pathway. Cutoff for the p-value is <0.05; fold change cutoff is >1.5 or <0.66. All p-values were adjusted using the False Discovery Rate method. All points were annotated on the graph when feature identification was possible using GNPS or SIRIUS.

Discussion
The gut microbiota produce a wide variety of metabolites that can be absorbed directly from the colon or reach systemic circulation. Some of these metabolites appear to play a key role in host health and disease. Accumulating evidence shows that bioactive molecules, such as PAs, act directly on host cells to perform physiological functions [35]. Research on spermidine has mainly focused on its anti-aging, neurologic and cardiovascular benefits, but convincing evidence has highlighted its impact on the epithelium of the intestinal mucosa and intestinal immunity [15,17]. In humans, spermidine intake is mainly dependent on diet and microbiota. Two strategies were then proposed: the first one consists of proposing supplementation in spermidine or a diet rich in spermidine while the Cutoff for the p-value is <0.05; fold change cutoff is >1.5 or <0.66. All p-values were adjusted using the False Discovery Rate method. All points were annotated on the graph when feature identification was possible using GNPS or SIRIUS.

Discussion
The gut microbiota produce a wide variety of metabolites that can be absorbed directly from the colon or reach systemic circulation. Some of these metabolites appear to play a key role in host health and disease. Accumulating evidence shows that bioactive molecules, such as PAs, act directly on host cells to perform physiological functions [35]. Research on spermidine has mainly focused on its anti-aging, neurologic and cardiovascular benefits, but convincing evidence has highlighted its impact on the epithelium of the intestinal mucosa and intestinal immunity [15,17]. In humans, spermidine intake is mainly dependent on diet and microbiota. Two strategies were then proposed: the first one consists of proposing supplementation in spermidine or a diet rich in spermidine while the other one proposes to modulate intakes of spermidine in the colon by modulating the gut microbiota. Both strategies have shown that increased spermidine intake results in an anti-inflammatory response and improved gut homeostasis.
We hypothesize that modifying the spermidine supply with genetically engineered bacterial strains is an attractive strategy. In this study, we used the probiotic EcN, recognized for its anti-inflammatory properties, to perform metabolic engineering that will increase its spermidine production. In a first attempt, we increased the expression of speD and speE and were able to show that this overexpression already resulted in a considerable increase in the amount of spermidine produced and transported to the extracellular compartment. One of the main problems encountered with this strategy was the accumulation of acetylated forms of spermidine, which was, therefore, an important issue to maximize the amount of spermidine produced and bioavailable in the exometabolome. Based on accumulated knowledge of Pas' metabolic pathways in E. coli, we hypothesized that this accumulation was related to a normal response of the bacterium to limit PA toxicity. We have abolished this effect by increasing the expression of the two-component exporter system MdtI/MdtJ. This strategy significantly increased the amount of spermidine present in the extracellular space while limiting the acetylation of spermidine by EcN. We have also showed that spermidine production was influenced by the availability of putrescine in the environment of the bacteria. Finally, we evaluated the impact of our engineering on the metabolomes of the strains. The data generated showed that the metabolic engineering we performed has a limited impact on the overall metabolome of the strains and that this impact seems to target the metabolites related to the PA pathway. Although it provides valuable insights to answer these questions, additional experiments are needed to maximize the metabolic coverage being detected.
To our knowledge, this study provides the first evidence that spermidine can be overproduced in a probiotic bacterial strain. This strategy will allow researchers to address complex questions regarding the impact of targeted variations in a metabolic pathway on the anti-inflammatory effects of EcN and, more importantly, the impact of these engineered strains on the microbial ecosystem of the gut. As PA are bioactive molecules that are important in maintaining human health, this engineered probiotic will be useful to explore more deeply the importance of the PA on the resolution of pathological inflammatory processes, as is the case in chronic inflammatory bowel diseases.

Patents
We have filed a patent application with the registration number FR2209825.
Author Contributions: C.C., A.L.G. and B.T. conceived the study. C.C. performed the experiments with support from A.L.G. and B.T., C.C. performed data processing and analysis with support from A.L.G. In particular, C.C., C.P., V.C. and A.L.G. contributed to metabolomics analysis experiments and interpretation. C.C. wrote the first original draft of the manuscript, which was reviewed and validated by all authors. The funding acquisition was achieved thanks to A.L.G. and B.T. All authors have read and agreed to the published version of the manuscript.