Carbon monoxide-producing probiotic Escherichia coli attenuates HSV-1 skin disease Authors: Omar P Vallejos, Guillermo Hoppe-Elsholz, Pablo Reyes-Bustos, Eduardo I Tognarelli, Yaneisi Vázquez, Alejandro Piña-Iturbe, Patricia Pereira-Sánchez, Lucas Vásquez-Herrera, Javiera P Bahamondes-Abrigo, Valentina A Vergara-Espinoza, Francisco B Quero, Sofía Campos-Gajardo, Javiera Sepúlveda-Alfaro, José Acevedo-López, Isidora D Suazo, Pedro H Silva, Esteban Caamaño, Valentina Scaff, Franco Bassi, Irenice Coronado-Arrázola, Felipe Melo-González, Alejandra Mateluna, Manuel Álvarez-Lobos, Aniela Wozniak, Alexis M Kalergis, Pablo A González, Susan M Bueno Published: 2026 ABSTRACT SummaryPersistent infections and inflammatory diseases remain major health challenges due to their prevalence and limited prophylactic and therapeutic options. Heme oxygenases (HOs) degrade heme into bioactive molecules, including carbon monoxide (CO), a traditionally toxic gas yet naturally produced in mammals to regulate inflammation. Here, we investigated whether bacterial CO production via bacterial heme oxygenase ChuS, present in commensal Escherichia coli, could confer prophylactic effects against HSV-1 skin infection. Oral administration of E. coli Nissle 1917 (EcN) increased serum carboxyhemoglobin in a ChuS-dependent manner. Moreover, isolated commensal E. coli strains carrying chuS, but lacking the genotoxin colibactin, enhanced CO levels and reduced HSV-1 severity. Combined administration of these strains elicited a trend toward reduced disease severity and a significant reduction in viral load compared to controls. These findings identify ChuS as a microbial driver of CO-mediated anti-inflammation and highlight probiotic E. coli as a safe, natural platform for CO delivery.Subject areas: molecular biology, microbiology Introduction Inflammation is an evolutionarily conserved process in which immune and non-immune cells protect the host from damaging agents and promote tissue repair. This process involves the recruitment of leukocytes to sites of infection and/or damage through signaling cascades of proinflammatory cytokines, including tumor necrosis factor-alpha (TNF-α), interleukin-1β (IL-1β), and interleukin-6 (IL-6).1 Although inflammation is essential for host defense, prolonged, uncontrolled, or exacerbated inflammatory responses can cause additional tissue damage, contributing to the development of enhanced diseases and chronic pathologies.2,3 Carbon monoxide (CO) is a diatomic, colorless, and odorless gas that can bind tightly to heme-containing proteins. CO is primarily known for its toxicity at elevated concentrations due to its high affinity to hemoglobin and has also gained attention for its anti-inflammatory properties when present or administered at low levels to cells and tissues.4 Endogenously, CO is produced as a byproduct of heme oxygenase (HO) activity, which catalyzes heme degradation into CO, free iron (Fe2+), and biliverdin, which is rapidly converted to bilirubin.5 Among the existing HO isoforms in mammalian cells, inducible HO-1 and constitutive HO-2 play crucial roles in oxidative stress responses and cellular homeostasis, with HO-1 expression being upregulated in response to chemical and physical stress.6 CO modulates immune responses by suppressing proinflammatory cytokines such as TNF-α, reducing inflammation locally and systemically,7,8 and increasing anti-inflammatory cytokines such as IL-10, and modulates different inflammation-related processes.9,10,11,12 Studies have highlighted that HO-1 plays essential roles during viral infections, and its induction may have antiviral effects. This has been reported for numerous viruses,13,14,15 most recently in the context of herpes simplex virus (HSV) infection of immune and non-immune cells.16,17 Pharmacological induction of HO-1 using cobalt protoporphyrin (CoPP) reduced viral propagation in epithelial cells and mitigated cytopathic effects, an outcome partially attributed to CO-mediated interference with viral replication cycles.16 In dendritic cells (DCs) infected with HSV-1, pharmacological induction of HO-1 with CoPP restored cellular functions related to migration and the activation of virus-specific T cells, thereby reducing herpetic skin disease.17 Despite these findings, the translational application of CO-based strategies remains limited due to challenges associated with safe and controlled delivery of CO, thus motivating the exploration of alternative platforms, such as organisms that produce or stimulate endogenous CO generation. Beyond mammalian systems, structurally simpler bacterial HOs also catalyze heme degradation.18,19 Several bacterial HOs have been described to date, mainly in bacterial pathogens, including HemO from Neisseria meningitidis,20 PhuS from Pseudomonas aeruginosa,21 HugZ from Helicobacter pylori,22 MhuD from Mycobacterium tuberculosis,23 IsdG and IsdI from Staphylococcus aureus,24 and ChuS from Escherichia coli.25 These enzymes often facilitate iron acquisition, a critical cofactor for bacterial survival, as 67% of mammalian Fe2+ is bound to heme.26,27,28 Although these systems are primarily found in pathogenic bacteria, ChuS has also been identified in E. coli Nissle 1917 (EcN), the primary component of the widely used probiotic Mutaflor (Ardeypharm, Germany). This bacterium possesses a heme utilization system encoded in the chu operon, which includes ChuS.29 EcN exhibits immunomodulatory properties that are associated with its ability to colonize the intestinal mucosa and establish sustained host-microbe interactions.30,31 EcN has been shown to modulate T cell activation and expansion through Toll-like receptor 2 (TLR-2)-dependent signaling,32 and to induce both pro- and anti-inflammatory cytokine responses, including IL-10, in immune cells.33 In addition, EcN limits intestinal colonization by enteric pathogens such as Salmonella enterica serovar Typhimurium through competition for essential nutrients, including iron.34 Preventive administration of EcN attenuates intestinal inflammation and barrier dysfunction in colitis murine models,35 supporting its relevance in both inflammatory and infectious contexts. Recent studies have demonstrated that microbial-based platforms capable of locally producing bioactive gases can effectively modulate inflammation, supporting the feasibility of gas-mediated immunoregulation through biologically compatible delivery systems.36,37 In contrast to complex engineered approaches, probiotic strategies offer a simpler and physiologically relevant means to achieve localized gas production in vivo. Together, these observations raise the possibility that bacterial HO-expressing probiotics may extend the immunomodulatory benefits of CO beyond inflammatory disorders to viral infections in which immune cells contribute to tissue pathology and disease resolution.38 Preventive measures for HSV-1 infections include antiviral prophylaxis with drugs such as acyclovir or valacyclovir, particularly in individuals with recurrent outbreaks or immunosuppression.39 However, conventional treatments often show limited efficacy and fail to address the underlying mechanisms of the disease, highlighting the urgent need for innovative and complementary prophylactic and therapeutic strategies.38 Notably, few strategies have explored microbiota-based or immunomodulatory approaches as complementary interventions against HSV-1 skin disease. In this context, CO-producing probiotics represent a promising alternative strategy. By harnessing the anti-inflammatory and immunomodulatory properties of endogenously produced CO derived from bacterial HO, such as ChuS, these formulations may help restore immune balance and mitigate disease severity in both inflammatory and infectious settings. Given the immunomodulatory properties of CO and the contribution of inflammation to HSV-1 skin pathology, we investigated whether ChuS-mediated CO production by commensal Escherichia coli is associated with modulation of disease outcomes in an HSV-1 skin infection model. Specifically, we evaluate the contribution of ChuS to bacterial CO production in vivo and its relationship with disease severity and viral load under the experimental conditions studied. Results A complete chu operon is present in the genome of the Escherichia coli strains Sequence analysis of the EcN genome revealed a gene cluster of 8 genes that resembles the operon chu (Figure 1A) in E. coli O157:H7 (EHEC). Closer inspection of the operon shows that the encoded proteins are highly conserved in both sequence identity and length (Figure 1A). Of note, ChuS from EcN (ChuSEcN) shares 98.25% sequence identity with ChuS from EHEC (ChuSEHEC), as the proteins differ in only six residues (Figure 1B), of which two are conservative replacements (T180S and Y297H). To further isolate distinct indigenous E. coli strains that possess the chu operon and naturally lack genes encoding the genotoxin colibactin, as described in EcN, a total of 401 fecal isolates from healthy Chilean volunteers were screened for chuS by PCR, primers sequences used in this study are listed in Table 1. Isolates exhibiting chuS amplification and lacking hemolytic activity were further assessed for pH and bile salt tolerance, resulting in the selection of UCPROB-001 and UCPROB-002 (data not shown). These functional criteria ensured that the final UCPROB formulation consisted of robust, safe, and chuS-positive bacterial mix. The chu operon in the UCPROB isolates is almost identical to that found on EHEC, and ChuS from UCPROB (ChuSUCPROB) has only three residue substitutions, from which two are also found in ChuSEcN (T78A and H46R) (Figure 1B). Moreover, a key residue for the catalysis of heme in ChuSEHEC, His-193,32 is present in both ChuSEcN and ChuSUCPROB, which suggests that the enzymes may be catalytically active. Furthermore, the residue substitutions do not seem to alter the predicted structure of ChuSEcN or ChuSUCPROB significantly, as the models had a high degree of structural similarity to the crystal structure of ChuSEHEC (0.670 Å and 0.683 Å, respectively) (Figures 1C and 1D). Notably, this is the first report describing the presence and structure of the chu operon in EcN, as no previous studies have documented this locus in this widely used probiotic strain. The chu operon from EcN, UCPROB-001, and UCPROB-002 is conserved (A) Schematic representation of the chu operon in EcN. chuS is depicted in purple, and the other known operon genes in light blue. The percentage identity and coverage of each protein encoded by the chu operon in EcN and UCPROB, compared to those in EHEC, are shown below each gene (Identity%/Coverage%). Since the chu operon from UCPROB-001 and UCPROB-002 are identical, only one coverage and residue identity is displayed for each protein. (B) MUSCLE alignment of ChuS from UCPROB-001, UCPROB-002, EcN, and EHEC colored by percentage identity in Jalview 2.11.4.1. The Alphafold2 predicted structures of (C) ChuSEcN and (D) ChuSUCPROB, along with their RMSD alignments to the crystal structure of ChuSEHEC (PDB: 1U9T), are shown. The red residues indicate the amino acids that are different between ChuSEHEC and ChuSEcN or ChuSUCPROB. Table 1List of primers used in this workUsePrimerSequenceAmpliconchuS screeningGIA_FwATC TGC ACG CAG CCA CGG TT180–200 bpGIA_RvTGG CGA GCA GAT GAC CGAGIB_FwATC TGC ACG CAG CCA CGA TTGIB_RvTGG CGA GCA GAT GAC TGAGIIA_FwGCG AGC CAT GAC CGA CGT345 bpGIIA_RvAGG CTG GTG ACG AAA CCG TCGIIB_FwTGG CGA GCC ATG ACC GAT GTGIIB_RvAGG CTG GTG ACG AAA CCA TCGIIIA_FwCGG GCG ATG ACC GAC GT180–200 bpGIIIA_RvGGT TGC CAA CAA ACA CCA TGA TTT CGIIIB_FwTGG CGG GCG ATG ACC GAT GTGIIIB_RvGGT TGC CAA CAA ACA CCA TGA TCT CAmplification of chuSchuS_FwTTA TGC AGT AAC GCC TTC CGG TAT C1029 bpchuS_RvATG AAC CAC TAC ACA CGC TGG CTT GEcN ΔchuS mutant generationKan_chuS (H1+P1)ATA AAC TCA AAC GCA ACG AGG TAA ATT GCG GAC GTG ACA Tgt gta ggc tgg agc tgc ttcKan_chuS (H2+P2)TCT GAA TTT TAA ATA AAT TTA TTT TTC GCC AGC CAC CCC Gca tat gaa tat cct cct tagLinearization of pKK233.3pKK_lin_FwGGG AAT TCT GTT TCC TGT GTG AAA4577 bppKK_lin_RvTCG ACC TGC AGC CAA GCT TCT GAmplification of chuS with overhangspKK_chuS_FwGAA GCT TGG CTG CAG GTC GAT TAT GCA GTA ACG CCT TCC GGT ATC1069 bppKK_chuS_RvACA CAG GAA ACA GAA TTC CCA TGA ACC ACT ACA CAC GCT GGC TTGConfirmation of insertioncon_NEB_FwGAG GAT CAA TCC GGC ATG TCC666 bpcon_NEB_RvAAC AAG CGC TCA TGA GCC CGAchuS qPCRq_chuS_FwATG AAG GGT AAA CGT CGG GTq_chuS_RvTGC GTA CAG ATT TTC ACC GGrpoD qPCRq_rpoD_FwTGA TGC TGG CTG AAA ACA CCq_rpoD_RvAGT TCA ACG GTG CCC ATT TCUL30 qPCRUL30_FwGGC CAG GCG CTT GTT GGT GTAUL30_RvATC ACC GAC CCG GAG AGG GAUL30_Probe56-FAM/CCG CCG AAC/ZEN/TGA GCA GAC ACC CGC/3IABkFQ List of primers used in this work Nucleotide and amino acid wide screening reveals high prevalence of chuS in Escherichia coli To determine the prevalence of the chuS gene in E. coli, we used Blastn and Blastp on a database curated from available E. coli genomes. The nucleotide database comprised 6,921 distinct E. coli genomes, whereas the protein database contained 5,602 protein sequences. Due to the large size of the database, some files were corrupted during downloading, which explains the differences in the final database sizes between genomes and proteins. According to Blastn, the chuS nucleotide sequence was found in 3,326 distinct genome sequences with high identity (the lowest being 87%), making chuS prevalent in 48.1% of our database (Figure S1). The range of positive chuS strains included identified E. coli strains and non-identified E. coli isolates. The same pattern can be seen with the protein sequence, where ChuS was found in 2,463 of the protein data files with high identity (lowest being 77%), making ChuS prevalent in 48.7% of our database (Figure S1). According to the database, most positive hits corresponded to “hematinate-forming HO ChuS”, “Hemin-degrading factor”, and very few corresponded to “Hemin transporter” or “putative Hemin-degrading factor”. These results suggest that the chuS gene is widely distributed across different E. coli strains. chuS overexpression reduces bacterial viability As chuS encodes a HO that degrades heme and generates carbon monoxide (CO), we sought to assess the fitness of these strains. We cultured them in LB broth for 24 h and monitored growth (OD600) and viability (CFU/mL) at multiple time points for E. coli UCPROB-001, E. coli UCPROB-002, and E. coli Nissle 1917 (EcN). To further evaluate the role of chuS in the physiology of E. coli, we generated a chuS knockout mutant (EcN ΔchuS) and the complemented strain (EcN ΔchuS pKK233.3chuS). As shown in Figures 2A and 2B, all strains exhibited similar growth kinetics and comparable areas under the OD600 curve (AUC), indicating that neither chuS deletion nor overexpression significantly altered bacterial proliferation under these conditions. However, viability measurements (Figures 2C and 2D) revealed a statistically significant reduction in CFU counts and AUC for EcN ΔchuS:chuS compared to the other E. coli strains. These results indicate that overexpression of chuS is associated with reduced bacterial viability, whereas basal expression of chuS does not compromise bacterial cell survival. The observed reduction in viability may reflect a combination of factors linked to high-level extrachromosomal expression, including CO accumulation, plasmid burden, and metabolic stress, without implying a single causal mechanism. chuS overexpression reduces bacterial viability under standard growth conditions UCPROB-001, UCPROB-002, EcN, EcN ΔchuS, and EcN ΔchuS::chuS were cultured under standard growth conditions in LB broth for 24 h, starting at an initial OD600 of 0.01. Growth and viability were measured at selected time points. (A and B) Bacterial growth was monitored using OD600 readings, and the area under the curve (AUC) was calculated to compare strains. (C and D) Viability was assessed by colony-forming unit (CFU) counts using spot plating, and AUC was calculated for comparisons. Data are shown as mean and SD in three independent experiments, except for UCPROB-001 and UCPROB-002, with two independent experiments. The data shown are the mean and SD. Statistical analyses were performed with one-way or two-way ANOVA test; ∗p < 0.05. To determine the effect of chuS expression on CO production, bacterial strains were cultured in LB broth under sealed and static conditions for 24 h to prevent CO loss. Then, CO levels were measured using an electrochemical detector immediately after opening the cultures. As shown in Figure S2A, EcN ΔchuS::chuS produced significantly higher CO levels than EcN and EcN ΔchuS, while no differences were observed between EcN and EcN ΔchuS. These results confirm that the overexpression of chuS in EcN ΔchuS::chuS leads to a substantial increase in CO production. Interestingly, all strains, regardless of genetic background, produced detectable CO levels in the absence of measurable transcription of chuS, suggesting the existence of chuS-independent mechanisms of basal CO production in standard laboratory conditions. To normalize CO production relative to bacterial viability, CO levels were expressed per 108 viable cells (Figure S2B). EcN ΔchuS::chuS exhibited a mean CO production per 108 cells approximately 200 times higher than EcN and EcN ΔchuS, further highlighting the impact of chuS overexpression on CO output. Viability (CFU counts) and growth (OD600) were also assessed to evaluate the fitness of the strains under these conditions. EcN ΔchuS::chuS exhibited significantly lower CFU counts than EcN and EcN ΔchuS (Figure S2C), consistent with the reduced viability previously observed in this strain. In contrast, no significant differences in OD600 were observed among the strains (Figure S2D), indicating that growth dynamics were unaffected despite the observed differences in viability. Relative transcription levels of chuS were quantified using qPCR and normalized to rpoD (Figure S2E). As anticipated, EcN ΔchuS::chuS exhibited markedly higher chuS transcription levels compared to EcN and EcN ΔchuS, confirming successful overexpression in this strain. Interestingly, no chuS transcription was detected in EcN under the evaluated conditions. The production of CO in vivo by E. coli strains is chuS-dependent and does not alter physiological parameters To evaluate the role of chuS in the probiotic properties of UCPROB and EcN, male and female C57BL/6 mice were monitored over 28 days of probiotic administration twice a week. Weight gain and clinical scores were assessed throughout the experiment. No significant weight loss was observed in any group during the treatment period, and no increase in clinical score was observed in any group. Although mice treated with the EcN ΔchuS exhibited slightly lower average weight gain than the wild-type strain, this difference was not statistically significant (Figure 3A). At the end of the 28-day treatment, mice were euthanized. Blood samples, liver, spleen, and mesenteric lymph nodes were collected to assess bacterial dissemination to deep organs and obtain serum from the blood. No bacterial dissemination to blood or internal organs was detected, confirming the safety of all strains, including the UCPROB formulation. Effects on weight gain, carboxyhemoglobin production, and cytokine serum levels of the intragastrical administration of Escherichia coli strains in mice Six-week-old C57BL/6 male and female mice received intragastrical administration of 3.5 × 107 CFU of UCPROB, EcN, EcN 1917 ΔchuS, and 50 μL of PBS (as a control) twice a week for 28 days. (A) Percentage of body weight change relative to day 0. (B) Carboxyhemoglobin (HbCO) levels in serum were quantified by ELISA. (C) IL-10 (D) IFN-γ levels in serum were measured by ELISA. The assay included 6 mice per group in two independent experiments. The data shown are the mean and SD. Statistical analyses were performed with one-way or two-way ANOVA test; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗∗p < 0.0001. The concentration of HbCO in mice’s serum was measured as an indirect indicator of CO production following bacterial administration. This analysis utilized serum collected from the treated mice at the end of the experimental period, providing a reliable approach to evaluating in vivo CO production. As anticipated, mice treated with EcN exhibited significantly higher HbCO concentrations than untreated controls, highlighting EcN’s ability to produce CO in vivo (Figure 3B). Remarkably, mice treated with UCPROB also showed high HbCO levels, statistically comparable to those of EcN WT. Notably, the administration of Escherichia coli K12 (EcK12), a strain that does not harbor the chu operon on its chromosome, resulted in weight dynamics (Figure S3A) and HbCO production levels (Figure S3B) comparable to those observed in the EcN ΔchuS mutant. These results demonstrate that oral administration of EcN and UCPROB strains increases carboxyhemoglobin levels in vivo, suggesting that these strains increase CO production in vivo. This is consistent with bioinformatics analysis results showing that UCPROB isolates are homologous to EcN. To assess whether the oral administration of E. coli strains affects the systemic immune response, serum levels of IL-10 and IFN-γ were quantified in C57BL/6 mice treated intragastrically with EcN, EcN ΔchuS, UCPROB, or PBS as a control. As shown in Figure 3C, no significant differences were observed in the serum levels of IL-10. For IFN-γ, we observed increased serum levels in mice treated with EcN and UCPROB (Figure 3D), although the levels detected were low. No significant change in production of this cytokine was observed in the serum of mice treated with EcN ΔchuS, demonstrating that the administration of ChuS-producing E. coli strains results in increased levels of HbCO and IFN-γ in serum. Finally, relative transcriptional levels of hmox1 (HO-1) were evaluated in the colon, as a local intestinal readout, and in the spleen, as a distal systemic tissue associated with immune regulation. As shown in Figures S4A and S4B, mice treated with UCPROB and EcN exhibited a consistent trend toward increased hmox1 expression in both tissues compared to PBS and EcN ΔchuS controls, although these differences did not reach statistical significance. ChuS-dependent CO production by UCPROB modulates viral replication and disease severity during HSV-1 skin infection To evaluate the role of chuS-expressing commensal bacteria in antiviral protection, female C57BL/6 mice were treated intragastrically with the UCPROB mix, EcN, EcN ΔchuS, or PBS as a control twice a week for 28 days. After the probiotic treatment period, mice were topically infected with HSV-1 in the skin and monitored for 11 days post-infection (dpi). A group without probiotic treatment or HSV-1 infection (UNINF) was included. Clinical scores, viral loads, cytokine levels, and HbCO were measured to assess the impact of chuS-dependent CO production on HSV-1 skin disease outcomes. The HSV-1-specific skin disease score, monitored daily, demonstrated that the mice treated with ChuS-producing E. coli showed a marked reduction in disease severity, particularly between days 2 and 5 post-infection, compared to the PBS and EcN ΔchuS groups (Figure 4A). This reduction in clinical signs was consistently lower in the UCPROB-treated group than that in the EcN group. An area under the curve (AUC) analysis that integrates the severity and duration of clinical scores confirmed this effect, showing that UCPROB significantly reduced cumulative disease burden relative to PBS and ΔchuS, and outperformed EcN (Figure 4B). Effect of UCPROB, EcN, and EcN ΔchuS treatment in mice challenged with HSV-1 in the skin Six-week-old C57BL/6 male and female mice received intragastrical administration of 3.5 × 107 CFU of UCPROB, EcN, EcN 1917 ΔchuS, and 50 μL of PBS (as a control) twice a week for 28 days. At day 28, mice were topically infected in the skin with 1 × 106 PFU of HSV-1 and observed for 11 days. (A) Clinical score of HSV-1 skin disease. (B) Area under the curve (AUC) of the HSV-1 clinical score. (C) Viral load was determined by quantification of the HSV-1 UL30 gene copy number. (D) IL-10 and (E) IFN-γ concentrations in serum at day 11 post-infection, measured by ELISA. There were 6 mice per group in two independent experiments. The data shown are the mean and SD. Statistical analyses were performed with one-way or two-way ANOVA test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. Viral load was assessed by qPCR quantifying HSV-1 UL30 gene copies in infected skin samples at 11 dpi. UCPROB-treated mice exhibited significantly lower viral loads than PBS and EcN ΔchuS, and not only outperformed EcN WT but also showed viral titers that approached those of the uninfected (UNINF) group (Figure 4C). This reduction supports a robust antiviral effect mediated by the UCPROB formulation, likely linked to its ability to produce CO in vivo. Serum HbCO concentration, measured at 11 dpi, was significantly higher in both UCPROB and EcN WT groups, compared to PBS, EcN ΔchuS, and UNINF (Data not shown). Serum levels of the anti-inflammatory cytokine IL-10 were significantly increased in mice treated with UCPROB and EcN compared to PBS, EcN ΔchuS, and UNINF groups (Figure 4D). In parallel, IFN-γ levels were significantly elevated in the UCPROB, EcN, and EcN ΔchuS groups relative to UNINF. Notably, a statistically significant increase compared to the PBS-treated group was observed exclusively in the UCPROB-treated mice (Figure 4E). Discussion In this study, we report that human commensal Escherichia coli strains, isolated from healthy donors UCPROB-001 and UCPROB-002, and the commercial probiotic EcN, carry a functional chu operon encoding the bacterial HO ChuS, enabling them to utilize heme and produce carbon monoxide (CO) as a metabolic byproduct. The chu operon identified in the E. coli strains included in this study is highly conserved and shows similarity to the pathogenic strain E. coli O157:H7. Although the chu operon is essential for pathogenic E. coli to survive during infection,40 our findings in EcN and UCPROB strains suggest that the chu operon is not intrinsically a virulence factor and may play additional roles in non-pathogenic strains. Comparative genomic and protein sequence analysis revealed that the chu operon present in EcN is similar to the chu operon E. coli O157:H7, and the chu operon of UCPROB-001 and UCPROB-002 is nearly identical to that of EcN, exhibiting high alignment coverage and nucleotide sequence across all genes. Furthermore, the ChuS enzyme from both UCPROB strains exhibited near-complete sequence and residue identity with ChuS from EcN and E. coli O157:H7. These findings highlight that ChuS-mediated CO production is not exclusive to model probiotics like EcN but is naturally present in human commensal E. coli strains, such as those selected for the UCPROB formulation, emphasizing their potential role in immune regulation. Our analysis shows that chuS is present in nearly half of the analyzed E. coli genomes and proteomes, suggesting it is not restricted to pathogenic strains. In the gut environment, where heme levels can rise during inflammation, chuS-mediated heme degradation may provide a fitness advantage to certain E. coli strains, supporting their persistence under dysbiotic conditions. This metabolic trait could play a role in shaping microbial community dynamics and influencing host-microbiota interactions. Its presence in probiotic and commensal strains like UCPROB and E. coli Nissle 1917 further suggests potential implications for immunomodulation and mucosal homeostasis, reinforcing the relevance of exploring chuS-positive commensals in microbiome-driven therapies. The expression of chuS in trans in the EcN strain lacking the gene EcN ΔchuS::chuS significantly reduced bacterial viability compared to the wild-type EcN strain. This reduction was evidenced by lower CFU counts despite similar growth rates, indicating that high-level extrachromosomal expression of chuS is associated with reduced bacterial viability. Similarly, a study showed that treating E. coli with exogenous CO reduced its growth rate and CFUs.41 Although we did not detect a reduction in the growth rate, the decreased CFU count after chuS induction may reflect multiple contributing factors, including increased intracellular CO levels, plasmid burden, and metabolic stress. RT-qPCR analysis revealed that chuS transcript levels in EcN ΔchuS::chuS were over 1,000-fold higher than those in the wild-type strain; this marked overexpression likely imposes a substantial metabolic cost relative to the wild-type strain. Conversely, the deletion of chuS did not affect the growth or viability of the ΔchuS mutant, indicating that the absence of chuS does not inherently alter bacterial fitness under the tested conditions. These findings highlight a relationship between chuS expression and bacterial viability, underscoring the potential costs of excessive ChuS activity. The selection process for UCPROB strains incorporated not only genomic and functional chuS conservation but also excluded potentially hemolytic isolates and prioritized strains with high tolerance to gastrointestinal stressors such as bile salts and acidic or alkaline pH. This methodology ensured the safety and resilience of UCPROB bacteria under physiological conditions, reinforcing their suitability as a potential therapeutic formulation. Overexpression of chuS in EcN ΔchuS::chuS resulted in significantly elevated CO production in vitro and in vivo. These results confirmed the pivotal role of chuS in CO synthesis. The elevated CO levels were consistent with the high transcriptional activity of chuS in this strain, reinforcing the association between chuS expression and CO output. Importantly, in vivo measurements of HbCO corroborated these findings, with significantly higher HbCO levels observed in UCPROB and EcN-treated mice compared to controls. Similarly, a study showed that deletion of chuS in E. coli NC101 decreases the accumulation of CO in infected mice livers, which is reversed after the trans expression of chuS.42 Interestingly, they observed increased mammalian HO 1 gene (hmox1) expression in the livers of mice infected with a strain overexpressing chuS. Therefore, the increased blood CO concentration observed in mice inoculated with UCPROB and EcN may reflect a combination of bacterial ChuS activity and host-derived CO production pathways, including hmox1 induction. However, treatment with E. coli strains lacking chuS did not increase the levels of HbCO in mice, supporting the notion that the production of CO by E. coli is due to the activity of ChuS. Measuring HbCO levels is particularly significant, and we describe it as a biomarker for endogenous CO generation, offering insight into the functional activity of chuS and its potential role in modulating host physiology, while acknowledging that HbCO represents a systemic surrogate marker and does not provide information on local CO concentrations at the site of infection. These data underscore the physiological relevance of chuS-mediated CO production and its potential immunomodulatory relevance under the experimental conditions tested. Remarkably, the UCPROB formulation induced HbCO levels in vivo that were comparable to EcN, further validating its ability to produce CO under physiological conditions. Given that UCPROB strains are derived from a healthy human microbiome and require no genetic manipulation, they offer a promising and naturally equipped alternative for CO-based probiotic interventions. This lack of gene modification positions UCPROB as a non-genetically modified probiotic platform; however, any potential regulatory or translational advantages remain prospective, as safety and efficacy in humans have not yet been evaluated. Their natural origin and capacity to produce CO under physiological conditions support further investigation of UCPROB as a candidate for microbiota-based, CO-associated interventions in inflammatory and infectious disease contexts. In vivo, UCPROB demonstrated safety and immunomodulatory effects in a murine model, as evidenced by stable weight gain, unchanged clinical scores, and the absence of bacterial dissemination to internal organs. Additionally, cytokine profiling revealed no significant alterations in systemic IL-10 levels and a slight but significant increase in IFN-γ levels compared to control groups, confirming the safety of the probiotic treatments under experimental conditions. The safety of EcN has been well documented in vivo, with multiple studies showing that its administration does not cause clinical symptoms, weight loss, or bacterial dissemination under homeostatic conditions, supporting its classification as an immunologically neutral and non-pathogenic probiotic.43,44 Similarly, UCPROB administration was well tolerated in mice, with no evidence of clinical symptoms, colitis, or bacterial dissemination. Safety conclusions are therefore limited to the preclinical conditions evaluated in this study, including dose, duration, and experimental model. In this context, the analysis of host hmox1 transcription provides additional insight into the potential spatial dimensions of CO-associated host responses. We evaluated hmox1 expression both in the colon, as a local intestinal compartment directly exposed to the administered bacteria, and in the spleen, as a distal systemic immune organ. Although no statistically significant differences were detected, both UCPROB- and EcN-treated mice exhibited a consistent trend toward increased hmox1 transcription in the colon and spleen compared to PBS and EcN ΔchuS controls. Notably, the concordant direction of the response in the colon and the spleen suggests that ChuS-mediated CO production may influence HO-1 transcription both locally at the intestinal level and distally at the systemic level. While these data do not establish causality, they are consistent with a model in which bacterially derived CO contributes to host HO-1 pathway engagement across tissue compartments, warranting further tissue- and compartment-resolved analyses. In the HSV-1 skin infection model, an inflammatory viral disease characterized by usually painful, fluid-filled skin lesions, treatment with the UCPROB formulation significantly reduced disease severity, as evidenced by lower clinical scores and viral loads. These findings indicate an association between UCPROB treatment and reduced viral burden and tissue pathology, likely mediated through host immune modulation, as supported by previous studies using pharmacological inducers of HO-1 in immune and non-immune cells.16 Interestingly, UCPROB treatment led to a significant elevation in IL-10 and IFN-γ, suggesting a coordinated anti-inflammatory and antiviral immune response. This balance may contribute to the reduced tissue pathology and enhanced viral control observed in UCPROB-treated animals. The dynamic interplay between IL-10 and IFN-γ highlights UCPROB formulation capacity to fine-tune immune responses in a manner that promotes viral clearance while reducing tissue damage. IFN-γ is a central mediator of antiviral immunity, supporting viral control through activation of innate and adaptive effector mechanisms, whereas IL-10 plays a key role in constraining excessive inflammation and limiting immunopathology. In this context, the concomitant increase in IL-10 may act to modulate IFN-γ-driven inflammatory responses without compromising antiviral efficacy, thereby preserving tissue integrity while maintaining effective viral control. Previous studies have shown that some E. coli strains can modulate antiviral immunity by enhancing host resistance while controlling inflammation.45 Their administration has been associated with reduced viral replication and tissue damage, likely through the balanced regulation of cytokines such as IL-10 and IFN-γ.45 Moreover, CO has been reported to limit viral replication by interfering with host cell pathways essential for viral propagation,46,47 suggesting a potential role for CO in UCPROB’s antiviral effects. Therefore, the observed effects may result from a combination of ChuS-mediated bacterial CO production and indirect mechanisms, including host responses to bacterial exposure and activation of endogenous pathways, and not necessarily a direct production of CO in the skin. The findings of this study underscore the versatility of UCPROB formulation as a probiotic with anti-inflammatory and antiviral properties. The role of chuS in CO production is particularly important, supporting an association between this metabolic byproduct and the immunomodulatory effects observed in the HSV-1 skin infection model assessed herein. The delicate balance between chuS expression, bacterial viability, and CO output presents challenges and opportunities for optimizing probiotic strains as a therapeutic agent. Further mechanistic and translational studies will be required to determine how ChuS-mediated CO production can be leveraged while preserving bacterial fitness, and if this probiotic formulation could be effective after an HSV-1 infection. We conclude that chuS is a critical determinant of CO production in Escherichia coli probiotic strains, highlighting its role in modulating immune responses during HSV-1 skin infection. The anti-inflammatory and antiviral-associated effects of UCPROB, mediated in part by chuS-dependent CO production, position this probiotic as a promising prophylactic tool for inflammatory and infectious diseases. Importantly, UCPROB-treated mice exhibited a significant reduction in viral load compared to control and EcN-treated groups, further supporting its prophylactic potential. The compensatory not significant increase in IL-10 observed during HSV-1 infection may illustrate the potential of UCPROB’s ability to balance inflammatory and regulatory pathways, preventing excessive immune activation. These results suggest that similar mechanisms may naturally occur in the human gut microbiota, where commensal E. coli strains expressing chuS could contribute to the local regulation of inflammation and immune homeostasis. Future research should focus on optimizing chuS expression to enhance therapeutic efficacy while ensuring bacterial viability. In addition, while the present study evaluates a prophylactic/pre-conditioning administration strategy, an important next step will be to determine whether UCPROB or EcN retain efficacy when administered after HSV-1 infection, which is essential to support therapeutic relevance. Addressing post-infection dosing paradigms will be critical for defining the translational scope of this approach. Together, these findings pave the way for developing innovative, CO-mediated strategies to combat various pathological conditions. Limitations of the study Carbon monoxide production in vivo was assessed indirectly using serum carboxyhemoglobin as a systemic surrogate marker. While this supports increased CO exposure, it does not provide direct or tissue-specific quantification of CO levels. In particular, the presence and distribution of CO at the skin infection site were not evaluated. In addition, the experimental design does not allow discrimination between bacterially derived CO and CO generated through host HO-1 activity, which are therefore considered potentially concurrent mechanisms. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Susan M. Bueno (sbuenor@uc.cl). Materials availability All unique/stable reagents generated in this study, including the Escherichia coli UCPROB-001 and UCPROB-002 strains and derived bacterial constructs, are available from the lead contact upon reasonable request. A completed materials transfer agreement (MTA) may be required for their distribution, particularly in cases with potential commercial applications. Data and code availability Whole genome sequencing data generated in this study have been deposited in GenBank and are publicly available as of the date of publication. Accession numbers are listed in the key resources table (PRJNA1357318 and PRJNA1357320). Reference genomes of Escherichia coli Nissle 1917 and Escherichia coli O157:H7 (EHEC) were obtained from GenBank under accession numbers CP007799.1 and CP008957.1, respectively. Additional data supporting the findings of this study are available from the lead contact upon reasonable request. This study does not report original code. Acknowledgments This research was supported by the FONDEF IDeA grant no. ID20I10082 (S.M.B. and A.W.) and FONDECYT Regular grants no. 1231905 (S.M.B.), no. 1240971 (P.A.G.), no. 1231851 (A.M.K.), and no. 1250827 (F.M.-G) from the Chilean National Agency for Research and Development (ANID); the Millennium Institute on Immunology and Immunotherapy (ICN 2021_045); the Chile-France Scientific Cooperation Program ECOS-ANID no. ECOS22027 (S.M.B.); Avanza UC grant no. AV25137 (P.A.G.); the PhD Fellowship from the Research Vice-Presidency at Pontificia Universidad Católica de Chile (O.P.V.) and the ANID National PhD fellowship no. 21251772 (O.P.V.) and no. 21202008 (P.H.S.). We also thank the members of the Microbial Pathogenesis, Biomedical Molecular Immunology, and Virus and Inflammation laboratories at the Pontificia Universidad Católica de Chile for their discussions and advice. We also thank the Sequencing and Omics Technologies Unit, the Bioinformatics and Computational Biology Unit, and the Teaching Laboratory at the Pontificia Universidad Católica de Chile. Author contributions O.P.V. and S.M.B. conceived and planned the experiments. O.P.V., G.H.-E., P.R.-B., and E.I.T. carried out the experiments. Y.V., A.P.-I., P.P.-S., L.V.-H., J.P.B.-A., V.A.V.-E., F.B.Q., S.C.-G., J.S.-A., J.A.-L., I.D.S., P.H.S., E.C., V.S., F.B., I.C.-A., and F.M.-G. contributed to sample processing and experimental support. M.A.-L. and A.M. performed the evaluation and enrollment of healthy volunteers for E. coli isolation from human microbiota. O.P.V., G.H.-E., P.R.-B., E.I.T., A.W., A.K.M., P.A.G., and S.M.B. contributed to the interpretation of the results. O.P.V., G.H.-E., P.R.-B., and S.M.B. took the lead in writing the manuscript. All authors provided critical feedback and contributed to shaping the research, data analysis, and final manuscript. Declaration of interests Pontificia Universidad Católica de Chile has filed a provisional patent application protecting the UCPROB strains. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERBacterial and virus strainsEscherichia coli Nissle 1917Ardeypharm GmbHN/AEscherichia coli UCPROB-001This studyN/AEscherichia coli UCPROB-002This studyN/AEscherichia coli ΔchuSThis studyN/AEscherichia coli ΔchuS pkk233.3:chuSThis studyN/AEscherichia coli K12Laboratory stockATCC 10798HSV-1 (strain KOS)Donated by Dr. Betsy Herold and Dr. Natalia Cheshenko from the Albert Einstein College of Medicine, USAATCC VR-733Streptococcus pneumoniae D39UK Health Security AgencyNCTC 7466Bacillus cereusPontificia Universidad Católica de ChileNCTC 2599Biological samplesMouse serumThis studyN/AMouse skin tissueThis studyN/AMouse bloodThis studyN/AMouse spleen tissueThis studyN/AMouse liver tissueThis studyN/AMouse mesenteric lymph nodes tissueThis studyN/AChemicals, peptides, and recombinant proteinsTRIzol ReagentInvitrogenCat# 15596026BactoTM TryptoneGibcoCat# 211705BactoTM Yeast ExtractGibcoCat# 212750Sodium chlorideWinklerCat# SO-1455Potassium chlorideWinklerCat# PO-1260Magnesium chlorideWinklerCat# MA-0960D-(+)-GlucoseSigmaCat# G7021-1 KGL-(+)-ArabinoseMerck MilliporeCat# 178680Agar AgarMerck MilliporeCat# 1016141000MacConkey AgarMerck MilliporeCat# 100205Columbia Base Sheep Blood AgarValtekCat# 285-085Bile saltsMerck MilliporeCat# B8756-50 GRIPTGSigma-AldrichCat# I6758AmpicillinSigma-AldrichCat# A9518Kanamycin sulfateSigma-AldrichCat# K4000ChloroformMerck MilliporeCat# C2432-2.5LIsopropanolMerck MilliporeCat# 190764-2.5LEthanolWinklerCat# 2223Phosphate Buffered Saline 10XCorningCat# 46-013-CMIsofluraneBaxterCat# 210190Collagenase DSigma-AldrichCat# 11088858001VEET® depilatory agentVeetCat# 346604Critical commercial assaysiScript cDNA kitBio-RadCat# 1708891TaqMan RNA-to-CT kitThermo ScientificCat# 4392938GoTaq G2 Flexi DNA PolymerasePromegaCat# M7801iScriptTM cDNA SynthesisBio-RadCat# 1708891Mouse IL-10 ELISABD BiosciencesCat# 555252Mouse IFN-γ ELISABD BiosciencesCat# 555138Carboxyhemoglobin ELISAMyBioSourceCat# MBS261419NEBuilder HiFi DNA Assembly Master MixNew England BiolabsCat# E2621Deposited dataEscherichia coli UCPROB-001 genomeThis studyGenBank: PRJNA1357318Escherichia coli UCPROB-002 genomeThis studyGenBank: PRJNA1357320Escherichia coli Nissle 1917 genomeThis studyGenBank: CP007799.1Escherichia coli O157:H7 genomeThis studyGenBank: CP008957.1Experimental models: Organisms/strainsMouse C57BL/6 WTCIBEM, Pontificia Universidad Católica de Chile from The Jackson LaboratoriesCat# 000664OligonucleotidesGapdh assay (Mm99999915_g1)Thermo FisherCat# 4331182Hmox1 assay (Mm00516005_m1)Thermo FisherCat# 4331182UL30_Probe (56-FAM/CCG CCG AAC/ZEN/TGA GCA GAC ACC CGC/3IABkFQ)Thermo FisherN/AUL30_Fw (GGC CAG GCG CTT GTT GGT GTA)IDTN/AUL30_Rv (ATC ACC GAC CCG GAG AGG GA)IDTN/AGIA_Fw (ATC TGC ACG CAG CCA CGG TT)IDTN/AGIA_Rv (TGG CGA GCA GAT GAC CGA)IDTN/AGIB_Fw (ATC TGC ACG CAG CCA CGA TT)IDTN/AGIB_Rv (TGG CGA GCA GAT GAC TGA)IDTN/AGIIA_Fw (GCG AGC CAT GAC CGA CGT)IDTN/AGIIA_Rev (AGG CTG GTG ACG AAA CCG TC)IDTN/AGIIB_Fw (TGG CGA GCC ATG ACC GAT GT)IDTN/AGIIB_Rv (AGG CTG GTG ACG AAA CCA TC)IDTN/AGIIIA_Fw (CGG GCG ATG ACC GAC GT)IDTN/AGIIIA_Rv (GGT TGC CAA CAA ACA CCA TGA TTT C)IDTN/AGIIIB_Fw (TGG CGG GCG ATG ACC GAT GT)IDTN/AGIIIB_Rv (GGT TGC CAA CAA ACA CCA TGA TCT C)IDTN/AchuS_Fw (TTA TGC AGT AAC GCC TTC CGG TAT C)IDTN/AchuS_Rv (ATG AAC CAC TAC ACA CGC TGG CTT G)IDTN/AKan_chuS (H1+P1) (ATA AAC TCA AAC GCA ACG AGG TAA ATT GCG GAC GTG ACA Tgt gta ggc tgg agc tgc ttc)IDTN/AKan_chuS (H2+P2) (TCT GAA TTT TAA ATA AAT TTA TTT TTC GCC AGC CAC CCC Gca tat gaa tat cct cct tag)IDTN/ApKK_lin_Fw (GGG AAT TCT GTT TCC TGT GTG AAA)IDTN/ApKK_lin_Rv (TCG ACC TGC AGC CAA GCT TCT G)IDTN/ApKK_chuS_Fw (GAA GCT TGG CTG CAG GTC GAT TAT GCA GTA ACG CCT TCC GGT ATC)IDTN/ApKK_chuS_Rv (ACA CAG GAA ACA GAA TTC CCA TGA ACC ACT ACA CAC GCT GGC TTG)IDTN/Acon_NEB_Fw (GAG GAT CAA TCC GGC ATG TCC)IDTN/Acon_NEB_Rv (AAC AAG CGC TCA TGA GCC CGA)IDTN/Aq_chuS_Fw (ATG AAG GGT AAA CGT CGG GT)IDTN/Aq_chuS_Rv (TGC GTA CAG ATT TTC ACC GG)IDTN/Aq_rpoD_Fw (TGA TGC TGG CTG AAA ACA CC)IDTN/Aq_rpoD_Rv (AGT TCA ACG GTG CCC ATT TC)IDTN/ARecombinant DNApKD4 plasmidDonatedN/ApKD46 plasmidDonatedN/ApKK233.3 plasmidDonatedN/ApKK233.3:chuS plasmidThis studyN/ASoftware and algorithmsGraphPad Prism 10GraphPad Softwarehttps://www.graphpad.comKraken2Johns Hopkins Universityhttps://ccb.jhu.edu/software/kraken2FlyeGenome Assembly Toolkithttps://github.com/fenderglass/FlyeRavenGenome Assembly Toolkithttps://github.com/lbcb-sci/ravenMiniasmHeng Li Labhttps://github.com/lh3/miniasmMedakaOxford Nanopore Technologieshttps://github.com/nanoporetech/medakaBUSCOOrthoDBhttps://busco.ezlab.orgSAMtoolsWellcome Sanger Institutehttp://www.htslib.orgProkkaTorsten Seemann Labhttps://github.com/tseemann/prokkaEggNOG-mapperEMBLhttp://eggnog-mapper.embl.deMEGA11MEGA Softwarehttps://www.megasoftware.netJalviewUniversity of Dundeehttps://www.jalview.orgPyMOLSchrödingerhttps://pymol.orgColabFoldGoogle Research/DeepMind-basedhttps://github.com/sokrypton/ColabFoldOtherCO electrochemical detectorSmart SensorAS8700A Experimental model and study participant details Ethics Approval and consent to participate Fecal samples were obtained from three healthy adult Chilean volunteers, who consented to participate by signing an informed consent form, as approved by the Institutional Scientific Ethical Committee for Health under protocol number 200316004. The volunteers were evaluated by a physician and met strict inclusion and exclusion criteria. Animal use All in vivo experiments were performed in six-week-old C57BL/6 male and female mice. The mice were kept at the pathogen-specific free animal facility at Pontificia Universidad Católica de Chile. The Institutional Scientific Ethical Committee for Animal Care and Environment approved the mouse experiments under protocol numbers 200316004 and 240131006. Method details Bacterial and viral strains Ardeypharm GmbH provided the Escherichia coli Nissle 1917 (EcN) strain used in this study. The UCPROB formulation consists of Escherichia coli UCPROB-001 and Escherichia coli UCPROB-002, two bacterial strains isolated from fecal samples of healthy Chilean adult volunteers. Streptococcus pneumoniae D39 (NCTC 7466) was obtained commercially from the UK Health Security Agency. Bacillus cereus (NCTC 2599) was provided by the Teaching Laboratory at Pontificia Universidad Católica de Chile. Herpes simplex virus type 1 strain KOS (ATCC VR-733) was provided by Dr. Betsy Herold and Dr. Natalia Cheshenko from the Albert Einstein College of Medicine, USA, to Dr. Pablo González, and was prepared as previously described.48 Bacterial isolation and screening for chuS Under sterile conditions, commensal bacteria were isolated from fecal samples of Chilean volunteers, resulting in a total of 401 bacterial isolates from three volunteers. Genomic DNA was extracted from each isolate and screened for the presence of the chuS gene, which encodes a bacterial heme oxygenase, using a standardized PCR protocol. PCR was performed using the GoTaq G2 Flexi DNA Polymerase kit (Promega). To increase detection sensitivity and account for potential sequence diversity, multiple primer pairs were employed (Table S1), each designed to amplify conserved regions of phylogenetically distinct chuS-like genes identified in enterobacteria. PCR products were resolved on 1.2% agarose gels and visualized under UV illumination. Out of the 401 isolates, 271 were chuS-positive. Hemolysis assay Hemolytic activity of chuS-positive isolates was assessed on Columbia Base Sheep Blood Agar (Valtek). Plates were incubated at 37°C for 24 h under aerobic conditions, and hemolysis was classified as α (partial), β (complete), or γ (none). As controls, Streptococcus pneumoniae and Bacillus cereus were used to represent α- and β-hemolysis, respectively, while Escherichia coli Nissle 1917 (EcN) and Escherichia coli K12 served as γ-hemolysis (negative) controls. Isolates displaying α- or β-hemolysis were excluded from further analysis. Out of the 271 chuS-positive isolates, 11 isolates exhibited α-hemolysis and were excluded, resulting in a final total of 260 isolates retained for subsequent characterization. Bile salt and pH tolerance assays To assess the robustness of isolated E. coli strains as potential probiotics, the 260 chuS-positive and γ-hemolytic isolates were subjected to stress tolerance testing. Tolerance to acidic and alkaline conditions was assessed by culturing bacteria in LB broth (prepared with 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L sodium chloride in distillated water) adjusted to pH 2, 4, 6, 8, or 10 using 1 M HCl or 1 M NaOH. Resistance to bile salts was assessed by culturing bacteria in LB broth supplemented with 0.25, 0.5, 1, and 2% weight/volume bile salts (Merck). Overnight bacterial cultures were normalized to OD600= 0.1 and inoculated into 96-well plates containing the various media. Cultures were incubated at 37°C for 5 h, and growth was quantified by measuring OD600 using a plate reader. All conditions were tested in technical duplicates. Growth under each stress condition was normalized to growth in standard LB at pH 7.0 and expressed as a relative tolerance index. For each pH value and bile salt concentration, isolates were considered tolerant when their OD600 after 5 h of incubation was equal to or higher than that measured for the reference probiotic strain, Escherichia coli Nissle 1917 (EcN), evaluated in parallel under identical conditions. Whole-genome sequencing using Nanopore Technology Bacterial genomes were sequenced using Oxford Nanopore Technologies (GridION platform) at the Sequencing and Omic Technologies Unit at Pontificia Universidad Católica de Chile. Raw reads were processed for quality assessment and adapter trimming using NanoPlot, Porechop, and NanoFilt. Taxonomic classification was performed with Kraken2 to confirm sample identity and detect contaminants. Genome assemblies were generated using Flye, Raven, and Miniasm, and the best assembly was selected based on contiguity and coverage metrics. The chosen assemblies were polished using Medaka and evaluated with BUSCO and SAMtools for completeness and depth. Gene prediction and annotation were performed using Prokka and EggNOG-mapper. All bioinformatic analyses were conducted by the Bioinformatics and Computational Biology Unit of the same institution. Protein modeling and sequence alignment The crystal structure of ChuS from E. coli O157:H7 was obtained from PDB: 1U9T. The predicted structure of ChuS from UCPROB-001/002 and EcN was modeled using ColabFold v1.5.5. The proteins were aligned in PyMOL 3.1.6.1 using the “align” command with predetermined settings, which calculated the Root Mean Square Deviation (RMSD). The protein sequences encoded in the chu operon of UCPROB-001 and UCPROB-002 were aligned to their homologs in E. coli O157:H7 with BLASTp to calculate coverage and sequence identity, and multiple sequence alignments were performed with MEGA11 (version 11.0.13). The resulting alignments were visualized using Jalview (version 10.0.7). Analysis of chuS prevalence in nucleotide and protein sequences of Escherichia coli chuS gene prevalence in E. coli was determined using Blastn and Blastp of chuS nucleotide and aminoacidic sequences from E. coli Nissle 1917 (CP007799) using a curated database. The database was constructed using sequences extracted from the NCBI corresponding to complete genomes and the corresponding protein data of each genome in the taxa Escherichia coli. The construction of the nucleotide49 and protein50,51 databases and Blastn/Blastp alignments were done using Blast+ v2.17 package running on a local machine. Genetic manipulation of EcN A mutant EcN strain lacking the chuS gene (EcN ΔchuS::aph) was constructed following Datsenko & Wanner’s protocol.52 Briefly, the kanamycin resistance gene from pKD4 was amplified by PCR using 60-nucleotide primers, including a 40-nucleotide extension homologous to the loci flanking the chuS coding sequence and 20-nucleotide primers for pKD4. The PCR product was purified from agarose gel using the Wizard SV gel and PCR cleanup system and electroporated into EcN harboring pKD46 in 60% glycerol in 0.2 cm cuvettes at 2.5 kV. After transformation, the cells were recovered in SOC broth (consisting of 20 g/L tryptone, 5 g/L yeast extract, 0.5 g/L NaCl, 0.186 g/L KCl, supplemented with 10 mM MgCl2 and 20 mM glucose) containing 10% arabinose at 37°C for 3 h, after which they were seeded on LB agar supplemented with 50 μg/mL kanamycin. The next day, transformants were picked and confirmed by PCR. The chuS complementation plasmid was based on the pKK233.3 plasmid. The plasmid was designed using SnapGene v.7.2. Insertion of chuS was carried out through Gibson Assembly methodology as follows: pKK233.3 was linearized using the Platinum SuperFi II DNA Polymerase and primers with overhangs homologous to the coding sequence of chuS. The open reading frame (ORF) of chuS was amplified from EcN genomic DNA (gDNA) using primers that annealed at the 5’ and 3’ ends of the ORF that included overhangs complementary to the linearized plasmid. ChuS was inserted using the NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs), following the manufacturer’s instructions. Confirmation of the correct insertion of chuS was made using PCR. All the primers used are listed in Table S1. After the obtention of the plasmid, it was stored in an E. coli DH5α at -80°C. Complementation was carried out by electroporation of electrocompetent EcN ΔchuS::aph as described above. 50 μL of electrocompetent EcN ΔchuS::aph cells were electroporated with 100 ng of pKK233.3_chuS. Electroporated bacteria were recovered in 1 mL of SOC medium for 1 h at 37°C. Subsequently, 100 μL were plated on LB agar containing 100 μg/mL of ampicillin for mutant selection. The transformants were confirmed using PCR. RNA Extraction RNA was obtained using TRIzol™ reagent (Invitrogen) for every sample studied. Samples stored at -80°C in 500 μL of TRIzol reagent were thawed on ice. Subsequently, 200 μL of chloroform was added to each sample and shaken vigorously for 15 s to homogenize. The samples were incubated for 3 min at room temperature and centrifuged at 12,000 x g for 15 min at 4°C. The aqueous phase of each sample was removed, avoiding contact with the other phases, and transferred to clean tubes. 500 μL of 100% isopropanol was added to the new tubes, which were then incubated at -20°C for 30 min. The samples were centrifuged at 12,000 x g for 10 min at 4°C. The supernatant was discarded, and the pellet was washed with 1 mL of 75% ethanol (prepared in nuclease-free water), homogenized by inversion, and centrifuged at 7,500 x g for 5 min at 4°C. The supernatants were discarded, and the tubes were left open for the pellet to dry for 10-15 min. Finally, the pellets were resuspended in 50 μL of nuclease-free water and stored at -80°C. cDNA synthesis Reverse transcription of RNA samples was performed using an iScriptTM cDNA Synthesis Kit (Bio-Rad) following the manufacturer's instructions. Briefly, 1 μg of total RNA (variable volume), 4 μL of 5x iScript Reaction Mix, 1 μL of iScript Reverse Transcriptase, and nuclease-free water (variable volume) were mixed to a final volume of 20 μL. The mix was then incubated in a thermal cycler using the following protocol: Priming: 5 min at 25°C; Reverse Transcription: 20 min at 46°C; RT inactivation: 1 min at 95°C; and Hold: 4°C indefinitely. Then, the samples were stored frozen at -20°C. Bacterial fitness assays Growth rates were determined in LB broth. Briefly, overnight bacterial cultures were diluted to a starting 0.01 OD600 in 100 mL of LB broth in 250 mL sterile flasks. Kanamycin (50 μg/mL) was added for EcN ΔchuS growth. Ampicillin (100 μg/mL) was included for EcN ΔchuS pKK233.3 and EcN ΔchuS pKK233.3::chuS (EcN ΔchuS::chuS). The cultures were incubated at 37°C and 150 rpm. OD600 of each bacterial culture was measured at the beginning of incubation and every h for the first 5 h. A final OD600 measurement was performed at 24 h. Colony-forming units (CFU) counts were done in the same manner as OD600 measurements. Briefly, serial dilutions of bacterial culture were done on a 96-well plate at the previously stated time points. Dilutions were performed in 1:10 until 10-7, and 5 μL of each dilution were placed on LB agar plates by spot plating. Plates were incubated for 18 h at 37°C, and CFUs were counted. Each CFU count was performed in triplicate. In vitro carbon monoxide detection Carbon monoxide production was determined in LB broth with 0.1 mM IPTG (Sigma-Aldrich). Briefly, overnight bacterial cultures were diluted to an initial optical density at 600 nm (OD600) of 0.01 in 35 mL of supplemented LB broth in 50 mL Falcon tubes. All cultures were tightly closed and incubated at 37°C for 24 h, with no agitation. Carbon monoxide levels were measured by opening the tubes and immediately placing a stabilized electrochemical CO detector (Smart Sensor AS8700A, CO-specific electrochemical sensor) on the lid opening. The detector has a measurement range of 0–1000 ppm, an accuracy of ±5% or ±10 ppm, and a response time of <60 s. Prior to measurements, the detector was zeroed in ambient air to establish baseline calibration. CO readings were recorded using the MAX value function to capture peak concentrations immediately after tube opening. The effective detection limit was defined by the instrument resolution (10 ppm). Negative control measurements were performed using uninoculated LB medium and bacterial cultures lacking chuS expression, which consistently yielded baseline readings. The highest concentration recorded was used to compare bacterial strains and growth conditions. Accordingly, CO measurements obtained with this approach were interpreted as semi-quantitative and comparative rather than absolute. Escherichia coli probiotics administration to mice For probiotic administration, before treatment, male or female C57BL/6 mice were anesthetized with 5% isoflurane (Baxter) with 1 L of oxygen per min. Once adequate anesthesia was confirmed, a dose of 3.5 × 107 CFU in 50 μL of sterile PBS of either UCPROB, EcN, or EcN ΔchuS was administered through gavage. Control mice were treated with 50 μL PBS alone (vehicle). This treatment was performed twice a week for 4 weeks. This dose was selected by previous experiments performed in our laboratory, being 3.5 × 107 CFU the higher dose that does not alters any clinical parameter in mice. Cytokine profile evaluation by ELISA Serum levels of IL-10 and IFN-γ were quantified using commercially available ELISA kits according to the manufacturer’s instructions (BD OptEIA™ Mouse IL-10 ELISA Set and BD OptEIA™ Mouse IFN-γ ELISA Set, BD). At the end of each treatment, mice were injected intraperitoneally with a lethal dose of ketamine (300 mg/kg) and xylazine (30 mg/kg). After confirmation of adequate anesthesia, blood was collected by cardiac puncture, allowed to clot for 1 h at 37°C, and centrifuged at 4,400 × g for 15 min. Serum samples were stored at ≤ −80°C until analysis. Prior to cytokine quantification, samples were centrifuged at 16,000 × g for 4 min to remove particulate material. Standard curves were generated using recombinant standards provided with each kit following serial dilutions; the standard curve working range was 31.3–2,000 pg/mL, as specified by the manufacturer. Samples and standards were measured in duplicate, and concentrations were calculated from the standard curve. Intra-assay variability was assessed from duplicate wells within the same plate, and results were reported as the mean of duplicates. According to the manufacturer’s specifications, the expected intra-assay precision for these assays is <10% CV under recommended conditions. Absorbance was measured at 450 nm with wavelength correction at 570 nm. Carboxyhemoglobin detection Mouse serum samples were analyzed for carboxyhemoglobin (HbCO) levels using the Mouse HbCO ELISA Kit (MyBioSource) based on a double-antibody sandwich ELISA technique. Blood samples were collected from mice, allowed to clot for 24 h at 4°C, and centrifuged at 4,400 rpm for 15 min to separate the serum. The serum was collected, stored at -80°C, and thawed on ice before analysis. Reagents were prepared according to the manufacturer’s protocol. For the assay, 100 μL of standard or serum samples were loaded into the pre-coated 96-well plate wells, with blanks receiving only standard diluent. The plate was sealed and incubated at 37°C for 90 min. After two washes, 100 μL of biotinylated antibody was added to each well and incubated at 37°C for 60 min. Following three additional washes, 100 μL of enzyme conjugate was added, excluding blank wells, and the plate was incubated at 37°C for 30 min. After five washes, 100 μL of color reagent was added and incubated at 37°C for up to 30 min in the dark. The reaction was stopped by adding 100 μL of stop solution, and absorbance was measured at 450 nm using a microplate reader within 10 min. Gene expression analysis by RT–qPCR Gene expression analysis was performed by one-step reverse transcription quantitative PCR (RT–qPCR) using the TaqMan™ RNA-to-CT™ 1-Step Kit (Thermo Scientific) according to the manufacturer’s instructions. Reactions were prepared in a final volume of 10 μL per well, containing 5.0 μL of 2× TaqMan™ RT-PCR Mix, 0.25 μL of 40× TaqMan™ RT Enzyme Mix, 0.5 μL of 20× TaqMan™ Gene Expression Assay, 2.0 μL of total RNA (100 ng/μL; 200 ng per reaction), and 2.25 μL of nuclease-free water. The following TaqMan™ Gene Expression Assays were used: Gapdh (Mm99999915_g1, FAM-MGB) as the endogenous control and hmox1 (HO-1) (Mm00516005_m1, FAM-MGB) as the target gene. Each gene was analyzed in singleplex reactions. Thermal cycling was performed under standard conditions consisting of reverse transcription at 48°C for 15 min, polymerase activation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. All reactions were performed in duplicate. Relative gene expression levels were calculated using the 2-ΔΔCt method, normalizing hmox1 Ct values to gapdh. Results are presented as relative expression compared to the corresponding control group. Bacterial loads in deep organs Spleen, liver, mesenteric lymph nodes, and blood were obtained from the inoculated mice and weighed before downstream processing. Then, the samples were homogenized by a 70 nm cell strainer in sterile PBS and cultured by spot plating on LB agar (consisting of 10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, and 15 g/L agar agar) and MacConkey agar (Merck Millipore) plates. Plates were incubated overnight at 37°C, and after 18 h, the CFUs were registered. HSV-1 skin infection in mice The experimental procedure for HSV-1 infection in mice pretreated with the probiotics was conducted as previously described.53 Briefly, mice were anesthetized via intraperitoneal injection with a ketamine/xylazine solution. The right flank of each mouse was prepared by scissor shaving and then depilation with VEET® depilatory agent. The following day, the epidermis was abraded using a nail file over the region where mice were infected with the HSV-1 strain KOS at a dose of 1x106 PFU or mock-treated in 10 μL volume. Disease progression was monitored daily for 11 days, with clinical scores assigned based on the severity of zosteriform lesions: (1) weight loss, (2) erythema or primary lesion, (3) mild erythema/edema with distant zosteriform lesions, (4) ulceration and edema with increased epidermal involvement, and (5) hind limb paralysis. For more details, please see Table S2. HSV-1 viral load HSV-1 infection in mouse skin was assessed using a quantitative PCR (qPCR) to detect the viral genome. Skin samples were collected from infected mice, treated with collagenase-D (Sigma-Aldrich) for 30 min at 37°C, and disaggregated using a mortar. Total DNA was directly extracted from the homogenized skin tissue using TRIzol Reagent for qPCR analysis. 200 ng of DNA was used per reaction to quantify HSV-1 viral load, targeting the UL30 gene as previously described54 and using the primers and probe listed in Table 1. The qPCR was performed on an Applied Biosystems StepOnePlus thermocycler. Quantification and statistical analysis Statistical analysis and Software All statistical analyses were performed using GraphPad Prism 10 (GraphPad Software). Data are presented as mean ± SD, as indicated in the corresponding figure legends. One-way or two-way ANOVA was used as appropriate, depending on the number of variables evaluated, followed by Tukey’s multiple comparisons test. The number of biological replicates (n) and independent experiments are specified in each figure legend. A p value ≤ 0.05 was considered statistically significant. Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.isci.2026.116042. Supplemental information Document S1. Figures S1–S4 and Tables S1 and S2 References Associated Data Supplementary Materials Document S1. Figures S1–S4 and Tables S1 and S2 Data Availability Statement Whole genome sequencing data generated in this study have been deposited in GenBank and are publicly available as of the date of publication. Accession numbers are listed in the key resources table (PRJNA1357318 and PRJNA1357320). Reference genomes of Escherichia coli Nissle 1917 and Escherichia coli O157:H7 (EHEC) were obtained from GenBank under accession numbers CP007799.1 and CP008957.1, respectively. Additional data supporting the findings of this study are available from the lead contact upon reasonable request. This study does not report original code.