Low-dose cephaloridine induces CA beyond temperature restriction to prolong life span

Colanic acid (CA) is an extracellular polysaccharide synthesized in E. coli and other Enterobacteriaceae [17]. Its biosynthesis is catalyzed by a series of enzymes encoded in the capsular polysaccharide synthesis (cps) operon [18]. Our previous studies have revealed the longevity-promoting effect of CA in C. elegans and Drosophila melanogaster [19,20]. We also showed that the optogenetic induction of the cps operon is sufficient to prolong C. elegans life span [20]. To examine the relevance of the cps operon in the human gut microbiome, we have searched the Unified Human Gastrointestinal Genome (UHGG) [21] using DIAMOND [22]. In this database, 47 out of 4,744 species (approximately 0.99%) contain at least one genome that carries the cps operon, corresponding to 5,578 genomes out of 289,232 total genomes. These include 4,598 in Escherichia, 466 in Salmonella, 341 in Enterobacter, 147 in Citrobacter, and 26 in other species (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Chemical induction of bacterial colanic acid (CA) promotes longevity.

(A) Phylogenetic tree of the cps operon from 5,578 genomes across 47 species within the gut microbiota. Tree leaves are colored by genus, with the legend on the right showing the number of genomes for each genus. The largest clade, comprising 4,577 E. coli genomes, is collapsed into a red triangle. In addition to Escherichia, dominant clades from Salmonella, Enterobacter, and Citrobacter are highlighted in their respective genus-specific colors. Also see the tree file in S2 Data. (B) Comparison of the percentage of reads mapping to the cps operon between centenarians and elderly individuals in the Japanese cohort (PRJNA675598). The p value was obtained using the Welch’s t test. (C) Same as (B), but showing the comparison in the Italian cohort (PRJEB25514). (D) The disk-diffusion assay shows E. coli treated by chemical compounds presenting various blue colorations, indicating the induction of the cps operon responsible for CA biosynthesis. Ctrl (LB medium), Cepha (cephaloridine), CephT (cephalothin), Cefaz (Cefazolin), Cefu (cefuroxime), PenG (benzylpenicillin), Amp (ampicillin), Carb (carbenicillin), Col (colistin), Cipro (ciprofloxacin), Azit (Azithromycin), Tet (tetracycline), and Chlor (chloramphenicol); N = 2. (E) Quantification of CA from filtered bacterial cultures shows the lon deletion mutant E. coli (Δlon) produces an increased level of CA by 14-fold as compared to the wild-type (WT) BW25113 E. coli. Cepha at 1.8 μg/mL and PenG at 5 μg/mL induce CA production by 7-fold and 5-fold, respectively; CephT at 0.625 μg/mL and Amp at 1.5 μg/mL do not induce CA significantly (p > 0.05). N ≥ 3 (F) Quantification of the relative CA induction in WT BW25113 E. coli when responding to 0, 0.45, 0.9, 1.8, and 3.6 μg/mL Cepha. A low dose of cephaloridine at 1.8 μg/mL is defined as the Colanic Acid-Inducing dose (Cepha-CAI dose), N ≥ 3. (G) Bacterial growth curves for WT E. coli with the vehicle and Cepha-CAI dose treatments and the Δlon mutant. WT E. coli treated by Cepha-CAI dose and the Δlon mutant E. coli increase CA levels during the log phase (solid lines) and do not affect bacterial growth (dotted lines). N = 2 biologically independent replicates. (H) Measurement of carbon-deuterium levels in WT E. coli cultured in M9 medium supplemented with deuterium-labeled glucose using stimulated Raman scattering (SRS) microscopy at 2180 cm⁻¹, indicating no metabolic suppression by the Cepha-CAI dose. Independent biological replicates are shown in S1 Fig. (I) Representative SRS images of WT E. coli cultured with deuterium-labeled glucose, scale bar = 1 µm. (J) The life span of C. elegans is extended by 14% with Cepha-CAI dose-treated WT BW25113 E. coli. N = 3, 60–100 worms per replicate. In (E) and (F), error bars represent the mean ± standard error of the mean (s.e.m.). **p < 0.01, *p < 0.05, ns p > 0.05 by two-tailed Student t test. In (J), ***p < 0.001 by log-rank test, also seen in S1 Table. In (H), Error bars represent the mean ± standard deviation (s.d.), ns p > 0.05 by Two-way ANOVA. (B, C, E, F, G, H, J) Source data available in S1 Data.

https://doi.org/10.1371/journal.pbio.3002749.g001

Next, we analyzed two publicly available metagenomic datasets of human centenarian gut microbiome (PRJNA675598, PRJEB25514) [23,24]. A recent study reported that members of the Enterobacteriaceae family typically comprise less than 1% of the healthy human gut microbiota [25]. In our analysis, Enterobacteriaceae accounted for approximately 1.43% of total reads in the Italian cohort (PRJEB25514) and 3.02% in the Japanese cohort (PRJNA675598). We further compared the presence of the cps operon between centenarians and their elderly controls in these two cohorts. On average, 9.16% of total reads in the Japanese cohort and 7.25% in the Italian cohort mapped to the representative cps-positive genomes from 47 identified species.

In the Japanese cohort, which includes over 100 individuals per group, we observed a statistically significant difference in cps operon abundance between centenarians and controls (Fig 1B, Welch’s t test, p = 0.0071), with a small-to-moderate effect size (Cohen’s d = 0.37). In the Italian cohort, which includes approximately 20 individuals per group, a statistically significant difference was also observed (Fig 1C, Welch’s t test, p = 0.0436), with a moderate effect size (Cohen’s d = 0.64). Notably, a post hoc power analysis based on this effect size suggests that approximately 39 individuals per group would be required to achieve 80% power at α = 0.05, indicating that the Italian dataset may have been underpowered in the original analysis. Despite this limitation, these findings collectively support a small-to-modest increase of cps operon abundance in centenarian individuals. Unfortunately, neither cohort studies included information on individual diet, medication use, or lifestyle factors, preventing us from controlling these potential confounding variables. Moreover, this association analysis cannot clarity whether the increased cps operon abundance contributes to longevity or is simply a consequence of advanced age in centenarians. Nevertheless, these observations motivated us to investigate whether the cps operon can be induced in human gut-derived commensal bacteria.

Escherichia represents the largest genus carrying the cps operon, and E. coli is known to be among the first commensal bacteria to inhabit the human gut [26]. However, CA production from the E. coli cps operon becomes restricted when the environmental temperature exceeds 30 °C [27]. A previous study indicated that certain β-lactam antibiotics trigger the overexpression of the cps operon in E. coli at sub-minimum inhibitory concentrations (sub-MIC) at 37 °C [28]. We thus utilized an E. coli strain that harbors a LacZ reporter for the cps operon transcriptional regulation to screen multiple β-lactam antibiotics and antibiotics that disrupt the outer membrane, protein synthesis, and genome replication. Using a colorimetric assay, we examined the levels of β-galactosidase from the transcriptional up-regulation of the cps operon, resulting in intensified blue coloration. We confirmed that cephaloridine (Cepha) and cefazolin (Cafez) strongly induce the cps operon, while cephalothin (CephT), cefuroxime (Cefu), and penicillins—benzylpenicillin (PenG), ampicillin (Amp), and carbenicillin (Carb) show various levels of moderate effects based on the blue coloration (Figs 1D and S1A).

Next, we quantified CA production in the bacteria culture medium upon the treatment of different antibiotics. The medium from E. coli treated with cephaloridine at sub-MIC showed the highest CA level, while penicillin and cephalothin showed lower or no increase (Fig 1E). Furthermore, we found that the most effective CA-inducing concentration of cephaloridine was 1.8 μg/mL (Fig 1F). Unlike the growth delay commonly observed with sub-MIC antibiotic exposure [29], wild-type E. coli treated with 1.8 μg/mL cephaloridine exhibited growth rates comparable to untreated controls and to CA-overproducing Δlon mutants [19] (Fig 1G).

We referred to this low dose of cephaloridine as the Colanic Acid-Inducing dose (Cepha-CAI dose).

We further examined whether the treatment at the Cepha-CAI dose affects E. coli metabolic activity. To this end, we have applied stimulated Raman scattering (SRS) microscopy to monitor the incorporation of deuterium-labeled glucose into individual bacteria, enabling assessment of their metabolic activity and viability at single-cell resolution. Compared to the control, the Cepha-CAI treatment did not alter glucose incorporation during the 1–6 h growth window (Figs 1H, 1I, and S1B). Together, these results demonstrate that the Cepha-CAI dose overrides the temperature restriction of CA production at 37 °C without inhibiting E. coli growth or causing cell inviability.

We further revealed that, similar to the Δlon mutant, wild-type E. coli treated by Cepha-CAI dose showed increased CA production during the log phase, and the elevated levels persisted into the stationary phase (Fig 1G). Importantly, wild-type C. elegans with E. coli treated with the Cepha-CAI dose showed a 14% extension in life span compared to the controls without the treatment (Fig 1J and S1 Table). This demonstrates that chemically induced CA effectively confers longevity-promoting benefits in the host C. elegans.

CA-induction by low-dose cephaloridine improves metabolic health during mouse aging

Due to its poor oral absorption, cephaloridine as an antibiotic agent was terminated in the clinic [30]. Even at a concentration of 1 mM, cephaloridine cannot be sufficiently absorbed by the GI tract due to a high efflux rate [30]. Since we successfully induced CA using the Cepha-CAI dose that bypassed the temperature constraint at 37 °C, we hypothesized that cephaloridine would specifically target commensal E. coli and stimulate CA production in the mammalian gut, where the average body temperature ranges from 36 °C to 40 °C. To test this hypothesis, we first treated MG1655, a K-12 E. coli strain derived from human microbiota, with the Cepha-CAI dose at 37 °C and detected increased CA levels in the culture medium (Fig 2A). Furthermore, wild-type C. elegans hosting the Cepha-CAI dose-treated MG1655 showed a 13% life span extension (Fig 2B and S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. A low dose of cephaloridine up-regulates colanic acid (CA) biosynthesis operon in the mouse gut microbiota.

(A) Cepha-CAI dose-treated MG1655 E. coli showed increased CA levels. N > 3. (B) Cepha-CAI dose-treated MG1655 E. coli increases C. elegans life span by 13.3%. N = 3, 60–100 worms per replicate. (C) Schematic diagram showing the analysis using the cps G-R reporter E. coli strain. (D) Representative ratio matric images of E. coli treated with vehicle or Cepha-CAI dose, scale bar = 5 µm. (E) Compared to the vehicle-treated control, Cepha-CAI dose-treated cps G-R reporter strain increase the GFP-to-RFP ratio from 0.87 to 1.05 (0.18 increase from the baseline). (F) Schematic diagram presenting the drug administration procedure in wild-type B6 mice. (G) Compared to the vehicle control, cephaloridine increases the GFP-to-RFP ratio of the cps G-R reporter E. coli harvested from the mouse gut from 2.01 to 2.16 at 2 μg/mL (orange) and from 2.19 to 2.41 at 4 μg/mL (dark blue), with 0.15 and 0.22 increases from the baseline, respectively. The cephaloridine at 8 μg/mL (light blue) and tetracycline treatment at 10 μg/mL (yellow) made no significant changes on the GFP-to-RFP ratio. (H) Plasma levels of low-density lipoprotein (LDL) increase by 2-fold between 12 and 18 months of age in control male mice, but remain nearly unchanged in mice treated orally with cephaloridine (2 µg/mL). (I) Between 12 and 18 months of age in female mice, the increase in plasma insulin levels is lower in mice treated orally with cephaloridine (2 µg/mL) compared to controls. (J) The age-related increase in the LDL/HDL ratio is suppressed by the cephaloridine treatment in male mice. In (A), error bars represent the mean ± s.e.m., * p < 0.05 by Student t test. In (B), ***p < 0.001 by log-rank test, also seen in S1 Table. In (E) and (G), black bars show the median of the data set, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns p > 0.05 by Student t test. In (H–J), error bars represent the mean ± s.e.m., **p < 0.01, *p < 0.05, ns, p > 0.05 by t test with Welch’s correction. (A, B, E, G, H–J) Source data available in S1 Data.

https://doi.org/10.1371/journal.pbio.3002749.g002

To monitor the induction of CA biosynthesis within the mouse gut, we genetically engineered the MG1655 strain to express GFP under the control of the cps promoter and RFP under a constitutive promoter syn135 [31] (cpsG-R reporter, Fig 2C). In this strain, constitutively expressed RFP indicates the baseline transcription, while inducible GFP reflects the transcriptional level of the cps operon. We found that the Cepha-CAI dose increases the GFP-to-RFP intensity ratio from 0.87 to 1.05 in the cpsG-R reporter strain at 37 °C (Fig 2D and 2E). Next, we introduced this cpsG-R reporter strain to wild-type C57B6 (B6) mice that were raised in a specific pathogen-free environment through drinking water to examine chemical induction in the murine gut. After supplementing cpsG-R reporter E. coli through drinking water for 3 days, we administered cephaloridine or other drugs to mice in a single low dose through oral gavage, followed by supplementing drug-containing water for 3 h (Fig 2F), accounting and allowing for sufficient intestinal transit time [32] and E. coli turnover time [33].

Leveraging the minimal intestinal absorption and utilizing low doses of cephaloridine (2 μg/mL or 4.8 μM, 4 μg/mL or 9.6 μM, and 8 μg/mL or 19.2 μM), we examined its effects on gut bacteria in inducing the cps operon. By imaging the RFP-labeled E. coli harvested from the luminal content of mouse ileum, we observed increased GFP-to-RFP ratios in mice treated with 2 μg/mL (2.01 to 2.16) and 4 μg/mL (2.19 to 2.41) cephaloridine (Figs 2G and S2A). This increase was not observed in mice treated with 8 μg/mL cephaloridine (2.19 to 2.14) (Figs 2G and S2A). In the supplement group with tetracycline (10 μg/mL) that showed no CA induction (Fig 1D), we observed that the GFP-to-RFP ratio was not increased (2.19 to 2.08) (Figs 2G and S2A). These findings suggest that a low dose of cephaloridine effectively stimulates the cps operon of E. coli in the mammalian gut microbiota.

Next, we treated one-year-old wild-type mice with 2 μg/mL cephaloridine through drinking water for 6 months (S2B Fig) and profiled a series of metabolic parameters in plasma samples (S2C–S2K Fig). We found that low-density lipoprotein (LDL) levels were increased in 18-month-old male mice (Figs 2H and S2C), whereas insulin levels were elevated in 18-month-old female mice (Figs 2I and S2H). Notably, these age-related increases were attenuated by the cephaloridine treatment (Fig 2H and 2I). Furthermore, the cephaloridine treatment did not affect high-density lipoprotein (HDL) levels (S2D Fig), thereby leading to a reduction in the LDL/HDL ratio (Fig 2J). Together, these findings suggest that cephaloridine-induced CA production mitigates age-related metabolic dysfunction in mice.

We also collected fecal samples, and metagenomic analysis of these samples revealed that microbiome richness was not reduced by the cephaloridine treatment (S2L–S2N Fig), supporting the idea that low-dose cephaloridine does not cause bacterial lethality in the microbiota.

Chemical induction of colanic acid independent of RcsC-RcsD

Next, we investigated the molecular mechanism underlying the CA induction by the Cepha-CAI dose. At nonrestricted temperatures below 30 °C, the induction of CA biosynthesis is known to be mediated by the RCS two-component system (TCS) [18]. This system involves the RcsC sensor histidine kinase situated in the inner membrane, which autophosphorylates in response to environmental cues. The RcsD phosphotransferase interacts with RcsC to transmit the phosphoryl group to the cytosolic response regulator, RcsB (Fig 3A). RcsA, interacting with RcsB, enhances the heterodimers’ binding with the promoter and functions as transcription factor together to up-regulate the cps operon, thereby stimulating transcriptional activation (Fig 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Chemical induction of bacterial colanic acid (CA) acts independently of RCS inner membrane sensors.

(A) A schematic diagram of the canonical RCS activation to induce the cps operon and CA biosynthesis. (B) Treatments with Cepha-CAI dose increase CA levels in WT, ∆rcsC, and ∆rcsD mutant E. coli but not in the ∆rcsA or ∆rcsB mutant E. coli. N > 3. (C) Cepha-CAI dose-treated—∆rcsA mutant E. coli fails to increase C. elegans life span. N = 3, 60–100 worms per replicate. (D) Cepha-CAI dose-treated—∆rcsD mutant E. coli increases C. elegans life span by 30%. N = 3, 60-100 worms per replicate. In (B), error bars represent mean ± s.e.m., **p < 0.01, * p < 0.05, ns p > 0.05, by two-tailed Student t test. In (C) and (D), **** p < 0.0001, ns p > 0.05 by log-rank test, also seen in S1 Table. (B, C, D) Source data available in S1 Data.

https://doi.org/10.1371/journal.pbio.3002749.g003

To assess the necessity of the RCS system in the chemical induction of CA at 37 °C, we treated the E. coli deletion mutants of rcsA, rcsB, rcsC, and rcsD with the Cepha-CAI dose. By measuring CA levels in their respective culture mediums, we found that deleting either rcsA or rcsB completely abolished the CA induction by the Cepha-CAI dose (Fig 3B). To our surprise, deleting either rcsC or rcsD did not suppress the induction (Fig 3B). These results suggest that cephaloridine relies on the RcsA-RcsB transcriptional control of the cps operon but operates independently of the RcsC-RcsD TCS on the bacterial inner membrane.

Further investigation of the host life span in C. elegans revealed that, unlike cephaloridine-treated wild-type E. coli (Fig 1J), the cephaloridine-treated rcsA deletion mutant failed to extend the host’s life span (Fig 3C and S1 Table). In contrast, the cephaloridine-treated rcsD deletion mutant remained effective in causing life span extension (Fig 3D and S1 Table). These results further support that the chemical induction of CA in bacteria underlies the pro-longevity effect of low-dose cephaloridine on the host and suggest that the RcsC-RcsD inner membrane kinase complex is dispensable for the cephaloridine-mediated induction of pro-longevity CA.

PBP1a and ZraS mediate the chemical induction of CA

We then hypothesized that the Cepha-CAI dose acts through other bacterial inner membrane proteins. Considering β-lactam antibiotics show preferential affinities to various penicillin-binding proteins (PBPs) [34,35], we initially explored the role of these inner membrane proteins in regulating the chemical induction of CA. In E. coli, cephaloridine, cefazolin, and cephalothin primarily bind to PBP1a, a protein encoded by mrcA [36]. This contrasts with penicillin and ampicillin, which exhibit a higher affinity for PBP4, a protein encoded by DacB [35] (Fig 4A). Notably, the Cepha-CAI dose failed to induce CA biosynthesis in the mrcA deletion mutant (∆PBP1a), but this dose remained effective in the dacB deletion mutant (∆PBP4) (Fig 4B). Thus, low-dose cephaloridine specifically acts through PBP1a to induce CA. These findings also support that the inducing effect of the Cepha-CAI dose on CA biosynthesis is distinct from the general β-lactam antibiotic effect mediated through any PBPs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. PBP1a and histidine kinase ZraS mediate colanic acid (CA) chemical induction.

(A) A schematic diagram of PBP protein binding preference of different antibiotics. (B) The loss of PBP1a encoded by mrcA in E. coli suppresses CA induction upon the Cepha-CAI dose treatment. N > 3. (C) A volcano plot summarizing differentially expressed genes between the Cepha-CAI dose- and vehicle-treated E. coli. (Fold change > 2, FDR adjusted p value < 0.05). (D) Known target genes regulated by the RcsB, BtsSR, ZraSR, GlrKR, and RstBA two-component systems are enriched among the genes differentially expressed in the Cepha-CAI dose-treated group. Differentially expressed genes with log₂ fold change > 0.5 in red and with log₂ fold change < - 0.5 in blue, * p < 0.05; genes with - 0.5 < log₂ fold change < 0.5 or undetectable in gray. (E) The loss of zraS increases the CA level in the absence of cephaloridine. The Cepha-CAI dose treatment does not further increase the CA level in the zraS deletion mutant, but remains effective in either btsS or glrK deletion mutant. N > 3. (F) Illustration of the proposed model. At 37 °C, ZraS inhibits RcsAB-mediated transcription of the cps operon. Upon cephaloridine interacting with PBP1a, it acts through the activated ZraS to up-regulate RcsAB-mediated cps transcription. Ultimately, this activation promotes the production of longevity-promoting CA at 37 °C. In (B) and (E), error bars represent mean ± s.e.m., **p < 0.01, *p < 0.05, ns p > 0.05 by two-tailed Student t test. In (C), DESeq2 |log₂ fold change| ≥ 0.5; p < 0.05 by two-sided Wald test (cephaloridine vs. vehicle). (B, E) Source data available in S1 Data.

https://doi.org/10.1371/journal.pbio.3002749.g004

In parallel, we conducted comprehensive transcriptional profiling using wild-type MG1655 E. coli in response to the Cepha-CAI dose. Using RNA-seq analysis, we identified genes with significant differential expression between the Cepha-CAI dose-treated E. coli and the untreated control (FDR-adjusted p value < 0.05, Fig 4C). Among these genes, we observed an enrichment of RcsB target genes (Fig 4D), confirming the pivotal role of the RcsB transcription factor in the cephaloridine-mediated induction of CA. Furthermore, integrating information from transcription factor and TCS databases from RegulonDB led to the identification of four additional TCSs—BtsS-BtsR, RstB-RstA, GlrK-GlrR, and ZraS-ZraR—whose target genes exhibited enrichments among the differentially expressed candidate genes in response to the Cepha-CAI dose (Fig 4D).

To investigate the potential involvement of these TCSs, we treated the following single-gene deletion mutant E. coli of btsS, glrK, and zraS with the Cepha-CAI dose at 37 °C, and then measured the CA levels in their respective culture media. We found that the CA level in the zraS deletion mutant is higher than wild-type E. coli in the absence of the cephaloridine treatment (Fig 4E). This finding suggests that ZraS normally suppresses CA biosynthesis under baseline conditions. Furthermore, treatment with the Cepha-CAI dose did not further elevate the CA level in the zraS mutant (Fig 4E), indicating that ZraS is required for mediating the CA-inducing effect of cephaloridine. In contrast, deletion of either btsS or glrK did not affect CA induction by the Cepha-CAI dose (Fig 4E). The rstB deletion mutant could not be evaluated in this analysis due to its impaired growth in M9 medium, even when supplemented with both carbon and nitrogen sources. Together, these results suggest ZraS functions as the inner membrane kinase that typically inhibits CA biosynthesis at 37 °C; however, exposure to low-dose cephaloridine appears to convert ZraS from an inhibitor into an activator of CA production.

In parallel, we have systematically examined the expression levels of genes from the β-lactam resistance pathway reported by KEGG pathway analysis. We found that only ompC, which encodes a general outer membrane porin mediating the nonspecific diffusion of small solutes, and oppA, which encodes the periplasmic binding protein mediating the transport of oligopeptides, are differentially up-regulated and down-regulated, respectively, in response to the Cepha-CAI dose (S3 Fig). These results suggest that the Cepha-CAI dose does not stimulate general β-lactam antibiotic stress response, which further supports that the observed induction of CA is not a secondary consequence of general antibiotic-mediated growth inhibition, but acts through a specific signaling mechanism.

Together, we propose a model (Fig 4F) that ZraS inhibits RcsAB at 37 °C, thereby silencing the cps operon at this temperature. Upon binding of cephaloridine to PBP1a, the activation of ZraS, which is likely associated with its autophosphorylation, triggers the transcriptional up-regulation of the cps operon through RcsAB and, in turn, the production of longevity-promoting CA at 37 °C.

Ethics statement

All animal procedures were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. For short-term Cephaloridine treatment, the protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine, under protocol number (Protocol #AN-6975). For long-term Cephaloridine treatment, the protocol was approved by the IACUC of the Janelia Research Campus, Howard Hughes Medical Institute (Protocol #25-276). Every effort was made to minimize animal suffering and reduce the number of animals used.

Bacteria strains

For antibiotic screening targeting cps transcriptional expression, we utilized the SG20781 strain (MC4100 lon+cpsB10::lacZ Mu-immλ) originally generated by S. Gottesman. The bacteria were cultured overnight at 37 °C in Luria–Broth (LB) medium and evenly plated on X-gal (40 μg/mL) supplemented LB agar plates, incubated overnight.

For CA quantification, E. coli Keio mutants and their parental strain BW25113 were cultured overnight at 37 °C in M9 minimum medium [40]. Bacteria were then removed by centrifugation and filtration, and the culture medium was kept for CA measurement.

For longitudinal assays, E. coli Keio mutants were cultivated in M9 minimum medium at 37 °C for 16 h. Subsequently, 10^7 bacteria cells from the stationary phases were reinoculated in every 50 mL fresh M9 medium with a low dose of cephaloridine at 37 °C for another 16 h as a seeding bacterial culture. Fresh seeding bacterial cultures were made weekly. A 200 μL aliquot of the seeding bacteria culture was plated onto each 6-cm nematode growth medium (NGM) plate and maintained at 20 °C to transfer live worms every other day.

For high-resolution imaging, rph1 ilvG rfb-50 attL(Psyn135::mcherry <FRT>) attHKPR218 (cps G-R reporter strain), a genetically modified strain derived from MG1655 (genotype rph1 ilvG rfb-50), was generated by Dr. Patricia Rohs. pPR218, integrated at the HK site, encodes Pcps::sfGFP and chloramphenicol resistance.

C. elegans strain

C. elegans var. Bristol N2 strain was obtained from the Caenorhabditis Genome Center. The strain was maintained on standard NGM agar plates seeded with corresponding bacteria at 20 °C. Age synchronization of C. elegans was achieved by isolating eggs as previously described [41].

Disk-diffusion assay

E. coli strain SG20781 was used to assess antibiotic susceptibility conducted via disk diffusion, previously described by the Sailer group [28]. In brief, a 16-h overnight culture at 37 °C of bacteria in LB medium was spread onto LB agar containing 0.1 mg/mL 5-bromo-4-chloro-3-indolyl-β-D-galactoside (X-gal). Filter papers were saturated by 100 µL of the following antibiotic solutions (cephaloridine C258600 Toronto Research Chemicals; Cephalothin, Sigma; Cefazolin, Sigma; Cefuroxime, Sigma; penicillin, Sigma; Ampicillin, Sigma; Carbenicillin, Sigma; Colistin, Sigma; Ciprofloxacin, Sigma; Azithromycin, Sigma; Tetracycline, Sigma; Chloramphenicol, Sigma) at concentrations of 10 µg/mL and air-dried. The soaked filter papers were placed onto the bacteria-seeding plates and incubated for 24 h at 37 °C.

Colanic acid measurement

The desired bacterial strain was first inoculated in M9 minimum medium overnight at 37 °C as the seeding culture. The OD 600 was measured, and 10^7 cells were reinoculated in 50 mL M9 minimum medium or antibiotic-containing M9 medium for another 16 h. After the incubation, bacterial cultures were spun down at 4,000g at 4 °C for 30 min. The bacterial pellet was removed, and the culture medium was filtered through a 0.2 μm syringe filter (Corning). Each 25 mL of supernatant was collected into a 50 mL conical tube, and then an equal volume of ice-cold 100% ethanol (EtOH) was used to make a final 50% EtOH for precipitation. The mixture was kept at 4 °C overnight. Once precipitation occurred, the liquid was carefully removed after centrifugation at 4,000g at 4 °C for 50 min. The pellet was washed once with cold 80% EtOH and then air-dried in a hood. The pellet from every 50 mL culture was resuspended in 500 μL distilled H₂O. The solution was sonicated in a water bath for 60 min at 37 °C, followed by centrifugation at 4,000g at 4 °C for 10 min. A 200 μL sample was boiled with 30 μL hydrochloric acid (320,331-500mL, Sigma) for 2 h. Subsequently, 60 μL 5M NaOH and 150 μL NaHCO₃/NaOH buffer pH 10 were added to reach pH 6–8, calibrated by 1M HCl or NaOH solutions. The fucose quantification assay was then carried out using a K-FUCOSE kit (Megazyme).

Lifespan assay

Lifespan assays were performed as previously described [42]. In brief, age-synchronized C. elegans at the L1 stage were seeded onto designated bacteria lawns. Fresh bacteria-containing plates were made every other day for live C. elegans transfer. Once reaching adulthood as day 0, worms were transferred to new bacteria plates. Death events were scored upon each transfer. Bagged worms and vulva protruding worms were censored through the analysis. Each assay contained 80–100 animals with 30–40 animals per 6 cm plate.

High-resolution bacteria imaging

SRS imaging of D-glucose incorporation

For SRS imaging [45], E. coli MG1655 cells were cultured overnight in M9 minimum medium at 37 °C. The overnight culture was then diluted 1:100 into 5 mL of pre-warmed M9 medium supplemented with 2% glucose-d₇ (Sigma-Aldrich 552003). Following incubation at 37 °C for 1, 2, 3, 4, or 6 h, 1 mL samples were collected, centrifuged, and washed twice with purified water. Approximately 6 µL of the resulting pellet was retained, resuspended, and transferred onto a microscope slide prepared with an agarose gel pad. After drying for about 2 min, a coverslip was placed on top, and the sample was ready for imaging.

SRS imaging was carried out using a custom-built SRS microscope, integrating an APE picoEmerald laser with an Olympus FV3000 confocal microscope, similar to our previously reported setup [46]. The system was carefully optimized to enhance sensitivity for the weak C-D Raman signal, enabling visualization of glucose-d₇ labeled bacteria. All SRS imaging was performed using a 100×/1.35 silicone oil immersion objective (Olympus UPlanSApo 100×). The pump laser was tuned to 842.2 nm and paired with a 1031.7 nm Stokes beam to excite the C-D vibration mode at 2180 cm⁻¹. The MG1655 cells labeled with glucose-d₇ were imaged at 2180 cm⁻¹ after various incubation times, with both control and cephaloridine-treated groups included for comparison.

Bacterial transcriptome analysis

Bacterial cells were cultured in M9 minimal media from an initial optical density (OD₆₀₀) of 0.03 to mid-exponential phase at an OD₆₀₀ of 0.6. 1 mL of each culture was pelleted, and RNA was extracted using RNAsnap [47] and column-purified on Zymo Clean and Concentrator columns with an off-column Dnase I digestion step. Total mRNA was measured using Qubit, and the quality of RNA samples was assessed (RINe for all samples ranged from 7.9 to 8.2). Ribosomal depletion was performed with the Ribominus Transcriptome Isolation Kit. The RNA libraries were prepared for Illumina sequencing using the Illumina stranded total RNA prep kit (Catalog No. 20040525). Libraries were sequenced on an Illumina NextSeq 550 platform with 2 × 37-cycle paired-end reads. Bcl2fastq2 v2.20.0.422 was used to generate raw.fastq files, which were then filtered using FastP v0.12.4 -g –poly_g_min_len 11 -l 25. TPM calculations were done on filtered.fastq’s using Salmon v1.10.0 –seqBias –gcBias –allowDoveTails to the E. coli MG1655 genome (NC_000913.3) with an index Kmer size of 17. DEG analysis was done on count tables using Deseq2 v1.36.0, with size factors being estimated using estimateSizeFactors() and significance assessed with nbinomWaldTest(). DEGs were visualized with pheatmap v1.0.12 and EnhancedVolcano v1.14.0 for heatmaps and volcano plots, respectively.

Identification of cps operons in the gut microbiome

1. 1. Searching cps operons in gut microbial genomes

The Unified Human Gastrointestinal Genome (UHGG) collection (v2.0.2) [21], comprising 289,232 nonredundant genomes from 4,744 gut prokaryotes, was downloaded from (https://www.ebi.ac.uk/metagenomics). Coding DNA sequences were extracted from the GFF files using (v0.12.8) [48] and translated into protein sequences, resulting in a redundant protein catalog containing 630,333,479 sequences. The amino acid sequences of 19 genes comprising the cps operon [49] were used as queries to search the protein catalog using the “blastx” subcommand of DIAMOND (v2.1.10.164) [22] with an E-value threshold of 0.001. A total of 5,578 genomes from 47 species were found to encode all 19 cps genes. Protein homologs of these genes were extracted from each genome and aligned using MAFFT (v7.526) [50]. The resulting 19 multiple sequence alignments (MSAs) were trimmed with trimAl (v1.5. rev0) [51] and concatenated using the “concat” subcommand of SeqKit (v2.9.0) [52]. The concatenated MSA was used to construct a phylogenetic tree in IQ-TREE (v2.3.6) [53] with the parameters “-bb 1000 -alrt 1000 –asr.” The tree was visualized using the ggtreeExtra R package (v1.6.1) [54].

1. 2. Searching cps operons in gut metagenomes

Two publicly available metagenomic datasets of centenarians were obtained from the European Nucleotide Archive (ENA) under accession numbers PRJNA675598 and PRJEB25514. The first dataset, from a Japanese cohort (N = 265), included 119 centenarians, 107 elderly, and 39 young individuals [23]. The second dataset, from an Italian cohort (N = 59), included 19 centenarians, 23 elderly, and 17 young individuals [24]. Metagenomic reads were aligned to the cps operon using BWA (v0.7.18) [arXiv:1303.3997v2]. SAM files were converted to BAM format, and the number of mapped reads was quantified using Samtools (v1.21) [55].

Metagenomic sequencing

Microbial genomic DNA was extracted from 48 mice fecal pellets using the ZymoBiomics DNm Mini Prep kit (p/n D4300), with bead beating performed using the tube adapter from Zymo for the Vortex Genie 2 (p/n S5001-7), along with a 75 µL aliquot of the ZymoBIOMICS Microbial Community Standard (p/n D6300) as a positive control, and a 75 µL aliquot of Zymo Shield (p/n R1100) as a negative control. DNA were quantified using a Qubit 3.0 fluorometer (p/n Q33216), broad-range DNA assay (p/n Q32853) from Thermo Fisher. Samples were prepared for sequencing using 1 ng gDNA inputs into the Illumina Nextera XT library preparation kit following the manufacturer’s recommendations. Resulting libraries were quantified using the Qubit 3.0 fluorometer high-sensitivity DNA assay (p/n Q32854) and size-qualified using the Agilent Bioanalyzer high-sensitivity DNA assay (p/n 5067−4626). Libraries were equimolarly pooled and then sequenced using a NextSeq 2000 P3-300 XLEAP kit (p/n 20100988).

Metagenomic data analyis

1. 1. Data processing and taxonomic profiling

Raw sequencing data in FASTQ format were processed using Taxprofiler (v1.0.0), a reproducible workflow for taxonomic profiling of metagenomic samples (https://www.biorxiv.org/content/10.1101/2023.10.20.563221v1; https://github.com/nf-core/taxprofiler). This pipeline facilitates standardized and efficient processing of metagenomic data through an automated workflow that integrates quality control, preprocessing, and multiple taxonomic classification tools.

For taxonomic classification, we employed MetaPhlAn4 [56], which uses clade-specific marker genes to identify and quantify microbial species present in the samples. This approach enables accurate resolution of bacterial, archaeal, viral, and eukaryotic microbes at species-level taxonomic resolution. The resulting taxonomic profiles were organized into abundance tables for downstream analysis.

1. 2. Microbiome diversity analysis

The taxonomic abundance data were analyzed using the phyloseq package [57] in R. Alpha diversity metrics were calculated using the vegan package(https://vegandevs.github.io/vegan/), which is integrated with phyloseq for comprehensive microbiome analysis.

From the taxonomic profiles, we calculated three alpha diversity metrics to assess the diversity within each sample:

Shannon Diversity Index: Calculated using the vegan::diversity function, this composite measure accounts for both species richness and evenness, with higher values indicating greater diversity.

Pielou’s Evenness: Derived by dividing the Shannon diversity index by the natural logarithm of species richness (Shannon/log(S)), this metric quantifies how evenly individuals are distributed among different species, with values ranging from 0 (complete dominance by a single taxon) to 1 (perfect evenness).

Richness: Determined using vegan::estimateR, this represents the observed number of unique taxa in each sample.

1. 3. Statistical analysis

To evaluate the effects of experimental conditions on microbial diversity during analysis, mice were assigned to either Control or Experimental groups, with measurements taken at both pretreatment and posttreatment timepoints. This created four distinct groups for comparison: Control-Pre treatment, Control-Post treatment, Experimental-Pre treatment, and Experimental-Post treatment. We performed Tukey HSD (Honest Significant Difference) tests following analysis of variance (ANOVA) to evaluate differences across all four treatment groups, with appropriate corrections for multiple comparisons.

To assess differences in cps operon abundance between centenarians and elderly controls, Welch’s t tests were used to compare log₁₀-transformed cps abundance between centenarians and elderly controls, following confirmation of normality (Shapiro–Wilk test p > 0.05). Effect sizes were quantified using Cohen’s d, and statistical power was evaluated by estimating the required sample size to detect the observed effect size with 80% power at a significance level of 0.05.

Long-term cephaloridine studies in mice

1. 1. Treatment procedure

C57BL/6J mice (Strain #000664) were obtained from The Jackson Laboratory and maintained on PicoLab Rodent Diet 20 (LabDiet, 5053). Cephaloridine (2 μg/mL cephaloridine C258600 Toronto Research Chemicals) and E. coli (1 × 10⁸ CFU/mL, MG1655) were administered via drinking water. The control group followed a weekly schedule consisting of 24 h on E. coli-supplemented water, 54 hours on regular water, another 24 h on E. coli-supplemented water, and 66 h on regular water. In contrast, the cephaloridine treatment group followed an identical schedule, with cephaloridine-supplemented water substituted for regular water during the 54- and 66-h intervals.

1. 2. Fecal collection

Fresh fecal samples were collected from individual mice, placed in pre-labeled sterile tubes. All fecal samples were immediately snap-frozen in liquid nitrogen and stored at −80 °C until processing for microbial DNA extraction and further shotgun-metagenomic sequencing.

1. 3. Blood collection

Blood was collected via submandibular vein bleeding using lithium heparin tubes (BD 365965). Samples were centrifuged at 2,000g for 15 min at 4 °C to separate plasma, which was stored at −0 °C for subsequent metabolic analysis.

1. 4. Plasma metabolic panel analysis

All the assays were performed according to the manufacturer’s protocol using either ELISA-based or end point colorimetric assay system in a 96-well microplate format. Standard curves were generated for individual assays using the provided standards to determine the concentration of unknown samples. Plasma levels of adiponectin, leptin, and insulin were quantified using ELISA kit (Millipore). Non-esterified free fatty acids (NEFAs) were determined using Wako NEFA colorimetric assay system (Fujifilm). Glucose, triglycerides, and total cholesterol were detected using colorimetric assay kits (Abcam). Plasma HDL and LDL levels were detected using ELISA kit (LS Bio). Glycerol concentrations were measured using an endpoint colorimetric assay (Sigma-Aldrich). Absorbance readings were acquired using Benchmark Plus microplate reader (Bio-Rad). Concentrations of the samples were automatically calculated based on the corresponding standard curve used for the assays.

Statistical analysis

For all figure legends, asterisks indicate statistical significance as follows: NS, not significant (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Data were obtained by performing independently at least three biological replicates unless specified in the figure legends. No statistical method was used to pre-determine the sample size. No data were excluded from the analyses. Two-tailed Student t test or one-way or two-way ANOVA with Holm–Sidak corrections were used as indicated in the corresponding figure legends. N indicates the number of biological replicates. N indicates the number of animals or technical replicates within each biological replicate. For survival analysis, statistical analyses were performed with SPSS software (IBM) using Kaplan–Meier survival analysis and log-rank test, as well as using GraphPad Prism 10 survival analysis. For RNA-seq, a two-sided Wald test in R package Deseq2 was used. Figures and graphs were constructed using BioRender.com, ImageJ, GraphPad Prism 10 (GraphPad Software), and Illustrator (CC 2019; Adobe).