Abstract
Interactions between gut bacterial polyamines and intestinal cells have been proposed to contribute to inflammatory bowel diseases, but the underlying molecular mechanisms are often unclear. Here, we use a derivatization-based LC-MS approach and the model animal Caenorhabditis elegans to study microbiome-derived polyamine bioactivity. We show that aberrant polyamine metabolism in two diverse bacterial species (Escherichia coli K12 and Bacillus subtilis 168) can result in the accumulation of a noncanonical polyamine intermediate, N1-aminopropylagmatine (N1-APA). N1-APA is produced via spermidine synthase (SpeE) and is bioactive in C. elegans intestinal cells and mouse bone marrow macrophages. Specifically, bacterial N1-APA can be transported into intestinal cells via the polyamine transporter CATP-5, where it antagonizes C. elegans development and activates the mitochondrial unfolded protein response. N1-APA functions analogously to the deoxyhypusine synthase inhibitor GC7 in C. elegans and, like GC7, it antagonizes eIF5A hypusination and inhibits the alternative activation of mouse macrophages in vitro. Our results indicate that bacterial N1-APA is a bioactive metabolite that functions similarly to deoxyhypusine synthase inhibitors but has other unidentified targets that likely play roles in mitochondrial stress responses. We hypothesize that N1-APA production by the gut microbiome, caused by either high dietary agmatine or loss of agmatinase activity, might contribute to inflammatory bowel diseases.
Similar content being viewed by others
Introduction
Polyamines are abundant, aliphatic polycations that are synthesized from either agmatine or ornithine and contain two or more amino groups1. The canonical polyamine metabolites—putrescine, spermidine, and spermine—are found in nearly all domains of life2. Polyamines function as signaling molecules3,4,5, influence oxidative stress6,7, stabilize negatively charged marcomolecules8,9, and regulate mRNA translation via the hypusination of the transcription factor, eIF5A10. Dysregulation, altered transport, and mutations in the polyamine biosynthetic pathway have been linked to a wide variety of human pathologies. For example, overaccumulation of polyamines caused by mutations in the ATP-dependent P5-type transporter, ATP13A2, causes cytotoxicity and contributes to a heritable form of juvenile-onset Parkinson’s disease11,12,13. Whereas, mutations in ATP13A3 are associated with pulmonary arterial hypertension14. Within the polyamine biosynthetic pathway, mutations in the macrophage arginase, ARG1, which converts arginine to ornithine, enhance tumor progression15,16. In addition, mutations in spermine synthase (SMS) are implicated in the intellectual disability, Snyder-Robinson Syndrome (SRS)17 and mutations in ornithine decarboxylase (ODC-1) cause the neurodevelopmental disorder, Bachmann-Bupp syndrome (BABS)18. Thus, polyamine metabolism plays a pivotal role in maintaining human health.
Aberrant polyamine metabolism by the gut microbiome has been associated with both intestinal diseases and host health benefits; however, mechanistic insight into how microbiome-derived polyamines influence host biology remains elusive19,20,21,22,23. For example, polyamines accumulated by the probiotic Bifidobacterium animalis subsp. lactis have been associated with decreased gut inflammation and improved longevity24. Similarly, the loss of putrescine synthesis, by genetic deletion of agmatinase (speB) and ornithine decarboxylases (speC and speF) in E. coli, has been associated with mouse colitis. These findings suggest that the loss of bacterially-derived putrescine plays a role in colitis, however, the mechanism by which the loss of putrescine might lead to colitis is unclear25,26. Conversely, agmatine accumulation by the gut microbiome after exposure to metformin or after loss of agmatinase (speB) was suggested to contribute to gastrointestinal complications26. Similarly, in a study of the available IBDMDB data, several fecal polyamines, including putrescine, spermidine, and the polyamine precursor, agmatine, were associated with IBD in human patients27. These studies suggest that the same polyamines that have previously been found to benefit animal intestinal health25,26 can also drive animal intestinal diseases in other contexts. Lastly, although limited in scope, a few studies have found noncanonical bacterially produced polyamines antagonize host intestinal physiology. For example, N1, N12-diacetylspermine was correlated with biofilm formation on colon cancers28 and IBD27. Also, norspermidine, thought to be a major polyamine intermediate in the gut microbiome, antagonizes C. elegans development29,30. Despite the growing evidence that suggests polyamines from intestinal microbiota play a role in host intestinal dysfunction, the molecular mechanisms by which altered microbiome polyamine metabolism influences host physiology remain unclear.
Studying the impact of bacterially derived polyamines on animal intestinal health is difficult for many reasons. For example, animal intestinal microbiomes are comprised of numerous, diverse, and variable species of bacteria. Further, polyamines are imported, metabolized, and exported amongst the different species found in the gut microbiome31. This makes genetic studies of bacterial mutants that deplete specific polyamines challenging unless costly germ-free models are utilized. To overcome such limitations, we developed a method to use the model animal Caenorhabditis elegans to conduct high-throughput screens for bacterial genetic mutants that perturb animal physiology. We used Bacillus subtilis 168 as a model microbe because Firmicutes account for a large percentage of the bacterial species found in the gut microbiome32. Using our screening approach, we found a diet of B. subtilis ∆speB caused animals to completely arrest their development. speB encodes an agmatinase which plays a key role in polyamine biosynthesis. We further found that this diet-dependent arrest was not due to a requirement for bacterial putrescine or due to excess agmatine accumulation, as previously suggested26,33. Instead, we found that both B. subtilis and E. coli can produce the non-canonical polyamine N1-aminopropylagmatine (N1-APA) via the activity of spermidine synthase (SpeE). When N1-APA is accumulated in bacteria lacking speB, we found that it is transported into C. elegans intestinal cells via the conserved polyamine transporter, CATP-5, where it causes developmental arrest and increased expression and protein abundance of several canonical markers of mitochondrial stress. Lastly, we show N1-APA acts analogously to GC7 (a synthetic inhibitor of deoxyhypusine synthase) as these two structurally similar molecules inhibit C. elegans development, hypusination of eIF5A, and differentiation of mouse bone marrow-derived macrophages (BMMՓ). Our results show that N1-APA is bioactive in eukaryotic model organisms and that a polyamine metabolite derived from bacteria functions analogously to the deoxyhypusine synthase inhibitor, GC7, in vivo. The work presented here suggests speB mutations in gut bacteria may lead to accumulation of N1-APA, which likely would contribute to exacerbated IBD symptoms in both mice and humans25,34.
Results
Loss of speB in B. subtilis antagonizes C. elegans development
To identify new links between bacterial metabolism and animal physiology, we screened a complete B. subtilis non-essential gene knockout library for changes in animal development, using germ-free Caenorhabditis elegans as a model (Fig. 1A – See “Methods” for screen details)35. From our screen of 3,984 B. subtilis mutants, we identified only a single B. subtilis gene, speB (BSU37490), that was required for C. elegans to develop to adulthood (Fig. 1B). We confirmed that speB is required for animals to develop to adulthood by complementing ∆speB with a wild-type (WT) copy of speB (Fig. 1B). speB encodes agmatinase, which converts agmatine to putrescine for subsequent production of spermidine by spermidine synthase (SpeE)(Fig. 1C)36,37. We conclude that speB is required in B. subtilis for C. elegans to development to adulthood.
A Graphical representation of the platform used to screen for mutants of B. subtilis that did not support animal development. Germ-free embryos were arrayed on the library of 3984 B. subtilis mutants and screened for development beyond the L4 stage. Candidate mutants were retested on 60 mm plates with ~ 500 embryos. Image created in BioRender [Gates, D. (2025) https://BioRender.com/sph72q5]. B Representative images of WT C. elegans on E. coli OP50, B. subtilis 168 (Bs)(NOBb513), Bs ∆speB (NOBb514), and ∆speB thrC::speB (NOBb475). The animals were incubated at 20 °C for 3 days on their respective diets, washed in M9 buffer, and paralyzed with tetramisole (~ 10 µg/mL). The scale bar is 1 mm. Images are representative of three biological replicates. C Canonically, in B. subtilis, spermidine is synthesized from L-arginine by the consecutive activities of SpeA, SpeB, and SpeE. D B. subtilis speA and speEB genomic regions and operon structures. E Percentage of WT C. elegans that developed pasts L4 + on diets of B. subtilis 168 (WT)(NOBb513), ∆speB (NOBb514), ∆speB thrC::speB (NOBb475), ∆speE (NOBb517), ∆speEB (NOBb515), ∆speEB thrC::speE (NOBb501), ∆speA (NOBb516), ∆speB ∆speA (NOBb518), and ∆speA ∆speB thrC::speA (NOBb503). The animals were incubated at 20 °C for 3 days on their respective diets, and L4 + animals were quantified. Experiments were performed in triplicate with 500 animals per replicate. An ordinary one-way ANOVA followed by a Sidak’s multiple comparisons test was used to determine statistical significance. Error bars = s.d. from the mean, **** = p < 0.0001, ** = p = 0.0065, ns = not significant. F Relative quantification of agmatine, putrescine, and spermidine by LC-MS. Relative quantifications were normalized to WT to determine baseline-corrected fold changes. P-values were calculated using the values for pre-normalized areas. Experiments were performed in biological triplicate, starting from three single colonies. A one-way ANOVA followed by pairwise t tests and the BH procedure to correct for FDR were performed to determine statistical significance. Error bars = s.d. from the mean, *** = adj.p < 0.0001, ns = not significant. Source data are provided as a Source Data file.
In B. subtilis, arginine is converted to spermidine by the sequential activities of speA (arginine decarboxylase), speB (agmatinase), and speE (spermidine synthase) (Fig. 1C, D)36. To determine if nonspecific dysregulation of polyamine metabolism in B. subtilis causes C. elegans arrest, we fed WT C. elegans a diet of B. subtilis ∆speA and ∆speE. We found that deletion of speA and speE did not cause C. elegans developmental arrest (Fig. 1E and Supplementary Fig. 1). Since speA converts L-arginine into agmatine36, we generated a double ∆speA ∆speB deletion strain to test if preventing the aberrant accumulation of agmatine prevented animal developmental arrest. We found deletion of speA in the ∆speB background rescued C. elegans development, suggesting high agmatine may cause development arrest (Fig. 1E and Supplementary Fig. 1B). We confirmed this phenotype by complementing ∆speA ∆speB with speA (Fig. 1E and Supplementary Fig. 1B). However, counterintuitively, when we profiled the canonical polyamines in each mutant strain by LC-MS, we did not find agmatine accumulation in ∆speB. Rather, the mutant strain appeared to accumulate less agmatine that WT (Fig. 1F). We conclude that C. elegans developmental arrest when feeding on B. subtilis ∆speB is likely not caused by the accumulation of agmatine.
Since speB and speE are encoded in an operon in B. subtilis, we knocked out the entire operon to test if the operon deletion (∆speEB) phenocopied ∆speB (Fig. 1D). Surprisingly, we found that knocking out the speEB operon did not phenocopy ∆speB, instead C. elegans developed to adulthood on a diet of B. subtilis ∆speEB (Fig. 1E and Supplementary Fig. 1B). We confirmed this result by complementing ∆speEB with speE and found that the strain once again led to C. elegans arrested development (Fig. 1E and Supplementary Fig. 1B). Since the relative amounts of putrescine and spermidine did not vary between ∆speB and ∆speEB (Fig. 1F), we conclude the loss of putrescine and spermidine is not sufficient to cause C. elegans developmental arrest. Thus, we conclude that the deletion of speB leads to arrested C. elegans development, and that arrest is dependent on both speA and speE activity.
Loss of speB causes diverse species of bacteria to accumulate N1-APA
We hypothesized that B. subtilis ∆speB produces a noncanonical polyamine that antagonizes animal development. To detect and quantify the relative abundance of noncanonical polyamines, we developed an isobutyl chloroformate derivatization LC-MS method to profile polyamines in B. subtilis single mutants (ΔspeA, ΔspeB, ΔspeE), double mutants (ΔspeEB and ΔspeB ΔspeA), and WT cells38. Polyamines are not well retained or ionized in typical LC-MS metabolomics approaches38. So, we generated carbamyl derivatives of metabolite extracts and performed untargeted metabolomics. Using this approach, we detected 2641 blank- and peak quality-filtered LC-MS features. Differential abundance analysis revealed a consistent feature, 388.29207 m/z at retention time of 3.781 min, that was significantly increased in B. subtilis ΔspeB cells compared to WT and the other mutants (Fig. 2A). Evaluation of extracted ion chromatograms (EIC) indicated that this compound was only detectable in ΔspeB (Fig. 2B). Using accurate mass MS1 measurements, we found three possible chemical formulae within 5ppm (Fig. 2C). Of these, only one contained oxygen atoms, of which there are two per carbamyl derivative (C5H9O2; Fig. 2C). From this we deduced that the unknown compound contained two carbamyl-derivatized sites, and subtraction of 2x(C5H9O2) from the predicted chemical formula of the unknown (C18H37N5O4) yielded a parent compound formula of C8H21N5. This is consistent with the non-canonical polyamine, N1-APA, that was originally reported to accumulate in speB mutant Thermus thermophilus39 (N1-APA; (Fig. 2D). To confirm this putative identification, we spiked in a synthetic standard to ΔspeB lysate and observed an increase in the 388.2921 EIC over unspiked ΔspeB controls (Fig. 2E). MS2 fragmentation spectra of the endogenous unknown and the synthetic standard further confirmed the identity of the ΔspeB-specific polyamine as N1-APA (Fig. 2F). Collectively, our findings corroborate previous work that found N1-APA is synthesized by SpeE as an intermediate in the synthesis of spermidine by subsequent SpeB activity39,40 (Fig. 2G).
A Volcano plot showing differential abundance analysis of untargeted metabolomics on carbamylated-derivatives between ΔspeB cell lysates and the other strains evaluated. Red dots are LC-MS features with an FDR-adjusted P-value < 0.05, calculated in Compound Discoverer 3.3.3.200. Note the feature 388.29207 m/z at retention time 3.781 min is the most significantly elevated compound in each comparison. B Representative extracted ion chromatograms of 388.29207 m/z (+/− 5ppm). C Top, chemical formula predictions for m/z 388.29207 using accurate mass and fine isotopic structure in Compound Discoverer 3.3.3.200. Bottom, the structure of carbamylated-derivatives indicates two oxygens are present in each derivatized moiety. D Structure of N1-APA with sites of carbamylated-derivatives denoted with a blue asterisk. Note the chemical formula of N1-APA (C8H21N5) with carbamylated-derivatives (C5H9O2 x2) at the indicated sites matches the predicted chemical formula of the unknown (C18H37N5O4). (E) Overlaid extracted ion chromatograms of ΔspeB and ΔspeB + N1APA spike-in provide accurate mass and chromatographic profile confirmation of N1APA as the unknown feature 388.29207@3.7981 min. F Data dependent MS2 fragmentation mirror plot of parent ion 388.29207@3.7981 min in ΔspeB (top) and neat N1-APA standard (bottom) provides further confirmation that the unknown is N1-APA. G Canonical spermidine synthesis pathway (1.) compared to the alternative pathway (2.) found in Thermus thermophilus, which includes N1-APA as an intermediate metabolite. H Absolute quantification of N1-APA accumulated in B. subtilis ∆speB (NOBb369; ∆B; blue) and in E. coli ∆speB (NOBb806; ∆B; pink) and ∆speBCF (NOBb910; ∆BCF; pink) grown in the presence (+) and absence (-) high exogenous agmatine. Concentrations are reported as nanograms measured in 1 mL of culture at a cell density of 20 McFarland units (20 McF). Experiments were performed in biological triplicate, starting from three single colonies. A one-way ANOVA followed by a Tukey multiple comparisons test was performed to determine statistical significance. Error bars = s.d., ND = Not Detected. **** = p < 0.0001. Source data are provided as a Source Data file.
To determine if N1-APA accumulation is evolutionarily conserved, we quantified N1-APA abundance in the model Gram-negative bacterium, E. coli K12. Because a previous study found that C. elegans fed an E. coli ∆speB diet supplemented with high dietary agmatine caused a similar developmental arrest as we observed for C. elegans fed B. subtilis ∆speB, we quantified N1-APA with and without dietary agmatine supplementation26. We found that E. coli ∆speB accumulated N1-APA in an agmatine-dependent manner (Fig. 2H), however the accumulation was ~ 10-fold less than what is accumulated in B. subtilis ∆speB (Fig. 2H). The E. coli genome encodes SpeC and SpeF, which are ornithine decarboxylases that have not been identified in B. subtilis. We hypothesized a triple ∆speBCF mutant may accumulate additional N1-APA by converting accumulated ornithine to agmatine41. Furthermore, E. coli ∆speBCF was previously found to cause IBD in a germ-free mouse model25, thus the E. coli ∆speBCF might have specific clinical relevance for studying how polyamine metabolism in gut bacteria affects animal intestinal function. We found the triple E. coli ∆speBCF mutant accumulated N1-APA in both the presence and absence of agmatine (50 mM), albeit the concentration detected in ∆speBCF in the absence of agmatine is near the detection limit of our mass spectrometry method (Fig. 2H). We conclude that E. coli ∆speB, like B. subtilis ∆speB, accumulates N1-APA. However, in E. coli the accumulation of N1-APA is dependent on agmatine, indicating N1-APA is either further metabolized or less is produced in E. coli than in B. subtilis.
N1-APA drives mitochondrial dysfunction in a CATP-5 transporter-dependent manner
To determine if N1-APA is sufficient to cause C. elegans arrest, we grew C. elegans on plates supplemented with a concentration gradient of N1-APA and seeded with WT B. subtilis. We found exogenous N1-APA was sufficient to cause 100% developmental arrest in C. elegans at 0.25 mM (Fig. 3A). Developmental arrest was not observed when plates were supplemented with agmatine at the highest concentration tested (1 mM) (Fig. 3A and Supplementary Fig. 2). To determine if N1-APA inhibits the function of canonical polyamines in C. elegans, we performed a competitive inhibition experiment by supplementing the media with the lowest inhibitory concentration of N1-APA (0.25 mM) and with increasing concentrations of spermidine (0.125-1 mM). We found the addition of exogenous spermidine to the media is sufficient to suppress exogenous N1-APA activity (Fig. 3B) or B. subtilis ∆speB (Fig. 3C and Supplementary Fig. 3) in a dose-dependent manner. By contrast, we found exogenous putrescine and agmatine were not sufficient to suppress developmental arrest caused by B. subtilis ∆speB at the highest tested concentration (Fig. 3C and Supplementary Fig. 3). We conclude that N1-APA production by B. subtilis antagonizes animal development, possibly by competitively inhibiting the normal function of spermidine.
A Percent of WT C. elegans (N2) that developed past L4 when grown on media supplemented with a range of concentrations of N1-APA (blue) and seeded with wild type B. subtilis after incubating for 3 days at 20 °C. Control plates were supplemented with 1 mM agmatine (purple). Experiments were performed in triplicate with 200 animals per replicate. An ordinary one-way ANOVA followed by a Sidak’s multiple comparisons test was used to determine statistical significance. Error bars = s.d. from the mean. **** = p < 0.0001. B Percent of WT C. elegans that develop past L4 on modified NGM (see “Methods”) supplemented with 0.25 mM N1-APA and a concentration gradient of spermidine (0.125 —1 mM). Experiments were performed in triplicate with 300 animals per replicate. A one-way ANOVA followed by a Tukey multiple comparisons test was performed to determine statistical significance. Error bars = s.d. from the mean. **** = p < 0.0001. C Percent of C. elegans (N2) that developed past L4 on modified NGM (see methods) seeded with B. subtilis ∆speB (NOBb369) and supplemented with agmatine (1 mM), putrescine (1 mM), or spermidine (1 mM). See Supplementary Fig. 3 for representative images. Experiments were performed in triplicate with 300 animals per replicate. Error bars = s.d. from the mean. D Fluorescent (GFP) images of growth stage-matched irg-1p::GFP and hsp-6p::GFP nematodes fed a diet of B. subtilis (NOBb513) or B. subtilis ∆speB (NOBb514). The C. elegans strains were incubated with the respective B. subtilis strains for ~ 30 h at 20 °C. Representative images were pseudo-colored using ImageJ v1.54 f. Scale bars are 0.5 mm (left panel). Average fluorescence was quantified for 10 animals fed either diet using ImageJ v1.54 f (right panel). Error bars = s.d. from the mean, *** = p = 0.0003, **** = p < 0.0001. E Global proteomics of C. elegans hsp-6p::GFP (GL347) raised on B. subtilis WT (Bs168) or ∆speB (NOBb369) for 24 hours. A volcano plot showing the log(fold change) of ∆speB/WT (top). The plot depicts the inverse log10 transformed adjusted p-values < 0.05 vs. the log2 fold-change of the ∆speB vs. WT comparison. The plot coloring is split for significant and non-significant adjusted p-values as well as by up and down-regulated proteins (significant and upregulated = magenta; significant and downregulated = blue, non-significant = gray). HSP-6 and IRG-1 are labeled in black. Worm Enrichr Go Term Analysis are shown (bottom). F Representative images of WT C. elegans and catp-5(bur50) on B. subtilis 168 (Bs)(NOBb513) and B. subtilis 168 ∆speB (NOBb514). The animals were incubated at 20 °C for 3 days on their respective diets, washed in M9 buffer, and paralyzed with tetramisole. The scale bar is 1 mm. Images are representative of three biological replicates. G Percent of WT C. elegans (N2) and catp-5(bur50-59) that developed past L4 on B. subtilis 168 ∆speB (NOBb514) after 5 days at 26 °C. WT C. elegans on B. subtilis 168(Bs)(NOBb513) was included as a control. Bar graph colors correspond to mutations labeled in Fig. 3H. Experiments were performed in triplicate with 500 animals per replicate. An ordinary one-way ANOVA followed by a Dunnett’s multiple comparisons test was performed to determine statistical significance. Error bars = s.d. from the mean, **** = p < 0.0001. H Graphical representation of mutations in catp-5 that restored animal development per allele (bur50-59). Red = nonsense mutation. Blue = nonsynonymous mutation. Green = insertion mutation. Purple = splice adaptor mutation. Numbers 50–59 correspond to allele bur50-bur59. Source data are provided as a Source Data file.
To determine how N1-APA antagonizes animal development, we screened 11 C. elegans GFP stress reporters on a diet of either B. subtilis WT or ∆speB. We found that GFP expression under the control of the hsp-6 and irg-1 promoters was increased specifically in the intestine (Supplementary Fig. 4) when animals were fed B. subtilis ∆speB compared to WT (Fig. 3D) or other polyamine synthesis mutants (Supplementary Fig. 5). The activation of hsp-6 and irg-1 promoters specifically in intestinal cells suggests that N1-APA exerts its effect on animals after bacteria are ingested. hsp-6 expression is part of the mitochondrial unfolded protein response(mitoUPR) and is activated in response to mitochondrial stress42. irg-1 was originally used as a reporter for bacterial infections, but was later found to also be activated in response to mitochondrial stress43. Thus, the activation of both hsp-6 and irg-1 is consistent with mitochondrial stress and the activation of the mitoUPR. Importantly, the activation of the mitoUPR is likely not due to general proteostasis stress because we did not observe activation of hsp-4::GFP, which is activated as part of the endoplasmic reticulum unfolded protein response (Supplementary Fig. 6). We did not observe a difference in fluorescence between ∆speB and WT in the other nine reporters that were tested: nlp-29, clec-60, gcs-1, hsp-60, hsp-16.2, gst-4, hsp-4, sod-3, bvIs5 (Table 1). Similar to B. subtilis ∆speB, we also found E. coli ∆speB and ∆speBCF activated C. elegans hsp-6p::GFP expression in the presence of exogenous agmatine (Supplementary Fig. 7B). We hypothesize this is not due to agmatine accumulation because agmatine also accumulates in E. coli WT cells grown in the presence of agmatine but, hsp-6 is not activated (Supplementary Fig. 7A, B). Second, we hypothesize hsp-6 is not activated by a loss of putrescine, because putrescine is decreased in E. coli ∆speB cells in the absence and presence of agmatine, but hsp-6 is only activated with agmatine (Supplementary Fig. 7A, B). We also found hsp-6 is likely not activated by a loss of spermidine (Supplementary Fig. 6A). Importantly, we found that activation of hsp-6 by E. coli ∆speB and ∆speBCF was dependent on speE (Supplementary Fig. 6B). These data collectively suggest that N1-APA synthesized by speE is required for activation of hsp-6. We conclude that bacterially produced N1-APA, produced by either B. subtilis or E. coli, activates the C. elegans mitoUPR, likely by causing mitochondrial stress.
To further define how N1-APA alters mitochondrial function and the abundance of mitochondrial stress response proteins in C. elegans, we exposed WT C. elegans to B. subtilis WT and ∆speB for 24 h and profiled global differences in protein abundance by proteomics (Supplementary Data 1). We confirmed that IRG-1 and HSP-6 proteins were more abundant in C. elegans exposed to B. subtilis ∆speB when compared to C. elegans exposed to WT B. subtilis (Fig. 3E), corroborating our previous results using GFP reporters (Fig. 3D). In addition to HSP-6, we found that mitochondrial proteins were the most significantly enriched class of proteins that exhibited increased abundance in this comparison by WormEnrichr44 (Fig. 3E, lower panels, Supplementary Data 2). For example, we found cox-17, involved in cytochrome c oxidase assembly, was among the proteins that exhibited the largest increase in abundance in C. elegans exposed to B. subtilis ∆speB when compared to C. elegans exposed to WT B. subtilis (Supplementary Data 1). Similarly, we found that 5 proteins that were previously found to increase after exposure to the trichothecene mycotoxin, deoxynivalenol, (irg-1, ugt-31, oac-32, T16G1.5, and ugt-24)45 were also increased in abundance in response to B. subtilis ∆speB (Supplementary Data 1). The mechanism of action of trichothecene toxins includes the generation of ROS and inhibition of mitochondrial translation46. Notably, several additional proteins were dysregulated by B. subtilis ∆speB, but their role in N1-APA-mediated arrest remains unclear (Supplementary Data 1). Collectively, our proteomics findings describe a broad stress response to a bacterial toxin (N1-APA) that likely perturbs C. elegans mitochondrial function.
To determine how bacterially produced N1-APA antagonizes C. elegans development, we mutagenized WT C. elegans using ethyl methanesulfonate (EMS) and screened for mutant animals that developed to adulthood despite exposure to N1-APA. We isolated 10 lines of mutant animals that developed to adulthood when fed B. subtilis ∆speB (Supplementary Fig. 8A and Table 2). Following short read whole genome sequencing and variant calling, we found all the mutant lines harbored predicted loss-of-function SNPs in the catp-5 polyamine transporter (Fig. 3F–H). CATP-5 is an ortholog of human ATP13A3 and is only expressed in C. elegans intestinal cells and the excretory cell. CATP-5 was also previously reported to import polyamines from the intestinal lumen into the intestinal cells29. These findings, along with our findings that hsp-6 and irg-1 reporter expression turn on specifically in intestinal cells, are consistent with a model in which gut microbiome-produced N1-APA is imported by CATP-5 into C. elegans intestinal cells, where it antagonizes animal development.
N1-APA functions analogously to the deoxyhypusine synthase inhibitor, GC7, in both invertebrates and mouse bone marrow macrophages (BMMΦ)
To determine if N1-APA antagonizes development and mitochondrial function across species, we exposed the nematode Pristionchus pacificus to B. subtilis WT and ∆speB (Fig. 4A). P. pacificus and C. elegans shared a last common ancestor approximately 100 million years ago47,48. We found that B. subtilis ∆speB also antagonized the development of P. pacificus and led to a > 90% reduction animal fertility (Fig. 4A, B). We conclude that N1-APA broadly affects nematode development and physiology and that this effect is not specific to C. elegans.
A Representative images of Pristionchus pacificus on a diet of E. coli OP50, B. subtilis, and B. subtilis ΔspeB (NOBb369). Scale bars are 1 mm. B Quantification of brood size in P. pacificus on a diet of E. coli OP50 (pink), B. subtilis 168 (blue), and B. subtilis ΔspeB (NOBb369; grey). Brood size was recorded for nine animals fed each respective diet. An ordinary one-way ANOVA followed by a Sidak’s multiple comparisons test was used to determine statistical significance. Error bars = s.d. from the mean, **** = p < 0.0001. C Fluorescent (GFP) images of growth stage matched C. elegans encoding hsp-6p::GFP fed a diet of B. subtilis (NOBb513) on NGM agar plates supplemented with 0.125 mM spermidine, N1-APA, and GC7. Chemical structures are shown to the right. The C. elegans strain were incubated with B. subtilis strain for ~ 30 h at 20 °C. Representative images were pseudo-colored using ImageJ v1.54 f. Scale bars are 0.5 mm. The average fluorescence of 10 animals was quantified using ImageJ v1.54 f. error bars = s.d. from the mean, **** = p < 0.0001. D Percent of WT C. elegans that developed past L4 on B. subtilis 168 (NOBb513) after exposure to 0.125 mM spermidine, norspermidine, N1-APA, and GC7 for 3 days at 20 °C. Experiments were performed in triplicate with 500 animals per replicate. An ordinary one-way ANOVA followed by a Sidak’s multiple comparisons test was used to determine statistical significance. Error bars = s.d. from the mean, **** = p < 0.0001. ns = not significant. E Representative flow plots, gates of RELM-α + BMMΦ. Proportions of (F) RELM-α and (G) CD301 BMMΦ (IL-4) following treatment with 200 µM N1-APA or GC7 for 24 h. Experiment performed in triplicate; representative of 2 independent experiments. A Brown-Forsythe and Welch ANOVA with Dunnett’s T3 multiple comparisons correction was performed to determine statistical significance. Error bars = S.E.M., **** = p < 0.0001, *** = p = 0.009, ** = p = 0.001. H N1-APA inhibits eIF5a hypusination in BMMΦ. Representative images of 3 independent experiments. Source data are provided as a Source Data file.
In the polyamine-eIF5A-hypusine axis, which is conserved in most animals from nematodes to humans49, deoxyhypusine synthase (DHPS) modifies eIF5A by transferring putrescine from spermidine to lysine5050,51,52. This post-translational modification is hypothesized to alter the translation of a subset of mRNAs, including many mitochondrial proteins53. Previous chemical inhibition studies found that GC7, a synthetic polyamine, was a potent inhibitor of deoxyhypusine synthase53,54. Furthermore, this study tested many derivatives of GC7 and found that N1-APA could also inhibit deoxyhypusine synthase in vitro, but that it was less potent than GC754. Given that deoxyhypusine synthase is evolutionarily conserved, we hypothesized that N1-APA causes both mitochondrial stress and developmental arrest in C. elegans, in part, by inhibiting deoxyhypusine synthase. To test this, we first exposed WT C. elegans expressing the mitochondrial stress reporter hsp-6p::GFP to either N1-APA or GC7 and found that hsp-6p::GFP expression is activated by both molecules (Fig. 4C). We also found that both GC7 and N1-APA caused animals to arrest their development (Fig. 4D). Importantly, neither spermidine nor norspermidine activated hsp-6p::GFP expression nor caused developmental arrest at the relevant concentrations, suggesting these results are not broadly caused by excess exogenous polyamines (Figs. 4C, D and Supplementary Fig. 9). We conclude that GC7 and N1-APA cause similar activation of a mitochondrial stress reporter and developmental arrest in C. elegans.
In mammals, polyamines are important for the regulation of macrophage activation53. Activation of macrophages by IL-4 leads to an anti-inflammatory phenotype that can be inhibited by GC7 and promoted by putrescine25,53,55. Because we found that GC7 and N1-APA activated HSP-6 and caused arrest in C. elegans, we tested whether N1-APA also altered the activation of mammalian macrophages in response to IL-4. Consistent with our hypothesis that GC7 and N1-APA function similarly, we found N1-APA inhibits IL-4 induced expression of M2-like macrophage markers RELM-α and CD301 (Figs. 4E–G and Supplementary Fig. 10). We also found N1-APA treatment decreased hypusination of eIF5A, suggesting that both GC7 and N1-APA inhibit deoxyhypusine synthase in macrophages (Fig. 4H and Supplementary Fig. 10). Lastly, to test if N1-APA alters mitochondrial function in mammalian macrophages, as predicted by our C. elegans studies, we quantified mitochondrial oxygen consumption ratios (OCRs) in IL-4 treated macrophages in the presence and absence of N1-APA. Consistent with our findings in C. elegans, we found that mitochondrial oxygen consumption was impaired in N1-APA treated cells (Supplementary Fig. 10D). These findings demonstrate that N1-APA inhibits the hypusination of eIF5A in vivo and are consistent with a model in which N1-APA interferes with a normal function of spermidine in the polyamine-eIF5A-hypusine axis53.
While GC7 has been reported to alter macrophage differentiation, the exact mechanism by which GC7 exerts its effects on cells remains debated. While some findings suggest that GC7 decreases macrophage mitochondrial function by inhibiting deoxyhypusine synthase, others suggest it functions via an unknown target or pathway, independent of deoxyhypusine synthease55. To test if B. subtilis ∆speB caused C. elegans to arrest their development due to the inhibition of deoxyhypusine synthase (DHPS-1), we grew C. elegans lacking dhps-1 (deletion) on WT B. subtilis and ∆speB. As previously reported, the homozygous dhps-1(ve590) mutants were sterile and smaller than wild-type animals, suggesting dhps-1 plays a role in C. elegans development and fertility (Supplementary Fig. 11). We found that feeding dhps-1 mutants B. subtilis ∆speB resulted in early developmental arrest, similar to heterozygous dhps-1 mutants or WT C. elegans (Supplementary Fig. 11A). These findings suggest that, while N1-APA inhibits deoxyhypusine synthase (Fig. 4H and Supplementary Fig. 11A), N1-APA likely also causes C. elegans development arrest by an unknown, dhps-1-independent mechanism55. Further, we found that hsp-6::GFP is not activated by the loss of dhps-1, suggesting hsp-6::GFP is also activated by a dhps-1 independent mechanism (Supplementary Fig. 11B). Future work aims to identify additional targets of GC7 and N1-APA and will be critical for better understanding how they regulate diverse aspects of animal physiology across species.
Discussion
In summary, our findings establish that N1-APA, produced via an alternative spermidine synthesis pathway in bacteria, inhibits nematode development and mouse macrophage differentiation. Our results demonstrate that N1-APA is bioactive in animals (both invertebrates and mammalian cells) and that diverse bacteria synthesize an inhibitor that functions similarly to GC7 by inhibiting deoxyhypusine synthase. Thus, we establish a link between a bacterial intermediate metabolite and animal development. However, since we found dhps-1 mutants further arrest when feeding on B. subtilis ∆speB, we hypothesize N1-APA has additional unknown targets (Fig. 5). Our findings add further evidence that GC7 and N1-APA have bioactive functions beyond inhibiting deoxyhypusine synthase and that these activities can alter animal mitochondrial function.
N1-Aminopropylagmatine accumulates in the C. elegans intestinal lumen when the animals are grown on a diet of B. subtilis ∆speB. CATP-5 transports N1-APA into intestinal cells where it inhibits eIF5a hypusination. N1-APA antagonizes mitochondrial and immune function. Image was created in BioRender [Nauta, K. (2025) https://BioRender.com/wh83sde] and ChemDraw.
Our findings suggest that the role of N1-APA in biology might be more profound that previously appreciated. Specifically, our findings indicate that N1-APA not only is synthesized in both B. subtilis ∆speB and E. coli ∆speB, but also that it can be produced at significant enough quantitates to alter animal physiology. Several recent publications also suggest N1-APA synthesis in bacteria may be widely conserved. First, Li et al., 2024 reported the surprising finding that 17 out of 18 bacterial spermidine synthases, including speE from B. subtilis, are sufficient to produce N1-APA when expressed in E. coli BL2140. These unexpected findings led the authors to conclude that N1-APA production may be the favored intermediate between agmatine and spermidine for some bacterial species. Second, recent work also found the arginine-agmatine-spermidine synthesis pathway contributes to spermidine synthesis in mammalian gut microbiomes27. These findings suggest N1-APA may be produced in the human gut microbiome as an intermediate. Third, carboxyaminopropylagmatine (CAPA) dehydrogenase has also been found to synthesize CAPA from agmatine and l-aspartate-β-semialdehyde, which is then dehydrogenated to form N1-aminopropylagmatine. This pathway was found to be conserved in 15 bacterial phyla56. Taken together, we conclude N1-APA is likely present in several environmental niches where it could antagonize host mitochondrial function.
We found ~ 10x more N1-APA accumulated in B. subtilis ∆speB than in E. coli ∆speB, and that N1-APA production in E. coli required high exogenous agmatine. While we cannot conclusively address this difference, we hypothesize it may be due to variabilities in polyamine metabolism across species. For example, E. coli encodes SpeC and SpeF, two ornithine decarboxylases that produce putrescine from ornithine57,58. These enzymes have not been identified in B. subtilis. Rather, B. subtilis metabolizes ornithine into L-glutamate semialdehyde via the ornithine transaminase, RocD59. In addition, previous work found speE from both E. coli K12 and B. subtilis 168 were sufficient to produce N1-APA in an E. coli BL21 ∆speB model40. These findings suggest that E. coli and B. subtilis spermidine synthases do not inherently encode variable abilities to convert agmatine to N1-APA. Therefore, we hypothesize that E. coli encodes enzymes that further modify or export N1-APA. Alternatively, we hypothesize E. coli converts excess agmatine to ornithine via the urea cycle41. However, why B. subtilis does not also modify or export accumulated N1-APA remains to be determined. Future work will be critical to define mechanistic requirements for the bacterial synthesis and accumulation of N1-APA.
A recent study found that bacteria-derived agmatine underlies metformin-associated gastrointestinal complications in patients and leads to C. elegans developmental arrest26. This was, in part, based on the finding that supplementing C. elegans feeding on E. coli ∆speB with agmatine resulted in developmental arrest. In our hands, N1-APA produced by E. coli, not dietary agmatine, was sufficient to activate hsp-6 in C. elegans. Several of our findings support this claim: (1) N1-APA accumulates in E. coli ∆speB, (2) we found activation of hsp-6 in the presence of agmatine (1 mM) required deletion of E. coli speB, and 3) we found hsp-6 was not activated when the animals were fed a diet of E. coli ∆speB ∆speE, suggesting SpeE is required to convert agmatine into N1-APA. We hypothesize agmatine did not cause developmental arrest in our hands due to lower levels of N1-APA accumulation, possibly caused by slight differences in strains or experimental conditions. Nevertheless, our findings suggest N1-APA could contribute to the gastrointestinal complications caused by metformin treatment in animal models and Type II Diabetes patients26. In this model, the effects of metformin, a Complex I inhibitor, may be altered because N1-APA is produced by ∆speB (with high dietary agmatine) and disrupts animal mitochondrial function26. Intriguingly, metformin was recently proposed as an inhibitor of bacterial agmatinases such as speB due to its polyamine-like structure60. Thus, we hypothesize N1-APA may accumulate in gut microbiota that have been exposed to metformin.
Lastly, studies in mice and humans have linked loss of speB in the gut microbiome to IBD in mammals34. For example, colonizing germ-free mice with E. coli ∆speBCF promoted the development of severe colitis in mice25. The development of colitis was proposed to be due to a lack of bacterial putrescine. However, our data suggest N1-APA might also contribute to colitis and could explain why colitis pathologies caused by E. coli ∆speBCF in mice are similar to those caused by GC725. Specifically, our findings suggest that the reason both E. coli ∆speBCF and GC7 were found to drive colitis-like changes in mice or mouse cells might be because E. coli ∆speBCF produces N1-APA, which inhibits the development of anti-inflammatory M2-like macrophages, like GC7 (i.e., “M2-like” macrophages, characterized by the production of RELM-α and CD301 (Fig. 4), are anti-inflammatory, and if N1-APA inhibits their activation, it might promote inflammatory bowel disease like symptoms in the intestine). Accumulation of N1-APA might also provide an explanation for why the absence of gut microbiome speB was associated with IBD in humans34. Our data suggest that existing inhibitors of the bacterial enzymes SpeA or SpeE, or excess dietary spermidine, could potentially be useful in the treatment of IBD because they would block N1-APA production or alleviate its deleterious effects. Future work will be critical to determine if N1-APA contributes to IBD development in humans or if inhibiting N1-APA production in intestinal bacteria alleviates signs of IBD.
Methods
This research complies with all relevant ethical regulations as recommended by the Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the IACUC committee of Van Andel Research Institute. Research with C. elegans does not require IACUC approval.
Quantification and statistical analysis
An ordinary one-way ANOVA followed by a Sidak’s multiple comparisons test were used to determine statistical significance for figures: 1E, 3 A, 4B, and 4D. 2 A adjusted p-values were calculated using Compound Discoverer 3.3.3.200. An ordinary one-way ANOVA followed by a Dunnett’s multiple comparisons test were performed to determine statistical significance for Fig. 3G. A one-way ANOVA followed by a Tukey multiple comparisons test was performed to determine statistical significance for Figs. 3B and 2H. A Brown-Forsythe and Welch ANOVA with Dunnett’s T3 multiple comparisons correction was performed for Figs. 4F, G, and Supplementary Fig. 10. A 2-way ANOVA was performed to determine statistical significance for Supplementary Fig. 11A. All p-value calculations were performed in GraphPad Prism 10.6.1 (892)61. Adjusted p-values to correct for false discovery rates due to multiple testing were calculated in R version 4.5.0(2025-04-11) and RStudio 2025.05.0.496 using pairwise.t.test (version 3.6.2) with the parameter p.adjust.method = “BH”62. Adjusted p-values are reported for the relative quantification of polyamines (Fig. 1F and Supplementary Fig. 7A). No statistical method was used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.
C. elegans growth media and conditions
C. elegans strains were maintained at 15 °C or 20 °C on NGM agar plates (3 g/L NaCl, 17 g/L Agar, 2.5 g/L peptone (Bacto), 25 mM KPO4 (pH 6), 5 mg/L Cholesterol, 1 mM MgSO4, 1 mM CaCl2) seeded with E. coli OP50 (CGC, DA837). All C. elegans strains used in this study are hermaphrodites and are listed in Table 3. C. elegans that were fed a diet of B. subtilis, B. subtilis mutants, E. coli K12, or E. coli K12 mutants for experimentation were grown on modified NGM (3 g/L NaCl, 17 g/L Agar, 5.0 g/L peptone (Bacto), 25 mM KPO4 (pH 9.2), 5 mg/L Cholesterol, 1 mM MgSO4, 1 mM CaCl2).
Mice
C57BL/6 J (000664) mice were purchased from Jackson Laboratory and bred in-house. All mice were bred and maintained in grouped housing at Van Andel Research Institute Vivarium under specific pathogen-free conditions. All procedures involving mice were completed as recommended in the Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the IACUC committee of Van Andel Research Institute. Female mice were used for this study between the ages of 8-10 weeks. Three mice were used per experiment, nine total. Flow cytometry, Western Blots, and OCRs are from the same mice for each experiment.
Egg Prep protocol
Adult C. elegans were washed from NGM agar plates with M9 buffer (3 g/L KH2PO4, 6 g/L Na2HPO4, 5 g/L NaCl, 1 mM MgSO4). C. elegans were lysed with Egg Preparation Solution (30% Sodium hypochlorite, 1 M NaOH) and monitored by dissection microscope for ~ 1–3 min. The embryos were washed 2-3 times in M9 buffer.
Bacteria growth media and conditions
All bacterial strains were grown on Lennox LB Agar (5 g/L NaCl, 10 g/L SELECT Peptone 140, 5 g/L SELECT Yeast Extract) (Invitrogen, 12780029) at 30 °C or 37 °C unless specified otherwise and are listed in Table 3. Liquid cultures were grown at 30 °C and shaken in < 5 mL Lennox LB at 220 rpm in 18 mm glass test tubes. Bacillus strains containing pDG1664 and its derivatives inserted at thrC were grown in Lennox LB containing MLS (25 µg/mL erythromycin (Thermo Fisher Scientific, 50-213-296) and 5 µg/mL lincomycin (Thermo Fisher Scientific, 50-213-395)). Bacillus non-essential gene knock out library mutants were grown in kanamycin (7.5 µg/mL (Sigma, K1377)). Plasmid selection in E. coli was performed on Lennox LB containing kanamycin (25 µg/mL, Sigma Aldrich, K1377) or ampicillin (100 µg/mL, Sigma Aldrich, A0166).
Bacteria strains and plasmid construction
All plasmids were designed in Benchling, constructed by isothermal assembly (ITA), and are listed in Table 4. DNA regions for plasmid assembly were identified using Biocyc63 and amplified by touchdown PCR from the bacterial chromosomes using Q5 DNA polymerase (New England Biolabs, M0492) and oligonucleotides from Integrated DNA Technologies (Coralville, IA). All oligonucleotides used in this study are listed in Supplementary Data 3. PCR fragments were purified using the Monarch DNA Gel Extraction Kit (New England Biolabs, T1020), and assemblies were performed using the NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs, E2621). All regions amplified by PCR were confirmed by DNA sequencing at Genewiz from Azenta Life Sciences (South Plainfield, NJ). Plasmids were transformed according to the manufacturer’s instructions and maintained in E. coli DH5α competent cells (New England Biolabs, C2988J). Plasmids or purified chromosomal DNA were transformed into B. subtilis strains by natural competency35.
All Bacillus subtilis strains reported in this study are isogenic derivatives of B. subtilis 168 (BGSCID: 1A1) and are reported in Table 3. The operon encoding speEB was knocked out by homologous recombination using the suicide vector, pMAD64. Approximately 1 kb upstream and downstream of speEB were amplified and assembled via ITA into pMAD digested with BglII and EcoRI (New England Biolabs, R0144 and R3101, respectively), resulting in plasmid pNOBb41 (Table 4). The deletion was confirmed by PCR, sanger sequencing, and whole genome sequencing. Strain NOBb518 (ΔspeB speA::kanR) was constructed by transforming purified chromosomal DNA from strain BKK14630 (speA::kanR)35 into NOBb369 (∆speB). Chromosomal DNA was transformed by natural competency35.
Plasmids for complementation studies (pNOB47, pNOB48, and pNOB49) were constructed by amplifying target genes from the Bacillus subtilis 168 chromosomes using the primers listed in Supplementary Data 3. The amplicons were assembled in pDG1664 digested with EcoRI and HindIII (New England Biolabs, R3101 and R3104, respectively) by ITA and were expressed under the control of their respective native promoters (Table 4). When necessary kanR was removed from the Bacillus subtilis chromosome using pDR24435 (Bacillus Genetic Stock Center, BGSCID ECE274).
E. coli K12 knock-out strains were constructed by P1 transduction using E.coli K12 donor strain lysate65. The lysate was prepared using E. coli P1 bacteriophage (ATCC, 25404-B1). Briefly, E. coli K12 donor strains was grown overnight at 37 oC in LB (30 μg/mL kanamycin). The culture was diluted 1:100 in LB supplemented with 0.2% glucose and 5 mM CaCl2, and incubated at 37 oC for 45 min, 220 rpm. 100 μL of E. coli bacteriophage P1 (ATCC 25404B1TM) was added to the donor bacteria culture and incubated at 37 oC for 3 hr. 200 µL of chloroform were added and incubated at 37 oC for 5 min. The mixture was centrifuged at 9200 x g for 5 min at 4 oC. Lysate was further purified using a 0.45 μm filter before collecting the P1 phage-containing supernatant. Recipient strain E. coli K12 strains were grown overnight at 37 oC in 3 mL LB broth. E. coli recipient cells were resuspended in a P1 salt solution (10 mM MgCl2, 5 mM CaCl2). Next, 10 μL of the P1 lysate was mixed with 100 μL of the recipient cells and incubated with shaking (220 rpm) at 37 oC for 30 min. Then, 1 mL of LB and 200 μL of 1 M sodium citrate was added and incubated at 37 oC for 1 hr, before being centrifuged at 13,000 x g for 2 min. Transformants were selected on kanamycin (30 µg/mL) and sodium citrate (5 mM sodium citrate) LB65,66. The kanamycin cassettes were removed using the temperature-sensitive plasmid, pCP20 (NovoPro, V001036).
Screen of B. subtilis 168 knockout library for loss of C. elegans development
The B. subtilis single gene deletion library--kanamycin was acquired from Addgene (Watertown, MA, 1000000115). The parent strain was acquired from the Bacillus Genetic Stock Center (Columbus, OH, BGSCID: 1A1). The library was replicated into 96-well plates for further use. Each 96-well plate was replicated and incubated overnight at 30 °C in LB. Each well was seeded on NGMB in a 12-well plate and dried in a laminar flow hood. 50–100 C. elegans daf-2 (DR1572) embryos were added to each well. The plates were incubated for 5 days at 26 °C. Each B. subtilis mutant that did not promote animal development was retested on a larger scale using 90 mm plates.
EMS mutagenesis of WT C. elegans
L4 staged WT C. elegans were washed in M9 buffer (recipe: 3 g/L KH2PO4, 6 g/L Na2HPO4, 5 g/L NaCl, 1 mM MgSO4) 3 times and resuspended in 4 mL M9 buffer. 20 µL of Ethyl methanesulfonate (EMS) (Sigma Aldrich, M0880) were added, and the animals were incubated for 4 h with a rotary mixer, at 20 °C. The animals were washed three times and plated in 20 pools of ~ 200 on E. coli OP50-seeded NGM. The animals were screened on B. subtilis ∆speB for development beyond L4. Approximately one animal from each pool that developed beyond L4 was isolated and whole genome sequenced.
C. elegans genomic DNA extraction and whole genome sequencing
WT C. elegans were incubated on NGM for 3 days at 20 °C. The animals were washed in M9 buffer. and suspended 1:1 in Worm Lysis Buffer (50 mM KCl, 10 mM Tris (pH 8.3), 2.5 mM MgCl2, 0.45% Tween 20, 0.01% Gelatin, 0.45% Nonidet P 40 Substitute (Sigma Aldrich, 74385). The animals were incubated at 56 °C for 1 h with 10 µL of proteinase K (20 mg/mL). After lysis, genomic DNA was purified by phenol-chloroform extraction and ethanol precipitation. DNA pellets were resuspended in sterile DI water.
Whole genome Illumina sequencing and variant calling was performed by SeqCenter (Pittsburg, PA). Briefly, Illumina sequencing libraries were prepared using the tagmentation-based and PCR-based Illumina DNA Prep kit and custom IDT 10 bp unique dual indices (UDI) with a target insert size of 320 bp. No additional DNA fragmentation or size selection steps were performed. Illumina sequencing was performed on an Illumina NovaSeq 6000 sequencer in one or more multiplexed shared-flow-cell runs, producing 2 x 151bp paired-end reads. Demultiplexing, quality control and adapter trimming was performed with bcl-convert1 (v4.1.5). The fastq paired end reads were prepared for variant calling using Samtools 1.17 and then indexed using bwa-mem index. Raw reads were then mapped to the indexed reference using bwamem2 2.2.1 and subsequently sorted and converted to BAM files using Samtools 1.17. GATK’s3(4.0.12) MarkDuplicates functionality was used to remove duplicate reads from the alignment file. GATK’s HaplotypeCaller was then used to call variants on the alignment file. BcfTools 1.17 was used to filter out variants with a QD < 2, or MQ < 40 or MQRankSum < − 12.5, or ReadPosRankSum < − 8, or FS > 60.0, or SOR > 3. Variant effect predictions were made using snpEff 5.1.
Microscopy
C. elegans strains were washed with M9 buffer prior to imaging to remove excess bacteria. The animals were paralyzed with tetramisole ( ~ 10 µg/mL). and incubated for ~ 2 min. Images were taken with a Leica DM6 B upright microscope with 4.00 x or 10.00 x objective lenses and a Leica DFC9000 GT camera equipped with 4.2 megapixel (2048 x 2048 pixels) resolution; 6.5 µm x 6.5 µm pixel size. Leica FLUO Filter cubes were used for GFP and TXR excitation. Leica Application Suite X (3.7.5.24914) was used for image acquisition. Images were pseudo-colored and merged using FIJI 1.54p67. The display lookup table (LUT) is linear and covers the full range of the data for all pseudo-colored images. Images were assembled and analyzed in Adobe Illustrator 2024 28.4.1 64-bit68. Fluorescence was quantified for 10 whole animals per strain or condition. Averages and standard deviations were calculated in GraphPad Prism 10.6.1 (892).
Metabolomics
Extractions
Metabolites from bacterial cell pellets were extracted via the addition of ice-cold 2:2:1 (v/v) mixture of acetonitrile:methanol:water (#A456, #A955, #W6, respectively; Thermo Fisher Scientific) at a ratio of 30 µl sample per mL of extraction solvent. The samples were vortexed for 10 sec, sonicated for 5 minutes in a water bath sonicator, and incubated on wet ice for 60 min. After incubation, samples were centrifuged at 17,000 × g and 4 °C to pellet the insoluble fraction. The metabolite-containing supernatant was collected, dried in a vacuum evaporator, and stored at -80°C until derivatization69.
Polyamine derivatization
Carbamylated-derivatives were generated by derivatizing the dried metabolite fraction with isobutylchloroformate38. Dried metabolite extracts were resuspended in 200 µL of water containing 1 µg/mL of 1,6-Diaminohexane as an internal standard. 5 µL of sodium bicarbonate (1 M, pH 9.0) was added to each sample vial, followed by the addition 20 µL of isobutyl chloroformate (177989, Sigma). Samples were vortexed for 10 s then incubated at 37 °C for 15 min. After cooling to room temperature, 1 mL of diethyl ether (309966, Sigma) was added, and samples were vortexed for 10 s. Samples were then centrifuged for 1 min to separate the organic and aqueous layers. 800 µL of the carbamylated metabolite-containing organic layer was transferred to an autosampler vial, and subsequently dried under a gentle stream of nitrogen gas. Dried samples were resuspended in 100 µL of 1:1 (v/v) acetonitrile:water, vortexed, and 90 µL transferred to a fresh autosampler vial for LC-MS analysis. For LC-MS analysis of N1-APA, a chemical standard (EN300-47248193, Enamine) was prepared at 2 µg/mL in water and derivatized as described above as either a neat standard or spiked into dried metabolite extracts.
Untargeted LC-MS
LC-MS analysis of samples for Fig. 2A–F was performed on derivatized metabolite extracts analyzed by high-resolution accurate mass spectrometry using an ID-X Orbitrap mass spectrometer (Thermo Fisher Scientific) coupled to a Thermo Vanquish Horizon liquid chromatography system as previously reported38. 2 μL of sample volume was injected on column. Chromatographic separations were accomplished with a Cortecs T3 (120Å, 1.6 μm, 2.1 mm × 150 mm) analytical column (#186008500, Waters, Eschborn, Germany) fitted with a pre-guard column (120Å, 1.6 μm, 2.1 mm × 5 mm; #186008508, Waters, Eschborn, Germany) using an elution gradient with a binary solvent system. Solvent A consisted of LC/MS grade water (W6–4, Fisher), and Solvent B was 99% LC-MS grade acetonitrile (A955, Fisher). Both mobile phases contained 0.1% (v/v) formic acid (A11710X1, Fisher). The 15-min analytical gradient at a flow rate of 400 μL/min was: 0–0.5 min hold at 100% A, 0.5–2.0 min ramp from 0% B to 40% B, 2.0–10.0 min from 40% B to 99% B, then held at 99% B from 10.0–15.0 min. Following the analytical separation, the column was re-equilibrated for 3.5 min as follows: 0–0.5 min hold at 100% A at 600 μL/min, 0.5–0.6 min decrease from 600 μL/min to 400 μL/min, 0.6–3.5 min hold at 100% A and 400 μL/min. The column temperature was maintained at 50 °C. The H-ESI source was operated at a spray voltage of 2500 V(negative mode)/3500 V(positive mode), sheath gas: 70 a.u., aux gas: 25 a.u., sweep gas: 1 a.u., ion transfer tube: 300 °C, vaporizer: 250 °C. Sample data were collected via data-dependent MS2 (ddMS2) fragmentation using MS1 resolution at 60,000, MS2 resolution at 7500, intensity threshold at 2.0 × 104, and dynamic exclusion after two triggers for 10 s. MS2 fragmentation was completed first with HCD using stepped collision energies at 15, 30, and 45% and was followed on the next scan by CID fragmentation in assisted collision energy mode at 15, 30, and 45% with an activation Q of 0.25. MS1 scans used a custom AGC target of 25% in positive mode and a maximum injection time of 50 ms. MS2 scans used a standard AGC target and a maximum injection time of 22 ms. The total cycle time of the MS1 and ddMS2 scans was 0.6 s. Untargeted metabolomics data were collected using Orbitrap ID-X Tune Application (4.1.4244) and Thermo Scientific XCalibur (4.7.102.25). Untargeted metabolomics data were analyzed in Compound Discoverer (3.3.3.200, Thermo). Figures were prepared in GraphPad Prism 10.6.1(892) and FreeStyle™ 1.8 SP2 QF1 Version 1.8.65.0.
Absolute quantification and targeted polyamine metabolomics LC-MS
Samples for absolute quantification were normalized to a cell density of 20 McFarland units per mL before pelleting at max speed for 30 s. See “Sample Extraction” and “Polyamine Derivatization” for details on sample preparation. For absolute quantification of N1-APA, a concentration gradient of N1-APA standard (EN300-47248193, Enamine) was prepared in B. subtilis ∆speE (NOBb439) extracted lysate. N1-APA was suspended in HPLC-grade H2O at 1 mg/mL for subsequent serial dilutions. Log and half-log dilutions were performed from the starting concentration of 10,000 ng/mL for a total of 10 dilutions. The standard curve and cell-extracted metabolites were derivatized using isobutylchloroformate and analyzed using an Agilent 6470 LC-QQQ mass spectrometer coupled to an Agilent 1290 UHPLC system as previously reported38. Data was acquired using MassHunter Workstation Version 12.1(12.1.179) (Agilent). Skyline (64-bit) 24.1.0.414 was used to normalize peak integration and export the relative abundances. The absolute concentration of N1-APA was calculated in Microsoft Excel(version 2405) using linear regression analysis from one concentration gradient point above and below the sample relative abundances. Linear regression yielded an intercept of 1842.9 and a slope of 316.03. R2 = 0.97. The detection limit was defined as 5 x the process blank. This detection limit was selected to filter out the samples and concentration gradient points in which all three ions (215.0000 + , 159.0000 + , and 115.0000 +) were not detected. Samples that fell below the detection limit were filtered out and are labeled as Not Detected (ND). Figures were prepared in GraphPad Prism 10.6.1 (892).
Targeted metabolomics data analysis
Skyline (64-bit) 24.1.0.414 was used to normalize peak integration and export the relative abundances. Relative abundances were filtered and analyzed using R (4.5.0(2025-04-11 ucrt)) base packages and RStudio (2025.05.0.496). Relative abundances were filtered by 5x the process blank. Any value below the process blank filter is reported as Not Detected (ND). Unfiltered data were used for downstream statistical analyses. Data were log10 transformed and fit to a linear model using lm70,71. A one-way ANOVA was preformed using base R anova72. P-values for multiple comparisons between strains for each metabolite were calculated using pairwise.t.test62, followed by a Benjamini-Hochberg procedure (p.adjust.method = BH) to correct for FDR. Adjusted p-values are reported in Fig. 1F and Supplementary Fig. 7A. Code and additional package information is publicly available at the GitHub repo: https://github.com/nautakel/nauta_et_al_2025_polyamine_metabolomics.git.
Brood size count
Pristionchus pacificus PS312 grew to adulthood on NGM plates seeded with E. coli OP50 at 20 °C. Embryos were isolated using egg preparation solution and plated on NGMB plates seeded with the respective bacterial strains. Bacterial strains were grown overnight at 30 °C before seeding. Once the animals reached the L4 growth stage, 10 animals were singled onto NGMB seeded with the respective bacterial strains, prepared from overnight cultures grown at 30 °C. The plates were incubated at 20 °C for the duration of the experiment. Offspring were counted each day for approximately 5 days. Animals that died on day 1 or 2 were removed from the data set.
Cell culture
Bone marrow cells were isolated from C57BL/6 female mice femurs, RBC lysed, pooled (n = 3-6 mice), counted, and resuspended at 1 x 10^7/ml, and plated at 1.5 x 10^6 per well. To generate Bone Marrow Macrophages (BMMΦ), cells were cultured in RPMI supplemented with 10% FBS (Hyclone), 2mM L-glutamine (Gibco), 100 µ/mL penicillin/streptomycin (Gibco), with 20 ng/ml M-CSF (R&D; 416-ML-050) for 7 days. BMMΦ(IL4) were generated with 20 ng/ml IL-4 (Peprotech 214-14-20µg) overnight from day 7; N1-guanyl-1 7-diaminoheptane (200 µM) and N1-APA (200 µM) treatment began on day 7.
BMMΦ Flow cytometry
Following treatment on D7, BMMΦ were harvested, washed and stained for flow cytometry. Samples were incubated with Fc block and eFluor 506 Fixable Viability Dye (ThermoFisher) in PBS, followed by an antibody cocktail prepared in wash buffer (PBS with 1% FBS, 1 mM EDTA, and 0.05% NaN3)(See Supplementary Data 4 for a list of all antibodies). For intracellular staining, cells were fixed with IC fixation buffer (eBioscience/ThermoFisher) for 30 min, permeabilized using Permeabilization Buffer (eBioscience/ThermoFisher) and incubated for at least one hour with antibodies targeting intracellular proteins. Samples were acquired on the Cytek Aurora spectral cytometer and data analyzed using FlowJo v10.
BMMΦ Western blot for hypusine and eIF5A
After 24 hours of treatment, BMMΦ were removed from culture by washing with warm 1 x PBS followed by a wash with chilled 1xPBS and placed on ice for 5 min then gently scraped, resuspended and put into 15 ml falcon tubes. Cells were pelleted at 4 °C for 5 min at 400 g. Pellets were lysed with 1x CHAPS lysis buffer ( + PI; + PMSF) freeze, thawed twice at −80°C, sonicated by 10 seconds in an ice slurry followed by 1 min rest (repeated 5 times). Supernatant was clarified by centrifuging for 15 minutes 4 °C at 17,000 x g. Lysed samples were stored at − 80 °C. Protein quantification was done with the Pierce 660 nm Protein Assay Reagent (ThermoFisher; 22660); samples were diluted 1:2 in 1 x CHAPS and read on a plate reader along with Pre-diluted Protein Assay Standards: Bovine Serum Albumin Set (ThermoFisher; 23208). For Hypusination and eIF5A blots,1 to 5 µg of protein for each sample was removed, brought up to volume with 1xCHAPS and 6 x LB. Samples were heated at 70 °C for 10 min. For Western Blot, 4–20% mini-protean gels were used and ran at 70 V for 30 min and 100 V until desired (30 min). Proteins were transferred to PVDF membranes, blocked for 1 hr 4 °C in 5% milk. Membranes were incubated with primary antibodies for anti-hypusine (1:10000) (Millipore ABS1064-I-25ul) and anti-eIF-5a (1:10000) (BD Biosciences 611976) overnight at 4°C in 5% BSA + TBST. Secondary antibodies, 1 hr RT 1:10000 in 5% milk + TBST. Blots were imaged using a BioRad Chemidoc MP Imaging system type 2.0.0.9, software version 3.0.1.14. See Supplementary Data 4 for a list of all antibodies.
Proteomics
Proteins were extracted from C. elegans samples using a mortar and pestle, followed by Bead Ruptor and sonication. The resulting supernatant was collected, and the protein concentration was measured using the Pierce BCA Protein Assay. The extracted protein was then subjected to reduction and alkylation before SDS removal using the S-Trap platform. The proteins were digested with Trypsin/Lys-C and then desalted using C18 micro spin columns before injection onto an Orbitrap Eclipse attached to a Vanquish Neo nano UPLC (Thermo Scientific) for Data-interdependent Acquisition (DIA) Analysis. The DIA approach provided high reproducibility and consistency of protein quantification across different samples and experiments. In addition, DIA can enable the detection of low-abundance proteins that may be missed using other conventional methods, such as Data-dependent acquisition. Thermo Scientific XCalibur (4.7.102.25). The resulting data was analyzed using Spectronaut 17 (Biognosys) for protein quantification.
Proteomics analysis
Before any analysis is performed, data are log2 transformed to improve the normality of model residuals. Before removing proteins with missing values, we looked for group-specific patterns in the missingness. Any proteins with group-specific missingness patterns were removed from the analysis. In addition, proteins missing from more than 75% of samples were removed from the analysis. The remaining missing values were imputed using quantile regression imputation, which is designed for left-censored, heteroscedastic, non-normal missing data (impute.QRILC).
Differential abundance is performed using limma 3.60.6 and eBayes. Limma was used with default parameters and was followed by eBayes with both the robust and trend options set to ‘TRUE’. All p-values were adjusted using the Benjamini-Hochberg method to control for the FDR. The results were evaluated for accuracy by exploring plots of the raw data of the top results from the limma eBayes analysis. The volcano plot (Fig. 3E) is created via ggplot2_3.5.1 using the results of the limma eBayes analysis. The proteomics code and additional detailed package information is publicly available on GitHub (https://github.com/vari-bbc/Nauta_2025).
Seahorse assay
Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of cultured BMMΦ were measured using the Seahorse XF96 Extracellular Flux Analyzer. Briefly, 5 x 104 BMMΦ were seeded to XF96 plates on day 7 of culture. After overnight rest, cells were treated with cytokine and N1-APA for 24 h. Following treatment, cellular bioenergetics were assessed using the Seahorse XF Cell Mito Stress Test Kit with sequential addition of 1.5 μM oligomycin, 3 μM fluoro-carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), 0.5 μM rotenone/antimycin A, and 10 mM monensin. Measurements of ECAR and OCR were done on the Seahorse XF96 Extracellular Flux Analyzer. Basal mitochondrial respiration = (Basal OCR) – (Non-mitochondrial respiration (post rotenone/antimycin A)).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Source data are provided with this paper. The raw proteomics data generated in this study have been deposited in the MassIVE database under accession code MSV00009965473 (https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?accession=MSV000099654). The metabolomics data generated in this study have been deposited in the MassIVE database under accession codes MSV000099660(Fig. 1F)74, MSV000099661(Fig. 2A, B)75, MSV000099662(Fig. 2E, F)76, MSV000099663(Fig. 2E, F)77, and MSV000099664(Fig. 2H and Supplementary Fig. 7A)78 (https://massive.ucsd.edu/ProteoSAFe/static/massive.jsp). RAW and LUT image files are available in the Source Data file. Source data are provided in this paper.
References
Chen, D., Shao, Q., Yin, L., Younis, A. & Zheng, B. Polyamine function in plants: metabolism, regulation on development, and roles in abiotic stress responses. Front. Plant Sci. 9, https://doi.org/10.3389/fpls.2018.01945 (2019).
Watanabe, S., Kusama-Eguchi, K., Kobayashi, H. & Igarashi, K. Estimation of polyamine binding to macromolecules and ATP in bovine lymphocytes and rat liver. J. Biol. Chem. 266, 20803–20809 (1991).
Bachrach, U., Wang, Y.-C. & Tabib, A. Polyamines: new cues in cellular signal transduction. Physiology 16, 106–109 (2001).
Alcázar, R. et al. Polyamines: molecules with regulatory functions in plant abiotic stress tolerance. Planta 231, 1237–1249 (2010).
Rao, J. N. Polyamines regulate intestinal epithelial restitution through TRPC1-mediated Ca2+ signaling by differentially modulating STIM1 and STIM2. Am. J. Physiol. Cell Physiol. 303, https://doi.org/10.1152/ajpcell.00120.2012 (2012).
Toro-Funes, N., Bosch-Fusté, J., Veciana-Nogués, M. T., Izquierdo-Pulido, M. & Vidal-Carou, M. C. In vitro antioxidant activity of dietary polyamines. Food Res. Int. 51, 141–147 (2013).
Muñoz-Esparza, N. C. et al. Polyamines in food. Front. Nutr. 6, https://doi.org/10.3389/fnut.2019.00108 (2019).
Acosta-Andrade, C. et al. Polyamine-RNA-membrane interactions: From the past to the future in biology. Colloids Surf. B Biointerfaces 155, 173–181 (2017).
Dever, T. E. & Ivanov, I. P. Roles of polyamines in translation. J. Biol. Chem. 293, 18719–18729 (2018).
Igarashi, K. & Kashiwagi, K. The functional role of polyamines in eukaryotic cells. Int. J. Biochem. Cell Biol. 107, 104–115 (2019).
van Veen, S. et al. Cellular function and pathological role of ATP13A2 and related P-type transport ATPases in Parkinson’s disease and other neurological disorders. Front. Mol. Neurosci. 7, https://doi.org/10.3389/fnmol.2014.00048 (2014).
Vrijsen, S. et al. ATP13A2-mediated endo-lysosomal polyamine export counters mitochondrial oxidative stress. Proc. Natl. Acad. Sci. USA 117, 31198–31207 (2020).
Dang, T. et al. ATP13A2 protects dopaminergic neurons in Parkinson’s disease: from biology to pathology. J. Biomed. Res. 36, 98–108 (2022).
Hamouda, N. N. et al. ATP13A3 is a major component of the enigmatic mammalian polyamine transport system. J. Biol. Chem. 296, 100182 (2020).
Colegio, O. R. et al. Functional polarization of tumour-associated macrophages by tumour-derived lactic acid. Nature 513, 559–563 (2014).
Chang, C.-I., Liao, J. C. & Kuo, L. Macrophage Arginase Promotes Tumor Cell Growth and Suppresses Nitric Oxide-mediated Tumor Cytotoxicity1. Cancer Res. 61, 1100–1106 (2001).
Li, C. et al. Spermine synthase deficiency causes lysosomal dysfunction and oxidative stress in models of Snyder-Robinson syndrome. Nat. Commun. 8, 1257 (2017).
Bachmann, A. S., VanSickle, E. A., Michael, J., Vipond, M. & Bupp, C. P. Bachmann-Bupp syndrome and treatment. Dev. Med. Child Neurol. 66, 445–455 (2024).
Ramos-Molina, B., Queipo-Ortuño, M. I., Lambertos, A., Tinahones, F. J. & Peñafiel, R. Dietary and gut microbiota polyamines in obesity- and age-related diseases. Front. Nutr. 6, https://doi.org/10.3389/fnut.2019.00024 (2019).
Feng, Y.-L. et al. Microbiome–metabolomics reveals gut microbiota associated with glycine-conjugated metabolites and polyamine metabolism in chronic kidney disease. Cell. Mol. Life Sci. 76, 4961–4978 (2019).
Mendez, R. et al. Microbial dysbiosis and polyamine metabolism as predictive markers for early detection of pancreatic cancer. Carcinogenesis 41, 561–570 (2020).
Matsumoto, M., Kakizoe, K. & Benno, Y. Comparison of fecal microbiota and polyamine concentration in adult patients with intractable atopic dermatitis and healthy adults. Microbiol. Immunol. 51, 37–46 (2007).
Postler, T. S. & Ghosh, S. Understanding the Holobiont: How microbial metabolites affect human health and shape the immune system. Cell Metab. 26, 110–130 (2017).
Matsumoto, M., Kurihara, S., Kibe, R., Ashida, H. & Benno, Y. Longevity in mice is promoted by probiotic-induced suppression of colonic senescence dependent on upregulation of gut bacterial polyamine production. PLoS ONE 6, e23652 (2011).
Nakamura, A. et al. Symbiotic polyamine metabolism regulates epithelial proliferation and macrophage differentiation in the colon. Nat. Commun. 12, 2105 (2021).
Pryor, R. et al. Host-microbe-drug-nutrient screen identifies bacterial effectors of metformin therapy. Cell 178, 1299–1312 (2019).
Li, X. et al. Uncovering de novo polyamine biosynthesis in the gut microbiome and its alteration in inflammatory bowel disease. Gut Microbes 17, 2464225 (2025).
Johnson, C. H. et al. Metabolism links bacterial biofilms and colon carcinogenesis. Cell Metab. 21, 891–897 (2015).
Heinick, A. et al. Caenorhabditis elegans P5B-type ATPase CATP-5 operates in polyamine transport and is crucial for norspermidine-mediated suppression of RNA interference. FASEB J. 24, 206–217 (2010).
Hanfrey, C. C. et al. Alternative spermidine Biosynthetic route is critical for growth of campylobacter jejuni and is the dominant polyamine pathway in human gut microbiota *. J. Biol. Chem. 286, 43301–43312 (2011).
Zhang, C. et al. Research progress on the microbial metabolism and transport of polyamines and their roles in animal gut homeostasis. J. Anim. Sci. Biotechnol. 16, 57 (2025).
Zhernakova, A. et al. Population-based metagenomics analysis reveals markers for gut microbiome composition and diversity. Science 352, 565–569 (2016).
Nakamura, A. Ooga,Takushi & and Matsumoto, M. Intestinal luminal putrescine is produced by collective biosynthetic pathways of the commensal microbiome. Gut Microbes 10, 159–171 (2019).
Kumbhari, A. et al. Discovery of disease-adapted bacterial lineages in inflammatory bowel diseases: Cell Host Microbe. 32, https://www.cell.com/cell-host-microbe/fulltext/S1931-3128(24)00194-X (2019).
Koo, B.-M. et al. Construction and analysis of two genome-scale deletion libraries for Bacillus subtilis. Cell Syst. 4, 291–305 (2017).
Sekowska, A., Bertin, P. & Danchin, A. Characterization of polyamine synthesis pathway in Bacillus subtilis 168. Mol. Microbiol. 29, 851–858 (1998).
Sekowska, A., Danchin, A. & Risler, J.-L. Phylogeny of related functions: the case of polyamine biosynthetic enzymes. Microbiology 146, 1815–1828 (2000).
Isaguirre, C., Gendjar, M., Nauta, K.M., Burton, NO, Sheldon RD. Polyamine quantitation by LC-MS using isobutyl chloroformate derivatives. Methods Enzymol. 715, 437–458. (2025).
Ohnuma, M. et al. N1-Aminopropylagmatine, a new Polyamine Produced as a Key Intermediate in Polyamine Biosynthesis of an Extreme Thermophile, Thermus thermophilus *. J. Biol. Chem. 280, 30073–30082 (2005).
Li, B. et al. Functional identification of bacterial spermine, thermospermine, norspermine, norspermidine, spermidine, and N1-aminopropylagmatine synthases. J. Biol. Chem. 300, (2024).
Tuchman, M., Rajagopal, B.S., McCann, M.T. & Malamy, M.H. Enhanced production of arginine and urea by genetically engineered Escherichia coli K-12 strains. Appl. Evviron. Microbiol. 63, 33–38 (1997).
Yoneda, T. et al. Compartment-specific perturbation of protein handling activates genes encoding mitochondrial chaperones. J. Cell Sci. 117, 4055–4066 (2004).
Pellegrino, M. W. et al. Mitochondrial UPR-regulated innate immunity provides resistance to pathogen infection. Nature 516, 414–417 (2014).
WormEnrichr. https://maayanlab.cloud/WormEnrichr/account.html.
Di, R., Zhang, H. & Lawton, M. A. Transcriptome analysis of C. elegans reveals novel targets for DON cytotoxicity. Toxins 10, 262 (2018).
Bin-Umer, M. A., McLaughlin, J. E., Basu, D., McCormick, S. & Tumer, N. E. Trichothecene mycotoxins inhibit mitochondrial translation—implication for the mechanism of toxicity. Toxins 3, 1484–1501 (2011).
Prabh, N. et al. Deep taxon sampling reveals the evolutionary dynamics of novel gene families in Pristionchus nematodes. Genome Res. 28, 1664–1674 (2018).
Rödelsperger, C. et al. Phylotranscriptomics of pristionchus nematodes reveals parallel gene loss in six hermaphroditic lineages. Curr. Biol. 28, 3123–3127 (2018).
Posttranslational synthesis of hypusine: evolutionary progression and specificity of the hypusine modification. Amino Acids. 33, 341–350 (2007).
Park, M. H. & Wolff, E. C. Hypusine, a polyamine-derived amino acid critical for eukaryotic translation. J. Biol. Chem. 293, 18710–18718 (2018).
Park, M. H., Kar, R. K., Banka, S., Ziegler, A. & Chung, W. K. Post-translational formation of hypusine in eIF5A: implications in human neurodevelopment. Amino Acids 54, 485–499 (2022).
Park, J.-H. et al. Molecular cloning, expression, and structural prediction of deoxyhypusine hydroxylase: A HEAT-repeat-containing metalloenzyme. Proc. Natl. Acad. Sci. USA 103, 51–56 (2006).
Puleston, D. J. et al. Polyamines and eIF5A hypusination modulate mitochondrial respiration and macrophage activation. Cell Metab. 30, 352–363 (2019).
Jakus, J., Wolff, E. C., Park, M. H. & Folk, J. E. Features of the spermidine-binding site of deoxyhypusine synthase as derived from inhibition studies. Effective inhibition by bis- and mono-guanylated diamines and polyamines. J. Biol. Chem. 268, 13151–13159 (1993).
Anderson-Baucum, E. et al. Deoxyhypusine synthase promotes a pro-inflammatory macrophage phenotype. Cell Metab. 33, 1883–1893 (2021).
Xi, H. et al. A bacterial spermidine biosynthetic pathway via carboxyaminopropylagmatine. Sci. Adv. 9, eadj9075 (2023).
Applebaum, D. M., Dunlap, J. C. & Morris, D. R. Comparison of the biosynthetic and biodegradative ornithine decarboxylases of Escherichia coli. Biochemistry 16, 1580–1584 (1977).
Applebaum, D., Sabo, D. L., Fischer, E. H. & Morris, D. R. Biodegradative ornithine decarboxylase of Escherichia coli. Purification, properties, and pyridoxal 5’-phosphate binding site. Biochemistry 14, 3675–3681 (2002).
Belitsky, B. R., Brill, J., Bremer, E. & Sonenshein, A. L. Multiple genes for the last step of proline biosynthesis in Bacillus subtilis. J. Bacteriol. 183, 4389–4392 (2001).
Tassoulas, L. J. & Wackett, L. P. Insights into the action of the pharmaceutical metformin: Targeted inhibition of the gut microbial enzyme agmatinase. iScience 27, 108900 (2024).
Korbie, D. J. & Mattick, J. S. Touchdown PCR for increased specificity and sensitivity in PCR amplification. Nat. Protoc. 3, 1452–1456 (2008).
pairwise.t.test function - RDocumentation. https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/pairwise.t.test.
Karp, P. D. et al. The BioCyc collection of microbial genomes and metabolic pathways. Brief. Bioinform. 20, 1085–1093 (2019).
Arnaud, M., Chastanet, A. & Débarbouillé, M. New vector for efficient allelic replacement in naturally nontransformable, low-GC-content, gram-positive bacteria. Appl. Environ. Microbiol. 70, 6887–6891 (2004).
Baba, T. et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol. Syst. Biol. https://doi.org/10.1038/msb4100050 (2006).
Thomason, L. C., Costantino, N. & Court, D. L. Genome Manipulation by P1 Transduction. Curr. Protoc. Mol. Biol. 79, 17.1–1.17.8 (2007).
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 9, 676–682 (2012).
Industry-leading vector graphics software. Adobe Illustrator. https://www.adobe.com/products/illustrator/campaign/pricing.html.
House, R. et al. A diverse proteome is present and enzymatically active in metabolite extracts. Nat. Commun. 15, 5796 (2024).
Wilkinson, G. N. & Rogers, C. E. Symbolic Description of Factorial Models for Analysis of Variance. J. R. Stat. Soc. Ser. C. Appl. Stat. 22, 392–399 (1973).
Chambers, J. M. & Hastie, T. J. Linear Models. in Statistical Models in S eds (Wadsworth & Brooks/Cole., 1992).
Chambers, J. M. & Hastie, T. J. Statistical Models in S. (Wadsworth & Brooks/Cole., 1992).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - proteomics for Fig. 3E. MassIVE MSV000099654 https://doi.org/10.25345/C5FB4X06K (2025).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - Fig. 1F. MassIVE MSV000099660 https://doi.org/10.25345/C5NV99Q31 (2025).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - Figs. 2A and 2B polyamine metabolomics. MassIVE MSV000099661 https://doi.org/10.25345/C5J38KX15 (2025).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - Figs. 2E and 2F. MassIVE MSV000099662 https://doi.org/10.25345/C5DF6KG85 (2025).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - Figs. 2E and 2F. MassIVE MSV000099663 https://doi.org/10.25345/C58P5VP60 (2025).
A noncanonical polyamine from bacteria antagonizes host mitochondrial function - Fig. 2H and SFig. 7A. MassIVE MSV000099664 https://doi.org/10.25345/C54X54W1R (2025).
Xiao, R. et al. RNAi Interrogation of dietary modulation of development, metabolism, behavior, and aging in C. elegans. Cell Rep. 11, 1123–1133 (2015).
Guérout-Fleury, A. M., Frandsen, N. & Stragier, P. Plasmids for ectopic integration in Bacillus subtilis. Gene 180, 57–61 (1996).
Acknowledgements
We would like to acknowledge Dr. Craig Ellermeier for sharing Bacillus strains and plasmids. This research was supported in part by the Van Andel Institute Mass Spectrometry Core (RRID:SCR_024903; Grand Rapids, MI), particularly Colt Capan, and the Metabolism and Nutrition (MeNu) Program (RRID: SCR_027494). We thank the Van Andel Research Institute Bioinformatics and Biostatistics Core for support (RRID: SCR_024762), particularly Emily Wolfrum. We thank D. Chandler for helpful comments on the manuscript text, and we thank Margene Brewer for administrative assistance. We thank the Caenorhabditis Genetics Stock Center (funded by the NIH National Center for Research Resources) and National BioResource Project (NRBP) for C. elegans strains, and WormBase as a repository of C. elegans data. This work was supported by funds from the NIH grant DP2DK139569.
Author information
Authors and Affiliations
Contributions
K. Nauta, D. Gates, and N. Burton formulated the project. K. Nauta, D. Gates, M. Mechan, X. Wang performed the C. elegans experiments. K. Nguyen maintained C. elegans strains and provided support. K. Nauta, M. Mechan, X. Wang, and S. Barton cloned the bacterial knock-out and rescue strains. C. Isaguirre, M. Gendjar, K. Nauta, and R. Sheldon developed and performed the mass spectrometry experiments and data analysis. C. Krawczyk and M. Weiland formulated the bone marrow macrophage experiments, and M. Weiland performed the experiments and downstream procedures. J. Cooper performed the proteomics experiments. K. Nauta, R. Sheldon, and N. Burton wrote the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Diversity and inclusion statement
All individuals involved in this study fulfilled the criteria for authorship required by Nature Portfolio journals and thus have been included as authors. Roles and responsibilities of all collaborators were agreed to in advance of the research. All efforts were made to ensure no one involved in this study was hindered from participation based on background, ethnicity, or gender. This research also does not result in any known stigmatization, incrimination, discrimination or personal risk to anyone involved in this study.
Peer review
Peer review information
Nature Communications thanks Mitsuharu Matsumoto, and the other anonymous reviewer for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Nauta, K.M., Gates, D.R., Weiland, M. et al. A noncanonical polyamine from bacteria antagonizes host mitochondrial function. Nat Commun 16, 11638 (2025). https://doi.org/10.1038/s41467-025-66499-w
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41467-025-66499-w
