Polyketide synthase-derived sphingolipids mediate microbiota protection against a bacterial pathogen in C. elegans

Abstract

Protection against pathogens is a major function of the gut microbiota. Although bacterial natural products have emerged as crucial components of host-microbiota interactions, their exact role in microbiota-mediated protection is largely unexplored. We addressed this knowledge gap with the nematode Caenorhabditis elegans and its microbiota isolate Pseudomonas fluorescens MYb115 that is known to protect against Bacillus thuringiensis (Bt) infection. We find that MYb115-mediated protection depends on sphingolipids (SLs) that are derived from an iterative type I polyketide synthase (PKS) cluster PfSgaAB, thereby revealing a non-canonical pathway for the production of bacterial SLs as secondary metabolites. SL production is common in eukaryotes but was thought to be limited to a few bacterial phyla that encode the serine palmitoyltransferase (SPT) enzyme, which catalyses the initial step in SL synthesis. We demonstrate that PfSgaB encodes a pyridoxal 5'-phosphate-dependent alpha-oxoamine synthase with SPT activity, and find homologous putative PKS clusters present across host-associated bacteria that are so far unknown SL producers. Moreover, we provide evidence that MYb115-derived SLs affect C. elegans defence against Bt infection by altering SL metabolism in the nematode host. This work establishes SLs as structural outputs of bacterial PKS and highlights the role of microbiota-derived SLs in host protection against pathogens.

Fig. 1. MYb115 PKS cluster-derived SLs mediate protection against B. thuringiensis infection.

A, B Survival proportion of C. elegans N2 on P. fluorescens MYb115 P_BADsga (A) or MYb115 P_BADnrpA (B) induced with arabinose (solid line) or in a non-induced state without arabinose supplementation (dashed line) 24 h post infection with B. thuringiensis Bt247. Bt407 was used as a non-pathogenic control. The data shown is representative of three independent runs with four replicates each (see Supplementary Data 1). C LC-MS chromatogram of MYb115 P_BADsga extracts from cultures with (solid line) and without (dashed line) arabinose supplementation. Upon induction with arabinose, three compounds (1–3) are produced. D Schematic representation of the MYb115 PKS gene cluster and its modifications. Polyketide synthase (PKS) SgaA, alpha-oxoamine synthase (AOS) SgaB and inducible arabinose promoter (P_BAD). E Survival proportion of N2 on E. coli OP50, MYb115, or MYb115 knockout mutants. C. elegans on both tested mutants MYb115 ΔsgaA, and MYb115 ΔsgaB were significantly more susceptible (p = 1.17E-09 or p = 2.00E-16, respectively) to infection with Bt247 than worms on wildtype MYb115. Means ± standard deviation (SD) of n = 4, are shown in survival assays (A, B, E), n = 3 in (F). Statistical analyses were carried out using the generalized linear model (GLM) framework with a binomial distribution. All tests were two-sided, and p-values were adjusted for multiple comparisons using the Bonferroni correction. Significance is indicated as ***p < 0.001. F Survival proportion of N2 on MYb115 ΔsgaA/P_vanCCsgaAB, which expresses SgaAB under the vanillic acid-inducible P_vanCC promoter on the pSEVA631 plasmid. Survival was assessed 24 h post-infection with Bt247, comparing vanillic acid-induced (solid line) and non-induced (dashed line) conditions (p = 3.53E-11). G LC-MS chromatogram of MYb115 wt, ΔsgaA and ΔsgaB. H Correlation of area under the C. elegans survival curve (AUC) and peak intensity, representing bacterial SL abundance. Each facet represents the correlation for a specific bacterial SL compound (1–6), with different bacterial treatments indicated by colour. Correlations were calculated using the two-sided Spearman method, correlation coefficients are shown with 95% confidence intervals. Source data and additional survival runs are provided in Supplementary Data 1.

Fig. 2. Proposed biosynthesis of P. fluorescens MYb115 PKS cluster PfSgaAB-derived SLs.

A Biosynthesis scheme of MYb115-derived PG-sphingolipids 4–6. The production of 3-ketodihydrosphinganines (KDSs) is catalysed by the iterative PKS (iPKS) PfSgaA and PLP-dependent serine palmitoyltransferase (SPT) PfSgaB. The reduction of KDSs to dihydrosphinganines 1–3 is presumably catalysed by the KDS reductase homologue PfSgaC. B PLP external aldimine formation following the addition of up to 10 mM L-serine (L-ser), monitored by UV–vis spectroscopy. External aldimine formation is signified by an increase in absorbance at 413 nm. C Schematic representation of PfSgaB-catalysed decarboxylative condensation between acyl-CoAs 7–9 and L-ser to give 3-ketodihydrosphinganines 10–12. D Relative activity of PfSgaB in the presence of C₁₆-CoA 7 and L-serine, L-alanine or glycine, determined using the DTNB assay (412 nm). UV–vis measurements were recorded after 20 min of incubation. Error bars represent the standard deviation of three technical replicates. All measurements were corrected for non-specific background absorbance. E Extracted ion (EI) chromatograms of PfSgaB-derived products 10–12, detected by LC/ESI-MS. F [M + H]⁺ ions of PfSgaB-derived products 10–12, detected by LC/ESI-MS. The theoretical m/z is shown for each product.

Fig. 3. Distribution of P. fluorescens MYb115 PKS cluster SgaAB homologues in bacteria.

The monomodular PKS (KW062_RS19805) and the alpha-oxoamine synthase (KW062_RS19800) in P. fluorescens MYb115 (NZ_CP078138) were searched against the NR NCBI database (https://www.ncbi.nlm.nih.gov/) using cblaster (1.8.1). A Five representative PKS cluster SgaAB homologs from various bacterial genera aligned and visualised using clinker. B Total distribution of 6101 PKS cluster SgaAB homologs across different bacterial genera. The width of each box represents the percentage of all identified PKS cluster SgaAB homologs, found in each bacterial genus are provided as source data in Supplementary Data 4.

Fig. 4. MYb115-mediated protection is independent of known C. elegans pathogen defense pathways.

A Transcriptional response of C. elegans to MYb115-derived SLs. Enrichment analysis of genes differentially regulated between worms exposed to SL-producing MYb115 and worms exposed to non-SL producing MYb115 ΔsgaA in the presence of pathogenic Bt247 (Supplementary Data 7). B, C Survival of p38 and JNK MAPK pathway mutants. Means ± standard deviation (SD) of n = 4 (p38 MAPK pathway (B)); n = 3 (kbg-1(ums3) survival (C)) are shown in all survival assays. Statistical analyses were carried out using the generalized linear model (GLM) framework with a binomial distribution. All tests were two-sided, and p-values were adjusted for multiple comparisons using the Bonferroni correction. Significance is indicated as, ***p < 0.001, **p < 0.01, *p < 0.05. All p-values can be found in Supplementary Data 8. nsy-1(ag3) and sek-1(km4) share the same N2 control since the experiment was conducted in parallel, with statistical analysis adjusted accordingly, as highlighted in Supplementary Data 8. Source data are provided in Supplementary Data 8.

Fig. 5. MYb115-derived SL contribute to intestinal barrier protection.

Visualisation and quantification of vesicular structures following Bt247 infection. Worms were raised on either E. coli OP50, P. fluorescens MYb115 ΔsgaA or P. fluorescens MYb115 P_BADsga + arabinose for 72 h and then infected with Bt247. Confocal images of PGP-1::GFP were captured 4 h after exposure to Bt247 mixed with either OP50, MYb115 ΔsgaA or MYb115 P_BADsga + arabinose. For each worm all PGP-1::GFP positive vesicles were scored and categorised into either of the three groups “0 vesicles”, “$\le$ 10 vesicles” or “$>$ 10 vesicles”. Representative images of worms are shown, highlighting magnified regions of PGP-1::GFP positive vesicles (indicated by white arrows) following Bt247 infection. Scale bar: 100 µm. The proportions of worms in each category are displayed as stacked bar plots for each replicate. Population size varied between 14 and 25 individuals (n = 3). Source data are provided in Supplementary Data 9.

Fig. 6. MYb115-derived SLs modulate host SL metabolism.

A Enriched metabolic pathways if the C. elegans (iCEL1314) metabolic model were identified following a comparison of worm models integrated with transcriptome data from worms treated with MYb115 with worms treated with MYb115 ΔsgaA. Significant reactions obtained by calculating two-sided p-values from linear regression models (data ~ treatment) of FVA centres and OFD data layers were used for Flux Enrichment Analaysis (FEA) against the background of all reactions within the iCEL1314 C. elegans metabolic model. Benjamini-Hochberg was applied only for FEA output due to high pathway/reaction collinearity. Source data are provided in Supplementary Data 10. B Reduced SL contents in worms exposed to MYb115 compared to worms exposed to MYb115 ΔsgaA. The heatmap shows the differences in ratio of detected SLs between the mean of MYb115 ΔsgaA and the mean of MYb115. The boxplot shows the difference in ratio of Sphingomyelin (t43:1) in worms exposed to MYb115 ΔsgaA and MYb115, all remaining boxplots can be found in Fig. S16. Boxplots display the median (line), the first and third quartiles (box edges), and whiskers extending to the smallest and largest values within 1.5× the interquartile range. Points beyond this range are shown as outliers. Statistical analysis was done with a two-sided Welch’s t- test (n = 5), * p-value < 0.05, ** p-value < 0.01. Dihydroceramides (DhCer), Ceramides (Cer), Sphingomyelins (SM), Hexosylceramides (HexCer), with hydroxylated fatty acyls (t) or non-hydroxylated fatty acyls (d), Hexosylceramides with phytosphingosine base and hydroxylated fatty acyls (HexCer(q)), monomethyl phosphoethanolamine glucosylceramide (mmPEGC(q)). Source data are provided in Supplementary Data 11.

Fig. 7. Modulations in C. elegans SL metabolism affect survival after Bt247 infection.

A Overview of SL metabolism in C. elegans. C. elegans produces sphingoid bases which are derived from a C17 iso-branched fatty acid and are thus structurally distinct from those of other animals with mainly straight-chain C18 bases. C. elegans SLs consist of a sphingoid base backbone derived from C15iso-CoA and serine, which is N-acylated with fatty acids of different lengths as well as different functional groups at the terminal hydroxyl group. Dihydroceramides (DhCers) are formed from C17iso sphinganine and fatty acids or 2-hydroxy fatty acids. Desaturation at the 4th carbon yields ceramides (Cers), which are the precursors of complex SLs such as sphingomyelin (SM) and glucosylceramide (HexCer). Mutants of SL metabolism genes in bold were tested in survival assays shown in (C). B Schematic survival comparing N2 wildtype (solid line) versus mutant strains (dashed lines), the difference of the area under the survival curve (AUC) is shaded in brown when the mutants are more susceptible to the infection than the control and in green when the mutants are more resistant to the infection. C Heatmap represents the ΔAUC of the survival of the C. elegans SL metabolism mutants versus average of the wildtype N2 strain. Each box represents an independent experiment, consisting of three to four technical replicates (individual bars). The intensity of the bar colour reflects the overall summary across all experiments, while the statistical analysis was performed separately for each experiment. Statistical analyses were carried out using the GLM framework with a binomial distribution. All tests were two-sided, and p-values were adjusted for multiple comparisons using the Bonferroni correction. Significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001. Each individual survival curve can be found in Fig. S17A, B. Source data and exact p-values are provided in Supplementary Data 12.