BMP-dependent mobilization of fatty acid metabolism promotes Caenorhabditis elegans survival on a bacterial pathogen

ABSTRACT

The bone morphogenetic proteins (BMPs) are secreted peptide ligands of the transforming growth factor beta (TGF-β) family, initially identified for their roles in development and differentiation across animal species. They are now increasingly recognized for their roles in physiology and infectious disease. In the nematode Caenorhabditis elegans, the BMP ligand DBL-1 controls fat metabolism and immune response, in addition to its roles in body size regulation and development. DBL-1 regulates classical aspects of innate immunity, including the induction of anti-microbial peptides. We theorized that BMP-dependent regulation of fat metabolism could also promote resilience against microbial pathogens. We found that exposure to a bacterial pathogen alters total fat stores, lipid droplet dynamics and lipid metabolism gene expression in a BMP-dependent manner. We further showed that fatty acid desaturation plays a major role in survival on a bacterial pathogen, whereas fatty acid β-oxidation plays a more minor role. We conclude that C. elegans mobilizes fatty acid metabolism in response to pathogen exposure to promote survival. Our investigation provides a framework to study potential metabolic interventions that could support therapeutics that are complementary to antibiotic strategies.

Editor's choice: Host metabolism may impact response to infection. In Caenorhabditis elegans, a conserved cell signaling pathway regulates fatty acid metabolism upon pathogen exposure, leading to improved host survival.

INTRODUCTION

Signaling pathways enable organisms to respond to environmental threats and avoid disease. The organismal phenotypes that result from the action or dysfunction of these pathways is determined by gene–environment interactions. An example of this is host–pathogen interactions, where organisms face many microbes in their environment, requiring the host immune system to defend against infection. Immunity includes both antibody-based adaptive immunity and innate immunity, which is the first line of defense. Most research on innate immunity has focused on mechanisms that reduce pathogen load, such as the regulation of anti-microbial peptides (AMPs) (Kim and Ewbank, 2018). However, less is known about the role of host metabolism in supporting survival independently of anti-bacterial responses.

Caenorhabditis elegans, a small free-living nematode, has been used as a model organism for decades owing to its short lifespan, easy laboratory maintenance, genetic tractability and physical features (Brenner, 1974; C. elegans Sequencing Consortium, 1998). The C. elegans diet consists of available bacteria in their environment, which in the laboratory is a non-pathogenic strain of Escherichia coli. This system is easily modified to study immunity, because the food source can be replaced with pathogenic bacteria, making it an excellent model system for the study of immunity (Nicholas and Hodgkin, 2002). In the wild, this soil-dwelling nematode encounters numerous pathogens, relying on its immune system for survival. Although C. elegans lack adaptive immunity, they have several other mechanisms for immune defense. These include pathogen-avoidance behaviors and innate immunity, including physical barriers and antimicrobial peptide expression. Another potential mechanism of defense is immune tolerance, which aims to reduce how an organism withstands the negative side-effects of an infection, rather than reducing the infection directly, but whether this mechanism contributes to C. elegans survival on pathogens has yet to be determined.

Immunity, like other aspects of metazoan physiology and development, is dependent on cell signaling. In particular, transforming growth factor-β (TGF-β) signaling in multicellular animals is widely conserved and has been shown to regulate many aspects of cell function (Robertis, 2008). Bone morphogenetic proteins (BMPs) are a major group within the TGF-β superfamily, first identified for regulating bone and cartilage development (Urist and Strates, 1971), but emerging as modulators of homeostasis. The BMP-like DBL-1 signaling pathway in C. elegans regulates innate immunity, lipid metabolism, body size and male tail development, among other functions (Gumienny and Savage-Dunn, 2013; Yamamoto and Savage-Dunn, 2023). This signaling pathway begins with the ligand DBL-1 (Morita et al., 1999; Suzuki et al., 1999), which binds a heterotetrameric receptor complex composed of the type I receptor SMA-6 (Krishna et al., 1999) and the type II receptor DAF-4 (Estevez et al., 1993). The signal is then transduced by the receptor-regulated Smads SMA-2 and SMA-3, and the common mediator Smad SMA-4 (Savage et al., 1996). This pathway was first identified as a major regulator of innate immunity when mutants of the DBL-1/BMP pathway exhibited decreased survival when exposed to the pathogenic bacteria Serratia marcescens (Mallo et al., 2002). Defects in immunity have been seen on many other pathogens, including bacteria E. coli, Enterococcus fecalis, Pseudomonas aeruginosa strain PA14, Salmonella enterica, Salmonella typhimurium strain SL1344, Photorhabdus luminescens and the nematophagous fungus Drechmeria coniospora (Ciccarelli et al., 2024a; Portal-Celhay et al., 2012; So et al., 2011; Tenor and Aballay, 2008; Zugasti and Ewbank, 2009). The DBL-1/BMP signaling pathway also regulates lipid metabolism in C. elegans, as BMP mutants show reduced fat stores (Clark et al., 2018, 2021; Yu et al., 2017).

Immune response in C. elegans has been shown to modulate lipid metabolism. A study found that animals exposed to some pathogenic bacteria undergo lipolysis and rapidly utilize lipid droplets, regulated by the nuclear hormone receptor NHR-49 (Dasgupta et al., 2020). This study observed a pathogen-specific alteration of fat stores in wild-type animals, with P. aeruginosa, Staphylococcus aureus, E. faecalis and Cryptococcus neoformans depleting neutral lipids after 8 h of exposure, whereas S. typhimurium after similar exposure did not. Another publication supported the involvement of NHR-49 in lipid regulation after infection, showing that it promotes immune resistance to P. aeruginosa, and its pathogen-induced targets include lipid metabolism genes and immune response genes (Naim et al., 2021). The transcription factor SKN-1 is also involved in regulating fat stores after infection. Previously, SKN-1 activation was shown to re-distribute lipids from somatic tissue to germline cells under oxidative and nutrient stress (Lynn et al., 2015). Subsequent work showed that SKN-1-dependent re-distribution of lipids also happens after pathogen exposure, and SKN-1 activation contributes to immune resistance to P. aeruginosa (Nhan et al., 2019). Further support for lipid metabolism gene expression changing in response to pathogen exposure was seen in another study, in which exposure to Bacillus thuringiensis resulted in differentially expressed genes enriched for lipid metabolism (Zárate-Potes et al., 2020).

Immunity-linked genes, genes upregulated in response to infection, are also often upregulated in response to lipid metabolism disruptions, both considered environmental stresses (Fanelli et al., 2023). The connecting mechanism is that these immunity-linked genes support secretory functions under stressful conditions. Certain lipid species have also been identified as necessary for a normal immune response. For example, oleic acid, a monounsaturated fatty acid (MUFA) was found to be required, and mutants deficient in oleic acid had decreased survival when exposed to bacterial pathogens, such as P. aeruginosa and S. marcescens (Anderson et al., 2019). Two 18-carbon polyunsaturated fatty acids (PUFAs), gamma-linoleic acid and stearidonic acid, were also found to be required to maintain p38 MAP kinase pathway activity and, when deficient, resulted in an increased susceptibility to infection (Nandakumar and Tan, 2008).

The DBL-1 regulation of immune response and lipid metabolism have, thus far, been seen as separate. However, here, we explore whether these two activities are connected and whether DBL-1 regulation of lipid metabolism has implications in immune response. We found that pathogen exposure affects fat storage, expression of genes involved in fatty acid desaturation and β-oxidation, and lipid droplet dynamics, in a DBL-1/BMP-dependent manner. Thus, BMP signaling regulates fatty acid metabolism after bacterial pathogen exposure for improved survival.

RESULTS

Pathogen exposure causes BMP-dependent alterations in fat storage

The first step in determining whether the immune response and lipid metabolism are connected was to identify whether there is a change in fat accumulation after pathogen exposure. Although we know that mutants of the DBL-1/BMP signaling pathway have low fat accumulation (Clark et al., 2018), whether the levels change after infection was unknown. dbl-1 mutants and wild-type controls at the fourth larval stage (L4) were placed on either non-pathogenic control E. coli bacteria or onto pathogenic bacteria, and, after 24 h, when animals became young adults, we conducted Oil Red O (ORO) fat staining to quantify their fat stores (Fig. 1A). We selected ORO because this stain accurately corresponds to triglyceride levels measured in biochemical studies (O'Rourke et al., 2009). The 24 h timepoint was selected as this is sufficient time for intact bacteria to proliferate in the intestinal lumen and for some antimicrobial peptides to increase in expression (Mallo et al., 2002). We focused on the anterior intestine based on our prior results (Clark et al., 2018) showing that ORO staining intensity is higher in the anterior intestine and gradually diminishes through the mid-body and posterior intestine, except for a small patch of higher intensity at the extreme posterior. Although the relationship between genotypes is maintained across these anatomical regions, the anterior intestine provides the highest resolution for ORO quantification. In wild-type animals, we found that 24 h exposure to a pathogen of moderate virulence, S. marcescens, resulted in a small, but not significant, decrease in fat accumulation (Fig. 1B). A slight decrease in fat accumulation was seen consistently, but significance varied, as it reached significance in some trials but not others (Fig. 1B). However, 24 h exposure to a pathogen of severe virulence, P. luminescens, resulted in a significant decrease in fat accumulation, with an average magnitude change of 10% (Fig. 1B). In contrast to wild-type animals, dbl-1 mutants had a dramatic and highly significant loss in total fat stores when exposed to S. marcescens and P. luminescens, with magnitudes of 25-50% (Fig. 1C). Images of the stained dbl-1 mutant animals show a substantial decrease in fat stores (Fig. 1E,F). To validate these results in another BMP signaling mutant, we repeated the experiment with sma-3/Smad mutant animals. Similar to dbl-1 mutants, sma-3 mutants also had a substantial decrease in fat stores when exposed to either S. marcescens or P. luminescens (Fig. 1D). We conclude that the BMP signaling pathway regulates lipid stores during infection with a pathogen. For all further experiments, we used S. marcescens as a pathogen because wild-type animals were better able to maintain levels of stored fat when exposed to S. marcescens than to P. luminescens.

Fig. 1.

Pathogen exposure causes BMP-dependent alterations in fat storage. Quantification is presented as normalized to the internal control, the same strain on non-pathogenic bacteria. For results normalized to the wild-type control, see Fig. S1. (A) Experimental schematic: L4 animals are transferred to either control or pathogenic bacteria. After 24 h, the animals are stained with Oil Red O (ORO). (B-D) Lipid accumulation of wild-type (WT; B), dbl-1 (C) and sma-3 (D) animals after 24 h pathogen exposure, stained with ORO. ORO experiments were repeated in triplicate on independent biological samples, with at least 15 animals per condition. Data points represent individual animals. Brown–Forsythe and Welch ANOVA multiple comparisons tests were used to determine significance. Boxes show second and third quartiles; whiskers show minimum and maximum values (E,F) Representative images of lipid accumulation in dbl-1 mutants after 24 h exposure to E. coli (E) or S. marcescens (F). ns, P>0.01; **P≤0.01; ***P≤0.001; ****P≤0.0001.

Alterations in fat storage have complex temporal dynamics and occur in the absence of pathogen ingestion

We were surprised to see the dramatic fat loss in the BMP mutants in only 24 h of pathogen exposure. We wondered whether 24 h was the earliest timepoint at which the decrease would be observed, and, thus, we investigated the temporal dynamics of lipid accumulation upon infection prior to 24 h. We repeated the ORO fat staining after pathogen exposure and assayed lipid levels at 6 h, 12 h, 18 h, and 24 h in wild types and dbl-1 mutants. In wild-type animals, across all time points, there was either no significant change in fat after pathogen exposure or there was a small decrease, indicating that wild-type animals are able to maintain lipid homeostasis (Fig. 2A). However, in dbl-1 mutants, we saw a more dynamic trend unfold (Fig. 2B). At 6 h, animals exposed to pathogen showed a highly significant increase in fat stores. At 12 h, there was still an increase in fat stores; however, the magnitude was smaller. At 18 h, it appeared there was no difference between animals on control bacteria and the pathogenic S. marcescens. At 24 h, we saw the significant decrease we had originally observed. We conclude that, in dbl-1 mutants, there is an immediate increase in fat stores in response to pathogen exposure, which are depleted by 24 h. These results suggest that, upon pathogen exposure, the loss of BMP signaling results in lipid dysregulation, perhaps having consequences for these animals' survival.

Fig. 2.

Alterations in fat storage have complex temporal dynamics and occur in the absence of pathogen ingestion. Quantification is presented as normalized to the internal control, the same strain at the same timepoint on non-pathogenic bacteria. (A,B) Lipid accumulation of wild type (A) and dbl-1 (B) after exposure to E. coli or S. marcescens for 6 h, 12 h, 18 h or 24 h. ORO experiments were repeated in triplicate on independent biological samples, with at least 30 animals per condition. Data points represent individual animals. Brown–Forsythe and Welch ANOVA multiple comparisons tests were used to determine significance. (C,D) Lipid accumulation of wild-type (C) and dbl-1 mutant (D) animals after 24 h exposure to E. coli bacteria, with either E. coli supernatant or S. marcescens supernatant, stained by ORO. ORO experiment was repeated in duplicate on independent biological samples, with 30 animals per condition. Data points represent individual animals. Boxes show second and third quartiles; whiskers show minimum and maximum values. Brown–Forsythe and Welch ANOVA multiple comparisons tests were used to determine significance. ns, P>0.01; *P≤0.05; **P≤0.01; ***P≤0.001; ****P≤0.0001.

Given that C. elegans are bacteriotrophs and consume bacteria in their environment as food, we wondered whether the dramatic changes in fat exhibited by BMP mutants after pathogen exposure were simply due to nutritional differences between E. coli and bacterial pathogens. To test this, we determined the effect of exposure to the supernatant of the bacterial culture, made by centrifuging and filtering the overnight bacterial culture. We expected that this would remove all bacterial cells, and thus the majority of the nutrition, while leaving supernatant with secreted peptides, a potential source of pathogenicity. All the plates had a lawn of E. coli as a food source for the animals, supplemented with either the E. coli or S. marcescens filtered supernatant on top (Mallo et al., 2002). We repeated the 24 h exposure using these plates, followed by ORO fat staining. Both wild-type and dbl-1 mutants had changes in fat accumulation following the filtered supernatant exposure, suggesting that it triggered an organismal response. Wild-type animals showed a decrease in fat levels when exposed to the S. marcescens supernatant, similar to that seen after 6 h of pathogen exposure (Fig. 2C). In dbl-1 mutants, we observed an increase in fat stores at 24 h, similar to that observed after 6-12 h of exposure to a lawn of S. marcescens (Fig. 2D). In both genotypes, 24 h of supernatant exposure caused changes similar to that seen in a shorter exposure to pathogen, consistent with a milder response to supernatant. Taken together, these results confirmed that the changes in fat accumulation cannot solely be due to nutritional changes.

RNA sequencing reveals that lipid metabolism genes are highly over-represented among differentially expressed genes in response to S. marcescens

Our results suggested an active organismal response to pathogen exposure leading to changes in lipid stores; thus, we hypothesized that transcriptional changes could be responsible for the changes in lipid accumulation observed after pathogen exposure. We conducted whole-animal RNA sequencing (RNA-seq) of wild types and dbl-1 mutants after 24 h exposure to either E. coli or S. marcescens. For each genotype, we analyzed gene expression in S. marcescens versus E. coli to identify genes induced or repressed upon pathogen exposure. We then compared the transcriptional response on pathogen between the two genotypes, as we aimed to identify BMP-dependent genes that were differentially expressed under pathogenic conditions. Differentially expressed genes (DEGs) were identified as those upregulated or downregulated in response to pathogen [false discovery rate (FDR)<0.05] (Tables S1 and S2). We did not see a strong pattern of activation or repression in either genotype (Fig. 3A,B). Some DEGs were shared in both genotypes, which are likely part of a conserved response that is not BMP dependent (Fig. 3C-E). The largest category of DEGs were those upregulated in wild types, but not in dbl-1 mutants (122 genes; Fig. 3C; N2 up in Fig. 4; Table S3). This category represents pathogen-induced genes that are BMP dependent.

Fig. 3.

Transcriptional changes in wild types and dbl-1 mutants after pathogen exposure. (A,B) Volcano plots of RNA-seq log fold change versus average expression for individual genes on pathogen compared to control bacteria, in wild-type animals (A) and dbl-1 mutants (B). Data points labeled in red are upregulated differentially expressed genes (DEGs), and data points labeled in blue are downregulated DEGs. (C,D) Venn diagrams of upregulated (C) and downregulated (D) DEGs, between wild types and dbl-1 mutants. (E) Heatmap of RNA-seq results, with each row being the relative expression of an individual gene. Genes with similar patterns of expression are grouped by K-means clustering. Figure generated using Morpheus by Broad Institute (RRID: SCR_017386).

Fig. 4.

Differential gene expression for key lipid metabolism genes in C. elegans. (A) WormCat analysis of identified gene category enrichments, showing results per genotype. As indicted in the key, the size of the circle represents the number of genes in the category, and the color represents the P-value. (B) WormCat visualization of identified gene category enrichments, showing results for direction-specific groups of DEGs. (C) Visualization of DEGs identified by RNA-seq in the context of key genes in C. elegans lipid metabolism, grouped by process. There are two bubbles to the right of every gene name, indicating whether they are upregulated or downregulated in wild-type animals (left) or dbl-1 mutants (right). Bubbles in red indicate upregulated DEGs and bubbles in blue indicate downregulated DEGs.

We investigated the DEGs further by employing WormCat (Holdorf et al., 2020) to identify enriched gene sets and to generate a heatmap for visualization. WormCat is a tool for gene set enrichment analysis specifically for C. elegans. The software structures annotations into three categories that become increasingly specific. We mostly focused on Category 2, which provides insight into molecular functions. We anticipated that the stress response, specifically the pathogen stress response, would be highly enriched, which we observed in both genotypes. We also hypothesized that several lipid metabolism genes would appear in the DEGs, given our previous experiments. Strikingly, lipid metabolism was the most enriched in both genotypes (Fig. 4A). If we examined a subset of the DEGs, the 122 genes that are BMP dependent, we observed that lipid metabolism was highly enriched under infection conditions, more so than any other molecular function category across all columns (Fig. 4B). This suggests that the first 24 h of pathogen exposure elicits a greater transcriptional response of genes involved in lipid metabolism genes than that of genes involved in the immune response.

We were interested in exploring which specific lipid metabolism genes were differentially expressed in wild-type and dbl-1 mutants. First, we created a schematic to help our visualization that contained many fundamental lipid metabolism processes, with rate-limiting or important genes listed below each process. We listed the lipid metabolism genes identified in WormCat and labeled our table with whether they were differentially expressed and, if so, in which direction (Fig. 4C). Some processes seemed to be unaffected by the short exposure to pathogen, such as glycolysis and the tricarboxylic acid cycle. In contrast, several processes were more affected, particularly in wild types, such as fatty acid desaturation and elongation, and β-oxidation. We hypothesized that, in the presence of BMP signaling, there is an increase in these processes in response to pathogen exposure.

Lipid droplets experience flux after pathogen exposure

We wanted to test whether changes to these processes were evident on a cellular level. Lipid droplets are sensitive to changes in both fatty acid desaturation and β-oxidation. In fatty acid desaturase mutants, lipid droplets are smaller and decreased in number, owing to impaired storage (Brock et al., 2007). In β-oxidation mutants, lipid droplets are larger owing to a disruption in fatty acid breakdown (Zhang et al., 2010). We selected the lipid droplet reporter DHS-3::GFP and crossed in the lipid droplet reporter to dbl-1 loss-of-function mutants. We compared DHS-3-positive lipid droplets in wild-type and dbl-1 animals after 24 h of bacteria exposure. We found that lipid droplets in the anterior intestine of wild-type animals on S. marcescens displayed a significant increase in the number of lipid droplets, but no significant change in droplet diameter (Fig. 5A,C). Because the posterior intestine contains a region of high lipid droplet density (Fig. 5E), we also analyzed this anatomical region. In contrast to the anterior intestine, there was no significant change in the number of lipid droplets in the posterior intestine, but there was an increase in droplet diameter (Fig. 5B,D,F). dbl-1 animals displayed no significant change in the anterior or posterior intestine, neither in lipid droplet quantity nor diameter (Fig. 5A-D,G,H). We therefore conclude that lipid droplet dynamics seen in wild-type animals are dependent on DBL-1 activity. We were surprised that pathogen exposure did not cause a reduction in lipid droplet number in dbl-1 mutants. Because DHS-3 is a dehydrogenase that functions in β-oxidation, it is possible that expression of DHS-3::GFP in these strains lowers the baseline level of lipid droplets such that further depletion in dbl-1 mutants is not easily quantifiable.

Fig. 5.

Lipid droplet dynamics are in flux after pathogen exposure. (A,B) Box-and-whisker plots of the number of lipid droplets per 400 µm² in the anterior (A) and posterior (B) intestine. Data points represent individual animals. Brown–Forsythe and Welch ANOVA multiple comparisons tests were used to determine significance. Confocal experiment was repeated in duplicate on independent biological samples, with five animals per condition. (C,D) Box-and-whisker plots of lipid droplet diameter (µm) in the anterior (C) and posterior (D) intestine. Data points represent individual animals. Brown–Forsythe and Welch ANOVA multiple comparisons tests were used to determine significance. In A-D, boxes show second and third quartiles; whiskers show minimum and maximum values. (E,F) Representative confocal images of DHS-3::GFP after 24 h of control bacteria exposure (E) and S. marcescens exposure (F). Red arrowheads point to enlarged lipid droplets. Scale bars: 10 µm. (G,H) Representative confocal images of dbl-1; DHS-3::GFP after 24 h of control bacteria exposure (G) and S. marcescens exposure (H). Scale bars: 10 µm. ns, P>0.01; *P≤0.05; **P≤0.01; ***P≤0.001; ****P≤0.0001.

β-oxidation plays a limited role in pathogen survival

We had found that pathogen exposure results in a lipid metabolism response, which upregulates both fatty acid desaturation and β-oxidation, and causes significant changes in lipid droplet dynamics. We next wanted to determine whether these changes impact survival on pathogenic bacteria. We first focused on β-oxidation, which we reasoned could be upregulated to convert lipid stores to the energy necessary for the immune response, as previously suggested (Dasgupta et al., 2020). We tested whether the impairment of key β-oxidation genes impacted C. elegans survival. We investigated three key β-oxidation genes: maoc-1, daf-22 and dhs-28. Mutations in any of these three genes result in an increase in lipid droplet size (Li et al., 2016). maoc-1 mutant animals showed minimal defects in survival during a S. marcescens survival assay (Fig. 6A). daf-22 and dhs-28 mutants, by contrast, had significant defects in survival (Fig. 6B,C). maoc-1 encodes the enzyme at the upstream rate-limiting step, while DAF-22 and DHS-28 function downstream of MAOC-1 in β-oxidation. DAF-22 and DHS-28 also have pleiotropic roles; for example, DAF-22 and DHS-28 are both involved in the synthesis of the dauer pheromone (Butcher et al., 2009). In addition, both daf-22 and dhs-28 have reduced survival on E. coli (Joo et al., 2009; Park and Paik, 2017), so the survival defect may not be specific to pathogen sensitivity. If β-oxidation is needed for energy generation in response to the infection, then we would expect that ATP levels would be lowered upon infection, similarly to the effect we see on fat stores. We quantified ATP levels in each experimental condition, after 24 h of pathogen exposure. Owing to the body size phenotype of dbl-1 mutants, we normalized each ATP concentration to the concentration of protein (PRO) in that sample. We found that there was no significant change in ATP/PRO in either genotype (Fig. 6D). Thus, ATP levels are maintained to a similar degree in wild-type and mutant animals, despite the significant differences in lipid metabolism between these two genotypes.

Fig. 6.

Fatty acid β-oxidation plays a minor role in survival after pathogen exposure. (A) Survival analysis of wild-type animals and maoc-1 mutants on S. marcescens bacteria. n values: wild type (89), maoc-1 (83). One representative trial of three replicates shown. (B) Survival analysis of wild-type, dbl-1 and daf-22 animals on S. marcescens bacteria. n values: wild type (106), dbl-1 (106), daf-22 (101). One representative trial of three replicates shown. (C) Survival analysis of wild-type, dbl-1 and dhs-28 animals on S. marcescens bacteria. n values: wild type (106), dbl-1 (106), dhs-28 (93). One representative trial of three replicates shown. (D) Ratio of ATP concentration to protein (PRO) concentration in wild types and dbl-1 mutants after 24 h exposure to either control E. coli or pathogenic S. marcescens. The experiment was conducted with three biological replicates for wild types and four biological replicates for dbl-1 mutants. Each sample contained ∼1000 age-synchronized animals. ns, P>0.01; *P≤0.05; ***P≤0.001; ****P≤0.0001. Black denotes significance relative to wild-type control; pink denotes significance relative to dbl-1.

Fatty acid desaturases promote survival in response to pathogen

Our results demonstrated that β-oxidation does not play a major role in the survival of animals after pathogen exposure; thus, we next tested the role of genes involved in lipid synthesis. We focused on genes that function in fatty acid desaturation, hypothesizing that these genes are upregulated in wild types to increase lipid synthesis under pathogenic conditions. FAT-6 and FAT-7 encode redundant Δ9 desaturases and are responsible for converting stearic acid into oleic acid, a rate-limiting step in lipid synthesis (Brock et al., 2006). fat-6;fat-7 mutants showed a severe decrease in survival when exposed to pathogen, very similar to the level observed in dbl-1 mutants (Fig. 7A). These mutants are defective in converting stearic acid to oleic acid (Brock et al., 2007); thus, we hypothesized that supplementing animals with oleic acid would rescue the survival defect. In wild types, the addition of 0.8 mM oleic acid had no effect on survival (Fig. 7B). However, dbl-1 mutants displayed improved survival when exposed to pathogen, and the addition of 0.8 mM oleic acid improved survival, consistent with our hypothesis (Fig. 7C).

Fig. 7.

Fatty acid desaturation plays a major role in survival after pathogen exposure. (A) Survival analysis of wild-type, dbl-1 and fat-6;fat-7 animals on S. marcescens bacteria. n values: wild type (100), dbl-1 (74), fat-6;fat-7 (80). One representative trial of three replicates shown. (B,C) Survival analysis of wild-type (B) and dbl-1 animals (C) on S. marcescens bacteria with and without supplementation of 0.8 mM oleic acid (OA). n values: wild-type control (78), WT OA (74), dbl-1 control (62), dbl-1 OA (48). One representative trial of two replicates shown. (D) Survival analysis of wild-type, fat-1 and fat-4 fat-1 animals on S. marcescens bacteria. n values: wild type (110), fat-1 (118), fat-4 fat-1 (94). One representative trial of two replicates shown. (E,F) Survival analysis of wild-type (E) and dbl-1 (F) animals on S. marcescens bacteria with and without supplementation of 0.8 mM eicosapentanoic acid (EPA) or 0.8 mM arachidonic acid (AA). n values: wild-type control (87), wild type EPA (84), wild type AA (70), dbl-1 control (90), dbl-1 EPA (94), dbl-1 AA (94). One representative trial of two replicates shown. ns, P>0.01; *P≤0.05; **P≤0.01; ****P≤0.0001. Black denotes significance relative to wild-type control; pink denotes significance relative to dbl-1; blue denotes significance relative to fat-1.

Our experiments suggest that MUFAs can partially, but not fully, rescue the survival of BMP mutants. We considered the possibility that PUFAs are also required for survival upon exposure to pathogen. In support of that, FAT-4 and several elongases are also induced in animals exposed to pathogen, and this response is BMP dependent. We chose to look at the fat-4 fat-1 double mutant, which would eliminate most PUFAs. We found that fat-4 fat-1 had a significant defect in survival after pathogen exposure (Fig. 7D). We concurrently assayed the survival of fat-4 fat-1 and fat-6;fat-7 mutants to determine whether one was more severe, and we found no significant difference between the two double mutants (Fig. S2). We conclude that disruption at any point in fatty acid desaturation has a strong impact on survival to pathogen exposure. We were curious whether supplementing with single PUFAs could also rescue the survival defect. Because FAT-4 is a Δ5 desaturase that produces the PUFAs arachidonic acid and eicosapentanoic acid (Watts and Browse, 2002), we supplemented with 0.8 mM arachidonic acid or 0.8 mM eicosapentanoic acid. Surprisingly, we observed no change in the survival of wild-type or dbl-1 animals (Fig. 7E,F). Therefore, unlike MUFAs, single PUFAs are not sufficient for rescuing the impaired immune response of dbl-1 animals.

DISCUSSION

We have found that, after a short exposure to the bacterial pathogen Serratia marcescens, C. elegans undergo BMP-dependent changes in lipid metabolism flux, both in the synthesis and breakdown of lipids. We find that these changes are associated with the induction of genes encoding proteins involved in β-oxidation and fatty acid desaturation. Furthermore, pathogen exposure of wild-type animals induces an increase in lipid droplet diameter, but a decrease in lipid droplet number. In wild-type animals, the net effect of these two changes is the maintenance of total fat stores, whereas dbl-1/BMP loss-of-function mutants exhibit first an increase and then a decrease in fat stores. Because both genetic backgrounds maintain a constant concentration of ATP following pathogen exposure, we believe that these changes may allow for the generation of MUFAs rather than the generation of ATP. Interestingly, dbl-1 mutants had reduced ATP content compared to wild types, consistent with mitochondrial defects, as previously reported (Vora et al., 2025; Yu et al., 2017). Another BMP-like ligand, TIG-2, has been shown to have reduced ATP levels, supporting that BMP-like ligands may have reduced mitochondrial function (Cheng et al., 2022). Finally, a recent report showed that sma-4/Smad acts downstream of mitochondrial translation inhibition to regulate lipid remodeling and survival on pathogen, further underscoring the complex web of interactions between these functions (Hu et al., 2025).

Our findings suggest that the changes in fat homeostasis likely occur to support the generation of MUFAs. In support of this, supplementation of dbl-1 mutants with the MUFA oleic acid partially rescues survival on S. marcescens. Consistent with our results, oleic acid was found to be required for a normal immune response, with fat-6;fat-7 mutants deficient in oleic acid having decreased survival when exposed to bacterial pathogens, such as P. aeruginosa and S. marcescens (Anderson et al., 2019). However, supplementation with the individual PUFAs eicosapentanoic acid and arachidonic acid did not rescue the pathogen survival phenotype. This suggests that survival may result from MUFAs or their products and not from PUFAs or their products (which include eicosanoids). Support for this hypothesis is seen in the RNA-seq results, which showed a decrease in FAT-3 expression on pathogen exposure in both wild types and dbl-1 mutants, indicating that it may be difficult to generate PUFAs under these conditions. One possible role for MUFAs is to serve as fatty acid ligands for a nuclear hormone receptor, such as NHR-49 (Tillman et al., 2019).

An organism's immune response employs many strategies in concert to fight illness and infection. These strategies can be clustered into three approaches: pathogen avoidance, immune resistance and immune tolerance (Medzhitov et al., 2012). Pathogen avoidance occurs prior to the organism making contact with a pathogen and aims to reduce the risk of exposure to infection. This typically manifests as a physical distancing of an organism from a potential pathogen. Early work in pathogen avoidance can be attributed to rodent models and even wild populations of lobster (Behringer et al., 2006; Kavaliers et al., 2004). In humans, the emotion of disgust is central to pathogen avoidance, as this core emotion is triggered by potential pathogens or pathogen-harboring substances (Curtis, 2014). In C. elegans, pathogen avoidance manifests as animals physically distancing themselves from pathogenic bacteria in their Petri dish environment, often climbing up the plastic sides and desiccating or burrowing into the agar. dbl-1 mutants have increased avoidance of E. coli, suggesting that the standard laboratory food source for C. elegans may have increased pathogenicity in these animals compared to that in wild-type animals (Madhu et al., 2023; Olofsson, 2014). dbl-1 mutants also show increased avoidance to three Gram-negative bacteria, compared to wild-type animals (Madhu et al., 2023), demonstrating that the DBL-1 pathway is required to suppress avoidance behavior.

The next strategy is immune resistance, which encompasses most traditional notions of disease fighting and can be found, to some extent, in all organisms. Immune resistance aims to reduce the pathogen burden once an infection is already established. This approach includes both innate immunity, such as physical barriers (skin, etc.) and the upregulation of AMPs, as well as adaptive immunity, such as antibodies. C. elegans only have innate immunity; thus, the primary method of resistance is the upregulation of AMPs in response to pathogen exposure. DBL-1/BMP signaling regulates the expression of many immune response genes, including those encoding lectins, lysozymes, lipases, P-glycoproteins of the ATP-binding cassette transporter family, caenacin AMPs and saposin-like proteins (Alper et al., 2007; Liang et al., 2007; Madhu et al., 2020; Mallo et al., 2002; Mochii et al., 1999; Roberts et al., 2010; Zugasti and Ewbank, 2009). Among these, caenacins play a critical role in the immune response and are induced upon infection. Notably, DBL-1 signaling promotes cnc-2 expression in the epidermis in a dose-dependent manner (Zugasti and Ewbank, 2009). Recent studies also revealed that CNC-2, along with another AMP, ABF-2, are regulated by SMA-3 activity in the pharynx (Ciccarelli et al., 2024a). Conversely, DBL-1 signaling negatively regulates the expression of the saposin-like protein SPP-9 (Madhu et al., 2020; Roberts et al., 2010).

The last strategy in the immune response, and the least understood, is immune tolerance, which aims to reduce the negative impacts of infection on host fitness. While the presence of a pathogen in a host has direct consequences such as cell death, there are also indirect consequences that can hinder an effective immune response, such as high inflammation. Foundational work in maize and wheat (Caldwell et al., 1958; Schafer, 1971), as well as in rodent models, has differentiated tolerance from resistance (Ayres and Schneider, 2008; Råberg et al., 2009; Read et al., 2008). This work allowed the hypothesis to form that some genotypes are more capable of withstanding the side-effects of infection and thus more tolerant. In C. elegans, processes with a role in immune tolerance may include the microbiome and lipid metabolism; however, studies on these topics are less abundant than those on immune resistance. In this study, we sought to explore whether DBL-1 signaling regulates lipid metabolism under pathogenic conditions and whether this is protective, potentially relating its effects to immune tolerance.

Based on our results, we see two possible models, which are not mutually exclusive, for the identified changes in lipid metabolism: (1) the lipid changes feed into the immune response in a way that is intended to directly combat bacteria, perhaps through some sort of systemic signaling, or (2) the lipid changes contribute to immune tolerance. To differentiate between these models, expression levels of AMPs, or bacterial load, in fat metabolism mutants could be determined.

In conclusion, our research has established a direct link between lipid metabolism and survival of C. elegans on pathogen, revealing that BMP-dependent regulation of lipid stores contributes to S. marcescens resistance. The connection between the immune response and lipid metabolism is likely to be conserved. Our findings are consistent with work in Drosophila melanogaster, where a study found that infection activates mobilization of host lipid stores, improving survival (Deshpande et al., 2022a,b). Similar observations have been made in grapevines, in which lipid signaling regulates pathogen response (Laureano et al., 2021), demonstrating how widespread the relationship may be across phylogenetic kingdoms. Furthermore, in mammalian cells, there is an established requirement for lipids in fighting infection. Adipose tissue has been identified as a key contributor to the immune system by storing immune cells (Desruisseaux et al., 2007). Individual adipocytes have been implicated owing to their potential regulatory role on the immune system through the secretion of hormones. Fat cells also dynamically move to wound sites and act collaboratively with macrophages to prevent infection (Franz et al., 2018). This relationship may explain the findings from human patients, as individuals that have metabolic syndromes, including 450 million patients with diabetes, have increased risk of severe infection (Carey et al., 2018). Our findings thus have broad implications for understanding host–pathogen interactions and may pave the way for the development of therapies that improve outcomes against infectious diseases, particularly in the context of metabolic diseases.

MATERIALS AND METHODS

C. elegans strains and growth conditions

C. elegans strains were grown on EZ worm plates containing streptomycin (550 mg Tris-Cl, 240 mg Tris-OH, 3.1 g bactopeptone, 8 mg cholesterol, 2.0 g NaCl, 200 mg streptomycin sulfate and 20 g agar per liter) to be consistent with previous studies from the laboratory (Ciccarelli et al., 2024a; Clark et al., 2018). All strains were maintained on E. coli DA837, a commonly used streptomycin-resistant variant of OP50, at 20°C. N2 was used as a wild-type control in all experiments. BMP mutant strains used were LT207 sma-3(wk30) and LT121 dbl-1(wk70). Lipid metabolism mutant strains used were BX156 fat-6(tm331); fat-7(wa36), BX17 fat-4(wa14), BX24 fat-1(wa9), BX52 fat-4(wa14) fat-1(wa9), VS18 maoc-1(hj13), DR476 daf-22(m130) and VS8 dhs-28(hj8). All mutations are strong loss-of-function or null alleles. Fluorescent lipid droplet reporter strain was LIU1 IdrIs1 [dhs-3p::dhs-3::GFP+unc-76(+)], CS772 dbl-1(wk70); IdrIs1 [dhs-3p::dhs-3::GFP+unc-76(+)]. Genetic information was obtained from WormBase (Sternberg et al., 2024).

Bacteria

Control bacteria used in all experiments were E. coli strain DA837, cultured at 37°C. Two pathogens were used for pathogenic bacteria exposure: S. marcescens strain Db11 (ATCC #13880) cultured at 37°C and P. luminescens (ATCC #29999) cultured at 30°C. All experiments involving pathogens were conducted on EZ worm plates without streptomycin.

ORO neutral lipid staining

ORO staining was done as previously described (Clark et al., 2018). ORO stock solution was prepared by dissolving 0.25 g ORO powder in 50 ml isopropanol. Animals were collected after the desired time of pathogen exposure in PCR tube caps and washed three times in PBS to remove excess bacteria. Worms were fixed for 1 h in 60% isopropanol while rocking at room temperature with caps covered with PCR tubes. While worms were fixing, the ORO working solution was made and allowed to rock at room temperature for 1 h. After the working solution had equilibrated for 1 h, it was filtered using a 10 ml syringe through a 0.45 µm filter, then through a 0.2 filter. The 60% isopropanol was removed and replaced with ORO working solution. The caps were covered with tubes and left overnight to stain while rocking at room temperature. The next day, the ORO was removed, and worms were washed once with PBS with 0.01% Triton X-100 and then left in PBS while preparing slides for imaging. Worms were mounted on 2% agar pads on glass slides and imaged on a Zeiss Axioscope 2 using a Gryphax camera with Gryphax software. Images were taken using a 40× objective. ImageJ software was used to turn the z-stack images into a single composite image so the number of lipid droplets and their diameters could be quantified. For every animal, a 50 pixel by 50 pixel region was selected for quantification, ensuring the region was anatomically consistent across all images. Measurements were taken for each visible droplet. Any circular or spherical body was counted as a droplet (Clark et al., 2018).

Survival analysis

Survival analysis was done as previously described (Ciccarelli et al., 2024a,b). Each survival plate was seeded with 500 µl Db11. 50 µM 5-fluoro-2′-deoxyuridine was added to each plate to prevent progeny and reduce the incidence of matricide by internal hatching of embryos. All survival experiments were carried out at 20°C. For each genotype, 120 L4 animals were picked for the experiment, and 20 animals were plated per survival plate. The numbers of alive and dead animals were counted at least 4 days per week. During the experiment, some animals were lost owing to burrowing, desiccation, etc. These animals were censored as their deaths were not observed. All survivals were repeated. Statistical analysis was done using log-rank (Mantel–Cox) test.

Bacterial supernatant preparation

Overnight bacterial culture of DA837 and Db11 was prepared according to the temperatures specified above. Cultures were centrifuged for 5 min at 4000 g, after which the supernatant should be relatively clear and the bacteria should be separated into a pellet. The supernatant was filtered through a 0.45 µm pore syringe filter to remove any remaining bacterial cells in the supernatant; however, secreted peptides should be able to pass through the filter. The supernatant was added to plates already seeded with DA837 in a 1:1 volume ratio and allowed to dry.

RNA-seq

Worms were synchronized with an overnight egg lay and 4 h timed hatch. Animals were grown on DA837 until L4, at which point they were washed with M9 buffer and transferred either to Db11 plates or new DA837 plates. After 24 h, they were washed with M9 again and collected in 15 ml tubes. They were washed 3× with M9, removing supernatant each time. RNA was extracted using a Trizol and chloroform precipitation, followed by Qiagen RNeasy mini kit. RNA concentration was measured using a Qubit with the RNA Broad Range kit. Samples were frozen at −80°C until ready to send for sequencing. Three biologically independent replicates were collected. Sequencing was done at Azenta, resulting in a range of 22-40 M single-end reads per sample, with phred scores of 38-39. Reads were mapped to the C. elegans genome (WS273). Gene counts were generated with STAR. Approximately 92% of reads aligned. EdgeR was used to determine differentially expressed genes (FDR<0.05). Analysis was done using pandas, and Venn diagrams were generated with matplotlib.

Fluorescence microscopy and image analysis

Animals were mounted on 2% agarose pads containing a 3 µl drop of 2.5 mM levamisole for immobilization. Images were taken on Zeiss LSM 900 with Airyscan 2 with Zen System software and a 63× objective. The anterior and posterior regions of the intestine were imaged as z-stacks. In Fiji, the z-stacks were converted to a maximum intensity projection, and the diameter and count of all visible lipid droplets in a 400 μm² area were measured. For each experiment, n=5 per condition was repeated in duplicate.

Fatty acid supplementation

Supplementation plates were prepared by making a base of EZ worm plates with no antibiotics and adding 0.1% tergitol (NP40). They were autoclaved as usual; then, after the medium had cooled to 50°C, 0.8 mM fatty acid stock solution was added and stirred until homogeneous. Plates were then poured as usual.

ATP/PRO

Worms were synchronized by bleaching. Animals were grown on DA837 until L4, at which point they were washed with M9 buffer and transferred either to Db11 plates or to new DA837 plates. After 24 h, they were washed with M9 again and collected in 15 ml tubes. They were washed 3× with M9, removing supernatant each time. For each sample, 20 µl worm pellet was transferred into a labeled Eppendorf tube, 180 µl boiled Tris-EDTA buffer was added, and the tube was incubated at 100°C for 2 min. Tubes were then sonicated on ice by pulsing for 4 min at 60%, then centrifuged at 14,440 g for 10 min. The supernatant was transferred to new tubes and used for measuring ATP and protein. ATP measurement was done using a Roche ATP Bioluminescence Assay Kit. Protein measurement was done using a Thermo Fisher Scientific Pierce BCA Protein Assay Kit. Each assay was done in 96-well plates and measured with a Tecan Spark plate reader.

Statistical analyses

Statistical analysis was performed in GraphPad Prism 10.