Dietary S. maltophilia induces supersized lipid droplets by enhancing lipogenesis and ER-LD contacts in C. elegans

ABSTRACT Dietary and symbiotic bacteria can exert powerful influence on metazoan lipid metabolism. Recent studies have emerged that microbiota have a role in animal obesity and related health disorders, but the mechanisms by which bacteria influence lipid storage in their host are unknown. To reduce the complexity of the relationship between gut microbiota and the host, Caenorhabditis elegans (C. elegans) has been chosen as a model organism to study interspecies interaction. Here, we demonstrate that feeding C. elegans with an opportunistic pathogenic bacterium Stenotrophomonas maltophilia (S. maltophilia) retards growth and promotes excessive neutral lipid storage. Gene expression analysis reveals that dietary S. maltophilia induces a lipogenic transcriptional response that includes the SREBP ortholog SBP-1, and fatty acid desaturases FAT-6 and FAT-7. Live imaging and ultrastructural analysis suggest that excess neutral lipid is stored in greatly expanded lipid droplets (LDs), as a result of enhanced endoplasmic reticulum (ER)-LD interaction. We also report that loss of function mutations in dpy-9 in C. elegans confers resistance to S. maltophilia. Dietary S. maltophilia induces supersized LDs by enhancing lipogenesis and ER-LD contacts in C. elegans. This work delineates a new model for understanding microbial regulation of metazoan physiology.


Introduction
In the wild, the nematode C. elegans feeds on a variety of soil bacteria including Pseudomonas mendocina, Bacillus megaterium, Comamonas sp. [1][2][3][4][5][6] In contrast, the E. coli OP50 is commonly used as the standard laboratory diet. 7,8 Therefore, the full range of physiological response in C. elegans toward dietary bacteria remains obscure. Nevertheless, increasing evidence suggests that C. elegans is capable of integrating olfactory, mechanical and nutritional cues to identify its preferred bacterial diet. Such dietary choices can subsequently exert significant impact on C. elegans lifespan and metabolism. [9][10][11][12][13][14][15][16][17][18] Lipid droplets (LDs) are conserved membranebound organelles for cellular neutral lipid storage. Over the last decade, to reveal LD conservation from bacteria to humans, our laboratory has focused on using mass spectrometry to determine the proteome of isolated LDs from a wide range of uni-and multi-cellular organisms, including C. elegans. Accordingly, we identified three major LD resident proteins in C. elegans, DHS-3, MDT-28, and F22F7.1. [19][20][21] When expressed as fluorescent fusion proteins in transgenic worms, DHS-3 and MDT-28 serve as faithful markers for monitoring LD morphology in live animals. Importantly, we found a strong correlation between LD size and number with organismal neutral lipid storage.
Metabolic syndromes are in fact a lipid storage disorder, ectopic lipid storage that originates from obesity. Obesity is a symptom of excessive accumulation of neutral lipids such as triacylglycerol (TAG) in white adipocytes. Adipose tissue governs the individual's lipid homeostasis at the tissue level. However, at the cellular level, TAG is stored in LDs. The distribution of LDs in non-adipose tissues, the size and number of LDs in these cells, as well as the type and amount of fatty acids in TAG, are the key factors of metabolic disorders. In white adipose tissue, adipocytes contain unilocular lipid droplets with diameter between 50 μm and 150 μm and their cytoplasm only takes less than 15% of cell volume. Thereby, the LD size is typically considered synonymous for the adipocyte size. 22,23 In other cells, such as liver cells, the number and size of LDs vary with the state of the cell. [24][25][26] Therefore, morphological study of lipid droplets is of great significance for understanding obesity and ectopic lipid storage.
Here we apply our suite of fluorescence markers for the ER and LDs, to study the effect of an environmental bacterium S. maltophilia on neutral lipid storage in C. elegans. Interestingly, clinical isolates of S. maltophilia have been regarded as pathogenic in immunocompromised humans, but not necessarily to nematodes. [27][28][29] We found that C. elegans grew slower and accumulated significantly more triacylglycerol (TAG) when fed S. maltophilia, instead of the standard laboratory diet, E. coli OP50. By genetic analysis, we uncovered a transcriptional regulatory network that promoted lipogenesis in response to S. maltophilia. Coupled with the remodeling of ER-LD interaction, massive LD expansion ensued. Accordingly, attenuation of ER-LD interaction partially suppressed the effect of dietary S. maltophilia on neutral lipid storage. Our results help establishing a new C. elegans-microbe experimental paradigm for the study of dietary factors that modulate lipid metabolism.

Size of lipid droplet in C. elegans is regulated by dietary microbe
Two bacterial species were previously shown to reduce C. elegans lipid accumulation. 30,31 To recapitulate these results, we fed Lactobacillus plantarum and Pseudomonas aeruginosa (strain UCBPP-PA14, PA14) to transgenic worms that express the intestinal LD marker, DHS-3::GFP (Figure 1a-1d). We found that Lactobacillus plantarum and Pseudomonas aeruginosa reduced the average LD diameter by 26.7% and 83.3%, respectively (Figure 1a-e). We were therefore encouraged to use our DHS-3::GFP and MDT-28::mCherry reporter strains for further exploration of the effect of bacterial diets on C. elegans neutral lipid storage. 9 We performed a small-scale screen to identify bacteria that modulate C. elegans lipid metabolism. To capture environmental bacteria, Nematode Growth Medium (NGM) plates were left open in the laboratory. Bacteria that yielded individual colonies were isolated, cultured in liquid medium, and subsequently re-introduced to NGM plates to form single-species bacterial lawns. Synchronized L1 stage transgenic reporter worms that express DHS-3::GFP were allowed to develop with the environmental bacteria as food. We subsequently quantified the LD size in larval L4 stage worms. Bacteria that significantly altered LD size of worms were taken through multiple cloning steps to obtain a pure clone. The cultures were then fed again to worms to verify the effect and the bacteria were identified by 16S rRNA sequencing ( Figure  1f). Worms fed these positive clones invariably showed growth retardation or arrest (Table 1). We then focused on S. maltophilia because it was the only bacterial species so far tested that significantly increased the LD size of worms.

S. maltophilia promotes lipid droplet expansion in C. elegans
The bacterium S. maltophilia induced a striking increase in LD diameter after 2 days of feeding (Table 1). To confirm the effect, we repeated the experiment using S. maltophilia strains isolated in other laboratories 32,33 (Table 2). Similar to our original isolate, all strains tested reproducibly promoted LD expansion in worms (Figure 2a). Our results suggest that S. maltophilia harbors a speciesspecific factor that potently modulates neutral lipid storage in C. elegans.
Next, we asked if the effect of S. maltophilia feeding on LD size was restricted to C. elegans intestine. As indicated by the vha-6 promoter driven, intestine-specific DHS-3::GFP marker, the LD diameter distribution of worms fed S. maltophilia had an increased percentage of larger LDs (>2 μm), but a decreased percentage of smaller LDs (<2 μm) compared to those of WT worms (Figure 2b-d). To ensure the reliability of the results of DHS-3::GFP, DGAT-2::GFP was used for further verification. The results showed that S. maltophilia feeding also induced DGAT-2::GFP-labeled LD expansion (Fig.  S9). To determine the change of LD size in the hypodermis, we used MDT-28::mCherry that was expressed under the control of the mdt-28 promoter ( Figure 2e). 21 Similar to our observations in the intestine, S. maltophilia feeding induced LD expansion in the hypodermis. The maximum LD diameter reached 6 μm (Figure 2e, Figure 2f and Figure 2g). Finally, we compared LD morphology in embryos from parents that were fed either OP50 or S. maltophilia (Figure 2h-2i). The embryos from S. maltophilia-fed worms again showed enlarged LDs (Figure 2i, white arrows). These results were further confirmed by examining isolated LDs from worms fed OP50 or S. maltophilia. The LDs isolated from S. maltophilia-fed larval L4 worms, were enlarged relative to OP50-fed worms (Figure 2j  (e) Distribution of the LD size (% Lipid droplets), as measured from DHS-3::GFP labeled worms from (a, b, d). Distribution of the LD size was measured for least 10 animals from at least three biological replicates. Data are shown as mean ± SEM, student's t-test, **P < .01, ***P < .001, n = 30. (f) Schematic representation of the screening method to identify bacteria that affect host LDs.
increase in organismal lipid content, we used biochemical methods to measure TAG levels. Total lipids were extracted from worms fed either OP50 or S. maltophilia from L1 to L4 stage. We found that S. maltophilia-fed worms had 3.7-fold more TAG than those fed OP50 (Figure 2n). Taken together, these morphological and biochemical analyses indicate that S. maltophilia can potently induce neutral lipid accumulation in C. elegans.
Induction of C.elegans neutral lipid storage by S. maltophilia was not due to its; species-specific fatty acid composition Differences in lipid composition between bacterial food sources could potentially underlie the effect of dietary S. maltophilia on host lipids, perhaps due to some lipids being more readily absorbed, processed, and incorporated than others. We examined the fatty acid compositions of E. coli OP50 and S. maltophilia by gas chromatography-mass spectrometry (GC-MS), and found substantial differences between them (Fig. S1A), and branched fatty acid iso 17(C17iso) was used to feed worm but no enlargement of LDs in the nematode intestinal cells was detected (Fig. S1F). However, it is not clear from this observation alone if lipid compositional difference accounted for the ability of S. maltophilia to promote neutral lipid accumulation in worms. Therefore, we conducted a transposon-based forward genetic screen in S. maltophilia to identify mutants that failed to induce LD expansion when fed to C. elegans (Fig.   S1B). Through this approach, we isolated SMa9, a mutant S. maltophilia strain that could not increase LD size and number in C. elegans ( Fig.  S1C and S1D). However, there was no difference in fatty acid composition between the mutant SMa9 and the parental wild type (WT) bacterium (Fig.  S1E). In addition, to detect the absorption rate of food fatty acids by nematodes, we fed worm with BODIPY 556/568 C12 as a free fatty acid in the presence OP50 or S. maltophilia, and the result showed that S. maltophilia did not affect uptake and incorporation of BODIPY 556/568 C12 (Fig.  S8). Therefore, the results clearly suggest that one or more factors other than fatty acids, were responsible for the ability of S. maltophilia to promote neutral lipid storage in C. elegans.

S. maltophilia-induced neutral lipid storage was not part of an innate immune response in C. elegans
We then examined whether the phenotype of excess neutral lipid accumulation was due to a pathological process associated with live S. maltophilia. Worms were fed ultraviolet (UV)killed, culture supernatant of S. maltophilia or high temperature-killed S. maltophilia and their LDs were examined ( Fig. S2A and S6). The killed bacteria retained the ability to increase LD size in C. elegans, compared with OP50 feeding (Fig.  S2A). As the effect of the bacterial culture supernatant was very small, we concentrated the bacterial culture supernatant using freeze-drying method and fed the nematodes, and found that the bacterial culture supernatant could only induce the formation of nematode large LDs slightly (Fig. S6). Next, we measured the expression level of NLP-29, 34 which is the nematode immune response marker. We found that NLP-29 could be significantly  Data are presented as mean ± SD of 10 animals for each worm strain, the assay increased by S. maltophilia feeding (Fig. S2B). However, when we measured LD size in the worm defective in the p38 MAPK pathway (sek-1, pmk-1) that is critical for the innate immune response, we found no significant change in the LD phenotype in these mutant worms (Fig. S2C). We concluded that the elevated neutral lipid storage in worms upon S. maltophilia feeding was independent of the p38 MAPK pathway.

S. maltophilia modulates lipid metabolism in C. elegans by upregulating sbp-1, fat-6,and fat-7
To understand the pathways that S. maltophilia utilize to enlarge LDs in the nematode, we explored other possible mechanisms of this process. To determine the effects of bacteria on C. elegans physiology, we measured developmental rate, fecundity, and hatching rate of worms fed by either E. coli OP50 or by S. maltophilia. To assess developmental rate, we synchronized animals at the L1 stage and monitored the developmental age of the population over time (Fig. S3A). Development to the L4 stage was delayed in worms fed by S. maltophilia, compared with OP50 (Fig. S3A). The number of offspring produced by S. maltophilia-fed worms was reduced by 30% compared with the OP50 group (Fig. S3B).
There was no significant difference in hatching rate between OP50-and S. maltophilia-fed worms (Fig.  S3C). The S. maltophilia bacteria were strongly preferred by worms over the E. coli OP50 ( Fig. S3D-F). We measured pharyngeal pumping in worms after 1 h and 3 h of feeding and found no differences (Fig.  S3G). Lastly, we investigated the effect of S. maltophilia feeding on LD size in eat-4(ky5) mutant worms, which exhibited reduced pharyngeal pumping and therefore ate less than wild type worms. 35 There were no differences in LD size between wild type and eat-4(ky5) mutant worms when they were fed S. maltophilia (Fig. S3H). These results suggest that the effect of S. maltophilia on C. elegans LD size was not dependent on food intake. Next, we turned our attention to examine lipid metabolic pathways in C. elegans. The enlargement of LDs or increasing of TAG accumulation mainly depends on the increasing of TAG synthesis or/and decreasing of TAG hydrolysis. 36,37 Therefore, we compared the levels of mRNA expression of TAG metabolism-related genes [38][39][40][41] in OP50 vs. S. maltophilia fed worms, and the results are presented in Figure 3a. Among them the genes involved in synthesis of fatty acids especially unsaturated fatty acids were significantly upregulated. Previous studies also found that feeding fatty acids especially unsaturated fatty acids can increase LD size and TAG content. 42,43 Wild type worms were fed OP50 or S. maltophilia from L1 to L4 stage and the relative expression level of selected metabolic genes was determined by real-time PCR. This analysis clearly indicated that dietary S. maltophilia significantly increased the expression of sbp-1, fat-5, fat-6, and fat-7, all of which are critical for lipogenesis in C. elegans, but there is no significant change in lipid synthesis key genes (mdt-15, nhr-49, dgat-2, fasn-1) and lipid hydrolysis key genes (atgl-1) (Figure 3a).

dpy-9 and acs-13 suppress the S. maltophilia-induced lipid droplet expansion
A forward genetic screen was conducted to identify C. elegans host factors that mediate the S. maltophilia effect on LDs. We used DHS-3::GFP or MDT-28:: mCherry as LD markers ( Fig. 4a and 4b) and screened 7,000 haploid genomes after chemical mutagenesis with ethyl-methane sulfonate (EMS), and isolated 7 mutant strains ( Figure 4a). Genetic mapping based on single nucleotide polymorphisms (SNPs) eventually led to the molecular cloning of two genes: acs-13 and dpy-9 ( Figure 4c-Figure 4h). The acs-13 gene encodes an ortholog of human ACSL1, 5, 6 (acyl-CoA synthetase long-chain family member 1, 5, 6). The dpy-9 gene encodes a cuticular collagen family member with similarity to human collagen alpha 5, type IV. Using complementation tests and RNAi, we confirmed that the loss of acs-13 and dpy-9 function blocked the ability of S. maltophilia to induce LD expansion in C. elegans (Figure 4a-Figure 4i).
We next investigated if the loss of acs-13 or dpy-9 function could normalize neutral lipid storage in C. elegans mutants that are known to accumulate excess fat. Notably, mutations in dhs-28 and daf-22 are known to impair peroxisomal β-oxidation and induce LD expansion in the intestine in C. elegans. 49,50 We analyzed OP50-and S. maltophilia-fed worms and found no significant difference in peroxisome morphology ( Fig. S4A  and S4B). Feeding daf-22(m130) mutants with S. maltophilia resulted in a significant increase in the number and size of LDs in the intestine and hypodermis (Fig. S4C and S4D), suggesting that dietary S. maltophilia acted in parallel of the peroxisomal β-oxidation pathway, which is responsible for fat catabolism. Finally, we recapitulated the LD expansion phenotype by knocking down dhs-28 by RNAi in wild type worms, the enlarged LDs were detected in the intestine (Fig. S4E and S4F). Next, we knocked down dhs-28 or daf-22 in acs-13 and dpy-9 mutant worms. The acs-13 mutation suppressed the formation of enlarged LDs in dhs-28 or daf-22 deficient worms. In contrast, mutations in dpy-9 had no effect (Fig. S4G). Therefore, we concluded that DPY-9 is a host factor that is required for dietary S. maltophilia to promote LD expansion.

S. maltophilia induces lipid droplet expansion by enhancing ER-LD interaction
The endoplasmic reticulum (ER) is the primary site where TAG is synthesized. 51,52 Therefore, the effects of S. maltophilia feeding on the ER were investigated. We used two reporters, Phyp-7:: TRAM-1::GFP ( Fig. S5A and S5B) and Pvha-6:: SEL-1 ::mCherry::HDEL, to mark the ER membrane and lumen, respectively. After 2 days of S. maltophilia feeding, ring-shaped structures were observed in the hypodermis of the TRAM-1:: GFP worms (Fig. S5C-5F). We also observed ERwrapped LDs in the intestine of the mCherry:: HDEL worms (Fig. S5G-5L). It is plausible that the remodeling of the ER morphology upon S. maltophilia feeding supports LD expansion. We further investigated the ER-LD interaction using electron microscopy (EM). In worms fed by E. coli OP50 diet, few connections between the ER and LDs were observed (Figure 5a and Figure 5d), and there is no obvious structure between LD and ER. In contrast, in worms fed by S. maltophilia diet, we readily noted that LDs were connected to the ER through small tubular structures (Figure 5b, Figure  5c, Figure 5e, and Figure 5f). To observe the contact sites in detail, electron tomography was performed. The images show that the connection between LDs and the ER was approximately hollow (Figure 5g-Figure 5j). Since similar structures have previously been predicted by William A Prinz, 53 we continue to name this structure ER-LD bridge.
To further understand how S. maltophiliainduced LD expansion was instigated by ER remodeling, we examined the localization of SEIP-1 that is highly enriched in an ER subdomain that associates tightly with LDs. 54 SEIP-1 is the C. elegans ortholog of seipin that functions at ER-LD contact sites. [55][56][57][58][59][60] To visualize SEIP-1 positive ER subdomain (peri-LD cages), we used two reporter strains that expressed SEIP-1::GFP fusion proteins either ubiquitously (hjSi189) or specifically in the intestine (hjSi3). 54 When hjSi189; ldrIs2 worms were fed E. coli OP50, few intestinal LDs were associated with SEIP-1::GFP cages (Figure 6a, Figure 6b and Figure  6c). In contrast, almost all intestinal LDs were enwrapped by SEIP-1::GFP cages when hjSi189; ldrIs2 transgenic worms were fed S. maltophilia, coincident of LD expansion (Figure 6d, Figure 6e and Figure 6f). We also found that more than 95% of LDs were associated with SEIP-1::GFP positive structures in worms on S. maltophilia diet (Figure 6d,  Figure 6e, and Figure 6f). To further quantified the fraction of LDs that were decorated with SEIP-1:: GFP, LDs were isolated from hjSi189; ldrIs2 worms fed by OP50 or by S. maltophilia for 2.5 days and analyzed using confocal microscopy. Again, few LDs were found to retain SEIP-1::GFP in the OP50-fed group while 100% of LDs from the S. maltophilia-fed group showed SEIP-1::GFP signals (The green SEIP-1::GFP signals appears more or less on the red MDT-28:: mCherry signal) (Figure 6g-Figure 6m). These results are consistent with the model that the SEIP-1 positive ER subdomain promotes LD expansion, and   Data are presented as mean ± SD of 30 animals for each worm strain, student's t-test and one-way ANOVA, 0.01<*P < .05, *** P < .01, *** P < .001, ns, no significance, n = 30. that S. maltophilia feeding significantly increases the coverage of such subdomain, thereby supporting widespread LD expansion.
Next, we investigated if loss of DPY-9 function could suppress S. maltophilia-induced LD expansion by interfering with SEIP-1 enrichment in peri-LD cages. Accordingly, we introduced dpy-9, and dyp-5 (negative control) mutations into the SEIP-1::GFP reporter strain hjSi189. After feeding S. maltophilia for 2 days, no SEIP-1::GFP positive, peri-LD cages were observed in dpy-9 mutant animals ( Figure 6n, and Figure 6o). It is conceivable that DPY-9 directly or indirectly act in pathways that remodel the membrane environment of peri-LD cages. Such remodeling is crucial for the recruitment of SEIP-1 and additional proteins that are critical for LD expansion.

Requirement of proteins involved in ER-LD interactions for S. maltophilia-induced lipid droplet expansion
To determine if SEIP-1 is important for LD expansion induced by dietary S. maltophilia, we fed S. maltophilia to seip-1(tm4221) mutant animals (Figure 7a, Figure 7b, Figure 7c, and Figure  7d). The results revealed that loss of SEIP-1 function only partially suppressed the effect of S. maltophilia feeding on LDs (Figure 7e), which hints that additional proteins may reside in ER-LD contacts that mediate the effect of dietary S. maltophilia. To this end, we investigated rab-18 and dgat-2 since they have been reported to be involved in ER-LD interaction. [61][62][63][64][65] We examined the effect of rab-18, and dgat-2 mutations on the size of LDs in C. elegans after S. maltophilia feeding. As shown in Figure 7f-Figure 7j, loss of DGAT-2 function partially suppressed LD expansion. However, loss of Rab-18 function had no effect in this experimental context.

Discussion
In this paper, we show that dietary S. maltophilia promoted organismal neutral lipid storage in C. elegans through a lipogenic transcriptional program that consists of SBP-1, FAT-6 and FAT-7. In addition, S. maltophilia feeding enhanced DPY-9-dependent ER-LD interaction in the intestine and hypodermis, which was partially dependent on DPY-9. Collectively, we hypothesize that the increased contact, via ER-LD bridges, facilitates the transfer of lipid from the ER to LDs (Figure 8).
To investigate the physiological significance of S. maltophilia-enlarged LDs in C. elegans, we conducted experiments to determine if they affect survival. The lifespan of worms fed S. maltophilia was not significantly changed compared with worms fed OP50 (Fig. S7). However, blocking the S. maltophilia-induced increase in LD size by knocking out sbp-1 or fat-6; fat-7 reduced the lifespan of the worms (Fig. S7B and S7C). Next, we fed N2 worms S. maltophilia for 2 days to induce large LDs. Subsequently the animals were infected with PA14 (Fig. S7D). S. maltophilia-fed worms were able to resist PA14 infection better than the OP50fed worms (Fig. S7E). This indicates that the emergence of enlarged LDs may protect nematodes from environmental stress, which is important for their survival.
C. elegans and its diet has become a powerful system to study host-microbiota interactions. 5,9,18 However, it is still challenging to study the relationship between bacteria and lipid metabolism in nematodes because bacteria not only provide nutrients, but also can be pathogenic to the worms. 5 The observation that UV-killed S. maltophilia still generated the LD phenotype suggests that the effect is not due to pathogenicity. To examine the role of bacterial lipids in the phenotype, gas chromatography-mass spectroscopy was used to measure the fatty acid composition in S. maltophilia and OP50. The fatty acids of S. maltophilia are mainly 13methyl myristic acid (C15iso) and 15-methyl hexadecanoic acid (C17iso), which differs substantially from those of OP50. We investigated further using a transposon-based forward genetic screen in S. maltophilia and found that a mutation in SMa9 lost the effect in C. elegans. We compared the fatty acid composition of S. maltophilia and SMa9 and found that their fatty acids were not significantly different, which rules out a causal role of fatty acid composition.
The use of C. elegans to study host-pathogen interaction was pioneered by the Ausubel lab, which stemmed from the observations that Pseudomonas Aeruginosa PA14 could effectively kill worms. 66 C. elegans genetic and molecular analyses led to the discovery of conserved innate immune response pathways that counter P. Aeruginosa infection. 67,68 The systematic response appears to require environmental sensing and inter-tissue communication between the nervous system, hypodermis and intestine. [69][70][71] Ultimately, multiple transcriptional networks coordinately mount gene expression programs that support the defense against the pathogen. 72,73 Besides P. Aeruginosa, many other pathogenic bacteria, such as E. faecalis, S. aureus, S. typhimurium and C. neoformans, have subsequently been shown to effectively kill worms. 74 However, relatively less is known about the impact of pathogenic bacteria on C. elegans physiology during chronic infection. Here, we discovered that environmental isolates of the pathogenic bacterium S. maltophilia did not kill worms, and may be able to colonize the intestines. 28 Instead, worms maintained on a S. maltophilia diet grew slower and accumulated excess neutral lipid, but for unc-76 or daf-2 mutants, the slow-growing mutant, S. maltophilia still had the ability to induce LD expansion (data unpublished), this excluded the effect of growth on the enlargement of LDs. Accordingly, the p38 MAP kinase pathway, which is central to innate immunity against lethal pathogens in C. elegans, did not play a major role in the induction of neutral lipid storage in response to dietary S. maltophilia. Our results indicate that S. maltophilia may exert metabolic burden to its host. Further investigation is needed to elucidate the casual factors from S. maltophilia that modulate host metabolism.
Since S. maltophilia is commonly found in immunocompromised, hospitalized patients, 75 there is an urgent need to understand the host response to this opportunistic pathogen, in order to identify cellular proteins that may serve as therapeutic targets. From a forward genetic screen, we found that loss of function mutations in dpy-9 rendered the worms resistant to S. maltophiliainduced neutral lipid accumulation. The dpy-9 gene encodes an extracellular matrix collagen that is presumably expressed in the hypodermis. The lack of dpy-9 compromises cuticle integrity that may give rise to osmotic stress. 76 Based on the functional annotation of this protein, it is currently unclear how the loss of DPY-9 function in distinct tissues confers resistance to S. maltophilia. To further understand its function, we further examined the epidermal structure of the nematode and found that S. maltophilia feeding only increased the fluorescence intensity of COL-19::GFP, but it could not change its texture (Fig. S10). It is plausible that DPY-9 is part of a wider genetic network that facilitates inter-tissue communication in response to S. maltophilia. Phosphatidylcholine (PC) and phosphatidylethanolamine (PE) are the most abundant phospholipids in cell membranes and the relative abundance of PC and PE regulates the size and dynamics of LDs. 54,77,78 Based on our results, the ratio of PC to PE was not the reason for S. maltophilia-induced large LDs. Although further analysis is needed to identify other components of this network, our results clearly suggest that one of its outputs is required to sustain ER-LD contacts, which in turn supports neutral lipid accumulation in LDs.
Our work provides a basis for further investigation into the mechanisms governing ER-LD interaction. Under normal conditions, it is difficult to identify and study proteins involved in this process, because ER-LD contact is rare and not easily detected. However, the ER-LD interactions induced by S. maltophilia provide an unambiguous, robust, and easily measured phenotype dependent on the machinery of this interorganelle contact. This discovery sets the stage for further investigation using genetic screening and mass spectrometry to further elucidate the mechanistic basis for this phenomenon.
To further investigate the role of enlarged LDs in S. maltophilia-fed worms, we examined the percent survival of the nematodes. This experiment showed that the emergence of highly enlarged LDs enhanced the resistance of nematodes to stress. Based on this finding, we hypothesize that LDs may sequester or eliminate some toxic substances, giving the animals a survival advantage in extreme conditions.
In summary, our work has established a new host-microbe experimental paradigm. Chronic exposure of C. elegans to S. maltophilia reproducibly caused metabolic remodeling in multiple tissues. Future experiments will be directed to studying microbial factors that are responsible for such remodeling via DPY-9 and additional yet-tobe identified host factors.

1)
Bacterium S. maltophilia stimulates the formation of supersized LDs in C. elegans, 2) The effect of S. maltophilia on LDs is mediated by the sbp-1, fat-6, and fat-7 pathway, 3) an ER-LD bridge is involved in the S. maltophilia-induced effect on LDs. These findings suggest that C. elegans can serve as a model organism for the study of gut microbiota-mediated human obesity.

Nematode strains and growth conditions
All C. elegans strains were handled and maintained following standard procedures. 8 The N2 Bristol strain, dpy-9(e12), and dpy-5(e61) were obtained from the Caenorhabditis Genetics Center S. maltophilia and OP50 were cultured on LB and NGM plates.
All experimental animals were maintained at 20°C.

Forward genetic screen and mutant mapping
To screen the mutant animals with a suppressed phenotype of LD induced by S. maltophilia, ldrIs3 was mutagenized with ethyl methane sulfonate (EMS) as previously described. 79,80 7,000 haploid genomes were screened and mutant phenotype in the F2 worms were selected to observe the LD phenotype. The mutant animals were selected with a suppressed phenotype of LD induced by S. maltophilia (Mutants which LDs could not become larger on a S. maltophilia diet). Mutant worms were backcrossed with ldrIs3 at least four times and their LD phenotypes were studied using a ZEISS Imager M2. The stable mutant animals were mapped using single-nucleotide polymorphism (SNP) mapping. 81

Isolation of lipid droplets
LDs were isolated using a method described previously. 20,82 Briefly, synchronized L1 worms were cultured on NGM plates and the L4 larval stage worms were collected in M9 buffer (22 mM KH 2 PO 4 , 42 mM Na 2 HPO 4 , 86 mM NaCl, 1 M MgSO 4 ). The worms were washed three times in Buffer A (25 mM Tricine, pH 7.6, 250 mM Sucrose, and 0.2 mM Phenylmethylsulfonyl fluoride) and were then homogenized to obtain whole body lysates. The lysates were centrifuged at different speeds (2,020 g, 4,546 g and 12,628 g) in SW40 tubes to obtain LDs of different diameters. The LD fraction was collected from the top layer and was washed three times with Buffer B (20 mM HEPES, 100 mM KCl, 2 mM MgCl 2 , pH 7.4). Images of the isolated LDs were obtained using a ZEISS Imager M2 or LSM880.

Measurement of triacylglycerol
Synchronized L4 worms were washed three times with M9, and were dissolved in 200 μl 1% Triton X-100 by sonication. Then, the worm lysates were centrifuged at 13,210 g for 3 min. The triacylglycerol (TAG) content of the supernatants was measured using the Triacylglycerol Assay Kit. 19 The Pierce BCA Protein Assay Kit (Thermo, USA) was used to quantify the proteins. 19 The results are shown in the ratio of TAG to protein (PRO) (mg/mg).

Lipid extraction and analysis
Bacterial lipid extraction, separation, and analysis were conducted as described previously. 43 In general, bacteria were grown at 20°C on the NGM plates, after 24 h, 3 ~ 5 mg bacteria were collected into glass tubes and water was removed with a Pasteur pipet. To each bacterial pellet 1 ml of MeOH and 2.5% H 2 SO 4 was added. Fatty acids were extracted and converted into methyl esters by heating at 70°C for 60 min. Then the extractions were incubated at 25°C for 5 min followed by addition of 0.2 ml of hexane and 1.5 ml of water. The mixtures were shaken vigorously and then centrifuged 12,000 g for 1 min in a clinical centrifuge. The fatty acid methyl esters contained in the top hexane-rich fraction were collected and 2 μl were analyzed using an Agilent 6890 series gas chromatograph equipped with a 20 m x 0.25 mm SP-2380 column (Supelco, Bellefonte, PA) and a flame ionization detector. The Gas Chromatography (GC) was programmed for an initial temperature of 120°C for 1 min followed by an increase of 10°C per min up to 190°C followed by an increase of 2°C per min to 200°C. Peak identity was determined by mass spectroscopy.

Transposon-based forward genetic screen
The S. maltophilia transposon insertion mutant library was constructed using an EZ::TN <KAN-2> Tnp transposome kit (Epicenter) as described previously and in accordance with the manufacturer's protocol. 83 The kanamycin-resistant transformants were selected and fed the worms. To identify mutants which did not result in enlarged LDs when fed to C. elegans.

Fluorescence imaging of C.elegans
Worms were immobilized with 0.5 mg/ml levamisole in M9 buffer and then transferred to a small glass slide or dish, then covered by a 2-4% (w/v) agar pad. For GFP or LipidTOX (G), a 488 nm laser was used for excitation and signals were collected with a 500-550 nm emission filter. For mCherry, a 561 nm laser was used for excitation and signals were collected with a 585-650 nm emission filter. For auto-fluorescence in the intestine, a 405 nm laser was used for excitation and signals were collected with a 417-477 nm emission filter. To obtain optimal images, immersion oils with refractive indices of 1.520 were used for LD on glass coverslips. Fluorescence images of L4 larval animals were obtained using a laser scanning confocal microscope (LSM 710 Meta, LSM880 Meta, ZEISS). Fluorescence images were taken and used to measure the diameter of the LDs in the posterior of the intestine with the same area (100 μm × 80 μm) by Image J and ZEN 2011 (ZEISS). At least 10 worms were utilized to measure LD size for each worm strain. All LDs were manually counted in the images. The Image J software usage method is as follows: select "Analyze"-"Set Scale"-"Known distance input the length of the drawn straight line"-"Unit of length input unit, check the Global checkbox (this standard is used for all pictures)"click "OK"-"Ctrl+M (Measurement), the result will be displayed and recorded in the Results window"-"Length displays the converted length value", and the measurement method was done blinded. According to the diameter of the LDs, the LDs were divided into 0-1 μm, 1-2 μm, 2-3 μm, and >3 μm. % LDs represents Number of LDs with certain size/Total number of LDs. All experiments are repeated three times.
LD size measurements were made at the L4 larval stage because: (1) the L4 stage can be easily identified; (2) earlier larval stages are too small for imaging; and (3) at the adult stage, embryos will interfere with imaging and may affect LD morphology.

Lifespan assay for C. elegans
Lifespan analysis was conducted at 20°C according to a protocol modified from a previous method. 84,85 Briefly, synchronized L1 animals were seeded onto NGM plates and grown until the L4 stage. On day 0, 20 L4 worms per plate (five plates, about 100 worms in total per condition) were transferred onto different diets plates. The statistical analyses were performed using GraphPad Prism 5.

Quantitative RT-PCR analysis
Total RNA was isolated using Trizol reagent. 86 Moloney murine leukemia virus (M-MuLV) reverse transcriptase with random hexamer primers was used to synthesize the cDNA. RT-PCR was performed on a CFX96 real-time system with SYBR green. Relative expression levels of all mRNAs were normalized to ama-1 mRNA.

Behavioral experiment
OP50 and S. maltophilia were cultured and dropped on NGM plates. The bacteria were allowed to grow for 24 h. The synchronized L4 larval nematodes were placed equidistant between the OP50 and S. maltophilia cultures. After 12 h, the number of nematodes on the bacterial colonies was determined.
RNAi assay of C.elegans L4440 (HT115) was used as the control for the RNAi assay. 19 The RNAi for dhs-28 and daf-22 were from the Ahringer RNAi library. The synchronized L1 (P0) worms were cultured on the RNAi NGM plates at 20°C to generate F1 or F2 worms for phenotypic analysis.

Brood size analysis
Approximately 20 L4 worms were removed from synchronized mothers, fed with OP50, and transferred to NGM plates seeded with the appropriate bacteria, in triplicate. The worms were transferred to new plates every 24 h three times until no additional embryos were produced. The number of embryos was counted every 24 h. 9,87 Growth rate Two-day old adults fed with the relevant bacteria were treated with hypochlorite to obtain embryos. The embryos were plated on NGM plates (∼100/ plate) seeded with different bacteria. The percentage of L4 worms out of the total worms was determined at various time points. 19 Pharyngeal pumping measurements L4 stage worms were picked from NGM plates. After 24 h, the pumping rate on different food was recorded. Worms were detected under an optical microscope and pumping rates were measured by visual observation. One pharyngeal pump was defined as a complete forward and backward movement of the grinder in the pharynx. The rate was counted for a period of 30 sec. 88

Dietary supplements
Fatty acid C17iso (From Prof. Huanhu Zhu) was prepared as a 10 mM stock in DMSO. A stock solution was mixed with 500 μl of OP50 overnight bacterial suspension in a 1:10 ratio and spotted on the appropriate plates according to the experiment. 89

Electron microscopy assay
Samples were prepared using previously described methods. 90 2D images were acquired using transmission electron microscope (FEI Tecnai Sprit 120 kV). 3D images were acquired using transmission electron microscope (FEI Tecnai Sprit 120 kV and Serial EM software).

Data analysis
All numerical data were plotted as mean ± SEM unless otherwise indicated. The statistical analysis was performed using GraphPad Prism 8 and Image J (NIH, USA). Data represent biological replicates. Appropriate statistical tests were used for every figure. Data met the assumptions of the statistical tests described for each figure. Statistical parameters, including the exact value of n and descriptive statistics (mean ± SEM) and statistical significance, are reported in the figures and the figure legends. Data are judged to be statistically significant when P < .05 by twotailed Student's t test. In figures, asterisks denote statistical significance as calculated by Student's t test (*P < .05, **P < .01, and ***P < .0001), as compared to appropriate controls.

Availability of Data and Material
All data generated or analyzed during this study are included in this published article and its supplementary information files.

Consent for publication
All authors have given consent for publication.