Bacterial carotenoids suppress Caenorhabditis elegans surveillance and defense of translational dysfunction

Microbial toxins and virulence factors often target the eukaryotic translation machinery. Caenorhabditis elegans surveils for such microbial attacks by monitoring translational competence, and if a deficit is detected, particular drug detoxification and bacterial defense genes are induced. The bacteria Kocuria rhizophila has evolved countermeasures to animal translational surveillance and defense pathways. Here, we used comprehensive genetic analysis of Kocuria rhizophila to identify the bacterial genetic pathways that inhibit C. elegans translational toxin surveillance and defense. Kocuria rhizophila mutations that disrupt its ability to disable animal immunity and defense map to multiple steps in the biosynthesis of a 50-carbon bacterial carotenoid from 5 carbon precursors. Extracts of the C50 carotenoid from wild type K. rhizophila could restore this bacterial anti-immunity activity to K. rhizophila carotenoid biosynthetic mutant. Corynebacterium glutamicum, also inhibits the C. elegans translation detoxification response by producing the C50 carotenoid decaprenoxanthin, and C. glutamicum carotenoid mutants are defective in this suppression of C. elegans detoxification. Consistent with the salience of these bacterial countermeasures to animal drug responses, bacterial carotenoids sensitize C. elegans to drugs that target translation and inhibit food aversion behaviors normally induced by protein translation toxins or mutations. The surveillance and response to toxins is mediated by signaling pathways conserved across animal phylogeny, suggesting that these bacterial carotenoids may also suppress such human immune and toxin responses.


Introduction
Many bacterial toxins and virulence factors target the highly conserved RNAs and proteins of the ribosome and associated translation factors. In response to such toxin attacks, eukaryotic defense responses include the induction of specific suites of drug detoxification and anti-bacterial immunity genes. The textbook view is that these detoxification responses are triggered by chemical recognition or virulence factor receptor proteins (guard proteins) that then couple to signaling pathways for the induction of immunity and detoxification gene responses. Such a toxin recognition surveillance system can be defeated by the evolution of novel bacterial toxins with distinct chemical or protein sequence signatures. But a system that surveils for toxins by their toxicity to, for example, the eukaryotic system targeted, for example, translation of proteins, may be superior for the detection of novel bacterial toxins. The nematode Caenorhabditis elegans surveils a wide range of core cellular pathways for such deficits, and whether it is caused by a toxin, a mutation or RNA interference, detoxification and immunity genes are induced (Melo and . By monitoring decrements in core cellular functions rather than the molecular structure of an unknown toxin, C. elegans can detect unanticipated pathogens and toxins. Many of the components of this signaling cascade, for example, the MAP kinase and bile acid biosynthetic pathway, are conserved across animals, suggesting that this system of toxin surveillance and response may not be parochial to C.

elegans.
Such an animal defensive strategy may drive the evolution of bacterial pathogen countermeasures to thwart these defense responses (Melo and Ruvkun, 2012)(Samuel et al., 2016). Commensal bacteria may also seek to silence such animal defense responses to establish a benign or symbiotic relationship. Because bacteria synthesize a wide palette of chemical toxins and virulence factors that target the ribosome and associated translation factors (Bérdy, 2005), we reasoned that a screen of diverse bacterial strains could detect counter-surveillance drugs or virulence produced by bacteria. We screened bacterial species by growth of individual bacterial strains with a C. elegans strain carrying a mutation in a translation factor gene for bacterial activities that disrupt the induction of xenobiotic detoxification genes. This screen identified an activity of Kocuria rhizophila that suppresses C. elegans surveillance of translation (Govindan et al., 2015). Here we identify by genetic analysis of K. rhizophila that C50 carotenoids are the bacterial activity that suppresses the C. elegans translational toxin defense response. This K. rhizophila activity suppresses the induction of xenobiotic detoxification and food aversion responses caused by deficits in protein synthesis. Another bacterial species that also produces the C50 carotenoid decaprenoxanthin, Corynebacterium glutamicum, also suppressed the induction of xenobiotic detoxification genes in a C. elegans translation factor mutant, and C. glutamicum carotenoid biosynthesis mutants also fail to suppress C. elegans detoxification induction. The bacterial inhibition of this detoxification response increases the potency of bacterial toxins that target translation: treatment of wild type C. elegans with purified K. rhizophila C50 carotenoids caused hypersensitivity to drugs that inhibit translation. Thus, the evolution and lateral transfer between bacterial strains of these carotenoids may have been selected because they increase in the potency of bacterial toxins. Bile acid signaling mediates the signaling between C. elegans translational surveillance and induction of drug detoxification genes (Govindan et al., 2015). Addition of mammalian bile acids to the K. rhizophila inhibited pgp-5 response pathway bypassed the carotenoid inhibition of the response, showing that carotenoids act upstream of bile acid production.

Results
Bacterial carotenoids suppress C. elegans surveillance and response to translational deficits Like many animal species, C. elegans encodes about 500 xenobiotic detoxification genes ---for example, cytochrome p450, ABC transporter, UDP-glycosyl transferase genes-that are induced by particular toxins or genetic deficits (Melo and Ruvkun, 2012). Particular suites of C. elegans detoxification genes are strongly induced by inhibition of ribosomal proteins, tRNA synthetases, and other genes implicated in translation (Govindan et al., 2015).
The C. elegans ABC transporter gene pgp-5 is strongly induced by disruption of ribosomal and other translation factor genes but not by defects in other conserved cellular components such as the mitochondrion or the cytoskeleton (Govindan et al., 2015). A pgp-5::gfp fusion gene is therefore a reporter of C. elegans translational dysfunction and not just general poor health or poor growth conditions (Govindan et al., 2015). This ABC transporter protein may eliminate toxins that target the ribosome from the cell, but this has not been established. The translational defect can even be limited to the germline using the C. elegans eft-3(q145) germline-translation defective mutant, which is sterile, but with normal translation in somatic cells. In the eft-3(q145) mutant, pgp-5p::gfp is strongly induced in the intestine when the animals are fed the benign E. coli OP50 ( Figure S1A; Figure 1A). Feeding K. rhizophila rather than E. coli to C. elegans eft-3(q145); pgp-5p::gfp animals disrupts the induction of pgp-5p::gfp ( Figure S1A; Figure 1A To establish that K. rhizophila can suppress the surveillance of a range of defects in translation, we tested the induction of the pgp-5p::gfp xenobiotic detoxification gene reporter after gene inactivation by RNAi of other genes that encode ribosomal protein subunits or translation factors. For example, growth of wild type C. elegans carrying pgp-5p::gfp on E. coli engineered to produce rpl-1 dsRNA or vrs-2 dsRNA, which inactivate the RPL-1 ribosomal protein or the VRS-2 tRNA synthetase genes, caused induction of pgp-5p::gfp. By contrast, in animals fed on either rpl-1 dsRNA or vrs-2 dsRNA and transferred to K. rhizophila, pgp-5p::gfp expression was decreased ( Figure S1B-C). The suppression of surveillance pathways by K. rhizophila is specific for translational defects because K. rhizophila does not suppress the induction of the mitochondrial stress response or endoplasmic reticulum stress response (Govindan et al., 2015). Constant exposure to K. rhizophila is necessary to suppress C. elegans translational surveillance: eft-3(q145);pgp-5p::gfp animals fed on K. rhizophila and transferred after various times to E. coli OP50 plates re-expressed pgp-5p::gfp within 12 hours of transfer ( Figure S1D). To identify the K. rhizophila genetic pathways responsible for the inhibition of pgp-5p::gfp induction by a C. elegans deficit in translation, we conducted a forward genetic screen after EMS mutagenesis of K. rhizophila for bacterial mutant strains that can no longer inhibit pgp-5p::gfp induction in eft-3(q145). ~2000 individual K. rhizophila strains that grew normally on bacterial LB plates after EMS mutagenesis were fed to C. elegans eft-3(q145); pgp-5p::gfp animals, one bacterial mutant per well, and screened for K. rhizophila mutants that failed to suppress C. elegans pgp-5p::gfp induction in a population of 50-100 eft-3(q145); pgp-5p::gfp animals. We identified six K. rhizophila mutant strains that failed to inhibit the induction of pgp-5p::gfp in the eft-3(q145); pgp-5p::gfp strain (Figure S1E-F). All these Kocuria mutant strains had defects in colony pigmentation compared to wild type K. rhizophila ( Figure S2A). Wildtype K.
Genome sequencing of 23 of these pigmentation mutants revealed that each carried a mutation in one of six carotenoid biosynthetic cluster genes ( Figure S2D-E). Carotenoids are yellow to red colored pigments, which are produced by a terpenoid biosynthetic pathway (Takarada et al., 2008). The reaction catalyzed by GGPP synthase CrtE, phytoene synthase CrtB and phytoene desaturase CrtI mediate steps in the production of lycopene (Klassen, 2010) (Krubasik et al., 2001a). CrtEb and CrtYe/f cyclases catalyze the biosynthesis of C50 carotenoid from lycopene. C50 carotenoids are synthesized more rarely than other carotenoids (Krubasik et al., 2001a). Six missense mutations (e21, e11, e23, e14, e5, and e13) and two nonsense mutations (e15 and e10) were in crtI, which encodes phytoene desaturase ( Figure 1C; Figure S2D-E). CrtI catalyzes the conversion of the non-colored phytoene to lycopene, which is red. All these crtI mutants are white colored colonies ( Figure   S2D; Figure 1C Figure S2F) and thus are predicted to be defective in lycopene synthesis. The six missense mutations are in highly conserved residues suggesting that these are strong loss of function mutations ( Figure S3). The e4, e6, and e8 are missense mutations in the crtB gene, which encodes phytoene synthase ( Figure 1C; Figure S2D-E). These missense mutations are in highly conserved residues, suggesting that these are strong loss of function mutations ( Figure S4). The mutants produce white bacterial colonies ( Figure S2D; Figure 1C), as was observed in the C. glutamicum DcrtB mutant (Heider et al., 2012) ( Figure S2F). We obtained four mutations in crtEb; two nonsense mutations (e16 and e17) and two missense mutations (e3 and e19) in highly conserved residues ( Figure 1C; Figure S2E; Figure S5). Mutations in crtEb are likely to be defective in the conversion of lycopene to flavuxanthin ( Figure 1C).
These mutants form pale red colonies (Supplementary Fig. 2d; Fig. 1c glutamicum crtYe and crtYf respectively. Therefore, the yellow pigment produced by K.
One trivial explanation for the failure of the K. rhizophila carotenoid mutants to suppress pgp-5p::gfp in a translation-defective C. elegans mutant would be that these K. rhizophila pigmentation mutants might induce pgp-5p::gfp even in a wildtype C. elegans background. But wild type C. elegans carrying pgp-5p::gfp grown on the K. rhizophila crtEb(e17), crtYe(e22), crtYf(e18), crtB(e6) or crtI(e10) mutants do not induce pgp-5p::gfp (Table S1; Figure S7A). Another possible interpretation was that K. rhizophila feeding might induce other stress responses in C. elegans that somehow "distract" the animal from surveillance of translation. We tested the induction of other GFP fusion reporters of stress on wild type and various K. rhizophila mutants, and found that K. rhizophila wildtype or carotenoid mutants did not induce hsp-4p::gfp (( Figure S7B), hsp-6p::gfp ( Figure S7C (Sutthiwong and Dufossé, 2014). For three of the elution peaks at 19, 16, and 9 min, high resolution mass spectra contained ions that were consistent with decaprenoxanthin and its mono-and di-glucosides, respectively, with mass accuracy within 2 ppm ( Figure 3C). Three other peaks with similar absorbance spectra were not assignable by mass.
To analyze the carotenoid production in the K. rhizophila mutants, we extracted carotenoids from crtI(e10), crtEb(e17), crtYe(e22), and crtYf(e18), which are nonsense mutant alleles and crtB(e6), a missense mutant. Spectrophotometric analysis of methanol extracts from wildtype K. rhizophila showed absorption maxima of 415-425nm, whereas the crtEb(e17) extract showed absorption maxima at 445-455nm ( Figure S9). K. rhizophila crtEb(e17) and crtYe(e22) mutant methanol extract show similar absorption spectra while the extract from K. rhizophila crtI(e10) and crtb(e6) showed no absorption at all. The methanol extracts from crtYf(e18) showed two separate absorption peaks one at ~400 nm and another minor one at ~500nm. To determine the identities of these absorption peaks in the mutants, we conducted LC-MS analysis of carotenoid extracts from crtYe(e22) and crtEb(e17) ( Figure   S10A-C). Mutations in crtEb are predicted to block the synthesis of carotenoids at the flavuxanthin biosynthesis step and thus are likely to accumulate lycopene ( Figure S10C). Consistent with this prediction, MS spectral analysis revealed that crtEb(e17) mutants accumulate lycopene. Mutations in crtYe are predicted to block the synthesis of carotenoids at the decaprenoxanthin biosynthesis step and thus likely to accumulate flavuxanthin ( Figure   S10C). However, we found that crtYe(e22) mutants accumulate lycopene as well as other unidentified peaks.
We tested whether the crude methanol extracts from wildtype K. rhizophila, containing pigmented carotenoids, could suppress GFP induction in the germline-translation defective C. elegans eft-3(q145);pgp-5p::gfp animals. Animals fed on E. coli with the K. rhizophila carotenoid extract exhibited significantly reduced pgp-5p::gfp expression compared to eft-3(q145);pgp-5p::gfp animals fed on E. coli with control methanol extract ( Figure 3D). The wild type K. rhizophila carotenoid methanol extract could also rescue the failure to suppress C. elegans surveillance by K. rhizophila carotenoid biosynthetic mutants: when eft-3(q145);pgp-5p::gfp animals were fed on K. rhizophila crtEb(e17), crtB(e6) or crtI(e10) mutants supplemented with wildtype K. rhizophila methanol extract, GFP was not induced while in the animals fed on control extract, the GFP expression was induced by the C. Consistent with the genetic analysis of K. rhizophila and C. glutamicum highlighting the production of C50 decaprenoxanthin as key to suppression of C. elegans immune responses, we found that other commercially available 40 carbon carotenoids in the pathway to decaprenoxanthin C50 biosynthesis cannot suppress the induction of pgp-5p::gfp in a C. elegans translation defective mutant. For example, treating eft-3(q145);pgp-5p::gfp animals grown on E. coli OP50 with either C40 beta-carotene or C40 astaxanthin did not suppress pgp-5p::gfp induction ( Figure S11B). C40 zeaxanthin, C40 neurosporene, C40 violaxanthin, C40 delta-carotene, or C40 alpha-carotene also did not suppress pgp-5p::gfp induction ( Figure   S11C). These C40 carotenoids cannot suppress the response to translational deficits like the C50 carotenoids produced by K. rhizophila and C. glutamicum.
Suppression of C. elegans translational surveillance by Kocuria rhizophila carotenoids enhances the toxicity of ribosomal toxins K. rhizophila also suppresses the C. elegans detoxification response to translation inhibiting drugs. Hygromycin is a bacterially produced antibiotic (from Streptomyces hygroscopicus) that inhibits translation and induces xenobiotic detoxification in C. elegans.
While 10 µg/ml of hygromycin induces pgp-5p::gfp expression in wild type animals grown on E. coli OP50, wild type animals grown on K. rhizophila and 10µg/ml of hygromycin fail to induce pgp-5p::gfp ( Figure S7A; Figure S11D). However, at high concentrations of hygromycin, pgp-5p::gfp is induced both in animals fed on E. coli OP50 or K. rhizophila ( Figure S11D). By contrast, wild type C. elegans carrying pgp-5p::gfp grown on the K. rhizophila carotenoid biosynthesis mutants crtI(e10) or crtEb(e17) induced GFP at the low concentration of 10µg/ml of hygromycin unlike animals grown on wild type K. rhizophila that produces the carotenoid ( Figure S7A). Similar results were obtained with emetine which blocks protein synthesis by binding to the 40S subunit of the ribosome; wild type K. rhizophila suppressed the pgp-5p::gfp induction by a low dose of emetine but the carotenoid mutant K. rhizophila now allowed induction of pgp-5p::gfp by a low dose of emetine ( Figure S12A-B).
We tested whether the suppression of drug detoxification responses by K. rhizophila carotenoids increase C. elegans sensitivity to translational inhibitors. While wildtype animals grown on 10µg/ml hygromycin when fed E. coli OP50 are not growth inhibited, animals grown on 10µg/ml hygromycin plus K. rhizophila carotenoid extract when fed E. coli OP50 are strongly growth inhibited ( Figure 4B; Figure S12C). Carotenoids without hygromycin are not toxic to the worms ( Figure 4B; Figure S12C). Similar results were obtained with emetine: animals treated with K. rhizophila extracts were hypersensitive to emetine compared to animals fed on control extract ( Figure 4C). The xenobiotic hypersensitivity phenotype is specific to translation defects because K. rhizophila carotenoid extract does not alter the sensitivity of animals to antimycin, a mitochondrial poison ( Figure S12E).
pgp-5p::gfp is also induced in response to genotoxic stress: cisplatin, which interferes with DNA replication, induces pgp-5p::gfp. While 1 mM cisplatin induces pgp-5p::gfp expression in wild type C. elegans grown on E. coli OP50, wild type C. elegans grown on K.
C. elegans food aversion behaviors are induced when animals are exposed to xenobiotics or essential gene inactivations (Melo and Ruvkun, 2012). Exposing animals to hygromycin or cisplatin induces strong food aversion, and this food aversion is modulated by K. rhizophila carotenoid extract ( Figure 4E-F). While ~40% of animals exposed to 25µg/ml hygromycin display food aversion behavior, only ~15% of animals exposed to 25µg/ml hygromycin and K. rhizophila carotenoid extract show aversion ( Figure 4E). Similar results were obtained with cisplatin: ~50% of animals exposed to 1mM cisplatin display aversion behavior while <20% of animals exposed to 1mM cisplatin and K. rhizophila carotenoid extract display food aversion ( Figure 4F). . Constitutive expression of ZIP-2::mCherry from an intestine-specific promoter fusion is sufficient to induce pgp-5p::gfp expression in wildtype C. elegans without translation inhibition (Fig. 5a,b). pgp-5p::gfp induction in this ZIP-2::mCherry constitutive expression strain was similar in animals fed on E. coli OP50 or wild type K. rhizophila ( Figure   A

Discussion
We have found that multiple actinobacterial species produce C50 carotenoid pigments that are potent inhibitors of the animal surveillance pathways for bacterial toxins and virulence factors which target translation. We genetically dissected how the actinobacteria Kocuria rhizophila inhibits the surveillance and detoxification response of C. elegans to translational deficits. Our bacterial genetic analysis identified the carotenoid biosynthetic pathway as the key bacterial countermeasure. We isolated mutants in the K. rhizophila C50 carotenoid biosynthetic pathway that failed to inhibit this C. elegans xenobiotic detoxification response ( Figure 1A-C). Hydrophobic and pigmented extracts of K. rhizophila enriched for C50 carotenoid suppressed C. elegans xenobiotic detoxification response and restored the ability of K. rhizophila carotenoid mutants to suppress the C. elegans response to defects in translation ( Figure 2B-C). The K. rhizophila extracts also suppress C. elegans detoxification of toxins that target translation or DNA damage, rendering the animal far more sensitive to these toxins ( Figure 4B-D, Figure S12C). Thus, carotenoid biosynthesis increases the potency of other bacterial toxins or virulence factors that might target translation or DNA.
A distantly related bacterium, Corynebacterium glutamicum, also suppresses this C. It is possible that the pigmentation of the C50 carotenoids, the particular wavelengths of light absorbance, are key to their anti-surveillance function. For example, our genetic analysis showed that the carotenoids of K. rhizophila or C. glutamicum do not function at lengths less than C50, suggesting that simple hydrophobicity of C40 or smaller carotenoids is not sufficient to mediate K. rhizophila or C. glutamicum inhibition of C. elegans translational surveillance. The specific color absorbance of the C50 carotenoids (their red color) may mediate their C. elegans anti-surveillance activity. For example, C. elegans with a translational deficit may produce blue light (Ohmiya and Hirano, 1996), and the absorption of such light by C50 carotenoids could suppress a light-responsive C. elegans surveillance pathway. Such a system might function in the dark of the soil or rotting fruit or night, but might be overwhelmed by sunlight.
Translation surveillance not only induces detoxification but also food aversion behaviors in C. elegans. Aversion to food associated with a toxin is an appropriate animal response since many toxins originate from bacterial pathogens that cause the rotting of food.
Aversion to toxins is a common animal response since many toxins originate from a pathogen, and this response is likely to be an animal program derived from this evolutionary history. K. rhizophila C50 carotenoids suppressed the food aversion induced by emetogenic toxins, emetine or cisplatin. Cisplatin, which is used to block DNA replication in cancer patients, also has high emetogenic potential. Emetine, which is an antibiotic that targets eukaryotic protein synthesis, is also highly emetogenic as the name itself implies. These two drugs not only induce C. elegans xenobiotic detoxification but also strong food aversion behavior, both of which are strongly suppressed by bacterial C50 carotenoids.
Chemotherapy-induced nausea and vomiting (CINV) in humans may be related to these xenobiotic aversion programs. The K. rhizophila C50 carotenoid decaprenoxanthin might for example suppress CINV, perhaps as an adjuvant in chemotherapy. We have hypothesized previously that autoimmune disorders may be triggered by reduction of function mutations in core cellular pathways such as translation that activate these surveillance and detoxification pathways (Govindan et al., 2015). Because the C50 carotenoid decaprenoxanthin robustly suppresses C. elegans translational surveillance, it may suppress the inappropriate autoimmune responses to such mutations in mammals. More broadly, bacterial gene activities that suppress animal toxin surveillance pathways could neutralize the toxicity that often plagues drug development.

Acknowledgments
We thank members of the G.R. laboratory for helpful discussions. The work was supported by NIH Grant AG043184 (to G.R.). Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440).      Synchronized wildtype animals treated with solvent control or K. rhizophila extract from L1-larval stage were exposed to antimycin at L3-larval stage and scored for development 60 hours later at 20 o C.
Supplementary Table 1 Analysis of the effects of K. rhizophila mutants on gfp induction in eft-3(q145); pgp-5p::gfp animals and pgp-5p::gfp animals. Growth and handling of microbes used:

Materials and methods
16S ribosomal sequence was amplified using specific primers and sequenced to identify the microbes. LB media as well as plates was used for culturing Kocuria rhizophila, Arthrobacter arilaitensis and Corynebacterium glutamicum and its mutants. 500 µl of overnight culture was seeded onto SK media plates and incubated at room temperature for 2 days before initiating the experiments. For experiments involving Kocuria rhizophila wildtype and mutants, Arthrobacter arilaitensis, and Corynebacterium glutamicum wildtype and as well as mutants, synchronized L1-larval stage animals grown until L4-larval stage or day one of adulthood in E. coli OP50 seeded plates and was washed in M9 buffer at least five times before transferring to the appropriate bacterial food.

Drug treatments:
Hygromycin diluted in M9 solution to the desired concentration was added onto preseeded E.coli OP50 bacteria containing NGM plates. Stock solution of emetine or cisplatin was diluted in M9 and the desired concentration was added onto preseeded E.coli OP50 bacteria containing NGM plates. 750µg/ml of K. rhizophila extracts was added onto preseeded E.coli OP50 bacteria containing NGM plates containing appropriate concentrations of hygromycin or cisplatin or emetine. For the xenobiotic experiments, synchronized L1-stage animals were dropped onto the drug containing plates and scored 4 days later.
Food aversion assay was performed by adding either hygromycin or cisplatin to E. coli OP50 lawns that contained synchronized L4 larval stage worms that were treated with 750µg/ml of K. rhizophila extracts. Aversion index was calculated by counting number of animals off lawn over number of total animls counted at desired time point (Aversion index=N(animals off lawn)/N(total animals). Each experiment was conducted in triplicate and the experiments were repeated in three independent trials.

RNAi Assays:
For RNAi assays synchronized L1 larval stage animals of the appropriate genotype were fed with appropriate RNAi clones until they reach day one of adulthood. Subsequently, the RNAitreated animals were washed in M9 at least five times to remove the E. coli bacteria and transferred to K. rhizophila seeded plates or E. coli OP50 seeded plates.

Microscopy
Nematodes were mounted onto agar pads and images were acquired using a Zeiss AXIO Imager Z1 microscope fitted with a Zeiss AxioCam HRm camera and Axiovision 4.6 (Zeiss) software. All the fluorescence images shown within the same figure panel were collected together using the same exposure time. Images were converted to 8-bit image, thresholded and quantified using ImageJ. Student's t test was used determine statistical significance.
Low-magnification bright-field and GFP fluorescence images were acquired using a Zeiss AxioZoom V16, equipped with a Hammamatsu Orca flash 4.0 digital camera, and using Axiovision ZEN software.

K. rhizophila EMS mutagenesis screen
Mutagenesis was performed by treatment of overnight culture of K. rhizophila in PBS solution with 50 mM EMS for 45 minutes at 37°C. Serial dilutions of the mutagenized K. rhizophila cultures were plated onto LB media plates and ~2000 mutagenized bacterial colonies were picked and grown in LB solution. 500 µl of overnight culture was seeded onto SK media plates and incubated at room temperature for two days before initiating the experiments.
Synchronized L1-larval stage in eft-3(q145);pgp-5p::gfp animals grown until L4-larval stage or day one of adulthood in E. coli OP50 seeded plates and was washed in M9 buffer at least five times before transferring to the K. rhizophila mutant bacterial food. The plates were visually screened after 24 hours for GFP induction.

Isolation of K. rhizophila carotenoids
Carotenoid isolation from K. rhizophila was isolated as described 1 with the following modifications. K. rhizophila cultures grown in LB solution was washed with equal volume of water after centrifugation at 4000RPM for 15 min. After centrifugation to remove water, equal volume of acetone was added and centrifuged again at 4000RPM for 15 min. After removal of acetone, the bacterial pellets were extracted with methanol at 65 o C in water bath after wrapping the samples with aluminum foil to protect from light. The samples were extracted with methanol multiple times. The supernatant was filtered with Whatman filter paper No1.
Two-volumes of 15% sodium chloride was added to the methanol extract and after mixing equal volume of hexane was added. The yellow carotenoids were separated from the methanol-salt mix and accumulated in the hexane fraction. The hexane fraction was removed and washed at least three times with water. The hexane fraction was evaporated and the resultant carotenoid pellet was dissolved in methanol.

High performance liquid chromatography
Crude methanolic extracts were separated over an Agilent Eclipse Plus C18 4.6 x 250 mm column with a 5-micron particle size using an Agilent 1200 HPLC equipped with a diode array detector, autosampler, column oven, solvent degasser, and binary pump. The mobile phases were (A) water vs. (B) methanol at a flow rate of 2 mL/min. The column was preequilibrated at 40°C with 90% B prior to sample injection. Following injection, the column was washed isocratically for 5 min at 90% B before ramping to 100% B over 5 min. Eluate absorbance spectra were monitored from 300-700 nm.

Liquid chromatography mass spectrometry (LC-MS)
Methanol (MeOH) extracts were diluted in dichloromethane (DCM) and filtered over prewashed silica gel using 15:85 methanol/DCM. The visibly colored eluate was collected, aliquoted, and dried under vacuum. Dry aliquots were stored under Ar at -30 °C and were resuspended just prior to analysis in a minimal volume of 1:9 water/MeOH. The material was separated over Waters XBridge C18 1 x 100 mm column with a 3.5-micron particle size at 25 °C using an Agilent 1200 HPLC equipped with a diode array detector, autosampler, column oven, solvent degasser, and binary pump. Compound peaks were identified by searching high resolution mass spectra by chemical formula from a database of known carotenoids, using a stringent mass error threshold (5 ppm), and then correlating extracted ion chromatograms (EIC) of candidate hits to 440 nm absorbance elution peak profiles.