Quantitative proteomics by amino acid labeling in C. elegans - Nature

brief communications

© 2011 Nature America, Inc. All rights reserved.

Julius Fredens, Kasper Engholm-Keller, Anders Giessing, Dennis Pultz, Martin Røssel Larsen, Peter Højrup, Jakob Møller-Jensen & Nils J Færgeman We demonstrate labeling of Caenorhabditis elegans with heavy isotope–labeled lysine by feeding them with heavy isotope–labeled Escherichia coli. Using heavy isotope– labeled worms and quantitative proteomics methods, we identified several proteins that are regulated in response to loss or RNAi-mediated knockdown of the nuclear hormone receptor 49 in C. elegans. The combined use of quantitative proteomics and selective gene knockdown is a powerful tool for C. elegans biology.

Quantitative proteomics is increasingly being applied to examine and understand how cells and organisms regulate their metabolism to support growth, proliferation, differentiation, development and survival1. Several strategies for quantitative proteomics have been developed, including chemical labeling such as isobaric tags for relative and absolute quantitation (iTRAQ) and dimethyl labeling, or metabolic labeling using heavy isotopes2. Particularly, metabolic labeling with stable isotopes has become the prevailing strategy to quantitatively compare proteomes of cells and organisms. Stable-isotope labeling by amino acids in cell culture3, known as SILAC, is a precise mass spectrometry–based quantitative strategy, which provides a defined number of labels per identified peptide and therefore enables easier and more comprehensive peptide identification and data analysis compared to metabolic labeling with 15N (refs. 4,5). Although labeled amino acid–based approaches have been applied primarily to cell cultures, SILAC has proven to be highly valuable for studies in in vivo systems such as yeast6,7, fruit flies4, plants8 and mice9.

Quantitative proteomics has been applied only to a limited extent to study signaling events governing metabolism in C. elegans5,10,11. Previously, metabolic 15N-labeling of nematodes has been achieved by feeding them uniformly labeled E. coli5. Therefore, we hypothesized that C. elegans feeding on an auxotroph bacteria could incorporate an exogenously supplemented labeled amino acid into their proteome, which subsequently could be monitored by mass spectrometry (Fig. 1). Usually SILAC is performed using labeled lysine and arginine because all peptides resulting from a tryptic digestion will be labeled. To label C. elegans with lysine and arginine, we attempted to create a double auxotroph E. coli strain but could not create and maintain such a strain. As labeling with arginine can compromise the sensi tivity because of metabolic conversion of arginine to proline12, we decided to label with lysine only. We prepared overnight cultures of a lysine auxotroph E. coli strain (ET505) by growing this strain in defined medium supplemented with lysine labeled with either light isotope (Lys0; 12C6,14N2) or heavy isotope (Lys8; 13C ,15N ), which we used to seed nutrient-deprived agar plates 6 2 (Fig. 1). We subsequently placed unlabeled day 1 larvae (L1) (N2 strain) on these plates and grew them for 2–3 d until early adult stage. We did not observe any developmental or phenotypic differences of C. elegans grown on E. coli labeled with either light or heavy lysine. We assayed the extent of heavy lysine incorporation in total protein extracts of N2-strain worms grown on Lys8-labeled bacteria to different stages by liquid chromatography–tandem mass spectrometry (LC-MS/MS) after digestion of a C. elegans protein lysate with lysyl endopeptidase (Lys-C) (Supplementary Fig. 1a). The use of Lys-C, which cleaves the protein backbone on the C-terminal side of lysine, ensured the generation of peptides

Figure 1 | Labeling of C. elegans with heavy isotope–labeled lysine. A lysine auxotroph E. coli strain was grown in defined medium containing Lys0 or Lys8 (light and heavy isotopes, respectively). Synchronized L1 worms were grown for 2–3 d on light isotope– or heavy isotope–labeled E. coli. L4 worm populations were mixed 1:1 and analyzed by LC-MS/MS. Relative protein expression was determined by comparing the intensity of light and heavy peptide doublets with a mass difference of 8 Da (heavy: light ratio).

Unlabeled wild type

Unlabeled mutant E. coli (Lys8)

E. coli (Lys0)

‘Light’ wild type

Relative intensity

Quantitative proteomics by amino acid labeling in C. elegans

‘Heavy’ mutant

8 Da

m/z

Department of Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark. Correspondence should be addressed to N.J.F. ([email protected]). Received 2 February; accepted 26 July; published online 28 august 2011; doi:10.1038/nmeth.1675

nature methods | VOL.8 NO.10 | OCTOBER 2011 | 845

carrying at least one label except for the C-terminal peptide. Digestion with Lys-C yields longer peptides than digestion with trypsin, but as seen in this study, this strategy is sufficient for large-scale protein quantification. Based on the average of the 100 most intense proteins, we found almost complete incorporation (94.6 ± 2.8% (± s.d.; n = 100)) of Lys8 in L4 worms in the first generation, which increased to 97.5 ± 2.2% in L4 worms grown for an additional day. Labeling worms for two generations did not increase the extent of labeling (Supplementary Fig. 1a). In addition, we mixed labeled and unlabeled worms in known ratios to evaluate the accuracy of the method (Supplementary Fig. 1b). We obtained s.d. less than 0.22, showing that 68.2% of the quantified ratios deviated less than log2 0.22-fold change from the expected ratio. RNAi-mediated knockdown of gene expression has been proven to be a highly powerful way to examine gene function in C. elegans13. We combined quantitative proteomics and RNAi by making the lysine auxotroph E. coli strain ET505 RNAi-compatible. First, to express the T7 polymerase in presence of isopropyl β-d-1thiogalactopyranoside (IPTG), we integrated a DE3 construct into the genome of E. coli strain ET505. Next, by P1 phage– mediated transduction of a deletion-insertion mutation (∆rnc-38), we deleted the gene encoding dsRNA-specific endonuclease RNaseIII to prevent degradation of the overexpressed dsRNA. The modified E. coli strain named NJF01 grew slightly slower than the parental strain but remained a lysine auxotroph (Supplementary Fig. 2). The effect on growth may be attributed to the genetic manipulation carried 6 a out to allow efficient expression of dsRNA. The extent of lysine incorporation into L4 4 C. elegans grown on E. coli strain NJF01 was virtually unchanged compared to that 2 on strain ET505 (Supplementary Fig. 1a and Supplementary Fig. 3). –6 –4 –2 We found that RNAi-mediated knockdown of dpy-13 resulted in almost com–2 plete penetrance of the expected shortened phenotype (dumpy) when using either the –4

commonly used RNAi bacteria strain HT115 or NJF01 irrespective of whether we used regular RNAi–nematode growth media (NGM) plates or minimal RNA-plates (Supplementary Fig. 4a). We also knocked down gfp in worms expressing a GFP–RAB-7 fusion protein primarily in the intestine and monitored GFP signal by fluorescence microscopy and by western blotting (Supplementary Fig. 4b–e). As above, we found that gfp expression was knocked down efficiently in C. elegans using either E. coli strain NJF01 or HT115 independently of the plates used (Supplementary Fig. 4d). To combine RNAi with quantitative proteomics, we fed transgenic L4-stage worms expressing intestinal GFP–RAB-7 on Lys8-labeled E. coli strain NJF01 expressing dsRNA to gfp and mixed them 1:1 with GFP::RAB-7 transgenic worms fed on Lys0-labeled E. coli strain NJF01 transformed with empty L4440 expression vector (Supplementary Fig. 4f,g). Next, we separated worm proteins by SDS-PAGE and excised gel bands containing proteins with an approximate molecular weight corresponding to the GFP–RAB-7 fusion protein, which we then in-gel–digested with Lys-C. By LC-MS/MS analysis of the extracted peptides we only detected light peptides of GFP and RAB-7, whereas peptides not related to GFP and RAB-7 were present in a heavy:light ratio of ~1:1 (Supplementary Figs. 4 and 5), showing the combination of heavy lysine–labeling and RNAi in C. elegans. Expression of genes involved in lipid metabolism in C. elegans is coordinately controlled by several transcription factors, including the nuclear hormone receptor 49 (NHR-49) 14.

nhr-49 RNAi versus control (log2-fold change)

Figure 2 | Quantification of protein changes upon loss or knockdown of nhr-49. (a) In a label-switch experiment, heavy isotope– and light isotope–labeled nhr-49 and wild-type worms, respectively, were mixed, and proteomes were analyzed by LC-MS/MS. The log2-fold change values of all quantified proteins were compared to log2-fold change of the same proteins quantified in a similar experiment, except that nhr-49 was knocked down by RNAi, revealing a significant correlation (Spearman, r = 0.4764, P < 0.0001) between the two experiments. (b) Log2fold change of heavy:light isotope ratios in duplicate nhr-49 RNAi experiments were plotted each against the other, revealing a significant correlation (Pearson, r = 0.6474, P < 0.0001) between the duplicate experiments. (c) GOterm distribution of proteins identified to be upregulated (162 proteins) or downregulated (168 proteins), in response to RNAi-mediated knockdown of nhr-49.

b

2

4

6

6 Duplicate 2 (log2-fold change)


brief communications

–6

–4

nhr-49 mutant versus wild type (log2-fold change)

846 | VOL.8 NO.10 | OCTOBER 2011 | nature methods

–2

2

4

6

Duplicate 1 (log2-fold change)

–4

–6

Quantified proteins Regulated proteins

Downregulated

Upregulated 6

2

–2

–6

c

4

5

9 24

13

43 51

7

12

3 5

3 9 5

16

21

22 37

39 Cellular processes and signalling

Information storage and processing

Not assigned

Poorly characterized

Amino acid metabolism

Carbohydrate metabolism

Energy production conversion

Lipid metabolism

Inorganic ion, coenzyme and nucleotide metabolism

Secondary metabolism

Inorganic ion, nucleotide and secondary metabolites


brief communications We examined the proteome changes in response to either functional loss or RNAi-mediated knockdown of the transcript encoding NHR-49. First, in a label-switch experiment, we fed wild-type worms and nhr-49 mutants (nr2041) on ET505 bacteria prelabeled with either Lys0 or Lys8 as shown in Figure 1. When worms reached the L4 stage, we mixed equal amounts of Lys0- and Lys8-labeled worms, digested the total protein extract, separated it into 15 fractions by hydrophilic interaction liquid chromatography and analyzed the fractions by LC-MS/MS. We analyzed label-switch samples (Lys0-labeled wild type mixed with Lys8-labeled nhr-49 and Lys8-labeled wild type mixed with Lys0-labeled nhr-49) individually and analyzed each sample twice. We identified 3,949 proteins, of which we quantified 2,903 proteins with high confidence in both label-switch experiments, among which we identified 143 proteins to be significantly upor downregulated (B ≤ 0.05) (Fig. 2, Supplementary Fig. 6 and Supplementary Table 1). Next, to test the potential broad use of our combined quantitative proteomics and RNAi knockdown approach, we knocked down nhr-49 by RNAi in C. elegans by feeding worms for two generations with E. coli strain NJF01 carrying either RNAi control plasmid L4440 or an nhr-49 RNAi plasmid. In duplicate experiments, we identified 4,627 proteins, of which we quantified 3,470 proteins with high confidence in both replicates and identified 330 proteins to be significantly up- or downregulated (B ≤ 0.05). We found that the quantified proteins in the nhr-49 (nr2041) mutant experiments significantly (Spearman, P < 0.0001) correlated with the quantified proteins in the nhr-49 RNAi experiments (Fig. 2a) and found that 99 proteins were significantly (B ≤ 0.05) regulated in both nhr-49 mutants and in the worms subjected to RNAi knockdown. The two replicates in the RNAi experiments (Fig. 2b) or in the nhr-49 mutant experiments (data not shown) correlated significantly (Pearson, P < 0.0001). Lists of quantified and regulated proteins from both types of experiments are available in Supplementary Table 1. We categorized the proteins we identified to be regulated in response to RNAi knockdown (Fig. 2c and Supplementary Table 2) or functional loss of nhr-49 (data not shown) according to gene ontology (GO) terms. We manually assigned the regulated proteins using WormBase and clusters of euKaryotic Orthologous Groups (KOGs) at US National Center for Biotechnology Information. Both analyses revealed substantial overrepresentation of downregulated proteins involved in lipid metabolism, including fatty acid desaturases such as FAT-1, FAT-2, FAT-5 and FAT-6 and additional proteins predicted or shown to be involved in fatty acid metabolism (Supplementary Tables 1 and 2). By quantitative PCR it has recently been found that among 110 genes examined, the expression of 13 genes had been altered notably in response to functional loss of NHR-49 (ref. 15) (Supplementary Table 1). We found that the protein

abundance in nhr-49 mutants or nhr-49 knockdowns relative to control worms significantly correlated (r = −0.3826, P < 0.0018 and r = −0.4338, P = 0.0002, respectively) with the reported mRNA amounts 14 (Supplementary Fig. 6e,f). Consistently, a large part of the downregulated proteins we identified in both experiments have either been shown or predicted to be involved in β-oxidation of fatty acids or energy metabolism (Supplementary Tables 1 and 2). We believe that quantitative proteomics combined with mol ecular genetics to systematically examine gene function in C. elegans will greatly advance our understanding of how cellular signaling and metabolism are coordinated and hence how perturbations of cellular functions can lead to human diseases. Methods Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturemethods/. Note: Supplementary information is available on the Nature Methods website. Acknowledgments We thank K.E. Budtz and L. Jakobsen for technical assistance, S.R. Kushner (University of Georgia, Athens) for bacterial strains, and The Danish Research Council and the Novo Nordisk A/S Foundation for financial support. AUTHOR CONTRIBUTIONS J.F., K.E.-K., A.G., P.H., M.R.L., J.M.-J. and N.J.F. designed research; J.F., K.E.-K., A.G., D.P., P.H., J.M.-J. and N.J.F. performed research and analyzed data; J.F. and N.F. wrote the manuscript, which K.E.K., P.H., M.R.L. and J.M.J. improved. N.F., J.M.J., M.R.L. and P.H. provided funding. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/naturemethods/. Reprints and permissions information is available online at http://www.nature. com/reprints/index.html.

Markaki, M. & Tavernarakis, N. Biotechnol. J. 5, 1261–1276 (2010). Leitner, A. & Lindner, W. Methods Mol. Biol. 527, 229–243 (2009). Ong, S.E. et al. Mol. Cell. Proteomics 1, 376–386 (2002). Sury, M.D., Chen, J.X. & Selbach, M. Mol. Cell. Proteomics 9, 2173–2183 (2010). 5. Krijgsveld, J. et al. Nat. Biotechnol. 21, 927–931 (2003). 6. Gruhler, A. et al. Mol. Cell. Proteomics 4, 310–327 (2005). 7. Jiang, H. & English, A.M. J. Proteome Res. 1, 345–350 (2002). 8. Gruhler, A., Schulze, W.X., Matthiesen, R., Mann, M. & Jensen, O.N. Mol. Cell. Proteomics 4, 1697–1709 (2005). 9. Kruger, M. et al. Cell 134, 353–364 (2008). 10. Dong, M.Q. et al. Science 317, 660–663 (2007). 11. Schrimpf, S.P. & Hengartner, M.O. J. Proteomics 73, 2186–2197 (2010). 12. Ong, S.E. & Mann, M. Nat. Protoc. 1, 2650–2660 (2006). 13. Kamath, R.S. et al. Nature 421, 231–237 (2003). 14. Van Gilst, M.R., Hadjivassiliou, H. & Yamamoto, K.R. Proc. Natl. Acad. Sci. USA 102, 13496–13501 (2005). 15. Van Gilst, M.R., Hadjivassiliou, H., Jolly, A. & Yamamoto, K.R. PLoS Biol. 3, e53 (2005). 1. 2. 3. 4.

nature methods | VOL.8 NO.10 | OCTOBER 2011 | 847


ONLINE METHODS Growing C. elegans in culture. Strains used were N2-Bristol (wild type), nhr-49 (nr2041) and GH197 (glo-1(zu437)X;pwIs170[vha6::GFP::rab-7;unc-119(+)]. Strains were kept on standard NGM plates and maintained in culture as described elsewhere 16. For labeling studies worms were grown at 20 °C on minimal plates consisting of 12 mg ml−1 agarose, 3 mg ml−1 NaCl, 5 µg ml−1 cholesterol, 1 mM CaCl2, 1 mM MgCl2 and 25 mM K2HPO4 (pH 6.0). For RNAi studies, plates were supplemented with 1 mM IPTG and 25 mM carbenicillin. Plates were seeded with 50-times concentrated overnight culture of E. coli strain ET505 or NJF01. Bacteria cultures were prepared in EZ rich defined medium 17 containing appropriate antibiotics and 0.4 mM lysine (Lys0), 96–98% enriched [4,4,5,5-D4]lysine (Lys4) or 97–99% enriched [13C6,15N2]lysine (Lys8) (Cambridge Isotope Laboratories). Bacterial strains. E. coli strain ET505 (F−, λ− lysA0::Tn10 IN(rrnD-rrnE)1) was obtained from the Coli Genetic Stock Center (CGSC number 7088). A DE3 construct was integrated using a λDE3 lysogenization kit18 (Novagen). rnc was mutated by P1 phage transduction of ∆rnc-38 from E. coli strain SK7621 (ref. 19) (gift from S.R. Kushner, University of Georgia). The resulting strain ET505(DE3) ∆rnc-38 was named NJF01. The strain is available upon request. We have not been able to construct a double lysine and arginine auxotrophic strain as the viability of the strain was compromised. Protein extraction. Worms were washed off the plates and incubated in 0.9% NaCl for 20 min to empty the intestine. Heavy isotope–labeled (‘heavy’) and light isotope–labeled (‘light’) populations were mixed 1:1, and lysed in 2% SDS, 50 mM TrisHCl (pH 6.9), 10% glycerol and 1 mM DTT by sonication in Bioruptor (Diagenode). We reduced 200 µg protein with DTT, and alkylated the sample with iodoacetamide. Detergents were removed by methanol-chloroform precipitation20. Resulting protein pellet was resuspended in 20 µl 6 M urea, 2 M thiourea and digested with 2 µg Lys-C (Wako Pure Chemical Industries) in 20 µl triethyl-ammonia bicarbonate. Remaining fatty acids were precipitated by adding trifluoroacetic acid (TFA) to 1%. Peptides were desalted on R2 microcolumns21. Half of the eluate was dried and redissolved in 90% acetonitrile, 0.1% TFA before hydrophilic interaction liquid chromatography (HILIC) fractionation before reverse-phase LC-MS/MS. SDS-PAGE and in-gel digestion. Whole worms were lysed in loading buffer, subjected to SDS-PAGE on 4–12% Bis-Tris gels, and protein bands were excised. Proteins were reduced, alkylated and digested with Lys-C. Extracted peptides were subjected to microcolumn purification as described above, resuspended in 5% formic acid and analyzed by LC-MS/MS. Western blotting. Sixty worms were picked directly into electro phoresis buffer and subjected to SDS-PAGE. After blotting, proteins were monitored using antibodies to GFP (Abcam). Antibodies to β-actin (Sigma) were used as control. Amino acid LC-MS. LC-MS analysis was performed on an Agilent XCT Ultra 6340 Ion-Trap using a 0.32 mm × 150 mm, five micrometer (average diameter) particle Hypercarb (Thermo nature methods

Scientific) column as stationary phase. Gradient separation of amino acids was achieved using 1 mM perfluoroheptanoic acid (PFHA, referred to as buffer A) and acetonitrile (referred to as buffer B) at a flowrate of 5 µl min−1 at 30 °C. Gradient profile was with an initial 3-min isocratic step at 0% buffer B followed by increases to 15% buffer B at 10 min, 26% buffer B at 20 min, 50% buffer B at 25 min and hold at 50% buffer B for 5 min. Column was subsequently re-equilibrated at 0% buffer B with a total runtime of 40 min. Samples of hydrolyzed proteins were dissolved in buffer A, and injection volumes were kept below 1 µl. The IonTrap mass spectrometer was operated in positive ion electrospray mode, and mass spectra were recorded as averages of 3 spectra by scanning in the 50–250 m/z range. Mass spectrometry parameters were optimized for best sensitivity by infusion of amino acid solutions at 5 µl min−1. For routine analysis, the individual amino acids were identified by their unique retention time and pseudomolecular ion, [MH]+ (Supplementary Fig. 3a). The method was initially verified by recording MS/MS spectra and comparing them to amino acid fragmentation patterns found in literature22. Using this setup we could separate lysine (m/z of 147) and heavy lysine (m/z of 151) from glutamate (m/z of 148) and methionine (m/z of 150), thereby avoiding potential isotopic overlap. HILIC fractionation of Lys-C peptides. Peptides were fractionated using an in-house–made capillary column of fused silica capillary tubing (0.32 mm × 180 mm) (Polymicro Technologies) and a polyether ether ketone (PEEK) inline microfilter (Upchurch Scientific), packed with 3 µm TSKGel Amide 80 HILIC resin (Tosoh Bioscience). A capillary flow HPLC system, model 1200 (Agilent) was used to deliver a gradient of 100–60% solvent B (solvent A = 0.1% TFA; solvent B = 90% can and 0.1% TFA) in 30 min at a flow rate of 6 µl min−1. Fractions were collected every minute and combined based on intensity measured by UV-light detection to yield 15 fractions in total, which were lyophilized. nanoLC-electrospray ionization MS/MS. Peptide fractions were analyzed using automated data-dependent acquisition on an LTQ Orbitrap XL (Hybrid-2D-Linear Quadrupole Ion Trap-Orbitrap) mass spectrometer (Thermo Scientific). Each mass spectrometry scan was acquired without lock-mass at a resolution of 30,000 and was followed by five MS/MS scans using collision-induced dissociation, except that seven MS/MS scans were carried out for the nhr-49 RNAi experiment. A nanoflow HPLC system (EASY-nLC, Proxeon) was used for online reverse phase chromatographic separation of the peptide sample before nano-electrospray ionization (ESI) and mass spectrometry detection. Peptides were eluted at 400 nl min−1 using solvent A (0.1% formic acid) and solvent B (95% acetonitrile and 0.1% formic acid) on a 0.1 mm × 180 mm fused silica capillary column packed with ReproSil-Pur C18 AQ 3 µm reverse phase material (Dr. Maisch GmbH) and the following gradients: for the nhr-49 mutant experiment 0–34% solvent B in 100 min and for the nhr-49 RNAi experiment 10–34% solvent B in 120 min. Data analysis. Raw data from the LTQ Orbitrap XL instrument was processed using the Quant.exe module in the MaxQuant software version 1.0.13.13 (ref. 23). The generated peak list was searched against a concatenated forward-reverse C. elegans SwissProt database (release date 8 June 2010) using an in-house Mascot doi:10.1038/nmeth.1675


server (version 2.3.2, Matrix Science). The database searches were performed with carbamidomethyl (C) as fixed modification and acetylation (protein N terminus) and oxidation (M) as variable modifications. Lys8 was chosen as heavy label. Enzyme specificity was selected as Lys-C with up to two missed cleavages allowed using 7 p.p.m. peptide tolerance ion and 0.6 Da MS/MS tolerance. The peptide and protein identifications from Mascot were further processed in the second MaxQuant module Identify.exe using default settings (protein and peptide false discovery rate ≤ 1%). The resulting proteingroups.txt output file from MaxQuant containing the peptide identifications was imported into Microsoft Excel, in which additional filtering was performed (protein posterior error probability (PEP) ≤ 0.01). In control experiments, at least two ratio counts were required for quantification. In studies of nhr-49 mutants or nhr-49 RNAi, at least two ratio counts were required in each duplicate or label-switch experiment for quantification, and proteins regulated similarly and with significance B level ≤ 0.05 in both duplicates or label-switch experiments were considered to be regulated. High-abundance proteins have high, intense signals and accordingly will be determined more accurately than low-abundance proteins. Significance B values group proteins of similar intensities into bins of equal occupancy and estimate the s.d. of protein ratios using the subset grouped into the same intensity bins. To functionally assign the identified and regulated proteins we manually assigned genes by searching the gene names and extracted the GO terms applied in NCBI KOGs (a eukaryotespecific version of Conserved Orthologous Groups) via WormBase. For LC-MS/MS of gfp knockdown sample, the peptides were identified using Mascot, and the spectra were manually

doi:10.1038/nmeth.1675

validated. For quantification, area under curve was calculated for the respective extracted ion chromatograms. RNAi-mediated knockdown. Standard RNAi was performed using E. coli strain HT115 transformed with empty control plasmid (L4440) or RNAi clones to dpy-13, gfp or nhr-49 (from either A. Fire (Stanford University) or from Ahringer RNAi library collection)13 grown in LB at 37 °C over night and seeded onto NGM plates containing 1 mM IPTG and 25 µg ml−1 carbenicillin. E. coli strain NJF01 was rendered competent by rubidium chloride to transform it with the same control or RNAi plasmids. When combining RNAi with stable-isotope labeling, E. coli strain NJF01 containing same plasmids was grown in EZ medium and seeded onto minimal media plates (see above). Worms were subjected to RNAi from the L1 to L4 stage at 20 °C. Fluorescence microscopy. L4 worms were mounted on an agarose pad in a drop of 10 mM tetramisole and monitored by fluorescence and differential interference contrast microscopy using a Leica DMI6000B (FITC channel). 16. Brenner, S. Genetics 77, 71–94 (1974). 17. Neidhardt, F.C., Bloch, P.L. & Smith, D.F. J. Bacteriol. 119, 736–747 (1974). 18. Timmons, L., Court, D.L. & Fire, A. Gene 263, 103–112 (2001). 19. Babitzke, P., Granger, L., Olszewski, J. & Kushner, S.R. J. Bacteriol. 175, 229–239 (1993). 20. Wessel, D. & Flugge, U.I. Anal. Biochem. 138, 141–143 (1984). 21. Gobom, J., Nordhoff, E., Mirgorodskaya, E., Ekman, R. & Roepstorff, P. J. Mass Spectrom. 34, 105–116 (1999). 22. Liu, D.L., Beegle, L.W. & Kanik, I. Astrobiology 8, 229–241 (2008). 23. Cox, J. & Mann, M. Nat. Biotechnol. 26, 1367–1372 (2008).

nature methods