Stable-isotope labeling with amino acids in nematodes.

We describe an approach for accurate quantitation of global protein dynamics in Caenorhabditis elegans. We adapted stable-isotope labeling with amino acids in cell culture (SILAC) for nematodes by feeding worms a heavy lysine- and heavy arginine-labeled Escherichia coli strain and report a genetic solution to elminate the problem of arginine-to-proline conversion. Combining our approach with quantitative proteomics methods, we characterized the heat-shock response in worms.

Results

The SILAC mass spectrometry approach facilitates the accurate and reproducible quantitation of large numbers of proteins[1]. In contrast, 15N-labeling which has been applied previously for proteomic analysis in C. elegans[2], has drawbacks for data analysis, including the unpredictable mass shifts induced by total 15N labeling[3]. SILAC is typically used with tissue culture cells, and SILAC-based methods have not been reported previously for C. elegans. While SILAC has been used in a small number of multicellular organisms, problems associated with the conversion of isotope-labeled arginine to proline and other amino acids have complicated such studies[4]. Arginine-to-proline conversion reduces the signal from all heavy SILAC labeled peptides containing proline. The increased number of MS/MS target peptides resulting from arginine-to-proline conversion reduces the overall sensitivity of global MS/MS analysis. These additional peptide peaks also increase the likelihood of overlapping isotopic envelopes reducing peak discrimination.

We aimed to adapt SILAC methodology for C. elegans and to eliminate the arginine-to-proline conversion problem. Initially, we employed an arginine and lysine auxotrophic version of E. coli BL21 (DE3), but C. elegans did not survive well on this strain (data not shown). We therefore modified E. coli HT115, a strain commonly used for the RNAi-feeding procedure[5], by generating an arginine and lysine auxotroph termed SLE1 (Supplementary Fig. 1a). Analysis of the egg laying rate and embryonic development/survival of C. elegans grown on the E. coli SLE1 strain indicated no difference (Supplementary Fig. 1b) compared to E. coli op50, which is generally used as a food source in C. elegans studies. For C. elegans labeling the SLE1 strain was grown in minimal media containing 15N4-13C6-arginine (heavy arginine) and 15N2-13C6-lysine (heavy lysine), and plated onto agarose petri dishes. C. elegans eggs were placed onto a lawn of SLE1 and the incorporation of heavy arginine and lysine into protein in the F1 generation was analysed by mass spectrometry (MS).

We observed approximately 93 % heavy isotope incorporation as illustrated for a representative peptide (Supplementary Fig. 2a). However, arginine-to-proline conversion led to ~20 % of the heavy peptide signal being diverted to a larger m/z signal. This diversion of signal intensity could be seen more clearly when equal portions of heavy and light protein were mixed (Fig. 1). Analysis of each peak shown in Supplementary Fig. 2a by tandem mass spectrometry (MS/MS) confirmed that the heavier peak contained 15N-13C5-proline (heavy proline) and no other labeled amino acids besides arginine (Supplementary Fig. 2b). The conversion of heavy arginine to heavy proline occurs in the nematode and not in SLE1 because the E. coli did not generate heavy proline containing peptides (Supplementary Table 1). Arginine-to-proline conversion occurs through the urea cycle[6] (Supplementary Fig. 3, see below).

Figure 1

Elimination of arginine-to-proline conversion using orn-1 RNAi-feeding facilitates stable isotope labeling with amino acids. Arginine-to-proline conversion, which occurs in wild type worms (upper panel) is abolished upon orn-1 depletion (lower panel). The peak intensity of the depicted ‘light’ model peptide approximately equals the combined intensities of the corresponding heavy peptide and the peptide that contains heavy proline (n = 3). Peptides are highlighted as ‘Light’, ‘Heavy’ and ‘Heavy+ Heavy proline’. Lysates from light or heavy labeled C. elegans were mixed in equal proportions, fractionated by denaturing SEC and fraction 12 was analysed by trypsin digestion and LC-MS/MS. A representative proline containing peptide derived from EF-1α was examined.

To eliminate the arginine-to-proline conversion problem we employed the RNAi-feeding procedure, targeting the ornithine transaminase enzyme orn-1 (C16A3.10). This enzyme converts ornithine to L-glutamate-5-semialdehyde (Supplementary Fig. 3) and is required for arginine-to-proline conversion in S. pombe[6]. RNAi-feeding of C. elegans was performed according to the schemes in Fig. 2 and Supplementary Fig. 4a, and the extent of orn-1 transcript depletion was evaluated by qPCR (Supplementary Fig. 4b). Worms grown with either control or orn-1 RNAi knock-down showed similar egg laying rate and embryonic development, which indicated that orn-1 RNAi did not have a major effect on viability (Supplementary Fig. 4c). orn-1 RNAi knock-down worms were labeled with both heavy arginine and heavy lysine, or their respective light amino acids, for one generation, to determine the effect of RNAi feeding on arginine-to-proline conversion. MS analysis showed again approximately 93 % heavy arginine and lysine isotope incorporation (Fig. 1 and Supplementary Fig. 2c), and that greater than 98% of total proline was 14N-12C5-proline (light proline), indicative of a near complete elimination of arginine-to-proline conversion (Fig. 1 and Supplementary Fig. 2a). In addition, when equal portions of untreated heavy and light protein were mixed and analysed by MS, more than 95 % of the proteins had Log2 ratios that were approximately 0 (Supplementary Table 1). These data also indicate that fold-changes greater than +/− 50 % would differentiate proteins whose expression level has been altered from non-affected proteins.

Figure 2

Flowchart for SILAC in nematodes, taking the analysis of the heatshock response as an example. Light medium (blue) is M9 minimal medium and arginine (R0) and lysine (K0). Heavy medium (red) is M9 minimal medium and 15N4-13C6-arginine (R10) and 15N2-13C6-lysine (K8). The RNAi procedure is described in more detail in Supplementary Fig. 4.

A large number of temperature sensitive mutants have been generated in C. elegans, allowing for a transient disruption of often essential proteins. One problem associated with this experimental procedure is background perturbations generated by the heatshock response. We therefore used the SILAC in nematodes approach to examine the extreme end of the heatshock response by shifting worms to 30 °C, according to the scheme shown in Fig. 2. The C. elegans proteome was fractionated using a detergent-free denaturing size exclusion chromatography method (Online Methods). This fractionation method proved effective as shown in Fig. 2.

LC-MS/MS analysis of each fraction yielded a dataset of ~19,000 peptides corresponding to > 1,400 proteins, each identified with at least two peptides (Supplementary Table 1 and Supplementary Table 2). Four small heat shock proteins were amongst the 9 proteins up-regulated more than four-fold and had Maxquant Significance B values[1] less than 0.05 (Fig. 3, Supplementary Table 1, and Supplementary Table 2). These data validate our technique and provide a global overview describing changes in protein abundance upon heatshock treatment. Strikingly, three cathepsin-like aspartic acid proteases of both lysozomal and non-lysozomal origin (ASP-1, ASP-2, and ASP-6), were down-regulated more than three-fold after heatshock and had Maxquant Significance B values[1] less than 0.05 (Fig. 3, Supplementary Table 1, and Supplementary Table 2). ASP-1 is a lysozomal aspartic acid protease of the cathepsin-D family that is mainly expressed in intestinal cells and has been observed previously to be down-regulated in response to heatshock[7]. Little is known about the role of ASP-1, ASP-2, and ASP-6 in nematode biology and future studies will be needed to reveal their role in the heat stress response.

Figure 3

Analysis the C. elegans heatshock response using SILAC in nematodes. The abundance of ~1,400 proteins is indicated on the y-axis using a log2 scale. The abundance of each protein indicated by the position of the dot on the y-axis was determined by summing up all individual light and heavy peptide intensities detected for each protein. The relative fold decrease or increase upon heat shock treatment is indicated on the x-axis. Heatshock treated worms were grown on heavy-labeled SLE1 bacteria, while untreated worms were grown on light bacteria. Proteins highlighted in solid black are up-regulated heatshock proteins, those highlighted with white fill and black outlines are down-regulated aspartic acid proteases (n = 2).

We have carried out further studies using sub-cellular fractionation of unlabelled worms. This allowed approximately three times the number of proteins to be identified by MS, as well as provided valuable information as to the sub-cellular distribution of proteins in untreated cells (Supplementary Table 1). Subcellular fractionation, in conjunction with the SILAC-based method presented here, will allow future studies to examine the dynamic re-localisation of the C. elegans proteome in response to various conditions, as recently demonstrated in human cells[8].

Using the SILAC in nematodes approach we have also characterised changes in the C. elegans proteome in response to orn-1 RNAi (Supplementary Table 1). These experiments compared the proteome of worms labeled with light amino acids and treated with a control RNAi (targeting GFP), with worms labeled with heavy amino acids and orn-1 targeted RNAi (n = 1). These data showed the reduction of total ORN-1 protein by ~60 % (Supplementary Table 1). The reduced knock-down could be the result of incomplete RNAi in C. elegans neurons, which has been observed previously[9]. However, the residual ORN-1 activity in RNAi resistant cells appears to be minimal, because arginine-to-proline conversion was mostly abolished (see above). orn-1 RNAi also generated some changes in protein expression compared with control RNAi treated cells, as expected. However, as orn-1 RNAi should be used in both heavy and light labelled worms in the SILAC in nematodes method, these differential effects will be negated. One caveat may be that analysis of the urea cycle enzymes could be altered by orn-1 RNAi, and this should be taken into account for any experiments specifically targeting that pathway. The generation of a null mutant for the orn-1 gene by the C. elegans Knock-out Consortium has been requested by our labortatory, and will facilitate future studies without the need for RNAi-mediated orn-1 knock-down.

Our SILAC in nematodes methodology opens up many opportunities for research in C. elegans using quantitative MS-based proteomic strategies. By eliminating arginine-to-proline conversion SILAC experiments can be performed more efficiently. It will also allow the application of methods previously used in tissue culture cells, for example the relative quantitation of protein-protein interactions and the elimination of contaminants in pull-down experiments, in worms[10]. The generation of the E. coli SLE1 strain for simultaneous stable isotope labeling with amino acids and RNAi feeding could also be used for double RNAi experiments where possible. For example, orn-1 can be knocked-down to eliminate arginine-to-proline conversion, and another C. elegans gene could be targeted to determine the effects on the proteome. This will be especially useful for the analysis of genes where null mutations are embryonic lethal but where RNAi leaves sufficient protein product for survival. The SILAC in nematodes technique will also help to determine global proteome changes during development, aging and stress responses.

METHODS

Methods are available in the online version of the paper.