Innate immunity in C. elegans


Our general aim is to provide an integrated view of innate immune mechanisms in C. elegans. In the coming years we hope to:

Currently, most of our effort is going into the first goal, and only this part is described in any detail below.

1. Identifying genes that intervene in the response of C. elegans to a pathogen.

A better understanding of innate immune mechanisms in C. elegans requires cataloguing and then functionally characterising the genes that are involved in the response to infection. Almost all of our efforts are focused on dissecting the interaction between C. elegans and the nematode-specific fungus Drechmeria coniospora


Spores of D. coniospora adhering to the cuticle of an adult C. elegans. Electron-micrograph kindly provided by Jürgen Berger, MPI, Tübingen, from samples prepared by former post-doc Kwang-Zin Lee.

We are taking three complementary approaches to address this:

1.1. Transcriptome-based methods.

Using microarrays, we identified two families of structurally-related antimicrobial peptide (AMP) genes, members of the nlp (for neuro-peptide-like protein) and cnc gene families. Many of them are strongly up-regulated by fungal infection, and in some cases by physical injury or osmotic stress. More recently, thanks to a collaboration with the ModENCODE consortium, and using RNAseq, we have extended these studies to obtain a comprehensive and quantitative view of the transcriptional changes that accompany infection with D. coniospora. The power of the approach lies in the remarkable dynamic range of these measurements, over more than 6 orders of magnitude, as illustrated below.


Comparison of the number of normalised sequence reads for all transcripts (u) from control worms and worms 12 h after infection with the fungus D. coniospora. The expression of inducible nlp (n) and cnc (l) genes is highlighted. The expression of undetected transcripts is set as 0.00001 dcpm (average depth of coverage per base per million reads) for the sake of representation.

While the nlp and cnc genes are among the most induced genes, they are far from being the only genes to show a marked up-regulation. Additionally the expression of many genes is specifically down-regulated upon infection. It will be a challenge for the coming years to understand the function of the many genes that are regulated as part of the response to D. coniospora. We are fortunate enough to have RNAseq data from worms infected with D. coniospora at different time-points and, importantly, with other pathogens, including the Gram-negative bacteria Photorhabdus luminescens, Serratia marcescens, the Gram-positive bacterium Enterococcus faecalis and the fungus Harposporium spp. Unlike D. coniospora that infects C. elegans via its epidermis, these latter pathogens all infect C. elegans via the intestine. This data will allow us to concentrate on the elements that are specific to the response to D. coniospora.

1.2. Genetic screens.

Our previous genetic screens, based on EMS-induced random mutagenesis, proved to be a good means for identifying genes that are involved in the regulation of AMP gene expression [3, 10, 11]. The principal is simple, as we have generated transgenic strains that light up upon infection.


Control worms (left) and worms 12 h after infection with the fungus D. coniospora. These worms carry two reporter transgenes. One drives dsRed expression constitutively in the epidemis, the other, with the promoter of the AMP gene nlp-29, drives GFP expression only upon infection.

While generating and isolating mutants based on their failure to express GFP is relatively straight forward, in several cases the bottleneck has been cloning the underlying mutated gene. Although for some screens, mutagenesis with the Mos1 transposon has proved to be an efficient method, we have not had success with the method. With the advent of whole-genome resequencing, however, identifying EMS-induced mutations has become simpler and much more rapid. We have now conducted a saturating EMS screen for genes required for the normal induction of an nlp‑29 reporter gene. We isolated some 60 independent alleles. We are currently applying a battery of different tests to each mutant in order to categorise them phenotypically. For example, we test whether each mutant has a defect in nlp‑29 expression after wounding or exposure to high salt. Our preliminary results are extremely encouraging as it appears that we have isolated mutants that specifically affect the response to D. coniospora, as well as genes that act upstream of the protein kinase C gene tpa‑1, and hence which potentially represent genes involved in pathogen recognition.

Epistasis analyses will allow us to place the different mutants into known pathways, or to define uncharacterised parallel pathways. We expect to characterise the role of genes that are involved in the response both to infection and injury together with Andrew Chisholm (UCSD), with whom we have a long-standing collaboration, as he is particularly interested in wounding healing in C. elegans.

1.3. Whole-genome quantitative RNAi screens.

While the direct genetic approach is a powerful way to dissect the signalling pathways that control antimicrobial peptide (AMP) gene expression,it does have a number of drawbacks. For example, we may not be able to isolate alleles of genes that play an essential role during development. For this reason, together with our partners Union Biometrica and ModulBio we have invested considerable effort in establishing automated RNAi screens to look for novel signalling components. We currently have two RNAi feeding libraries that together allow the individual inactivation of ca. 18,500 genes (close to 90% of the predicted C. elegans genes). The screens are based on quantifying changes in the induction of an AMP reporter gene seen after D. coniospora infection. The Union Biometrica COPAS Biosort allows us to measure the fluorescence intensity of each worm in a population.


Quantification of GFP expression in control worms (blue) and worms 12 h after infection with the fungus D. coniospora (orange), using the COPAS Biosort. GFP expression is plotted against length for each worm.

Together with Thomas Richardson (WashU), we are developing statistical tools to allow robust hit identification. Our first screen is underway using the same reporter strain for nlp‑29 as we used in the genetic screen. Our current aim is to be able to perform a whole-genome screen in less than 3 months. If this is achieved, we would be able to extend such screens to other anti-microbial peptide genes and to other pathogens. The combined results promise to provide a near exhaustive list of the genes that influence the transcriptional response to infection.