Introduction

The soil nematode Caenorhabditis elegans has been used for over a decade to study host-pathogen interactions. Such studies provided detailed information on pathogen-specific innate immune responses, and the signal transduction pathways and transcription factors that activated them (reviewed in [1]). However, what initiates innate immune responses in the worm remains unknown.

In vertebrates, innate immune responses are initiated mainly by recognition of pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide, peptidoglycan, or microbially-modified nucleic acids; PAMPs are occasionally referred to as MAMPs–microbe-associated molecular patterns, to acknowledge their existence in non-pathogenic microbes. In addition, recognition of damage-associated molecular patterns (DAMPs), such as ATP and monsodium urate crystals, can enhance activation by PAMPs, but also may be sufficient for initiation of innate immune responses [2]. The pattern recognition receptors (PRRs) responsible for recognizing PAMPs or DAMPs include members of several protein families including the Toll/Toll-like receptors (TLRs), NOD-like receptors (NLRs) and RIG-I-like nucleotide recognition receptors (RLRs) [3]. Innate immune recognition is conserved through evolution, both in terms of its general strategy as well as the proteins involved: The first TLR, Toll, originally identified as essential for Drosophila's development, was subsequently shown to also have a crucial role in innate immunity [4], [5], NLRs take part in antimicrobial defenses both in animals and in plants [6] and a RIG-I homolog was recently shown to function in C. elegans anti-viral defenses [7]. Although C. elegans mounts specific responses to infections with different pathogenic bacteria, manifested in a robust gene induction [8], it is yet unknown how it discriminates between different bacteria. C. elegans has one Toll homolog gene, tol-1, which appears to be is largely dispensable for immune protection ([9], [10], but see [11])).; Additionally, FSHR-1, a heterotrimeric G protein with leucine-rich repeats (a motif shared among all vertebrate TLRs and NLRs), is necessary for immune protection, but is equally protective against Gram-negative and–positive pathogens, suggesting that it might not function in the proximal events of pathogen recognition. [12]. The inability to identify C. elegans PRRs based on orthology raises the possibility that C. elegans may use novel modes of pathogen recognition. Alternatively, C. elegans may respond to pathogen-specific damage caused in the course of infection. To discriminate between these two possibilities we used an infection model of C. elegans intestinal colonization by Pseudomonas aeruginosa [13].

Methods

Statistical analysis

Gene expression profiles obtained with microarrays were analyzed by a multi-class t-test using Significance Analysis of Microarrays (SAM; [15]), implemented as part of the TMEV software package. Based on T statistics the test retrieves genes with a T value above a cutoff score estimated to give the desired false discovery rate (selected to be 10%). This analysis was used to identify C. elegans genes differentially expressed in either one of the three analyzed groups: worms exposed to E. coli, worms exposed and colonized by P. aeruginosa PA14-GFP, or worms exposed to PA14-GFP, but not colonized. Since no difference was observed between expression profiles in worms collected manually or with the worm sorter, data from the three repeats for each of the three experimental groups were pooled. The resulting list of genes responding to PA14 contained 359 genes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. C. elegans immune responses to Pseudomonas aeruginosa are independent of colonization.

A) 2-day old adult C.elegans exposed to GFP-expressing P. aeruginosa for 18 hrs show variability in colonization, allowing isolation of colonized and non-colonized worms. B) Gene expression profiles of C. elegans fed with E. coli or with GFP-expressing P. aeruginosa (colonized and non-colonized) for 18 hours; separation was achieved either by picking under a fluorescent stereoscope (1 experiment; grey bar), or using the COPAS^TM Worm Sorter (two independent experiments ; black bar). Shown are genes responding to the pathogen, as identified with a multi-class t-test analysis (10% false discovery rate).

https://doi.org/10.1371/journal.pone.0035400.g001

Results

The PA14 strain of Pseudomonas aeruginosa, made to express GFP (PA14-GFP from here on), can be followed as it colonizes the worm intestine, leading to death within three days. This colonization requires live bacteria and depends on bacterial regulators of virulence [16]. Furthermore, once reaching a significantly visible level of colonization (e.g. following 18 hours of exposure), most worms remain colonized, even when transferred to E. coli (65% of N = 96). This suggests active interactions between the pathogen and its host, which enable the pathogen to persist in most cases. When a genetically-homogenous and age-synchronized population of worms is exposed to PA14-GFP, significant heterogeneity is seen in colonization by the pathogen (Fig. 1A). We reasoned that if colonization-associated damage elicited immune responses in C. elegans, then immune responses would be correlated to the degree of colonization. To test this, we chose the extreme case of comparing significantly colonized worms with worms that were not visibly colonized. We exposed wild-type worms to PA14-GFP under standard conditions (or to E. coli control) and following 18 hours, separated significantly colonized worms (green) from visibly non-colonized (dark). Subsequently, gene expression was examined using microarrays in both groups as well as in those exposed to the E. coli control. Genes responding to P. aeruginosa were identified using a multi-class t-test with a false discovery rate of 10%. This analysis identified gene classes previously reported as being induced by P. aeruginosa, including lysozymes, lectins [14], [17], several neuropeptide-like genes, and detoxification genes [17] (Fig. 1B and Table S1). The analyzed dataset is attached as Table S2. Importantly, infection responses were largely independent of the degree of colonization. Thus, although worm death from PA14 infection is correlated with colonization, the responses against it were not.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Pseudomonas secreted factors are not sufficient to induce immune responses, while conserved cell-associated factors are.

Gene expression measured by qRT-PCR, following exposure of young-adult C. elegans (at T₀) to E. coli, alone or with P. aeruginosa supernatant (A) or P. aeruginosa filtrate (B,C), or to the non-pathogenic P. mendocina (D).

https://doi.org/10.1371/journal.pone.0035400.g002

Several possible explanations may account for immune responses without visible colonization. The first that we wished to be able to rule out was that worms that are not visibly colonized were previously colonized, but managed to clear the infection. To test this possibility under similar conditions as those of the experiment we picked visibly colonized worms (N = 80) following a 12-hour exposure to PA14-GFP, and continuing the exposure to PA14-GFP on new plates, we examined whether any of these worms appeared dark by the time of collection (6 hours later); this turned out not to be the case, as all worms remained green. This does not mean that ‘dark’ worms are completely non-colonized, as they may be colonized with an undetectably small number of bacteria, but the similarity between their immune responses and those of fully-colonized worms implies that the immune response is not correlated with the degree of colonization.

A second possibility is that molecules secreted by P. aeruginosa cause damage to the worm, which independent of colonization can induce the immune response. This is particularly plausible since P. aeruginosa is known to secrete a wide array of small-molecule and proteinaceous exotoxins [18]. To address this possibility, we used quantitative (q)RT-PCR to follow the expression of F55G11.2, a gene with a yet uncharacterized function, which was previously identified as part of the earliest responses to P. aeruginosa [14]. A closer examination showed that F55G11.2 was strongly induced within two hours of exposure to the pathogen (Fig. 2B) and as early as one hour following exposure (not shown). However, F55G11.2 was not induced when worms were exposed to a conditioned solution from a P. aeruginosa 24-hour culture (the ‘secretome’), laced onto dead E. coli (Fig. 2A). Induction of F55G11.2 was similarly missing when worms were grown on plates conditioned with P. aeruginosa grown on a filter, which was removed prior to transfer of worms and replaced with dead E. coli serving as food (Fig. 2B). Similar lack of induction was observed for lys-2 and pgp-5, two additional infection response genes that respond to P. aeruginosa, but with a slower time course than F55G11.2 (Fig. 2C and results not shown). Thus, secreted factors are not sufficient to induce immune responses against P. aeruginosa.

A third possibility that we are unable to rule out is that volatile toxins released by P. aeruginosa cause damage to the worm, which induces immune responses. P. aeruginosa has a distinctive smell produced by a combination of volatile compounds. Of these, one is hydrogen cyanide, a potent toxin. However, cyanide production was not found to take part in PA14 pathogenicity in C. elegans (unlike the PA01 strain [19]) and without any additional known toxic volatile compounds released from P. aeruginosa, this is unlikely.

The fourth possible explanation for induction of immune responses prior to detectable colonization is that C. elegans can recognize molecules associated with the pathogen (i.e. PAMPs/MAMPs), with a sensitivity that allows it to respond to a small number of bacteria. The failure of the P. aeruginosa ‘secretome’ to induce immune responses, supported this possibility. To decouple structural features of Pseudomonas from its pathogenicity, we examined immune responses to Pseudomonas mendocina, a recently identified C. elegans commensal that shows no pathogenicity, both in terms of survival/lifespan as well as with regards to early symptoms of infection (i.e. muscle function and movement) (Fig. S1 and Montalvo-Katz, unpublished results). Exposure to intact P. mendocina lead to F55G11.2 gene induction, smaller than the response to P. aeruginosa, but reproducible (Fig. 2D). This is consistent with the hypothesis that C. elegans can recognize cell-associated moieties that are shared between P. aeruginosa and P. mendocina, and that such recognition plays a role in the initiation of early immune responses. Altogether, our data suggest that C. elegans immune responses against P. aeruginosa are initiated by PAMP recognition, or, since the recognized pattern is shared with the non-pathogenic P. mendocina, MAMP recognition.

Discussion

We found that C. elegans immune response occurs prior to any visible colonization. Death, and presumably damage due to P. aeruginosa infection, is correlated with the extent of colonization. That this correlation does not hold for immune responses suggests sensitive detection of molecular patterns, apparently cell-associated and furthermore, shared among pathogenic and non-pathogenic Pseudomonas species. Recognition of MAMPs by C. elegans does not exclude the possibility that it can also respond to additional types of stimuli. In fact, the smaller magnitude of the response to P. mendocina compared to P. aeruginosa may be indicative of multiple signals, some cell-associated, but others associated with pathogenesis (perhaps DAMPs), leading to a full-blown immune response. The results described here provide evidence for MAMP recognition in C. elegans, but the nature of these MAMPs remains to be identified. Furthermore, since no PRR orthologs have been identified to date in C. elegans, these results further encourage a search for C. elegans PRRs, as they may represent novel mechanisms of pathogen recognition.

Supporting Information

Acknowledgments

We wish to thank Gordon Lithgow and Matthew Gill of the Buck Institute for the use and help with the worm sorter, and Terry Machen from UC Berkeley for insightful discussions.