Antimicrobial effectors in the nematode Caenorhabditis elegans: an outgroup to the Arthropoda

Abstract

Nematodes and arthropods likely form the taxon Ecdysozoa. Information on antimicrobial effectors from the model nematode Caenorhabditis elegans may thus shed light on the evolutionary origin of these defences in arthropods. This nematode species possesses an extensive armory of putative antimicrobial effector proteins, such as lysozymes, caenopores (or saposin-like proteins), defensin-like peptides, caenacins and neuropeptide-like proteins, in addition to the production of reactive oxygen species and autophagy. As C. elegans is a bacterivore that lives in microbe-rich environments, some of its effector peptides and proteins likely function in both digestion of bacterial food and pathogen elimination. In this review, we provide an overview of C. elegans immune effector proteins and mechanisms. We summarize the experimental evidence of their antimicrobial function and involvement in the response to pathogen infection. We further evaluate the microbe-induced expression of effector genes using WormExp, a recently established database for C. elegans gene expression analysis. We emphasize the need for further analysis at the protein level to demonstrate an antimicrobial activity of these molecules both in vitro and in vivo.

This article is part of the themed issue ‘Evolutionary ecology of arthropod antimicrobial peptides'.

1. Brief overview of the Caenorhabditis elegans immune system

Arthropods and nematodes are likely members of related evolutionary lineages and form the clade of the moulting animals, the Ecdysozoa [1,2]. Owing to their common ancestry, comparison between these lineages may enhance our understanding of the evolutionary origin of specific traits in either of the taxa. Such comparisons have previously revealed important differences in the general organization of nematode and insect immune systems (see below). They complement broader evolutionary comparisons, for example between insects and more distantly related animal lineages such as molluscs or cnidarians [3]. In the current review, we focus on the model nematode Caenorhabditis elegans, which has become a central model organism for studying the genetics of invertebrate immunity. First work in this field was published in 1999 [4,5]. Since then, the nematode's response to a variety of pathogens has been assessed. These include Gram-negative bacteria (e.g. Pseudomonas aeruginosa, Serratia marcescens, Salmonella enterica), Gram-positive bacteria (e.g. Microbacterium nematophilum, Leucobacter sp., Enterococcus faecalis, Staphylococcus aureus and Bacillus thuringiensis), fungi (e.g. Drechmeria coniospora, Nematocida parisii, Candida albicans) and also a nodavirus (Orsay virus) [6]. Some of the used pathogens interact with C. elegans in nature, especially the microsporidian N. parisii [7], Orsay virus [8], P. aeruginosa [9], Leucobacter sp. [10], and possibly also M. nematophilum and B. thuringiensis. These are thus likely to elicit more specific defence responses. Most of the used pathogens infect the worm's intestine. Thus, the immune response mediated by intestinal epithelial cells has been examined in detail. Nevertheless, a few pathogens show different infection characteristics. The microsporidian N. parisii and Orsay virus are intracellular pathogens [7,8]. The fungus D. coniospora and the bacteria Leucobacter sp. and M. nematophilum infect via the cuticle; the latter two via the anal region and tail, and the former mainly via the mouth, the vulva and the anus [10–12]. These pathogens therefore target distinct tissues to those infecting the gut.

Analysis of the response of C. elegans to these various pathogens revealed the presence of a complex interconnected immune system the structure of which has been described in several reviews (e.g. [13–17]). Thus, for the purpose of the current review, we only indicate some of the main elements. Most curiously, the nematode lacks specialized immune cells and also homologues of central components of insect immune systems such as the transcriptional regulator NFκB or genes belonging to the prophenoloxidase cascade (reviewed in [18]). Caenorhabditis elegans immune defence relies on behavioural and physiological responses. Of these, behavioural responses represent an important, fine-tuned first line defence against a variety of bacterial pathogens (reviewed in [15,19]). The physiological immune systems is based on several conserved core signalling cascades, including three mitogen-activated protein kinase (MAPK) pathways (i.e. the p38, the extracellular signal regulated kinase (ERK) and the c-Jun N-terminal kinase (JNK) MAPK pathways), the insulin-like receptor (ILR) pathway, and a transforming growth factor β (TGF-β) cascade. In addition, the RNA interference (RNAi) machinery mediates defence against Orsay virus, which seems to be recognized by the retinoic acid inducible gene I (RIG-I) helicase DHR-1 [20]. Recognition of bacterial and fungal pathogens is less clear. A G protein-coupled receptor was implicated in the indirect recognition of the fungal pathogen D. coniospora via the perception of a so-called damage-associated molecular pattern (DAMP) [21]. Indirect pathogen detection also seems to be achieved through a cellular surveillance system, which activates pathogen defence responses when central cellular processes are disrupted [17,22]. While the upstream activators of C. elegans immune signalling cascades are less well understood, several downstream transcription factors that regulate the activation of immune effector gene expression following infection with different pathogens have been identified. For example, the basic helix–loop–helix (bHLH) transcription factor HLH-30 (TFEB in mammals) was demonstrated to regulate expression of effector genes in response to S. aureus [23], the GATA transcription factor ELT-2 (homologous to human GATA-4, -5 and -6) in response to P. aeruginosa [24], and the signal transducer and activator of transcription (STAT)-like transcription factor STA-2 in response to D. coniospora [25]. Caenorhabditis elegans immune effector mechanisms, which function in pathogen elimination, have been characterized in several cases, as explained in more detail in §2.

The aim of the current review is to provide an overview of antimicrobial immune effectors and effector mechanisms in the model nematode C. elegans. We focus on mechanisms, genes and gene families, for which there is evidence of an antimicrobial function in the worm, or which are at least implicated to be part of nematode immunity because of homology with characterized antimicrobial effectors from other organisms. We explore the functional diversity between and within effector types. We summarize the available evidence for their role as effectors, including the demonstration of an antimicrobial effect at the protein or peptide level, an immune phenotype upon experimental manipulation of the gene (knock-out (KO), RNAi knock-down, or gene overexpression), and also induced gene expression after pathogen exposure. For the latter, we take advantage of a recently established database, WormExp, which combines all available gene expression studies for C. elegans [26]. We conclude by highlighting promising avenues for future research on this topic.

2. Caenorhabditis elegans immune effectors

A variety of immune effectors and mechanisms have been described for C. elegans. These include the production of reactive oxygen species (ROS), the process of autophagy, and also the expression of putative antimicrobial peptides and proteolytic enzymes (see overview of the latter in table 1). The evidence for an immune role of these effector types will be discussed in detail below.

Table 1.

Overview of putative antimicrobial effector gene families in the nematode Caenorhabditis elegans.^a

gene family	abbr.	comment	no. of genes	antimicrobial examplesb
caenacins/neuropeptide-like proteins	cnc/nlp	short peptides, rich in glycine and aromatic acid	12	NLP-31
caenopores	spp	SAPLIPs (with saposin domain)	23	SPP-1, SPP-3, SPP-5, SPP-12
lysozymes	lys/ilys	2 lysozyme types (entamoeba- and invertebrate-types)	16	none tested
defensin-like AMPs	abf	sequence homology to insect and mammalian defensins	6	ABF-2
C-type lectin domain-containing proteins	clec	diverse family with C-type lectin domain	283	none found
fungal-induced peptides and fip-related peptides	fip/fipr	short peptides induced upon fungal exposure	36	none tested
thaumatin-like proteins	thn	homologies to anti-fungal thaumatins from plants	8	none tested

^aThe overview only includes the gene families with members for which an antimicrobial function was demonstrated experimentally or which show homologies to known antimicrobial effectors from other taxa or for which an antimicrobial function was proposed. In addition to these gene families, antimicrobial effector mechanisms covered by this review also include the production of ROS and autophagy.

^bExamples, for which an antimicrobial function was demonstrated in vitro at the protein/peptide level.

(a). Reactive oxygen species

ROS are chemically highly reactive molecules such as superoxide and hydrogen peroxide (H₂O₂) that are generated by the partial reduction of oxygen, mainly during mitochondrial oxidative metabolism, but also during cellular response to xenobiotics and cytokines [27]. Superoxide is generated by nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) enzymes, coenzyme Q10, and complexes I and III of the mitochondrial electron transport chain [28]. While ROS are produced as toxic by-products of normal metabolism, they also function as signalling molecules and represent an efficient, highly conserved effector mechanism to eliminate pathogens in animals and plants. For example, ROS are generated by NOX2 in human macrophages to directly kill ingested pathogens [29]. In Drosophila melanogaster, ROS are produced by the NOX enzyme dDuox in the gut to limit microbial proliferation [30]. In C. elegans, ROS can activate protective cellular mechanisms to promote longevity, pathogen defence responses and wound healing [31–34]. In addition to ROS functioning as signalling mediators, C. elegans responds to infection with E. faecalis by producing ROS via the NOX Duox1/BLI-3 in the intestine [35,36]. ROS production by intestinal cells represents a protective antimicrobial response: worms are more susceptible to E. faecalis infection when ROS production is impaired by reducing bli-3 expression via RNAi, or when ROS is eliminated by the addition of antioxidants to the medium [36]. The only other pathogen which is as yet known to induce ROS production in the C. elegans intestine is the yeast Saccharomyces cerevisiae. As for E. faecalis, a decrease in ROS production, in this case in a bli-3 mutant, leads to enhanced susceptibility to S. cerevisiae infection [37]. These studies produced experimental evidence for an important role of ROS as microbicidal effectors in C. elegans. It remains, however, unclear if the generation of ROS is a general defence mechanism to protect the worm's intestinal epithelium against pathogenic attack and how exactly the generation of ROS is activated upon pathogen exposure.

(b). Autophagy

Autophagy is not a classical effector mechanism in sensu stricto, such as the production of ROS or the expression of antimicrobial peptides (AMPs). It may still be involved in the elimination of pathogens, and thus it can provide an additional immune effector process. In the following, we will summarize the evidence for its contribution to C. elegans immune defence. Autophagy is a process during which intracellular material is sequestered within double-membrane vesicles (autophagosomes) and then targeted for lysosomal degradation. In this way, autophagy recycles intracellular components to produce energy during nutrient depletion, or removes potentially toxic material which is generated during, for example, oxidative stress, to prevent cellular damage. Autophagy is thus an important part of the protective surveillance machinery of the cell. In addition, autophagy plays a role in defence against pathogens by delivering intracellular microorganisms to lysosomes in both specialized immune cells and epithelial cells [38].

In C. elegans, autophagy is required for host defence against Salmonella enterica serovar Typhimurium [39,40], N. parisii and S. aureus [23], and mitophagy (the specific elimination of mitochondria by autophagy) for resistance to Pseudomonas aeruginosa [41]. Its contribution to defence against these four pathogens shows some variation, as explained in more detail in the following: (i) S. enterica, a facultative intracellular pathogen, invades the C. elegans intestinal epithelial cells only in worms in which the autophagy gene bec-1 (homologous to human BECN1) is silenced by RNAi. In the bec-1 RNAi animals S. enterica replicates intracellularly, leading to cytoplasmic destruction and premature death of the animal [39]. Similar to bec-1 RNAi, silencing the expression of two other autophagy genes, lgg-1 (encoding the homologue of mammalian MAP-LC3) and atg-7, resulted in animals being more susceptible to S. enterica infection. These data provide evidence that in wild-type animals S. enterica is efficiently degraded by the autophagy pathway and thus represents an important effector mechanism against an intracellular pathogen. (ii) Infection with the obligate intracellular pathogen N. parisii activates expression of several autophagy genes. Knock-down of lgg-1 and atg-18 (encoding the homolog of human WIPI1) by RNAi resulted in increased bacterial load, indicating that autophagy is required for controlling N. parisii infection. In addition, examination of transgenic worms carrying a translational GFP reporter for LGG-1 revealed that the autophagy machinery is targeted to N. parisii cells [42]. (iii) Infection with the extracellular pathogen S. aureus also activates the expression of several autophagy genes and this activation is dependent on the bHLH transcription factor HLH-30, which is required for the expression of 80% of the transcriptional C. elegans response to S. aureus, including AMP genes [23]. Staphylococcus aureus infection-induced autophagosome formation could be observed by analysis of the localization of GFP-tagged LGG-1 in the C. elegans intestine [23]. Worms in which the autophagy genes lgg-1, unc-51 (homologous to human ULK1) and vps-34 (homologous to human phosphoinositide 3-kinaseVPS34) are knocked down by RNAi are more susceptible to S. aureus-mediated killing. (iv) In the case of P. aeruginosa infection, iron chelation through pyoverdin, an iron-binding virulence factor of PA14, causes mitochondrial damage, which in turn triggers mitophagy. Several genes involved in autophagy (e.g. bec-1, lgg-1, mboa-7 (homologous to human MBOAT7)) and mitophagy (e.g. pink-1 (homologous to human PINK1), and pdr-1 (homologous to human PARK2)) are required for resistance to PA14 in a liquid-based killing assay [41].

Together, these data indicate that the autophagy machinery is required for C. elegans defence against infection with the facultative and obligate intracellular pathogens S. enterica and N. parisii, respectively, and the extracellular pathogens S. aureus and P. aeruginosa. The exact protective mechanism of autophagy and mitophagy still needs to be determined. While autophagy might act in the direct elimination of intracellular pathogens (as proposed by [39,42]), both autophagy and mitophagy might help the host in coping with the cellular damage caused by infection (as proposed by Visvikis et al. [23]), thus indirectly increasing tolerance against extracellular pathogens. The latter idea is supported by the fact that autophagy was shown to protect C. elegans against necrosis during P. aeruginosa infection [43].

(c). Caenacins and related peptides

The families of neuropeptide-like proteins (NLPs) and caenacins (CNCs) are antimicrobial peptides that are phylogenetically closely related. They were first discovered in a microarray analysis of the C. elegans transcriptional response to infection with the fungus D. coniospora [44]. Unlike most other C. elegans pathogens, which establish an infection in the intestine, D. coniospora spores infect the worm by attaching to its cuticle. After spore germination, the fungus penetrates the cuticle and epidermis to invade the worm. The NLPs and CNCs are small proteins (51–82 amino acids) with signal peptides at their N-terminus. The mature peptides are basic and rich in glycine and aromatic acid [44]. Two AMP genes that are similar to the C. elegans nlps and cncs can be found in D. melanogaster; otherwise the gene family seems to be restricted to nematodes. Six nlp (nlp-27, nlp-28, nlp-29, nlp-30, nlp-31 and nlp-34) and six cnc genes (cnc-1, cnc-2, cnc-3, cnc-4, cnc-5 and cnc-11) are localized in two separate clusters on chromosome V (named nlp-29 cluster and cnc-2 cluster, respectively [45,46]). All nlp genes of the nlp-29 cluster and five of the cnc genes in the cnc-2 cluster (all except cnc-3) are highly upregulated after D. coniospora infection [44–46]. At the time of discovery, these nlp genes had already been annotated as neuropeptide-like genes, as they show some sequence similarity with known C. elegans neuropeptides [47]. They form, however, a monophyletic group that is distinct from the other nlp genes encoding characterized neuropeptides of the worm's nervous system. Moreover, as described further below, there is evidence for NLP and CNC function as genuine AMPs. Analysis of the genomic distribution and evolutionary history of C. elegans nlp genes and their respective orthologues from C. briggsae and C. remanei provided evidence of recent expansion by gene duplications and of positive selection likely driving the nlp gene diversification [45]. Several nlp and cnc genes, such as nlp-29, nlp-30, nlp-31 and cnc-2 [44–46], were shown to be expressed in the epidermis of the worm where the fungus attacks. Interestingly, the expression of cnc-2 is only activated by fungal infection, while nlp-29 expression is also induced by sterile wounding, osmotic stress and in worms with epidermal defects, such as the dpy-9 and dpy-10 mutants [45,46,48]. There is experimental evidence from protein and genetic analyses for an important role of NLPs and CNCs in resistance to infection. The chemically synthesized NLP-31 protein inhibited fungal growth in infected worms [44] (figure 1a,b), although it may not be produced by nematodes themselves at very high concentrations. In addition, overexpression of the complete nlp-29 or cnc-2 cluster renders worms more resistant to D. coniospora infection [45,46].

Figure 1.

Antimicrobial effects of two C. elegans antimicrobial peptides. (a,b) Inhibitory effect of synthesized NLP-31 on hyphal growth of D. coniospora in vitro. (a) Worms are overgrown by D. coniospora hyphae 48 h post infection. (b) The addition of synthetic NLP-31 at 200 µM to infected worms completely inhibits fungal growth. These two images are from [44] and are shown here with permission from Jonathan Ewbank. (c) Knock-down of spp-5 (caenopores-5) leads to increased survival of E. coli in the C. elegans intestine. Homogenates of worms treated with spp-1, spp-3, spp-4, spp-5, spp-6 RNAi or an empty vector control, exposed to ampicillin-resistant E. coli for 30 min, and then thoroughly washed, were plated on LB agar plates to count the number of growing bacterial colonies. (d) SPP-5 exhibits pore-forming activity. Pore-forming activity was measured by fluorimetrically monitoring the dissipation of a valinomycin-induced diffusion potential in liposomes. The increase in fluorescence over time after addition (arrow) of 0.2 nmol SPP-5 (trace 1) or a control pore-forming peptide (trace 2), but not of the peptide solvent control (trace 3), reflects pore-forming activity. Panels (c,d) are from [49] and are shown here with permission from Thomas Roeder and Matthias Leippe.

The Ewbank lab developed an elegant AMP reporter gene-based approach to decipher the molecular signalling pathways that underlie regulation of nlp-29 and cnc-2 expression [46,48]. Using a pnlp-29::GFP reporter strain as a tool for candidate gene, forward genetic and RNAi screen approaches, it was possible to characterize a complex signalling pathway, from the upstream infection signal and its receptor to the downstream transcription factor that is required for AMP gene expression. In particular, nlp-29 expression is upregulated following sterile wounding and fungal infection via detection of the endogenous signal 4-hydroxyphenyllactic acid (HPLA) by the G protein-coupled receptor DCAR-1, which acts upstream of the Gα protein GPA-12, the protein kinase C TPA-1, the TIR-domain adaptor protein TIR-1, a central p38 MAPK cascade and the STAT-like transcription factor STA-2. While the p38 MAPK pathway is indispensable for activation of nlp-29 expression [48], a non-canonical TGF-β pathway is essential for the induction of cnc-2 expression [46].

Using WormExp [26], we looked for microorganism-dependent expression of nlp/cnc genes. For this analysis and also all further assessments below, we considered a total of 111 differentially expressed gene sets from 35 transcriptome studies, which examined the C. elegans response to bacteria, fungi and one virus at various exposure time periods (electronic supplementary material, table S1). For convenience, the main figures (figures 2 and 3) show the results for the most responsive gene set per pathogen (i.e. the gene set with the largest number of differentially expressed putative immune effector genes), while the full results across all gene sets are shown in the electronic supplementary material, figure S1. Most remarkably, expression of nlp and cnc genes is mainly induced by infection with the fungal pathogens D. coniospora and Harposporium sp. (in the case of cnc-4 and cnc-7 also with Candida albicans), while their expression is mainly downregulated and only upregulated in a few cases towards pathogenic or non-pathogenic bacteria (figure 2). This confirms previous observations [50] and supports the role of NLP and CNC peptides as inducible anti-fungal effectors. One exception is infection with the Gram-positive bacterium Xenorhabdus nematophila, which induces the expression of several cnc and nlp genes (figure 2). Interestingly, X. nematophila is one of the few bacterial pathogens that does not infect the C. elegans intestine, but adheres to the cuticle and forms biofilms on the head of the worm [51,52]. The tissue affected by the infection is thus the cuticle and the underlying epidermis, as in the case of D. coniospora. The fungus Harposporium sp., however, establishes an intestinal infection. The expression pattern of the nlp genes may thus suggest that both pathogen type (i.e. fungus versus bacterium) and site of infection (i.e. intestine versus epidermis) can influence effector gene expression in the worm [50].

Figure 2.

Expression of caenacins and related peptides in response to microbe exposure. Red and blue colours indicate up- and downregulated expression after microbe exposure, respectively. The panels show exemplary results for the considered 25 microbes, chosen from the total of 111 gene sets available for these taxa. The list of all gene sets is given in the electronic supplementary material, table S1 and the full results in the electronic supplementary material, figure S1. OR, Orsay virus; NP, Nematocida parisii; CA, Candida albicans; DC, Drechmeria coniospora; HA, Harposporium; YP, Yersinia pestis; XN, Xenorhabdus nematophila; YPS, Yersinia pseudotuberculosis; SE, Salmonella enterica; PA, Pseudomonas aeruginosa; MA, Microcystis aeruginosa; ECA, Erwinia carotovora; PL, Photorhabdus luminescens; SM, Serratia marcescens; VC, Vibrio cholerae; BT, Bacillus thuringiensis; SA, Staphylococcus aureus; EF, Enterococcus faecalis; MN, Microbacterium nematophilum; LR, Lactobacillus rhamnosus; ML, Micrococcus luteus; BM, Bacillus megaterium; PS, Pseudomonas sp.; EC, Escherichia coli.

Figure 3.

Expression pattern of caenopores, lysozyme and defensin-like AMP genes after microbe exposure. Red and blue colours indicate up- and downregulation of genes, respectively. The figure shows exemplary results of the most responsive gene sets. Full results are given in the electronic supplementary material, figure S1 and the list of considered gene sets in the electronic supplementary material, table S1. OR, Orsay virus; NP, Nematocida parisii; CA, Candida albicans; DC, Drechmeria coniospora; HA, Harposporium; YP, Yersinia pestis; XN, Xenorhabdus nematophila; YPS, Yersinia pseudotuberculosis; SE, Salmonella enterica; PA, Pseudomonas aeruginosa; MA, Microcystis aeruginosa; ECA, Erwinia carotovora; PL, Photorhabdus luminescens; SM, Serratia marcescens; VC, Vibrio cholerae; BT, Bacillus thuringiensis; SA, Staphylococcus aureus; EF, Enterococcus faecalis; MN, Microbacterium nematophilum; LR, Lactobacillus rhamnosus; ML, Micrococcus luteus; BM, Bacillus megaterium; PS, Pseudomonas sp.; EC, Escherichia coli.

(d). Caenopores

The caenopores belong to an ancient family of antimicrobial peptides with a saposin domain. They share similarities with the saposin-like proteins (SAPLIPs) such as the amoebapores first characterized for Entamoeba histolytica [53,54] and the mammalian NK lysin and granulysin [55,56]. In C. elegans, the caenopores or saposins (spps) form a gene family of currently 23 members. At the moment, the most comprehensive evidence for an antimicrobial function of C. elegans effector molecules is available for this group of peptides. The first caenopore characterized in more detail was SPP-1(T07C4.4) [57]. The recombinant SPP-1 protein was shown to have a helix bundle structure characteristic of saposin-like domains and to exhibit an antibacterial effect on E. coli [57]. A very detailed analysis of SPP-5 or caenopore-5 function at the peptide level was performed by Roeder, Leippe and co-workers, including the first and only inference of the structure of a C. elegans antimicrobial effector [49,58]. This caenopore is exclusively expressed in the intestine. Silencing spp-5 by RNAi leads to reduced worm fitness (e.g. reduced offspring production) and highly increased numbers of the food bacterium E. coli OP50 in the nematode intestine (figure 1c) [49]. This suggests that spp-5 is required for killing ingested bacteria, which in C. elegans can represent both food and pathogenic bacteria. Heterologously expressed SPP-5 was used to demonstrate the protein's ability to permeabilize the membranes of viable bacteria and to induce pores in phospholipid vesicles (figure 1d) [49]. Recently, SPP-5 was synthesized through native chemical ligation based on Boc solid-phase peptide synthesis, allowing further structural analysis of the antimicrobial function. This analysis revealed that a 35 residue long N-terminal fragment is sufficient for cell permeability activity [59]. A detailed peptide-level functional analysis was additionally performed for SPP-1, SPP-3 and SPP-12 [60,61]. SPP-1 is expressed in the intestine, SPP-3 in both intestines and a head neuron, while SPP-12 is exclusively expressed in two pharyngeal neurons. Importantly, all three peptides are able to permeabilize a variety of different microorganisms, they can form pores into phospholipid vesicles, and they may therefore contribute to the worm's interaction with microbes. Indeed, when spp-1 or spp-12 expression is knocked down by RNAi, C. elegans lifespan on E. coli is reduced [62]. Moreover, expression of spp-1 is induced by infection with S. enterica serovar Typhimurium [63]. Furthermore, S. enterica strains which lack certain virulence genes and are thus not able to persist in the C. elegans intestine, are capable of colonizing spp-1 RNAi-treated worms.

Despite the considerable knowledge, we have on SPP function on the peptide level, the regulation of spp gene expression is much less understood. spp-1 and spp-12 were identified as downstream targets of the forkhead box O (FOXO) homologue DAF-16 [62]. Pseudomonas aeruginosa is able to manipulate the C. elegans immune response by downregulation of immune effector genes, such as spp-1, and this repression is dependent on the DAF-2/DAF-16 (ILR/FOXO) pathway [64]. It thus seems that the ILR pathway induces expression of at least some spp genes. In addition, expression of several spp genes (spp-1, spp-2, spp-8, spp-15 and spp-17) is strongly upregulated in response to S. aureus and this upregulation requires the transcription factor HLH-30 [23].

Based on WormExp [26], we further explored differential expression of this gene family and find that all but one of the considered caenopore genes responds to microbe exposure (figure 3a; full results given in the electronic supplementary material, figure S1). The only exception refers to spp-19. The strongest gene activation response is shown against fungi, especially the intestinal fungal pathogen Harposporium sp., which induces expression of 10 genes. Strong downregulation of spp genes is observed after exposure to the Gram-negative bacteria X. nematophila, Photorhabdus luminescens, the Gram-positive B. thuringiensis, E. faecalis, and also the intracellular microsporidian fungus N. parisii. In general, spp genes show differential regulation by all of the considered types of microorganisms, including both pathogenic and also non-pathogenic taxa. The gene spp-12 appears most responsive to the microbes considered, followed by spp-1, spp- 2, spp-4, spp-8, spp-14, spp-20 and spp-23. None of the genes shows an identical pattern of differential expression. Taken together, this gene expression pattern suggests that spp genes contribute to both digestion and immune defence in a microbe-specific manner.

(e). Lysozymes

Lysozymes are known to contribute to both digestion and immunity in a wide variety of organisms, ranging from bacteria, phages, protists and plants to invertebrate and vertebrate animals [65]. The nematode C. elegans is unique in that its 16 lysozyme genes fall into two very distinct lysozyme types with an enormous level of sequence divergence within a single species. Ten of the genes are most closely related to the lysozymes found in protists (lys), while the remaining six are of the invertebrate type (ilys) [66]. Within these types, the lysozyme genes most likely arose through repeated duplication events across the phylogeny of the genus Caenorhabditis and show repeated episodes of positive selection [66]. To date, their exact function in C. elegans has not yet been studied at the protein level. Therefore, it is not known whether and how they interact with microbial molecules such as peptidoglycan and thus directly contribute to destruction of microbial membranes. Nevertheless, a likely role in worm immunity was inferred from functional analysis at the gene level, including analysis of knock-out mutants, gene silencing through RNAi, or gene expression analysis in transgenic worms. Most of the genes were shown to be expressed in the intestine, while lys-1 is additionally expressed in neurons, lys-7 in larval muscles and lys-8 in the pharynx (reviewed in [66]). Manipulation of gene activity indicated a role in the interaction with microbes for several lysozyme genes. The overexpression of lys-1 increased resistance against pathogenic S. marcescens [67]. Silencing of either lys-7 or ilys-2 led to higher susceptibility towards the pathogen M. nematophilum [68]. Gene knock-out of lys-2, lys-5 and lys-7 enhanced susceptibility to pathogenic B. thuringiensis, while overexpression of lys-5 and lys-7 but not lys-2 enhanced resistance relative to a non-pathogenic control [69]. Furthermore, ilys-2 and lys-5 RNAi-treated worms are more susceptible to S. aureus infection [23].

The expression of several lysozyme genes was shown to be controlled by C. elegans immune signalling pathways. For example, lys-7 and lys-8 are known targets of DAF-16 (FOXO) [62] and the TGF-β pathway [70,71]. Expression of lys-2 is regulated by the GATA transcription factor ELT-2 and the p38 MAPK PMK-1 [72]. The transcription factor HLH-30 influences expression of lys-2, ilys-3, ilys-4, lys-3, lys-5 and lys-10 following S. aureus infection [23]. Based on WormExp [26], we find that lysozyme gene expression is differentially regulated by almost all of the considered microbes (figure 3b; full results in the electronic supplementary material, figure S1). The only exceptions refer to Orsay virus and the pathogenic bacterium Yersinia pseudotuberculosis. Otherwise, lysozymes respond to both pathogenic and also non-pathogenic microorganisms in a highly taxon-specific pattern. The most responsive genes are lys-6 and ilys-3, followed by lys-4, lys-5, ilys-2 and ilys-5 (figure 3b). The gene lys-9 is the only gene which does not show any differential gene expression upon microbe exposure (figure 3b). As this gene is very different in sequence from all remaining C. elegans lysozymes and found at the end of a long branch within the lysozyme phylogeny [66], lys-9 may represent a pseudogene or at least show a function unrelated to that of the other lysozymes. Overall, the observed pattern of differential gene expression for the remaining lysozyme genes suggests that they contribute to both defence and digestion in a microbe-specific form.

(f). Defensin-like antimicrobial peptides

The C. elegans genome contains six genes with high similarity to the defensin-type antimicrobial peptides, well known from insects and vertebrates to contribute to immune defence [73]. These genes have been named antibacterial factor (abf) genes in the nematode. ABF-2 has been characterized at the peptide-level in two studies [74,75]. The heterologously expressed peptide showed high in vitro activity against a diversity of microbes, ranging from Gram-negative to Gram-positive bacteria and yeasts. The exact mode of action still requires further examination. abf-2 is mainly expressed in the pharynx, and, as a secretory peptide, it is likely present in the lumen of the pharynx and the gut. Its expression is induced by infection with S. enterica serovar Typhimurium, as monitored by qRT PCR, and abf-2 knock-down by RNAi resulted in an increased infection load [63]. The signalling pathways that control abf gene expression have not been fully deciphered. abf-2 expression is regulated by the transcription factor HLH-30 following S. aureus infection [23]. Upregulation of abf-1 and abf-2 by infection with Cryptococcus neoformans requires the scavenger receptor CED-1 (homologous to human SCARF1) and C03F11.3 (homologous to human CD36) [76]. Moreover, expression of abf-1 was demonstrated to be dependent on the gene npr-1 in the context of P. aeruginosa infection [77]. The transcriptome database WormExp [26] revealed that five out of the six abf genes respond to microbes and, if so, to only very few microbial taxa (figure 3c; electronic supplementary material, figure S1). In particular, abf-3 does not show any differential gene expression after exposure to the various microorganisms and abf-4 is downregulated by only P. luminescens. The remaining four genes can be activated by a total of seven pathogen strains (including for example S. aureus) and repressed by two other pathogens (E. faecalis and S. marcescens). Taken together, the abf genes appear to play a less prominent role in the inducible defence against pathogens or the inducible response to food microbes than the other above highlighted gene families (also see further discussion in §3). It is still possible that they are important in the constitutively expressed protection of the worm against pathogens and/or enhance the pathogen-specific response mediated by one of the other gene families with antimicrobial functions.

3. Future challenges: functional evidence for worm immune effectors and the involvement of the Caenorhabditis elegans microbiota

The model nematode C. elegans possesses a large repertoire of potential effector proteins and mechanisms to defend itself against pathogen attack. Our review assessed the involvement of putative effector gene families through their differential expression upon pathogen exposure, using the database WormExp [26]. To enhance comparability of the very diverse individual transcriptome datasets, this database includes only those genes in the gene sets that were found to be significantly differentially expressed, and it then only uses their presence/absence in the gene sets for all further analyses. This approach may have less sensitivity than analyses based on exact expression fold changes for all nematode genes. Yet, it also reduces the level of noise, often prevalent in transcriptome datasets, and thus it may help to identify the more robust overlaps in gene expression among conditions. A more sensitive analysis of pathogen-induced gene expression should be based on parallel assessment of various pathogens under exactly the same conditions, as previously performed for some pathogenic strains by the Ewbank lab [50]. Similarly, the C. elegans transcriptome studies usually use entire worm populations of whole animals, characterized at only one or two time points and usually one or two developmental stages. Although strong expression responses of putative effector genes should be visible in whole animal samples, subtle expression variations that are only shown by certain tissues or life stages may go undetected in such a crude approach. In the future, it would thus be of particular value to perform tissue-specific gene expression analyses across different life stages, taking into account the different modes of infection (intestinal infection, intracellular infections, infection of the cuticle; see §1). Based on our current approach, we can nevertheless conclude that there are at least some differences in microbe-responsiveness among the considered gene families, as discussed in more detail below.

It is additionally important to emphasize that differential expression after pathogen exposure does not suffice to prove an antimicrobial function of these genes. Such functional evidence needs to be obtained at the peptide or protein level. On the one hand, this may be achieved in vivo by silencing of the gene. A common approach for this relies on RNAi, which may, however, not always work with high efficiency, especially if genes are expressed in specific cell types such as neurons [78,79]. Another approach is to use knock-out mutants, which are already available for a large number of C. elegans genes [80] and which can be produced through application of CRISPR/Cas technology [81,82]. On the other hand, additional information on antimicrobial functions can be obtained through analysis of synthesized proteins and peptides. To date, such a protein/peptide-level analysis has only been performed for six C. elegans immune effectors: NLP-31, SPP-1, SPP-3, SPP-5, SPP-12 and ABF-2 [44,49,59–61,74]. To complicate matters further, evidence for an antimicrobial effect is still insufficient for demonstrating the protein's role in C. elegans pathogen defence. As this nematode feeds on a variety of microorganisms in the wild [83] and on E. coli bacteria in the laboratory, peptides and proteins which can damage or break down bacterial cell walls might also function in digestion. In C. elegans, this is most likely the case for those effectors which we here show to be inducible by non-pathogenic microbes (e.g. lys-4, ilys-3 and spp-18; figure 3), and also those which are constitutively expressed, as previously shown for spp-5 [49], abf-1, abf-2 and abf-3 [63,71,74]. As suggested for spp-5 [49], constitutively expressed effector genes might function in both defence and digestion, because they could enable the worm to access bacteria-derived nutrients and at the same time eliminate potential pathogens. Such a dual function of effectors is similarly discussed for Drosophila fruitflies, which also inhabit microbe-rich environments (see review by Broderick [84]).

Microbe responsive genes are especially found among the nlp/cnc, lysozyme and spp gene families (figures 2 and 3). These might thus play more general roles in immunity, as most clearly demonstrated thus far for the nlp and cnc genes and their particular contribution to anti-fungal defence (e.g. [21,44,45]). It is worth noting that none of the considered effector genes show a change in expression after infection with Orsay virus (except downregulation of nlp-28). This may suggest that the protective cellular response to viral infection mainly depends on the RNAi pathway in C. elegans [8], or that other as yet uncharacterized effector mechanisms contribute to virus elimination. Similarly, infection by the other known intracellular pathogen N. parisii only represses but does not activate expression of the considered effector genes. In this case, defence may also rely on as yet uncharacterized effector genes or different types of protective mechanisms (e.g. [16,42]). These observations also suggest that the immune effectors considered here mainly act as secreted proteins targeting extracellular microbes in the intestinal lumen, the pseudocoel or the cuticle. In comparison to the other C. elegans effector genes, the abf genes are inducible by only a few pathogens (figures 2 and 3). This may suggest that they contribute to defence and/or digestion in a form clearly distinct from the other microbe-inducible effectors. These abf genes may thus also have a less prominent role as inducible immune effectors than the homologous defensin-like genes in arthropods (see [3,85,86] in this issue).

Next to the immune effector genes which we have explored in the current review, additional gene families and/or processes may play an as yet undiscovered role in eliminating pathogens. Four gene families have been repeatedly highlighted in this context, although an antimicrobial function of these genes has not yet been demonstrated. One of these families comprises the C-type lectins, originally described as Ca²⁺-dependent (C-type) glycan binding (lectin) proteins, but now known as a diverse group of proteins which may bind to proteins, lipids and nucleic acids in both Ca²⁺-dependent and -independent ways. All C-type lectins share a highly conserved domain, the carbohydrate recognition (CRD) or C-type lectin-like domain (CTLD). The CTLD gene family is highly diverse in C. elegans, comprising more than 280 genes and being the seventh most abundant gene family in the worm (reviewed in [87]). Although vertebrate CTLD proteins are known to be involved in pathogen recognition, some mammalian CTLD proteins of the RegIII family possess antibacterial activity and function in pathogen elimination [88,89]. Similarly, an in vitro bactericidal activity was also described for several crustacean CTLD proteins [87]. In C. elegans, the majority of CTLD proteins contain a signal peptide and are thus predicted to be secreted. Moreover, the expression of the majority of C. elegans CTLD (clec) genes is induced by pathogen infection, showing a highly specific pattern of regulation, as recently evaluated by us with a similar approach to that used here [87]. In addition, several clec genes are required for resistance to infection as demonstrated in functional genetic analyses using mutant strains or RNAi (reviewed in [87]). The exact function of CTLD proteins in C. elegans immunity is still unclear. To date, only one study has assessed the function of these genes in the context of an immune response at the protein level, demonstrating that the two CTLD proteins CLEC-39 and CLEC-49 are able to bind to a bacterial pathogen, in this case S. marcescens [90]. Although clec-39 and clec-49 mutant worms are more susceptible to S. marcescens infection, the proteins CLEC-39 and CLEC-49 do not have an inhibitory effect on S. marcescens growth in vitro. These results suggest that the genes either function in recognition or do not mediate pathogen elimination alone but perhaps in collaboration with other effectors. A more detailed discussion of possible immune functions of these proteins has recently been published elsewhere [87].

Two additional groups of putative effectors are the fungal-induced peptides (fip) and fip-related peptides (fipr). These genes are induced in expression upon infection with fungal pathogens such as D. coniospora or Harposporium sp. [50]. Caenorhabditis elegans possesses seven fip genes and 29 fipr genes which generally vary in their expression upon pathogen exposure (electronic supplementary material, figures S1 and S2). As they encode proteins that are less than 100 amino acids in size and are predicted to have signal peptides, it is likely that fip and fipr genes encode AMPs. Experimental evidence for their contribution to C. elegans anti-fungal defence or antimicrobial activity is so far missing [91].

Yet another group are the thaumatin-like proteins. These are small proteins around 200 amino acids in length which act as anti-fungal defence proteins in plants. Their anti-fungal properties are likely based on beta-1,3-glucanase activity, alpha-amylase inhibiting properties (reviewed in [92]) and/or membrane-permeabilizing activity [93,94]. There are eight homologues of thaumatin encoding (thn) genes in C. elegans. While it is not known if C. elegans thaumatins exhibit anti-fungal activity, three of the eight thn genes (thn-1, thn-2 and thn-3) are both induced and repressed by infection with fungal as well as bacterial pathogens, whereas their expression upon exposure to non-pathogenic microorganisms remains unchanged (electronic supplementary material, figures S1 and S3). A possible immune function of thn genes was further suggested by altered resistance to P. aeruginosa infection after RNAi knock-down of thn-1 and thn-2 [24,64]. Moreover, thn-2 expression depends on the ILR pathway and can also be directly manipulated by pathogenic P. aeruginosa [24,64]. It remains to be determined whether C. elegans thaumatins indeed act as bona fide antimicrobial effectors. Moreover, it is possible that in future, still other C. elegans gene families may be discovered to possess antimicrobial activity.

In conclusion, the nematode C. elegans possesses a variety of putative antimicrobial peptides and additional antimicrobial mechanisms, some of which have been characterized in depth at both genetic and protein level, especially certain caenopores. Nevertheless, the exact immune function of most of these putative effectors still needs further clarification. In future, it would be essential to demonstrate at the protein level in vitro that these effectors are able to interact with either non-pathogenic and/or pathogenic microbes and break up bacterial cell membranes or inhibit bacterial growth or viability in some other way. Moreover, it is similarly important that the function of the genes is studied in vivo in the worm, using the available tools for C. elegans gene manipulation at the cellular level in combination with microscopic dissection of the resulting infection pattern produced by various microbes.

In this context, it would also be of particular interest to assess how different immune effectors interact with each other to affect microbe proliferation. Effector molecules are usually studied in isolation, although several effector genes are simultaneously expressed in response to pathogen infection and some effector proteins are known to exert their antimicrobial activity in synergy with other immune effectors to enhance their potency ([95]; see also [85,86,96]). Interestingly, the mixture of expressed effector genes seems to be highly specific (figures 2 and 3). In fact, we do not find identical patterns of co-expressed genes in response to the various microbes, possibly suggesting specifically fine-tuned immune defences. Such fine-tuning is likely orchestrated by interconnected signalling processes which integrate information from various stimuli, including microbial molecules and also the cellular consequences of pathogen infection (i.e. cellular damage; reviewed in [17]). However, the exact regulation of C. elegans effector gene expression is largely unexplored. The main exception refers to nlp-29, for which an endogenous danger signal triggers gene expression [46] and which is regulated through a complex signalling network (e.g. [21,25,45,46]). Further analysis of the exact regulation of antimicrobial effector genes would be of great value for understanding their role in defence. The screening approach developed by the Ewbank lab, based on an AMP gene reporter strain, may be of particular promise in this context.

Last but not least, it is conceivable that antimicrobial effectors in C. elegans are also used to control the worm's microbiome, in analogy to what is known for example for weevils (see review by Masson et al. [97]) and proposed for fruitflies (see review by Broderick [84]). At the same time, it is possible that members of the worm's microbial associates themselves produce protective antimicrobial factors, thus increasing the nematode's arsenal of effector molecules. This nematode seems to contain a rich microbial flora [83,98,99], yet the exact species composition and functions of the microbiome of natural C. elegans isolates have not yet been published. The production of antimicrobial compounds, for example bacteriocins or specific anti-fungal proteins, is known for microbiota members of various host taxa, ranging for example from Hydra polyps [100] to humans (e.g. [101]). For C. elegans, co-cultivation with bacteria such as Lactobacillus acidophilus or Pseudomonas mendocina enhanced resistance against pathogens [102,103]. In these two cases, it is unclear whether the tested bacteria produce antimicrobial factors themselves or stimulate their production in the worm. Similarly, we lack any information on possible antimicrobial functions of bacteria associated with natural C. elegans isolates. Future characterization of the worm's microbiome should thus specifically assess to what extent individual bacterial strains may enhance nematode immune defence.

Data accessibility

All used transcriptome datasets are available from WormExp, http://wormexp.zoologie.uni-kiel.de/.

Authors' contributions

All authors jointly wrote the manuscript.

Competing interests

We have no competing interests.

Funding

We are grateful for support from the German Science Foundation to K.D. (DI 1687/1 and project A1 of CRC 1182) and H.S. (SCHU 1415/8; SCHU 1415/9, and project A1 of CRC 1182). W.Y. is additionally supported by the International Max-Planck Research School (IMPRS) for Evolutionary Biology at the University of Kiel.