Effector and regulator: Diverse functions of C. elegans C-type lectin-like domain proteins

In C. elegans, 283 clec genes encode a highly diverse family of C-type lectin-like domain (CTLD) proteins. Since vertebrate CTLD proteins have characterized functions in defense responses against pathogens and since expression of C. elegans clec genes is pathogen-dependent, it is generally assumed that clec genes function in C. elegans immune defenses. However, little is known about the relative contribution and exact function of CLEC proteins in C. elegans immunity. Here, we focused on the C. elegans clec gene clec-4, whose expression is highly upregulated by pathogen infection, and its paralogs clec-41 and clec-42. We found that, while mutation of clec-4 resulted in enhanced resistance to the Gram-positive pathogen Bacillus thuringiensis MYBt18247 (Bt247), inactivation of clec-41 and clec-42 by RNAi enhanced susceptibility to Bt247. Further analyses revealed that enhanced resistance of clec-4 mutants to Bt247 was due to an increase in feeding cessation on the pathogen and consequently a decrease in pathogen load. Moreover, clec-4 mutants exhibited feeding deficits also on non-pathogenic bacteria that were in part reflected in the clec-4 gene expression profile, which overlapped with gene sets affected by starvation or mutation in nutrient sensing pathways. However, loss of CLEC-4 function only mildly affected life-history traits such as fertility, indicating that clec-4 mutants are not subjected to dietary restriction. While CLEC-4 function appears to be associated with the regulation of feeding behavior, we show that CLEC-41 and CLEC-42 proteins likely function as bona fide immune effector proteins that have bacterial binding and antimicrobial capacities. Together, our results exemplify functional diversification within clec gene paralogs.


Introduction
Vertebrate C-type lectin-like domain (CTLD) proteins play an essential role in pathogen recognition and subsequent activation of the immune response to fungal and bacterial infection [1]. CTLD proteins are characterized by a conserved carbohydrate-recognition domain (CRD), which may bind sugar in a calcium-dependent (C-type) manner. But not all of the proteins carrying a CRD bind glycans or require calcium for binding, which is why the more general term CTLD proteins was introduced [2]. In the vertebrate immune system CTLD proteins mainly act as antimicrobial effector proteins, classical pattern recognition receptors (PRRs), binding ligands derived from fungi, bacteria or viruses, and as dead and cancerous cell sensors enhancing tumor killing activities of natural killer cells (reviewed in [3]). In invertebrates CTLD proteins also seem to play a role in immune defenses, but their exact functions are much less clear [4].
C. elegans has a highly diversified CTLD encoding gene (clec) repertoire (reviewed in [4]). 288 CTLD proteins are encoded in the nematode's genome with half of them carrying an additional CTLD or other domains such as CUB (Complement C1r/C1s Uegf Bmp1), CW (conserved cysteine and tryptophan residues), or VWA (Von Willebrand factor type A) [4,5]. Several C. elegans transcriptome analyses revealed that the majority of clec genes is highly upregulated upon pathogen exposure (e.g. [6][7][8]) and this, in a highly specific pattern [5]. By reverse genetic analyses some clec genes were shown to be required for defense against infection with bacterial pathogens, e.g. clec-17, clec-60, and clec-86 for Microbacterium nematophilum [9], clec-70 for Staphylococcus aureus [10], clec-65 for the pathogenic Escherichia coli strain LF82 [11], and clec-174 for Vibrio cholerae [12]. It is however unclear, what exact role these CLEC proteins play in immune defenses. There are only four studies providing further evidence, revealing that functions of C. elegans CLEC proteins in response to pathogens are likely to be manifold: First, CLEC-39 and CLEC-49 are required for C. elegans resistance to Serratia marcescens infection. As both proteins bind to S. marcescens in vitro, but do not exhibit antimicrobial activity, CLEC-39 and CLE-49 were suggested to act as classical PRRs mediating pathogen recognition [13]. Second, CLEC-1 is secreted by body muscle cells and was identified as regulator of protein homeostasis in the extracellular space in response to exposure with the B. thuringiensis-derived pore forming toxin Cry5B [14]. Protein homeostasis in the extracellular space was suggested to be important for sustaining a systemic immune response by keeping secreted immune effectors functional. CLEC-1 was shown to be highly effective in preventing aggregation of LYS-7 [14], which contributes to C. elegans resistance to B. thuringiensis [6]. Third, the CTLD containing protein IRG-7 affects C. elegans pathogen resistance by activating the p38 MAPK-ATF-7 immune defense pathway [15]. Moreover, IRG-7 was identified as mediator of longevity, being a component of the reproductive longevity pathway thus revealing complex effects of a CLEC protein on C. elegans physiology [15]. Finally, the CTLD containing protein C54G4.4 was implicated in the regulation of C. elegans behavioral immune defenses. C54G4.4 mutants were more resistant to pathogen infection and exhibited enhanced pathogen avoidance responses [16].
Although experimental evidence for a role of C. elegans clec genes in immune responses is scarce, clec genes are generally assumed to be important innate immune genes based on their pathogen-dependent expression patterns. The aim of this study was to explore to what extent pathogen-responsive clec genes function in C. elegans immunity. We focused on the C. elegans clec genes clec-4, whose expression is highly upregulated by infection with a broad array of pathogens, and its paralogs clec-41 and clec-42. We explored clec-4, clec-41, and clec-42 expression, redundancy, and their function on the gene and protein level. Moreover, we did an explorative transcriptome analysis of the clec-4(ok2050) mutant. We found that while clec-41 and clec-42 seem to encode bona fide immunity proteins that function in C. elegans resistance to B. thuringiensis MYBt18247 (Bt247) infection, clec-4 indirectly affects pathogen resistance by regulating feeding behavior. Together, we demonstrate for the first time an antimicrobial function of C. elegans CLEC proteins and identified clec-4 as novel regulator of feeding behavior.

Bacterial strains and culture conditions
E. coli OP50 was obtained from the CGC. The pathogenic Pseudomonas aeruginosa strain PA14 (provided by Dennis Kim) was grown first on LB plates and then in LB broth at 37˚C overnight prior to inoculation of assay plates. We used Bacillus thuringiensis strains MYBt18247 and MYBt18679 (in the following Bt247 and Bt679; our lab strains) and Bt407 (provided by Christina Nielsen-LeRoux, INRA, France) as a non-pathogenic control. Sporetoxin mixtures were generated following previous protocols [21][22][23] and frozen at -20˚C in aliquots with a spore concentration ranging from 3 � 10 9 to 8 � 10 9 particles/ml, depending on the culture, for Bt247 and 1.1 � 10 10 particles/ml for Bt407. Stocks were thawed and then immediately applied in infection assays. Serratia rubidaea MYb239 was co-isolated with C. elegans from a compost heap in Kiel, Germany ( [24], Carola Petersen and Hinrich Schulenburg). Serratia marcescens Db11 (provided by Jonathan Ewbank), S. aureus SA113 (provided by Andreas Peschel), and Rhodococcus erythropolis MYb53, which is part of the natural microbiota of C. elegans [25], were used for in vitro bacterial binding assays.

Generation of transgenic C. elegans strains
The gene reporter and rescue constructs were generated by PCR fusion as previously described [26]. The promotor regions of the clec genes (1.0-1.6 kb upstream of start codon) were amplified from genomic DNA by PCR with primer A and primer B of which the latter contains an overlap to the sequence of the gfp vector (S1B Table). The gfp or mCherry coding sequence plus the 3'-UTR of unc-54 was amplified from the Fire vector pPD95.75 with primer C (5'agcttgcatgcctgcaggtcgact-3') and D (5'-aagggcccgtacggccgactagtagg-3'). The transgenic constructs were finally synthesized using PCR fusion with primer A � (S1B Table) and D � (5'ggaaacagttatgtttggtatattggg-3') and directly injected at a concentration of 10 ng/μl. The plasmids carrying ttx-3p::RFP (40 ng/μl), which is expressed in the AIY interneuron pair of successfully transformed animals, or myo-2p::RFP (25 ng/μl), expressed in pharyngeal muscle, were used as co-injection markers in lines carrying the clec-4 promotor construct or in lines carrying the clec-41 or clec-43 promotor construct, respectively. The fusion constructs with the clec-41 and clec-42 promotors were injected into the unc-119(ed3) background together with plasmid pPK605 (gift from Patricia Kuwabara, Addgene plasmid # 38148) which served as rescue for the unc-119 phenotype (S1B Table). At least three lines were generated per reporter construct and microscopically evaluated. As the lines injected with the same fluorescent reporter construct showed similar expression patterns we focused on the ones listed in S1B Table for further analyses. For microscopy worms were mounted on slides with a 2% agarose patch and immobilized with sodium azide. All pictures were taken with the confocal microscope LSM 700 or the Axio Observer Z.1 by Zeiss (Carl Zeiss AG, Jena, Germany).
Transgenic strains MY1121 and MY1122 for rescuing clec-4(ya1) were generated by injecting 50 ng/μl of a fusion construct with 0.5 kb intestinal promotor of mtl-2 [27] and the coding region of clec-4, or the complete coding region of clec-4 including 1.4 kb of the upstream promotor into the unc-119(ed3) background (S1C Table). The rescue for the unc-119 mutation and myo-2p::RFP served as co-injection markers as described above.

Survival assays and lifespan analysis
PA14 requires enriched NGM (0.35% instead of 0.25% peptone) for efficient killing of worms. Agar plates seeded with a mixture of PA14 (OD 600 1) and E. coli OP50 (OD 600 5) in PBS at a concentration of 1:3 were incubated for 24 h at 37˚C, followed by 24 h at 25˚C. The PA14 survival assay was conducted at 25˚C. For Bt infection peptone-free medium (PFM) agar plates were inoculated with a mixture of E. coli OP50 at OD 600 5 in PBS and Bt at different concentrations. The survival plates were kept at 20˚C overnight. As different preparations of Bt247 spore-toxin mixtures may vary in pathogenicity, the exact killing levels of Bt247 may vary between experimental runs when different spore-toxin preparations are used. However, effects of knockout or knockdown of central immune system components on worm survival can still be consistently identified in comparison to simultaneously characterized controls. NGM plates for S. rubidaea infection were inoculated with an overnight culture adjusted to OD 600 5 in PBS and left at 20˚C overnight. Plates seeded solely with E. coli OP50 or non-pathogenic Bt407 were used as controls in Bt survival assays, and E. coli OP50 plates in survival assays with PA14 and S. rubidaea. In lifespan experiments NGM plates were seeded with an overnight culture of E. coli OP50.
30 synchronized L4 larvae were picked onto each plate and worm survival was scored once after 24 h (Bt) or regularly across time until all worms on the pathogen were dead (PA14 and S. rubidaea). Worms were considered to be dead if they did not respond to light touch.

Behavioral assays
For scoring the avoidance behavior on Bt 9 cm PFM plates were prepared as described above for the survival assays, but with a 30 μl spot of the bacterial mixture in the middle. For Bt exposure mild pathogen concentrations were chosen in order to challenge, but not to kill the worms. Ten synchronized L4 hermaphrodites were picked onto each bacterial spot in the middle of the plate at time point 0 hours post infection (hpi) and the worms residing on that spot were scored every second hour. The leaving index was calculated as (total number of wormsworms on bacterial spot) / total number of worms. Dead individuals were excluded from the total number of worms per plate.
For scoring the pumping rate on Bt PFM plates were prepared as described above for the survival assays. For Bt exposure a mild pathogen concentration was chosen. Ten synchronized L4 hermaphrodites were pipetted onto each plate to ensure that five worms are on the spot by the time the pumping rate was scored. After 6 and 24 h of exposure to Bt or OP50, the pumping of five worms per plate was counted for a period of 20 s. A total of four plates for each treatment was scored.

Bacterial load assay
In the bacterial load assay L4 larvae were placed on PFM plates seeded with either Bt247, Bt407, or OP50 OD 600 5 for 24 h. All of the alive worms were picked into M9 + 0.025% Triton-X, and gravity washed with 1 ml of M9 + 0.025% Triton-X for 5 times, before worms were paralyzed with 10 mM tetramizole. Worms were then bleached with a soft bleach protocol, following [29]. Worms were then washed twice in PBS + 0.025% Triton-X, before the exact number of worms was determined (~10-25). After the final washing step, worms were homogenized in the GenoGrinder 2000 by adding sterile zirconia beads (1 mm diameter, 3 min, 3000 strokes/ min). Importantly, 100 μl of washing buffer were transferred to a separate tube before grinding to be treated as the supernatant control. Worm homogenate and supernatant control were serially diluted and plated onto LB plates. After two overnights at 25˚C, colonies were counted at the appropriate dilutions and colony forming units (CFUs) per worm were calculated. The soft bleaching treatment sterilized the worm surface sufficiently as the supernatant control was almost always free of viable bacteria.

Brood size assay
For the brood size assay, single worms were picked onto NGM plates seeded with OP50, transferred daily, and the hatched offspring was scored until the end of the reproduction period. The assay was conducted at 20˚C.
The bacterial binding assay was done following published protocols [30]. In detail, four types of Gram-negative (E. coli OP50, P. aeruginosa PA14, S. marcescens Db11, S. rubidaea MYb239) and 4 types of Gram-positive bacteria (S. aureus SA113, B. thuringensis Bt247 and MYBt18679, and R. erythropolis MYb53) were grown at 37˚C or 28˚C overnight in LB to midlogarithmic phase, pelleted, washed, and resuspended in TBS buffer with CaCl 2 (50 mM Tris, 150 mM NaCl, 2 mM CaCl 2 , pH 7.5) at OD 600 2. 100 μl bacterial solution was incubated with 6 μg recombinant protein at gentle rotation for 1 h at room temperature. The bacteria were washed three times with 1 ml TBS-CaCl 2 and eluted with 100 μl of 2% SDS. The whole lysates plus 5x loading buffer were heated at 95˚C for 5 min, equally loaded onto a 12% SDS-PAGE, analyzed by Coomassie staining, and then transferred to a PVDF transfer membranes (BIO-RAD, Cat. #1704272). After blocking in 5% non-fat milk at room temperature for 1 h the membranes were incubated with Mouse-anti-His mAb (BIO-RAD, Cat. #MCA1396GA) overnight at 4˚C, followed by HRP-linked secondary antibody (advansta, Cat. #R-05071-500). The signals were detected using the chemiluminescence phototype-HRP kit (BIO-RAD, Cat. #1705060S) according to the manufacturer's instructions.

Antimicrobial activity assay
Assessing the antimicrobial activity of CLEC proteins in vitro was adapted from a previously published broth dilution method (Protocol (E), [31]).
CLEC proteins were serially diluted in LB in a polypropylene 96-well plate leaving 50 μl protein dilution per well. An E. coli OP50 culture in logarithmic phase was diluted in LB to a final concentration of 100-1,000 CFUs/well, 50 μl were added to each well, and the plates were incubated at 37˚C overnight. The control wells contained either only serial dilutions of the CLEC's native buffer and OP50 (buffer control), only LB (sterility control), or OP50 in LB (growth control). Melittin (Sigma-Alrich, Cat. M2272), an antimicrobial peptide of the honeybee venom, served as positive control. The minimal inhibitory concentration (MIC) was defined as the lowest protein concentration that inhibited visibly bacterial growth, i.e. no bacterial pellet or a diffuse bacterial pellet without defined border at the bottom of the well. The wells with the resulting MICs were plated onto LB plates and incubated overnight in order to determine the CLEC's bactericidal or bacteriostatic activity.
For testing the synergistic effect of two CLEC proteins one CLEC protein was mixed with either the second CLEC protein or with the native CLEC buffer in a polypropylene 96-well plate and serially diluted in the CLEC buffer, leaving 50 μl protein dilution per well. A Bt407 or Bt247 culture in logarithmic phase was diluted in LB to a final concentration of approx. 2,000 CFUs/well, 50 μl were added to each well, and the plates were incubated at 28˚C overnight.

clec-4(ok2050) transcriptome analysis by RNA-Seq
Transcriptomic responses were assessed 6 and 12 h after exposure to the respective bacteria. At the respective time points, worms were washed off the assay plates with PBS containing 0.3% Tween20, and subsequently centrifuged. The worm pellet was resuspended in 800 μl TRIzol (Life Technologies) reagent and worms were broken up prior to RNA extraction by treating the worm suspension five times with a freeze-and-thaw cycle using liquid nitrogen and a thermo block at 45˚C. RNA was extracted using a NucleoSpin miRNA extraction kit (Macherey-Nagel), treated with DNAse, and stored at -80˚C. RNA libraries were prepared for sequencing using standard Illumina protocols. Libraries were sequenced on an Illumina HiSeq 2000 sequencing machine with paired-end strategy at read length of 100 nucleotides. The raw data is available from the GEO database [34,35] under GSE110913.
RNA-Seq reads were firstly trimmed for adaptor sequence, masked for low quality sequence via Trimmomatic [36] and then mapped to the C. elegans genome (WormBase web site, http:// www.wormbase.org, release WS235) by STAR 2.5.3a [37] under default setting. Transcription abundance (read counts per gene) was extracted via HTSeq [38]. Differential expression analysis was performed by aFold from ABSSeq [39]. We only considered genes with a significant change between conditions (clec-4 mutant vs. N2; adjusted p-value < 0.01). The log 2 transformed fold-changes were taken as input for k-means cluster analysis using cluster 3.0 [40]. A heatmap was generated by TreeView version 1.1.4r3 [41].

Gene ontology and gene set enrichment analysis
Gene ontology (GO) analysis was performed using DAVID with a cutoff of FDR < 0.05 [42]. Gene set enrichment analysis was performed by WormExp [33]. A gene set with FDR < 0.05 was considered significant.

Statistical analyses
Statistical analyses were done with RStudio (Version 1.0.136), graphs created with its package ggplot2 (Version 2.2.1) and edited with Inkscape (Version 0.91). General statistical tests and appropriate corrections for multiple testing were applied and can be found in S2 Table.

Results & discussion
Expression of clec-4 and its paralog clec-41 is highly upregulated upon exposure to various pathogenic bacteria C. elegans clec genes have repeatedly been suggested to be involved in pathogen defense because they are always among the genes that are highly upregulated by pathogen infection. The repertoire of induced clec genes differs from pathogen to pathogen, suggesting a highly specific regulation (reviewed in [4]). However, the expression of a few clec genes is activated by infection with several different pathogens, indicating a more general role of these genes in C. elegans defense responses. We performed a transcriptome meta-analysis of pathogen-dependent clec gene expression using WormExp [33] and identified clec-4 as one of the clec genes, which is highly upregulated upon infection with a broad array of pathogens, including Gramnegative and Gram-positive bacteria, as well as the fungal intestinal pathogen Harposporium spec. (Fig 1A). The pathogen-dependent upregulation of clec-4 was previously confirmed for the Gram-negative P. aeruginosa strain PA14 by qRT-PCR [43] and on the protein level for the Gram-positive B. thuringiensis strain Bt247 by a quantitative proteome analysis [44,45].

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins expression of clec-4 in five out of seven pathogen data sets (Fig 1A), in case of Bt infection at both the transcriptome and proteome level [44,45]. In addition to clec-41, we decided to include clec-42, the closest paralog to clec-41, in our analysis. clec-42 is the only other gene among the 39 paralogs that encodes a protein with the same domain architecture as CLEC-41, consisting of two CTL and two CUB domains (Fig 1B). We therefore focused our functional analysis on these three genes.

clec-4, clec-41, and clec-42 are expressed in the intestine and co-expressed in int1
To elucidate in vivo expression patterns of clec-4, clec-41, and clec-42 we generated transgenic strains carrying the transcriptional reporter construct clec-4p::GFP (Fig 1C and 1D) and strains carrying the clec-41p::GFP reporter alone (Fig 1F) or together with clec-42p::mCherry ( Fig 1E). Under standard laboratory culture conditions, clec-4p::GFP was constitutively expressed in the intestine of worms throughout all developmental stages (S1A Fig). However, expression of clec-4p::GFP seemed to be decreased in L4s and adults, in which a strong GFP signal could only be observed in the first intestinal ring (int1) and the posterior intestine ( Fig  1C). Localized intestinal expression of infection-induced genes was observed previously: For example, expression of the caenopore gene spp-7 is stronger in the posterior intestine than in the anterior [50]. Mallo et al reported expression of the lysozyme gene lys-1 throughout the intestine, but also observed lys-1::GFP in vesicles in a single posterior intestinal cell [51]. In addition to its intestinal expression, clec-4p::GFP is constitutively expressed in amphid neurons from the first larval to the adult stage (Fig 1C and 1D). Expression of the clec-4p::GFP reporter gene was not inducible after an infection with pathogenic Bt or PA14, which is in contrast to the strong upregulation shown in the gene expression analyses. It was, however, consistently observed across experiments and pathogens (Fig 1A), including both transcriptomic and also independently performed proteomic analyses on the same Bt247 pathogen [44,45,52] (Fig 1A). This discrepancy indicates that the extrachromosomal array does not faithfully recapitulate endogenous expression of clec-4. One possible reason might be that the reporter gene construct does not contain all relevant regulatory sequences.
Similar to clec-4, clec-41p::GFP is weakly but constitutively expressed throughout the intestine of worms at all life stages under standard laboratory culture conditions. In contrast, clec-42p::mCherry is constitutively expressed only in int1 and only at the four larval stages, not in adults ( Fig 1E). Interestingly, a transcriptional GFP reporter of another clec-4 paralog, clec-43, was also exclusively expressed in int1 (S1B Fig). That some clec genes are specifically expressed in int1 is an intriguing observation. The intestine of the worm is a tube consisting of 20 epithelial cells, of which the most anterior ring (int1) directly behind the pharynx is comprised of four cells, whereas the subsequent rings (int2-int9) are comprised of two cells each (https:// wormatlas.org/). It was already noted by John Sulston and colleagues that "The anterior ring of four cells (int1) is specialized in having shorter microvilli than the rest of the intestine" [53]. Moreover, the pH in the lumen of int1 differs from the pH in the remaining intestine and the int1-associated part of the gut was suggested to act as mediator between the basic pharynx and the acidic intestine [54]. Our striking observation of exclusive expression of several clec genes in int1 provide further evidence of a specialization of int1. The secretion of CLEC proteins and other potential immune effectors specifically by int1 might create a distinct microenvironment that is important for host-microbe interactions at the 'entry gate' of the intestine.
Although clec-41p::GFP is already constitutively expressed, we observed an even stronger expression after infection with Bt247 (Fig 1F), which confirms the infection-dependent upregulation seen in the expression data sets. The expression of clec-42p::mCherry was not induced by Bt247 infection, which also is in line with the expression data. Together, clec-4 and clec-41 are both expressed in the C. elegans intestine and their expression co-localizes with clec-42 expression in the anterior intestinal ring (int1).

clec-4 mutants exhibit enhanced survival on B. thuringiensis
clec-4 expression is highly upregulated in response to infection with several different pathogens, including P. aeruginosa PA14, S. marcescens Db10, and B. thuringiensis Bt247 (Fig 1A). As a first step toward understanding the potential functional role of clec-4 in C. elegans immune defense responses, we tested if clec-4 is required for resistance to infection with the Gram-negative pathogens PA14 and S. rubidaea MYb239, and the Gram-positive pathogen Bt247, using the clec-4(ok2050) deletion mutant. The clec-4(ok2050) mutant contains a deletion of 1610 bp, which removes most of the coding region (S2G Fig), comprising part of one CTLD at the N-terminus and the complete remaining CTLD and CUB domain, resulting in the complete absence of any mRNA product and thus represents a null deletion allele (Figs 1B and  S2A). We found that clec-4(ok2050) mutant animals survived as well as wildtype worms on PA14 and S. rubidaea (Fig 2A and 2B), being consistent with previous results on the unaffected susceptibility of the clec-4(ok2050) mutant to infection with S. marcescens [13]. These results indicate that clec-4 does not function in C. elegans defense responses to these pathogens. On the Gram-positive pathogens Bt247 however, clec-4(ok2050) mutant animals were unexpectedly more resistant than wildtype worms (Fig 2C).
To confirm that the knock-out of clec-4 is the causative factor of the resistance phenotype we used CRISPR/Cas9 to generate the clec-4(ya1) allele that contains a large in-frame deletion, which removes a 2071 bp long fragment of the clec-4 ORF (S2G Fig), i.e. the complete intermediate CTLD, the majority of the N-terminal CTLD, and part of the CUB domain, resulting in a very short mRNA product (Figs 1B and S2B). clec-4(ya1) animals were also more resistant to Bt247 infection (Fig 2C). It is important to note that the survival phenotypes of both clec-4 (ok2050) and clec-4(ya1) mutant animals were variable across several experimental runs (S3A and S3B Fig). Taking all data into account, the resistance phenotype was, however, the most prevalent clec-4 mutant phenotype, observed in 7 out of 12 experimental runs for clec-4 (ok2050) and 3 out of 4 runs for clec-4(ya1). Also, we generated two strains, in which we reintroduced copies of the clec-4 sequence expressed from its endogenous promoter and from an intestinal promoter, respectively, into the clec-4(ya1) knock-out mutant background and Fig 2. clec-4(ok2050) and clec-4(ya1) increase C. elegans resistance to Bt247 infection. Survival of clec-4(ok2050) and N2 wildtype worms on (A) P. aeruginosa PA14 and (B) S. rubidaea MYb239 over time. (A, B) Alive, dead, and missing worms were scored until all individuals on the pathogen were dead. No significant differences in survival between clec-4 mutants and N2 wildtype worms as determined by Kaplan-Meier analysis [55] and log-rank test [56]. Horizontal ticks represent censored data (missing worms), n = 5, (A) data are representative of two independent experiments, (B) data represent the mean of three independent experiments. (C) Difference in survival on serial dilutions of Bt247 and a dilution of 1:10 of the non-pathogenic Bt407 control 24 hpi (hours post infection) between clec-4(ok2050), clec-4(ya1), and N2. Means ± standard deviation (SD) of n = 4 are shown, data are representative of at least four independent experiments. �� p < 0.01, and ��� p < 0.001, according to a generalized linear model (GLM) [57]. Also see S3A and S3B Fig, and S2A

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins obtained partial rescue of the resistance phenotype observed in clec-4(ya1) mutant animals (S3C Fig). As this particular Bt247 infection assay produces highly reproducible results for gene mutants with a central role in resistance (see e.g., our previous results with the assay for RNAi-silenced elt-2 or jun-1 knock-out mutants [52]), we conclude that mutations in clec-4 contribute partially albeit not essentially to C. elegans resistance to Bt247 infection.

Functional loss of CLEC-4 increases feeding cessation on pathogenic Bt247 and consequently decreases intestinal pathogen load
Despite the upregulation of clec-4 gene expression [44] and the higher abundance of CLEC-4 protein [45] upon Bt247 infection, clec-4 deficiency unexpectedly led to increased resistance (Fig 2C). It is possible that pathogen resistance results from defense behaviors such as increased pathogen avoidance or decreased pharyngeal pumping (i.e. pathogen up-take) as previously shown for the CTLD containing gene C54G4.4 mutant [16]. Also, we observed clec-4 expression in C. elegans amphid neurons (Fig 1D), which are chemosensory and thermosensory neurons with openings to the exterior that play a role in detecting microbial cues and in regulating pathogen avoidance behavior [58]. To understand if clec-4 functions in behavioral defense, we assessed avoidance and pumping activity of the clec-4 mutants on Bt247. We found that clec-4 mutant animals avoided the pathogenic Bt247 strain as much as wildtype worms ( Fig 3A). However, clec-4 mutants exhibited a prolonged decrease in pharyngeal pumping on Bt247: As expected N2 wildtype animals decreased their pumping activity on pathogenic Bt247 compared to the non-pathogenic control 6 hpi. Pumping activity of clec-4 mutants  . clec-4(ok2050) and clec-4(ya1) decrease C. elegans feeding leading to a reduced bacterial load. (A) Avoidance behavior of clec-4(ok2050) and N2 wildtype worms on Bt247. Worms were exposed to a mild concentration (1:400) of non-pathogenic Bt407 and pathogenic Bt247. The avoidance index is defined as (initial total number of worms-worms on bacterial spot) / initial total number of worms. Means ± SD of n = 5, no significant differences between clec-4(ok2050) and N2 were determined by a generalized linear mixed model (GLMM) [59] per bacterial treatment, p-values corrected for multiple comparisons with Bonferroni [60]. (B) Pumping rate of clec-4(ok2050), clec-4(ya1), and N2 wildtype worms on a 1:100 dilution of Bt247 and Bt407, n = 4, data are representative of three independent experiments. � p < 0.05, �� p < 0.01, ��� p < 0.001, according to GLMM per bacterial treatment and per worm strain, Bonferroni corrected [60]. (C, D) Bacterial load on (C) a Bt dilution of 1:100 or (D) OP50 measured in colony forming units (CFUs) per worm. Shown are pooled data, (C) n = 18 or (D) n = 16 combined from three independent experiments. � p < 0.05, ��� p < 0.001, as determined by Wilcoxon rank sum test, Bonferroni corrected [60]. Also see S2B and S3D-S3F Tables. https://doi.org/10.1371/journal.ppat.1009454.g003

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins was even significantly lower than that of wildtype animals and while the wildtype N2 strain resumed feeding at 24 hpi, the clec-4 mutants remained at a low feeding rate (Fig 3B).
We then examined whether clec-4 deficiency affects pathogen accumulation in the intestine. To this end, we quantified intestinal pathogen load by counting colony forming units (CFUs) of live bacterial cells recovered from the intestines of wildtype and mutant animals. To avoid any contamination with bacterial cells that stick to the surface of the worm, we applied a mild bleaching protocol that efficiently removed all bacteria from the outside (see [29] and materials and methods). We observed a substantial reduction in pathogen load in clec-4(ok2050) and clec-4(ya1) animals compared to wildtype worms (Fig 3C).
These results indicate that the clec-4 mutant animals' enhanced survival on Bt247 is due to a reduction in pumping and consequently a reduction in pathogen load. As feeding cessation on pathogens is a behavioral response and as behavior of animals can be dramatically affected by small changes or variations in culture or assay conditions [61], this may explain the variation we observed in clec(ok2050) and clec(ya1) resistance to Bt247 (S3A and S3B Fig). clec-4 mutants were as resistant as wildtype worms to infection with S. rubidaea MYb239 (Fig 2B) and PA14 (Fig 2A). Interestingly, C. elegans does not exhibit feeding cessation on all pathogenic bacteria: C. elegans feeds normally on the pathogens P. aeruginosa PA01 and Salmonella Typhimurium MST1 [62]. On PA14, worms exhibit pathogen avoidance behavior, but do not decrease pharyngeal pumping [63]. Moreover, eat-1 mutants, which also exhibit a reduced rate of pharyngeal pumping [64] are indistinguishable from wildtype for resistance to PA14 [63] and pharyngeal pumping does not affect PA14 accumulation in the intestine [65]. Consequently, killing by PA14 appears to depend on the establishment and proliferation of bacteria within the gut and to be independent of the rate at which bacteria enter the gut [63]. Thus, prolonged feeding cessation as exhibited by the clec-4 mutants only leads to an increase in resistance to infection when the rate of pathogen intake determines the rate of pathogen accumulation in the gut, as is the case for Bt.
If clec-4 regulates feeding behavior, why is its expression upregulated during pathogen infection (Fig 1A)? Most transcriptome analyses of C. elegans pathogen responses determine pathogen-induced genes through a comparison with the expression pattern shown on E. coli OP50. However, in contrast to E. coli, pathogenic bacteria colonize the intestine of young worms and live bacteria accumulate in the gut. We thus hypothesized that clec-4 expression is not only induced by pathogens, but by live, colonizing bacteria in general. Members of the beneficial natural C. elegans microbiota are known to colonize the C. elegans intestine and we thus looked for clec-4 expression in two transcriptome data sets on the C. elegans response to its natural microbiota. clec-4 expression is indeed also upregulated on these non-pathogenic bacteria [66]. This observation points to a modulation of clec-4 expression depending on the quantity or nutritional composition of bacterial food in the gut and is in agreement with a role of clec-4 in regulating feeding also on non-pathogenic bacteria (see below).

Functional loss of CLEC-4 affects feeding
During the pumping assays we observed that clec-4 depletion also led to a pumping/feeding phenotype on Bt407 that we used as non-pathogenic control and on the food bacterium E. coli OP50 (Fig 3B). However, the pumping phenotype was more subtle and we detected a statistically significant difference in pumping between wildtype and clec-4 mutants in two out of three runs at only one of two time points on Bt407 and in one of three runs on E. coli (S2B Table). We then measured bacterial load by counting CFUs and found that bacterial load was significantly reduced in clec-4 mutants for both Bt407 (Fig 3C) and E. coli (Fig 3D). However, as E. coli OP50 bacteria are efficiently broken up by the grinder in the C. elegans pharynx [64] and live E. coli cells thus do usually not accumulate in the gut of young worms (day 1 adults in our experiments), the CFU counts were very low (ranging from 0 to 19 for clec-4 mutant animals and from 0 to 75 for wildtype animals). Also, clec-4 mutants showed normal development and morphology. It is thus difficult to draw any conclusions on differences in the nutritional status of the animals from these experiments.

Gene expression profile links clec-4 to nutrient sensing
To explore the possibility that the restrictions in normal feeding observed for the clec-4 mutants result in a transcriptional response similar to the response to dietary restriction or starvation, we performed gene expression profiling of the clec-4(ok2050) mutant and wildtype animals exposed to E. coli OP50 and non-pathogenic Bt407. We confirmed that the Bt407 and E. coli OP50 treatment showed only little variation to each other [52], validating that Bt407 is non-pathogenic. 435 genes were differentially regulated in clec-4(ok2050) mutant worms on Bt407 when compared to wildtype worms, all, except 11 of them, downregulated (Fig 4A, S5A  Table). We used WormExp [49] for an enrichment analysis to assess the overlap between the clec-4-dependent gene set and other, previously published C. elegans gene sets and found that one common denominator within the most significantly enriched gene sets was nutrient

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins sensing. For example, targets of the p38 MAPK pathway ("down in pmk-1 mutant") and insulin/insulin-like growth factor-1 (IGF-1) signaling ("up by DAF-16", "up in daf-2 mutant") were enriched in the genes downregulated in clec-4 mutants (Fig 4B and S5B Table). The p38 MAPK pathway is a crucial C. elegans innate immunity pathway [67] and has more recently been proposed to act as an immunometabolic pathway that senses bacterial and nutrient signals [68]. daf-2, which encodes the C. elegans insulin/IGF-1 receptor ortholog, and oga-1, which encodes a O-GlcNAc cycling enzyme that is a component of carbohydrate metabolism/ hexosamine signaling, play central roles in nutrient sensing [69]. Furthermore, genes regulated by starvation, glucose, rapamycin, the dietary and environmental sensor DAF-12, the C. elegans AMP/energy signaling AMP-activated protein kinase alpha subunit AAK-2, and the homolog of the cyclic AMP-response element binding protein (CREB) CRH-1, were enriched in genes downregulated in clec-4 mutants. While the CREB gene crh-1 has not as yet been directly linked to dietary restriction and/or starvation, activation of the energy-sensing AMPK AAK-2 and inhibition of the nutrient-sensing target of rapamycin (TOR) pathway by rapamycin are perturbations mimicking dietary restriction. Together, the results of the enrichment analysis indicate an involvement of clec-4 in energy metabolism and/or nutrient sensing.

CLEC-4 loss of function has subtle effects on reproduction
Dietary restriction affects life history traits in many organisms, including C. elegans [78]. The C. elegans dietary restriction mutant eat-2 has a reduced food intake due to a mutational defect in pharyngeal pumping [64,79]. eat-2 mutation has been shown to increase lifespan [79], extend the self-fertile reproductive period [80,81], and reduce lifetime fertility [82]. To assess if the restrictions in food intake similarly affect life history traits in clec-4 mutants, we measured lifespan, fertility, and reproductive timing in clec-4(ok2050) and clec-4(ya1) mutants. The data on lifespan were inconclusive, as the curves of the clec-4 mutants crossed that of wildtype worms in two experimental runs and showed no difference in another experimental run (Figs 4C, S4A and S4B). However, the brood size of clec-4 mutants was smaller than that of wildtype worms (Fig 4D), albeit the difference was statistically significant only during early reproduction (S4C Fig) and there was no statistically significant difference in overall lifetime brood size (Fig 4D). In comparison to wildtype worms, early progeny production was slightly decreased and late reproduction was increased in clec-4 mutants, but there was no extension of the reproductive period (S4D Fig). Thus, mutations in clec-4 do only have an effect on fertility during early reproduction, but do not subject worms to dietary restriction as mutations in eat-2. In this context it is important to note that the magnitude of the effect of mutations in eat-2 on life history traits such as lifespan are correlated with the severity of the eating defect [64,79] and that longevity induced by mutation in eat-2 has been shown to be variable [83]. As clec-4 mutants only have a milder pharyngeal pumping phenotype, the defect in normal feeding behavior may alter the nutritional state of the animals that is reflected in their gene expression profile (Fig 4A and 4B), but food intake appears to be sufficient to support normal lifespan and fertility. C. elegans feeding behavior is influenced by the presence and quality of bacterial food and by internal nutrient status [84,85]. How nutrient status is sensed and transduced is not well understood. CLEC-4 is predicted to have a signal sequence (as 81% of all C. elegans CLEC proteins) and may thus be secreted into the intestinal lumen, where it may bind bacterial compounds or internal nutrients such as glucose or other glycans. Only one C. elegans CLEC protein was found to have sugar-binding activity so far: CLEC-79 binds to the non-reducing terminal galactose residues of glycans [86]. We thus aimed at further investigating CLEC-4 function also on the protein level and first analyzed binding of a recombinant CLEC-4 protein (production of the protein was outsourced to a specialized company; see materials and methods) to Gram-positive and Gram-negative bacteria. Recombinant CLEC-4 did not bind to any of the tested bacteria (see below). Also, we conducted a natural C. elegans N-glycan microarray [87] to investigate CLEC-4-carbohydrate interactions, but CLEC-4 did not bind to any carbohydrates included in the glycan array. It is difficult to interpret these negative results. CLEC-4 may still be a carbohydrate-binding protein, targeting motifs on other glycoconjugates (e.g., O-glycan and glycolipids) that were not included in the N-glycan array. We also cannot exclude that the recombinant protein is inactive due to misfolding, the lack of post-translational modifications, which may be required for its proper function, in the E. coli expression system, or missing co-factors. Thus, we were unable to determine the role, if any, of CLEC-4 in binding bacteria or glycans.

Simultaneous knock-down of clec-41 and clec-42 increases susceptibility to Bt247
The meta-analysis of pathogen-dependent expression of clec-4 and its paralogs revealed that clec-41 is co-expressed with clec-4 ( Fig 1A) As co-expression might indicate similar function, we explored function of clec-41 and its closest paralog clec-42. First, we used two RNAi clones from the Ahringer library, which both simultaneously target clec-41 and clec-42 and reduce mRNA levels of both genes (S2F Fig). In contrast to clec-4 deficiency, silencing clec-41;clec-42 expression in wildtype worms caused increased susceptibility to Bt247 infection (Figs 5A, 5B, S5A and S5B), but did not affect susceptibility to infection with PA14 or S. rubidaea MYb239  [55] and log-rank test [56]. Horizontal ticks represent censored data (missing worms), n = 5. Also see S5A and S5B Fig, and S2D

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins (Fig 5C and 5D). This indicates that clec-4 and clec-41/clec-42 have distinct functions and that certain C. elegans clec genes indeed function in defense against specific pathogens, as previously suggested [4].
To disentangle the roles of clec-41 and clec-42 in mediating resistance to Bt247 and to confirm the results of the RNAi experiments, we assessed survival of clec-41(tm6722) and clec-42 (tm6526) single mutants and a clec-41(tm6722);clec-42(tm6526) double mutant, respectively. However, the results were inconclusive. While the clec-41 single mutant was as resistant as wildtype worms (S5C Fig), clec-42(tm6526) animals were significantly more resistant than wildtype worms in two runs and more susceptible in three experimental runs (S5D Fig). Similarly, the clec-41(tm6722);clec-42(tm6526) double mutant was as resistant as wildtype animals in one run and more resistant in another run (S5E Fig). Thus, there is a discrepancy between clec-41;clec-42 double knock-out and knock-down phenotypes. This phenotypic difference between the clec-41(tm6722);clec-42(tm6526) double knock-out mutant and simultaneous knock-down of clec-41 and clec-42 may be due to off-target effects of the RNAi treatment or to genetic compensation. We searched for potential off-targets of the RNAi clones but could not identify any targets in addition to clec-41 and clec-42. Thus, it is possible that gene expression changes in other clec genes that mitigate the consequences of the clec-41 and clec-42 mutations and consequent compensation in the double mutant could be the reason for the observed differences. Genetic compensation or transcriptional adaptation in response to gene knock-out (but not gene knock-down) is a widespread phenomenon that has been observed in several model systems [88]. In C. elegans, genetic compensation was recently demonstrated for knockdown of act-5 and unc-89 [89]. If genetic compensation is indeed involved in the clec-41 (tm6722);clec-42(tm6526) double knock-out mutant needs to be determined in future. However, as the clec gene family is C. elegans 7 th largest gene family with 283 members, potential redundancy and complex interactions between different clec paralogs within the family may complicate the matter. Together, we conclude that analysis of knock-out mutants for genes of large gene families, such as the clec family, can have limited power for inferring gene functions, possibly because of gene compensation and/or redundancy. Therefore, we decided to further investigate the function of CLEC-4, CLEC-41, and CLEC-42 on the protein rather than the gene level.

CLEC-41 and CLEC-42 bind a broad range of bacteria and exhibit antimicrobial activity in vitro
Vertebrate CTLD proteins play important roles in pathogen recognition, acting as transmembrane PRRs or soluble PRRs and mediating intracellular signaling, but can also function as secreted antimicrobial proteins that kill bacteria [90]. Both functions imply direct binding to bacteria. As CLEC-4, CLEC-41 and CLEC-42 both have a signal peptide and are predicted to be secreted. To test the role of CLEC-4, CLEC-41, and CLEC-42 in bacterial recognition or elimination, we assessed binding of recombinant proteins (production of the individual proteins was outsourced to a specialized company; see materials and methods) to Gram-positive and Gram-negative bacteria. While the recombinant CLEC-4 protein did not bind any bacteria, CLEC-41 and CLEC-42 proteins bound to all tested bacteria (Fig 6A, 6B, and 6C). We next asked whether CLEC-4, CLEC-41, and CLEC-42 exhibit antimicrobial activity in vitro. While the CLEC-4 protein did not inhibit visible bacterial growth, CLEC-41 and CLEC-42 both exhibited antimicrobial activity against E. coli OP50, although only at higher concentrations ( Fig 6D). Moreover, CLEC-41 inhibited visible growth of the pathogenic Bt247, but CLEC-42 did not (Figs 6E and S6B). Combining both proteins had a mild synergistic effect on Bt247 growth inhibition (Figs 6E and S6B), which is in line with the effect on survival upon joint silencing of clec-41;clec-42 by RNAi (Fig 5A and 5B). In all cases, we were able to recover live bacteria from the wells without visible bacterial pellets, indicating that the CLEC proteins have bacteriostatic and not bactericidal activity. Interestingly, neither CLEC-41, nor CLEC-42 inhibited growth of Bt407 (S6A Fig), indicating specific interactions with different bacterial strains. While this is the first demonstration of antimicrobial function of C. elegans CLEC proteins, an in vitro bactericidal activity was previously described for several crustacean CTLD proteins, which also possess broad bacterial binding properties, for example Fc-hsL of the Chinese white shrimp Fenneropenaeus chinensis and EsLecA and EsLecG of the Chinese mitten crab Eriocheir sinensis that inhibit growth of Gram-positive as well as Gram-negative bacteria [91,92].
In summary, CLEC-41 and CLEC-42 contribute to C. elegans resistance to Bt247 infection (Fig 5A and 5B), bind to a broad range of bacteria (Fig 6B and 6C) and exhibit weak antimicrobial activity against E. coli and pathogenic Bt247 in vitro (Fig 6D and 6E). Although it is possible that the native proteins may behave differently in vivo, our in vitro results for the recombinant proteins suggest that CLEC-41 and CLEC-42 are bona fide antimicrobial immune effector proteins.

Conclusion
The exact functions of the extremely diversified C. elegans clec genes are largely unknown. Of the 283 clec gene family members only few have been studied at a functional genetic or protein level. Here, we explored the functions of CLEC-4, CLEC-41, and CLEC-42. We identified clec-4 as a novel regulator of C. elegans feeding behavior and provide evidence of a link between  Table. https://doi.org/10.1371/journal.ppat.1009454.g006

PLOS PATHOGENS
Diverse functions of C. elegans C-type lectin-like domain proteins clec-4 function and nutrient sensing (Fig 7). Further, we show that clec-41 and clec-42 are required for resistance against Bt247 infection and demonstrate antimicrobial activity of CLEC-41 and CLEC-42 in vitro (Fig 7). Our work reveals a novel function of C. elegans clec genes in regulating feeding behavior, it defines a role for CLEC-41 and CLEC-42 as bona fide immune effector proteins, and thus, it extends the current knowledge on the functional diversification of this large gene family.