Natural variation in the roles of C. elegans autophagy components during microsporidia infection

Natural genetic variation can determine the outcome of an infection, and often reflects the co-evolutionary battle between hosts and pathogens. We previously found that a natural variant of the nematode Caenorhabditis elegans from Hawaii (HW) has increased resistance against natural microsporidian pathogens in the Nematocida genus, when compared to the standard laboratory strain of N2. In particular, HW animals can clear infection, while N2 animals cannot. In addition, HW animals have lower levels of initial colonization of Nematocida inside intestinal cells, compared to N2. Here we investigate how this natural variation in resistance relates to autophagy. We found that there is much better targeting of autophagy-related machinery to parasites under conditions where they are cleared. In particular, ubiquitin targeting to Nematocida cells correlates very well with their subsequent clearance in terms of timing, host strain and age, as well as species of Nematocida. Furthermore, clearance correlates with targeting of the LGG-2/LC3 autophagy protein to parasite cells, with HW animals having much more efficient targeting of LGG-2 to parasite cells than N2 animals. Surprisingly, however, we found that LGG-2 is not required to clear infection. Instead, we found that LGG-2/LC3 regulates Nematocida colonization inside intestinal cells. Interestingly, LGG-2/LC3 regulates intracellular colonization only in the HW strain, and not in N2. Altogether these results demonstrate that there is natural genetic variation in an LGG-2-dependent process that regulates microsporidia colonization inside intestinal cells, although not microsporidia clearance.


Introduction
Natural genetic variation underlies differences in susceptibility to infection and inflammation among individuals [1]. Genome-wide association studies in humans have revealed genetic variation in innate immune genes that predispose individuals to increased risk of infection and autoimmune disease. For example, polymorphisms in autophagy genes Nod2, ATG16L, and IRGM are associated with increased risk for Crohn's disease, which is an inflammatory bowel disease characterized by a dysregulated gut microbiome [2,3]. The discovery that natural PLOS  during the L1 stage. Quantitative genetic analysis indicated that the genomic regions that confer increased immunity contain genes predicted to encode ubiquitin ligase adaptor proteins [28]. Therefore, we hypothesized that HW C. elegans may be better than N2 at targeting Nematocida cells with ubiquitin to direct autophagy machinery for clearance of these parasites via xenophagy. Indeed, because C. elegans lacks professional immune cells, and its intestinal epithelial cells are non-renewable, xenophagy provides an attractive candidate for a process that could explain how C. elegans can clear microsporidia infection from their intestinal cells [30,31].
Here we identify natural variation in both C. elegans and Nematocida in the targeting of host autophagy-related machinery to pathogen cells as well as the importance of autophagyrelated machinery for regulating infection levels. We found a strong correlation between targeting of autophagy-related machinery to Nematocida cells and the ability to clear these pathogens from host intestinal cells. For example, we found that HW animals have much higher levels of ubiquitin targeting to Nematocida cells compared to N2 animals. Furthermore, these differences are specific to the developmental stage at which HW animals can clear infection. We also observed higher frequencies of ubiquitin targeting around cells of a Nematocida species that is cleared, compared to two Nematocida species that are not cleared. In addition to ubiquitin, we observed higher frequencies of the autophagy protein LGG-2/LC3 targeted to Nematocida cells in HW animals compared to N2 animals. Surprisingly however, we found that LGG-2 is not required for clearance in HW animals, indicating that there are other proteins responsible for clearing this co-evolved pathogen. Instead, we found that LGG-2/LC3 regulates pathogen colonization inside intestinal cells. Interestingly, LGG-2/LC3 regulates intracellular colonization only in the HW strain, and not in N2. Previous work mapped HW resistance to genomic regions other than the region containing lgg-2 [28], and the LGG-2 protein does not appear to vary in amino acid sequence between N2 and HW (www.wormbase. org). Therefore, the lgg-2 gene itself does not appear to underlie the genetic variation in pathogen resistance between N2 and HW. Rather, these findings indicate that an lgg-2-dependent process appears to naturally vary between N2 and HW, and is required to control levels of Nematocida parasites at a very early stage of infection.

Nematocida infection assays and imaging
Infection experiments were performed with N. ironsii strain ERTm5, N. parisii strain ERTm1, and N. ausubeli strain ERTm2 [18,29,36]. Spores were prepared and quantified as previously described [37]. Host animals were either continuously exposed to Nematocida spores or pulseinoculated with spores for 3 hours on plates as described in the figure legends. For S7 Fig, animals were incubated in liquid culture for 15 minutes. Infections of L1 stage animals or L4 stage animals were initiated as described previously [28]. Experiments were carried out at 20˚C (  Each experiment included analysis of at least 10 animals, and each animal was infected by one to nine parasite cells. Data shown are compiled from more than three independent experiments per strain. (C) Abundant GFP::UBQ-1 (green) accumulation in close proximity to N. ironsii cell (red). Scale bar = 5 μm. From 4 independent experiments totaling 117 N2 animals with 296 Nematocida cells and 169 HW animals with 234 Nematocida cells, extreme amounts of ubiquitin targeting were observed in 10 out of 234 cells in HW, and were never observed in N2 animals. (D) Analysis of GFP:: UBQ-1 expression in N2 and HW strains using a COPAS Biosrt machine, with expression measured on an individual animal basis. AU, arbitrary units. Boxplots show the interquartile range (IQR) from 25th to 75th percentile with horizontal lines indicating medians. The range bars encompass all data within 1.5 IQR above and below the upper and lower IQRs, respectively. Dots show individuals outside of that range. Only GFP levels in GFP + animals are plotted, where GFP + includes all animals with GFP levels above non-transgenic N2 animals. More than 1000 animals were measured per strain and condition. MicroB FISH probe conjugated to the CAL Fluor Red 610 dye at 5ng/μl overnight, then resuspended in PBST (clearance experiments) or Vectashield with DAPI (Vector Labs) to stain DNA (colocalization experiments). mCherry::GFP::LGG-1 animals were stained with the MicroB FISH probe conjugated to the Pacific Blue dye at 10ng/μl overnight, with no staining of DNA. For infection clearance experiments, the percentage of infected animals out of 100 total animals per sample was calculated as described previously [28]. To image the localization of transgenic proteins during Nematocida infections, fixed and stained animals were mounted  on 5% agarose pads and imaged using a 40X oil immersion objective on a Zeiss LSM700 confocal microscope run by ZEN2010 software. At least 10 infected transgenic animals were imaged per sample and Nematocida cells were counted as having transgenic proteins localized around them or not.
Blue dots represent the frequencies observed in HW animals; red dots represent frequencies in N2 animals. Data shown are compiled from more than three independent experiments per strain. (C) GFP-fusion autophagy protein analysis in transgenic N2 and HW strains. Transgenic GFP expression measured on an individual animal basis. AU, arbitrary units. As in Fig 1, boxplots show the interquartile range (IQR) from 25th to 75th percentile with horizontal lines indicating medians. The range bars encompass all data within 1.5 IQR above and below the upper and lower IQRs, respectively. Dots show individuals outside of that range. Only GFP levels in GFP + animals are plotted, where GFP + includes all animals with GFP levels above non-transgenic N2 animals. More than 1000 animals were measured per strain and condition. https://doi.org/10.1371/journal.pone.0216011.g003

Measurements of transgenic protein levels
For Western blot measurements of protein levels, animals were washed off NGM plates with M9 buffer, and then washed once with PBS with 0.1% Tween 20 to remove bacteria. Samples were then resuspended in sample buffer with 1% SDS and 50 mM DTT and boiled at 95˚C for 10 minutes. Lysates were then run on a 4-20% gradient SDS-PAGE gel (Bio-Rad) and transferred to PVDF membrane (Bio-Rad). The blots were stained with Ponseau S stain (Sigma-Aldrich) and imaged for total protein before staining with an anti-GFP antibody made in rabbits (a gift from Arshad Desai and Karen Oegema labs at UC San Diego) diluted at 1:5000 overnight at 4˚C, then staining with an antirabbit HRP antibody at 1:10000 for 45 minutes at room temperature. The blots were treated with ECL reagent (Amersham GE Healthcare Life Sciences) and imaged on a Bio-Rad ChemiDoc. For measurements of GFP protein levels in live animals, uninfected and N. ironsii infected animals were collected 15 hour post-inoculation (hpi) and measured on a COPAS Biosort machine (Union Biometrica) for size (time-offlight, TOF) and GFP (green). More than 1000 animals were measured per sample. Animals expressing transgenic GFP were distinguished from non-expressing animals by measuring GFP levels in wild type N2 animals and only analyzing GFP levels in animals above this threshold.

CRISPR-Cas9 generation of lgg-2 deletion alleles
Using the CRISPR design tool (http://crispr.mit.edu), DNA sequences with high scores were chosen in regions upstream from the start and downstream from the stop codon of lgg-2 to create crRNAs (synthesized by Integrated DNA Technologies). crRNAs were annealed to the tracrRNA, and sgRNA products were co-injected with Cas9 protein and dpy-10 sgRNA LGG-2 expression in lgg-2(jy44) mutants is able to rescue the higher pathogen load phenotype. 300 animals were analyzed for each genotype. (C) lgg-2 deletion in N2 background does not affect N. ironsii colonization. 300 animals were analyzed for each strain. (A-C) Results from six (A) or three (B, C) independent experiments are shown as box-and-whisker plots, indicating the number of N. ironsii sporoplasms per animal at 3 hpi. Each box represents 50% of the data closest to the median value (line in the box). Whiskers span the values outside of the box. A student's t-test was used to calculate p values; p < 0.001 is indicated with four asterisks; p < 0.05 is indicated with one asterisk; ns indicates nonsignificant difference (p > 0.05). Experiments were performed at 25˚C. https://doi.org/10.1371/journal.pone.0216011.g005 Natural variation in the roles of autophagy components during microsporidia infection marker [38]. Dumpy F1 animals were screened for lgg-2 deletions using PCR analysis. Isolated homozygous lines were confirmed and characterized by DNA sequence analysis and backcrossed three times to the original strain, CB4856 for jy44 and jy45 alleles, and N2 for jy102 and jy103. The jy44 allele contains a 1376-nucleotide deletion, which removes the entire lgg-2 gene and an additional 132 nucleotides upstream and 146 nucleotides downstream of the gene. Similarly, the whole lgg-2 gene together with 138 nucleotides upstream and 146 nucleotides downstream (1382 nucleotides in total) are substituted for a single nucleotide insertion in jy45 allele. In the N2 background, the jy102 allele contains a 1356-nucleotide deletion, which removes the whole lgg-2 gene together with 125 nucleotides upstream and 133 nucleotides downstream of the gene; the jy103 allele is an indel with 7 nucleotides inserted in place of 1344 nucleotides, which spans the entire lgg-2 gene together with 122 nucleotides upstream and 124 nucleotides downstream of the gene.

Sequence of lgg-2 in N2 and CB4856
According to www.wormbase.org, there are no predicted amino acid changes in LGG-2 between N2 and CB4856. There is one nucleotide substitution in lgg-2 in CB4856 animals compared to N2 animals (WBVar00191911): A/T in the first intron. Also, there is one nucleotide substitution in a possible promoter region (WBVar00191912): G/A.

Gene expression measurements
To confirm the absence of lgg-2 mRNA transcripts in lgg-2(jy44) mutants, synchronized populations of CB4856 wt and HW lgg-2(jy44) animals were grown at 20˚C to reach L4 stage, and then collected in TriReagent (Molecular Research Center, Inc.), and RNA extraction was performed following manufacturer's guidelines. cDNA synthesis was performed using Super-Script VILO cDNA Synthesis Kit (Thermo Fisher Scientific) following manufacturer's guidelines. qRT-PCR was performed as using a BioRad CFX Connect. lgg-2 mRNA was amplified in two separate reactions using the same forward primer (5'-GAATCGTTCCATCGTTCA AGG-3') and two different reverse primers (5'-TGTTGGCTGCGGATTTCT-3' and 5-TTGGAGGCGTCGTCTAACA). Each of two replicate experiments was measured in duplicate and compared to the expression of snb-1 control gene.

N. ironsii cells are surrounded by ubiquitin more frequently inside the Hawaiian strain of C. elegans than the N2 strain
To investigate the hypothesis that HW C. elegans clear N. ironsii infections through improved targeting of ubiquitin to parasite cells followed by autophagic clearance, we examined the localization of ubiquitin to parasite cells in the HW host compared to the N2 host strain, which does not clear infections. First, we generated transgenic N2 and HW strains that express GFP-tagged ubiquitin under the control of the intestinal-specific promoter vha-6 (GFP::UBQ-1). We pulse-infected these GFP::UBQ-1-expressing strains at the first larval (L1) stage by feeding animals with N. ironsii spores for three hours, then removed them from spores and fixed them at different hours post-inoculation (hpi). We performed FISH staining on these fixed animals using a probe that targets Nematocida rRNA to label parasite cells and then determined the percentage of those cells that also have GFP::UBQ-1 localization (Fig 1A and 1B). With this method, we found that a much higher percentage of parasite cells had ubiquitin localization to parasite cells in HW animals than in N2 animals. For example in one experiment at 15 hpi, we found that 84% of Nematocida cells (out of 19 total parasite cells analyzed in 10 separate animals) were targeted with GFP::UBQ-1 in HW animals (Fig 1B), and every animal had at least one parasite cell with GFP::UBQ-1 localization. In contrast, only 11% of Nematocida cells (out of 37 total parasite cells analyzed in 10 separate animals) were targeted with GFP::UBQ-1 in N2 animals at 15 hpi ( Fig 1B). Interestingly, in some HW animals we observed exuberant accumulations of ubiquitin that formed long structures next to parasite cells, which were never observed in N2 animals ( Fig 1C).
Next we analyzed how ubiquitin targeting to parasites correlates with their clearance. In previous clearance experiments at 25˚C we had found parasite load much lower in HW animals at 20 hpi compared to 3 hpi, but did not measure when clearance began between those two timepoints [28]. Here we analyzed clearance at several timepoints after 3 hpi and saw that clearance begins around 10 hpi at 25˚C (S1A Fig). Because we have found that there are more protein aggregates at 25˚C, which can make localization experiments challenging [22], we performed our autophagy localization experiments at 20˚C (this paper and [22]). (We do note that previous studies indicate that autophagy does not appear to proceed differently at 20˚C compared to 25˚C [33].) Therefore, we also analyzed clearance at 20˚C to ensure that clearance is not just a high-temperature phenomenon, and to better compare the kinetics of clearance and ubiquitin targeting. We found that parasite clearance also occurs at 20˚C (S1B Fig). Furthermore, we were able to compare the kinetics of clearance with ubiquitin targeting at the same temperature, and found that the peak of ubiquitin targeting correlates with the start of clearance (compare Fig 1B and S1B Fig). These findings are consistent with ubiquitin targeting driving parasite clearance.
One potential explanation for increased targeting of ubiquitin to parasite cells in HW animals is increased expression from the GFP::UBQ-1 transgene in HW animals. To investigate this possibility, we compared protein levels of GFP::UBQ-1 with two methods. First, we performed Western blots with anti-GFP antibodies on lysates from transgenic HW and N2 animals to compare levels of GFP::UBQ-1 across populations of animals. Here we found that HW animals do not have increased levels of GFP::UBQ-1 compared to N2 animals (S2 Fig). Second, we analyzed GFP::UBQ-1 fluorescence on a per-animal level using a COPAS Biosort 'worm sorter'. While there was animal-to-animal variability in fluorescence levels of GFP-ubiquitin, the overall levels were not higher in HW compared to N2 animals ( Fig 1D). Therefore, the increased targeting of GFP-ubiquitin to N. ironsii cells in HW compared to N2 does not appear to be simply due to increased GFP::UBQ-1 levels in HW animals.

Localization of ubiquitin to parasite cells correlates with their clearance in terms of host age and parasite species
In previous work, we showed that HW animals can clear infection by N. ironsii only during the L1 larval stage, and that this ability is lost in later larval stages [28]. To characterize genetic variation in clearance on the parasite side, we infected HW at the L1 and L4 stages with three different Nematocida species. We observed variation in clearance of these Nematocida species by HW animals, where N. ironsii was cleared by L1, but not L4 animals while N. parisii and N. ausubeli were cleared poorly by both L1 and L4 animals (Fig 2A and 2B). Therefore, the clearance of microsporidia by HW animals is specific to N. ironsii infections at the L1 stage.
To determine how these levels of clearance relate to the association of host ubiquitin with parasite cells, we examined the frequency of ubiquitin localization to Nematocida cells in L1 HW animals over time. Here we observed a high frequency of ubiquitin localization around N. ironsii cells but not N. parisii or N. ausubeli cells (Fig 2C). Furthermore, the high frequency of ubiquitin localization to N. ironsii cells was observed in HW animals infected at the L1 stage but not in HW animals infected at the L4 stage or N2 animals at either stage ( Fig 2D). Thus, the frequency of host ubiquitin localization around microsporidia cells in C. elegans is highly correlated with clearance of infection.

Localization of LGG-2/LC3 but not LGG-1/GABARAP correlates with parasite clearance
Xenophagic clearance of intracellular pathogens involves targeting of autophagic machinery after ubiquitin targeting to pathogen cells. One of the most well-studied autophagic events is activation and conjugation of proteins in the LC3/GABARAP family to the developing autophagosome, which are required for fusion with the lysosome [39]. Therefore, we examined whether LGG-1 (homolog of GABARAP in mammals) or LGG-2 (homolog to LC3 in mammals), localized more frequently to parasite cells in HW worms compared to N2 worms. To analyze localization, we first generated N2 and HW transgenic animals with lgg-1p::mCherry:: GFP::LGG-1 or lgg-2p::GFP::LGG-2 transgenes. Then we performed experiments similar to the ubiquitin studies in Fig 1. We pulse-infected these animals for 3 hours, fixed them at various timepoints after inoculation, stained with a FISH probe to label N. ironsii cells and then quantified parasite localization by following GFP (Fig 3A). Here we found that there was a gradual increase in mCherry::GFP::LGG-1 targeting to N. ironsii cells over time in both N2 and HW animals, with no obvious difference between the two strains ( Fig 3B). In contrast, we found that there was a much greater percentage of parasite cells targeted with GFP::LGG-2 in HW animals, compared with N2 animals (Fig 3B). This increased localization of LGG-2 was comparable to the increased ubiquitin targeting seen in HW compared to N2 (Fig 1C).
To eliminate the possibility that the increased targeting of LGG-2 in HW animals was simply due to increased levels of GFP::LGG-2 expression, we performed Western blot analysis (S3 Fig) and COPAS Biosort 'worm sorter' analysis of GFP::LGG-2 (and mCherry::GFP::LGG-1) (Fig 3C). Similar to our analysis of ubiquitin levels, we did not see an increased level of GFP:: LGG-2 (or mCherry::GFP::LGG-1) expression in HW compared to N2 animals.

LGG-2/LC3 is not required for clearance of N. ironsii
The increased localization of LGG-2 to N. ironsii cells in HW, and the ability of this strain to clear infection led us to hypothesize that parasite cells associated with LGG-2 were destined to be cleared, and that LGG-2 was important for the clearance. Therefore, we examined a functional role for LGG-2 in defense against N. ironsii. To do this we used CRISPR/Cas9 editing to generate a complete deletion of the lgg-2 locus in the HW strain background. We isolated and characterized two independent alleles (called HW lgg-2(jy44) and HW lgg-2(jy45)) in which the entire lgg-2 locus is deleted and should produce no LGG-2 protein (Fig 4A). We confirmed via qRT-PCR that lgg-2 mRNA was not produced in lgg-2(jy44) mutant animals (S2 Table). lgg-2 mutant strains were backcrossed three times and then analyzed for their ability to clear N. ironsii. In several independent experiments, we saw that HW lgg-2 mutants were still able to clear N. ironsii between 3 hpi and 20 hpi (Fig 4B and S4A Fig). Therefore, despite the striking increase in localization of LGG-2 to N. ironsii cells in HW animals compared to N2 animals, LGG-2 is not required for clearance of N. ironsii under these conditions.

LGG-2 regulates N. ironsii colonization of intestinal cells in HW animals but not in N2
In addition to differences in clearance ability, our previous studies indicated that N2 animals have an increase in intracellular colonization at 3 hpi compared to HW animals, but only at the L1 stage [28]. Consistent with this observation, we noticed in our clearance assays here that N2 animals had a higher percentage of infected animals than HW animals (Fig 4B). Interestingly, we also saw that HW lgg-2 mutants had a higher percentage of animals infected with N. ironsii compared to HW wild-type animals, with HW lgg-2 mutants having a percentage of infected animals closer to N2 wild-type animals (Fig 4B and S4A Fig). When we analyzed pathogen load, we found that HW lgg-2 animals had more sporoplasms per animal than HW wildtype animals at 3 hpi, with levels similar to N2 wild-type animals (Fig 5A and S4B Fig). To ensure that the higher pathogen load phenotype was specific to lgg-2(jy44), we crossed the GFP::LGG-2 construct used for localization into HW lgg-2 mutant animals and found that this construct successfully rescued the higher pathogen load phenotype of HW lgg-2(jy44) mutants ( Fig 5B).
One potentially trivial explanation for the higher pathogen load in HW lgg-2 mutants could be that they have greater exposure to the pathogen, because it accumulates to a higher level within the intestinal lumen either through increased feeding or decreased defecation. To examine this possibility, we used a fluorescent bead-feeding assay and measured accumulation of beads inside the intestinal lumen in HW and HW lgg-2 mutants. Here we found that there was no increase in the accumulation of beads in HW lgg-2 mutants, indicating that their increased parasite load at 3hpi is likely not due to increased exposure to parasites (S5 Fig). Therefore, lgg-2 appears to be acting at a very early stage to regulate pathogen load of the natural intracellular parasite N. ironsii in HW animals.
Although clearance still occurred in HW lgg-2 mutants, these animals had slightly less efficient clearance between 3 hpi and 20 hpi compared to HW wild-type animals (S3 Table). To determine whether this was due to the difference in initial dose at 3 hpi between HW and HW lgg-2 mutants, we infected HW animals with a higher spore dose than HW lgg-2(jy44) animals such that both strains had roughly equivalent initial sporoplasm number at 3 hpi (S6 Fig). Under this condition, we found that clearance from 3 hpi to 20 hpi was similar in HW and HW lgg-2 mutants, indicating that lgg-2 is not required for clearance of N. ironsii, but is required for regulating the initial colonization of this natural parasite inside host intestinal cells.
We next considered the possibility that HW wild-type animals may already have cleared some infection before being analyzed for pathogen load at 3 hpi. If this were the case, then the increase in sporoplasm number in lgg-2(jy44) mutant animals at this timepoint would be a defect in clearance, rather than an increase in colonization. To test this hypothesis, we infected N2, HW wild-type and HW lgg-2(jy44) animals with a high dose of N. ironsii spores for 15 minutes and then analyzed the number of sporoplasms per animal. Interestingly, even after this very short-term incubation, HW lgg-2(jy44) mutants showed higher pathogen load than the HW wild type, suggesting that the intestinal cells in these mutants are more easily colonized by N. ironsii (S7 Fig). Because N2 wild-type animals have a higher initial colonization rate of Nematocida compared to HW wild-type animals, we hypothesized that some lgg-2-dependent process active in HW animals (that is inactive in N2) might be responsible for this difference. If this were the case, then loss of lgg-2 in the N2 background would not have an effect. To test this hypothesis, we generated two lgg-2 deletion alleles (jy102 and jy103) in the N2 genetic background and tested for pathogen load at 3 hpi. Notably, we observed similar infection rates in wild-type and lgg-2(jy103) mutant animals (Fig 5C). We confirmed this result using the other deletion allele, lgg-2(jy102) (S8A Fig), and also showed that there is no significant difference in bead accumulation between N2 wild type and lgg-2 mutant animals (S8B Fig). Therefore, the role of LGG-2 in pathogen colonization appears to be specific to the HW background, and the difference in colonization between N2 and HW appears to depend on LGG-2.

Discussion
These studies reveal natural variation in the roles of C. elegans autophagy-related machinery in response to natural Nematocida pathogens (Fig 6). We observe natural variation in localization of ubiquitin to parasite cells in different strains, and this localization correlates very well with clearance of parasite. In addition, there is a correlation between clearance and localization of the autophagy protein LGG-2/LC3 to N. ironsii cells, with HW animals having increased targeting compared to N2 animals. These findings led us to hypothesize that parasite cells associated with LGG-2/LC3 are cleared, which could explain the greater resistance of HW animals compared to N2. However, we found that HW lgg-2 mutants can still clear infection, which does not support this hypothesis. Instead, we found that lgg-2 regulates the level of microsporidia colonization inside intestinal cells in HW animals, but not in N2 animals. The LGG-2 predicted amino acid sequence does not vary between N2 and HW (see Materials and methods). Furthermore, quantitative genetic studies indicate that the basis for the N2 and HW difference maps to chromosomes II, III and V, and not to chromosome IV, where lgg-2 resides [28]. Therefore, we do not favor a hypothesis that lgg-2 itself is responsible for the genetic variation between these two strains. Instead, we propose that an lgg-2-dependent process (and not lgg-2 itself) is active in HW animals to regulate levels of colonization, and is not active in N2 animals. In addition, our findings indicate that an lgg-2-independent process mediates clearance.
What non-LGG-2 protein mediates clearance of Nematocida cells in HW animals? In mammals there are six proteins in the LC3/GABARAP family that contribute redundantly to autophagy-mediated clearance [39]. C. elegans has two proteins in this family, LGG-2 (homolog of LC3) and LGG-1 (homolog of GABARAP), and they do have redundant function in other contexts [34,35]. Therefore, it is possible that LGG-1 (GABARAP ortholog) is able to compensate for LGG-2 in clearance of microsporidia in HW C. elegans. It is difficult to test this model because LGG-1 is an essential gene, and HW worms are resistant to germline RNAi [40,41], which is required to perform RNAi knockdown treatments that will affect L1 larvae where we see the resistance phenotype. However, we find it unlikely that LGG-1 could substitute for LGG-2 given the extensive LGG-1 localization to parasite cells we observed in both resistant HW and susceptible N2 animals (see Fig 3).
Our findings do not rule out LGG-2-mediated xenophagy as playing a role in clearance, but they show it is not required. It is possible that LGG-2 does mediate clearance through xenophagy, and this process is redundant with another form of clearance (Fig 6). Given that C. elegans and Nematocida species appear to have a long co-evolutionary relationship [18], it is possible that C. elegans evolved additional forms of clearance as part of an arms race between this host and its natural pathogen. Studies in other systems have described pathogen virulence factors that suppress host autophagy, for example through removing LC3 conjugation from the autophagosomal membrane [10]. Perhaps because another pathogen suppressed LGG-2-mediated autophagy in its evolutionary past, C. elegans evolved a separate pathway for clearance that functions here in the absence of LGG-2 to clear N. ironsii. Regardless of the reason, it is clear that HW animals have a pathway separate from LGG-2 that mediates clearance of N. ironsii. Given the extremely close correlation between ubiquitin targeting and Nematocida clearance (Figs 1 and 2), perhaps ubiquitin is functionally relevant for this pathway, and serves to recruit another type of degradative machinery to the parasite in parallel to LGG-2. Unfortunately, because ubiquitin is an essential gene and functions in many processes, it is difficult to test a functional role in clearance specifically. It is also possible that the ubiquitin targeting we see is not functionally relevant but is simply correlative with clearance. Indeed, there are several other studies where targeting of autophagic machinery to microbes is not indicative of a functional role in clearance [11,42]. It is difficult to thoroughly test a role for autophagy in the clearance of Nematocida from HW animals, because many autophagy proteins are essential, and as mentioned above, germline RNAi knock-down in HW is ineffective. Nonetheless, our findings suggest C. elegans evolved additional, non-xenophagic strategies for clearance of natural pathogens.
Interestingly, our results point to a clearance-independent role for lgg-2 in controlling colonization of Nematocida inside of intestinal cells of HW animals, after the spores are ingested into the intestinal lumen. One explanation for these results is that lgg-2 has a role in regulating the viability of microsporidia spores in the extracellular lumenal space, or in regulating their ability to fire the polar tube for invasion of intestinal cells. If so, this effect would be consistent with LGG-2 acting through secretion of anti-microbial compounds, as has been shown for regulation of lysozyme secretion by the autophagy pathway in the mammalian intestine [13]. Another possibility is that lgg-2 regulates the intracellular establishment of sporoplasms after they have been delivered into the cytoplasm by the polar tube, and leads to their degradation so quickly that we do not detect them in the FISH assay. Nothing is known about whether host cells control this initial stage of microsporidia invasion, as polar tube firing is described as a mechanical rupture that 'forces' a parasite cell into the host cell [19]. However, it is exciting to consider that host cells may regulate this step via LGG-2/LC3. It is important to note that LC3 has non-autophagy roles in immunity, for example through LC3-associated phagocytosis or LAP [43,44]. During LAP, microbes that have been phagocytosed by host cells are targeted by LC3 and then degraded through phagosomal/lysosomal fusion. While we have not seen a separate membrane around N. parisii sporoplasms suggesting they are not phagocytosed [21], it is possible that LC3/LGG-2 is recruited to sporoplasms that are in direct contact with cytoplasm, and directs their immediate degradation through some process related to LAP.
Our observations that lgg-2 controls sporoplasm levels at both 15 minutes post-inoculation and 3 hpi are arguably the earliest stages at which a host factor has been shown to control microsporidia load in any system. Studies of microsporidia species that infect mammalian cells have implicated host glycosaminoglycans [45] and a transferrin receptor protein [46] in regulation of early steps, but the exact stage and mechanism by which they act is unknown. It will be interesting to further explore this early role for lgg-2 to shed light on the poorly understood question of how hosts control microsporidia infection. ironsii clearance in HW wild-type and lgg-2(jy44) animals is similar if HW wt animals are infected with a higher dose of microsporidia spores than HW lgg-2(jy44) mutants, to achieve similar initial intestinal colonization in both strains. (A) N. ironsii clearance in HW wild-type animals (green lines) and lgg-2(jy44) mutants (orange lines). Thick lines represent average values of two experiments; dotted lines indicate results from individual experiments. 100 animals were analyzed per strain at 3 hpi and 20 hpi (x-axis). (B) Box-and-whiskers plot shows similar N. ironsii infection rate between HW wild-type and lgg-2(jy44) mutant animals. Each box represents 50% of the data closest to the median value (line in the box). Note that the median value for HW lgg-2(jy44) sample is two sporoplasms per animal and that it overlaps with the upper boundary of the box. Whiskers span the values outside of the box. A student's t-test was used to calculate p values; ns indicates non-significant difference (p > 0.05). (TIF) S7 Fig. Short-term exposure to high dose of pathogen causes higher infection rate in HW lgg-2 deletion mutants. Significant difference in pathogen load is observed between HW wildtype and HW lgg-2 deletion mutant animals after 15 minutes incubation with 12 million N. ironsii spores (6 times more spores than in other clearance assays with lgg-2 mutants). Results from two independent experiments are shown as box-and-whisker plots, indicating the number of N. ironsii sporoplasms per animal at 3 hpi. Each box represents 50% of the data closest to the median value (line in the box). Whiskers span the values outside of the box. 400 animals were analyzed for each strain. A student's t-test was used to calculate p values; p < 0.001 is indicated with four asterisks; ns indicates non-significant difference (p > 0.05). Experiments were performed in liquid culture at 25˚C. (TIF) S8 Fig. Deletion of lgg-2 in the N2 background does not affect N. ironsii colonization. (A) lgg-2(jy102) mutants show similar infection rate as the wild type control. Results from two independent experiments are shown as box-and-whisker plots, indicating the number of N. ironsii sporoplasms per animal at 3 hpi. Each box represents 50% of the data closest to the median value (line in the box). Whiskers span the values outside of the box. 200 animals were analyzed for each strain. (B) N2 wild type and two N2 lgg-2 mutant strains (lgg-2(jy103) and lgg-2(102)) have similar feeding rate. Bead fluorescence is standardized to TOF. At least 450 animals were analyzed for each strain, combined from three replicate experiments. (A, B) A student's t-test was used to calculate p values; ns indicates non-significant difference (p > 0.05). Experiments were performed at 25˚C. (TIF) S1 Table. List of strains used in this study. (XLSX) S2 Table. qRT-PCR measurements of lgg-2 mRNA transcripts in HW wild-type and lgg-2 (jy44) animals. Ct (cycle threshold) values for primer-template combinations are indicated in the table. The Ct is defined as the number of cycles required for the fluorescent signal to cross the threshold (ie exceeds background level), with higher Ct values indicating lower levels of starting template. A Ct value of 40 shows there was no signal detected for the lgg-2 primers in the lgg-2(jy44) mutant. (Because there was no signal we did not provide a quantitative comparison among samples, which is the typical method for reporting qPCR data, e.g. [47]). noRT samples represent negative controls in which reverse transcriptase was not added during cDNA synthesis. (XLSX) S3 Table. Statistical analysis of N. ironsii clearance in HW wt and HW lgg-2(jy44) animals at 20 hpi. To compare clearance efficiencies between HW wild-type and lgg-2(jy44) animals in six separate experiments, infection rates in each experiment at 3 hpi were standardized to a starting value of 100%. Then infection rates at 20 hpi were proportionally re-calculated and subtracted from 100% to determine the clearance percentage for each strain. Student's t-test was used to compare clearance and calculate the p value.