An extracellular Argonaute protein mediates export of repeat-associated small RNAs into vesicles in parasitic nematodes

Mobile small RNAs are an integral component of the arms race between plants and fungal parasites, and several studies suggest microRNAs could similarly operate between parasitic nematodes and their animal hosts. However, whether and how specific sequences are selected for export by parasites is unknown. Here we use density gradient purification and proteinase K sensitivity analysis to demonstrate that a specific Argonaute protein (exWAGO) is secreted in extracellular vesicles (EVs) released by the gastrointestinal nematode Heligmosomodies bakeri, at multiple copies per EV. Phylogenetic and gene expression analyses demonstrate exWAGO is highly conserved and abundantly expressed in related parasites, including the human hookworm and proteomic analyses confirm this is the only Argonaute secreted by rodent parasites. In contrast, exWAGO orthologues in species from the free-living genus Caenorhabditis are highly diverged. By re-sequencing and re-annotating the H. bakeri genome, and sequencing multiple small RNA libraries, we determined that the most abundant small RNAs released from the nematode parasite are not microRNAs but rather secondary small interfering RNAs (siRNAs) that are produced by RNA-dependent RNA Polymerases. We further identify distinct evolutionary properties of the siRNAs resident in free-living or parasitic nematodes versus those exported in EVs by the parasite and show that the latter are specifically associated with exWAGO. Together this work identifies an Argonaute protein as a mediator of RNA export and suggests rhabditomorph nematode parasites may have co-opted a novel nematode-unique pathway to communicate with their hosts.


79
The majority of our understanding of nematode RNAi pathways is based on the 80 free-living model organism Caenorhabditis elegans, which has at least four types of 81 endogenous sRNAs and 25 Argonaute genes (11). In addition to miRNAs and 82 piRNAs, C. elegans produces small interfering RNAs (siRNAs) from exogenous or endogenous double-stranded RNAs (dsRNAs). There is also a mechanism for de 84 novo generation of siRNAs by RNA-dependent RNA polymerases (RdRPs), which 85 are recruited to sRNA-target transcripts to amplify the silencing signal through the 86 generation of secondary siRNAs. The secondary siRNAs dominate the sRNA 87 content of adult C. elegans and have also been documented in several parasitic 88 nematode species (12)(13)(14)(15). They are distinguished from other sRNAs by the 89 presence of a 5' triphosphate and a preference for a 5' guanine. In C. elegans, 90 secondary siRNAs associate with WAGOs and have been shown to be important in 91 self versus non-self-recognition in the germline (16)(17)(18)

99
Here we examine the molecular and evolutionary properties of exWAGO in 100 parasitic and free-living nematodes and demonstrate that exWAGO mediates the 101 selective export of specific siRNAs in EVs. We compare the genomic origin of 102 siRNAs exported in EVs by H. bakeri to the resident siRNAs expressed in adults of 103 both H. bakeri and C. elegans. Our results support a model where the resident 104 sRNAs are dominated by secondary siRNAs, which are used for endogenous gene 105 regulation and control of retrotransposons. In contrast, the parasite preferentially 106 exports secondary siRNAs that are produced from newly evolved repetitive 107 elements in the genome that associate with exWAGO. This adds evolutionary 108 breadth to the handful of reports in mammalian systems suggesting RNA-binding some cases under specific signalling conditions (24). However, Argonautes have 120 also been reported to be contaminants that co-purify with EVs (21). In order to 121 rigorously determine whether the exWAGO that we have identified exists within 122 EVs, we used ultracentrifugation followed by flotation on a sucrose gradient for 123 purification, quantification by nanoparticle tracking analysis and visualisation by 124 transmission electron microscopy. As shown in Figure 1, the EVs had a density of 125 1.16-1.18 g/cm3 and co-purified with exWAGO. We further subjected the sucrose-126 purified EV fractions to proteinase K treatment and confirmed that the exWAGO 127 was protected from degradation but became susceptible when the EVs were lysed 128 with detergent ( Figure 1D). We analysed a defined number of sucrose-gradient 129 purified EVs by western blot in comparison to recombinant exWAGO and found 130 that exWAGO was present at 3.4 ± 1.1 copies per EV ( Figure 1E).

131
An improved genome assembly and annotation to explore extracellular

152
Over half (58.3%) of the H. bakeri genome contains some type of repeat element, 153 including LINE elements (12.6% of the genome) and DNA elements (12.8%) ( Table   154 1). Of all the repeats, 33.3% were found within genes (mostly in introns, which themselves occupy 33.5% of the genome). Interestingly, 30.6% of the genome was 156 annotated as unclassified repeats, nearly two-thirds of which do not overlap any 157 other kind of annotation.

283
To further explore and quantify this selectivity, we calculated, for each polyP-284 enriched cluster, a measure of entropy-based Information Content (IC) using either 285 adult or EV reads (see Methods). The higher the IC value, the more concentrated 286 the reads are in a few peaks, while the lower the IC value, the more evenly 287 distributed the reads are across the cluster (e.g. Figure 5A, inset). Interestingly, the 288 IC values are consistently higher for reads coming from EVs than from adult 289 libraries ( Figure 5A), indicating that the EVs more often contain reads from specific 290 peaks and are not a random sampling of the adult sRNA pool. Figure 5B illustrates

325
Our immunoprecipitation experiments show that Y RNAs, which are also abundant 326 in EVs (6), do not associate with exWAGO, suggesting this protein is specific for 327 siRNAs. Mechanistically, we envision three processes that could contribute to the 328 total RNA present in EVs, acting independently or together. The EVs could be 329 passively loaded with the sRNAs present in the cell type from which EVs are 330 exported. It is likely that EVs are released from the intestine (6). Secondly, the sRNAs could be actively loaded by some intrinsic property, perhaps related to their 332 specific biogenesis pathway. Lastly, the sRNAs could associate with a specific 333 RNA-binding protein, as has been shown in some mammalian systems. Our 334 immunoprecipitation data suggest that associative binding occurs for the siRNAs 335 and we identify exWAGO as the mediator of this selective export. Intriguingly 336 exWAGO is highly conserved and abundant in all Clade V parasitic nematodes 337 examined, and we have further shown that it is also secreted in the rodent parasitic 338 nematode N.brasiliensis. We propose therefore that the mechanism of exWAGO-339 mediated siRNA export extends beyond the H. bakeri model.

340
The sRNAs selected for export with exWAGO derive from regions of the H. bakeri 341 genome that are repetitive and novel, which may reflect recent, dynamic evolution 342 of this putative host manipulation system. Rather than derive sRNAs from 343 conserved loci, and risk self-directed effects, selection may exploit the rapidly 344 evolving non-genic portion of the genome to generate evolutionarily novel but host-345 relevant sRNA loci. It will be informative to correlate these sequences with host 346 genes and to explore their evolution across parasites.

354
Very little is understood regarding the evolution of cross-species communication.

355
That pathogens use small RNAs to modulate their hosts is not unexpected, as it

Figure 3
Expression relative