Intestinal NF-κB and STAT signalling is important for uptake and clearance in a Drosophila-Herpetomonas interaction model

Dipteran insects transmit serious diseases to humans, often in the form of trypanosomatid parasites. To accelerate research in more difficult contexts of dipteran-parasite relationships, we studied the interaction of the model dipteran Drosophila melanogaster and its natural trypanosomatid Herpetomonas muscarum. Parasite infection reduced fecundity but not lifespan in NF-κB/Relish-deficient flies. Gene expression analysis implicated the two NF-κB pathways Toll and Imd as well as STAT signalling. Tissue specific knock-down of key components of these pathways in enterocytes (ECs) and intestinal stem cells (ISCs) influenced initial numbers, infection dynamics and time of clearance. Herpetomonas triggered STAT activation and proliferation of ISCs. Loss of Relish suppressed ISCs, resulting in increased parasite numbers and delayed clearance. Conversely, overexpression of Relish increased ISCs and reduced uptake. Finally, loss of Toll signalling decreased EC numbers and enabled parasite persistence. This network of signalling may represent a general mechanism with which dipteran respond to trypanosomatids.

Introduction Neglected Tropical Diseases (NTDs) like sleeping sickness, leishmaniasis, hookworm infections, river blindness and elephantiasis are the most common infections of the world's 1.4 billion poorest people and the leading causes of chronic disability and poverty [1,2]. NTDs are found mostly (but not only see [1]) in low and middle income countries [3]. For NTDs communicated to humans through an insect vector, the ability of the pathogen to overcome the insect's midgut defenses is absolutely central to transmission. This is clearly illustrated by the following examples.
African trypanosomes, responsible for sleeping sickness and nagana, encounter a severe barrier to their establishment in the midgut of their tsetse fly vectors [reviewed 4]. It has been shown that there is increased resistance to Trypanosoma brucei spp (T. brucei spp) from their first blood meal where 50% of T. brucei spp become established, to their third blood meal onwards (the fly may take 40-60 blood meals in its life) where less than 10% of challenged flies become infected [5]. Paradoxically therefore, given their importance as vectors, tsetse fly populations are overwhelmingly resistant to trypanosome infection and the resistance mechanisms are manifested largely in the fly midgut [reviewed in 6,7].
Leishmania parasites seem to have successfully overcome barriers to establishment in their sand fly hosts as they develop in large numbers in the midgut of challenged laboratory strains. Nevertheless, in the wild there is only 1% of caught sandflies infected with Leishmania indicating that most infected sandflies are able to clear the parasite [reviewed in 8]. In the case of filariasis, the numbers of microfilaria ingested by all vectors (black flies, mosquitoes etc.) decline dramatically in the midgut lumen with either none or only a small fraction managing to penetrate the midgut barrier. Again, permissiveness of the mosquito midgut for parasite invasion is a key factor in determining success of the infections [9] but we understand little at the molecular level of the mechanisms involved.
In mosquitoes also, establishment of the malaria parasite Plasmodium falsiparum is dependent on overcoming midgut defenses [10]. Unlike mosquitoes however, where technological development has been rapid, for some dipteran vectors challenged with kinetoplastid parasites the "tool box" required to tease out these interactions is being developed rather more slowly. For example, there is no realistic prospect of producing transgenic technology for tsetse flies because eggs are inaccessible due to intrauterine development of larvae; there is currently no transgenic technology for sandflies; maintenance of multiple lines of both flies permitting genetic studies is costly and complex; bioinformatics resources are in their infancy. In this context, the model dipteran insect D. melanogaster may be able to give answers on the possible existence of an evolutionary conserved component of the dipteran host response to kinetoplastid parasites.
Like all insects, Drosophila possesses a sophisticated antimicrobial defense. This is rapidly activated upon immune challenge by NF-κB-like transcription factors through two distinct signaling cascades, namely the Toll and IMD pathways. Sensing of β-1,3-glucan of fungi and peptidoglycan of Gram-positive bacteria principally trigger Toll signaling. This activation centers on the transmembrane receptor Toll, which is activated by the endogenous ligand Spz, a Nerve Growth Factor homologue [11]. Signal transduction through a receptor-proximal complex including Myd88, Tube and Pelle culminates in the proteolysis of the Drosophila IκB Cactus, which enables the translocation to the nucleus of the NF-κB homologue DIF [12]. Moreover, peptidoglycan from Gram-negative bacteria and Gram-positive bacilli primarily induce the IMD pathway, homologous to the TNFR1 pathway. Upon recognition by Peptidoglycan Recognition Proteins PGRP-LC and PGRP-LE, the signal is transmitted to Imd itself (a RIP-1 homologue) and then to TGF-β Activating Kinase 1 (TAK1), which phosphorylates and thus activates, the fly IκB Kinase (IKK) complex [13]. In turn, IKK phospholylates and the caspase-8 homologue Dredd cleaves the composite NF-κB/IκB Relish transcription factor, releasing the N-terminal DNA-binding part of the protein to translocate to the nucleus [14] and regulate hundreds of genes including several antimicrobial peptides (AMPs) [15]. In addition, the Drosophila Janus Kinase/Signal Transducer and Activator of Transcription (JAK-STAT) pathway has been implicated in defenses against viruses as well as a stress-response mechanism utilizing a set of core signaling components [reviewed in 16]. A transmembrane receptor encoded by domeless (dome), a single JAK tyrosine kinase encoded by hopscotch (hop), the transcription factor stat92E and unpaired (upd), as well as two related ligands encoded by upd2 and upd3. Binding of Upd ligands to the Dome receptor leads to activation of Hop, which phosphorylates itself and Dome. Cytoplasmic Stat92E can bind to phosphorylated Dome/Hop complexes. Once bound to the Dome/Hop complexes, Stat molecules are phosphorylated and forming Stat dimers that will translocate to the nucleus and regulate target genes [16]. Both Imd and JAK/STAT pathway are involved in systemic as well as in epithelial immunity, especially gut epithelial immunity. There, the Drosophila midgut contains pluripotent intestinal stem cells (ISCs) that have a simple lineage: each ISC divides asymmetrically to produce itself and a transient enteroblast (EB), which will undergo terminal differentiation into either an polyploid absorptive enterocyte (EC) or as a diploid secretory enteroendocrine cell (EE) [17,18]. Individual ISCs are scattered along a thin layer of basal lamina in the posterior midgut and are the only proliferating cells in the epithelium [17,18]. This proliferation can be marked with an antibody against the phosphorylated form of Histone-3 (PH3 + cells) [19,20].
Very few studies of kinetoplastid interactions with Drosophila have been published. One biochemical study has looked at AMP production in response to infection with Crithidia spp [21]. In addition to not being a natural parasite for Drosophila, Crithidia largely infect the rectum of flies and are not a good model for midgut vector-parasite interactions. Nevertheless, natural gut-dwelling kinetoplastid parasites of Drosophila do exist [22]. A potential model system of greater relevance is Herpetomonas ampelophilae, a natural kinetoplastid parasite of D. melanogaster, which establishes infection in the midgut of the fly and can go on to invade the salivary gland [23,24]. However, there are no studies that have examined the interaction between the adult D. melanogaster midgut and Herpetomonas beyond those initial papers. A recent study, has described the interaction between Drosophila falleni and Jaenimonas drosophilidae, a novel natural trypanosomatid parasite isolated from the wild [25]. There, D. falleni larvae were persistently infected throughout development, demonstrating persistent infection. However, there was a pronounced bottleneck in infection over metamorphosis with substantially lower rates of Infection in adults than in larvae (Hamilton et al, 2015). Using the rate of initial infectivity as a measure, these authors showed that J. drosophilidae infection in D. melanogaster larvae provoked an immune response that was not dependent on the IMD pathway [25].
In the present work we have developed a Drosophila-Herpetomonas system to be able to dissect insect-parasite interactions in a model dipteran insect. Using transcriptomics as well as tissue-specific RNAi assays, we have pinpointed the activation of the immune pathways responsible for gut defences upon parasite infection and documented the role of Toll, IMD and JAK-STAT signalling in modulating the number and time of parasite clearance. Linked to IMD-Relish, the timing of ISC proliferation had a pivotal role in the ability of the fly to clear the parasite quickly and to keep numbers down. Furthermore, our results provide a framework to establish the evolutionary conserved component of the response of dipteran insects against kinetoplastids.

A natural parasite for Drosophila melanogaster
We have isolated trypanosomatid parasites from D. melanogaster caught in the wild on fruit baits in and around Oxford, UK (see materials and methods). Following sequencing of 18S rRNA we concluded that these parasites belonged all to the same species namely, Herpetomonas megaseliae (99% sequence identity with strain accession number U01014) [26]. However, through the molecular redefinition of phylogenetic relationships in the Herpetomonas genus, Herpetomonas megaseliae is now included in Herpetomonas muscarum and henceforth we will call the isolated parasite H. muscarum [27]. In our hands, the prevalence of the infection in the wild was 5.01% in males (or 16 out of 212 male flies caught and identified as D. melanogaster), in accordance with reports of more extensive samplings [28]. The prevalence in females was approx. 6% (or 5 out of 91). Following species identification, males were assayed directly for the presence of the parasite and were used to isolate it. Female flies were kept to establish separate lines and verify inect species identification (see materials and methods).
We were able to culture the parasite (see materials and methods; Fig 1A) and therefore wanted to transfer this host-parasite system in the lab to determine various aspects of this interaction. Pioneering work by Rowton and McGee has previously characterized infection of Drosophila laboratory populations by Herpetomonas ampelophilae [23,24]. However, the experimental design did not preclude the infection of uninfected flies at different time points after the initial infection event. We therefore began by characterizing the likelihood, time course and fitness consequences of H. muscarum infection in laboratory D. melanogaster after a single exposure event. The infection protocol and scheme of time point sample collection for downstream experiments is illustrated in S1A Fig.

Effects of Herpetomonas infection on Drosophila survival and life expectancy
We were able to record 100% infection rates in D. melanogaster laboratory flies (Oregon R and w 1118 ) fed with log-phase (72h) H. muscarum in 10% sucrose (S1B Fig). In this and all subsequent experiments, control flies are those fed with just 10% sucrose. Parasite presence was confirmed by visual inspection using live stains in the posterior (6h post-infection; S1C Fig) and anterior midgut (24h post-infection) (S1D Fig). These positions were reminiscent of the "swim back" of Leishmania towards the mouthparts in sand flies after digestion of the blood meal [8]. The possible attachment of Herpetomonas to the intestinal epithelium from the inside of the peritrophic matrix was also documented using infection with pre-stained parasites (S1C Fig). Finally, the dynamics of infection was followed using quantitative real time PCR in reference to a standard curve that was made each time from a fresh parasite culture used in that specific infection experiment (S1E Fig). Infection dynamics showed that Oregon R and w 1118 flies were able to clear infection within 4-5 days (Fig 1B).
Parasite infection reduced egg-laying of w 1118 and w 1118 ; relish mutant flies ( Fig 1C) as well as Oregon R . This suggested a reduction in fecundity that was strategically placed to divert resources to parasite clearance [reviewed in 29]. Connected to the reduced energy spent for egg laying, Oregon R flies showed a statistically significant increase in their median lifespan when infected with the parasite compared to controls (Fig 1D) while lifespan of infected w 1118 flies was statistically indistinguishable from controls ( Fig 1E). In germ-free conditions, w 1118 flies exhibited a higher initial parasite uptake (6h post infection) but later inhibited parasite infected with H. muscarum was statistically indistinguishable compared to sucrose-only fed controls either conventionally reared or germ free. (F) In the absence of gut microbiota, more parasite intake was observed during the first 6 hours of infection compared to conventionally reared flies. However, at 18h post-infection germ-free flies exhibited significantly reduced parasite numbers. Two-way ANOVA was used to analyse all data. For fecundity assays, statistical difference was observed on D3 between w 1118 sucrose fed controls and proliferation, especially at 18h post-infection ( Fig 1F). Overall however, there was no difference in the lifespan of germ-free w 1118 when fed sucrose only (control) or parasite in sucrose ( Fig  1E).
Infected flies were not able to pass the parasite to non-infected adults when infected and non-infected flies were co-cultured in cages. This result showed that the parasite was able to survive inside one fruit fly host but could not subsequently transfer to an uninfected fly in contrast to H. ampelophilae [23]. In the latter case, transmission to adults and larvae was probably accomplished by feeding on substrates contaminated with host feces or "social digestion" of adult or larval cadavers by developing larvae. Our results however, leave open the possibility that flies killed all parasites prior to clearance or that an intermediate plant host is necessary for transmission, as some Herpetomonas species have been proposed to be associated with tomato plants [23]. Fast clearance and/or an intermediate plant host may also explain the low prevalence of infected fruit flies in the wild [30].

Transcriptomic analysis of host response to parasite infection
We next sought to determine gene expression that was specifically altered by the presence of the parasite. We investigated transcriptome variations in whole, sucrose-only fed and sucroseparasite fed flies following the same infection protocol shown in S1A Fig. Transcriptome data were generated using the Illumina RNA sequencing platform of sequencing poly-A RNA (thus avoiding bacterial contaminants) to capture both known and novel coding and noncoding RNA (see materials and methods). Taking into to account that the process of parasite clearance took about 105 hours (see Fig 1B) we followed infection dynamics of 4-day old flies from early time points (6, 12, 18h post infection), to 54h (as a mid-point) and finally to day 7, a time point after parasite clearance had been achieved. There were 1,556 genes that were significantly regulated following parasite infection (P<0.05). From these, our analysis identified 155 genes whose expression varied by at least a 2log fold change relative to expression in 10% sucroseonly fed flies (Fig 2A). The statistical confidence P-value was established in each of the three biological repeats. During the lifetime of infection those 155 genes included 66 upregulatedonly transcripts, 56 downregulated-only transcripts and 33 (or 20%) that were upregulated in certain time points while downregulated in others. In terms of gene expression, the majority of the transcripts were differentiated at the initial stages of the response (98 and 79 at 6h and 12h following infection) while as the response progressed there was less differential transcription (39 transcripts at 54h and 13 at Day 7, see Fig 2B).
Using a global classification of gene ontology (GO), nearly a quarter (23.8%) of the genes were assigned as "unknown function" (11.6% upregulated-only, 8.38% downregulated-only and 3.82% both). Of those, the ones with at least one homologue could be divided into three main categories namely, intracellular signalling molecules, collagen-like cuticle proteins and genes with unknown function (see full data set and all gene expression tables at the European Nucleotide Archive website, project accession number PRJEB30020; https://www.ebi.ac.uk/ ena/data/view/PRJEB30020). The rest of the transcripts were assigned to 17 functional categories ( Fig 2B). These included digestion related to proteolysis and lipid metabolism, oxidative stress responsive and serine protease inhibitors (all upregulated mostly at 6h post-infection), metal-ion metabolism (upregulated at 54h but suppressed at 18h and Day 7 post-infection), antimicrobial peptides (AMPs) and other known immune-related genes (including known pathways and stress responsive genes; gradually upregulated through to the mid-point at 54h w 1118 infected flies and on D3 and D4 between Rel E20 sucrose fed controls and Rel E20 infected flies. On D4, w 1118 sucrose fed controls, w 1118 infected and Rel E20 sucrose fed were statistically indistinguishable and significantly different from Rel E20 infected flies. https://doi.org/10.1371/journal.pgen.1007931.g001 post-infection), mucus/chitin synthesis related and cytoskeletal genes (upregulated at 54h and Day 7 but suppressed at earlier time points), neuropeptides (suppressed at all time points) and micro-RNAs (activated at 6h, 12h and 18h but suppressed at 54h and Day 7). Based on this analysis we concluded that parasite infection triggers high levels of transcriptional signatures of signalling associated to immune, stress and metabolic responses. In this work, we study the involvement of immune response genes.

Immune pathways related to Herpetomonas infection in Drosophila
We found that following parasite infection, there were several differentially expressed AMP genes that were targets of the IMD pathway, genes that coded for components of the oxidative stress response as well as the Toll receptor itself. Therefore, we addressed the functional relevance of these findings using the GAL4-GAL80 ts -UAS-RNAi system where we could knockdown genes in adulthood in a ubiquitous or tissue specific manner avoiding any developmental or long-term effects. At the permissive temperature (18˚C) GAL80 ts prevents GAL4 binding and therefore RNA-interference, whereas shifting to the restrictive temperature (30˚C) inactivates GAL80, releases GAL4 and thus induces RNAi [31]. We tested that the system was effectively induced using three GAL4 lines (see

Antimicrobials
We found that silencing the gene coding for Dual Oxidase, the enzyme responsible for ROS synthesis in ECs [32], significantly increased initial loads (6h post infection) and delayed parasite clearance (Fig 3A). Similarly, silencing caudal loss of which de-represses all Relish-regulated AMPs in ECs [33], significantly decreased parasite uptake ( Fig 3B). Conversely, silencing the AMPs Diptericin ( Fig 3C) or Cecropin (Fig 3D) increased uptake and delayed clearance while overexpressing Attacin had the opposite effect ( Fig 3E). These results are consistent with a role for ROS as well as AMPs (especially those transcriptionally regulated by Imd/Relish) in initial uptake and clearing infection.

Relish
Consistent with the results above, Rel E20 flies displayed significantly higher parasite numbers at early time points (6h, 12h) compared to controls (Fig 4A). In exploring the tissue-dependency for Relish in the control of parasite numbers, we found that silencing Relish in progenitor cells (ISCs and EBs) with esg-GAL4 had a significant influence in parasite numbers throughout the lifetime of the infection and delayed parasite clearance by 48h (Fig 4B). Conversely, overexpression in ISCs and EBs of a form of Relish that constitutively localises inside the nucleus [35], significantly decreased initial parasite loads and accelerated clearance with most flies clearing Herpetomonas at day 1 (Fig 4C). Therefore, Relish was necessary and sufficient to control parasite numbers in progenitor cells. Finally, silencing Relish in ECs   In contrast, a modest increase in parasite number intake was observed when Relish was blocked in ECs while (E) no effect when Relish was blocked in the fat body was observed. Two-way ANOVA was used to analyse all data ( � p<0.01, �� p<0.001, ��� p<0.0001).

Toll
Next, we wanted to identify the role of the Toll receptor, which was differentially expressed in parasite-infected flies. Silencing Toll in all immunocompetent tissues significantly increased initial and early parasite load (6h to 1 day post-infection) (Fig 5A). In contrast to Relish however, we found that Toll was not required in progenitor cells (Fig 5B) but in ECs (Fig 5C). In addition, silencing Toll expression in the fat body (using yolk-GAL4) indicated that Toll activity was also important there but only for controlling initial parasite uptake (Fig 5D). To corroborate the role of the Toll pathway in controlling parasite numbers, we tested dif 1 , a mutant in DIF [11]. We found that dif 1 mutant flies displayed a significant increase in the initial Herpetomonas uptake (Fig 5E). Moreover, dif 1 flies were not able to clear the infection at D3 in comparison to their yw genetic background (Fig 5E).

STAT
Targets and signaling components of the JAK-STAT were also differentially regulated following parasite infection. As the pathway has been implicated in gut physiology and immunity [16], we investigated the effect of silencing STAT in ECs, progenitor cells, hemocytes and fat body. Using J6-GAL4 we found that silencing expression of the Stat transcription factor (Stat92E) increased initial parasite numbers (6h post infection) and delayed clearance ( Fig  6A). However, there was no requirement for STAT in ECs (Fig 6B) or the fat body ( Fig 6C). However, silencing Stat in progenitor cells resulted in a significant difference in both parasite numbers in the early phase of infection (6h to day 1) and delay in clearance at day 3 ( Fig 6D). This result, along with the fact that Stat has been shown to be expressed in ISCs (see below) indicated that Stat activity was required in progenitor cells to control parasite numbers and clearance.

Parasite infection increases Stat-mediated transcription and proliferation of ISCs
The involvement of JAK-STAT signalling in ISC proliferation and intestinal regeneration upon Upd cytokine secretion from ECs, has been documented previously [19,34]. Using multimerized Stat92E consensus binding sites, which control expression of destabilised-GFP (10x-STAT-GFP) as a reporter of JAK-STAT signalling activity [36], our results indicated the involvement of STAT in parasite infection. Following parasite feeding, GFP-expressing cells were transiently increased early, at 6h and day 1 post infection compared to the sucrose-only control but there was no difference at day 3 ( Fig 7A). This correlated with a significant increase in proliferative cells measured with an antibody against phospho-histone-3 (PH3) in w 1118 flies (Fig 7B quantification). Moreover, there was persistent increase in GFP-positive progenitor cells (6-30h) when an esg-GAL4, UAS-GFP strain was infected with Herpetomonas ( Fig 7C).

ISC proliferation and Herpetomonas infection
Given the above, we next explored the signaling underlying proliferation of ISCs following parasite infection. When w 1118 were infected with H. muscarum, PH3 + cells were significantly increased continuously compared to the sucrose control, starting from 6h (Fig 8A), day 2 ( Fig  8B) and day 3 (Fig 8C). In contrast, sucrose-only fed controls showed a stable number of PH3 + cells in both 6h ( Fig 8A) and day 2 (Fig 8B) only increasing at day 3 ( Fig 8C). The number of PH3 + cells in sucrose-only fed imd R156 was significantly higher than w 1118 controls indicating a de-repression of progenitor cell proliferation in the absence of Imd (Fig 8A). Parasite infection significantly suppressed this phenomenon at 6h post-infection indicating a strategy from the parasite's side to achieve midgut establishment (Fig 8A). Nevertheless, this suppression was not evident at day 2 ( Fig 8B) or day 3 (Fig 8C). At these time-points, the numbers of PH3 + cells between infected and sucrose-fed imd R156 were statistically indistinguishable. Intestinal proliferation in rel E20 mutant flies (sucrose control) was also significantly increased compared to w 1118 controls but not as pronounced as imd R156 (Fig 8A). As with imd R156 , parasite infection of rel E20 significantly suppressed de-repression of intestinal proliferation (Fig 8A). The number of PH3 + cells in rel E20 infected intestines however, was kept significantly lower both at day 2 ( Fig 8B) and day 3 (Fig 8C) compared to both infected imd R156 and controls. This indicated an important and more general role for Relish in ISC proliferation compared to Imd. When silencing rel in progenitor cells, the majority of the GFP expressing cells in the guts of esg-CD8-GAL4, UAS-GFP; GAL80 ts flies were enlarged GFP-positive cells that have lost their proliferative capacity as no PH3 + was detected (Fig 8D). Conversely, overexpression of Relish increased the population of smaller GFP expressing ISC-looking cells, which correlated with an increase in proliferating PH3 + cells (Fig 8E). This PH3 + proliferation was suppressed by the parasite (Fig 8E).
Thus, independently of infection, Relish was sufficient for intestinal progenitor cell proliferation. High levels of Relish in the nucleus pushed cells to be PH3 + , with an ISC morphology whereas low levels of Relish provoked progression in the progenitor lineage to larger EBs. The role of Relish in proliferation of midgut progenitor cells and the fact that its loss resulted in sustained low numbers of PH3 + cells following infection, suggested an explanation for the delay in parasite clearance in rel E20 . In addition, STAT-mediated transcriptional activity correlated with ISC proliferation following infection and was important for parasite clearance. Finally, AMPs, Duox/ROS and Toll were required in ECs (Fig 8F).
Nevertheless, even in the absence of Relish the parasite was eventually cleared. This suggested that gut epithelium renewal was just one of the factors influencing parasite clearance and that most likely "flushing" of the parasite through the mechanical contraction of the gut was at play. Clearance by "flushing" would be possible if the parasite was failing to attach to the gut epithelium. More work is needed to define a possible interaction interface for this attachment.
Herpetomonas infection induced an acute early increase in ISC proliferation in w 1118 control flies as seen with PH3 + quantification ( Fig 8A). However, this was suppressed in immunedeficient rel E20 or imd R156 flies even though these flies displayed a higher than the control ISC proliferation in sucrose-only treatment (Fig 8A). This suggested that the parasite actively suppressed proliferation of ISCs and epithelial renewal as a means of establishing a stable midgut presence and the Imd/Relish pathway was paramount to resist this strategy. This has been also observed in Vibrio cholera [37] and Pseudomonas entomophila [38] intestinal infections where delamination of ECs is an important way with which "shedding" of cells removes the pathogen as well. It would be interesting to see whether suppression of ISC proliferation happens in the intestines of tsetse infected with trypanosomes or sand flies infected with Leishmania. Intriguingly, PH3 + suppression following infection was most pronounced in rel E20 where the number stayed significantly lower than the control for the duration of our experiment (day 3, Fig 8C). This was in contrast to imd R156 where numbers were significantly low early on (6h) but were indistinguishable to controls from day 1 onwards (Fig 8B and 8C). This suggested a more important role for Relish than Imd in ISC proliferation and does not conform to a linear pathway hypothesis. Therefore, Relish may regulate ISC proliferation by integrating signals from other signalling networks in addition to Imd.

Conclusions
We established for the first time, a framework to systematically study how the model dipteran insect Drosophila, responds to a trypanosomatid parasite infection. We showed that Relish, STAT, Toll, Duox and AMPs were all important for fast clearance and control of parasite numbers. Our results provide cellular context to similar data from tsetse where RNAi of Relish or of an AMP homologous to attacin increased T. brucei numbers in infected flies [39]. However, more work is needed to ascertain whether the Drosophila-Herpetomonas system could be used  Day 1 (B) and Day 2 (C) post-challenge; the parasite was able to suppress ISC proliferation in the absence of Relish (6h, D1 and D3) and to a lesser extend Imd (6h). (D) Relish also played a role in the proliferation of ISCs as loss of Relish blocked ISC proliferation (PH3-red channel) following parasite challenge while (E) Relish overexpression induced ISC proliferation even in the absence of the parasite (see PH3 in sucrose control) (F) A cartoon illustrating a working model of the H. muscarum oral infection model in Drosophila. Both STAT and Relish transcription factors are essential for ISC proliferation and differentiation whereas Toll signaling as well as ROS generation (through Duox) is important in ECs. Parasite infection can also induce Relish-dependent transcription as seen in the upregulation of IMD controlled AMPs. Our hypothesis is that the combinatorial effect from STAT-mediated signaling and Relish-dependent transcription encourages a hyper proliferation of ISCs which eventually results in a fast epithelium turn over in the Drosophila gut. Such a mechanism, together with the expression of effector molecules like AMPs and ROS would be able to remove the parasite quickly and efficiently. Two-way ANOVA was used to analyse all data ( � p<0.01, �� p<0.001, ��� p<0.0001).
https://doi.org/10.1371/journal.pgen.1007931.g008 to address questions in insect vectors (e.g. tsetse, sand flies) where the tool-box available for direct functional studies is not yet fully developed.

Drosophila collection and species identification
We collected Drosophila in a residential area using the corresponding author's back garden and compost tip (inside the garden) as places to trap fruit flies. As such, no consent was necessary. Collection was done through 5-day periods from late March to late June 2011. During this period, traps containing fermented banana were set every week. The nine species of the melanogaster species subgroup are morphologically very similar but male genitalia is a reliable characteristic to distinguish males [28]. For females we used as a discriminatory characteristic the area below the eye (cheek) along the long axis, which is broader in D. melanogaster than all other members of the group [40]. To confirm our identification following establishment of separate stocks from a single female founder we randomly selected flies identified morphologically and used a PCR diagnostic test for the antimicrobial peptide gene drosomycin as described [22]. Species identification was confirmed with first generation offspring.

Drosophila stock
Single trapped females were isolated in vials and their offspring cultured as separate "isofemale" lines to verify species identification [28]. Trapped males were used to isolate and culture the parasite. Starting from a single cross (one female-one male), Oregon R flies were used as a wild type laboratory strain, for establishment of laboratory oral feeding infection protocols and transcriptomics experiments thus ensuring a less variable, streamlined genetic background. w 1118 (BL #6326) and yw 67c23 (BL #6599), were used as controls and for the genetic background of all the other strains used in these studies and were obtained from the Bloomington Stock Centre. IMD, Toll and STAT signaling pathway mutants: yw 67c23 ,Dredd B118 (BL #55712); w 1118 ;relish E20 (BL #55714) and y 1 w � ; Dif 1 cn 1 bw 1 (BL #36559) were also obtained from the Bloomington Stock Centre. The GAL4 driver lines used were: w 1118 ; np1-GAL4 (ECs) was kindly provided by Heinrich Jasper (Buck Institute for Research on Aging, Novato, CA, USA), yolk-GAL4 (female fat body), w 1118 ; P{GawB}c564 (J6-GAL4, expressed in fat body, gut, hemocytes, (BL#6982) and esg-GAL4 (expressed in ISCs and EBs) was obtained by Bruno Lemaitre EPFL, Lausanne see ref [19].

Culture of H.muscarum and EM
The isolated H.muscarum used in the study was routinely cultured in 10% FBS (Gibco) supplemented Drosophila Schneider's-2 (S2) media (Sigma-Aldrich). The parasite culture was maintained by sub-culturing every 3 days at 1:100 dilution to a fresh 5 ml S2 media. To count the parasite cells, cultures were first spun down at 2,000rpm for 5 mins and the cell pellets were treated with 10% methanol for 15 minutes to kill and immobilize the parasite cells. The cells were subsequently precipitated by spinning at 2,000rpm for 5 mins and washed twice by 1xPBS. After washing, cell pellets were resuspended in 1XPBS and diluted by 100 times in PBS before being transferred to a hemocytometer to count under a bright field light microscope at 100x magnification.

TEM Microscopy of H. muscarum cells
5ml of culture (10 6 cells/ml) was fixed using 2ml of 16% PFA (w/v) and 0.8ml of 25% glutaraldehyde (v/v) at 28˚C. Cells were pelleted at 10000rpm for 5min and the fixative removed. Cell pellets were washed twice in 1ml 0.1M PIPES pH 7.2 and once in 1ml 0.1M PIPES pH 7.2 with 50mM Glycine. The supernatant was removed, and the pellets embedded in 0.5ml 2.5% agarose (w/v). Embedded pellets were fixed in 1% OsO 4 for 1h at 4˚C, washed three times in 2ml milli-Q water and then incubated overnight in 0.5% Uranyl acetate in milli-Q water at 4˚C. Following this, cell pellets were washed for 10 mins in 2ml of milli-Q water at room temperature on a rotor before a 30min incubation in 30% EtOH on ice. The pellets were then dehydrated by immersion in increasing concentrations of 2ml EtOH (50%, 70%, 80%, 90% and 95%) for 10 mins each on ice. The three final dehydrations were done for 30min in 2ml 100% EtOH. Pellets were then incubated in 2ml of the following mixes of (v/w) ethanol: agar100 resin; 3:1 (1h with rotation), 1:1 (2.5h with rotation), 1:3 (1h with rotation). Pellets were then incubated in 2ml 100% agar100 resin for 24 hours with the resin replaced with fresh resin at 18 hours and 22 hours. Finally, pellets were transferred into the base of beem capsules (Agar Scientific) and the resin set at 60˚C for 18 hours. Images were taken using the FEI Technai 12 Transmission electron microscope with the Gatan Ultrascan 1000 CCD camera and Gatan Digital micrograph and SerialEM image acquisition platforms.

Oral feeding infection and quantification of parasite by real-time PCR
For each independent infection of a group of 20-30 flies, 10e7 H. muscarum parasite cells were harvested from a 3 days-old culture (which showed the highest infectivity rate from our experience) and resuspended in 500ul 1% sucrose. The parasite solution was then transferred to a 21mm Whatman Grade GF/C glass microfibre filter circle (Fisher Scientific). Circles containing the parasite cells were placed into standard Drosophila small culture vial without any food. The flies used in the infections were 4-5 days old before they were starved overnight. After starvation, the flies were transferred to food vials that contained the Whatman circles with the parasite cells. After 6h of feeding, flies were moved and reared on standard yeast/molasses medium. At different time points post oral infection (including the 6h feeding as 6h point), infected flies were collected for downstream experiments.
An absolute quantification of real-time PCR based method was developed to quantify the parasite numbers in the flies. This was done every time an experimental infection was carried out as follows. H.muscarum parasite cells from 1ml of the 3 days old culture used for the infection, were harvested and genomic DNA (gDNA) was isolated. The purified gDNA was serially diluted and was used as template in RT-PCR reactions to obtain linear regression of gDNA verse Ct values using H. muscarum gene specific primers. The serial dilution of parasite gDNA was then correlated with the serial dilution of the absolute number of parasite cells from the same culture. Hence a standard curve of the parasite numbers versus the Ct values was established as shown (S1E Fig). Isolation of gDNA from infected whole flies was also carried out and used for RT-PCR measurements. We did not use any uninfected fly DNA (only sucrose fed controls) for RT-PCR as the primers were parasite gene specific and would not pick up any PCR products from uninfected flies. For each individual fly or group of flies infected, gDNA was exacted (containing both fly and parasite gDNA). By measuring the Ct value of the gDNA sample, the absolute number of parasite cells could be extrapolated from the standard curve.
The gDNA of the cells was extracted using cells and tissue genomic DNA isolation kit from Norgen Biotek no 53100 following the manufacturer's instruction. H. muscarum gene specific primers used in the RT-PCR was designed based on the paraflagellar rod protein (PFR2) gene (GenBank accession no: AY785780). The primer sequences were: HmegRodF 5'-GGACTGC TGGAACAAGATC-3', HmegRodR 5'-AGCTTCTTGTGCTGGGAG-3'.

RNAseq and data analysis
Total RNA of 8-10 flies at 6h, 12h, 18h, 30h, 54h and D7 (S1A Fig) and post H. muscarum oral infection was extracted with total RNA purification kit from Norgen Biotek no. 17200 following the manufacturer's instruction. Triplicate of samples were used for each time point. cDNA libraries were prepared with the Illumina TruSeq RNA Sample Prep Kit v2. All sequencing was performed on Illumina HiSeq 2000 plaftform using TruSeq v3 chemistry (Oxford Gene Technology, OGT). All sequence was paired-end and performed over 100 cycles. Read files (Fastq) were generated from the sequencing platform via the manufacturer's proprietary software. Reads were processed through the Tuxedo suite [41]. Reads were mapped to their location to the appropriate Illumina iGenomes build using Bowtie version 2.02. Splice junctions were identified using Tophat, version v2.0.9. Cufflinks was used to perform transcript assembly, abundance estimation and differential expression and regulation for the samples (version 2.1.1). RNA-Seq alignment metrics were generated using Picard.
At each corresponding time point, total RNA from triplicate of sucrose fed controls was extracted and the transcriptomes were sequenced and analyzed in the same manner for the subsequent comparison and data analysis. Differential expression of transcripts from three paired triplicates was statically analysed by Cufflinks. We have selected transcripts that showed log2 fold change below -2 or above +2 of absolute FPKM values (Fragments Per Kilobase of transcript per Million mapped reads) compared with sucrose fed samples to generate the heat map. An average of 26280113 paired-end reads were sequenced per sample. A total of 75.73 Gigabases (788403397 reads) of sequence data were read and aligned at high quality. The entirety of the RNA-seq data along with all the gene expression analysis tables can be found at The European Nucleotide Archive (ENA) with accession number PRJEB30020.

Lifespan monitoring
Life span of flies infected or sucrose fed controls was monitored by counting the number of live flies on a 3-day interval until the last single fly in the culture died. During that time live flies were put in fresh food every 2 days.

Generation of germ-free flies and RT-PCR to measure dynamics of gut microbiota
Germ free flies were generated following standard protocols [19]. Briefly, 200-300 embryos were collected from fly cages fitted with apple juice plates. The embryos were then treated with 50% bleach for 2-3 minutes until all embryos were dechorionated when observed under a stereoscope. Subsequently, embryos were washed twice with 70% ethanol followed with autoclaved miliQ water to remove residual ethanol. Treated embryos were then transferred to vials with standard yeast/molasses medium. Both vials and food were autoclaved forehand. From then on, aseptic procedures were employed in handling of vials with flies.

Dissection of fly guts and immune-staining
Dissection of the fly guts and immune staining was generally following a standard protocol as in [41]. Rabbit anti Phospho-Histone H3 (Ser10) (anti-PH3) and goat anti rabbit Alexa Fluor A568 secondary antibody were both obtained from Invitrogen.

Fluorescence microscope and image analysis
Images of the gut with GFP expressing cells, DAPI and anti-PH3 staining were obtained with either Zeiss Axioplan 2 (Carl Zeiss) or Zippy DeltaVision Elite (Applied Precision). Image J was used to quantify individual cells from the stacks taken by Zippy DeltaVision Elite. GFP expression cells and DAPI cells in the same designated region were counted separately and automatically. The relatively ratio of GFP expressing cells normalized by DAPI cell was used instead of absolute cell numbers to represent more accurately the change of the GFP expressing cells by eliminating factors such as physiological and anatomical change of the gut tissue in response to parasite infection and/or experimental handling.