Read mapping, de novo transcript assembly and differential expression analysis
Five to Nine-day old flies were fed blood or parasite-containing blood. This resulted in four treatment groups of P. papatasi females namely: a) blood-meal only, b) blood meal + L. major, c) blood meal + L. donovani and d) blood meal + Herpetomonas muscarum. Flies were left to develop an infection or digest their non-infected blood meal for 1, 4 or 9 days. Therefore, in terms of age, on the 9th day PBM sand flies would have been 14-18 days old. Across infections, we compared flies at the same timepoint (and so of the same age) unless we were comparing flies with the same infection.
RNA was purified from whole sand flies at 1, 4 and 9 days PBM, with the view to identify any systemic responses over and above tissue specific signals. Each time point included 20 flies and therefore 20 flies X 3 timepoints=60 flies per treatment group (as defined in a-d above) were processed. This experiment was independently repeated three times so that in the end 60x3=180 sand flies were processed for every treatment group in all three experiments (20X3=60 flies for every time point). The goal of our experimentation was to find if there was a difference between blood meal vs. infected blood meal. Therefore, the non-infected blood meal was our control.
As expected, only L. major was able to establish infection following blood meal digestion and defecation at day 4 PBM (Fig S1). 95% of these developed into heavy L. major infections at 8 days PBM while the other two parasites were cleared (Fig S1). At this time point, L. major parasites were found in both the anterior and thoracic midgut as well as the stomodeal valve (SV) (Fig S2). In mature Leishmania infections, the SV is forced open and becomes blocked with parasites embedded in the promastigote secretory gel, a viscous mixture of phosphoglycans secreted by the parasites. This opening of the SV is essential for colonization of the foregut and transmission by regurgitation [12].
The resulting reads sequenced and mapped against the P. papatasi genome (Ppapl1, vectorbase) [[17]]. The number of reads generated per sample ranged from 1.08-12.05 million reads with 69.7-79.3% mapping to the P. papatasi genome in each sample (Table S1). Upon visual inspection of read mapping using IGV [[22]] it appeared that over 20% of reads were mapping to regions which lacked annotated features. To include these potentially novel genes in our analysis we assembled de novo concordantly mapped read pairs (from all samples) into 16,025 transcripts. The assembled transcripts were then merged with the existing annotation of 11,834 transcripts, to give a final set of 18,592 unique transcripts (see Supplementary data files). This represents approximately 97.2Mb of P. papatasi transcriptome with an average transcript length of 4,190 bp. All reads were then counted against the final set of transcripts for differential expression analysis.
Principal component analysis (PCA) showed a high degree of difference between the fly transcriptomes at day 1 PBM and those at day 4 or 9 PBM (Figure 2) – with transcriptomes from days 4 and 9 PBM appearing similar. We also note that samples do not clearly group in accordance with trypanosomatid feeding status.
Differential expression associated with trypanosomatid presence in the bloodmeal
There are few differentially expressed transcripts which were specifically associated with trypanosomatids being present in the blood meal (Table S2). We found no significant difference in transcript abundances between blood-fed and L. major fed flies at any time point. Furthermore, we find in excess of twelve thousand genes for which we reject the hypotheses that expression has changed by 2-fold or more in either direction in pairwise comparisons (Wald test) between blood-only fed and blood+trypanosomatid fed flies (Figure 3 and Table 1).
We did however observe differential abundance for some transcripts after H. muscarum and L. donovani feeding compared to blood only fed control flies (Figs 1D and 1G respectively). There were significantly fewer transcripts for the gene PPAI009043, an orthologue to the D. melanogaster signalling protein Rho GTPase activating protein at 54D (RhoGAP54D), in flies fed H. muscarum than in blood fed controls at day 1 PBM (log2foldchange 1.13, p-adj = 0.005). The Aedes aegypti and Anopheles gambiae RhoGAP54D orthologues are upregulated in blood fed mosquitos compared to sugar fed controls [[23,24]]. Given this, and that this response was not seen after Leishmania feeding, this transcriptomic response may be H. muscarum specific. The biological significance of reduced RhoGAP54D transcription in this context remains unclear, however the protein is linked to epithelial morphogenesis during Drosophila development [[25]] and so may also play a role in the mature insect gut.
In L. donovani fed flies there were significantly fewer transcripts for the putative transporter TrpA1 (PPAI004036, log2foldchange 2.8, padj = 0.025) versus blood only fed flies at 9 days PBM. TrpA1 is more generally associated chemo- and thermo-sensing [[26,27]] in Drosophila, however a study by Du et al. 2006 links TrpA1 to the expulsion of food-borne pathogens by increased defecation and the DUOX pathway (discussed further below) [[28]]. Speculatively, reduction in TrpA1 transcripts after L. donovani feeding may hint at modification of host defensive pathways to promote survival. We also find significantly more CUFF.12679 transcripts (log2-foldchange 16.8, p-adj = 0.0001) in L. donovani fed flies than in blood-only fed flies. This novel transcript lacks conserved domains or sequence similarity to known Dipteran gene transcripts.
Direct comparisons between trypanosomatid infections yielded similarly few differentially expressed transcripts (Table S3). At day 1 PBM the only differentially expressed transcript between the three infections was that of trypsin 1 (PPAI010956, padj = 0.035) which was 2-fold enriched in H. muscarum fed flies compared to those fed L. donovani.
After defecation at around 4 days PBM it is thought only parasites able to establish in the ectoperitrophic space persist to develop mature infection [[16]]. Despite the differences in the infection outcome reported in laboratory infections across the three trypanosomatids [[14]], there were few differences in the host transcriptome at this critical time point. Two transcripts were found to be significantly differentially abundant - one corresponding to the PPAI000999 gene and the other a novel transcript CUFF.14170. Both transcripts were found at significantly higher levels (p-adjusted 0.04 and 2.27E-09 repectively, log2foldchanges of 4 and 18 respectively) in H. muscarum fed flies compared to those fed L. donovani. PPAI000999 encodes a protein predicted to bind to chitin (GO:0006030, GO:0008061 and smart00494). The novel transcript CUFF.14170 has no known conserved domains and BLAST searches against Dipteran sequences did not yield any significant hits.
The most variation between the three infections was found at 9 days PBM, where 6 transcripts were differentially expressed between Leishmania fed and H. muscarum fed flies. Flies fed L. donovani had significantly more transcripts for the previously discussed TrpA1 (PPAI004036), and significantly less for the putative zinc metalloprotease PPAI010164 and novel transcript CUFF.12679, than those fed H. muscarum. Flies fed L. major had significantly more transcripts for the hypothetical protein PPAI002947. Additionally, feeding with H. muscarum resulted in significantly more CUFF.14170 transcripts, a novel transcript from this study which lacks conserved domains, than both Leishmania infections (padj = 1.79E-09).
Overall, the above observations suggest that blood feeding status is the major source of transcriptional variation in these flies and not trypanosomatid infection. As such we further investigated transcriptomic changes after blood feeding alone in P. papatasi.
The Phlebotomus papatasi transcriptome after blood feeding
Shortly after blood feeding there were significant changes in transcription which may be a universal response to blood meal. The transcriptomes at day 1 PBM appeared very different to those at 4 (and 9) days PBM, with 12,289 significantly differentially regulated transcripts (Table S4). However, after defecation of the blood meal remnants the transcriptome was comparatively stable with 264 differentially regulated transcripts (4 vs. 9 days PBM, Table S5). Due to the large number of differentially expressed transcripts highlighted by these comparisons we first investigated transcripts whose log2 fold change was > 4 in either direction between timepoints. From this subset we were able to focus our analysis on a number of key genes and pathways which are discussed further below (Tables S6 and S7).
Early transcriptomic responses to blood meal ingestion are concerned with digestion, metabolism and immunity
Of the 217 transcripts differentially regulated > 4-fold between 1 and 4 days PBM, 197 transcripts were found to be comparatively enriched at day 1 PBM and 20 were comparatively enriched at day 4 PBM. Ninety-eight of these transcripts did not contain known conserved domains.
Transcripts for putative and known trypsins were one of the most highly represented groups differentially regulated between day 1 and day 4 PBM. We observed upregulation of 9 transcripts for putative trypsins and chymotrypsins – including the previously characterised chymotrypsins 1 (PPAI010833), chymotrypsin 3 (PPAI005023) and trypsin 4 (PPAI010456)[8,29,30]. We also observed upregulation of transcripts which may represent novel trypsins, based on conserved domains and similarity to other Dipteran trypsin/chymotrypsin sequences, as they are not included in the current genome annotation (Ppap v1) [[17]] (CUFF.11666, CUFF.9493, CUFF.6542) and chymotrypsins (CUFF.15058, CUFF.16005, CUFF.15086, CUFF.14587, CUFF.12454). In contrast, the transcript putatively encoding for trypsin 1 (PPAI010956) was shown to be enriched at day 4 PBM compared to the earlier timepoint. The roles of trypsin and chymotrypsin-like serine proteases during blood digestion in hematophagous insects are well characterised with expression levels varying according to type and the time since the last blood meal. Our findings agree with previous work, which showed upregulation of trypsins 3/4 and chymotrypsin 1 in response to the blood meal, as well as the decrease of trypsin 1 [[30]].
In addition to the trypsins themselves, five transcripts whose products are predicted to contain trypsin inhibitor like domains (PPAI003932, PPAI000270, PPAI000272, PPAI000274, PPAI003557) were also comparatively enriched at day 1 PBM (vs. day 4 PBM). It is possible the corresponding proteins play roles in the regulation of the trypsin 1 as well as other trypsins (e.g. trypsin 2), reported to be downregulated after blood feeding [[30]].
Several transcripts encoding for proteins with predicted serine protease/proteolytic activity, the sequences of which do not resemble trypsins/chymotrypsins were also comparatively enriched at day 1 PBM. These included two known genes (PPAI009419, PPAI009871) and three novel transcripts (CUFF.6132, CUFF.6133, CUFF.16132). Serine proteases are implicated in several other cellular processes including innate immune signalling – notably in Toll pathway activation [[31]] and the melanisation response [[32]]. The predicted protein for PPAI009419 shares approx. 51% identity with the Culex quinquefasciatus CLIPA15 (also known as masquerade) across its sequence. CLIPA proteases interact with and regulate other CLIPs, and the prophenoloxidases (PPO), involved in melanisation [[33,34]]. This response produces reactive quinones which then polymerise to form the dark insoluble pigment melanin. These molecules can encapsulate and isolate invading pathogens or toxic compounds. They also locally generate high local levels of cytotoxic reactive oxygen species and prevents gas diffusion, starving the invading pathogen of oxygen. In addition to the putative CLIPA transcript, four pro-phenoloxidase transcripts are upregulated in early blood meal (PPO1 - PPAI008831, PPAI010450; PPO2 - PPAI012836, PPAI012835). These zymogens are the rate limiting enzymes in the production of melanin. PPO1/2 and CLIPA15 were also upregulated immediately after blood feeding in Anopheles gambiae [[24]] - suggesting this is a conserved response to blood feeding in Dipterans.
We also observed differential transcription of another group of proteins reported to play vital roles in protection against invading pathogens - peritrophins. These core components of the peritrophic matrix (PM) have been shown to be a major barrier against infection establishment. Knockdown of Peritrophin 1 (Per1) in P. papatasi results in an approximately 40% increase in Leishmania major load at 48 hours after parasite ingestion [[35]]. In our study, Per1 transcripts were highly enriched at day 1 PBM (vs. day 4 PBM) with log2 fold change of 9.96. Of the 32 annotated peritrophins in the P. papatasi genome, 14 were found to be significantly differentially regulated between days 1 and 4 PBM (Table 2). The majority of transcripts were comparatively enriched at day 1 PBM, however Per2 and Per28 transcripts were more abundant at later timepoints. Ramalho-Ortigão et al. [10] showed that P. papatasi peritrophin 1 (Per1) transcripts were enriched in flies fed a blood meal compared to a sugar meal, whilst peritrophin 2 (Per2) transcripts were comparatively depleted in blood fed flies. Additionally, the group showed that transcripts for both Per1 and Per2 were depleted in L. major infected flies compared to those fed only blood [[29]]. Our data largely agree with these findings. However, transcript levels were not statistically significant different between trypanosomatid and blood-only fed flies - though we do observe fewer transcripts for Per2 (PPAI009723) in trypanosomatid-fed flies at day 4 PBM (Fig S3). Other than Per2, the patterns in peritrophin transcript abundance for trypanosomatid fed flies resembled those of the blood fed controls.
Additionally, transcripts for another chitin-binding protein, PPAI000188, were significantly more abundant at 4 days PBM than at day 1 PBM. The sequence of PPAI000188 resembles the Lutzomyia longipalpis protein ChiBi (EU124616.1 [[36]], 84% protein sequence identity). ChiBi was shown to be enriched in L. longipalpis fed with blood containing L. infantum chagasi [[36]]. Its upregulation here in P. papatasi in the absence of Leishmania may indicate this upregulation is a more general response to blood meal, rather than an infection-specific response.
In addition to trypsins, transcripts of several other groups of genes associated with digestion and nutrient uptake were differentially regulated PBM. Several transcripts for lipid metabolism associated genes were found to be upregulated at day 1 PBM. In addition, eight transcripts corresponded to known extracellular carboxylic ester hydrolases (PPAI002323, PPAI003061, PPAI003086, PPAI005115, PPAI005116, PPAI005680, PPAI009133, PPAI008993). Similarly, transcripts for a putative sterol transfer protein (PPAI008838), and two paralogous membrane fatty acid desaturase genes (PPAI008098 and PPAI002108) were shown to be comparatively enriched at day 1 PBM. One transcript, CUFF.7417, does not correspond to a known gene, however the transcript showed strong sequence similarity to the extracellular carboxylic ester hydrolases paralogues PPAI005115 and PPAI005116 mentioned above (90% identity). Additionally, CUFF.7417 is immediately downstream of PPAI005115/6 in the genome and as such we propose this represents a previously unknown paralogue.
Four transcripts coded for proteins with solute carrier domains (cl00456). These transcripts encode for the two paralogous sodium-coupled monocarboxylate transporters (SCMTs, PPAI005125 and PPAI007402) and two putative SCMTs (CUFF.14648 and CUFF.14649). The SCMTs are transmembrane proteins, which move molecules with a single carboxylate group including pyruvate and lactate, across the plasma membrane in a proton-dependent manner and are associated with the insect midgut brush boarder [[37]].
We found two transcripts, CUFF.17209 and CUFF.15972, whose products are predicted to contain the conserved insect allergen related repeat domain (pfam06757). These transcript sequences also showed similarity to reported cDNAs for P. papatasi microvillar proteins MVP1 and 2 respectively (>89% identity to mRNA sequences). These proteins were also found previously to be upregulated in sand flies upon ingestion of a blood meal compared to sucrose-fed flies [[29]]. These transcripts could not be assigned to an annotated gene in the current vector base genome (Ppal1) [[17]]. The function of these proteins is not well understood though they appear to have a conserved signal peptide at the n-terminus and lack transmembrane domains.
Finally, three olfactory (Or57 - PPAI013155, Or99- PPAI013290 and the putative protein PPAI002404) and a gustatory receptor orthologous to sweet taste receptors of Drosophila (Gr9 - PPAI010978), were upregulated at day 1 PBM compared to later timepoints. It is likely these sensory receptors were involved in sensing and acquisition of the blood meal and subsequent decreases in their transcript abundances, may indicate these sensors were not required after digestion.
The transcriptome after defecation of the blood meal is comparatively stable
The two later timepoints in this study had similar transcriptomic signatures, with only six transcripts comparatively enriched >2-fold at 9 days PBM (vs. 4 days). These transcripts corresponded to two glutamate receptors (PPAI003634, PPAI008275), apoptosis inhibitor survivin (PPAI002284), two histone methyltransferases (PPAI005539, PPAI005538) and a mucin (PPAI009152). Mucins have been implicated in the interaction with Leishmania parasites. Given that several immunity-related transcripts (including peritrophins, mucins and melanization pathway genes) were upregulated, we postulated that upon blood meal ingestion a general immune response was triggered. As such we investigated the transcription of the members of the two major innate pathways after a bloodmeal: Toll and Immunodeficiency (Imd). Both pathways have been shown to play a role in the response to trypanosomatids [[11,38–42]] . Furthermore, we also investigated members of the Dual-oxidase (DUOX) and JAK-STAT pathways, both of which were implicated in D. melanogaster-H. muscarum interaction [[11]]. Differential regulation statistics for these transcripts can be found in Table S8.
Blood ingestion alone is associated with increased innate immune gene transcription
In blood-fed flies, transcripts putatively encoding early Toll pathway genes (two Toll receptors, Spätzle and GNBP3) were found to be significantly enriched at day 1 PBM compared to days 4/9 PBM (Figure 3A). An exception to this was the spätzle processing enzyme (SPE) the putative transcript for which is enriched in the latter two timepoints along with several intracellular Toll pathway components. These trends were broadly consistent in blood-only fed flies as well as those fed with each of the trypanosomatids. However, only flies fed with blood containing L. major or L. donovani promastigotes had significantly higher levels of transcripts encoding Toll pathway inhibitor Cactus at day-1 PBM compared to day 4 PBM (> 2-fold, p-adj = 0.01 and 5.08E-05 respectively). Cactus transcript abundance was not significantly different between days 1 and 4 PBM in blood only or H. muscarum fed flies.
A similar pattern emerges for the IMD pathway (Figure 3B). Transcripts for putative peptidoglycan recognition proteins (PGRPs) were more prevalent at day 1 PBM compared to later timepoints (Figure 6B). However, only putative PGRP 2 (CUFF.5670) was found to be statistically significantly enriched (2.23-fold) at day 1 PBM (vs. day 4/9 PBM). The transcripts putatively encoding IMD, and several other proteins downstream of IMD in the pathway were found to be significantly enriched at 4 and 9 days PBM (vs. day 1 PBM), including: DREDD, TAK1 and IKKb. We also observed significant enrichment of transcripts putatively encoding negative regulators of the IMD pathway Caspar, dUSP36, Trabid at days 4 and 9 PBM. Interestingly, the IMD transcription factor Relish was not significantly differentially regulated in blood-only fed flies, however flies fed blood containing L. major or L. donovani promastigotes showed enrichment of putative Relish transcripts at day 1 PBM compared to at 4 and 9 days PBM. As such, whilst there is overall upregulation of IMD pathway transcription with or without trypanosomatids in the blood meal, there may be important differences in the expression levels of the innate effectors the meal regulates when Leishmania are present.
Both Toll and IMD result in the expression of a suite of antimicrobial peptides. Transcripts for these immune effectors were not significantly differentially regulated after blood feeding alone. However, flies fed with blood containing L. major or L. donovani promastigotes were found to have significantly more transcripts for the AMP defensin at day 1 PBM compared to day 4 PBM (2.3 and 1.75 respective log2foldchanges). This was not observed in H. muscarum fed flies. Whilst transcript levels for other antimicrobial peptides did change after trypanosomatid feeding, e.g. we observe elevated transcript levels for cecropin and attacin in some trypanosomatid infections (Figure 3C), overall these were not found to be statistically significant changes.
In addition to AMP expression, the IMD pathway can also result in the transcription of the NADPH oxidase, dual-oxidase (DUOX), through interaction of IMD with MEKK1 [[43]]. This transmembrane protein is responsible for production of ROS in the gut epithelium in response to microbes. We found that DUOX transcripts were significantly higher at days 4 and 9 PBM compared to day 1 PBM in all feeding conditions (log2foldchange 2.98-3.33, Figure 6C) – with no significant difference in DUOX transcript abundance between 4 and 9 days PBM in any infection condition. Similarly, we see significant increases in transcripts for genes upstream of DUOX across infection conditions including: the transcription factor ATF2, p38 kinase and MEKK1. As such, induction of DUOX pathway transcription appears to be a generalised response to blood feeding rather than an infection-specific response.
The JAK-STAT pathway is also associated with the dipteran response to trypanosomatids
Finally, given the association between the JAK-STAT pathway (Figure 4B), dipteran gut morphology and immunity [[44]], particularly in a trypanosomatid infection context [[11]], we also investigate the transcription of key components of this pathway after blood feeding. We observed higher abundance of putative Upd1 transcripts at day 1 PBM compared to later timepoints, however this change was only show to be statistically significant for flies fed with blood and L. major where there was a 2.2-fold enrichment of putative Upd1 transcripts. Furthermore, putative transcripts for the JAK-STAT transcription factor STAT92E were 2-fold enriched in flies in all infection conditions at the two later time points (vs. 1 day PBM). We also observed a modest enrichment of transcripts for cytokine Upd2 and the transmembrane receptor Domeless at days 4 and 9 PBM compared to earlier timepoints (fold changes 1.19 and 1.74 respectively). The transcription pattern for signalling protein hopscotch resembled that of Domeless, however these transcripts were only found to be statistically significantly enriched in trypanosomatid fed flies. Together these observations suggested an increase in JAK-STAT signalling a few days after a blood meal in P. papatasi. Further work to investigate if this signalling translates to changes in gut homeostasis, such as the increased stem cell proliferation observed in the Drosophila-Herpetomonas model, will be important. Currently however, as transcript abundance for STAT92E is enriched in blood only fed controls this response does not appear to be trypanosomatid-specific.
Given the magnitude of the transcriptomic changes associated with blood feeding alone, and the little variation between blood meals spiked with trypanosomatids that produce very different infections we speculate that aforementioned defensive responses are not infection specific. Such a strong response to the blood meal alone is not surprising given the additional stresses associated with the hematophagous habit [[45]]. The high-risk nutrient attainment method drives the insects to take large volumes of blood at each meal e.g. mosquitoes and tsetse flies expand up to 3 times their pre-meal size a blood meal [[45,46]] – putting enormous mechanical strain on the tissues. In addition to the volume, the content of their meal presents additional problems: excess water/ions [[47]], toxic compounds [[48]] and bacterial expansion in response to the rich meal [[49,50]]. Due to the warm-blooded nature of their victim’s blood temperature of blood-feeding arthropods can rapidly (< 60 seconds) increase by over 10°C during their meal [[51,52]]. All of which must be overcome even in bloomeals which do not contain parasites. Additionally, activation of immune pathways following a blood meal (but in the absence of infection) may be a strategy to ‘pre-emptively’ protect the host against pathogens/toxic compounds which may be present in the newly ingested bloodmeal. Such anticipatory responses have been reported in other hematophagous insects including malaria vector Anopheles gambiae [[53]].