Anopheles aquasalis transcriptome reveals autophagic responses to Plasmodium vivax midgut invasion

Background Elimination of malaria depends on mastering transmission and understanding the biological basis of Plasmodium infection in the vector. The first mosquito organ to interact with the parasite is the midgut and its transcriptomic characterization during infection can reveal effective antiplasmodial responses able to limit the survival of the parasite. The vector response to Plasmodium vivax is not fully characterized, and its specificities when compared with other malaria parasites can be of fundamental interest for specific control measures. Methods Experimental infections were performed using a membrane-feeding device. Three groups were used: P. vivax-blood-fed, blood-fed on inactivated gametocytes, and unfed mosquitoes. Twenty-four hours after feeding, the mosquitoes were dissected and the midgut collected for transcriptomic analysis using RNAseq. Nine cDNA libraries were generated and sequenced on an Illumina HiSeq2500. Readings were checked for quality control and analysed using the Trinity platform for de novo transcriptome assembly. Transcript quantification was performed and the transcriptome was functionally annotated. Differential expression gene analysis was carried out. The role of the identified mechanisms was further explored using functional approaches. Results Forty-nine genes were identified as being differentially expressed with P. vivax infection: 34 were upregulated and 15 were downregulated. Half of the P. vivax-related differentially expressed genes could be related to autophagy; therefore, the effect of the known inhibitor (wortmannin) and activator (spermidine) was tested on the infection outcome. Autophagic activation significantly reduced the intensity and prevalence of infection. This was associated with transcription alterations of the autophagy regulating genes Beclin, DRAM and Apg8. Conclusions Our data indicate that P. vivax invasion of An. aquasalis midgut epithelium triggers an autophagic response and its activation reduces infection. This suggests a novel mechanism that mosquitoes can use to fight Plasmodium infection. Electronic supplementary material The online version of this article (10.1186/s13071-019-3506-8) contains supplementary material, which is available to authorized users.


Background
Malaria is still an important public health problem in several tropical countries. In 2016, 216 million cases of malaria were diagnosed; Brazil had 18% of all malaria cases that were confirmed by the World Health Organization (WHO) for the region of the Americas [1], and 99.5% of the Brazilian cases were in the Legal Amazon [2] where Plasmodium vivax is the predominant species accounting for 85% of reported cases [3].
Anopheles aquasalis is an important vector in coastal areas of South America [4,5]. Since its colonization it has been used for Plasmodium experimental infections for research, revealing a robust model to study the interaction of American vectors with Plasmodium species [6].
Inside the mosquito midgut the Plasmodium gametocyte-to-ookinete-to-oocyst transition is completed within the first 24 hours. The ingested parasite populations suffer substantial losses during this process which corresponds to the most critical population bottleneck of the entire parasite life-cycle; more often than not, transmission is terminated at this stage [7][8][9]. Invasion of the malaria vector Anopheles gambiae midgut by Plasmodium parasites triggers transcriptional changes of genes that mediate the antiparasitic defence [10] and, thus, the ability of these mosquitoes to transmit malaria [11]. Moreover, several mechanisms are triggered by the mosquito in order to combat the intracellular pathogen. Apoptosis is one possibility that has been described during ookinete invasion of the midgut [12]. Another related mechanism is autophagy, an important and well-studied cytosolic response. During macroautophagy, a double membrane vesicle called autophagosome forms around cytosolic components, which subsequently fuse with lysosomes and degrade the vessel's content [13]. Under certain conditions in Drosophila, midgut and salivary gland tissues show both high caspase activity and the formation of autophagosomes [14,15], suggesting that apoptosis and autophagy may be highly integrated in arthropods.
The upregulation of autophagy can enhance resistance to pathogens, a phenomenon that has also been associated with resistance of the fruit fly Drosophila melanogaster to bacterial challenge [16] and of mammalian cells to bacteria, viruses and parasites such as Toxoplasma gondii [17][18][19][20][21]. Intriguingly, a P. falciparum infection in Anopheles stephensi induced a translation of autophagy (ATG) protein mRNAs, including those for key regulators ATG6 and ATG8, in the midgut epithelium at 24 hours after infection [22], which suggests that autophagy is induced early during sporogonic development in the mosquito host. While autophagy induction can control resistance, perhaps contributing to the large-scale death of parasites in the midgut, it is possible that highly conserved regulation of stem cell renewal and differentiation by autophagy could also impact the midgut's response to parasite infection [23].
Transcriptomic analyses of African and Asian mosquitoes in response to pathogens have generated a wealth of data that can facilitate the development of new anti-malaria tools [24,25]. More recently, An. aquasalis specimens have been analysed for functional annotation creating opportunities for further molecular characterization of genes. The An. aquasalis transcriptomes of larvae and adults fed on sugar and on blood revealed valuable information about protein-coding transcripts involved in biological processes relevant to mosquito physiology and development of this new world model [26].
Nevertheless, a deeper understanding of the processes taking part on this critical phase of malaria transmission remains unexplored. Here, we report the transcriptional profile of An. aquasalis midgut, in the early stage of P. vivax development and invasion of the midgut epithelium. Via this profile, we were able to gain insights on a molecular level of how to functionally characterize this critical phase of malaria transmission. Our results revealed the importance of alternative mechanisms, such as autophagy, for the control of Plasmodium infection in the mosquito.

Mosquito collections and rearing
An. aquasalis were obtained from a well-established colony at the Entomology Department's insectary at the Fundação de Medicina Tropical Dr Heitor Vieira Dourado, Manaus, Amazonas, Brazil (FMT-HVD). All mosquitoes were reared at 26 °C, 70-80% relative humidity under a 12/12 light/dark photoperiod. Larvae were fed with commercial fish food (Tetramin Gold ® ; Tetra GmbH, Melle, Germany) and adults were fed ad libitum on 10% sugar solution. Three-to five-day-old adult females were used in all experiments.

Blood collection
Adult volunteers (ages ≥ 18 years), residents from the region of Manaus (State of Amazonas, Brazil) with P. vivax malaria infection diagnosed by blood smears, were recruited at the Fundação de Medicina Tropical Dr Heitor Vieira Dourado (FMT HVD). All volunteers were instructed on the study objectives. A sample of about 10 ml of venous blood was drawn from each patient and placed in heparinized tubes. After blood collection all patients were treated according to Brazilian Health Ministry guidelines [27].

Plasmodium vivax infection of mosquitoes via membrane feeding assay
Adult mosquitoes were sugar starved overnight prior to infection and separated into two experimental groups. One group was offered whole blood from P. vivax patients for a period of 45-90 min via membrane feeding assay (MFA). Blood was held at 37 °C through a hose system connected to a thermal bath [6]. The second group was treated in similar way, but with inactivated gametocytes as described by Mendes et al. [28]: P. vivax infective blood was briefly heated for 15 min at 43 °C, chilled to 37 °C and then offered to the mosquitoes. Only fully engorged mosquitoes were transferred to rearing containers and maintained in the insectary at 26 °C with 70-80% relative humidity and fed ad libitum on 10% sugar solution.

Tissue collection and RNA isolation
Mosquito midguts were collected from pools of 30 mosquitoes, 18-24 h after the blood meal, from each of three groups: (i) P. vivax-blood-fed group (Pv); (ii) non-infective group (Bl; blood-fed using P. vivax-blood in which the gametocytes were inactivated); and (iii) unfed group (Unf; unfed mosquitoes). Tissues were dissected from mosquitoes submerged in ice-cold phosphate buffered saline (PBS) and transferred to RNAlater (Thermo Fisher Scientific, Massachusetts, USA). Samples were stored at − 20 °C until RNA extraction. Total RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific, Massachusetts, USA) following the manufacturer's protocol. To eliminate possible contaminant genomic DNA, the RNA samples were treated with RNase-Free DNase I according to manufacturer's protocol (Qiagen, Hilden, Germany). At 8-9 days post-infection, mosquito midguts were collected to determine the infection rate (number of infected mosquitoes over total number of mosquitoes observed) and infection intensity (mean number of oocysts per infected mosquito). Three independent biological replicates of each experiment were performed.
Differential expression (DE) analysis was performed using GLM test in the edgeR v.3.16.5 package [32] in R. Pairwise comparisons were made between the different group samples. In any given group, a transcript was considered differentially expressed if its adjusted P-value to control the false discovery rate (Benjamini-Hochberg adjustment) was less than 0.05 and if log fold change was higher than 1. Differentially expressed genes were further analysed for functional classification using gene ontology analysis on PANTHER (http://www.panth erdb.org) [33]. The dataset has been deposited at the Gene Expression Omnibus under the accession number GSE124997.

Validation of differentially expressed genes
In order to validate transcriptome analysis, a total of 8 differentially expressed genes between mosquitoes fed on blood with infective P. vivax (Pv) and fed on blood in which gametocytes were inactivated (Bl) were chosen for real-time quantitative PCR analysis which was performed as described in [34]. For this, total RNA was used, and first strand cDNA was synthesized using oligo dT and MMLV Reverse Transcriptase (Promega, Wisconsin, EUA) as described in [35]. The cDNA was used as a template for RT-qPCRs using the primer pairs reported in Additional file 1: Table S1. The primers were derived from the sequences identified in the transcriptome. For all groups, mosquito midguts were collected 18-24 h post-infection in order to determine the levels of expression of the genes in the midgut. Experiments were conducted with three biological replicates, each in triplicate.

Reverse transcription quantitative real-time PCR (RT-qPCR)
Real-time quantitative PCR was performed on a Fast 7500 instrument (Applied Biosystems, California, USA) with SYBR Green Power Master Mix (Applied Biosystems) using 2 μl of cDNA template in a final volume of 20 μl reaction mixture. Fold-changes of gene expression were analysed using the 2 −ΔΔCT method. The ribosomal protein S7 was used as the endogenous control.

Spermidine and wortmannin treatments
The transcriptome associated to P. vivax infection revealed a variety of transcripts that play a key role in autophagy. In order to evaluate the effect of the autophagy process in the outcome of infection, we inoculated mosquitoes with wortmannin (an inhibitor of phosphatidylinositol 3-kinase DPI3K) and spermidine (an autophagy activator) [36,37]. Three-to four-day-old female mosquitoes were cold-anesthetized and inoculated intrathoraxically with 69 nl of a 5 μM and 0.05 μM solution of wortmannin (Merck, Darmstadt, Germany) or with the same volume of H 2 O Ultra Pure and with 69 nl of a 100 μM solution of spermidine (Sigma) or DMSO (0.05%) using a Nanoject micro-injector (Drummond Scientific, Pennsylvania, USA). Twenty-four hours after injection with the solutions, the mosquitoes were fed with a P. vivax-infected blood meal as described above. Three independent biological replicates were performed for each experiment. Mosquitoes were dissected 18-24 h after feeding; batches of 20-30 midguts were dissected in cold DEPC-treated phosphate-buffered saline (PBS) and processed for RNA preparation and cDNA synthesis using the same protocols mentioned above. Mosquito midguts were also collected on the 8th day post-infection to determine the prevalence and intensity of infection.

Gene expression of autophagy related genes
The expression of genes that regulate autophagy (Beclin, DRAM and Apg8) was investigated 18-24 h after P. vivax infection and 24 h after inhibition and activation of autophagy (treatment with wortmannin or spermidine) as described above.
Anopheles aquasalis mosquitoes were dissected 18-24 h after infection; 20-30 midguts were collected, and RNA and cDNA were prepared as described above. Three independent experiments were performed. Gene expression analysis was carried out by quantitative real-time PCR following the same conditions described above.

Statistical analysis
For data not normally distributed (oocyst density), twosample comparisons were done using the non-parametric Mann-Whitney test. The differences in the infection rate between the control group and the tested groups were compared using Fisher's exact one-tailed test (F). Comparisons of mRNA expression levels obtained by RT-qPCR between the control and the tested groups were done using the Mann-Whitney one-tailed test. Statistical analyses were performed using the software GraphPad Prism v.6.00.

General characterization of midgut transcriptome
A total of 9 cDNA libraries from An. aquasalis midguts were constructed and sequenced, namely three libraries for each of the following groups: (i) P. vivax-blood-fed mosquitoes (Pv: groups Pv1 to Pv3); (ii) mosquitoes fed on P. vivax-blood from which gametocytes were inactivated (non-infected: groups Bl1 to Bl3); and (iii) unfed mosquitoes (unfed groups: Unf1 to Unf3). The obtained mean number of high quality paired-end short reads were: 56 (Table 1).
To examine differential expression between mosquitoes fed on P. vivax infected-blood (Pv) and mosquitoes fed on non-infective blood (Bl) or unfed mosquitoes, FDR < 0.05 and logFC > 1 (fold change) were used as the threshold to classify differentially expressed genes. The analyses showed a total of 12,942 expressed genes. Of these, 49 genes were identified as differentially expressed genes in the P. vivax infected-blood-fed group (Pv) in relation to non-infected-blood-fed group (Bl); 34 were upregulated and 15 were downregulated, which represents differential expression associated to P. vivax infection (Fig. 1). A total of 111 genes were differentially expressed in infectedblood-fed mosquitoes (Pv) when compared to the unfed group (Unf); of these, 65 were upregulated and 46 were downregulated (Fig. 2). The detailed gene lists are shown in Additional file 2: Tables S2-S5. About 49% (24 out of 49) of the differentially expressed genes in the Pv × Bl group were exclusive to this comparison (Fig. 3) and involved a large gene set related to autophagy. On the other hand, the Pv × Unf comparison presented 86 out of 111 (78%) of the differentially expressed genes exclusively in this group (Fig. 3).
To validate the robustness of RNAseq results, we analysed eight genes by real-time qRT-PCR and compared the expression of these genes in Pv × Bl (Additional file 3: Figure S1). These analyses revealed a significant correlation (Pearsonʼs correlation coefficient = 0.874, R 2 = 0.7663, slope = 0.04539) between the qRT-PCR and the RNAseq data.

Anopheles aquasalis midgut differential gene expression associated with Plasmodium vivax infection
Transcriptomic analysis of midgut infected mosquitoes revealed 49 differentially expressed genes in the P. vivax infected-blood-fed group (Pv) in relation to noninfected-blood-fed group (Bl); of these, 34 were upregulated and 15 were downregulated. From these, genes involved in cellular process, metabolic process (GO: 0008152), cellular component organization or biogenesis process (GO: 0050896) and biological regulation process (GO: 0065007) were predominant (Fig. 4). The results suggest that many of the upregulated genes are involved in several metabolic processes and molecular functions, among them, catalytic activity (GO: 0003824) and cofactor binding (GO: 0005488) were enriched with hits of 6 and 4 genes, respectively, followed by transporter and structural molecule activity.

Immunity
Activation of mosquito immunity genes has been traditionally associated with midgut-infected mosquitoes.
In the present study, a transcript coding for a leucinerich repeat protein, orthologue of LRR-containing protein 58 (TRINITY_DN6165_c5_g1_i4), was found upregulated in mosquitoes infected with P. vivax (Pv) when compared to the mosquitoes fed on inactivated gametocytes (Bl). LRR-containing protein 58 has been previously associated with the An. gambiae response to Plasmodium berghei infection [38,39]. LRR containing proteins are related to the control of the complementlike protein TEP1 function, and have other important roles in innate immune defence [40]. Information on all transcripts (TRINITY_DN0000_c00_g00_i00) are summarised in Additional file 2: Tables S2-S5.

Detoxification
Plasmodium vivax is probably able to modulate detoxification of free radicals while invading the midgut of An. aquasalis, as suggested by the increase of H 2 O 2 following artificial reduction of catalase activity which leads to increased parasite infection in the mosquito midgut. As gene silencing also decreases the midgut microbiome, Bahia et al. [41] suggest that this manipulation occurs through the control of competitive bacteria which allows better parasite development. Transcript coding for CYP315A1 (TRINITY_DN6118_ c3_g2_i14) was upregulated in the Pv × Bl group, as was its orthologue in deltamethrin-resistant An. gambiae mosquitoes when compared with a sensitive mosquito line from Kenya [42]. NAD+ kinase (TRIN-ITY_DN5956_c5_g5_i5) and senecionine N-oxygenase (TRINITY_DN5975_c0_g1_i2) were also upregulated with infection, which suggests that the enzyme activity was needed for the antioxidant activities of other enzymes. Autophagic responses can be modulated by radical species and NAD+ homeostasis and the midgut metabolism can be an important player in autophagy regulation [43].

Cytoskeleton remodelling
Anopheles gambiae midgut response to P. berghei ookinete invasion is characterized by profound alterations in the transcription of genes that modulate the architecture of the cytoskeleton [38]. Plasmodium parasites need to modify the cytoskeleton of mosquito epithelial cells to successfully complete their life-cycle. We found several downregulated genes (TRINITY_DN4493_c0_g1, TRINITY_DN5277_c0_g1, TRINITY_DN5389_c0_g1, TRINITY_DN6055_c0_g1, TRINITY_DN6090_c5_g3, TRINITY_DN6296_c2_g1) that could be associated with cytoskeleton remodelling, which reinforces the prominent role of this cellular mechanism in response to Plasmodium and extends it to P. vivax ookinete invasion of the mosquito midgut.

Autophagy
The present transcriptional analysis suggested that differential expression of autophagic genes is involved in An. aquasalis females following a P. vivax infected blood meal. Forty-nine percent of differentially expressed genes during invasion (60.0% of the upregulated and 44.1% of the downregulated genes) could be associated with autophagic processes (Table 2). Transcript coding for the GPCR Methuselah receptor 6 (TRINITY_DN5851_c0_g1_i2) was downregulated, as was β-arrestin (TRINITY_DN5911_c0_g1), which uncouples GPCRs from their G-proteins, and suggests that regulation of free radical production might occur through this molecule. In Drosophila, Methuselah receptors have been associated with lifespan and resistance to starvation and free radicals [44]. Wang et al. [45], using a specific agonist and antagonist, demonstrated that the TOR pathway is one of the major effectors underlying Methuselah. Blocking Methuselah reduced dTOR activity and promoted autophagy.
Trehalose is a natural sugar found in prokaryotes, yeast, fungi, plants and invertebrates, and serves not only as a reserve of carbohydrate, but can also protect organisms and cells against adverse environmental conditions. Some controversy exists on the real effect of trehalose on autophagy. In murine models, trehalose seems to induce autophagy, while in cultured cells it could inhibit fusions of autophagosomes and lysosomes, thus blocking the final stage of autophagy [46]. Our data suggest that the An. aquasalis midgut increases intracellular trehalose by upregulating trehalose 6-phosphate synthase/phosphatase (TRINITY_DN6177_c2_g2) and downregulating the TreT1-facilitated trehalose transporter (TRINITY_ DN5823_c0_g1), suggesting autophagy induction in the An. aquasalis midgut during P. vivax infection. Anopheles gambiae TreT1 RNA silencing reduces the number of P. falciparum oocysts in the mosquito midgut [47], suggesting it might exert parasite protection during midgut invasion.
Microtubules (MT) are important to autophagosome formation and motility. Dynein light chain 1 (TRINITY_ DN6473_c3_g4), a motor protein, was upregulated upon infection in our study. In vertebrates, Beclin-1 is sequestered in MT in complexes containing dynein light chain 1. When autophagy is stimulated, Beclin-1 is released from this complex. In parallel, c-Jun N-terminal kinase-1 (JNK1) is activated which allows phosphorylation of Bcl-2 and Bim, which, in turn, releases Beclin 1 and contributes to autophagosome formation [48]. Regarding the gamma-tubulin complex component 5 (TRIN-ITY_DN5389_c0_g1), involved in microtubule assembly [49], and T-complex protein 1 subunit gamma (TRIN-ITY_DN4493_c0_g1), a chaperonin for tubulin and actin [50], we found that these were downregulated, possibly confirming that (as in other Anopheline species) microtubule dynamics are altered during P. vivax invasion of the midgut epithelium, and it is possible that this is associated to autophagy. JNK signalling has been demonstrated to be involved in lifespan control and is required in differentiated cells of the intestinal epithelium in order to prevent excessive sensitivity of these cells to oxidative stress in Drosophila [51] and has been implicated in mosquito defense against malaria parasites, and altered expression patterns of autophagy biomarkers [22]. Garver et al. [52] showed that basal mRNA expression of the genes involved in JNK signalling were upregulated in the mosquito midgut and JNK silencing significantly increases the prevalence of infection. However, more recently, Souvannaseng et al. [53] demonstrated that moderate inhibition of JNK signaling in the An. stephensi midgut extended lifespan and enhanced resistance to P. falciparum. In the present study, we found that the AP-1 transcription factor (TRINITY_DN6454_c2_g2), a downstream product of this signalling pathway, was upregulated in mosquitoes fed on P. vivax blood.
The ubiquitin machinery regulates fundamental biological processes within eukaryotic cells. The enrichment of functional terms such as ubiquitin-dependent proteasome was also denoted for insects facing dehydration stress [54]. Nitric oxide synthase expression and nitric oxide increase in the midgut of An. aquasalis, An. stephensi and An. gambiae during Plasmodium parasite infection, which limits parasite development within the mosquito [55][56][57]. Ubiquitin carboxyl-terminal hydrolase (TRINITY_DN6536_c2_g8), ubiquitin-like-conjugating enzyme ATG3 (TRINITY_DN6039_c0_g1) and OTU domain-containing protein 5 (TRINITY_DN6021_c0_g3) were downregulated during midgut invasion by P. vivax. OTU domain-containing proteins are deubiquitinating enzymes and cleave distinct sets of ubiquitin chain types [58]. In vertebrates, OTU domain-containing protein 5 regulates interferon signalling [59]. USP10 regulates the deubiquitination of Beclin1 in Vps34 complexes, which in turn leads to a reduction in the levels of PtdIns3P and consequent inhibition of autophagy. Spautin-1, an inhibitor of USP10 and USP13, promotes the ubiquitination and degradation of Vps34 complexes, which in turn leads to a reduction in the levels of PtdIns3P and consequent inhibition of autophagy [60]. Since ubiquitination and deubiquitination are central to autophagy regulation, once more, our data indicated that P. vivax invasion of An. aquasalis midgut epithelium triggers an autophagic response. ATG3, among other enzymes, is involved in the maturation of the growing autophagosome, a process that occurs once autophagy is initiated. Recently, Frudd et al. [61] described a mechanism that associates reactive species to autophagy induction. The oxidation of Atg3 and Atg7 prevents phosphatidylethanolamine conjugation to LC3 (microtubule-associated protein 1A/1B-light chain 3), thus, associating reactive species to autophagy induction. The production of reactive oxygen/nitrogen species during ookinete invasion of An. aquasalis [41], together with the set of differentially displayed genes such as, is suggestive of an interplay between reactive species and autophagy during ookinete invasion of the midgut.
Overexpression of prominin 1 constitutively activates autophagy in the human retinal pigment epithelium via inhibition of mTORC1 and mTORC2, while it impairs autophagy via upregulation of mTORC1/2 activities. Prominin (TRINITY_DN6489_c3_g1) was upregulated during parasite invasion of the midgut epithelium, which suggests that autophagy might be activated during this stage of infection.
Lipid droplet (LD) homeostasis [62] also plays an important role in autophagy regulation. We observed that a fat storage-inducing transmembrane protein (TRINITY_DN6330_c2_g6), that was described to facilitate proper LD budding from the ER [63], was downregulated. Sphingolipids have also been associated with lipid droplet formation, and sphingomyelin phosphodiesterase 1 (TRINITY_DN6531_c1_g1) was upregulated. Although sphingomyelinases do not play a role in autophagy induction, the upregulation of sphingomyelin phosphodiesterase 1 (TRINITY_DN6531_c1_g1) can act in the autophago-lysosomal degradation [64], thus regulating the autophagosome formation. Vesicle transport protein SEC22 (TRINITY_DN6646_c8_g1), a protein implicated in autophagosome biogenesis [65], was downregulated, while endothelin-converting enzyme (TRIN-ITY_DN6321_c0_g2), previously detected in autophagic vesicles [66], was upregulated. Regulation of these genes suggests that P. vivax invasion affects regulation of different stages of the autophagic process and includes autophagosome maturation and degradation.
Plasmodium invasion of the midgut epithelial cell leads to a number of molecular and morphological changes, including cell death. Vlachou et al. [67] proposed that the first invaded cells undergo apoptosis and are expelled to the lumen, while adjacent cells extend lamellipodia to maintain epithelium continuity. The ookinete reinvades several cells until it reaches the extracellular matrix in order to develop into oocysts, which implies a substantial dynamic rearrangement of the actin cytoskeleton. Division of regenerative cells within the midgut epithelium of an adult female An. stephensi in response to P. falciparum invasion [68] has been reported. Even so, the extent of apoptosis and the mechanisms by which the integrity of the midgut epithelium is maintained are not yet understood. The detachment-induced apoptosis (anoikis), driven by these morphological changes, can be deleterious to the mosquito and can be compensated by extracellular matrix (ECM) detachment, which induces autophagy [69] mediated by integrin [70]. Integrin are cell surface proteins that interact with the external cellular matrix (ECM), and signal through the cell membrane in both directions. Plasmodium vivax infection of the midgut positively regulated the expression of this gene (TRINITY_DN6333_c5_g2).
The Drosophila epidermal growth factor receptor (EGFR) pathway has been implicated in the control of delamination and anoikis of damaged enterocytes following oral bacterial infection [71] and Serratia marcescens infection of An. gambiae activates the EGFR pathway by modulating the outcome, possibly through synergistic functions in gut homeostasis [72]. Spitz (TRINITY_DN1272_c0_g1), the ligand of EGFR, was upregulated with infection and Forkhead box K2 (TRIN-ITY_DN6571_c0_g10_i7) which can inhibit EGRF in 769-P cells [73], which suggests that the EGFR pathway is activated following P. vivax infection. This probably contributes to gut hemostasis through autophagy. Subcellular localization of the EGFR seems to be determinant on the effect on autophagy, being either an inhibitor or stimulant [74].

Treatment of mosquitoes with autophagy inhibitor
In order to evaluate the autophagy effect on the outcome of P. vivax infection in An. aquasalis, the mosquitoes were treated prior to infection with the autophagy inducer spermidine, or an autophagy inhibitor wortmannin.
The differences between the two doses of wortmannin are probably a consequence of the drug mode of action. Wortmannin is a PI3-kinase inhibitor, therefore, as autophagosome formation requires class III PI3kinase activity, it is normally used to study the effect of autophagy inhibition. Nonetheless, wortmannin can also inhibit class I PI3-kinase activity (which inhibits autophagy) and can also inhibit mTOR (an autophagyinhibitory molecule) [75]. Furthermore, wortmannin can also act on the parasite and interfere with its development [76].
Spermidine is a polyamine that stimulates autophagy, both through mTOR-independent or dependent mechanisms [77] and its administration to mosquitoes produces a significant reduction in P. vivax infection after treatment. Polyamine biosynthesis inhibitors cause growth arrest of P. falciparum blood stages in vitro but show no effect on survival of mice infected with P. berghei (reviewed in [78]). Despite these data, polyamide biosynthesis seems to be fundamental for sporogonic cycle completion. Targeted deletion of the enzyme AdoMetDC/ODC from Plasmodium yoelii blocks transmission to the mosquito An. stephensi, which could not be rescued by supplementation with spermidine [79]. This information reinforces that spermidine is acting on the mosquito rather than the parasite. Our data demonstrated a significant reduction in P. vivax infection after spermidine treatment in An. aquasalis, which, together with the data obtained using low wortmannin treatment, suggests that autophagy can control P. vivax infection in An. aquasalis.

Expression of autophagy genes following Plasmodium vivax infection and treatment with autophagy suppressor and inhibitor
To further characterize the role of autophagy in the mosquitoes treated with autophagy inhibitor in response to Plasmodium infection, qRT-PCR was used to quantify the changes in gene expression in response to a P. vivaxinfected blood meal. A differential expression analysis of several autophagy genes, including DRAM, Apg8 and Beclin, during inhibition and activation of autophagy, was performed. Atg8 protein, formerly known as Apg8/Aut7 is part of a group of proteins that control autophagy, many of which also participate in direct cytoplasmto-vacuole transport of proteins [80,81]. Among the genes that promote autophagy is the damage-regulated autophagy modulator (DRAM-1), which belongs to an evolutionarily conserved family of proteins that encodes for a lysosomal protein that is required in order to induce autophagy [82,83], and Beclin-1, which is part of a Class III phosphatidylinositol 3-kinase complex that is thought to be important in mediating localization of other Apg proteins to pre-autophagosomal structures [60].
No major differences were observed in expression of these genes after treatment with both drugs when compared with infection without treatment (Fig. 6). This is in-line with RNAseq data where transcription alterations of these genes were not detected. The major difference in expression was observed for beclin, which was downregulated after mosquitoes were treated with spermidine (P = 0.0635), suggesting that spermidine is

Conclusions
Our results clearly indicate that autophagy is regulated by P. vivax invasion of the mosquito midgut epithelium. A vast number of genes associated with autophagy were regulated by infection of which 60% were upregulated. Furthermore, when autophagy was inhibited by spermidine, we observed a significant reduction of the prevalence and intensity of infection. In view of our results, we propose that when ookinetes invade the midgut cells they trigger host cell morphological rearrangement, with actin and microtubule remodelling and production of nitrogen and oxygen radicals and possible cell death. To counterbalance invaded epithelial cell death/extrusion and other injuries parasites, could trigger an autophagic mechanism that would restrain parasite development, possibly through GPCR signalling Methuselah, the increase of intracellular trehalose, and detachment from the excellular matrix. This effect was apparent by the regulation of genes that could be assigned to the different stages of autophagy (initiation, nucleation, elongation/closure and maturation degradation) [84]. Autophagy triggered by Plasmodium invasion in epithelial midgut cells is a novel mechanism for mosquitos in order to fight Plasmodium infection.

Additional files
Additional file 1: Table S1. List of primers used in qRT-PCR analysis.
Additional file 2: Additional file 3: Figure S1. Validation of RNAseq analysis using qRT-PCR. Gene expression values for eight genes obtained by RNAseq were plotted against the corresponding averages of three qRT-PCR-derived gene expression values from biological replicates. The Pearsonʼs correlation coefficient (0.874) and the best-fit linear-regression analysis R 2 = 0.7663 demonstrated a good degree of correlation between gene expression determined by each assay. Genes used for validation: TRINITY_DN4493_ c0_g1_i2, TRINITY_DN5277_c0_g1_i2, TRINITY_DN5911_c0_g1_i4, TRIN-ITY_DN6055_c0_g1_i13, TRINITY_DN6039_c0_g1_i17, TRINITY_DN6296_ c2_g1_i5, TRINITY_DN6531_c1_g1_i4, TRINITY_DN6536_c2_g8_i1.