Since the abandonment of the wide-spread vector control programs in the late 1960's, the yellow fever mosquito, Aedes aegypti, has rapidly re-emerged across the globe. The current range of this important disease vector puts ~2.5 billion people at risk for dengue fever, with an estimated 50–100 million cases per year. A. aegypti is also the primary vector of the viruses which cause yellow fever, chikungunya, and Zika (Nene et al., 2007; CDC, 2009; WHO, 2011; Rodriguez-Morales, 2015; Benelli and Mehlhorn, 2016; Lazear et al., 2016).

Larval and adult mosquitoes inhabit very different environments with diverse requirements regarding osmoregulation and excretion. During the larval (and pupal) life stages, mosquitoes are aquatic organisms that are immersed in a hypotonic environment (Bradley, 1987; Marusalin et al., 2012). Post-eclosion, adult A. aegypti are terrestrial organisms that face the threat of rapid dehydration. Accordingly, the osmoregulatory system must change rapidly for the mosquito to acclimate to the new environment (Bradley, 1987; Piermarini, 2016). Adult females are presented with an additional challenge when feeding on vertebrate blood. Within minutes, a female A. aegypti can ingest more than her own body weight in blood which drastically reduces her mobility (Clements, 1992). In order to regain mobility and maintain osmotic balance, A. aegypti females excrete at least 40% of the volume contained in a blood meal within 1–2 h (Williams et al., 1983; Drake et al., 2010). Vertebrate blood is composed of ~85% water, and a mixture of ions, sugars, and proteins (Scanlon and Tina, 2007). Thus, ingesting large volumes of blood affects osmotic balance and leads to the release and production of toxic metabolic wastes (e.g., heme, NH₃) during digestion.

The Malpighian tubules (MT) are the key excretory tissues in mosquitoes for osmotic balance and excretion of wastes, such as xenobiotics and excess nitrogen. In A. aegypti, there are five MT that attach to the alimentary canal at the midgut/hindgut junction. The tubules are comprised of two main cell types (principal cells and stellate cells) that form a single epithelial layer around the lumen (Beyenbach and Piermarini, 2011). Stellate cells intercalate between principal cells along the distal segment of the tubule. Principal and stellate cells are thought to work in conjunction to maintain proper osmotic balance within the hemolymph (Beyenbach et al., 2010; Beyenbach and Piermarini, 2011).

In the current model of transepithelial fluid secretion in the MT of A. aegypti, a proton motive force is created by the principal cells through a V-type H⁺ ATPase (V-ATPase) located in the apical membrane (Beyenbach et al., 2010; Beyenbach and Piermarini, 2011; Piermarini, 2016). The V-ATPase pumps protons into the lumen creating proton and voltage gradients to power other electrogenic exchange transporters and ion channels that move Na⁺ and K⁺ from the hemolymph into the lumen of the tubules. Only some of the molecular mechanisms have been identified and characterized so far. The stellate cells exchange intracellular HCO₃ with Cl⁻ from the hemolymph, then the Cl⁻ is transported to the lumen, presumably by chloride channels in stellate cells and/or a K,Cl cotransporter (KCC) in principal cells (Beyenbach, 2003; Beyenbach et al., 2010; Beyenbach and Piermarini, 2011; Piermarini et al., 2011). It is unknown exactly how water moves from the hemolymph to the lumen, however in A. aegypti, several aquaporin (AQP) mRNAs are enriched in the MT (Pietrantonio et al., 2000; Drake et al., 2010, 2015). In MT of Anopheles gambiae, Prip (AQP2 in A. aegypti) immunoreactivity has been located on the basolateral membrane of stellate cells in the distal segment and principal cells of the proximal segment (Liu et al., 2011; Tsujimoto et al., 2013). RNAi-mediated knockdown of selected AQPs in adult female A. aegypti resulted in greatly reduced whole-mosquito diuresis capabilities (Drake et al., 2010), consistent with roles in transepithelial water transport in the MT. Mosquito diuresis is regulated by a number of neuropeptides including kinins, cardioacceleratory peptides (CAPs), adiokinetic hormones (AKHs), and corticotropin-releasing factor related (CRF related) hormones (Gade, 2004).

In addition to maintaining osmotic balance, the MT have a role in detoxification and excretion of metabolic waste products. Uric acid is thought to be transported into portions of the MT where it accumulates into crystals and assists with transport of water into the tubules by contributing to the osmotic gradient (O'Donnell et al., 1983). Enrichment of cytochrome p450's and glutathione S-transferases (GST) in the MT of Drosophila, A. gambiae, and Aedes albopictus indicates that this organ may also play a role in detoxification (Beyenbach et al., 2010; Ingham et al., 2014; Esquivel et al., 2016).

Previous studies have characterized the transcriptome of the MT in A. albopictus (before and after blood feeding) (Esquivel et al., 2014, 2016) and A. gambiae (larval, adult, before, and after blood feeding) (Overend et al., 2015), but transcriptomic studies in the MT of A. aegypti have not previously been performed. The goal of this study was to characterize the changes occurring in the MT of A. aegypti mosquitoes when switching from an aquatic to a terrestrial life. We conducted an in depth RNA-seq analysis of MT from 4th instar larvae and adult females 3 days post-eclosion.

We found a high number of transcripts differentially expressed in the MTs of adult mosquitoes and larvae. We also identified potential sodium and chloride channels that expand our current model of diuresis in mosquitoes.

Mosquitoes from the A. aegypti Rockefeller (ROCK) strain were raised as previously described (Price et al., 2011) with the exception that they were fed as larvae solely on cat food (Special Kitty Original, Wal-Mart stores, Bentonville, AR).

MT were isolated from ~30 adult female or 30 larval (4th instar) A. aegypti in PBS and placed in Trizol® (Thermo Fisher Science). This was done in duplicate, creating two replicates for each group. Total RNA was then extracted from these tubule samples according to the manufacturers protocol (Chomczynski, 1993). A Nanodrop 1000 (Thermo scientific) was used to quantify total RNA concentration. RNA quality was assessed visually by an RNA gel.

Four micrograms of total RNA (the recommended maximum) from each sample was used to prepare a cDNA library for each sample using the TruSeq RNA Sample Preparation Kit v2 (Illumina), according to the manufacturer's protocol for low-throughput sample preparation, with the following modifications (Tsujimoto et al., 2017). Differences in the protocol and our preparation procedure were: (1) using PCR strip tubes instead of PCR plates, (2) Elute, Prime, Fragment mix was thawed on ice and mixed into each well of the RBP plate on ice, and (3) Ligation mix was thawed on ice and mixed into each well of the ALP plate on ice.

The resulting libraries were quantified using a Nanodrop 1000 and Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA) at New Mexico State University, then sent for sequencing to the National Center for Genome Resources (Santa Fe, NM). The sequencing lab further analyzed the libraries and pooled them for sequencing on a single lane of Illumina HiSeq2000 1 × 100 bp reads.

Illumina reads were aligned using bowtie2 (v2.2.9) and Tophat2 (v2.1.1) to the A. aegypti reference transcripts available from Vectorbase (downloaded May. 5, 2016) (Lawson et al., 2009; Langmead and Salzberg, 2012). Each library was aligned separately. Alignments in each library for each transcript were tallied from bowtie2/Tophat2 results by htseq-count (Anders et al., 2015), then expression was compared between larvae and adults using the DESeq package in R (Anders and Huber, 2010). After alignment by bowtie2 and Tophat2, the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) values and expression comparison of transcript isoforms were also calculated using Cuffdiff (Trapnell et al., 2010). A Pearson correlation test (Galton, 1877; Pearson, 1895) was used to compare DESeq results and Cuffdiff results with a correlation value of 0.98 and a P-value of lower than 2.2e-16.

For the clustering analysis, we followed a similar protocol to Esquivel et al. (2014, 2016). In brief, transcripts were separated based on their expression in the MT of adults relative to MT of larvae (i.e., down-regulated or up-regulated in adult MT). Then, transcripts were submitted to a standalone BLASTn (version 2.2.31) analysis against the A. gambiae transcriptome (PEST strain transcript sequences, AgampP4.4 geneset, v1.00; https://www.vectorbase.org/). The “best hit ortholog” for each A. aegypti transcript was retrieved. A. gambiae orthologs were subjected to a Database for Annotation, Visualization, and Integrated Discovery (DAVID, version 6.7) functional clustering analysis (Niaid, 2006). Only those functional clusters with an enrichment value >1.3 (corresponding to P < 0.05) were retrieved and recorded for further analysis.

Two cDNA libraries for each group were sequenced, resulting in 26.3 and 26.2 million reads respectively for the adult MT libraries. All raw sequence data generated were submitted to the sequence read archive (Leinonen et al., 2011) and accepted under accession number SRX1884160 for larvae and SRX539939 for adult. Sequencing of larval libraries resulted in 31.0 and 31.2 million reads. Seventy-three to seventy-six percent of the reads in each library successfully aligned to the A. aegypti reference transcriptome. We calculated the Pearson correlation coefficient for our sample repeats. Coefficients between adult: adult and larval: larval libraries were 0.996 and 0.991, respectively. Our replicates exhibited a high degree of similarity, meaning that the transcriptome of A. aegypti Malpighian tubules was stable across biological repeats and not highly variable (Bonizzoni et al., 2011). The Pearson correlation between adult and larval libraries was 0.732, indicating potential changes in transcript expression between larval and adult MT. Figure 1 shows a volcano plot visualizing the differences in gene expression between the larval and adult transcriptomes. The overall sequence results are shown in Table S1.

Due to the vastly different osmoregulatory requirements of aquatic larvae and terrestrial adults, we expected to see significant differences between the transcript expression in their MT.

The pathway analysis performed using DAVID v6.7 (Huang et al., 2009) showed a total of 23 functional clusters among the transcripts up-regulated in adult MT, and 10 clusters among the transcripts down-regulated in adult MT (Table 1). Clusters were then manually categorized based on their putative biological role: transcription and translation (9 clusters), cell signaling and voltage-gated channels (10 clusters), protein sorting and trafficking (5 clusters), organelle sorting and trafficking (7 clusters), cell signaling (1 cluster), and redox and detoxification (1 cluster; Table 1).

The DAVID functional cluster analysis identified four groups associated with ion transport: “voltage-gated channel activity,” “transmembrane,” “ligand-gated ion channel activity,” and “voltage-gated potassium channel activity” (Table 1). Other transcripts associated with epithelial ion transport are also found within the “cell signaling and voltage-gated channels” clusters (Table 1). In addition, a manual search was performed to find transcripts associated with the current model of transepithelial fluid secretion in MT (Hine et al., 2014; Piermarini, 2016). We were able to identify at least one member of each gene family represented in this model in both larval as well as adult MT transcriptomes. Figure 2 shows the relative expression levels of some model-associated transcripts. This model includes sodium/potassium ATPases (Na/K ATPase), a chloride/bicarbonate anion exchanger (Cl/HCO₃), sodium/proton antiporters (NHA1, NHA2), a sodium/proton exchanger (NHE3), inward rectifier potassium channels (Kir1 & 3), a sodium channel (NaC), sodium/potassium/chloride cotransporters (NKCC), chloride channels (ClC), potassium/chloride cotransporters (KCC), and V-ATPase subunits. Our analysis resulted in the identification of several novel paralogous transcripts expressed in the MT and also allowed the identification of some genes that are dominantly expressed. We found that many transcripts show life stage-dependent expression. Most transcripts that we analyzed had a greater level of expression in the larval MTs with only a few exceptions. For example, the putative sodium channel (NaC) AAEL014228, the only one of three NaC genes that is highly expressed in MTs, is 3.1 fold increased in adult relative to larval MTs, suggesting that it may play a role in the post-prandial diuresis of females.

We found that Kir1 (AAEL008932) was up-regulated in adults while Kir3 (AAEL001646) was up-regulated in larvae which confirms qPCR results published earlier (Yang et al., 2017). In adults, Kir1 is expressed on the basolateral membrane of stellate cells where it contributes to most of the transepithelial secretion of K⁺ (Piermarini et al., 2015). On the other hand, Kir3 is expressed in intracellular compartments of principal and stellate cells in adult MT, but its physiological role is unknown (Piermarini et al., 2015).

NHA2 (AAEL000137) was highly up-regulated in larvae (6.46x), consistent with a putative role in the absorption of luminal Na⁺ as proposed by Xiang et al. (2012).

The NKCC transcripts AAEL009888 and AAEL006180 were not differentially expressed.

Catalytic subunit A of V-ATPase (AAEL008787) was 2.96x higher in larvae, while many other subunits show moderate up-regulation in larvae (see Table 6). This is in contrast to microarray data from A. gambiae which indicates that in this mosquito species V-ATPase subunits are generally higher expressed in adults compared to larvae (Overend et al., 2015).

These results suggest there may be potential differences in transport rates of certain ions between larval and adult MT, due not only to changes in expression levels but also of transporter types which may have different kinetics.

As mentioned earlier, we found a transcript encoding the alpha-subunit of the sodium/potassium ATPase (AAEL008787) to be 2.96x more abundant in MTs of larvae vs. adults (Figure 2). The Na/K-ATPase has recently been considered to play an important role in transepithelial fluid secretion by MT of A. aegypti (Hine et al., 2014). Immunoreactivity for the alpha subunit localizes to the basolateral membranes of stellate cells in the distal segment and principal cells in the proximal segment (Patrick et al., 2006), and ouabain significantly inhibits K⁺, Cl⁻, and fluid secretion in isolated MTs (Hine et al., 2014). Thus, our finding is consistent with a potentially important role of this ATPase in transepithelial fluid and cation secretion by adult MTs.

The larval challenge is to retain NaCl and KCl and to remain hypertonic by excreting a dilute urine, in the face of a constant hypotonic external force. In contrast, the adult female mosquito is challenged with the sudden need to excrete a large amount of fluid following a blood meal. The larvae need to get rid of water constantly and retain ions, whereas the adult female needs to retain water to prevent dehydration until the moment she takes a blood meal. At that point she needs to get rid of water and ions very quickly (Patrick et al., 2001; Beyenbach, 2003; Beyenbach and Piermarini, 2011). Thus, the transcriptome of adult female MTs is likely primed with molecular mechanisms to fulfill her diuretic needs in response to a blood meal.

The model hypothesizes the existence of a basolateral sodium channel (NaC), which allows the entry of Na⁺ into the principal cells from the hemolymph, and an apical chloride channel, which allows the movement of Cl⁻ from the stellate cells to the tubule lumen (Figure 2). In our analysis, we found one NaC (AAEL014228), and two ClC's (AAEL005950, AAEL001752) to be highly expressed in adult tubules (Figure 2). These two ClCs are orthologous to ClC-a and ClC-b in Drosophila, respectively. ClC-a (2x increase) and b (1.1x increase) were found to be more highly expressed in adults. ClC-a is known in Drosophila to be exclusively expressed in stellate cells and responds to diuretic hormone-induced secretion, while ClC-b is a housekeeping channel (Cabrero et al., 2014). There are three ClC-type channels in Drosophila, whereas in A. aegypti there are seven. The large expansion of genes in A. aegypti is potentially due to differences in selective pressure as a result of different feeding habits of fruit flies vs. mosquitoes (Wang et al., 2004).

Notably, a potassium-dependent sodium-calcium exchanger (NCKX1) was very highly expressed in larval and adult MT (AAEL004805). An ortholog of this transcript was also highly abundant in the MT of adult female A. albopictus and A. gambiae (Overend et al., 2015; Esquivel et al., 2016). NCKX1 may be associated with the response of MTs to kinins. That is, after stimulation with kinin peptides, there is an increase of intracellular Ca⁺², which stimulates the secretion of Cl⁻ (Yu and Beyenbach, 2002). A basolateral Na⁺/Ca²⁺ exchanger could provide a mechanism for lowering intracellular [Ca²⁺] after the influx of Ca²⁺ or maintaining low intracellular [Ca²⁺] before activation by kinins.

These previously unstudied transporters in A. aegypti represent potential mechanisms for disrupting diuresis, and exploiting as targets for novel insecticides development (Raphemot et al., 2013).

It is still unclear how the MT transport water from the hemolymph to the tubule lumen, but paracellular transport through septate junctions and transcellular transport through aquaporins are the most likely routes. RNAi studies by our group suggest that aquaporins play a major role (Drake et al., 2010, 2012; Benoit et al., 2014). In the present study, we found reads aligned against all six currently known aquaporins in the A. aegypti genome, as well as the B and C isoforms of aquaporin 5 (Eglp2) in both samples (Table 7). The pattern of expression observed for each aquaporin in the MT was very similar to the results of Drake et al. with aquaporins 1(Drip), 2 (Prip), 4 (Eglp1), and 5 (Eglp2) being the most highly expressed, while 3 (Bib) and 6 (Aqp12L) are expressed at very low levels (Drake et al., 2010). We found that aquaporins are generally upregulated in larvae with the exception of AQP6. However, our statistical analysis did not show significant changes except for AQP 1 and 3 (P < 0.05). These results may indicate that AQP expression overall is higher in MT of larvae vs. adults and that transcellular water transport is more important in MT of larvae than in those of adults.

In addition to maintaining fluid balance, the Malpighian tubules are a major site of detoxification and excretion of waste products, from pesticides to excess nitrogen and iron, through mechanisms such as cytochrome P450's and glutathione S-transferases (Folwell et al., 2006; Yang et al., 2007; Dow, 2009; Chahine and O'Donnell, 2011). We investigated which of these detoxification mechanisms are present by examining the expression of transcripts and comparing adult and larval expression.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.