The TRI Reagent^® (Cat: T9424) was purchased from Sigma-Aldrich Company (St. Louis, MO, United States). Qubit ssDNA Assay Kit (cat: Q10212) and SYBR^TM Green PCR Master Mix (cat: 4309155) were obtained from Thermo Fisher Scientific (Waltham, MA, United States). QuantiTect Reverse Transcription Kit (Cat: 205310) was obtained from Qiagen (Hilden, Germany). Sheep blood with Alsever’s anticoagulant (Cat: SB068) was purchased from TCS Biosciences (Buckingham, United Kingdom). Other reagents used in this work were analytical grade.

Sequences from GH13 and GH31 families from five different insect species – Aedes aegypti (AAEL006719, AAEL000647, AAEL008502, AAEL010602, AAEL009838, AAEL017226, AAEL022548), Culex quinquefasciatus (CPIJ005064, CPIJ007333, CPIJ011854, CPIJ001201, CPIJ009306), Anopheles gambiae (AGAP012230, AGAP010428, AGAP001200, AGAP001534, AGAP000862), Drosophila melanogaster (FBpp0311600, FBpp0071525, FBpp0070061, FBpp0084221), and Drosophila ananassae (FBpp0115773, FBpp0115136, FBpp0345570) – were retrieved from Vector Base¹, Fly base², and NCBI³ (Cantarel et al., 2009; Gabriško, 2013). These sequences were employed to perform ClustalW alignment and HMMER search in the Vector Base database to find similar sequences in the L. longipalpis genome (conserved domains on Lutzomyia longipalpis, LlonJ1.4, last updated 27 June 2017). Sequences recovered from L. longipalpis were annotated by similarity with proteins and conserved domains using the BLASTp (ref-seq protein) and Swiss-Prot/UniProt databases. Manual annotation was performed with the Apollo annotation tool (Dunn et al., 2019) and ClustalW (Thompson et al., 1994). Sequences were considered complete when initial methionine, correct exon/intron junction, stop codon were identified, and exon structures were complete based on the alignment with orthologous genes.

After sequences identification, characteristics like putative signal peptide (SignalIp) (Petersen et al., 2011), transmembrane domain (THMM) (Krogh et al., 2001), GPI-anchor (Big-PI predictor) (Eisenhaber et al., 1998), protein subcellular localization (DeepLoc) (Almagro Armenteros et al., 2017), N-glycosylation (NetNGlyc) (Gupta et al., 2002), O-glycosylation (NetoGlyc) (Steentoft et al., 2013), molecular mass and isoelectric point (pI) (Compute pI/Mw) (Bjellqvist et al., 1994; Gasteiger et al., 2005) were predicted in silico. The N-glycosylation sites were not considered if predicted inside transmembrane or cytoplasmic domains of membrane proteins. Genes and proteins representative structures were drawn using PROSITE MyDomains tool⁴ (Hulo et al., 2008).

The InterPro tool (Finn et al., 2017) was used to predict catalytic sites and, for α-amylases, also the calcium-binding site. The catalytic sites predicted by InterPro were confirmed by local alignment using ClustalW tool by comparing the L. longipalpis sequences with sequences of crystallized proteins deposited in the UniProt databank, and the analysis was based on conserved regions (CSRs) (Janecek, 1997; Kashiwabara et al., 2000; Lovering et al., 2005). This comparison was also used to predict the chloride binding sites in α-amylases. For GH13, the templates were the sequences of α-amylase from Aspergillus oryzae (Uniprot P0C1B3) (Toda et al., 1982), maltase from Apis cerana japonica (Uniprot A1IHL0) (Wongchawalit et al., 2006), amino acid transport protein (UniProt Q07837) (Bertran et al., 1993) from Homo sapiens, glucan branching enzyme from E. coli (UniProt P07762) (Baecker et al., 1986), and glycogen debranching enzyme from Candida glabrata (UniProt Q6FSK0) (Zhai et al., 2016). For GH31, the templates were the sequences of neutral α-glucosidase from Homo sapiens (UniProt O43451), lysosomal α-glucosidase from Homo sapiens (UniProt P10253) (Roig-Zamboni et al., 2017), and glycosidase NET37 from Homo sapiens (UniProt Q6NSJ0).

Using each of the GH13 and GH31 sequences retrieved from L. longipalpis genome as a query, we searched for orthologous proteins in six other insect species. Sequences from Aedes aegypti (LVP_AGWG strain, AaegL5.1 geneset), Anopheles gambiae (PEST strain, AgamP4.9 geneset), Culex quinquefasciatus (Johannesburg strain, CpipJ2.4 geneset) and Phlebotomus papatasi (Israel strain, PpapI1.4 geneset) were retrieved from Vector Base⁵. Sequences from Drosophila melanogaster and Drosophila ananassae were retrieved from Fly base⁶. For Phlebotomus papatasi, sequences were searched and annotated as described above at “Search for GH13 and GH31 sequences in L. longipalpis genome.” The orthologous sequences used in this analysis are described in Supplementary Tables 1,2.

For phylogenetic analysis, sequences were aligned using Clustal Omega algorithm (version 1.2.2) (Sievers et al., 2011; Sievers and Higgins, 2018), after removing the signal peptide. The maximum likelihood phylogenetic tree was obtained using MEGAX software (Kumar et al., 2018) using partial deletion (90% site coverage cutoff). The bootstrap consensus tree was inferred from 500 replicates with cutoff of 50% for bootstrap values for the nodes.

The experiments were performed using L. longipalpis (from Jacobina, Bahia, Brazil), maintained at Lancaster University (United Kingdom). Insects were kept under standard laboratory conditions of temperature (24 ± 2°C), and a photoperiod of 8 h of light/16 h of darkness. Adult sandflies were fed with 70% (w/v) autoclaved sucrose solution in cotton wool ad libitum, unless stated differently in the experiments. For oviposition, females were fed with sheep blood containing Alsever’s anticoagulant. For this, an artificial apparatus (Hemotek – Discovery Workshops, United Kingdom) at 37°C was used for 1 h using chicken skin as a membrane. Engorged females were transferred to rearing containers with a piece of cotton wool soaked in sugar solution. The eggs were separated from dead females after oviposition and reared to preserve the colony. For experiments, recently emerged females (0–3 h) were fed with 1.2 M sucrose for 2 days (SF females), with a piece of cotton wool soaked in the sugar solution, or were maintained for 3 days with a cotton wool soaked in water, before feeding on sheep blood. After blood-feeding (infected or not), they were kept with a piece of cotton wool soaked in water for 2 days.

For infections, axenic cultures of amastigotes from L. mexicana strain M379 were used. For maintaining parasites in the amastigote form, they were incubated at 32°C with Grace’s insect medium supplemented with 20% of fetal bovine serum, BME₁ vitamins, 2% human urine, and 25 μg/mL gentamycin sulfate. Parasites with a maximum of 26 passages in the amastigote maintaining medium, after isolation from the vertebrate host, were used for sandfly infections (Moraes et al., 2018). Infections were performed using a parasite concentration of 2 × 10⁶ parasites/mL, estimated using Neubauer chamber. Briefly, after centrifugation for 5 min at 2000 × g, the supernatant was removed, and parasites were mixed on sheep blood and offered to 4-day old unfed females as described in section “Maintenance of Insects.” Unfed females were discarded, and control insects were fed with uninfected blood. After blood-feeding, infected and control females were maintained as described above with water for 2 days.

Newly hatched females (0–3 h) (NF), females fed with 1.2 M sucrose for 2 days (SF females), control females fed with uninfected blood (BF females) and infected females (BF_INF females), 2 days after blood feeding, were dissected in sterile phosphate-buffered saline. Pools with 10 guts or rest of the body (RB) were transferred to polypropylene vials containing 50 μL of TRI Reagent^® (Cat. No. T9424, Sigma-Aldrich, Darmstadt, Germany) and kept at -80°C until further RNA extraction. For infected females, midgut was checked with light microscopy for parasite presence and only midguts containing parasites were used.

Total RNA was extracted following the supplier’s protocol and quantified using NanoDrop^® (NanoDrop Technologies, Wilmington, United States). The cDNA was synthesized by the Reverse Transcriptase (RT) reaction using the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. cDNA was quantified using Nanodrop^® and normalized to a concentration of 50 ng/μL.

GH13 and GH31 coding sequences identified in the L. longipalpis genome were used to design specific oligonucleotides (Thornton and Basu, 2011), using the Primer3 program (Rozen and Skaletsky, 2000), Beacon Designer^TM Free Edition⁷, and mFOLD Software⁸. The list of primers is described in Supplementary Table 3. Gene expression was initially analyzed by RT-PCR, and at least one gene representative of each cluster (A, B, or C) of the GH13 α-amylase group (see section Results, Figure 3) was analyzed. Genes LLOJ000566/LLOJ008156 (LlAglu1/LlAglu2), LLOJ002257 (LlAglu3), LLOJ004841 (LlAamyA4) and LLOJ003489 (LlNAglu2), were additionally analyzed by RT-qPCR. Samples obtained from the gut and the rest of the body of recently emerged, sugar, and blood-fed females were used for RT-PCR or RT-qPCR analysis. Samples from females infected with Leishmania mexicana parasites were used only for the RT-qPCR analysis.

For RT-PCR, we carried duplicate reactions for each cDNA sample obtained from a tissue pool of 10 sand flies. Three biological replicates were carried for each group, with two cDNA samples obtained for each condition tested. Each amplification reaction was performed in a volume of 25 μL containing 2× Biomix Red (Bioline, London, United Kingdom), 100 ηg cDNA, and 0.2 μM of each primer. The PCR parameters were: incubation at 95°C for 5 min, followed by 26 cycles of 94°C for 30 s, 55°C for 40 s, 72°C for 1 min, and a final incubation of 72°C for 5 min. Relative expression was normalized using a housekeeping gene (LLOJ001891, Glyceraldehyde 3-phosphate dehydrogenase, GAPDH) as control. RT-PCR products were analyzed by electrophoresis in a 1% (w/v) agarose gel containing gel red^® (Biotium, Fremont, United States). Expression level was determined by densitometry of bands using the software ImageJ (Schneider et al., 2012).

Reverse transcriptions followed by quantitative polymerase chain reactions (RT-qPCR) were conducted using a CFX96 Touch^TM Real-Time PCR Detection System (BioRad). We carried duplicate reactions for each cDNA sample obtained from a tissue pool of 10 sand flies. Three biological replicates were carried for each group, with two cDNA samples collected for each condition tested. The reaction was performed in 10 μL volume containing 2× of SYBR^TM Green Master Mix (Thermo Fisher Scientific), 20 ηg cDNA, and 0.5 μM of each primer. The parameters for PCR were: incubation at 95°C for 15 min, followed by 39 cycles of 94°C for 15 s, 56°C for 30 s, and 72°C for 30 sec. As a negative control, PCR reactions were carried out without cDNA template to assess primer dimer formation or contamination in the reactions. To ensure that only a single PCR product was amplified, an analysis of the melting curve was performed. The expression of target genes in each tissue was quantified by the comparative Ct (ΔCt) method (Schmittgen and Livak, 2008) normalized with the GAPDH housekeeping gene as an internal control.

For multiple comparisons, one-way ANOVA was used, followed by Tukey’s multiple comparison tests, and significance was considered when p < 0.05. Results are expressed as the means ± SEM. All statistical analysis were executed using the Software GraphPad Prism 6.0 (San Diego, CA, United States).

Proteins belonging to GH13 and GH31 are evolutionary conserved, being found in many classes of organisms. A total of 15 sequences belonging to the GH13 and four sequences from GH31 were found in the L. longipalpis genome, based on HMMER search in the Vector Base platform. For this, we used GH13 and GH31 know sequences of 5 dipteran species (Aedes aegypti, Culex quinquefasciatus, Anopheles gambiae, Drosophila melanogaster, Drosophila ananassae) as queries (Supplementary Tables 1,2).

Sequences of proteins found were classified based on similarity to other protein sequences and conserved domains using the BLASTp (ref-seq protein) and Swiss-Prot/UniProt databases according to their best hit. Results indicated that only 3 sequences from GH13 (LLOJ004841, LLOJ008156, LLOJ008629) and 1 from GH31 (LLOJ003489) were complete, considering the presence of initial methionine, stop codon, correct exon/intron junctions, and similar size of the exons when compared to orthologous proteins. Truncated sequences were analyzed and annotated for missing parts using the Apollo annotation tool and alignment by ClustalW. The transcriptomic data present in the Vector Base was also used to predict the correct size of exons. Modifications for each of the sequences retrieved are described in Supplementary Table 4. Some sequences could not be completed because they were in a region of poor sequencing quality or at the beginning or end of the scaffold. The nucleotide and protein sequences after annotation are presented in Supplementary Files 1 and 2, respectively.

Some nucleotide sequences retrieved were coding for more than one protein belonging to GH13 or GH31. These sequences were split up, so the number of genes and, consequently, proteins increased. From sequences initially retrieved from Vector Base, genes LLOJ004838, LLOJ004880, LLOJ004881, LLOJ005909 codified for four different alpha amylases. After annotation, it was possible to determine that these genes actually codify for ten different alpha-amylases. Gene LLOJ004838, for example, codified for two different alpha-amylases that seem to be transcribed at different moments (corroborated by transcriptomic data from Vector Base). Protein1 (LlAamyA1) was transcribed in both larvae and adults and Protein 2 (LlAamyA2) was transcribed only in immature forms. Annotation resulted in 21 complete sequences for GH13 and 6 for GH31. With these analyses, we identified 14 α-amylases, tree α-glucosidases, two amino acid transport proteins (heavy chain), one 1,4-α-glucan branching enzyme, and one glycogen debranching enzyme from GH13 (Supplementary Table 5). For GH31, we identified four glycosidases NET37 (nuclear envelope transmembrane glycosidase 37), one lysosomal α-glucosidase, and one neutral α-glucosidase (α-subunit) (Supplementary Table 6). Gene structures are demonstrated in Supplementary Figure 1, and protein structures are represented in Figure 1.

According to the genomic map available in the Vector Base, the α-amylases genes were organized in two clusters. LlAamyA1 to LlAamyA4 (4 α-amylases) composed the cluster A, and the cluster B was formed by LlAamyB1 to LlAamyB8 (8 α-amylases). Besides that, the α-amylases C1 and C2 were in tandem in the genome. All the putative α-amylases were putative soluble extracellular proteins, with a predicted molecular mass from 54.6 to 56 kDa, and estimated isoelectric points from 4.4 to 6.4 (only complete sequences were analyzed, see Supplementary Table 5). α-glucosidases were predicted as membrane proteins. LlAglu1 and LlAglu2 were putative transmembrane proteins, and for LlAglu3, both transmembrane domain and GPI anchor were predicted. These enzymes had molecular masses around 70 kDa and estimated isoelectric points from 4.7 to 5.6 (Supplementary Table 5). A lysosomal membrane α-glucosidase (LlLysAglu1, LLOJ000840) with 76.3 kDa and a soluble α-glucosidase (LlNAglu1, LLOJ003489), putatively active in the endoplasmic reticulum, were identified in the GH31 family (Supplementary Table 6).

Analysis with InterPro and amino acid sequences alignments of L. longipalpis with GH13 and GH31 representative sequences from different organisms allowed the identification of highly conserved regions, catalytic residues, and the calcium and chloride binding sites for α-amylases (Tables 1, 2). The active site containing the catalytic residues, the calcium-binding site, and the chloride-binding site (Supplementary Table 5 and Table 1) were correctly identified for α-amylases sequences belonging to cluster A and also for LlAamyB7 and LlAamyC2. The α-amylase B1 (LlAamyB1/LLOJ004880_1) had not a typical catalytic triad and presented only two residues that might serve as the nucleophile and the acid/base catalyst (D204 and D241, respectively). Moreover, this enzyme also showed two substitutions: i) one in the calcium-binding site, a conserved histidine residue by asparagine, and ii) one in the chloride-binding site at the β4 sheet, a conserved arginine by leucine (Supplementary Table 5 and Table 1). Some of the α-amylases might not function as active enzymes, since they lacked the correct nucleophile and the acid/base catalyst, as α-amylases B4 and B6 (LlAamyB4/LLOJ004881_1 and LlAamyB6/LLOJ004881_3).

The catalytic triad was D216, E290, D357 for both α-glucosidases 1 and 2 (LlAglu1, LLOJ000566 and LlAglu2, LLOJ0008156), and D236, E304, D375 for α-glucosidase 3 (LlAglu3, LLOJ0002257) with all highly conserved regions preserved (Supplementary Table 5 and Table 1).

The amino acid transport proteins (heavy chains) found in our screening shared a domain similarity with alpha-glucosidases, but the general protein primary structure was different (Figure 1), and they lacked the conserved catalytic residues (Table 1). By similarity, LLOJ008629 was described as “neutral and basic amino acid transport protein” (NBAThc), while LLOJ006803, due to the presence of the SLC3A2 domain, was classified as CD98hc (cluster of differentiation 98) (Supplementary Table 5 and Table 1).

The sequence of a 1,4-α-glucan branching enzyme (LlAGB1, LLOJ005533) had a central domain containing the active site, and the catalytic residues similar to the other enzymes from GH13. In the putative sequence of a glycogen debranching enzyme (LlGDE1, LLOJ008312), we identified both α-1,4 glucanotransferase (GT) and α-1,6 glucosidase (GC) domains, this enzyme was a soluble cytoplasm enzyme with an estimated size of 172.6 kDa. The GT domain at the N-terminal region of this protein, despite the low overall sequence identity, had similarity with five short conserved regions of the GH13; the catalytic triad was conserved with the D548 as the catalytic nucleophile, E577 as the proton donor, and the conserved residue D649. The C-terminal GC domain was similar to the catalytic domain of glucoamylases and other members from GH15 family. The general acid and general base for GC activity were identified as D1282 and E1515, respectively (Supplementary Table 5 and Table 1).

Regarding the members of GH31 family, the neutral α-glucosidase (α-subunit) (LlNAglu1, LLOJ003489) was described with both Galactose mutarotase domain at the N-terminal end and the GANAB_GANAC domain (Figure 1C). This enzyme shared a 46.8% identity with the human neutral α-glucosidase (GANB_HUMAN) and was predicted to be localized at the endoplasmic reticulum. The two aspartic residues involved in the catalytic domain were identified as D522 and D598 (Supplementary Table 6 and Table 2). The lysosomal α-glucosidase (LlLysAglu1, LLOJ006451) presented a substitution of both catalytic aspartate residues by glutamine (Supplementary Table 6 and Table 2). The glycosidases NET37 were predicted to have a single transmembrane domain at N-terminus and the GH31_NET37 domain at its C-terminus. The glucosidase domain was facing the non-cytoplasmic region (Figure 1C). Since these proteins are dependent on the catalytic residues to be active, glycosidase NET37 1 (LlGlyMyo1, LLOJ001847_1) and NET37 2 (LlGlyMyo2, LLOJ001847_1), seemed to lack hydrolytic activity. The proton donor residue was substituted for proline (Supplementary Table 6 and Table 2). Although the sequence for glycosidase NET37 3 (LlGlyMyo3, LLOJ001881) was not complete, it was possible to predict the two catalytic residues, 401D and 460D. Glycosidase NET37 4 (LlGlyMyo4, LLOJ000840) had glutamate as the conserved catalytic nucleophile, E408, and D467 as a proton donor.

A phylogenetic tree was constructed, aiming to support the classification and prediction of the functional role of the proteins. Using the L. longipalpis GH13 and GH31 protein sequences, we identified orthologs in other dipteran species and verified if gene amplification might have occurred in these GH families in the Phlebotominae branch. Phylogenetic analysis of GH13 protein sequences demonstrated the occurrence of 5 clades, each of them representing one of the activities described in this family (Figure 2). Within each clade, Phlebotominae proteins are more closely related to each other than to their orthologs in other dipteran species (Figure 2).

Clade I represented the α-amylases, and one of the subgroups of this clade had representatives only from the Phlebotominae subfamily. This subgroup was composed by 10 of 14 (71%) α-amylases of the L. longipalpis genome, suggesting a gene expansion in this protein family that included three sequences from cluster A (LlAamyA1, LlAamyA2, LlAamyA3) and seven sequences from cluster B (LlAamyB1, LlAamyB2, LlAamyB4, LlAamyB5, LlAamyB6, LlAamyB7, LlAamyB8) (Figure 2). The glucan branching enzymes were represented by a monophyletic group in clade II, demonstrating the sequence conservation among dipteran species. Enzymes from L. longipalpis and P. papatasi are grouped with the D. melanogaster and D. ananassae (Figure 2). Clade III represented the debranching enzyme, and was formed by a paraphyletic group, since the Phlebotominae subfamily, Drosophilidae, and Culicidae families established different branches (Figure 2). Clade IV was composed by the amino acid transport proteins, and formed a paraphyletic group divided into two subgroups. One subgroup represents the neutral and basic amino acid transporters (NBATs), and the other subgroup, the CD98 heavy chain, comprised the large neutral amino acid transport (LAT1). NBATs proteins missed the SLC3A2 domain, which is characteristic of CD98hc. The LlCD98hc grouped with orthologs from the Drosophilidae family, already classified as different isoforms of the CD98 heavy chain (Figure 2).

Clade V represented the α-glucosidase proteins, and L. longipalpis α-glucosidases were grouped with P. papatasi orthologs. Two subgroups had representatives only from the Drosophilidae family (Figure 2). As described in Table 3, P. papatasi seemed to have an expansion in the α-glucosidase genes when compared to L. longipalpis. One of the subgroups of this clade had representatives only from P. papatasi. Interestingly, the PpAgluA3 is ortholog to the CqMal1 (Figure 2), an enzyme from C. quinquefasciatus characterized as GPI membrane-bound α-glucosidase with toxin binding properties (do Nascimento et al., 2017).

Phylogenetic analysis of GH31 protein sequences demonstrated the occurrence of 4 clades representing each activity described in this family (Figure 3). Once more, within each clade, Phlebotominae proteins related more closely to each other when compared to their orthologs in other dipteran species, except for clade IV, because it had no representative from P. papatasi (Figure 3). Clade I and II represented the lysosomal and the neutral alpha-glucosidases, respectively. As described in Table 2, neutral alpha-glucosidases maintained their catalytic residues. The LlLysAglu1 sequence had no conserved catalytic residue, and the phylogeny analysis grouped it with lysosomal enzymes from other dipteran species that also had not catalytic residues (Table 2 and Figure 3). The other two clades, III and IV, contained the glycosidases NET37. For NET37 glycosidases, clade III represented proteins containing the conserved catalytic domain, while clade IV represented glycosidases without the catalytic residues (Table 2 and Figure 3).

When we compare the number of genes for GH13 and GH31 in the L. longipalpis genome with their orthologs in 6 dipteran species, a potential expansion of α-amylases genes was observed in L. longipalpis (Table 3). The total of 14 α-amylase sequences found was 75% and 16% higher when compared to Phlebotomus papatasi and Culex quinquefasciatus, respectively, and almost 3-fold the number of α-amylases found in the representatives of the Drosophilidae family. Interestingly, a contraction occurred for α-glucosidases genes in L. longipalpis, when compared to the other six species (Table 3). The number of genes coding for α-1,4-glucan branching and glycogen debranching enzymes (GDE) was conserved among the species analyzed (Table 3).

For GH31 family, the number of lysosomal and neutral α-glucosidase was conserved in phlebotomine and mosquitoes. Drosophila species had no lysosomal α-glucosidases. Glycosidase NET37 demonstrated a variable number of copies among the species analyzed and seemed to be duplicated in L. longipalpis when compared to P. papatasi (Table 3).

After analyzing the sequences using bioinformatics tools, primers were designed for selected sequences in the GH13 family and all sequences of GH31. In the case of α-amylases, we analyzed at least one representative for each phylogenetic cluster of Figure 2. Also, it was not possible to design specific primer pairs to independently analyze expression of genes coding for LlAglu1/LlAglu2 or LlMyoGly1/LlMyoGly2, due to the high sequence identity between these sequences.

Primers amplified all sequences producing only one amplicon with the expected size from genomic DNA (data not shown). The expression of the L. longipalpis GH transcripts was evaluated by RT-PCR using different tissues (rest of body or gut) as sources of RNA, and at different dietary conditions, namely newly hatched females, sugar or blood-fed females.

All tested tissues and conditions poorly expressed both LlGlyMyo4 and the lysosomal α-glucosidase (data not show). The LlAamyA1 and LlAamyA4 were the α-amylases with the highest levels of expression in the gut when compared to other α-amylases genes independently of dietary condition tested (Figure 4A). No significant difference was found for LlAamyA1 expression levels when comparing different dietary conditions inside the same tissue. However, when comparing different tissues, LlAmy1 was significantly more expressed in the gut of sugar-fed females (one way ANOVA, ^∗∗p < 0.01) and blood-fed females (One way ANOVA, ^∗p < 0.05) than in rest of the body (Figure 4A). LlAamyA4 expression was higher in the gut for all dietary conditions tested (One way ANOVA, ^****p < 0.0001) and unaltered considering the dietary condition inside the same tissue (Figure 4A). Interestingly, blood-fed females did not express LlAamyB2. LlAamyB4 was poorly expressed in all tested conditions, while the expression of LlAamyC2 only occurred in the rest of the body (Figure 4A). α-glucosidase transcripts (LlAglu1/LlAglu2 and LlAglu3; Figure 4B) presented an expression pattern similar to LlAamyA1 and LlAamyA4, and no significant differences were found in expression levels when comparing different dietary conditions inside the same tissue. For LlAglu1/LlAglu2, the level of expression in the gut was higher for sugar-fed (One way ANOVA, ^∗∗p < 0.01) and blood-fed females (One way ANOVA, ^****p < 0.0001), when compared to the rest of the body (Figure 4B). LlAglu3 expression was higher in the gut for all dietary conditions tested (One way ANOVA, NF ^∗∗p < 0.01 and SF/BF ^∗∗∗p < 0.001) (Figure 4B).

The expression of the α-1,4-glucan branching enzyme and the amino acid transporters is not linked to a specific tissue, and LlNBAThc and LlCD98hc presented no significant changes in expression (Figure 4C) in different feeding conditions. The expression of LlAGB1 (Figure 4D) was reduced when females feed on blood.

For the GH31, blood-fed females did not express glycosidases NET37 in any tissue. LlGlyMyo1/LlGlyMyo2 were expressed in the gut and carcass of newly emerged and sugar fed insects, but with no significant differences among these samples (Figure 4E). LlGlyMyo3 was also expressed in these same tissues and conditions, but it was more expressed in the gut when compared to the rest of body (p < 0.05, Figure 4E). There were no significant differences in the expression of LlGlyMyo3 in the gut and rest of body when comparing newly emerged flies with sugar-fed ones (Figure 4E). LlNAglu1 was significantly more expressed in the gut when compared to the rest of body, and its expression in each tissue was not affected by the feeding conditions (Figure 4E).

Since we have an interest in sandfly proteins that might be affected by Leishmania, we selected the genes coding for LlAamyA4, LlAglu1/LlAglu2, LlAglu3 (GH13) and LlNAglu1 (GH31) for further quantitative expression analysis after L. mexicana infections. These genes were chosen considering their preferential expression in the gut, the organ which was in direct contact with the parasite during the experimental infections.

RT-qPCR analysis of the genes LlAglu1/LlAglu2, LlAglu3, LlAamyA4 (GH13), and LlNAglu1 (GH31) confirmed higher gene expressions in gut tissues when compared to the rest of the body, in all conditions tested (One way ANOVA, ^∗p < 0.05) (Figure 5). LlAglu1/LlAglu2 and LlAglu3 genes showed higher quantitative expression values when compared to the genes LlAamyA4 and LlNAglu1. Expression values for LlAglu1/2 and LlAglu3 were about 40 X higher than LlAamyA4 and 10 X higher than those observed for LlNAglu1 (Figure 5).

Interestingly, the expression levels of the putative α-glucosidases LlAglu1/2 and LlAglu3 were significantly affected by the blood diet and Leishmania infection (Figure 5). We observed an induction in the expression levels of LlAglu1/LlAglu2 (1.14 ± 0.06) and LlAglu3 (1.2 ± 0.1) in the gut of blood-fed females in comparison to non-fed or sugar-fed females (One way ANOVA, ^****p < 0.0001; Figure 5A). Also, infection with L. mexicana parasite modulated negatively both the expression of LlAglu1/LlAglu2 (0.4 ± 0.1) and LlAglu3 (0.6 ± 0.1) (Figure 5A) (One way ANOVA, ^****p < 0.0001). The sugar diet did not significantly affect expression of these genes when compared to the non-fed condition.

LlAamyA4 had the lowest level of expression among the genes studied and presented the same pattern obtained previously using semi-quantitative RT-PCR, with no significant changes after sugar or blood feeding, when comparing to newly emerged flies (Figure 5B). Besides that, LlAamyA4 expression was not affected by parasites in the blood diet (Figure 5B).

LlNAglu1 is not highly expressed in the gut tissue compared to the rest of the body, but there is a significant difference between these tissues for all dietary conditions tested (Figure 5C). The expression level of LlNAglu1 was also induced by the blood diet in both tissues in comparison to non-fed condition or sugar diet (One way ANOVA, ^****p < 0.0001; Figure 5C), but in this case the expression was not modulated by infection with L. mexicana parasites (Figure 5C).