Abdominal elasticity varies between male and female adult mosquitoes
Before selecting adult mosquitoes for AeDHPAAS knockdown experiments, male and female adult mosquitoes were first compared for their ability to handle abdominal injections. The abdomens of both male and female adult mosquitoes were flat before the injection of Ringer’s solution (Fig. 1a, d). After injection with Ringer’s solution, the abdomens of both male and female adults swelled (Fig. 1b, c, e, f). These initial results showed that female adults could survive after injection of 3–5 µL solution, but male adults could only survive after being injected with 0.3 ± 0.05 µL solution at most. Figure 1b shows a surviving adult male after injection with 0.23 ± 0.05 µL of Ringer’s solution. In contrast, Figs. 1e and 1f show surviving adult females after injection with 3 µL and 4 µL of solution, respectively. Based on the abdominal injection results, female adult mosquitoes were selected for subsequent RNAi experiments.
DHPAAS knockdown increases mortality and reduces blood meal volume
Adult female mosquitoes were collected 12 h after RNAi microinjection to analyze gene expression. Real-time PCR (qPCR) showed no significant difference between negative and blank controls, while the relative expression of AeDHPAAS in the RNAi group was significantly lower (75% lower, P < 0.001) than that in the control groups (Fig. 2a). Meanwhile, almost no adults in the control groups died after 4 µL of Ringer’s solution injection, but 50% (P < 0.0001) of the adults with AeDHPAAS knockdown failed to survive (Fig. 2b). The weight of adult mosquitoes before and after blood meal was measured, as shown in Fig. 2c, and no significant weight differences were observed between control and experimental groups. The average weight of each female adult was 1.7–1.8 mg (Fig. 2c). Figure 2d illustrates blood meal intake weight, the blood meal weight of adults in the control groups ranged from 3.7 mg to 4 mg, while the blood meal weight of adults with AeDHPAAS knockdown was only 2 mg (P < 0.0001).
DHPAAS knockdown disrupts the abdominal cuticle
After RNAi knockdown of AeDHPAAS, followed by oviposition, 40 ± 3.3% of the adult female mosquitoes died within three days (Fig. S1). In contrast, only 5% failed to survive in the control groups under matching conditions. Furthermore, 25 ± 5% of the surviving mosquitoes with knocked-down AeDHPAAS showed lowered tolerance to the abdominal injection of Ringer’s solution, with a higher frequency abdominal rupturing in mosquitoes with knocked-down AeDHPAAS (Fig. 3).
DHPAAS knockdown reduces egg hatching
RNAi knockdown of AeDHPAAS had no significant effect on the number of eggs, where each mosquito produced 100 ± 20 eggs (Fig. 4a). However, hatching of eggs was significantly inhibited by AeDHPAAS knockdown. The egg hatching rates of the control groups and AeDHPAAS‑dsRNA treated group were 90 ± 5% and 53 ± 28%, respectively (Fig. 4b). Approximately 60% of the non-hatching eggs of the AeDHPAASdsRNA treated group contained no larvae, but all eggs of the control groups contained developed larvae. These non-hatching eggs without larvae in the AeDHPAASdsRNA treated group rapidly dehydrated after reducing humidity from 20–10%; control group eggs did not dehydrate with decreased humidity (Fig. 4c). The remaining (~ 40%) non-hatching eggs of the AeDHPAASdsRNA treated group contained formed larvae and did not dehydrate; however, none of these eggs could hatch.
DHPAAS knockdown alters cuticular structure
During mosquito cuticle formation, the epithelium first secretes cuticulin and then the lamellate endocuticle. Finally, the exocuticle is derived by quinone tanning of the outer lamellae of the endocuticle 18. To observe the effect of DHPAAS knockdown on cuticle formation, transmission electron microscopy (TEM) was used to observe abdominal cuticular structures 12h after microinjection. The resulting TEM images showed that the endocuticle was thicker than the exocuticle in the control groups (Fig. 5a, b). In contrast, the AeDHPAAS RNAi group showed a thinner endocuticle layer relative to the exocuticle (Fig. 5c).
Differential gene expression after DHPAAS knockdown
Knockdown of AeDHPAAS resulted in obvious abnormal exfoliation and delayed development of Ae. aegypti larvae. To further explore the functional role of AeDHPAAS in larval development and molting, gene expressions of AeDHPAAS RNAi and control groups were focally analyzed using RNA-seq. Genes that were significantly differentially expressed between the experimental group and both negative and blank control groups could then be analyzed. After AeDHPAAS knockdown, the development of larvae was significantly inhibited 12. The differentially expressed genes involved in development were analyzed according to predefined pathways annotated by Kyoto Encyclopedia of Genes and Genomes (KEGG). As a result, 60 development-related genes were found to be differentially expressed after AeDHPAAS knockdown, of which only five genes were up-regulated while the rest were down-regulated (Fig. 6a).
Our previous research has shown that knockdown of AeDHPAAS can cause abnormal larval molting 12. Accordingly, the expression of genes related to cuticle proteins, chitin metabolism and lipid metabolism were altered after knockdown of AeDHPAAS. Compared with the control groups, 118, 107 and 110 genes related to cuticle proteins, chitin metabolism and lipid metabolism, respectively, were differentially expressed (Dataset S1) (Fig. 6b-d). Most of these differentially expressed genes showed trends of up-regulated expression. Nine out of 118 differentially expressed cuticle genes related to cuticle proteins were downregulated (Fig. 6b), 24 out of 107 genes involved in chitin metabolism were downregulated (Fig. 6c), and 27 out of 110 genes involved in lipid metabolism were downregulated (Fig. 6d). In addition, interesting changes in the expression of sterol carrier protein 2 (SCP-2) and lipophorin receptor (LpR) genes were observed in Ae. aegypti. Five out of eight SCP-2 genes were significantly upregulated and a single LpR gene was significantly downregulated (Fig. 6e).
Crystal structure of insect DHPAAS
To provide insight into the structure-function relationship and catalytic mechanism of DHPAAS, DmDHPAAS was purified and crystallized for X-ray crystallographic analysis. After extensive optimizations, the enzyme could finally be well-crystallized. The structure of DmDHPAAS was solved using molecular replacement based on the DmDDC structure (PDB ID 3K40) 17. The crystal structure was refined to 2.2 Å resolution with good crystallographic statistics (Table S1) and the structure was deposited and released (PDB ID 6JRL). According to analysis of the protein interfaces, surfaces and assemblies service 19, the DmDHPAAS crystal structure is a dimer (Fig. S2). The DHPAAS structural architecture resembles other type II PLP-containing enzyme structures. Similar to the DmDDC crystal structure, which contains a disordered region spanning residues 322–348, residues 328–341 of the DmDHPAAS crystal structure are disordered.
The DHPAAS active site is located at the monomer-monomer interface of the homodimer rebuilt based on the symmetry, and is mainly composed of residues from a single subunit; however, the Phe103- and Leu353-containing loops support the formation of the active site of the neighboring subunit. The active site Lys303 is observed to form a Schiff base with the PLP cofactor, and this internal aldimine is referred to as lysine-pyridoxal-5-phosphate (LLP) (Fig. 7).
Similar to most PLP-dependent enzymes, the LLP protonated pyridine nitrogen forms a salt bridge with the Asp271 carboxylate. Furthermore, the LLP pyridine ring is anchored by the Ala273 methyl group as well as the catalytic Asn192 amide group. The LLP pyridine ring hydroxy group is adjacent to the Thr246 side chain and two water molecules. The LLP phosphate moiety is further stabilized by multiple interactions with Ser149, Asn300 and His302 from within the same subunit, and is 5.0 Å from Leu353 of the neighboring subunit (Fig. 7a).
Structural comparison of DHPAAS and DDC
There is a significant difference regarding the loop containing the key active site residue Phe103, when comparing DmDDC and DmDHPAAS structures. As shown in Fig. 7b (right panel), Phe103 is relatively close to LLP and His192 in DmDDC; however, in DmDHPAAS the Phe103-containing loop is oriented away from the active center. His192 or Asn192 are important for determining DmDDC or DmDHPAAS catalytic activity, respectively 9, 20; where His192 is much closer to Phe103 (4.2 Å) in the DmDDC complex compared to the corresponding Asn192 distance in DmDHPAAS.
The DmDHPAAS region spanning residues 320–350 displayed a more defined structure as compared to that of DmDDC (Fig. 7b, left panel). In the insect DmDDC structure, residues 322–348 (26 residues) are disordered and cannot be observed; however, in the DmDHPAAS structure, the disordered region only includes residues 328–341 (13 residues). The stabilized 328–341 loop is observed to influence the Leu353-containing loop adjacent to the DHPAAS active site; DmDHPAAS Leu353 residue is positioned just 3.4 Å away from the catalytic Asn192 residue, whereas the corresponding Leu350 residue in DmDDC is oriented much further away from the active site (Fig. 7b, left panel). The DmDHPAAS Leu353-containing loop also contains Gly354 which has been reported to be important for substrate specificity towards dopa 21.
AeDHPAAS full-length model
To construct a valid model, the crystal structure of DmDHPAAS (PDB ID 6JRL, chain A) was selected as the most satisfactory template, with sufficient homology to AeDHPAAS (61% identify and 92% coverage). The MODELLER program was used to build a homology model of AeDHPAAS. The plotted Discrete Optimized Protein Energy (DOPE) score profile shows regions of relatively high energy for the disordered region spanning from residues 330 to 345, and the long loop at the N-terminal (Fig. S3). This disordered region is located away from the active site with an approximately 40 Å distance between Gln341 and Asn195 throughout the model. PROCHECK showed that 91.4% of residues are located in most favored regions, 7.5% of residues are located in additional allowed regions, and 0.4% of residues are found in outlier regions (Fig. S4). Verify3D indicated that 80.8% of amino acids show an averaged 3D-1D score ≥ 0.2. The above statistics indicate a reliable structure model.
Molecular docking and simulation of AeDHPAAS substrate binding
To reveal possible ligand-enzyme interactions, detailed analysis of substrate binding to the AeDHPAAS active site was carried out with a molecular docking bound ligand. AutoDock Vina docked dopa, DHPAA and dopamine into the active site cavity (Fig. 8). The results show that dopa is a good ligand with the highest relative binding affinity among the three studied ligands (-5.6 kcal/mol) and the docking solutions of DHPAA and dopamine resulted in binding affinities of -5.0 kcal/mol and − 4.9 kcal/mol, respectively. The substrate-binding pocket of AeDHPAAS is composed of a set of conserved residues, as well as three variable residues (Phe82, Tyr83 and Asn195) that have diverged from the AeDDC sequence. The substrate specificity determining residues appear to be similar between DDC and DHPAAS. Phe82-Tyr83 is of particular interest as this motif is conserved as Phe-Tyr in AeDHPAAS, but substituted to Tyr-Phe in AeDDC. Asn195 is conserved in AeDHPAAS, but substituted to histidine in AeDDC. Substitution of these three variable residues is reported to tune bifunctional switching between DHPAAS and DDC activities 21, 22.
Considering the dimeric conformations in a physiological environment, molecular dynamics was used to calculate the atomical dynamic movements and conformational variations of AeDHPAAS dimer bound to ligands and LLP cofactor (Fig. 9). Root-mean-square deviation (RMSD) plots (Fig. 9a-c) indicate convergence in all simulations. The fluctuations occurring during simulations may be due to loop regions with higher flexibilities. The local motility properties of amino acid residues in AeDHPAAS complexed with ligands were further determined by the root-mean-square (RMS) fluctuation. Chain A experiences less conformational changes than chain B according to the RMSD and RMS values (Fig. 9a-f). Overall, it is observed that there are two main areas with high-RMS fluctuations, including the N-terminus and residues 328–341 (Fig. 9d-f).