Structural and functional comparisons of salivary α-glucosidases from the mosquito vectors Aedes aegypti, Anopheles gambiae, and Culex quinquefasciatus

Abstract

Mosquito vectors of medical importance both blood and sugar feed, and their saliva contains bioactive molecules that aid in both processes. Although it has been shown that the salivary glands of several mosquito species exhibit α-glucosidase activities, the specific enzymes responsible for sugar digestion remain understudied. We therefore expressed and purified three recombinant salivary α-glucosidases from the mosquito vectors Aedes aegypti, Anopheles gambiae, and Culex quinquefasciatus and compared their functions and structures. We found that all three enzymes were expressed in the salivary glands of their respective vectors and were secreted into the saliva. The proteins, as well as mosquito salivary gland extracts, exhibited α-glucosidase activity, and the recombinant enzymes displayed preference for sucrose compared to p-nitrophenyl-α-D-glucopyranoside. Finally, we solved the crystal structure of the Ae. aegypti α-glucosidase bound to two calcium ions at a 2.3 Ångstrom resolution. Molecular docking suggested that the Ae. aegypti α-glucosidase preferred di- or polysaccharides compared to monosaccharides, consistent with enzymatic activity assays. Comparing structural models between the three species revealed a high degree of similarity, suggesting similar functional properties. We conclude that the α-glucosidases studied herein are important enzymes for sugar digestion in three mosquito species.

Graphical Abstract

1. Introduction

Although most adult female mosquitoes blood feed to reproduce, both males and females require sugar as an energy source [1–3]. Sugars contain carbohydrates that support survival, flight, and metabolism, as well as contribute to female fertility and egg development [4–6]. In addition to its direct impacts on physiology, the act of sugar feeding influences behaviors such as blood feeding and mating that are important in mosquito ecology and pathogen transmission [7–9].

Mosquito saliva contains a diverse array of bioactive molecules that are important for both blood and sugar feeding [10–13]. Anti-hemostatic compounds secreted at the bite site facilitate the former [10,14], while enzymes involved in sugar metabolism assist in the latter [11,15]. Previous studies have shown that mosquito saliva, including from Aedes aegypti [5,11,15,16], Ae. albopictus [17], Anopheles darlingi [18], An. dirus [19], Culex quinquefasciatus [20,21], and Cx. tarsalis [15], contains α-glucosidases that break down polysaccharides into simple sugars. Although blood is ingested into the midgut, sugar is instead mainly routed to a non-secretory organ known as the crop (also called the diverticula) [22]. Saliva re-ingested during sugar feeding supports its initial digestion in the crop, and sugar is periodically released into the midgut, where rapid digestion and absorption ultimately occurs [2].

Salivary proteins related to sugar feeding mainly reside in the female proximal-lateral lobes (and in the entire male salivary gland), whereas those important in blood feeding tend to be in the medial and distal-lateral lobes [2,11,23,24]. Indeed, salivary α-glucosidase activities have been mainly reported in mosquito proximal-lateral lobes [11,18,19], whereas enzymes important in blood feeding such as apyrases are restricted to the distal lobes [11,17,18,25]. Such cellular compartmentalization suggests that salivary proteins are differentially secreted during blood or sugar feeding. Marinotti and colleagues (1990) addressed this hypothesis by analyzing apyrase and glucosidase activities and salivary protein levels after blood or sugar feeding [11]. The authors concluded Ae. aegypti selectively secrete an α-glucosidase during a sugar meal but secrete both α-glucosidase and apyrase after a blood meal; this suggested that sugar feeding solely affects the proximal lobes, while blood feeding triggers secretion from the whole gland [11]. A possible explanation may be that proteins necessary for sugar metabolism facilitate blood feeding as well – a sugar-rich substance itself.

Alpha-glucosidases (EC 3.2.1.20) are widely distributed across organisms [26]. For mosquitoes and other insects, these enzymes are important for hydrolyzing sucrose, a major component of plant nectar [27,28]. Alpha-glucosidases hydrolyze terminal glycosidic bonds of α-glucans, disaccharides, or oligosaccharides and release non-reducing α-glucoses [26,29]. There are three groups defined by their preferred substrate: (1) those that prefer phenyl-α-glucoside or sucrose, (2) maltases that prefer maltooligosaccharides but not α-glucoside or sucrose, and (3) maltases that can also hydroylze α-glucans such as starch and glycogen [26,29]. They can be further divided into two families based on primary amino acid sequence homology of five conserved regions [26,29], where yeast, bacilli, and insects harbor α-glucosidases from family I, while mammals and plants harbor those from family II [26].

Although α-glucosidase activity has been observed in salivary glands across several blood feeding arthropod species, we still lack comparative biochemical data on the molecules responsible for these observations. We therefore set out to study three salivary α-glucosidases from Ae. aegypti, An. gambiae, and Cx. quinquefasciatus (henceforth called Aegluc, Angluc, and Cxgluc, respectively). We successfully expressed and purified the three recombinant α-glucosidases, and we raised polyclonal antibodies against them that displayed minimal cross reactivities. We found that the proteins were present in salivary gland extracts (SGE) from their respective vectors, and that they were secreted into mosquito saliva. The recombinant proteins, as well as SGEs, displayed α-glucosidase activities. The recombinant α-glucosidase activities against sucrose were more pronounced compared to p-nitrophenyl-α-D-glucopyranoside (NPαGlu). Finally, we report the crystal structure of the Ae. aegypti α-glucosidase bound to two calcium ions at a 2.3 Ångstrom resolution. Comparing structural models between the three species revealed a high degree of similarity, suggesting similar functional properties.

2. Materials & Methods

2.1. Expression and purification of recombinant proteins

Coding α-glucosidase DNA sequences from Ae. aegypti (XP_001660189), An. gambiae (XP_320938), and Cx. quinquefasciatus (XP_001866573) were obtained from NCBI. Genes containing a C-terminal 6x-histidine tag were codon optimized for mammalian expression in the VR2001 vector and were synthesized by BioBasic Inc (Markham, Ontario, CA). Expi293F human embryonic kidney cells (Thermo Fisher Scientific; Cat. no.: A14527) were transfected with sterile plasmid DNA, prepared by Aldevron (Fargo, ND, USA), at the NCI Protein Expression Laboratory (Frederick, MD, USA), according to the manufacturer’s protocol and as described in [30]. Supernatants were collected 96 h after transfection.

Recombinant proteins were purified by nickel affinity chromatography followed by size exclusion chromatography using nickel charged HiTrap Chelating HP and HiLoad 16/600 Superdex 200 pg columns (Cytiva, Marlborough, MA, USA), respectively. Briefly, using a 400 ml Amicon Stirred Cell (Millipore, Burlington, MA, USA), HEK293 cell supernatants underwent buffer exchange into Tris Buffer A (10 mM Tris, 500 mM NaCl, 5 mM imidazole, pH 7.4) and were concentrated into ^~ 50 ml. Samples were then loaded onto a pre-equilibrated nickel-charged HiTrap Chelating HP column. Recombinant proteins were eluted by passing a gradient of Tris Buffer B (10 mM Tris, 500 mM NaCl, 1 M imidazole, pH 7.4) using an AKTA Start system (GE Healthcare, Chicago, IL, USA). Peak fractions were visualized by SDS-PAGE as described below, and fractions of interest were pooled and concentrated using Amicon Ultra-15 (10 kDa cutoff) centrifugal filter units (Millipore Sigma, Burlington, MA, USA). Size exclusion chromatography was performed on a HiLoad 16/600 Superdex 200 pg column using an AKTA pure 25 system (GE Healthcare, Chicago, IL, USA) into 20 mM Tris-HCl, 150 mM NaCl pH 7.4, and peak fractions were confirmed by SDS-PAGE.

2.2. SDS-PAGE

All proteins were heated to 95 °C for 5-10 min under reducing conditions in 1X LDS (Thermo Fisher Scientific, Waltham, MA, USA) and were separated using 4-12% Bis-Tris protein gels (Thermo Fisher Scientific, Waltham, MA, USA). Gels were stained with Coomassie Brilliant Blue (GenScript, Piscataway, NJ, USA). Protein concentrations were determined using the DeNovix DS-11 FX+ spectrophotometer (DeNovix, Wilmington, DE, USA) adjusted by the molar extinction coefficient (MEC; 138200, 136710, 127200 M⁻¹ cm⁻¹ for AeGluc, AnGluc, and CxGluc, respectively).

2.3. Polyclonal antibody generation

Polyclonal antibodies against AeGluc, AnGluc, and CxGluc were raised in mice. Mice (BALB/c; Charles River, Frederick, MD, USA) were IM immunized with 10 μg of recombinant protein in combination with Magic Mouse Adjuvant (CD Creative Diagnostics, Shirley, NY, USA). Negative control mice were immunized with Magic Mouse Adjuvant alone. At 21 d post-immunization, mice received a booster immunization with 10 μg of recombinant protein in combination with Magic Mouse Adjuvant (or adjuvant alone for negative control group). Blood was collected 35 d post-immunization. The antibody levels were confirmed by ELISA. These studies were carried out according to the NIAID-NIH animal study protocols (ASP) approved by the NIH Office of Animal Care and Use Committee (OACUC), with approval ID ASP-LMVR3.

Polyclonal antibodies against Ae. aegypti, An. gambiae, or Cx. quinquefasciatus SGE were raised in rabbits, and immunization was carried out in Noble Life Science facility (Woodbrine, MD) according to their standard protocol, as described in [14]. IgG from rabbit sera was purified using a 5 ml HiTrap protein A HP column following the manufacturer’s instructions (GE Healthcare, Piscataway, NJ) and as described in [14].

2.4. ELISA

ELISAs were performed to assess α-glucosidase secretion into mosquito saliva (two protocols are described in detail below), α-glucosidase immunogenicity after mouse immunizations, and to detect α-glucosidase-specific antibodies in patient convalescent plasma from the Biodefense and Emerging Infections Research Repository (BEI Resources, NIAID/NIH, Bethesda, MD, USA).

To assess α-glucosidase secretion into mosquito saliva, either a mosquito feeding solution or a pool of saliva collected by the forced salivation technique were used to directly coat a Corning Costar ELISA 96 well high binding plate (Corning, Kennebunk, ME, USA). To assess α-glucosidase immunogencity after mouse immunization, 1 μg/ml recombinant AeGluc, AnGluc, CxGluc in 0.05 M carbonate-bicarbonate buffer (Millipore Sigma, Burlington, MA, USA) were added to the same type of ELISA plates. In both cases, coated plates were left O/N at 4 °C. The next day, the plates were washed 3 X with 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.1% Tween 20 (TBST) and then blocked with 200 μl blocking solution (2.5% BSA, TBST) at room temperature (RT) for 1-2 h. Mouse serum raised against the recombinant glucosidases (1:200 in blocking solution) were added for 2 h at RT. Purified IgG from rabbit serum raised against Ae. aegypti, An. gambiae, or Cx. quinquefasciatus SGE were used as positive controls (1:500 in blocking solution). Plates were again washed 3 X with TBST before the addition of secondary goat anti-mouse (Seracare, Milford, MA, USA) or goat anti-rabbit (Invitrogen, Waltham, MA, USA) conjugated to HRP (1:2000 in blocking solution). Plates were washed 2 X with TBST and 1 X with TBS before the addition of color reagents A & B (1:1, as described in the manufacturer’s instructions; Bio-Techne Corporation, Minneapolis, MN, USA). Ten-to-twenty minutes later, plates were read after the addition of the STOP solution (as provided by the manufacturer, Bio-Techne Corporation, Minneapolis, MN, USA) on a VersaMax microplate reader (Molecular Devices, San Jose, CA, USA) at 450 nm.

To detect α-glucosidase-specific antibodies in patient convalescent plasma, ELISA plates were coated with 1 μg/ml recombinant AeGluc in 0.05 M carbonate-bicarbonate buffer (Millipore Sigma, Burlington, MA, USA O/N at 4 °C. Patient convalescent plasma known to harbor DENV antibodies from BEI Resources were added at 1:250 dilution in blocking solution for 2 h at RT. Human serum negative for Ae. aegypti SGE-specific antibodies [31] (prepared at 1:250 in blocking buffer) were used as negative controls. ELISA was performed as described above using anti-human IgG-peroxidase secondary antibody (Millipore Sigma, Burlington, MA, USA) for detection. For all assays, wells coated O/N with buffer only were considered baseline, whose absorbances were subtracted from all measurements in the final analyses, which were visualized on GraphPad Prism version 9.0.

2.5. Western blot

One-to-three μM recombinant proteins (^~ 5 μg total protein) and 20 μg mosquito SGE were separated by SDS-PAGE for western blots. Proteins were transferred to a PVDF membrane (iBlot, Invitrogen, Waltham, MA, USA) that was blocked for ^~ 2 h in blocking buffer (5% [w/v] instant non-fat dry milk (Carnation) in TBST). Membranes were incubated O/N at 4 °C with anti-glucosidase mouse serum (1:2000 in blocking buffer). The next day, membranes were washed with TBST (2 X for 10 min) and TBS (1 X for 10 min) and incubated at RT for 1-2 h with goat anti-mouse IgG conjugated to alkaline phosphatase (Sigma Aldrich, St. Louis, MO, USA; 1 mg/ml, diluted 1:10,000 in blocking buffer) or goat anti-mouse conjugated to HRP (Invitrogen, Waltham, MA, USA; 1:3000, diluted in blocking buffer). The membranes were again washed with TBST (2 X for 10 min) and TBS (1 X for 10 min). Blots were developed using Western Blue Stabilized alkaline phosphatase substrate (Promega, Madison, WI, USA) or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific, Waltham, MA, USA). All blots were imaged using an Azure 300 imaging system (Azure Biosystems, Dublin, CA, USA).

2.6. Mosquito rearing and salivary gland extract (SGE) preparation

Ae. aegypti, An. gambiae, and Cx. quinquefasciatus mosquitoes were reared in standard insectary conditions (27 °C, 80% humidity, 12 h light/dark cycle) at the Laboratory of Malaria and Vector Research, NIH/NIAID. Salivary glands from sugar-fed 5-7 d old female mosquitoes were dissected in phosphate-buffered saline (PBS) pH 7.4 using a stereomicroscope as described in [32,33]. SGE was obtained by sonication (Fisher Brand, model FB120). Sonication was carried out at power 30% duty cycle (36W/20kHz), with 3 cycles of 2 seconds pulse and one sec rest (for 40 sec) to allow dissipation of heat. Tubes were then centrifuged at 12,000 x g for 5 m. Supernatants were kept at −80 °C until use.

2.7. Mosquito secretion assay

To test for α-glucosidase secretion during probing and feeding, cartons of approximately 100-200 Ae. aegypti, An. gambiae, or Cx. quinquefasciatus mosquitoes were sugar starved O/N and fed a 15 ml bicarbonate feeding solution (150 mM NaCl, 100 mM NaHCO₃, 10 mM ATP pH 7.3; it had been previously reported in [34] that several anopheline species engorged this solution) in an artificial glass feeder lined with parafilm at 37 °C for 1 h. Upon visual inspection, the three mosquito species all appeared to engorge the solution. After 1 h, we collected the solution from the glass mosquito feeder, concentrated it to ^~ 500 μl with an Amicon Ultra-15 (10 kDa cuttoff) centrifugal filter unit (Millipore Sigma, Burlington, MA, USA), and used it to directly coat an ELISA plate to test for α-glucosidase secretion as described in the ELISA method section [34].

2.8. Forced salivation collection

Saliva was collected from female Ae. aegypti, An. gambiae, and Cx. quinquefasciatus by the forced salivation technique for ELISA and Mass Spectrometry analyses. Six-to-eight d old mosquitoes were sugar-starved O/N and water deprived for ^~ 1-2 h before saliva collection. After sedating them on ice, mosquito legs and wings were removed, and the hypopharynx was carefully separated from the outer sheath of the proboscis. The mosquito mouthparts were then inserted into 10 μl tips containing 2 μl 150 mM NaCl taped to a plastic petri dish. One hundred mosquitoes of each species were left to salivate for 1 h, and the saliva solutions were subsequently pooled into Protein Lo-Bind tubes by species (Eppendorf, Enfield, CT, USA).

2.9. Enzymatic activity assays

Alpha-glucosidase activity with p-nitrophenyl-α-D-glucopyranoside (NPαGlu) was tested via an α-glucosidase activity assay kit (Millipore Sigma, Burlington, MA, USA) following the manufacturer’s instructions. Alpha-glucosidase activity was calculated as the amount of enzyme that catalyzes the hydrolysis of 1 μmol p-nitrophenyl-α-D-glucopyranoside per minute. Specific activity was normalized by protein concentration or number of salivary gland pairs, and data corresponding to the optimum protein concentration (0.001 mg/ml) or salivary gland pair number (4) was reported based on where the absorbance curve remained linear but gave the largest signal. At least four enzyme concentrations/salivary gland pair numbers were determined for each incubation, and the initial rates of hydrolysis were calculated.

Activity toward sucrose was detected by the release of the reducing sugars glucose and fructose from sucrose that react with p-Hydroxybenzhydrazide (PAHBAH, 5 mg/ml in 0.5 M NaOH; Millipore Sigma, Burlington, MA, USA) to form a spectrophometric product after boiling – a protocol adapted from Millipore Sigma and described in detail in [35]. To determine the optimum protein concentration for the assay, ten-fold serial dilutions of 0.01 mg/ml recombinant proteins were carried out in PBS until a concentration of 0.0001 mg/ml was reached. One hundred μl of each concentration was then added to 900 μl sucrose (10 mg/ml in 100 mM sodium acetate, pH 5.5) and incubated at 37 °C for 30 min. One hundred μl of each reaction was then added to 2.9 ml p-Hydroxybenzhydrazide (PAHBAH, 5 mg/ml in 0.5 M NaOH; Millipore Sigma, Burlington, MA, USA), which was boiled for 5 m. The solutions were cooled to RT, and 100 μl were transferred to 96 well plates, where the absorbance at 410 nm were measured using a VersaMax microplate reader (Molecular Devices, San Jose, CA, USA). A standard curve was plotted by measuring the Abs₄₁₀ of 0-1 mM of glucose standard solution prepared in water. Enzymatic activity was calculated as mmoles glucose released divided by time of the assay (30 minutes) * 2 (conversion factor for 1 μmol of sucrose being hydrolyzed to glucose and fructose). Enzymatic activity was then determined by dividing by the final concentration of the protein used in the reaction.

To determine enzyme K_m, final concentrations of 0-500 mg/ml sucrose in 900 μl was added to 100 μl 0.001 mg/ml recombinant protein, and the sucrose activity assay was preformed as described above. A Michaelis-Menten curve was plotted by enzymatic activity (mM/min) and a nonlinear regression curve with a least squares regression curve fit was plotted on GraphPad Prism version 9.0.

2.10. Deglycosylation

AeGluc, AnGluc, CxGluc were deglycosylated using the Enzymatic DeglycoMx Kit (QA-Bio, Palm Desert, CA, USA) following the manufacturer’s instructions, omitting the denaturation step. Briefly, 1 mg/ml AeGluc, 1 mg/ml AnGluc, or 0.1 mg/ml CxGluc in 38 μl 20 mM Tris 150 mM NaCl pH 7.4 were mixed with 10 μl 5X reaction buffer and 2 μl DeGlycoMx enzyme cocktail in a 1.5 ml tube and were left at 37 °C O/N. The efficiency of deglycosylation was assessed by SDS/PAGE. Enzymatic assays performed using deglycosylated proteins are specified in the text and were standardized by protein concentration.

2.11. Crystallization and Crystal Structure Determination

Initial screening for crystallization conditions performed by the hanging drop vapor-diffusion method produced a hit in condition #11 from Crystal screen (HR2-110, Hampton Research). After optimization of the protein concentration, diffraction quality crystals were obtained in the original condition that consisted of 1.0 M ammonium phosphate monobasic in 0.1 M sodium citrate tribasic dihydrate buffer, pH 5.6. For data collection the crystals were mounted on a loop and flash frozen in a nitrogen gas stream at 95 K using 25% ethylene glycol as a cryoprotectant added to the mother liquor. Data were collected at a home source on a Rigaku FR-X high intensity microfocus rotating anode X-ray generator equipped with an EIGER2 R 4M detector (DECTRIS) using Cu radiation.

A crystal that diffracted to 2.32 Å resolution with cell dimensions (in Å) a = 48.76, b = 51.03 and c = 70.75 and belonging to the P1 space group was used to collect a data set (Table 1). The data were processed, reduced and scaled with XDS [36]. The structure of Ae. aegypti α-glucosidase was determined by molecular replacement using Phaser [37] by employing a search model obtained by Phyre2 [38]. The final model of Ae. aegypti was constructed by iterative cycles of manual model building using the program Coot [39] after each cycle of refinement with stepwise increase in the resolution using Phenix [40]. Protein models were visualized in PyMol version 1.3.

Table 1.

Catalytic efficiencies of AeGluc, AnGluc, and CxGluc for hydrolysis of sucrose or NPαGlu

	Sucrose
	kcat (x 103 min−1)	Km (mM)	Vmax (mM/min)	kcat/Km (mM−1 min−1)
AeGluc	48.3 ± 4.0	47.6 ± 16.3	0.74 ± 0.07	1014.4 ± 83.1
AnGluc	48.9 ± 3.6	20.4 ± 7.4	0.74 ± 0.06	2398.9 ± 175.4
CxGluc	45.0 ± 4.5	40.1 ± 15.1	0.69 ± 0.07	1123.1 ± 111.2

2.12. Molecular Docking

3D structures of α-glucose and sucrose were obtained from PubChem [41] (CIDs 79025 and 5988, respectively). Since 3D conformers of maltohexaose are not available in PubChem due to its large number of atoms, its starting 3D conformation was extracted from the PDB entry 3U2V. These structures were prepared for docking using the LigPrep [42] procedure in Maestro [43]. Protein structures were readied for docking by the Protein Preparation procedure in Maestro [44]. The ligand binding site was defined based on the superimposed structure of Bombyx mori GH13 sucrose hydrolase (PDB: 6LGE). Ligand docking was performed using Glide [45,46] in the Maestro suite. Due to the high number of rotational bonds in some of the ligands, Glide docking utilized expanded conformational sampling and considered a greater number of initial poses as well as post-docking poses for optimization. Post-docking minimization was performed using the 50 best scoring ligand poses (5 default), and the final number of poses reported was 20. Glide docking scores approximate binding energy, with more negative scores indicating greater stability. Figures of the docked poses and molecular lipophilicity potential surface maps were prepared using ChimeraX [47,48].

2.13. Circular dichroism

Purified AeGluc (0.1 mg/ml) in 20 mM, Tris 150 mM, NaCl pH 7.3 was used for circular dichroism spectroscopy (CD) analyses using a Jasco J-1100 CD spectropolarimeter instrument. Continuous measurements with a pitch of 0.2 nm were recorded over 200-250 nm wavelengths with a bandwidth of 1 nm in a 1 mm quartz cuvette. Mean residue ellipticity was calculated with the following equation: (protein molecular weight in daltons/(number of amino acids – 1) * θλ))/(10 * pathlength in cm * protein concentration in g/ml). All readings were normalized by subtracting with blank (buffer) mean residue ellipticity. Data were analyzed using CAPITO [49].

3. Results

3.1. Alpha-glucosidases present in the salivary glands of Ae. aegypti, An. gambiae, and Cx. quinquefasciatus are secreted into saliva

We aimed to compare three salivary α-glucosidase orthologs found in Ae. aegypti (NCBI accession: XP_001660189; “AeGluc”), An. gambiae (NCBI accession: XP_320938; “AnGluc”), or Cx. quinquefasciatus (NCBI accession: XP_001866573; “CxGluc”) in vitro and to compare their activities to those found in vivo. A 67 kDa α-glucosidase had been previously identified in Ae. aegypti salivary glands, where sucrose was found to be its best substrate [5,11,16]. Furthermore, the salivary α-glucosidase from Cx. quinquefasciatus had been expressed and purified from yeast previously and was found to rapidly hydrolyze maltotriose and sucrose [21]. To our knowledge, the activity of AnGluc has never been characterized. All three proteins displayed high degrees of similarity in their primary amino acid sequences (60-74%), notably in the four regions known to be conserved across family I α-glucosidases (Supplementary Figure 1). As such, they were also similar in various biophysical parameters, such as isoelectric points and molar extinction coefficients (Supplementary Table 1). To compare the three salivary α-glucosidases in standard conditions, we expressed each of them in a mammalian expression system and purified cell supernatants by nickel and size exclusion chromatographies. Soluble protein was stable in 20 mM Tris-HCl pH 7.4 and 150 mM NaCl and ran at the expected sizes (66.0 kDa, 66.5 kDa, 66.0 kDa for Aegluc, Angluc, and Cxgluc, respectively) on SDS-PAGE (Figure 1).

Figure 1. Expression and purification of recombinant α-glucosidase sequences from Aedes aegypti, Anopheles gambiae, and Culex quinquefasciatus.

Chromatographs (left) and Coomassie-stained SDS/PAGE (right) of recombinant α-glucosidases from Ae. aegypti, An. gambiae, and Cx. quinquefasciatus after size exclusion chromatography with a Superdex 16/600 HiLoad column. Red arrows indicate peak of interest.

To confirm that Aegluc, Angluc, and Cxgluc were present in the salivary glands of their respective mosquito vectors, we generated polyclonal antibodies (pAbs) by immunizing mice with the individual recombinant proteins. ELISA analyses using mouse serum against Aegluc, Angluc, or Cxgluc recombinant proteins indicated the presence of α-glucosidase-specific antibodies (Supplementary Figure 2). We then ran equal amounts of SGE from Ae. aegypti, An. gambiae, or Cx. quinquefasciatus on SDS-PAGE and performed western blots using the Aegluc, Angluc, or Cxgluc-specific pAbs. All three SGEs reacted with their respective α-glucosidase pAbs in a species-specific manner (Figure 2A-D). These results corroborated previously published immunofluorescent assays (IFAs) using Ae. aegypti, An. gambiae, or Cx. quinquefasciatus salivary glands and the Aegluc, Angluc, or Cxgluc-specific pAbs generated herein [32,33]. Confocal microscopy had revealed that Aegluc, Angluc, or Cxgluc were localized in the proximal lateral lobes, typical of salivary proteins involved in sugar metabolism [24,32,33]. We also compared the sizes of recombinant and native α-glucosidases side-by-side by western blot using recombinant protein and mosquito SGE (Supplementary Figure 3A). We found that although the recombinant protein ran at a slightly higher molecular weight than native protein on SDS-PAGE, their sizes were almost identical. We hypothesized this was due to similar glycosylation patterns that likely occur both in vitro (in the mammalian protein expression system) as well as in vivo (in the mosquito). Indeed, in silico analyses revealed that AeGluc, AnGluc, and CxGluc had 5, 8, and 4 predicted glycosylation sites, respectively [50]. Furthermore, deglycosylating the recombinant proteins and running them side-by-side with their glycosylated counterparts showed that the deglycosylated protein ran at a smaller size (< 70 kDa) than both the native and recombinant α-glucosidases (Supplementary Figure 3B).

Figure 2. Antibodies raised against recombinant α-glucosidases react to salivary gland extracts from (A) Aedes aegypti, (B) Anopheles gambiae, or (C) Culex quinquefasciatus.

Western blots with 10 μg salivary gland extract (SGE) from (lane 1) Ae. aegypti (Ae), (lane 2) An. gambiae (An), and (lane 3) Cx. quinquefasciatus (Cx) with antibodies raised in mice against recombinant (A) AeGluc, (B) AnGluc, or (C) CxGluc. D. Corresponding Coomassie-stained SDS-PAGE with 10 μg Ae. aegypti (Ae), An. gambiae (An), and Cx. quinquefasciatus (Cx) SGE.

Previously, Marinotti and colleagues found that AeGluc was secreted during mosquito feeding [11]. We too have previously found that peptides covering 34-39% of the AeGluc protein were detectable by mass spectrometry in duplicate Ae. aegypti saliva samples [51]. Furthermore, we re-analyzed mass spectrometry data from duplicate Cx. quinquefasciatus saliva samples and found that peptides covering 53-56% of CxGluc were present [52]. To complement these data sets, we performed mass spectrometry on An. gambiae saliva. We found that peptides covering 63% of AnGluc were detectable in duplicate saliva samples (Supplementary Dataset 1). To confirm the mass spectrometry data, we collected mosquito saliva by two methods – through forced salivation and artificial membrane feeding with a sodium bicarbonate feeding solution (described in detail in the Materials and Methods section) – and performed ELISAs using the α-glucosidase-specific antibodies. Antibodies raised against Ae. aegypti, An. gambiae, or Cx. quinquefasciatus SGE were used as positive controls to ensure that we could detect secreted salivary protein by both methods. As expected, SGE-specific antibodies reacted with the saliva samples by ELISA (Supplementary Figure 4). AeGluc-, AnGluc-, and CxGluc-specific pAbs also reacted with the saliva samples (Figure 3), indicating that the α-glucosidases were secreted into mosquito saliva.

Figure 3. Alpha-glucosidases are secreted into Aedes aegypti, Anopheles gambiae, or Culex quinquefasciatus saliva.

Saliva was collected from Ae. aegypti (purple), An. gambiae (green), or Cx. quinquefasciatus (orange) by the forced salivation technique (A) or after an artificial membrane feeding on a bicarbonate buffer (B). Primary antibodies raised against AeGluc, AnGluc, or CxGluc were used to detect respective α-glucosidase secretion by ELISA into saliva samples collected by both methods. Black bars indicate mean and standard deviation.

We next hypothesized that α-glucosidase-specific antibodies would be detectable in patients who had been bitten by mosquitoes in the field. This was because the α-glucosidases were present in mosquito saliva and because we were able to generate species-specific pAbs. To address this hypothesis, we obtained patient convalescent plasma from the Biodefense and Emerging Infections Research Repository (BEI Resources) that were positive for dengue virus (DENV) IgG antibodies, an indicator that these patients had been bitten by Ae. aegypti, to see if they also harbored AeGluc-specific antibodies. As a negative control, we used serum from patients that had never been bitten by Ae. aegypti mosquitoes. Of the 19 serum samples analyzed, none significantly reacted to recombinant AeGluc, compared to negative contrls, by ELISA (Supplementary Figure 5). Taken together, these data suggest that although α-glucosidases are likely secreted during mosquito feeding, AeGluc may not be immunogenic enough to be used as a marker for exposure to Ae. aegypti in the field.

3.2. Ae. aegypti, An. gambiae, and Cx. quinquefasciatus α-glucosidases display sucrose substrate preference

Alpha-glucosidases hydrolyze terminal glycosidic bonds of oligosaccharides or α-glucans and release non-reducing α-glucoses [26]. We determined α-glucosidase activity of recombinant AeGluc, AnGluc, and CxGluc by measuring the release of p-nitrophenolate from NPαGlu after 30 minutes at 37 °C. AeGluc displayed the highest activity for NPαGlu (76.77 ± 7.21 U/mg) compared to AnGluc (47.16 ± 2.76 U/mg; p = 0.036, unpaired two-tailed t-test) or CxGluc (41.75 ± 0.90 U/mg; p = 0.021, unpaired two-tailed t-test) (Figure 4A). We repeated the experiment for SGE from all three mosquito species under the same conditions. Activities were low overall, ranging between 0.01-0.02 U/salivary gland. Cx. quinquefasciatus SGE showed greater activity for NPαGlu compared to Ae. aegypti (p = 0.0011, unpaired two-tailed t-test) or An. gambiae (p = 0.014, unpaired two-tailed t-test), but the differences were minimal (Figure 4B).

Figure 4. Sugar activity assays for recombinant α-glucosidases or mosquito salivary gland extract.

A. AeGluc, AnGluc, and CxGluc activity for NPαGlu or sucrose and B. Ae. aegypti (AeSGE), An. gambiae (AnSGE), or Cx. quinquefasciatus (CxSGE) salivary gland extract activities for NPαGlu or sucrose. One unit (U) refers to the amount of enzyme that catalyzes the conversion of 1 μmol substrate per minute. Activities were normalized by milligrams (mg) of recombinant protein or number of salivary gland pairs to allow for comparisons across groups. * p < 0.05, ** p < 0.01 by unpaired two-tailed t-test. Black bars indicate mean and standard deviation.

Previously, Ae. aegypti SGE [5,11] and recombinant Cx. quinquefasciatus α-glucosidase expressed in yeast [21] both displayed preference toward sucrose, a major component of nectar [27], compared to NPαGlu. To compare the activities of recombinant AeGluc, AnGluc, and CxGluc for sucrose versus NPαGlu, we performed a sucrose activity assay that detected the release of the reducing sugars glucose and fructose from sucrose, which react with p-Hydroxybenzhydrazide to form a spectrophometric product, and quantified it based on a standard curve using increasing concentrations of D-(+)-glucose in the assay alone (Supplementary Figure 6). Specific activities were normalized by protein concentration (0.001 mg/ml). This value was chosen because it gave the largest signal while enzymatic activity remained linear (Supplementary Figure 7). AeGluc, AnGluc, and CxGluc all displayed more than 10-fold greater activity for sucrose compared to NPαGlu (Figure 4A). Specifically, AeGluc hydrolyzed sucrose at 220.45 ± 25.38 U/mg (p = 0.016, unpaired two-tailed t-test compared to its activity for NPαGlu), AnGluc hydrolyzed sucrose at 419.51 ± 23.57 U/mg (p = 0.002, unpaired two-tailed t-test compared to its activity for NPαGlu), and CxGluc hydrolyzed sucrose at 260.038 ± 34.89 U/mg (p = 0.013, unpaired two-tailed t-test compared to its activity for NPαGlu). When comparing the activities across recombinant α-glucosidase orthologs, AnGluc displayed the highest rates of activity compared to AeGluc (p = 0.015, unpaired two-tailed t-test) or CxGluc (p = 0.033, unpaired two-tailed t-test). We repeated the experiment for SGE from all three mosquito species under the same conditions. Activities were again low overall, ranging between 0.01-0.07 U/salivary gland pair, and there were no significant differences between those measured for NPαGlu, although sucrose activity tended to be greater for all SGEs tested (Figure 4B). There were also no significant differences in the activities for sucrose between SGEs from the different species – Ae. aegypti (0.031 ± 0.0086 U/pair), An. gambiae (0.042 ± 0.049 U/pair), or Cx. quinquefasciatus (0.043 ± 0.043 U/pair).

Because we observed that the recombinant α-glucosidases preferred sucrose as a substrate compared to NPαGlu, we determined kinetic parameters for the hydrolysis of sucrose for each recombinant enzyme by plotting Michaelis-Menten saturation curves (Table 1) (Figure 5). AnGluc exhibited the best catalytic efficiency compared to AeGluc and CxGluc, which showed decreased k_cat/K_m values by 42.3% and 46.8%, respectively, compared to that of AnGluc. Although the k_cat (AeGluc = 48.3 ± 4.0 x 10³ min⁻¹; AnGluc = 48.9 ± 3.6 x 10³ min⁻¹; CxGluc = 45.0 ± 4.5 x 10³ min⁻¹) and V_max (AeGluc = 0.74 ± 0.07 mM/min; AnGluc = 0.74 ± 0.06 mM/min; CxGluc = 0.69 ± 0.07 mM/min) values were similar, it was the K_m value for AnGluc (20.4 ± 7.4 mM) that outcompeted those of AeGluc (47.6 ± 16.3 mM) or CxGluc (40.1 ± 15.1 mM).

Figure 5. Catalytic efficiencies of AeGluc, AnGluc, or CxGluc for hydrolysis of sucrose.

Michaelis-Menten saturation curves for AeGluc (purple), AnGluc (green), and CxGluc (orange) with sucrose. Black bars indicate mean and standard deviation for duplicate experiments.

Taken together, these results indicate that the recombinant α-glucosidases display preference for sucrose over NPαGlu. Although the corresponding mosquito SGEs displayed similar activities against NPαGlu or sucrose, there was a also trend of higher activities for sucrose. Nonetheless, there may be other salivary enzymes involved in breaking down polysaccharides as well in addition to the α-glucosidases described herein.

3.3. Crystal structure of Ae. aegypti α-glucosidase reveals protein is bound to two calcium ions

To further characterize the α-glucosidases and to gain insights on their functional properties, we solved the crystal structure of the Ae. aegypti α-glucosidase. A crystal of AeGluc that belonged to P1 space group and defracted to a 2.3 Ångstrom (Å) resolution was used to collect a dataset (Table 2). The coordinates and structure factors have been deposited in the Protein Data Bank under the PDB ID 8SLV.

Table 2.

Data collection and refinement statistics for AeGluc

Data Collection
Space group	P 1
Cell dimensions a, b, c, α, β, γ (Å)	48.76, 51.03, 70.75, 101.82, 94.66, 95.76
Resolution (Å)	22.55 – 2.32
Completeness (%)	94.8 (22.55-2.32)
I/σ|1	1.66
R, Rfree	0.146, 0.188
Refinement
Resolution (Å)	24.79 – 2.32
Completeness (%)	94.9 (24.79-2.32)
R, Rfree	0.146, 0.188
Rfree test Set	2000 reflections (7.39%)
Wilson B-factor (Å2)	19.8
Anisotropy	0.164
Bulk solvent ksol(e/ Å3), Bsol(Å2)	0.36, 52.6
Fo, Fc correlation	0.96
Total number of atoms	4913
Average B, all atoms (Å2)	21.0

¹intensities measured from amplitudes

The AeGluc crystal structure is shown in Figure 6A. It revealed that AeGluc has 22 α-helices and 17 β-sheets (Figure 6A). Additional circular dichroism analyses had also suggested the structure contained more α-helices compared to β-sheets (Supplementary Figure 8). The crystal structure showed the protein was bound to two calciums (Figure 6B), which have been reported in other α-glucosidases to be necessary for proper folding [53]. The two calciums were on the outside of the protein 32.2 Å apart; one calcium bound D19, D21, D23, I25, and D27, while the second calcium bound N100, D171, Y205, L206, and E208. They flanked a core pocket within the protein structure, which, by an electrostatic density map appeared to be highly negatively charged (Figure 6C). In other α-glucosidases, the active site is located in a similarly large cleft [53,54]. We therefore superimposed the AeGluc structure with a silkworm sucrose hydrolase (PDB: 6LGB), the closest structure by homology available in PDB, which had been solved bound to glucose [53]. The two structures were highly similar, with a root-mean-square difference (RMSD) = 0.903 Å, and the glucose binding site overlaid with the cleft pocket in the AeGluc structure (Figure 6D). This suggested that the inner pocket within the AeGluc was the active site, distant from the calcium binding sites. We therefore performed molecular docking using the AeGluc crystal structure docked with α-glucose (PubChem CID: 79025; a monosaccharide), sucrose (PubChem CID: 5988; a disaccharide), and maltohexaose (PubChem CID: 439606; a polysaccharide) and definted the ligand binding site based on the superimposed structure of Bombyx mori GH13 sucrose hydrolase (PDB: 6LGE). Glide docking scores, where negative scores indicate greater stability, suggested that the polysaccharide maltohexaose bound AeGluc the best (−18.675), followed by sucrose (−8.916) and lastly glucose (−6.992), which had the worst binding score. Indeed, glucose and sucrose displayed 68 and 73 putative contacts with AeGluc, respectively, while maltohexaose displayed 182 putative contacts (Figure 7). Taken together, the docking analyses suggest that AeGluc harbors a deep binding pocket, where smaller sugars can bind weakly, as well as an outer pocket, where larger sugars are stabilized and able to bind strongly.

Figure 6. Crystal structure of AeGluc.

A. Ribbon representation of AeGluc with α-helices in violet, β-sheets in cyan, and loops in green. B. View of AeGluc with 2 calcium ions (black spheres) bound. C. Electrostatic density map where red = negatively charged and blue = positively charged and the greater the color intensity signifies the greater the charge. D. Bombyx mori sucrose hydrolase (PDB: 6LGA; red) superimposed with AeGluc (green).

Figure 7. Molecular docking with sugars of differing links and AeGluc.

Alpha-glucose (A) sucrose (B), or maltohexaose (C) docked to AeGluc. AeGluc is displayed by molecular lipophilicity potential (MLP), where dark cyan represents most hydrophilic and white-to-dark goldenrod represents most lipophilic. Ligands are shown as black stick models and are highlighted in green.

3.4. AeGluc structural model compared to AnGluc and CxGluc models suggests similar functional properties

To compare predicted structures of the AnGluc and CxGluc proteins with our AeGluc crystal structure, we used AlphaFold to model the three α-glucosidases. The coordinates of the predicted models for AeGluc (Figure 8A), AnGluc (Figure 8B), and CxGluc (Figure 8C), retrieved from the AlphaFold Protein Structure Data Base with accession codes AF-P13080, AF-Q7PYT9 and AF-B0XAA1 respectively, were aligned using the secondary structure matching algorithm [55] as implemented in COOT [39]. The RMSD between the AnGluc model (593 amino acid residues) and the AeGluc model (579 amino acid residues) was 0.67 Å with 60.5% identity, and the RMSD between the CxGluc model (584 amino acid residues) and the AeGluc model was 0.56 A with 74.02% identity, based on 562 aligned Cα atoms for both cases. The same alignment as above between the AeGluc crystal structure coordinates (PDB: 8SLV) with those of the AlphaFold predicted model (562 amino acid residues) resulted in an RMSD of 0.86 Å over 557 aligned Cα atoms. Based on these results, we conclude with a high degree of confidence that the structures of the α-glucosidases from the three species are highly similar and will likely possess similar functional properties.

Figure 8. Alpha-fold models of three mosquito α-glucosidases.

(A) AeGluc, (B) AnGluc, and (C) CxGluc predicted structures modeled with AlphaFold with α-helices in violet, β-sheets in cyan, and loops in green.

Because the AeGluc structure revealed AeGluc bound two calcium ions, and because our AlphaFold modeling suggested the three α-glucosidase orthologs possessed similar functional and structural properties, we were curious whether calcium impacted enzyme catalysis for AeGluc, AnGluc, and CxGluc. We therefore performed the sucrose activity assay in the presence of 0 mM, 10 mM or 20 mM CaCl₂ and measured enzymatic activity. We found that increasing concentrations of calcium had no significant impact on enzymatic activity for AeGluc or CxGluc and significantly inhibited enzymatic activity for AnGluc at 10 mM (p = 0.0097) and 20 mM (p = 0.0067) calcium chloride (Figure 9). We therefore conclude that the calcium binding sites are likely not essential for enzyme catalysis.

Figure 9. The impacts of increasing concentrations of calcium on enzymatic activities.

Enzymatic activities of AeGluc (purple), AnGluc (green), or CxGluc (orange) for the hydrolysis of sucrose with or without 10 mM or 20 mM calcium chloride. Black bars indicate mean and standard deviation. ** p < 0.01 by unpaired two-tailed t-test.

4. Discussion

In this study, we successfully expressed and purified three salivary α-glucosidase orthologs from Ae. aegypti, An. gambiae, and Cx. quinquefasciatus. We found that although the recombinant enzymes exhibited a pronounced preference for sucrose compared to NPαGlu, the respective mosquito SGEs displayed similar activities for both substrates. This led us to conclude that there are likely multiple enzymes in mosquito saliva involved in sugar metabolism. In Lutzomyia longipalpis sand flies, at least four α-glucosidases in different tissues have been found to have different biochemical properties, suggesting the enzymes are likely involved in different metabolic processes such as digestion of plant sugars or blood glycoproteins [28]. Along these lines, it would be interesting to study α-glucosidase activities with different substrates in saliva that is secreted after mosquito feedings on different sources. It has also been shown that Ae. aegypti salivary α-glucosidase activity varies during adult development [5]. Our mosquito SGEs were prepared from 5-7 day-old mosquitoes that were maintained on sugar and were never been blood fed, so we did not capture α-glucosidase activity over time or in the context of different life conditions. Perhaps α-glucosidase expression and/or activity is dependent on complex factors such as these.

Using western blot with SGE from the three different mosquito species, we found that mice immunized with recombinant AeGluc, AnGluc, or CxGluc produced sera that displayed minimal cross reactivities against the three α-glucosidase orthologs (Figure 2). Because of this, and because AeGluc, AnGluc, and CxGluc are secreted during mosquito feeding, they may be useful markers for exposure to different vectors in the field. Peng and colleagues previously found that AeGluc (named rAed a 4 in that manuscript) bound IgE in sera derived from 13 individuals who had histories of mosquito allergies and who were bitten by colony reared-Aedes aegypti during the study [16]. The mosquito-allergic individuals had significantly higher mean levels of AeGluc-specific IgE and IgG than unbitten controls [16]. Using serum from patients harboring DENV-specific antibodies, we did not find AeGluc-specific IgG was a reliable marker for previous Ae. aegypti exposure by ELISA (Supplemental Figure 5). Individuals in Peng et al. (2016) had recently been bitten by Ae. aegypti in a laboratory setting, whereas we did not have information about the the number of bites or the time frame when the patients had been bitten by Ae. aegypti in the field [16]. It could be that AeGluc-specific antibodies are not long lasting, or that there is great variability in AeGluc allergy in the field. Controlled studies where individuals are exposed to different species of mosquitoes over time could begin to better address whether the α-glucosidases could be reliable and specific markers for exposure to different vectors in the field.

Although presence of salivary glucosidase activities have been reported in mosquitoes, no experimentally determined high resolution structure exists for any mosquito glucosidse. We therefore successfully crystalized AeGluc and solved its structure at a 2.3 Å resolution and found that AeGluc bound two calcium ions more than 30 Å apart. The closest structure by homology available in PDB, a silkworm sucrose hydrolase (PDB: 6LGB), also bound a single calcium ion, as well as magnesium, and expression efforts suggested that the protein likely required calcium for proper folding [53]. Other α-glucosidases have been shown to bind calcium, including one from Saccharomyces cervisiae, where calcium actually inhibited enzymatic activity [56]. In that study, the authors concluded that calcium binding induced structural changes that resulted in the exposure of hydrophobic residues that likely inhibited sugar binding [56]. Calcium has also been shown to have inhibitory effects on other α-glucosidases from some bacteria [57] and rats [58]. This was similar to our studies with AnGluc, where increasing the concentration of calcium chloride significantly inhibited sucrase activity (Figure 9). On the other hand, we did not observe significant inhibition of catalytic activity with the addition of increasing concentrations of calcium to AeGluc or CxGluc with sucrose (Figure 9).

Another class of sugar digesting enzymes and a major component of human saliva, α-amylases (EC 3.2.1.1), generally require calcium for catalytic activity and are considered calcium metalloenzymes [59–61]. Calcium forms a tight metal-chelate structure necessary for structural integrity and rigidity [59]; for example, the Bacillus paralicheniformis α-amylase (PDB: 1BLI) has a Ca-Na-Ca metal triad that is essential for sugar hydrolysis [54]. Alpha-amylases digest polysaccharides such as starch and have been reported to have multiple carbohydrate surface binding sites (SBS) distinct from the active site, some of which may be close to the calciumbinding site [62–64]. This facilitates binding with starch, a lengthy and insoluble substrate, to the active site [64]. Alpha-glucosidases, on the other hand, preferably hydroylase oligosaccharides relative to polysaccharides, which may fit in the catalytic pocket in its entirety; therefore, calcium-mediated substrate-stabilizing effect seen in α-amylases may not be necessary for catalysis for α-glucosidases. Given that our molecular docking studies found that maltohexaose may bind AeGluc tightly at regions outside the putative inner pocket, it is possible that calcium also has a stabilizing effect with certain substrates.

In summary, we found that three salivary α-glucosidases – AeGluc, AnGluc, and CxGluc – efficiently digested sucrose, a dissacharide composed of glucose and fructose and the main component of nectar [27], with Michaelis-Menten constants in the millmolar range. The AeGluc crystal structure revealed that the enzyme binds calcium, which may stabilize the protein structure. Based on our modeling efforts combined with functional assays, the α-glucosidase orthologs studied herein are highly similar and likely possess similar functional properties.

• Mosquitoes blood and sugar feed

• Mosquito species of medical importance exhibit α-glucosidase activities

• Sugar digestion by mosquito saliva remains understudied

• Structural studies are instrumental in the understanding of biology of salivary glucosidases