Identification and Characterization of a Sialidase Released by the Salivary Gland of the Hematophagous Insect Triatoma infestans *

Sialidases (EC 3.2.1.18) are commonly found in viruses, bacteria, fungi, protozoa, and vertebrates, but not in invertebrates. We have previously reported the presence of a new sialidase activity in the gut of exclusively hematophagous insects of the Triatoma genus, which transmit Chagas’ disease (Amino, R., Acosta, A., Morita, O. M., Chioccola, V. L. P., and Schenkman, S. (1995) Glycobiology 5, 625–631). Here we show that this sialidase is present in the salivary gland ofTriatoma infestans, and it is released with the saliva during the insect bite. The sialidase was purified to homogeneity (>5000 times) to a specific activity of more than 20 units/mg. It elutes from a gel filtration column with a volume corresponding to the size of 33 kDa, and it migrates as a single 26-kDa band in SDS-polyacrylamide gel electrophoresis, which is unusually smaller when compared with other known sialidases. T. infestanssialidase hydrolyzes preferentially α2→3-linked sialic acids at pH 4–8, with maximal activity between pH 5.5 and 6.5, which is compatible with the optimal pH of secreted sialidases. The sialidase is competitively inhibited by 2-deoxy-2,3-dehydro-N-acetyl-neuraminic acid (K i = 0.075 mm) and differently from many sialidases, with exception of Salmonella typhimuriumsialidase, it is inhibited competitively by HEPES (K i = 15 mm). The fact that T. infestans sialidase is released with the saliva and can hydrolyze sialyl-LewisX blood groups, which are the ligands for selectins, suggests that it might have a role in the blood feeding.

The biological functions of sialidases (EC 3.2.1.18) are mainly related to the catabolic degradation of sialic acid residues linked to mono-or oligosaccharide chains of glycoconjugates. (1). Although sialidases are primarily present in the lysosomal compartments of vertebrates, these enzymes are also found in secreted forms, and at the surface of virus, bacteria, fungi, and some protozoa (2). Some of these microorganisms live in close contact with animals that contain sialic acids, suggesting that sialidases have a nutritional role, or can be virulence factors (2,3). Due to a certain degree of similarity between different sialidases from phylogenetically distant microorganisms, it has been proposed that these enzymes might have evolved by horizontal gene transfer during the contact between microorganisms and vertebrate host (4).
Sialidases have been rarely described in invertebrates, which also lack sialic acids. Exceptions are the sialidase of the leech Macrobdella decora, which feeds on blood (5), the sialidase of the starfish Asterias rubens (6), and the sialidase of the shrimp Peanaeus japonicus (7), enzymes involved in digestive processes. Recently, we have identified a sialidase activity in the gut of Triatoma genus, an hemiptera insect that exclusively feeds on blood (8). The presence of a sialidase in these insects, although expected due to the high content of sialylated molecules in blood, is unique. Other insects that feed on blood, including Rhodnius species, which are similar blood-sucking hemiptera, lack detectable sialidase activity (8). Interestingly, Triatoma and Rhodnius are definitive hosts of the protozoan parasite Trypanosoma cruzi that cause Chagas' disease in humans, and this parasite, as well as other Trypanosoma species, have sialidase activity, which is unusual among protozoa (9 -12). Trypanosoma sialidases contain the structural motives found in bacterial sialidases (11,13), suggesting that these enzymes could have a common ancestor. In addition, parsimony analysis of the conserved motifs of several Trypanosoma sialidases suggests that a common ancestor might have appeared in a protozoan living in the insect vector (14). Therefore, the parasitism of Triatoma species by Trypanosoma could be related to the acquisition of a sialidase by these species.
To investigate a possible evolutionary relationship between Triatoma and other sialidases, we started its purification and characterization. We found that T. infestans sialidase is produced and released by the insect salivary gland, suggesting that it might have a function during blood feeding. The enzyme was purified from the salivary gland, and based on its biochemical properties, we suggest that it is closer to Salmonella typhimurium sialidase than to that of Trypanosoma.

EXPERIMENTAL PROCEDURES
Insects-T. infestans were grown at the Laboratório de Xenodiagnóstico, Instituto Dante Pazzanese de Cardiologia, Sã o Paulo, Brazil, in glass cylinders at 28°C. The insects were fed on ducks between each molt. Adults were used in this study, and the salivary glands were obtained after removal of the head. The bodies were immersed in saline and the first thoracic segment excised under a binocular microscope. The salivary glands were removed and kept on ice. The luminal content of the glands were obtained from glands washed with phosphate saline buffer, pH 7.4. Five gland pairs were punctured with a needle, and the expelled material was collected in 50 l of the saline buffer. The glands were washed three times with 50 l of saline buffer, and the aqueous fractions pooled and used to measure sialidase activity.
Sialidase Activity Measurements-Activity was determined by the release of free sialic acid from sialyllactose. The free sialic acid was measured by the thiobarbituric acid procedure modified by Powell and Hart (15). The standard assay was done in final volume of 40 l containing 0.2 M sodium acetate, pH 5.5, and 0.5 mM siayllactose (Boeh-ringer Mannheim) for 20 -30 min at 37°C. After the thiobarbituric acid reaction, the products were detected spectrophotometrically at 549 nm or by high performance liquid chromatography (15). One enzymatic unit was defined by the release of 1 mol of sialic acid per min at 37°C, using 0.5 mM siayllactose at pH 5.5. Alternatively, the activity was measured by the hydrolysis of 0.2 mM of 2Ј-(4-methylumbelliferyl)-␣-D-N-acetylneuraminic acid (MuNANA) 1 at 37°C.
Histochemical Reactions for Sialidase Detection-Dissected glands were either fixed in 4% paraformaldehyde, 206 mM KCl adjusted to pH 7.4, or frozen in CO 2 . The fixed material was dehydrated, embedded in paraffin, sectioned, and stained with hematoxylin-eosin by conven-FIG. 1. Sialidase activity is present in salivary glands of T. infestans and released during feeding. a, the content of 10 D1, D2, and D3 salivary glands was collected and diluted in saline solution, and the sialyllactose hydrolysis was determined by the thiobarbituric acid assay as described under "Experimental Procedures." The values are means of two determinations. b, plates A show glands stained with hematoxylin and eosin, and plate B sections are stained for sialidase activity with X-NANA. The closed arrow indicates the absence of activity in the epithelium of a D1 gland, while the open arrow shows the activity in the epithelium of a D2 gland. c, hydrolysis of MuNANA onto a 35-mm Petri dish containing 0.8% agarose and bitten by three adult T. infestans. The white spots reveal the release of sialidase during the bites. tional procedures. In situ activity measurements were made using 10 -20-m frozen sections of frozen glands obtained in the cryostat. Sections were incubated for 1 h at 37°C in a solution containing 0.5 mg/ml 2-O-(5-bromo-4-chloro-3-indolyl)-␣-D-N-acetylneuraminic acid (X-NANA, Chemica Alta Ltd, Edmonton, Canada), 3.5 mM K 3 (F(CN) 6 ), 5 mM K 2 (F(CN) 6 ), 0.1 M sodium acetate, pH 5.5 (16). To detect sialidase activity released during the bite, insects were put in contact with 35-mm Petri dishes warmed at 37°C, containing 0.8% agarose in 0.2 M sodium acetate buffer, pH 5.5. The insects were removed and the plates covered with MuNANA. After 5 min, the agarose was removed from the plate, observed under short UV light, and photographed with a Polaroid 667 film and a Kodak yellow filter.
Sialidase Purification-The standard purification procedure was done with 50 dissected gland pairs (D1 and D2). The glands were homogenized in 2 ml of a solution of 2 mM Tris-HCl, 100 mM glycerol, pH 7.4, in the Potter at 4°C. The homogenate was centrifuged 1 h at 100,000 ϫ g, and the resulting supernatant was incubated for 1 h at 4°C with 0.5 ml of DEAE-Sephacel (Amersham Pharmacia Biotech) preequilibrated in the same buffer. The unbound fraction was collected and dialyzed overnight at 4°C against 50 mM sodium acetate, pH 5.2, and 100 mM glycerol. After dialysis, the material was subjected to chromatography in a Mono-S column equilibrated in the dialysis buffer. The enzyme was eluted with a linear gradient up to 0.4 M NaCl in the same buffer. Fractions containing activity were pooled, concentrated to 0.2 ml in a Centricon 10 spin filter (Amicon), and applied into a Superdex 75 column (Amersham Pharmacia Biotech) equilibrated in 50 mM sodium acetate, pH 5.2, 100 mM glycerol, and 0.15 M NaCl. The purified enzyme was stored frozen at Ϫ20°C. Protein was quantified by the Bradford method (Bio-Rad), using serum albumin as standard. The protein concentration obtained after the Mono S and Superdex was estimated by the absorbance at 280 nm, supposing an extinction coefficient of 1.0. In order to visualize the protein on SDS-PAGE, the gels were silver-stained (17). Alternatively, the pool obtained from the Mono S column was concentrated and labeled with 125 I (Amersham Phamacia Fractions 22 through 28 were then concentrated and fractionated in 15% SDS-PAGE, fixed, dried, and exposed to an x-ray film. The sizes indicated in panel a were deduced from standards (ovalbumin, carbonic anhydrase, and cytochrome c) and in b correspond to bovine serum albumin, carbonic anhydrase, trypsinogen, soybean trypsin inhibitor, and lysozyme. Biotech) by the iodogen method (18). Labeled proteins were dissolved in 0.1 M Gly-Tyr (Sigma) to remove free iodine, and filtered through a Sephadex G-25 column to remove labeled Gly-Tyr and free, or unreacted, iodine. After a new round of concentration, the sample was applied into the Superdex 75 column as above.
Selectin Binding Assays-Polyacrylamide-Sialyl-Lewis X (Syntesome, Munchen, Germany) was immobilized in 96-well plates and treated with, or without, the purified sialidase in 0.1 M sodium acetate, pH 5.5. After treatment, the wells were washed and blocked, and the binding of immunoglobulin chimeras of L-and P-selectins to the modified carbohydrates was measured by enzyme-linked immunosorbent assay as described in O'Connell et al. (19). Controls included the addition of 10 mM EDTA that inhibited selectin binding.

RESULTS
When we detected sialidase activity in the gut of T. infestans, we observed that most of the linked sialic acid was rapidly released from the blood ingested by the insect (8). As the anterior midgut of hematophagous hemiptera is supposed to contain few digestive enzymes (20,21), the origin of sialidase could be the salivary glands. Therefore, we tested for the presence of this activity in the complex of three glands found in T. infestans (22). As shown in Fig. 1a, the luminal content of the secretory glands D1 and D2 contained large amounts of sialidase activity. The D3 gland is thought to be a liquid storage compartment, and in fact contained much less activity. To confirm the presence of the sialidase and to localize the activity in the glandular tissue we stained frozen sections of D1 and D2 glands with the indigogenic substrate for sialidase X-NANA. A strong reaction was detected in both glands (Fig. 1b). While D2 gland displayed more activity in the epithelium, the activity of the D1 gland was found mostly in the lumen. Subsequently, to demonstrate that the enzyme is released with the saliva during the insect feeding, we took advantage of the fact that when starved T. infestans bite into any warmed surface, they release saliva. As shown in Fig. 1c, agarose plates bitten by a few specimens, and coated with MuNANA, revealed fluorescent spots in the positions where the insect maxilla was inserted, confirming that sialidase is released with the saliva.
Next we purified the sialidase from the salivary glands. After hypotonic shock, most of the enzymatic activity was released in a soluble form, and did not bind to a DEAE-Sephacel at pH 7.4. As the enzyme had a basic character, it also eluted at low ionic strength from a Mono-S column at pH 5.2 (Fig. 2a), being purified about 20 times at this step. The following purification step was a gel filtration column (Fig. 2b). After chromatography the sialidase was highly enriched (purification factor larger than 200 times), because the enzyme eluted with a size corresponding to 33 kDa, differently from most of the proteins of the salivary glands, which have between 14 and 20 kDa.
The above purification procedure was repeated several times with identical profiles, which were routinely followed by SDS-PAGE and silver staining. The initial supernatant contained a large number of bands from 12 to 100 kDa (Total, inset of Fig.  2a). After the DEAE and Mono S column the material was enriched with proteins of smaller size. The gel filtration step separated the sialidase activity from most of the 14-kDa contaminants and no protein could be detected by silver staining even loading an equivalent of a preparation of 50 salivary glands (inset of Fig. 2b). Table I summarizes the purification procedure, which yielded 20% of the initial activity with more than 20 units/mg of protein. This value is comparable to other purified sialidases. It might be underestimated, since we could not measure precisely the concentration of protein, and rarely we could detect a deflection in the absorbance at 280 nm during the elution of the Superdex column. The purification procedure was highly reproducible, including the enzyme isolated from the gut (not shown), suggesting that the sialidases found in the salivary gland and in the midgut may be similar.
To identify the sialidase and determine its degree of purity, activity containing fractions eluted from the Mono S column were concentrated and labeled with 125 I by the IODO-GEN method. The radiolabeled sample was then filtered through the Superdex column and the eluted fractions analyzed for sialidase activity, and radioactivity. As seen in Fig. 3a, the activity was found in fractions 23, 24 and 25, while most of the radiolabel was detected after fraction 26. The samples 21-26 were concentrated separately and analyzed in a 15% SDS-PAGE. A single band of 26 kDa co-eluted with the enzymatic activity as shown in the autoradiogram of Fig. 3b. Most of the contaminants had a size below 14 kDa, in agreement with the pattern found in the silver stained gel (see Fig. 2b, inset).
As shown in Fig. 4a, the purified sialidase is active in a broad pH interval with maximal activities from pH 5.0 to 7.0, confirming that it can hydrolyze sialic acids in the blood at physiological conditions. Double reciprocal plots of the rate of hydrolysis at different concentrations of ␣233-sialyllactose and MuNANA allowed us to estimate the K m values of 1.15 and 0.19 mM for these two substrates, respectively (Fig. 4b). Although, the estimated concentration of the protein is uncertain, the V max values (ϳ50 for sialyllactose and ϳ140 units/mg for Mu-NANA) are compatible with a high turnover number by this enzyme (66 s Ϫ1 for sialyllactose and 190 s Ϫ1 for MuNANA). T. infestans sialidase showed activity toward glycoproteins containing ␣233-linked sialic acid such as fetuin and was able to hydrolyze sialic acids of the mucin-like glycoproteins of T. cruzi. The enzyme, however, was unable to hydrolyze of ␣236containing molecules.
The purified sialidase was competitively inhibited by 2Јde- oxy-3Ј-dehydro-N-acetyl-␣-D-neuraminic acid (DANA), with a K i of 0.075 mM (Fig. 5a). Interestingly, HEPES buffer inhibited competitively the purified sialidase, although with a much higher K i (15 mM) (Fig. 5b). The inhibitory effect of HEPES was equally strong at pH 7 or 5.5, suggesting that protonation of the piperazine was not essential for inhibition. From all sialidases tested, only T. infestans and S. typhimurium sialidases were inhibited by HEPES. No inhibition was found for Clostridium perfringens, Arthrobacter ureafaciens, Vibrio cholerae, and T. cruzi sialidases (Fig. 6). HEPPS, which is similar to HEPES, but contains a propane sulfonic acid, instead of an ethane sulfonic acid, was less inhibitory, suggesting that the catalytic pocket might limit the entrance of large molecules.
The fact that T. infestans sialidase is released during the insect bite suggests that it could interfere with the blood intake or affect hemostasis and inflammation after the insect feeding. Thus, we studied the specificity of the enzyme against sialyl-Lewis X , known as ligands for selectins, which mediate cell adhesion during leukocyte migration, platelet adhesion/aggregation, and several inflammatory processes (23)(24)(25). Fig. 7 shows that incubation of sialyl-Lewis X adsorbed to 96-well plates with the purified sialidase decreased in a dose dependent manner the binding of chimeric P-and L-selectin as detected by an enzyme-linked immunosorbent assay. DISCUSSION We have identified and characterized a sialidase in the salivary gland of a hematophagous insect. A large amount of enzymatic activity is released during the insertion of the maxilla in the host tissue, when the insect is taking the blood meal. The enzyme is active at physiological pH and can remove sialic acid from molecules involved in cell migration and inflammatory reaction. Therefore, the sialidase may have an important role for the insect feeding.
We were very successful in purifying the sialidase from the salivary gland. Initially, the basic character of the enzyme (similar to S. typhimurium sialidase) (26), allowed us to remove several contaminants by two anion exchange columns. More impressive was that most of the proteins of the salivary gland have low molecular size (Ͻ20 kDa), and therefore, we could greatly enrich the enzyme in a single gel filtration step. The final specific activity was quite high and comparable to several other purified sialidases. Moreover, the enzyme appeared quite homogeneous after iodine labeling, eluting in the gel filtration column with a molecular size of 33 kDa, similar to the size found by SDS-PAGE (26 kDa). These sizes are small when compared with other known sialidases (2), and might distinguish T. infestans sialidase. The fact that the sialidase is found associated with the epithelium of D2 glands suggests that it is synthesized in this site, it accumulates in gland D1, and it is released with the saliva. Eventually the enzyme found in the midgut is originated in the salivary gland. We cannot exclude, however, that a similar (or identical) enzyme is also produced in the midgut.
In general two class of sialidases have been characterized. The large ones are represented by V. cholerae sialidase, which has a molecular size of 82 kDa and requires calcium. The second class is the calcium-independent sialidases, such as S. typhimurium sialidase, with molecular size around 42 kDa (27). Coincidentally the sialidase purified from the shrimp Penaeus japonicus, which is also an arthropod, is 32 kDa (7), and the leech sialidase, also from an invertebrate, is 42 kDa (5). We are working to obtain sequence information that will be important for testing whether the enzyme is related to other bacterial, viral, protozoan, or mammalian sialidases.
Our preliminary kinetic characterization of T. infestans sialidase also provided evidence that it is a unique enzyme. Its kinetic parameters resemble viral sialidases showing strong specificity to ␣233-linked sialic acids, and a relatively high K m value for sialyllactose (1.15 mM) when compared with the K m of several bacterial sialidases (3). It has activity in a broad pH range, and it is inhibited by HEPES, which does not inhibit V. cholerae, A. urefaciens, C. perfringens, and the T. cruzi enzyme. The exception was the S. typhimurium sialidase. However, differently from this latter enzyme, T. infestans sialidase is inhibited by relatively low DANA concentrations (K i ϭ 0.07 mM) while S. typhimurium sialidase has a K i ϭ 0.38 mM (26). T. infestans sialidase has a high turnover number, smaller than the values found for S. typhimurium (28), but higher than that of influenza neuraminidase. Unlike that from T. cruzi, it hydrolyzes sialyl-Lewis X but does not display detectable transsialidase activity. A more detailed kinetic analysis of T. infestans sialidase might contribute in revealing new features of catalysis by sialidases, which are shown to vary despite similar protein folding (29) and environment of the active site (30). For example, the finding that inhibition of sialidases by 2-hydroxyethyl piperazines is competitive, although of low affinity, may help to elucidate the differences in the active sites of these enzymes as well as to develop new sialidase inhibitors.
Several mechanisms are used by hematophagous insects to circumvent host reactions that could be deleterious for the insect. Most of these mechanisms appear to be redundant and variable from species to species (reviewed in Ribeiro (31)). In the case of triatomine insects, the feeding, which takes minutes, passes unperceived by the animal. Such insects feed directly from inside venules and arterioles by the food channel, a channel formed by mouth parts. At the same time that blood is sucked, saliva is released by the salivary channel (32). It is known that triatomine saliva prevents blood coagulation and platelet aggregation and promotes vasodilatation in order to obtain the maximal amount of blood and to avoid clogging inside the food channel (31). As sialic acids have been shown to participate in several processes involved in hemostasis, it is possible that the release of sialidase during blood feeding interferes with coagulation and platelet aggregation. For example, platelet aggregation induced by the Von Willebrand factor depends on the negative charge of glycoprotein Ib due to the presence of sialic acid in the hinge region (33). During acute inflammation, ␣-1 acid glycoprotein also inhibits platelet aggregation by thrombin, collagen, and ADP, acting on the second wave of aggregation. Removal of sialic acid from ␣-1 glycoprotein decreases its inhibitory effect (34). On the other hand, desialylation of plasminogen with a sialidase specific for ␣233linkages, promotes fibrin proteolysis without the requirement of tissue plasminogen activator (35). ␣233-linked sialic acids of fibrinogen are also low affinity calcium-binding sites for fibrin assembly (36). The degree of fibrin sialylation is shown to mediate the activity of plasminogen and to regulate fibrinolytic activity (37).
Sialic acid is also involved in several inflammatory processes. Injection of 0.1 milliunit of sialidase inhibits the migration of polymorphonuclear cells to the inflammatory site (38). This amount of sialidase could be injected by T. infestans, which contains about 0.3 milliunit of sialidase in the salivary glands (see Table I). Moreover, the finding that T. infestans sialidase can desialylate sialyl-Lewis X antigens, which participate as ligands in the interaction of leukocytes with endothelial cells, platelet degranulation, and lymphocyte homing through the binding of selectins (23)(24)(25), supports the idea that the sialidase release during blood feeding may have a relevant physiological role. Sialic acids also modulate the degree of histamine release by mast cells induced by the peptidergic pathway (39,40). As the purified sialidase did not affect blood coagulation and platelet aggregation (data not shown), it is quite possible that it might act in combination with other factors also present in the saliva. Indeed, a potent inhibitor of thrombin is described in the saliva of Triatoma pallidipennis (41). Thus, the presence of sialidase activity in the salivary gland of an hematophagous insect opens a new perspective in the understanding of the mechanism involved in hemostasis control during blood ingestion.