INTRODUCTION
Leishmaniasis is one of the most prevalent insect-transmitted human diseases. The vectors are sand fly females (Diptera: Phlebotominae), whose saliva contains a mixture of antihaemostatic and immunomodulatory molecules. In naive hosts, sand fly saliva enhances the initial phase of Leishmania infection increasing the parasite burden as well as the lesion size (reviewed by Gillespie, Mbow & Titus, 2000; Valenzuela, 2002). On the other hand, in pre-exposed hosts, sand fly saliva stimulates an immune response, which may block enhancing effects. The protection is provided by immunization with saliva (Belkaid et al. 1998) or by the bites of uninfected sand flies (Kamhawi et al. 2000). Accordingly, a usage of sand fly salivary proteins in transmission blocking vaccine has been discussed (reviewed by Reed, 2001; Valenzuela, 2002).
Experimental animals repeatedly exposed to sand flies develop anti-saliva antibodies (Belkaid et al. 1998; Ghosh & Mukhopadhyay, 1998; Valenzuela et al. 2001; Volf & Rohousova, 2001). Recently, Barral et al. (2000) and Gomes et al. (2002) demonstrated the production of antibodies against the sand fly saliva also in humans; they described the presence of specific IgG antibodies against the New World vector Lutzomyia longipalpis in sera of children living in an endemic area of visceral leishmaniasis in Brazil. In addition, a significant correlation was found between anti-L. longipalpis IgG levels and anti-Le. chagasi DTH response (Barral et al. 2000) indicating that the study of anti-sand fly saliva antibodies should be considered as an integral part of the leishmaniasis epidemiology. So far as we are aware, there is no study about human anti-sand fly antibody response from the Old World.
The composition of sand fly saliva varies between different species and, what is more important, the salivary antigens recognized by sera of experimentally bitten animals are species-specific (Volf & Rohousova, 2001). The purpose of this work was to investigate whether such specificity of antibody response could be demonstrated also in humans and whether human sera recognize the same salivary antigens as sera of the experimental mice. In a Leishmania tropica focus in Sanliurfa, Turkey, we compared the antibody response against Phlebotomus papatasi and P. sergenti in patients with active leishmania lesions and healthy individuals living in the same houses.
MATERIALS AND METHODS
Salivary gland lysate
Sand flies were reared in standard conditions as described previously (Volf & Rohousova, 2001). Three sand fly colonies were used: two Old World species, P. papatasi (origin from Sanliurfa, Turkey), P. sergenti (Sanliurfa, Turkey), and the New World sand fly L. longipalpis (Jacobina, Brazil). Previous study showed that salivary proteins differ between colonies of different geographical origin (Volf, Tesarova & Nohynkova, 2000); therefore, both P. papatasi and P. sergenti colonies used originated from the same locality as the human sera. Salivary glands were dissected out from 5 to 10-day-old females into cold Tris buffer (20 mM Tris, 150 mM NaCl, pH 7·6). Groups of 20 glands in 20 μl of Tris buffer were kept at −20 °C. Just before use, the salivary gland lysate (SGL) was prepared by disruption of glands by 3 freezing/thawing cycles.
Sera
Mice used in this study were housed in the animal house of the Faculty of Science; they were maintained and handled following the guidelines and legislation for the care and use of animals for research purposes. Mouse sera were obtained from BALB/c mice repeatedly bitten by P. papatasi, P. sergenti, and L. longipalpis, respectively. We used a similar protocol of mice exposure as previously described by Volf & Rohousova (2001). Mice were anaesthetized (ketamin 150 mg/kg and xylazin 15 mg/kg, intraperitoneally) and exposed to about 50–100 sand fly females, once a week in a total of 9–11 exposures. Mice were bled 1 week after the last exposure and individual sera were used in the experiments. Human sera were collected during August and September 1999 in the endemic area of Le. tropica in Sanliurfa, the provincial capital of the South-East Anatolia, Turkey. P. sergenti and P. papatasi are dominant species in this area and represented more than 99% of local sand fly fauna (Volf et al. 2002). Eighty-seven serum samples from one part of Sanliurfa were tested. Fifty-four sera were obtained from individuals with progressing Le. tropica lesions (group SU+). There was not any restriction on the age of the lesion and after serum sampling the patients were treated by Glucantime®. Thirty-three serum samples were from apparently healthy individuals without Leishmania lesion (group SU−). The control group consisted of 50 individuals from Prague, the Czech Republic (group PRG), the sand fly-free area in Central Europe.
ELISA
Specific anti-saliva IgG was measured by enzyme-linked immunosorbent assay (ELISA). In preliminary experiments, the assay was optimized using sera of experimentally bitten mice. Microtitre plate wells were coated with SGL (about 60 ng of protein per well) in 0·01 M carbonate-bicarbonate buffer (pH 9·6) overnight at 4 °C. The wells were then washed in PBS buffer with 0·05% Tween 20 (PBS-Tw) and incubated with 6% horse serum in PBS-Tw for 45 min at 37 °C to block free binding sites. After another washing, human sera were diluted 1[ratio ]50 in 3% horse serum/PBS-Tw and incubated in duplicate for 90 min at 37 °C. This step was followed by incubation with peroxidase-conjugated anti-human γ-chain specific antibodies (Sigma) at a 1[ratio ]2000 dilution for 45 min at 37 °C. Orthophenylendiamine and H2O2 in McIlwein phosphate-citrate buffer (pH 5·5) were used as the substrate solution. The absorbance was measured using the Multiskan RC ELISA reader (Labsystems) at 492 nm wavelengths. The data obtained by ELISA test were not normally distributed. Therefore, the non-parametric Wilcoxon Rank-Sum Test for Difference in Medians was used to analyse statistical significance (P-value=0·05). The cut-off values were determined by the addition of two standard deviations to the mean of optical densities obtained with sera from the control group.
Immunoblot
Sand fly salivary proteins were separated by SDS-PAGE on 13% gels, under reducing conditions using the Mini-Protean III apparatus (Bio-Rad). The equivalent of 10–15 glands was loaded in each lane. Separated proteins were electro-transferred onto nitrocellulose membrane by Semi-Phor equipment (Hoefer Scientific Instruments). After transfer, the first part of the membrane was stained by AmidoBlack (Bollag & Edelstein, 1991). On the second part free binding sites were blocked by 5% low-fat dried milk in Tris buffer with 0·05% Tween 20 (Tris-Tw) for 1 h and the membrane was cut into strips. The strips were then incubated with mouse or human sera (diluted in Tris-Tw, 1[ratio ]100 and 1[ratio ]50, respectively) for 1 h. Then, the strips were washed several times in Tris-Tw and incubated for 1 h with the peroxidase conjugate. Goat anti-mouse or goat anti-human γ-chain specific antibodies (both Sigma, dilution 1[ratio ]750 in Tris-Tw) were used. The colour reaction was developed using substrate solution containing diaminobenzidine and H2O2.
RESULTS
Immunoblot – mouse sera
In SDS-PAGE, sand fly SGL protein profiles differed significantly among species. From 14 to 15 prominent protein bands with molecular masses ranging from 12 to 70 kDa were visualized in each SGL (Fig. 1).

Fig. 1. (A) Salivary gland lysate (SGL) of Phlebotomus papatasi (Pap), P. sergenti (Ser), and Lutzomyia longipalpis (Lon): protein profiles stained by AmidoBlack. (B) Immunoblots with individual sera of mice repeatedly bitten by one of these species: anti-P. papatasi (1), anti-P. sergenti (2), and anti-L. longipalpis (3).
Phlebotomus papatasi, P. sergenti, and Lutzomyia longipalpis bites stimulated production of high levels of specific anti-saliva IgG antibodies. In all sand fly species, sera reacted strongly with homologous antigen (Fig. 1). Western blot analysis revealed 4–6 major antigenic bands in each species; the majority of them were species specific. There were one or two very faint bands in the strips where P. papatasi antigen was incubated with anti-P. sergenti serum and vice versa. No reaction was found in the control strips where normal mouse serum was used (Fig. 1).
ELISA – human sera
ELISA tests revealed high levels of specific anti-P. papatasi (Fig. 2A) and anti-P. sergenti IgG antibodies (Fig. 2B) in some individuals from the endemic area (Sanliurfa, Turkey). In most cases, individuals that gave a positive reaction with P. papatasi antigens, reacted positively also with P. sergenti SGL and vice versa (data not shown). The Wilcoxon test indicated that anti-Phlebotomus IgG levels in Sanliurfa sera were significantly higher (P<0·001) when compared with PRG sera. However, Sanliurfa sera presented antibodies at different levels; anti-P. papatasi IgG levels higher than the cut-off value were found in 40·2% (35/87) and anti-P. sergenti IgG levels in 45·2% (38/84) of sera from Sanliurfa (Fig. 2A,B). Despite the wide variation, individuals with leishmaniasis revealed significantly higher anti-P. sergenti IgG levels when compared with healthy individuals from the same place (P<0·001). Anti-P. papatasi IgG levels were equal in both Sanliurfa groups (P>0·05).

Map 1. Human IgG response against salivary gland lysate of the sand fly Phlebotomus papatasi (A) and P. sergenti (B). Human sera from the sand fly-free area in Prague, the Czech Rep. (PRG, □) and from the endemic area in Sanliurfa, Turkey: healthy individuals (SU−, ○) and patients with leishmaniasis (SU+, [bull ]). The lines represent cut-off values determined by addition of two standard deviations to the mean of optical densities obtained with sera from the control group. The bars represent median values of each group.
The specificity of ELISA was confirmed using salivary gland lysate of the New World sand fly, L. longipalpis. No positive reaction of human sera was observed when L. longipalpis SGL was used as an antigen and, moreover, the anti-Lutzomyia IgG level in Sanliurfa sera was significantly lower (P=0·013) when compared with control sera (data not shown).
Immunoblot – human sera
To confirm the specificity of the antibody response against P. papatasi and P. sergenti saliva, we tested sera of 15 individuals from the endemic area (either positive or negative by ELISA, from both SU− and SU+ groups) and five from the control group. Each serum was tested with the SGL of 3 sand fly species.
Positive sera of individuals from Sanliurfa recognized a variety of P. sergenti and P. papatasi salivary proteins. In each SGL they reacted with 1–6 bands ranging from 20 to 70 kDa. In P. papatasi SGL the prominent and consistently present was a single protein band of 30 kDa, whereas in P. sergenti SGL sera showed individual patterns (Fig. 3A and B). No differences in the reactivity between the SU− and SU+ groups were found (data not shown), some human sera, however, reacted strongly with P. papatasi SGL but faintly with P. sergenti SGL and vice versa. In contrast, sera from the Prague group did not react with any proteins in P. papatasi SGL, some faint bands were noticed with P. sergenti antigens (Fig. 3A and B). The specificity of anti-Phlebotomus IgG response was confirmed with salivary antigens of L. longipalpis. Sera of both human groups (Sanliurfa and Prague) gave only a very weak or no reaction with SGL of this New World sand fly (Fig. 3C).

Fig. 2. Immunoblots of Phlebotomus papatasi (A), P. sergenti (B), and Lutzomyia longipalpis (C) SGL with sera of individuals from the endemic area (Sanliurfa, Turkey, lanes 1–10) and sand fly-free area (Prague, the Czech Rep., lanes 11–15). The numbers refer to the same sera in parts A, B, and C.
Mouse versus human antibody response
Human and mice antibodies recognized similar salivary antigens; differences were, however, found in the intensity of reaction. In P. papatasi human IgG reacted strongly with a 30 kDa antigen, while mouse IgG recognized preferentially a 42 kDa protein band (Fig. 4A). Moreover, the 36 kDa antigen of P. papatasi was recognized by most human sera but none of the 3 mouse sera tested (Fig. 4A). In P. sergenti SGL the differences were found especially in the intensity of reaction with 12 and 13 kDa polypeptides (Fig. 4B). Despite the fact that in this test more concentrated SGL was used to highlight all antigenic bands, the control sera remained negative.

Map 2. Immunoblots of Phlebotomus papatasi (A) and P. sergenti (B) SGL with the positive human (1–6) and mouse (8–10) sera. Lane 0: total protein profile of SGL stained by AmidoBlack. Controls: human serum from Prague (lane 7) and normal mouse serum (lane 11).
DISCUSSION
This study showed that hosts bitten experimentally or naturally by sand flies produce antibodies that recognize the species-specific antigens in sand fly saliva.
In humans, the presence of anti-saliva antibodies was demonstrated in mosquito-allergic patients or in populations continuously exposed to mosquito bites; both IgG and IgE anti-mosquito antibodies were detected (Penneys et al. 1989; Konishi, 1990; Brummer-Korvenkontio et al. 1994, 1997b; Reunala et al. 1994; Peng, Yang & Simons, 1996; Peng et al. 2002). Recently, anti-sand fly antibodies have been found in sera of individuals living in the endemic area of visceral leishmaniasis in Brazil (Barral et al. 2000; Gomes et al. 2002). As far as we are aware, there exist no similar studies of human anti-sand fly antibody response in the Old World.
Approximately 40% of sera collected in Sanliurfa showed high levels of IgG antibodies against saliva of P. sergenti and P. papatasi, the 2 most abundant sand fly species in this area (Volf et al. 2002). P. sergenti is a proven vector of Le. tropica while P. papatasi is refractory to this parasite (reviewed by Sacks, 2001). Interestingly, Le. tropica patients possessed significantly higher anti-P. sergenti IgG levels than the healthy individuals from the same place while anti-P. papatasi IgG levels were equal in both groups. High levels of anti-P. sergenti antibodies may reflect more frequent exposure to vector bites and therefore the higher probability of Le. tropica transmission.
In visceral leishmaniasis focus in Brazil, Barral et al. (2000) and Gomes et al. (2002) found that individuals with positive DTH response (marker of protection against Le. chagasi in subclinical cases) had increased anti-L. longipalpis IgG levels while those who experienced seroconversion (marker of Le. chagasi infection) did not have increased anti-saliva antibody response. However, similar to our data, they found significantly higher levels of anti-saliva antibodies in individuals positive in both markers compared to individuals who did not experience Le. chagasi infection. This again may suggest a link between the infection and the frequency of exposure to vector bites.
Most data obtained in bloodsucking Diptera suggest that antigenic specificity is linked to the phylogenetic distance of the insect taxons and species-shared antigens are frequent between related species. Both species-specific and species-shared antigens have been found among Simulium species (Cross et al. 1993) and among different mosquitoes (Brummer-Korvenkontio et al. 1997a; Peng & Simons, 1997; Peng, Estelle & Simons, 1998). The specificity of the antibody response against mosquito saliva was less obvious in humans (Peng & Simons, 1997; Peng et al. 1998) than in experimentally bitten rabbits (Brummer-Korvenkontio et al. 1997a).
The anti-sand fly antibody response seems to be more species-specific. Results obtained in mice are in agreement with the previous work on P. papatasi, P. perniciosus, and P. halepensis (Volf & Rohousova, 2001). The present study, however, reports the first evidence that similar specificity could be demonstrated in sera of humans from an endemic area as well. In both methods used, human antibodies primarily recognised antigens from the local Phlebotomus species and did not react with Lutzomyia SGL. In contrast to the study of Barral et al. (2000), we did not observe cross-reactivity between L. longipalpis and P. papatasi antigens, not by ELISA nor by immunoblotting. It is possible that people in Brazil are exposed to some widespread antigen with similar epitopes to P. papatasi saliva and cross-reactivity observed by Barral et al. (2000) could be an artifact due to the non-specific reaction.
Among 14 main bands identified in P. papatasi and P. sergenti saliva using SDS-PAGE, most were antigenic and reacted with both, mouse and human sera. Differences were, however, found in the intensity of reaction. Studies done in mosquitoes, showed that sera of experimental mice reacted with more salivary proteins than did the human sera (Chen, Simons & Peng, 1998); the main salivary antigens detected by animal and human antibodies were, however, the same (Brummer-Korvenkontio et al. 1994, 1997a).
The present work showed that mouse anti-P. papatasi antibodies recognized preferentially a 42 kDa protein band. On the other hand, all positive human sera reacted strongly with the 30 kDa band, suggesting that this protein could be a major antigen of P. papatasi saliva for humans. In addition, the 36 kDa P. papatasi antigen was recognized exclusively by human sera but subsequent studies with more mouse sera are needed to validate this hypothesis. The low complexity of SGL and data obtained on P. papatasi by Valenzuela et al. (2001) allow prediction of the identity of these main antigens. The 30, 36 and 42 kDa protein bands were running to the molecular weights corresponded to D7 protein, apyrase, and Yellow protein, respectively (Valenzuela et al. 2001). Detailed characterization of these main antigens will be the aim of further study.
The history of human exposure (the intensity and the frequency of sand fly bites) was different from that of experimentally bitten mice. For all that, the differences between mouse and human antibody response suggest that results obtained by vaccination on mouse models could not be applied to humans indiscriminately. It would be interesting to know whether antibodies to P. papatasi 30 kDa protein confer the protective effect in Leishmania transmission and whether immunization by this antigen would inhibit the parasite infection in the way similar to SP15 (Valenzuela et al. 2001).
In conclusion, characterization of the immune response to sand fly saliva is useful in epidemiological studies. Considering the antigenic variability among sand fly species, we predict that the blocking effect of anti-sand fly immune response on Leishmania transmission is species-specific. Additionally, our results suggest that the antibody response to sand fly saliva could be used for monitoring the exposure of humans, domestic animals, and other hosts to sand flies and for the identification of sand fly species spectrum in a certain area. In endemic areas, the host responses to sand fly bites might be used as a marker of risks for Leishmania transmission.
The authors would like to thank Dr J. Flegr for providing control human sera, Dr V. Volfova for excellent dissection of sand fly salivary glands, Professor J. Vavra for comments on this manuscript, and Dr H. Kulikova and Mr D. Eremias for helpful technical assistance. The work was supported by the Ministry of Education of the Czech Republic (1131-00004 and FRVS 2807/2003) and by GACR 206/03/0325.