Am. J. Trop. Med. Hyg., 70(3), 2004, pp. 286–293 Copyright © 2004 by The American Society of Tropical Medicine and Hygiene

ANTIGENIC DIVERSITY IN MAXADILAN, A SALIVARY PROTEIN FROM THE SAND FLY VECTOR OF AMERICAN VISCERAL LEISHMANIASIS RANIA S. MILLERON, JOHN-PAUL MUTEBI, SONIA VALLE, ALBERTO MONTOYA, HUAIZHI YIN, LYNN SOONG,* AND GREGORY C. LANZARO* Department of Pathology and World Health Organization Center for Tropical Diseases, University of Texas Medical Branch, Galveston, Texas; Departamentos de Entomologia y Parasitologia, Centro Nacional de Diagnostico y Referencia, Ministerio de Salud, Managua, Nicaragua

Abstract. The salivary protein maxadilan (MAX) is a vasodilator and immunomodulator from the sand fly vector of the protozoan parasite Leishmania chagasi. Vaccinating BALB/c mice with sand fly salivary gland extracts or with MAX protects the host against L. major infection. Because of the potential use of MAX in an anti-Leishmania vaccine, we characterized the vertebrate host IgG response to MAX in the present study. Our immunochemical analysis indicated that antibodies to MAX were detected in BALB/c mice, as well as in pigs and humans, from a area in Nicaragua endemic for Lutzomyia longipalpis. Previous studies demonstrate that the MAX protein is polymorphic on the amino acid level. Our findings suggested that naturally occurring MAX variants were recognized specifically by the host immune system and antigenicity appeared to be associated with amino-acid sequence variability. Thus, antigenic diversity of MAX and possibly of other arthropod salivary proteins may dictate the development of vector-based vaccines(s). might represent antigenic polymorphism. The goal of this study is to test that hypothesis. Since immunization with MAX elicits both a humoral and a cellular response in mice,13 our study addresses whether amino acid variants may elicit specific IgG responses from the host. The ability of mammals to make antibodies against arthropod saliva is well established in studies with laboratory animals.15–17 Even humans living with endemic populations of blood-sucking arthropods have serum antibodies specific against saliva of ticks, mosquitoes, triatomes18–20 and Lu. longipalpis.21 In the present study, we detected a host antibody response to sand fly-injected MAX using immunochemistry. Binding assays revealed antigenic specificity to the C terminal end of variant MAX proteins, suggesting the latter are recognized specifically by the host immune system.

INTRODUCTION Members of the Lutzomyia longipalpis species complex1,2 are weak fliers that inhabit foci in semi-arid regions from southern Mexico to northern Argentina. This group of sand flies is the principal vector of the protozoan parasite Leishmania chagasi. The catholic feeding preferences of female Lu. longipalpis include mammals and birds. The mouthparts of sand flies are short, in comparison to those of mosquitoes, obliging them to lacerate dermal blood vessels and to feed from resultant pools of blood. However, damaged vessels rapidly initiate hemostatic processes, including vasoconstriction, coagulation, and platelet aggregation.3 The saliva that Lu. longipalpis spits into the host, while obtaining blood, contains a number of anti-hemostatic proteins. Among these is a small vasodilator protein, approximately 7 kD in size, known as maxadilan (MAX).4 A pair of salivary glands contains approximately 10 ng of MAX protein, which is equivalent to 1–2% of the total protein in saliva.5 The challenge posed by developing an effective antiLeishmania vaccine has led to innovative research on the ability of sand fly saliva from various species to enhance infectivity of Leishmania.6 For example, the saliva from Lu. longipalpis plays a role in establishment and exacerbation of L. major lesions.7 Furthermore MAX, identified only in Lu. longipalpis, is proving to be the most important immunomodulator in the saliva of this sand fly species complex.8–12 Both whole saliva and purified synthetic MAX equivalently increase the size of cutaneous lesions caused by L. major.13 Thus, Leishmania infection may be controlled by eliciting a host immune response against sand fly salivary proteins. Although we have eluded to MAX thus far as one protein, Lanzaro and others described high levels of amino acid polymorphism in the mature MAX protein among Lu. longipalpis sibling species, after comparing eight DNA sequences from four laboratory fly colony populations originating from Brazil, Colombia, and Costa Rica.14 Some variants differed by as much as 23% in amino acid composition, with a mean ± SD variability of 15.7 ± 4.2%. They hypothesize that variability

MATERIALS AND METHODS Study populations. Collections of pig and human sera from areas in Nicaragaua endemic for Lu. longipalpis were obtained and were stored at −20°C until analysis. Pig sera came from the communities of Apumpua and Ojochal in the departments of Chontales and Leon, respectively (n ⳱ 9). With the owner’s permission, a veterinarian obtained 2–4 mL of whole pig blood. Negative control pig serum samples were obtained from healthy pigs (n ⳱ 10) housed at Texas A & M University Veterinary School (College Station, TX). Human sera (n ⳱ 40), originating from the departments of Leon and Nueva Segovia were analyzed. These patients were diagnosed with visceral leishmaniasis (n ⳱ 3) or atypical cutaneous leishmaniasis (n ⳱ 37) at the Diagnostic and Reference Laboratory, Ministry of Health (Managua, Nicaragua) by an indirect immunofluorescent antibody test. These samples were then sent to Dr. Eva Harris (University of California, Berkeley, CA), who donated part of the remaining sera to the University of Texas Medical Branch (Galveston, TX) for the present study. Negative control human sera consisted of three lots of serum from ICN (Costa Mesa, CA) and Sigma (St. Louis, MO) biologic supply companies and four samples from individuals in Galveston, Texas with no previous exposure history to sand flies. An additional 31 individual human serum samples from a serosurvey for Trypanosoma cruzi in

* These authors contributed equally to this paper.

286

ANTIGENIC DIVERSITY IN MAXADILAN

Minas Gerais, Brazil (8 non-infected and 23 infected) were generously provided by Dr. Barbara Doughty (University of Texas Medical Branch). Institutional review board approval was obtained for the human sera, which were obtained without personal identifiers. Sensitization of mice by sand fly bites. BALB/c mice (eight weeks old) obtained from Harlan Sprague Dawley (Indianapolis, IN) were used and maintained in compliance with National Institutes of Health guidelines. Mice were exposed to Lu. longipalpis females by a modified protocol.17 Briefly, prior to exposing them to sand fly feeding, mice were sedated with an intraperitoneal injection of 60 mg/kg of ketamine and 5 mg/kg of xylazine. Each mouse was exposed to at least 50 female sand flies twice a week, followed by a two-week interval for a total of greater than or equal to four exposures. Serum was collected two weeks after the last exposure, and antibody titers were measured by an indirect enzyme-linked immunosorbent assay (ELISA). Collection of sand flies and preparation of salivary gland lysate (SGL). Lutzomyia longipalpis was collected via mouth aspirators from domestic animals and Centers for Disease Control (CDC, Atlanta, GA) light traps. In Nicaragua, the flies were collected in the communities of 1) Apumpua in the Department of Chontales and 2) Ojochal in the Department of Leon. Specimens were collected from the town of Baturite´ in the state of Ceará, Brazil. Field-collected sand flies were identified by species using standard morphologic characteristics.22 Lutzomyia longipalpis from Baturite´ were frozen in liquid nitrogen and were stored at −80°C. Field-collected female flies from Nicaragua were caged and provided a blood meal from hamsters. Gravid females were individually isolated in small containers and were transported to the laboratory after oviposition. The salivary glands of female progeny were used as an antigen in an indirect ELISA. Briefly, SGL was prepared from 2–8-day-old, laboratory-bred female flies, based on a previous protocol.5 The concentration of SGL protein was quantified by a Micro BCA (bicinchoninic acid) protein assay (Pierce Biotechnology, Rockford, IL) using bovine serum albumin as a standard. Cloning and sequencing of the MAX gene. Genomic DNA was isolated from individually dried Lu. longipalpis collected from Baturite´ , Brazil, Bucaramanga, Colombia and Liberia, Costa Rica14 by a previously described protocol.23 The MAX gene was amplified by a polymerase chain reaction (PCR) from DNA.24 The PCR products from individual flies were

287

cloned using the TOPO TA Cloning system (Invitrogen, Carlsbad, CA) and INV␣F’ competent Escherichia coli. Several white colonies were selected and plasmid DNA from individual clones was purified. The MAX gene was amplified from plasmid DNA using a BigDye™ terminator cycle sequencing ready reaction (Applied Biosystems, Foster City, CA), cleaned on a Micro Bio-Spin P-30 Tris chromatography column (Bio-Rad Laboratories, Hercules, CA), and sequenced on an ABI-377 automated sequencer (Applied Biosystems) according to the manufacturer’s instructions. DNA sequences were analyzed using DNASTAR Windows version 4.03 (DNASTAR, Inc., Madison, WI) for alignment and translation of DNA into protein sequence. Generation of synthetic MAX peptides and rabbit antisera. Seven peptides based on MAX sequence data from individual, field-collected Lu. longipalpis (Figure 1) were synthesized by SynPep Corp. (Dublin, CA). A Jameson-Wolf plot25 analysis of MAX predicted the relatively high antigenicity of the C terminal end. Six of the peptides were based on hypervariable 19 amino-acid residues from the C terminal end 4361 (vMAX), specifically, four vMAX peptides from Baturite´ , Brazil and one each from Liberia, Costa Rica and Bucaramanga, Colombia. One peptide, cMAX, was synthesized based on amino acid residues 23-40. The peptides were purified by high-performance liquid chromatography. As assessed by mass spectroscopy, the purity achieved was BZba1 ⳱ 93.6%, BZba9 ⳱ 79.8%, BZba15 ⳱ 88.3%, BZba17 ⳱ 80.8%, Liberia ⳱ 76.4%, BU12 ⳱ 84.2%, and cMAX ⳱ 61.7%. Proteins were cross-linked to a carrier protein, keyhole limpet hemocyanin, using maleimidobenzoic acid-Nhydroxysuccinimide ester. Female, age-matched New Zealand White rabbits were immunized subcutaneously five times with 500 ␮g of each peptide at two-week intervals by SynPep Corp. Freund’s complete adjuvant was used for the initial immunization, and Freund’s incomplete adjuvant was used for the remaining booster immunizations. Serum was collected 10 days after the fourth and the fifth booster immunizations. Rabbit anti-peptide titers were measured, and sera with comparable titers (1:3,200) were chosen for further study. Indirect ELISA to measure anti-MAX, anti-SGL, and antipeptide IgG. In general, antigen was coated (50 ␮L/well) on Immulon 4 microtiter plates (Dynex Technologies, Inc., Chantilly, VA). The ELISA was carried out using standard techniques26 with horseradish perioxidase–conjugated goat

FIGURE 1. Amino acid sequences for mature maxadilan (MAX) protein. Variable regions of the protein are highlighted in gray. Residues differing from the consensus sequence are in bold. Abbreviations of the MAX variants are as follows: BZba ⳱ Baturite´ , Brazil; Liberia ⳱ Liberia, Costa Rica; BU ⳱ Bucaramanga, Colombia; SMAX ⳱ synthetic MAX. Variant synthetic peptides (vMAX) were made from the 19 amino acid residues within the dotted line box. The conserved peptide (cMAX) was made from the consensus sequence of 18 amino acid residues within the solid line box.

288

MILLERON AND OTHERS

anti-mouse IgG, goat anti-rabbit IgG (Bio-Rad Laboratories), goat anti-human IgG, rabbit anti-pig IgG (Sigma), and ABTS (2,2’-azino-bis [3-ethylbenzothiazoline-6-sulfonic acid] diamonium salt) substrate (Pierce). Plates were read at 405 nm in a VERSAmax™ microplate reader (Molecular Devices, Sunnyvale, CA). Relative titers of antibodies were measured in triplicate by determining the dilution factor at which 50% maximal binding occurred. The specific details of individual experiments are given below. To measure levels of anti-SGL or synthetic MAX protein (SMAX), SGL equivalent to two pairs of salivary glands (⯝ 2.8 ␮g/mL) or SMAX (5 ␮g/mL) were used as coating antigens. SMAX was a generous gift of Dr. Richard Titus (Colorado State University, Fort Collins, CO) and was based on the published sequence of MAX from Lapinha Caves, Brazil.14 To establish a cut-off titer for each antibody assay, the IgG concentration of antibodies to saliva and antibodies to MAX was determined based upon pigs and humans living in an area where Lu. longipalpis are not found. To account for the possible cross-reactivity between IgG against another arthropod salivary protein and MAX, sera were analyzed from an area in Brazil endemic for triatomine bugs. The SGL coating antigen was derived from Lu. longipalpis from the same state. Negative control sera used in this ELISA originated from the United States. The optical density (OD) values (1:100) of control sera samples (mean + 3 SD) were used as the cut-off value for a positive response. The data obtained were verified by a minimum of three independent repeats. To determine the binding capacity of rabbit antibodies to vMAX peptides (5 ␮g/mL) or to the SMAX protein (5 ␮g/ mL), we compared binding of sera from immunized rabbits (n ⳱ 6) to that of pre-immune rabbit sera (n ⳱ 6). We then tested the binding capacity of six human and six pig sera samples to the vMAX peptides using six unexposed humans and pigs as controls. Data obtained were verified by a minmum of three independent repeats. To address whether the vMAX portion elicits a stronger antibody reponse than the cMAX portion of MAX, we determined the binding capacity of sera from six Texan and six Nicaraguan pigs to the coating antigens cMAX (5 ␮g/mL) and vMAX (5␮g of BZba17/mL). The OD readings from the Texas pigs were subtracted from the readings obtained for the Nicaraguan pigs. Data obtained were verified by two independent repeats. Western blot to measure anti-MAX IgG. Two pairs of salivary glands, or 3 ␮g of SMAX, were separated on a 4–20% gel by sodium dodecylsulfate–polyacrylamide gel electrophoresis (Bio-Rad Laboratories) under reducing conditions and transferred to a polyvinylidene fluoride membrane (Bio-Rad Laboratories). After blocking non-specific sites, membrane strips were incubated overnight at 4°C with pig sera from Texas or Nicaragua. Subsequently, strips were incubated in horseradish peroxidase–conjugated rabbit anti-pig IgG for one hour at room temperature. Blots were then washed and developed with ECL™ substrate (Amersham Biosciences Corp., Piscataway, NJ). Statistical analysis. An unpaired, two-sample t-test (twotailed) was used to compare mean OD levels. Pearson’s correlation coefficient (R2) described associations between the OD levels achieved with binding to SMAX or SGL. A general linear model (GLM) gave statistical significance to the differences in OD units when more than two variables were com-

pared. Exploratory data analyses were conducted to test for departure from normality, linearity, and independence, assumptions of the GLM, a model that also tests for analysis of variance (ANOVA). These descriptive analyses indicated no need to transform variables to meet assumptions. The ANOVA identified those variables responsible for observed differences in mean binding capacity between antisera and vMAX. A GLM was fitted in which the binding activity (OD level absorbance at 450 nm) was the dependent variable and independent variables (dilution factor, antibody × antigen interaction term, experimental replicate, population of origin of MAX peptide) were evaluated for their capacity to influence OD levels. RESULTS Detection of specific IgG antibodies to MAX and SGL. The IgG response to MAX delivered by sand fly bites was characterized in sera collected sera from laboratory and domestic animals exposed to sand flies. Antibodies to SMAX were detected in sera of mice (4 of 4) sensitized by bites of Lu. longipalpis, and in humans (8 of 40) and pigs (6 of 9) living in areas of Nicaragua endemic for Lu. longipalpis (Figure 2). A positive correlation was found between levels of SMAX and SGL IgG (Figure 3A). Western blot analysis also confirmed detection of specific IgG to SGL and SMAX in Nicaraguan pig sera (Figure 3B). The location of the 7-kD band (Figure 3B, lane 1) corresponds with the mature form of MAX. The faint bands at approximately 131, 100, 65, and 33 kD are other proteins found in the salivary gland (Figure 3B, lane 2). When pig serum from the United States was used to probe the blots, the sera did not bind to SMAX or to the proteins in SGL. None of the people at risk of or infected with T. cruzi had measurable IgG levels against sand fly SGL or against SMAX by the indirect ELISA using serum from the United States to determine the cutoff point of the ELISA. Variability in the MAX protein. Lanzaro and others reported high levels of amino acid variation in the MAX protein among four laboratory colonies of Lu. longipalpis.14 To test the hypothesis that this variability might be a mechanism to avoid host immune response to sand fly feeding, we chose four MAX variants recovered from a single population of Lu. longipalpis from Baturite´ , Brazil and compared these with

FIGURE 2. Levels of serum IgG against maxadilan (MAX) in animals exposed to feeding Lutzomyia longipalpis. Reactivity of sera from control or exposed BALB/c mice, pigs, or humans to MAX (synthetic MAX ⳱ 5 ␮g/mL) was determined by an indirect enzymelinked immunosorbent assay. Control and exposed pig and human sera were from Texas and Nicaragua, respectively. Results are displayed as symbols (•) representing individual animal serum (dilution ⳱ 1:100) in optical density (OD) units at 405 nm. Horizontal lines represent the mean OD values of each control group + 3 SD. Samples above the lines are considered positive.

ANTIGENIC DIVERSITY IN MAXADILAN

FIGURE 3. A, Serum IgG reactivity to both synthetic maxadilan (SMAX) and salivary gland lysate (SGL). Antibody levels were measured using an indirect enzyme-linked immunosorbent assay (ELISA). Plates were simultaneously coated with SMAX (5 ␮g/mL) or with SGL (10 pairs/plate). OD ⳱ optical density. B, Western blots of serum reactivity of Nicaraguan pig IgG. Sodium dodecyl sulfate– polyacrylamide gel electrophoresis under reducing conditions was used to separate SMAX (3 ␮g/mL) (lane 1) and Lutzomyia longipalpis SGL (2 salivary gland pairs) (lane 2). Proteins were electrotransferred to a polyvinylidene fluoride membrane and probed with serum reactive to both MAX and SGL by indirect ELISA, followed by anti-pig IgG conjugated to horseradish peroxidase. Blots were developed using the ECL™ reagent. Values on the right side of the blot are molecular weight (MW) standards (161-0324; Bio-Rad Laboratories) run in an adjacent lane and electrotransferred. The location of the 7-kDa band corresponds with the mature form of MAX. In lane 2, the faint bands at approximately 131, 100, 65, and 33 kDa are other proteins found in the salivary gland. kDa ⳱ kilodaltons.

MAX sequences from Lu. longipalpis originating in Liberia, Costa Rica and Bucaramanga, Colombia (Figure 1). Variable sites in the mature, 61-amino acid protein are found predominantly in residues 9-22 and 43-61, with relatively little variability in other regions of the protein. Specificity of antibodies against the C terminal variable region of MAX. To examine whether variant MAX sequences generate a specific IgG response, peptide sequences were chosen to assess whether a positive correlation exists between amino acid sequence divergence and antigenic specificity. Sequences were chosen from within a single population from Baturite´ , Brazil, as well as from populations representing two other Lu. longipalpis sibling species from Bucaramanga, Colombia and Liberia, Costa Rica (Figure 1). These six peptides based on 19 amino acid residues differ from 5.2% to 36.8% in amino acid composition within the Baturite´ population and from 15.7% to 42.1% between populations of Lu. longipalpis sibling species from Brazil, Colombia, and Costa Rica. The sequences varied by approximately as much within populations as between populations. Two interesting findings were obtained from anti-vMAX/

289

vMAX binding experiments (Figure 4). First, the panel of anti-peptide sera for six vMAX peptides bound to the peptide- coating antigens with specificity (F ⳱ 10.5–24.5, degrees of freedom ⳱ 5, P ⳱ 0.0005, by ANOVA). The six peptide sequences were distinguished as though they were two different sequences. Peptides BZba17 and BZba1 bound in a nearly identical fashion with the panel of antisera (Figure 4A and B). In addition, peptides Liberia, BZba15, BU12, and BZba9 reacted with the panel of antisera indistinguishably (Figure 4C–F). Second, for the antigen/antibody combinations tested, an increase in amino acid differences produced a decrease in binding affinity. A difference in peptide residues ⱖ 26.3% caused the panel of anti-peptide sera to show statistically significant differences in binding. The statistical significance of these observations was substantiated with a GLM analysis. Those independent factors having a significant influence on antibody/peptide binding were 1) the degree to which the serum was diluted (P ⳱ 0.0005) and 2) the antibody × antigen interaction term (P ⳱ 0.0005). These two factors produce 75% of the variability in mean OD levels shown in Figure 4. The term antigen × antibody suggests that there is a synergy between antigen and antibody that produces changes in the OD level. Based on the GLM, differences among independent experiments (P ⳱ 0.975) or between populations (P ⳱ 0.223) did not contribute significantly to changes in the OD level. The population was defined as either those sequences from Baturite´ , Brazil or Bucaramanga, Colombia and Liberia, Costa Rica. The data suggest there is antigenic variability between populations as well as within a single population. If pressure from the host immune system produces variability in MAX, antigenic variability must be found within a single population. Next, we determined that antibodies to vMAX specifically recognize full-length SMAX protein (Figure 5). Antibody to BZba15 showed the highest binding capacity to SMAX and

FIGURE 4. Rabbit immune sera reactivity to different maxadilan (MAX) peptides. A plate was coated with one of six variant synthetic (vMAX) peptides (5 ␮g/mL) shown in Figure 1 and rabbit immune serum was serially diluted 1:400–1:25,600 to measure anti-peptide IgG binding by an indirect enzyme-linked immunosorbent assay. Bars represent the mean reactivity (optical density [OD] at 405 nm) across the range of serial dilutions for three independent experiments. Vertical lines represent the SE. Statistically significant differences in the reactivity between the panel of antisera and each peptide were found by analysis of variance (P ⳱ 0.0005).

290

MILLERON AND OTHERS

FIGURE 5. Recognition of synthetic maxadilan (SMAX) by anti-peptide rabbit sera. Sera were collected from six rabbits before immunization or after the fourth booster immunization with one of the following synthetic peptides: BZba17, BZba15, BZba1, BZba9, BU12, or Liberia conjugated to keyhole limpet hemocyanin. The sera/peptide reactivity was individually tested by an indirect enzyme-linked immunosorbent assay using plates that were precoated with SMAX (5 ␮g/mL) and rabbit immune sera that was serially diluted as indicated. The data for BZba17 and BZba1 overlap. OD ⳱ optical density. For definitions of the synthetic peptides, see Figure 1.

antibody to BZba1/BZba17 showed the lowest binding capacity. This was an expected result since BZba15 was identical in amino acid sequence to the 19 residues at the C-terminal end of SMAX and BZba1/BZba17 peptide differed by greater than 26.3%. Antisera to the four remaining peptides bound with intermediate reactivity to SMAX, and their sequence divergence ranged from 10.5 to 15.7%. Recognition of different MAX variants by Nicaraguan human and pig sera. Since a single population of sand flies is comprised of numerous MAX variants, we predict that animals in the field should display differential serologic activity against different MAX variants. We tested our hypothesis by measuring levels of IgG antibody to vMAX in humans and pigs. The sera from individual Nicaraguan patients with atypical cutaneous leishmaniasis recognized BZba17 and BZba1 as one peptide and Liberia, BZba15, BU12 and BZba9 as another (Figure 6). The human sera from the United States did not show specific binding to any of these peptides. These results followed the same trend as those from rabbits immunized with these peptides. The differential serologic response was more pronounced in humans than in pigs. The sera from Nicaraguan pigs, with greater exposure to Lu. longipalpis, responded to the peptides with an OD greater than 3 SD above the level given by sera from pigs from the United States. However, the Nicaraguan pig serologic response showed less difference between peptides (Figure 7). Pig NP34 responded less to peptides BZba17 and BZba1 compared with the other peptides. Individual serum responded differently to the peptides, depending on the degree of exposure of these animals to Lu. longipalpis, and therefore to MAX variants. No significant differences in the concentration of either total protein or IgG between the Nicaraguan and U.S. pig or human sera were found. Therefore, the differences are likely due to the presence or absence of specific IgG antibody to MAX. A GLM analysis again statistically confirmed those factors responsible for the strength of binding between antisera and peptide in these experiments. The two factors explaining binding of the sera to vMAX peptides were 1) country of origin, i.e., Nicaragua or the United States (pig and human

sera; P ⳱ 0.0001), and 2) differences in peptide residues (pig sera; P ⳱ 0.035 and human sera; P ⳱ 0.011). The GLM explained 79% and 58% of the variability in anti-peptide binding capacity for pig and human sera, respectively. Binding capacity of Nicaraguan pig serum to the variable and conserved regions of MAX. The MAX amino acid residues 23–40 (cMAX) are conserved relative to residues 43–61 (vMAX). We sought to determine whether antigenicity is associated with variability. However, cMAX was difficult to produce and purify from other slightly degenerate, synthetic forms. Despite the degeneracy of cMAX, sera from pigs bitten by sand flies in Nicaragua were found to have a greater specificity to residues 43–61 compared with residues 23–-40 in an ELISA based on serum antibody capture (Figure 8). Texas pig serum did not bind with specificity to cMAX or to vMAX.

FIGURE 6. Specific recognition of variable synthetic peptides of maxadilan by human sera. The indirect enzyme-linked immunosorbent assay (ELISA) conditions are identical to those in Figure 4. Bars represent the mean number of ELISA units (optical density [OD] at 405 nm) generated by sera of individuals from Nicaragua or the United States (dilutions ⳱ 1:50–1:3,200). Vertical lines represent the SE. NH ⳱ serum of individuals from Nicaragua; USHu ⳱ sera from six individuals from the United States. For definitions of the synthetic peptides, see Figure 1.

ANTIGENIC DIVERSITY IN MAXADILAN

291

FIGURE 7. Specific recognition of variable synthetic peptides of maxadilan by pig sera. The indirect enzyme-linked immunosorbent assay (ELISA) conditions are identical to those in Figure 4. Bars represent the mean number of ELISA units (optical density [OD] at 405 nm) generated by sera of pigs from Nicaragua or the United States (dilutions ⳱ 1:100–1:400). Vertical lines represent the SE. NP ⳱ antiserum of an individual pig from Nicaragua; Tpig ⳱ antisera of six pigs from Texas. For definitions of the synthetic peptides, see Figure 1.

DISCUSSION The results presented in this report add a new dimension to our understanding of the interaction between blood-feeding insects and their vertebrate hosts. We provide the first evidence of antigenic polymorphism in a sand fly. Our results suggest that 1) MAX is immunogenic when delivered to animals via sand fly bite, 2) the MAX protein is variable within a single sand fly population, 3) vMAX peptides produce specific antibody responses in a natural population of hosts, and 4) the variable region (vMAX) appears to be more antigenic than the conserved region (cMAX). Extensive amino acid polymorphism in MAX from differ-

FIGURE 8. Antigenic responses elicited by a conserved peptide of maxadilan (cMAX) and a variant synthetic peptide of maxadilan (vMAX). The indirect enzyme-linked immunosorbent assay (ELISA) conditions are identical to those in Figure 4. One plate was coated with vMAX (BZba17) and another plate was coated with a cMAX whose sequence is shown in Figure 1. Bars represent the mean number of ELISA units (optical density [OD] at 405 nm) generated by sera from six Nicaraguan pigs that tested positive for anti-peptide IgG (dilution ⳱ 1:50). Vertical lines represent the SE. Mean ELISA units generated by binding of anti-serum from six Texas pigs to cMAX and to vMAX were subtracted from the Nicaraguan pig data.

ent sand fly populations, representing three sibling species from throughout Central and South America, was previously reported.14 In the present study, we found equally high levels of divergence among MAX variants from individuals collected from a single population in Baturite´ , Brazil. The variability of MAX is not explained largely by speciation because the Baturite´ population was a single gene pool with random mating among individuals.27 Polymorphism in MAX appears to be higher than the average degree of protein polymorphism in Lu. longipalpis. From analyses of enzyme-encoding loci, other Lu. longipalpis (Baturite´ Brazil) proteins do not have an unusual level of genetic polymorphism (37.5% polymorphic loci and 0.111 mean heterozygosity27) compared with values found in insects in general (38.0% polymorphic loci and 0.107 mean heterozygosity28). Therefore, the salient question is why is MAX protein sequence so diverse? Maintenance of the large number of MAX variants observed in the Baturite´ population cannot be explained by genetic drift alone. If positive diversifying selection has operated on a gene, the number of non-synonymous substitutions causing adaptive amino acid changes is greater than the number of synonymous substitutions.29 Our finding that nucleotide substitutions in the MAX gene described here are primarily non-synonymous suggests that MAX variability may be due to selection rather than genetic drift. We are beginning to shed light on the nature of the selection pressure that produces polymorphism in MAX. Lanzaro and others14 demonstrated that a series of MAX variants all had equivalent vasodilatory potency. Therefore, polymorphism in MAX appears to be constrained by the necessity to conserve vasodilatory function.14 These results prompted us to find that MAX variants are specifically recognized by the host immune system. Although both human and pig sera recognized vMAX, individual human serum samples showed a more dramatic differential response to MAX than did individual pig serum. These results are probably due to the finding that Lu. longipalpis seem to preferentially bite pigs compared with other vertebrate hosts in Costa Rica30 and Colombia.31 For pigs bitten by a substan-

292

MILLERON AND OTHERS

tial number of flies every night over time, we predict that a given animal may have been exposed to a higher degree to all six of the peptide variants in our study. Humans have less exposure to sand flies compared with pigs or other domestic animals because Lu. longipalpis are not particularly anthropophilic.31 We also present evidence that epitopes in vMAX elicit a greater antigenic response than do epitopes in cMAX. It can be inferred from previous work32,33 that the variable regions of MAX are involved in binding to IgG because these regions are essential for MAX binding to the pituitary adenylate cyclase activation polypeptide type 1 (PAC1) receptor. In contrast, the conserved middle region of MAX appears essential for MAX vasodilatory activity, but not for binding to the PAC1 receptor. It appears that either residues 24–42 are conformationally unexposed to the environment, and thus are not involved in IgG receptor ligand binding, or the physical properties of these residues make them less antigenic. Regardless of the reason, our studies suggest that amino acid sequence variability in MAX is associated with antigenicity. The MAX protein induces vasodilation in the skin of a mammal that has been exposed to multitudes of bites from a population of sand flies over time (Milleron RS, unpublished data). This situation may occur because the host immune system may not neutralize the vasodilatory function of MAX when a population of sand flies injects numerous variants. In light of this observation, we hypothesize that the sand fly may have evolved diversity in MAX as a strategy to evade the host immune response. Although a potentially large B cell repertoire can be produced in any given vertebrate host, selective recognition of specific epitopes occurs due to the immunogen itself or to host factors involved in its selection.34 Studies of strains of human immunodeficiency virus type 1 and other animal lentiviruses show that these viruses have evolved immunodominant epitopes capable of undergoing antigenic polymorphism. Because of their antigenic diversity, these pathogens are able to limit or fix the humoral and cellmediated responses to the initial resident pathogen; thus, the host seems unable to eliminate or control the agent that may be a closely related genetic variant.35 Whether the theory of deceptive imprinting or antigenic competition describes pressures encountered by MAX deserves further study. The possibility that the host immune system exerts selective pressure on a salivary protein from a hematophagous insect has never been previously investigated or described. The data in the present study lend support to this hypothesis. We are completing studies to address whether selection is acting on MAX by examining blood feeding behavior and fecundity of flies feeding on mice immunized with MAX. Our findings have implications for the general epidemiologic application of sand fly saliva. Antigenic diversity of MAX sequences is significant in light of the potential uses of salivary proteins in a diagnostic ELISA to test for sand fly or Leishmania exposure21 and a recombinant subunit vaccine.36,37 The observation that this antigen displays diversity could constrain these applications. Previously, a number of anti-parasite vaccines have failed based, in large part, on the high degree of amino acid diversity of a candidate vaccine antigen. Vaccinating mice against MAX elicits a protective Th1-type response against Leishmania, as well as a strong anti-MAX IgG response.13 However, the protective abilities of one MAX variant against exposure to another variant re-

main to be determined. Designing a successful vaccine may mean including all of the different immunogenic forms of MAX to give blanket protection. Alternatively, novel vaccination regimens will have to be developed to randomly reproduce the antigenic diversity displayed by these sand flies. Received May 14, 2002. Accepted for publication March 3, 2003. Acknowledgments: We thank Drs. Daniel Freeman and Frederic Tripet for discussing the statistical applications used in the manuscript, Dr. Dia Elnaiem for thoughtful comments, Dr. Richard Titus for his gift of synthetic MAX, Dr. Eva Harris for the human sera from Nicaragua, and Drs. Jose´ Ribeiro and Barbara Doughty for their critical reading of the manuscript and supply of essential reagents. Financial support: This work was funded by National Institutes of Health (grants T32 AI-075261 and AI-39540). Authors’ addresses: Rania S. Milleron, John-Paul Mutebi, Huaizhi Yin, and Lynn Soong, Department of Pathology and World Health Organization Center for Tropical Diseases, University of Texas Medical Branch, Galveston, TX 77555-0609. Sonia Valle and Alberto Montoya, Departamentos de Entomologia y Parasitologia, Centro Nacional de Diagnostico y Referencia, Ministerio de Salud, Managua, Nicaragua. Gregory C. Lanzaro, Department of Entomology, University of California, Davis, CA 95616-8579, Telephone: 530-7525833, Fax: 530-752-1537, E-mail: [email protected].

REFERENCES 1. Lanzaro GC, Ostrovska K, Herrero MV, Lawyer PG, Warburg A, 1993. Lutzomyia longipalpis is a species complex: genetic divergence and interspecific hybrid sterility among three populations. Am J Trop Med Hyg 48: 839–847. 2. Ward R, Phillips A, Burnet B, 1986. Genetic isolating mechanisms between different forms of the sandfly Lutzomyia longipalpis (Diptera: Psychodidae). Ann Ist Super Sanita 22: 69– 72. 3. Ribeiro JM, 1987. Role of saliva in blood-feeding by arthropods. Annu Rev Entomol 32: 463–478. 4. Lerner EA, Shoemaker CB, 1992. Maxadilan. Cloning and functional expression of the gene encoding this potent vasodilator peptide. J Biol Chem 267: 1062–1066. 5. Lerner EA, Ribeiro JM, Nelson RJ, Lerner MR, 1991. Isolation of maxadilan, a potent vasodilatory peptide from the salivary glands of the sand fly Lutzomyia longipalpis. J Biol Chem 266: 11234–11236. 6. Kamhawi S, 2000. The biological and immunomodulatory properties of sand fly saliva and its role in the establishment of Leishmania infections. Microbes Infect 2: 1765–1773. 7. Titus RG, Ribeiro JM, 1988. Salivary gland lysates from the sand fly Lutzomyia longipalpis enhance Leishmania infectivity. Science 239: 1306–1308. 8. Theodos CM, Titus RG, 1993. Salivary gland material from the sand fly Lutzomyia longipalpis has an inhibitory effect on macrophage function in vitro. Parasite Immunol 15: 481–487. 9. Qureshi AA, Asahina A, Ohnuma M, Tajima M, Granstein RD, Lerner EA, 1996. Immunomodulatory properties of maxadilan, the vasodilator peptide from sand fly salivary gland extracts. Am J Trop Med Hyg 54: 665–671. 10. Soares MB, Titus RG, Shoemaker CB, David JR, Bozza M, 1998. The vasoactive peptide maxadilan from sand fly saliva inhibits TNF- alpha and induces IL-6 by mouse macrophages through interaction with the pituitary adenylate cyclase-activating polypeptide (PACAP) receptor. J Immunol 160: 1811–1816. 11. Bozza M, Soares MB, Bozza PT, Satoskar AR, Diacovo TG, Brombacher F, Titus RG, Shoemaker CB, David JR, 1998. The PACAP-type I receptor agonist maxadilan from sand fly saliva protects mice against lethal endotoxemia by a mechanism partially dependent on IL-10. Eur J Immunol 28: 3120–3127. 12. Gillespie RD, Mbow ML, Titus RG, 2000. The immunomodulatory factors of bloodfeeding arthropod saliva. Parasite Immunol 22: 319–331.

ANTIGENIC DIVERSITY IN MAXADILAN

13. Morris RV, Shoemaker CB, David JR, Lanzaro GC, Titus RG, 2001. Sandfly maxadilan exacerbates infection with Leishmania major and vaccinating against it protects against L. major infection. J Immunol 167: 5226–5230. 14. Lanzaro GC, Lopes AH, Ribeiro JM, Shoemaker CB, Warburg A, Soares M, Titus RG, 1999. Variation in the salivary peptide, maxadilan, from species in the Lutzomyia longipalpis complex. Insect Mol Biol 8: 267–275. 15. Wikel S, 1982. Immune responses to arthropods and their products. Annu Rev Entomol 27: 21–48. 16. Wikel S, 1996. Host immunity to ticks. Annu Rev Entomol 41: 1–22. 17. Ghosh KN, Mukhopadhyay J, 1998. The effect of anti-sandfly saliva antibodies on Phlebotomus argentipes and Leishmania donovani. Int J Parasitol 28: 275–281. 18. Sanders ML, Glass GE, Scott AL, Schwartz BS, 1998. Kinetics and cross-species comparisons of host antibody responses to lone star ticks and American dog ticks (Acari: Ixodidae). J Med Entomol 35: 849–856. 19. Lane RS, Moss RB, Hsu YP, Wei T, Mesirow ML, Kuo MM, 1999. Anti-arthropod saliva antibodies among residents of a community at high risk for Lyme disease in California. Am J Trop Med Hyg 61: 850–859. 20. Peng Z, Simons FE, 1997. Cross-reactivity of skin and serum specific IgE responses and allergen analysis for three mosquito species with worldwide distribution. J Allergy Clin Immunol 100: 192–198. 21. Barral A, Honda E, Caldas A, Costa J, Vinhas V, Rowton ED, Valenzuela JG, Charlab R, Barral-Netto M, Ribeiro JM, 2000. Human immune response to sand fly salivary gland antigens: a useful epidemiological marker? Am J Trop Med Hyg 62: 740– 745. 22. Young DG, Duncan MA, 1994. Guide to the Identification and Geographic Distribution of Lutzomyia Sand Flies in Mexico, the West Indies, Central and South America (Diptera: Psychodidae). Gainesville, FL: Associated Publishers. 23. Post RJ, Flook PK, Millest AL, 1993. Methods for the preservation of insects for DNA studies. Biochem Syst Ecol 21: 85–92. 24. Warburg A, Saraiva E, Lanzaro GC, Titus RG, Neva F, 1994. Saliva of Lutzomyia longipalpis sibling species differs in its composition and capacity to enhance leishmaniasis. Philos Trans R Soc Lond B Biol Sci 345: 223–230. 25. Jameson BA, Wolf H, 1988. The antigenic index: a novel algorithm for predicting antigenic determinants. Comput Appl Biosci 4: 181–186.

293

26. Soong L, Duboise SM, Kima P, McMahon-Pratt D, 1995. Leishmania pifanoi amastigote antigens protect mice against cutaneous leishmaniasis. Infect Immun 63: 3559–3566. 27. Mutebi JP, Alexander B, Sherlock I, Wellington J, Souza AA, Shaw J, Rangel EF, Lanzaro GC, 1999. Breeding structure of the sand fly Lutzomyia longipalpis (Lutz & Neiva) in Brazil. Am J Trop Med Hyg 61: 149–157. 28. Nevo E, 1978. Genetic variation in natural population: patterns and theory. Theor Popul Biol 13: 121–177. 29. Endo T, Ikeo K, Gojobori T, 1996. Large-scale search for genes on which positive selection may operate. Mol Biol Evol 13: 685–690. 30. Zeledon R, Murillo J, Gutierrez H, 1984. Ecology of Lutzomyia longipalpis (Lutz & Neiva, 1912) and possibilities of the existence of visceral leishmaniasis in Costa Rica. Mem Inst Oswaldo Cruz 79: 455–459. 31. Morrison AC, Ferro C, Tesh RB, 1993. Host preferences of the sand fly Lutzomyia longipalpis at an endemic focus of American visceral leishmaniasis in Colombia. Am J Trop Med Hyg 49: 68–75. 32. Moro O, Wakita K, Ohnuma M, Denda S, Lerner EA, Tajima M, 1999. Functional characterization of structural alterations in the sequence of the vasodilatory peptide maxadilan yields a pituitary adenylate cyclase-activating peptide type 1 receptorspecific antagonist. J Biol Chem 274: 23103–23110. 33. Tatsuno I, Uchida D, Tanaka T, Saeki N, Hirai A, Saito Y, Moro O, Tajima M, 2001. Maxadilan specifically interacts with PAC1 receptor, which is a dominant form of PACAP/VIP family receptors in cultured rat cortical neurons. Brain Res 889: 138– 148. 34. Nara PL, 1999. Deceptive imprinting: insights into mechanisms of immune evasion and vaccine development. Adv Vet Res 41: 115–134. 35. Nara PL, Garrity R, 1998. Deceptive imprinting: a cosmopolitan strategy for complicating vaccination. Vaccine 16: 1780–1787. 36. Kamhawi S, Belkaid Y, Modi G, Rowton E, Sacks D, 2000. Protection against cutaneous leishmaniasis resulting from bites of uninfected sand flies. Science 290: 1351–1354. 37. Valenzuela JG, Belkaid Y, Garfield MK, Mendez S, Kamhawi S, Rowton ED, Sacks DL, Ribeiro JM, 2001. Toward a defined anti-Leishmania vaccine targeting vector antigens: characterization of a protective salivary protein. J Exp Med 194: 331– 342.