The SL1/PpSP15 group of proteins is similar to the SL1 salivary protein from L. longipalpis, which has no known function 13, and to PpSP15, a 15- kDa salivary protein from Phlebotomus papatasi that was previously shown to confer protection against L. major infection 14. The predicted molecular weight of these transcripts is approximately 14 kDa and is in agreement with the observed MW found through the proteome analysis (Figure 1). This group represents the most abundant transcripts in the salivary gland cDNA library of P. argentipes (Table 1). The fact that only one PpSP15 member was found in L. longipalpis, suggests that a number of lineage-specific gene expansions (gene duplication events) occurred in the Phlebotomus lineage at various periods in the evolution of these sandflies.

The PpSP15 family of proteins has only been found in species of sandflies suggesting that this family was a specific invention that occurred during sandfly evolution, most likely during their adaptation to a blood-feeding environment. Although PSI-BLAST analysis using PpSP15 proteins retrieved only members of the PpSP15 family, the PHYRE prediction servers indicated that members of the PpSP15 family possess an EF-hand fold most closely related to members of the odorant-binding protein (OBP) family to which the D7-proteins belong. It is thus likely that PpSP15 members were derived from an OBP ancestral protein. Characteristically, the OBP family in Drosophila has a low degree of sequence similarity among its members with only six conserved cysteines among the thirty-four members of this family 15.

The multiple pair-wise alignment analysis of sandfly PpSP15 shows a high degree of divergence among the sequences, 7.5% identity and 23.10% similarity (Figure 2A). The number of amino acids between the second and third cysteine (three) and between the fifth and the sixth cysteine (eight) were shown to be conserved among all the Drosophila OBP members 15. All of the sandflies sequences analysed from the PpSP15 protein family contained identical cysteine positioning. This data suggests that this family of proteins may be closely related to the short form of D7 because of the similarities to OBP and its small size of 15 kDa, which is similar to the MW found in the mosquitoes short D7 16.

Phylogenetic analysis of PpSP15 transcripts from the different sandflies including P. papatasi resulted in the formation of 5 distinct clades (Figure 2B). As such, three major clusters of orthologous groups of proteins (COGs), and hence gene duplication events, can be identified that possibly occurred in the ancestor to the Phlebotomus lineage. COG1 includes members from P. perniciosus, P. argentipes and P. papatasi with a second related clade that includes a lineage-specific expansion (LSE) in P. papatasi. COG2 includes members from P. perniciosus and P. argentipes. COG3 includes members from P. perniciosus, P. argentipes and P. ariasi. Another clade composed solely of members of P. argentipes indicates another LSE. Evolutionarily, the P. papatasi group was basal to the other Phlebotomus members analysed in this study, followed by P. argentipes, with P. perniciosus and P. ariasi forming the terminal clade. Given this, the topology of the cladogram obtained for the PpSP15 family suggests that COG2 resulted from gene duplication event that occurred in the ancestor to P. perniciosus and P. ariasi. COG1 follows the expected phylogenetic grouping but suggests that this gene was lost in P. ariasi or we failed to detect the ortholog in our library. COG3 again suggests that this specific gene duplication event occurred after divergence from the shared ancestor with P. papatasi. The PpSP15 family found in sandflies is thus characterised by both gene duplication and possibly gene loss events, both restricting an accurate reconstruction of its phylogeny. It is currently impossible to say which proteins share a conserved function with the PpSP15 from L. longipalpis as the major clades create a polytomy. As such, this family might bind related or similar pharmacologic components so that they all have, in fact, a similar function. This might explain the seemingly haphazard acquisition and loss of genes.

D7-related proteins are found in the saliva of different diptera, including Anopheles 17, Aedes 18 and Culex 19 mosquitoes as well as in the sandflies P. papatasi 19, L. longipalpis 13 and P. ariasi 21 and does not appear to occur in non-dipteran species. Two forms of the protein have been described; a long and short form 19; and appear to be distantly related to an odorant binding super family of proteins. Interestingly, the OBP family seems to be the ancestral molecule of the PpSP15 family (see above). Therefore, it may be possible that both D7 and PpSP15 related proteins have a common ancestor.

The D7 protein named hamadarin from Anopheles stephensi acts as an anticoagulant affecting the plasma contact system by inhibiting the activation of Factor XII and kallikrein 23. Recently, a biological function of four short members of the D7 family from A. gambiae and a long D7 from Aedes aegypti was described 16. These salivary proteins were shown to bind biogenic amines such as serotonin, histamine and norepinephrine. This function is relevant for blood-feeding because of the inhibition of the vasoconstrictor, platelet aggregating, and pain inducing properties of these biogenic amines 16. The exact function of D7 proteins in sandflies is largely unknown, but it may be related to the function observed in mosquito D7 proteins, either as an anticoagulant or binding biogenic amines.

The D7 family is represented in the P. argentipes cDNA library by two members, PagSP10 and PagSP25, and in the P. perniciosus cDNA library by three members, PpeSP04, PpeSP04B and PpeSP10. Only one member of this family is present in L. longipalpis sandfly, suggesting a case of gene duplication of this protein that probably occurred more recently in the Phlebotomus genus.

Comparative analysis of the D7 family of proteins from different sandflies reveals few conserved regions of identity, with only 16% identity and 23% similarity between the sandfly D7 proteins. There are 10 conserved cysteines throughout the molecule. The size of the sandfly D7 proteins is slightly smaller than the long D7 forms found in mosquitoes. Additionally, sandfly D7 is missing the last cysteine that is present at the carboxy terminal region of the mosquito long- and short-form D7 and, instead, have a cysteine present between conserved cysteines 8 and 19 19. Based on PHYRE prediction results, the long-form D7 proteins have 16 alpha helix domains, the short form have 8 alpha helix domains, while the sandfly form have 13 domains. All three forms are associated to OBP, as mentioned above. Interestingly, the sandfly D7 proteins are predicted to contain a beta-strand domain starting at the 7^th cysteine and characterised by a repeat of tyrosines at amino acid 188 (Figure 3A), replacing the alpha helix domain found in the mosquito D7 proteins at the same position. Due to these differences we are categorising the sandfly D7 as the medium form (Figure 3B).

Phylogenetic analysis of D7 proteins from different organisms shows 2 distinct clades with the sandfly proteins branching from the long form (Figure 3B). All of the sandfly D7 members are clustered within one clade, distinct from the long form members. The clade containing the mosquito D7 short-form proteins contains two clusters, one containing the Anopheles mosquitoes and the other containing the Culex and Aedes species. The sandfly clade subdivides into two distinct clades. The lower clade shows 5 distinct groups, two of these groups represent COGs of P. perniciosus and P. ariasi, and the group of P. papatasi seems to be a case of lineage expansion. The lower clade shows a COG that includes P. argentipes, P. ariasi and P. perniciosus proteins, PagSP25, ParSP07 and PpeSP10.

Both the P. perniciosus and P. argentipes libraries contained transcripts homologous to the Cimex family of apyrases 24, a protein also present in the saliva of P. papatasi 7 and L. longipalpis 13. Apyrases are enzymes that function as potent anti-platelet factors by destroying or hydrolysing the platelet activator ADP. An orthologue was found in humans and the recombinant protein was shown to hydrolyse a variety of nucleoside di- and triphosphates, preferentially UDP, followed by GDP, UTP, GTP, ADP, and ATP 2526.

Sequence alignment of the P. argentipes,P. perniciosus and P. ariasi apyrases show a 47% identity and 81% similarity at the amino acid level (Figure 4A). When L. longipalpis is included in the analysis, there is a considerable decrease in the identity (29%) as well as in the similarity (67%) (data not shown).

Phylogenetic analysis of apyrases from different organisms indicates three main clades with the sandfly apyrases in a distinct clade, apart from vertebrates, yet closely related to other insects (Figure 4B). Interestingly sandfly apyrases share a common ancestor with Cimex lectularius apyrase. The two insects appear to have evolved to the blood feeding mode independently 27. Within the clade containing the sandflies, one of the P. perniciosus apyrases, PpeSP01, is more closely related to the P. ariasi apyrase (ParSP01) than the second apyrase from P. perniciosus (PpSP01B). This may be the result of a gene duplication event in P. perniciosus and subsequent loss in P. ariasi. When searching databases we found a transcript from A. gambiae similar to sandfly apyrases. This is interesting because the known mosquito apyrases belong to the 5'-nucleotidase family of proteins. The known mosquito apyrase is very distinct, in size and sequence, from the Cimex family of apyrases also present in sandflies 28247, thus it is possible that mosquitoes in addition to a functional 5'-nucleotidase type apyrase, may have a non-functional sandfly/bedbug-like apyrase gene or it may have a house keeping function such as hydrolysing UDP formed after transglycosylation reactions in the Golgi 29. Alternatively, the mosquito may have a similar apyrase but with different substrate specificity or this protein is not present in their salivary gland.

Interestingly, the mosquito apyrase gene seems to be ancestral to the sandfly apyrase based on phylogenetic association (Figure 4B). Then it may be possible that mosquitoes have lost the function of this gene and kept the active form of the 5'-nucleotidase gene, which is the active apyrase in these insects.

The crystal structure of the Cimex family of apyrases was elucidated from the human counterpart and the amino acids relevant for calcium- and nucleotide-binding sites were determined 30. Several differences were noted among the amino acids relevant for nucleotide- or calcium-binding between sandflies, bedbugs and humans 30. Figure 5 shows the alignment of the different apyrases with amino acids relevant for calcium- and nucleotide-binding highlighted. We observed clear differences in some of these amino acids when sandflies were compared with other organisms including mosquitoes. The amino acids at position 124 are Ser (S), Thr (T) or Ala (A), in sandflies and Met (M) or Leu (L) in other organisms (Figure 5). Amino acids at position 126 is Met (M), Ile (I) or Leu (L) in sandflies and Lys (K) in other organisms; at position 129, sandflies have either Lys (K), Tyr (Y) or Leu (L) and other organisms have only Thr (T). At position 178, sandflies and bedbugs have a Trp (W) while other organisms have Ile (I). Amino acid substitutions may produce the specificity of the sandfly apyrases to ADP, a molecule that these insects must hydrolyse to overcome the hemostatic system and take a successful blood meal. In fact, Dai et al 30 showed that amino acid substitutions at some positions changed the substrate (GDP to ADP) in human apyrase. Human apyrases have more affinity for GDP substrate while sandfly apyrases have affinity for ADP 30.

The gene coding for the yellow protein was first described in Drosophila melanogaster 31. The proteins in this family appear to be derived from a common ancestor of the major royal jelly proteins (MRJPs) from honeybees and the yellow protein from Drosophila spp.

Drosophila yellow protein is related to pigmentation and male sexual behavior. In the family Culicidae a yellow protein was identified in Ae. aegypti whole-larvae extract and was associated to a dopachrome converting enzyme activity found in this insect 32. The function of this protein in the saliva of sandflies and its importance for blood feeding remains to be elucidated. The yellow protein family is one of the most abundant proteins found in the sandfly saliva.

We identified transcripts coding for secreted proteins of approximately 45 kDa, previously described in the saliva of L. longipalpis, P. papatasi and P. ariasi as yellow-related proteins 131421. In the P. argentipes cDNA library we found only one cluster (PagSP04) coding for this protein, yet in P. papatasi and L. longipalpis salivary glands there are multiple members of this family of proteins 33. Proteomic analysis (Figure 1A) revealed that the PagSP04 transcript (YHVEREYAWRNVTFEGVN) was one of the most abundant proteins found in the salivary glands of P. argentipes. Interestingly, three proteins with different mobilities coded for the same N-terminus sequence (Figure 1A) suggesting they may represent the same protein with different post-translational modifications.

Based on comparative analysis, we identified ten different sandfly salivary proteins that are members of the yellow family. Alignment of yellow proteins from Phlebotomus sandflies (P. argentipes, P. perniciosus, and P. ariasi), revealed a 43% identity and 79% similarity among the members of this protein family (Figure 6A). When the three yellow proteins from L. longipalpis were added, the identity was 21% identity and similarity was 57% among these proteins (Figure 6B). The phylogenetic analysis based on maximum likelihood using amino acid data of several MRJP/yellow proteins resulted in the formation of various clades (Figure 7), one clade containing yellow proteins from honeybees, a second clade containing mosquitoes and Drosophila, and a third clade containing the yellow proteins from sandflies. The sandfly clade was subdivided into three sub-clades, one containing the two yellow proteins from P. papatasi, the second clade containing the yellow proteins from P. ariasi and P. pernicious, and the third clade containing the L. longipalpis yellow proteins.

Based on their MW, the yellow proteins from L. longipalpis appear to be the most recognised proteins from the sera of individuals living in endemic areas of visceral leishmaniasis and from individuals that have anti-Leishmania immunity 34. The antibody response against these salivary proteins appears to correlate with protection against leishmaniasis.

This cluster codes for a secreted protein of 29 kDa similar to antigen 5-related protein found in wasp venom 35. Similar proteins have been isolated from the salivary glands of Aedes aegypti 20, An. gambiae 36 and from the salivary glands of L. longipalpis 13. We found only one cluster coding for this protein in the cDNA library of P. argentipes (PagSP05) and in the cDNA library of P. perniciosus (PpeSP07). The N-terminal sequence corresponding to these transcripts was identified in the salivary glands of P. argentipes (Figure 1A) and P. perniciosus (Figure 1B).

This family of proteins belong to the CAP family (CRISP, Ag5,PR-1) of proteins 3735. A remarkable feature of this family is the large number of cysteine residues, particularly at the carboxy-terminal region. X-ray structure of Na-AS-2, a member of this family from the human hookworm Necator americanus, was recently reported 38 and showed structural similarities to chemokines. Thus, it is possible that this type of protein in sandflies or other insects may bind cytokines with potential effects on the host immune response.

Multiple alignments of the antigen-5 protein from the sandflies indicated a 49% identity and 80% similarity (Figure 8A) with fourteen conserved cysteines. Phylogenetic analysis identified unique clades containing Hymenoptera, Culicidae, sandflies and mammals (Figure 8B). The only other organism included in the sandfly clade was the biting midge Culicoides sonorensis.

The 33-kDa protein family does not appear to be related to any other known proteins found in GenBank. Phlebotomus ariasi and P. perniciosus, 33-kDa proteins, are more closely related to each other than to P. argentipes, whereas the 33-kDa protein from L. longipalpis is distant to the three Phlebotomus species (data not shown). In general, all four sequences are somewhat similar with only 34% shared amino acids (Figure 9). Without further investigation, the function of this protein is unknown.

Transcripts coding for an endonuclease-like protein were found in P. argentipes (PagSP11) and P. perniciosus (PpeSP32) salivary gland cDNA libraries. These transcripts code for a protein of approximately 40 kDa with similarities to non-specific endonucleases from the sandfly L. longipalpis 33, tse tse fly Glossina morsitans 39, and the mosquito Culex pipiens quinquefasciatus 40. This transcript does not have a direct match with the NUC Smart motif, which is indicative DNA/RNA non-specific endonucleases and phosphodiesterases, but it does have a high homology with other known endonucleases from other arthropods such as D. melanogaster (AAL13973) as well as non-insect arthropods such as P. camtschaticus (king crab) (AAN86143) and M. japonicus (prawn) (ACB55635) (Figure 10B).

Endonuclease proteins were found in all sandflies studied. The multiple alignment of sandfly endonuclease showed various regions of identity among amino acids, and many regions of conserved amino acids, even when comparing endonucleases from different sandfly genera (Figure 10A). Phylogenetic analysis indicated that the sandfly endonucleases clustered (92% bootstrap support) with other arthropods, including two non-insect arthropods (Paralithode camtschaticus and Marsupenaeus japonicus) (Figure 10B), however, sandfly endonucleases formed a distinct clade within the arthropod cluster (90% bootstrap support). Additionally, these endonucleases were clearly distant from other endonucleases. This suggests that this endonuclease may represent a common antigen for different sandflies and that the distant relationship to other endonucleases may avoid potential cross reactivity with non-insect organisms. Since this type of enzyme can cleave double- and single-stranded DNA, the role of this protein in the saliva of sandflies should be investigated further.

The PpSP32-like family of proteins is similar to the 32.4-kDa protein first identified in P. papatasi salivary glands 14. We found only one cluster (PagSP06) coding for this protein in the P. argentipes cDNA library. BLAST analysis of PagSP06 identified significant homology to L. longipalpis and P. papatasi PpSP32-like proteins. The P. perniciosus cDNA library contained only one cluster (PpeSP05) sharing identity with the PpSP32-like protein. Interestingly, upon BLAST analysis the PpSP32-like protein from P. perniciosus was found to be highly homologous to a Type VII collagen protein from Canis familiaris (E = 10^-5) as well as a collagen from Mus musculus, Rattus norvegicus, Chinese hamster, and Bos taurus. This homology was found along approximately 74 amino acids (36% identity) and was dominated by conserved glycines and prolines (Figure 11).

The role of a collagen-like protein in sandfly salivary glands is unknown, yet intriguing. Because this protein may bind matrix protein, it is possible that this type of protein may form strong associations with basal matrix proteins. Within the Phlebotomus genera the relationship to collagen appears to be limited to PpSP32 kDa-like proteins from P. pernicious. P. papatasi and P. ariasi share only weak (non-significant) homology with Type VII collagen (E = 2.8 and 4.5, respectively) and P. argentipes did not match any collagen proteins upon multiple BLAST searches.

One explanation for the apparent homology between the four sandflies studied, yet lack of identity between collagen-related proteins in all sandflies, was revealed in the comparative analysis of the four sandflies. The majority of the homology between the four sandflies was found at the N- and carboxy terminus whereas the middle section of the protein appeared to be less conserved. The N-terminal and carboxy terminus has 43% and 23% identity between the four flies, respectively, whereas the midsection of the protein contains only 4% identity (Figure 11A). The region of homology between P. perniciosus and Type VII collagen was found in the non-homologous region of the protein (Figure 11B).

Additionally, BLAST analysis of L. longipalpis LJL04 (AAS16906) revealed a significant homology to collagen adhesion proteins from B. thuringiensis (ZP_00739782) and B. cereus (NP_830673) (Figure 11C). Again, the region of homology was found in the divergent section of the PpSP32-kDa protein. The most perplexing aspect of this family of proteins is that, although highly conserved among the four sandflies studied here, the N-terminal and carboxy regions of the protein have no homology to proteins of known function.