The transcriptome of the salivary glands of the female western black-legged tick Ixodes pacificus (Acari: Ixodidae)

Reagents

All water used was of 18 MΩ quality and was produced using a MilliQ apparatus (Millipore, Bedford, MA, USA). Organic compounds were obtained from Sigma (St. Louis, MO, USA) or as stated otherwise.

Group 1: Basic-tail proteins (BTP)

This family of proteins is highly represented in the salivary glands of both I. pacificus and I. scapularis ticks. Fig. 2A shows the alignments of the BTP of these ticks where a highly conserved signal peptide indicates their common origin from an ancestral gene. Fig. 2A also shows that the pattern of these sequences contain six cysteines (XnCX14CX3CX18CX9CX4CXn) followed by a basic tail with high content of lysines (Lys, K). On the other hand, some proteins from this family, such as gi 22652868 and gi 22164158, display a negatively charged tail composed of six glutamic acid (Glu, E) anionic residues (Fig. 2A). The evolutionary relationships of BTP were inferred by constructing the phylogenetic tree using the NJ algorithm; a cladogram is shown in Fig. 2B. Of interest, these proteins share sequence similarities to exogenous anticoagulants such as SALP 14 (gi 15428308) from I. scapularis. SALP 14 is a FXa inhibitor that appears to interact with the catalytic domain of FXa and with the so-called exosite (Narasimhan et al., 2002). Exosites—regions far from the catalytic site and known to determine specificity and affinity of blood coagulation factors toward substrates—also are critical for the assembly of the prothrombinase, a multimolecular complex that leads to thrombin generation (Krishnaswamy, 2005). Targeting these domains appears to be an effective strategy evolved by blood-feeding arthropods to effectively impair blood coagulation. In fact, we recently reported that ixolaris, a FX(a) scaffold-dependent inhibitor of Factor VIIa/tissue factor complex, specifically recognizes the FXa heparin-binding exosite (Monteiro et al., 2004).

The fact that BTP and SALP 14 contain a poly-Lys tail adds an additional layer of anticoagulation, as it directs the inhibitor to negatively-charged membranes (e.g., activated platelets) critical for productive blood coagulation complex assembly (Broze, 1995). As a result, the effective concentration of the inhibitor is increased at sites that are predominantly pro-coagulant. Also, we speculate that FXa—which is usually protected from physiologic inhibitors (e.g., TFPI, heparin/ATIII) when the prothrombinase is fully assembled (Mast and Broze, 1996; Rezaie, 2001)—would be more susceptible to these bifunctional molecules. Demonstration that proteins rich in positively charged residues effectively block the coagulation cascade comes from studies performed with a recombinant Rhodnius prolixus salivary lipocalin (nitrophorin-7, NP-7). NP-7 contains a cluster of positively charged residues in the N-terminus and specifically binds to anionic phospholipids, preventing thrombin formation by the prothrombinase (Andersen et al., 2004). Finally, a bifunctional fusion protein containing Kunitz and annexin domains was shown recently to inhibit the initiation of blood coagulation (Chen et al., 2005).

Group 2: Similar to Group 1, but without the basic tail

These sequences contain a cysteine pattern identical to Group 1 peptides except that, remarkably, the poly K tail is missing. Many other amino acids also are not conserved. Sequence alignment between the Group 1 peptides (containing poly K and poly E) and the peptides similar to Group 1 is shown in Fig. 3A. Fig. 3B shows that these proteins come from a common ancestor that appears to have evolved to display different functions. The function of the peptides of Group 2 deserves further investigation.

Group 3: Kunitz-containing proteins

Kunitz domains are about 60 residues and contain 6 specifically spaced cysteines (XnCX8CX15CX7CX12CX3CXn) that form disulfide bonds typically represented by bovine pancreatic trypsin inhibitor (BPTI). In most cases, they are reversible inhibitors of serine proteases that bind the active site (Laskowski and Kato, 1980); however, Kunitz inhibitors such as the dendrotoxins from Dendroaspis angusticeps snake venom block K+ channel but display negligible protease inhibitory properties (Harvey, 2001). Kunitz-containing proteins also interact with protease exosites (Monteiro et al., 2004) or platelets (Mans et al., 2002). Of note, sequencing the I. pacificus cDNA library yields a number of proteins containing Kunitz-like domains.

The alignment of BPTI, snake venom, and I. scapularis and I. pacificus single Kunitz-like proteins is shown in Fig. 4A. Some I. pacificus proteins contain one-Kunitz-like domain, here named the Monolaris-1 family (or “similar to 6.5- to 8.4-kDa proteins from I. scapularis”). These molecules display the following cysteine pattern: XnCX8CX18CX5CX12CX3CXn. Other single-Kunitz sequences present in I. scapularis belong to the Monolaris-2 family (or “similar to 7.9- to 8.7-kDa proteins from I. scapularis”) and display the sequence pattern XnCX8CX15CX8CX11CX3CXn. We could not, however, find members of the Monolaris-2 family sequences in our I. pacificus cDNA library. Fig. 4A also shows that the well-known tick anticoagulant peptide from the soft tick Ornithodoros moubata (Waxman et al., 1990) has Kunitz-like folding with the sequence pattern XnCX9CX17CX5CX15CX3CXn. At present, the functions of Monolaris-1 and -2 are unknown, but they may target specific proteases. The phylogenetic tree shown in Fig. 4B suggests that snake venom peptides containing Kunitz domains (non-neurotoxic or neurotoxic) and the tick families of Monolaris and basic tail peptides have diverged into two different main groups from a commom ancestor, suggesting that these proteins have evolved to perform different functions.

Additionally, cDNAs were sequenced coding for proteins containing two- or five-Kunitz domains. These proteins share sequence similarity to ixolaris (Francischetti et al., 2002b) and penthalaris (Francischetti et al., 2004a), two I. scapularis TFPI salivary proteins that prevent initiation of blood coagulation through specific inhibition of the Factor VIIa/tissue factor complex. It is possible that these proteins block other proteases (Ruf, 2004) or affect angiogenesis (Hembrough et al., 2004).

Fig. 4C depicts the predicted secondary folding of I. scapularis and I. pacificus Kunitz-like-containing proteins based on the crystal structure determined for BPTI (Huber et al., 1974).

Group 4: Proline-rich proteins

Group 3 cDNA sequences code for short peptides of mature molecular mass ranging from 3.5–4.8 kDa of both basic and acidic nature (Table 1). Alignments and cladograms (presented in Fig. 5A and 5B, respectively), show that all sequences are relatively glycine and proline rich in both I. pacificus and I. scapularis salivary glands. Some sequences display weak matches to proteins annotated as collagen in the NR database; these possess two conserved cysteine residues in the mature peptide and remarkable conservation of the secretory signal peptide (Fig. 5). Most amino acids of the predicted signal secretory peptide are conserved, versus few on the mature peptide, suggesting functional diversity. The possible function of these peptides remains to be characterized, but taking into account its similarity to collagen, it may somehow affect vascular biology through inhibition of cell-cell, cell-matrix, or cell-ligand interactions. These peptides may also function as adhesive molecules to cement the tick into their host’s skin.

Group 5: Similar to I. scapularis 18.7-Kda protein

This group of proteins (table 1) is similar to orthologs described in I. scapularis and code for an acidic putative protein of unknown function. Only low e values have been found when compared with proteins in the NR database including coagulation factor X (gi 9837158, e value 0.069), venom metalloprotease acurhagin precursor (gi 4689408; e value 3.8), and proprotein convertase subtilin (gi 51771463, e value 8.4). Accordingly, this family of proteins may have evolved from a protease precursor; however, any functional assignment will be possible only after testing the recombinant protein in screening assays.

Groups 6 and 7: Similar to I. scapularis 5-kDa protein and 9.4-kDa protein

Groups 6 and 7 code for basic proteins of ~ 5 kDa and 9.4 kDa that also are present in I. scapularis. No protein motif was identified for either protein; accordingly, the function of these proteins is not evident.

Group 8: Metalloprotease

These enzymes are capable of hydrolyzing various components of the extracellular matrix including fibrinogen and fibronectin and reportedly affect endothelial cells, leading to apoptosis. These enzymes are organized into four classes, PI through PIV, according to size and domain composition (Bjarnason and Fox, 1995).

Our library contains a truncated cDNA that codes for a mature metalloprotease similar to one described in I. scapularis (gi 31322779) (Francischetti et al., 2003) and I. ricinus (gi 5911708). The alignment of the mature metalloproteases from I. pacificus, I. scapularis, and I. ricinus, where the zinc-binding motif HExxHxxGxxH common to these enzymes was identified, is shown in Fig. 6A. Fig. 6B compares the PIII class of metalloproteases from snake venom and the I. pacificus, I. scapularis and I. ricinus enzymes. It is clear that enzymes from both genera have pre-, pro-, metalloprotease, disintegrin-like, and cysteine-rich-like domains; however, the Ixodidae disintegrin-like and cysteine-rich like domains are significantly shorter in the number of amino acid residues when compared with the corresponding domains of metalloproteases from the reprolysin family (Bjarnasson and Fox, 1995). We suggest that this pattern of cysteines confer a different specificity for these enzymes. This family of proteins also appears to account for the α-fibrinogenase and fibrinolytic activity recently reported for I. scapularis saliva (Francischetti et al., 2003). Degradation of fibrinogen and fibrin are associated with inhibition of platelet aggregation and clot formation. Metalloproteases also may interact with endothelial cell integrins, leading to apoptosis and inhibition of angiogenesis (Francischetti et al, 2005).

Group 9: GPIIb/IIIa antagonists from the short neurotoxin family

Inhibitors of platelet aggregation that targets the fibrinogen receptor (GPIIbIIIa, integrin αIIbβ3) have been described in the hard tick Dermacentor variabilis (variabilin) and the soft ticks, Ornithodoros moubata (disagregin) and O. savignyi (savignygrin) (Karczewski et al. 1994; Wang et al. 1996; Mans et al. 2002a). Savignygrin belongs to the Kunitz-BPTI family and presents the integrin RGD-recognition motif on the substrate binding loop of the Kunitz fold (Mans et al. 2002a). In contrast, variabilin possesses an RGD-motif in its C-terminal region that is not flanked by cysteines (Wang et al. 1996). A search for possible GPIIb/IIIa antagonists with RGD-motifs and flanking cysteines, termed the Ixodegrins, identified one candidate in I. pacificus and several homologs in I. scapularis (Table 1). It is clear that the Ixodegrins are related to variabilin, but do possess flanking cysteines. Variabilin probably possesses a flanking disulphide motif too, but was missed due to the technical difficulties in identifying cysteines correctly during N-terminal sequencing. Database searches using SAM-T99 (Karplus et al. 1998), a program that utilizes hidden Markov models to find remote homologous sequences, identified dendroaspin as the highest hit. Dendroaspin, also known as mambin, is part of the short neurotoxin family found in elapid snakes (McDowell et al. 1992; Williams et al. 1992; Sutcliffe et al. 1994). Strikingly, the RGD-active site loop (loop3) is conserved between snake and tick integrin antagonists (Fig. 7A). This includes the flanking cysteines involved in a disulphide bond that constricts the RGD-loop conformation and the flanking prolines that was shown to be important for presentation of the RGD sequence (Lu et al. 2001). The tick inhibitors maintain loops 2 and 3 of the short neurotoxin fold, but do not possess the N-terminal loop 1 and the C-terminal extension (Fig. 7A). This makes them the shortest members of the short neurotoxin family described to date, with only 39 amino acids forming the core active fold. Phylogenetic analysis of the neurotoxin family indicates that dendroaspin and tick inhibitors group within one clade to the exclusion of the other short neurotoxins (Fig. 7B). This suggests either an extreme form of convergent evolution, where ticks and elapid snakes used the same protein fold to evolve the same function or raises the possibility that ticks or snakes acquired the ancestral protein via a horizontal gene transfer event or that there is a true evolutionary relationship between the ixodegrins and short neurotoxins. The fact that orthologs of the Ixodegrins are present in both Ixodes (prostriate) and Dermacentor (metastriate) ticks, suggests that this inhibitor was present in the last common ancestor of hard ticks. Snakes evolved most of their venom properties approximately 60-80 million years ago (Fry, 2005), whereas most hard tick genera diverged at least 110 million years ago or earlier (Klompen et al. 1996). If tick and snake proteins are related, then the ancestral gene may have a platelet antagonist function and the neurotoxic properties (and the rest of the short neurotoxin fold - loop1 and the C-terminal extension) evolved later. In contrast, soft ticks in the genus Ornithodoros evolved integrin antagonists from the BPTI-fold which suggests that hard and soft ticks evolved different strategies to obtain a blood meal (Mans et al. 2002b; Mans and Neitz, 2004). Accordingly, ixodegrin may affect platelet or neutrophil integrin function or neutrophil function.

Group 10: Ixostatin family, or short-coding cysteine-rich peptides (“thrombospondin”)

The two sequences in Group 11 match a sequence deposited in the NR database from I. scapularis; alignments are shown in Fig. 8A. These sequences have been annotated “thrombospondin” (gi 15428290), but thrombospondin motifs are lacking. On the contrary, these short coding region cysteine-rich peptides—here named ixostatins—are remarkably similar to the cysteine-rich domain of ADAMTS (Fig.8B). Of note, ADAMTS-4 (a disintegrin and metalloproteinase with thrombospondin motifs), also known as aggrecanase, are enzymes involved in cartilage cleavage (Flannery et al., 2002). The role of the cysteine-rich domain of ADAMTS proteases is unknown, but it is postulated to interact with integrins and/or other attachment motifs of cells and matrix proteins (Porter et al., 2005). Accordingly, the ixostatin family of peptides could be involved in disruption of platelet aggregation or neutrophil function, cell-matrix interactions, or inhibition of angiogenesis (Porter et al., 2005). The protein modules of ixostatin and of ADAMST-4 are compared in Fig 8C.

Group 11: Histamine-binding proteins (lipocalins)

Group 11 contains sequences with similarities to histamine-binding proteins discovered in the saliva of Rhipicephalus appendiculatus ticks (Paesen et al., 1999). The alignments of these sequences (Fig. 9A) reveal that they do not display a highly conserved signal peptide which suggest that they may not share a common ancestor. In addition, the mature proteins contain few consensus sequences indicating that they may have diverged to perform distinct functions (Fig. 9B). This contention is also supported by the cladogram presented in Fig. 9B. It is likely that these proteins function by binding small ligands such as histamine, serotonine, and adrenaline (Andersen et al., 2005). Fig. 9C shows a predicted 3-D model for sequence gi 51011604 that has an e value of -768 for HBP from R. appendiculatus. The figure shows amino acid side chains of the histamine-binding protein from R. appendiculatus (red) surrounding the bound histamine ligand with the corresponding residues for gi 51011604 shown in cyan. In the histamine-binding protein, the imidazole ring of the ligand is stabilized by surrounding aromatic residues, while in the I. scapularis protein the binding pocket remains hydrophobic and fewer aromatic residues are present, suggesting different ligand specificity. Polar residues (Tyr 36 and Glu 135) forming electrostatic interactions with the aliphatic amino group of histamine in the histamine-binding protein are conserved in gi 51011604 suggesting the possibility of a similar role in this protein.

Group 12: Neuropeptide-like (npl-31) protein with GGY repeat

A cDNA coding for a protein that shows remarkable sequence homology to a neuropeptide-like protein (npl-21) described in Caenorhabditis elegans (Nathoo et al., 2001). This family of peptides displays a potent antimicrobial activity toward Drechmeria coniospora, Neurospora crassa, and Aspergillus fumigatus (Couillault et al., 2004). Identification of these peptides in ticks reinforces the notion that saliva contains a cocktail of antimicrobial peptides. These peptides may prevent growth of yeast and bacteria that, per se, can elicit an inflammatory/immune response that may be detrimental to the feeding behavior of the attached ticks. Expression of these molecules is particularly important vis-à-vis the remarkably immunosupressive property of the saliva (Wikel, 1999) that helps ticks to feed for days but otherwise creates an appropriate environment for pathogen overgrowth. The sequence alignments for C. elegans npl-21 and I. pacificus npl-21-like proteins are presented in Fig. 10A and the cladogram in Fig. 10B. This is the first time that this family of antimicrobial peptides has been identified in the salivary gland of a blood-sucking arthropod.

Group 13: Oxidant metabolism

Proteins with similarity to glutathione peroxidase and a putative secreted superoxide dismutase were found (Table 1). These sequences categorize the prominent salivary gland proteins in I. pacificus and demonstrate the presence of a potent antioxidant in tick saliva. Of interest, cluster F12_IPL_P23 has sequence similarity to SALP 25, a protein that catalyzes the reduction of hydrogen peroxide in the presence of reduced glutathione and glutathione reductase (Das et al., 2001). The functions of these proteins are likely related to maintenance of the physiologic redox of cellular intracellular milieu or to modulation of the extracellular levels of pro-oxidants often associated with inflammatory events.

Group 14: Similar to other ixodid proteins

A number of sequences show sequence homology to proteins from Ixodidae described before. We have found sequences similar to SALP 15, a immunodominat protein in I. scapularis (Das et al. 2001), and to ISAC, the anti-complement from I. scapularis (Valenzuela et al., 2002). We also have found sequences similar to domain 8 of human ADAMS and Factor VII.

Group 15: Novel, unknown

Some sequences containing a signal peptide and a stop codon and with a clear open reading frame were without database hits and were characterized as unknown-function proteins. Assignment of function for these proteins will only be possible after expressing and screening for testable biologic activities.

Group 16: Housekeeping cDNA

Thirty-seven sequences with homology to housekeeping protein are given in Table 1. They assign to ribosomal, glutathione S-transferase, vacuolar aasorting proteins, cytochrome, ATP synthase subunit, and NADH-ubiquinone oxidoreductase, among other molecules. In addition, housekeeping proteins may be useful in phylogenetic studies (Black and Piesman, 1994).