A total of 944 clones were sequenced and used to assemble a database [see additional file 1] that yielded 245 clusters of related sequences, 171 of which contained only one expressed sequence tag (EST). The consensus sequence of each cluster is named either a contig (deriving from two or more sequences) or a singleton (deriving from a single sequence). For simplicity sake, this paper uses 'cluster' to denote sequences deriving both from consensus sequences and from singletons. The 245 clusters were compared using the program BlastX, BlastN, or RPSBLAST 33 to the nonredundant protein database of the National Center of Biological Information (NCBI), to a gene ontology database 34, to the conserved domains database of the NCBI 35, and to a customprepared subset of the NCBI nucleotide database containing either mitochondrial or rRNA sequences.

Because the libraries used are unidirectional, the three-frame translations of the dataset were also derived, and open reading frames (ORF) starting with a methionine and longer than 40 amino acid (AA) residues were submitted to SignalP server 36 to help identify putative secreted proteins. The EST assembly, BLAST, and signal peptide results were loaded into an Excel spreadsheet for manual annotation.

Four categories of expressed genes derived from the manual annotation of the contigs (Table 1). The putatively secreted (S) category contained 34% of the clusters and 75% of the sequences, with an average number of 8.4 sequences per cluster. The housekeeping (H) category had 43% and 17% of the clusters and sequences, respectively, and an average of 1.5 sequences per cluster. Twenty two percent of the clusters, containing 8.4% of all sequences, were classified as unknown (U), because no functional assignment could be made. Similar to the Hgroup, this category also had an average of 1.5 sequences per cluster. A good proportion of these transcripts could derive from 3^/ or 5^/ untranslated regions of genes of the above two categories, as was recently indicated for a sialotranscriptome of An.gambiae 29. A possible transposable element originated 21 singletons representing either active transposition or, more likely, expression of transposable element regulatory transcripts in X.cheopis.

The 106 clusters (comprising 157 EST) attributed to Hgenes expressed in the salivary glands of X. cheopis were further characterized into 13 subgroups according to function (Table 2). Not surprisingly for an organ specialized in secreting polypeptides, the two larger sets were associated with protein synthesis machinery (82 EST in 44 clusters) and energy metabolism (15 clusters containing 19 EST), a pattern also observed in other sialotranscriptomes 373839. We have also included in the H category a group of 12 EST that grouped into eight clusters, representing conserved proteins of unknown function, presumably associated with cellular metabolism. The fourth most abundant group, attributed to the protein modification function, is of interest due to the specialized function of the salivary glands. This group, comprising five clusters, includes a chaperone protein and enzymes that are associated with disulfide bridge formation. Several transporters were also identified, including those coding for two subunits of the VATPase complex, which has been shown to be necessary for mosquito salivation 4041. Transcripts coding for VATPases are a common finding in mosquito sialomes 3739424344. The complete list of all 106 gene clusters, along with further information about each, is given in additional file 2.

Inspection of additional file 1 indicates the expression of several expanded gene families, including a family encoding proteins producing a match to KOG3720, indicative of lysosomal and prostatic acid phosphatases. This family alone is responsible for 35 clusters comprised of 268 sequences, or over 25% of the salivary library. Mucins are also represented, as well as an expansion of a peptide family unique to fleas that is related to the cat flea antigen annotated as FSH precursor (gi¦1575479). Several other novel peptide families, described in greater detail below, were also observed.

To obtain information on protein expression in the salivary glands of X. cheopis, salivary gland homogenates (SGH) of approximately 100 pairs of glands were separated by SDS-PAGE. The stained protein bands were excised from the gel and submitted to tryptic digestion followed by MS/MS identification. Note that the electropherogram is dominated by two broad bands in the molecular mass range of 40–43 kDa (Figure 1). Only five slices of the gel shown in Figure 1 yielded useful information by being assigned to predicted sequences from our clusterized database and to fulllength sequences disclosed in this work. These results are summarized in Figure 1. To explore the expression of lower molecular mass polypeptides that were not represented in the SDS-PAGE experiment, we filtered 51 μg of homogenized salivary glands through a 30-kDa cutoff filter and submitted this filtrate to direct LC-MS/MS analysis. In addition, an aliquot of this filtrate was digested with trypsin prior to LC-MS/MS analysis. The low molecular weight (MW) filtrate yielded 22 additional matches with secreted proteins, as indicated in the supplemental Tables S1 and S2. The description of the identified proteins are embedded in the following manuscript sections.

Several clusters of sequences coding for housekeeping and putative secreted polypeptides indicated in additional file 1 are abundant and complete enough to extract consensus sequences of novel sequences. Additionally, we have performed primer extension studies in several clones to obtain full-or near full-length sequences of products of interest. A total of 76 novel sequences, 48 of which code for putative secreted proteins, are grouped together in additional file 2.

Following is a detailed description of the full-length transcripts found in the salivary glands of adult X. cheopis:

Intact salivary gland pairs were collected from adult female X. cheopis fleas. Individual fleas (anesthetized by chilling on ice) were dissected in 10 μl of PBS on a glass microscope slide on the stage of a dissecting stereomicroscope. By grasping the dorsal half of the flea above the forelegs with forceps, pressing down on the abdomen just posterior to the midgut with a bent dissecting needle, and pulling, the two pairs of salivary glands and the midgut would usually remain attached to the head and be pulled free of the rest of the body. The common lateral salivary ducts were cut to release each pair, which were then hooked with a dissecting pin and placed in PBS at 4°C. Pools of 40 pairs of glands in 20 μl PBS were frozen at 70°C. Glands used for apyrase assays were dissected and stored in 10 mM TrisHCl and 150 mM NaCl, pH 7.4 rather than PBS.

X. cheopis salivary gland mRNA was isolated from 200 salivary gland pairs from adult fleas using the Micro-FastTrack mRNA isolation kit (Invitrogen, SanDiego, CA). The PCR-based cDNA library was made following the instructions for the SMART cDNA library construction kit (Clontech, Palo Alto, CA). This system utilizes oligoribonucleotide (SMART IV) to attach an identical sequence at the 5' end of each reverse-transcribed cDNA strand. This sequence is then utilized in subsequent PCR reactions and restriction digests.

First strand synthesis was carried out using PowerScript reverse transcriptase at 42°C for 1 hr in the presence of the SMART IV and CDS III (3') primers. Second strand synthesis was performed by a long distance (LD) PCR-based protocol, using Advantage™ Taq Polymerase (Clontech) mix in the presence of the 5' PCR primer and the CDS III (3') primer. The cDNA synthesis procedure resulted in the creation of SfiI A &B restriction enzyme sites at the ends of the PCR products that are used for cloning into the phage vector. The PCR conditions were: 95°C for 20 sec; 24 cycles of 95°C for 5 sec., 68°C for 6 min. A small portion of the cDNA obtained by PCR was analysed on a 1.1% agarose gel to check for the quality and range of cDNA synthesised. Double stranded cDNA was immediately treated with proteinase K (0.8 μg/ml) at 45°C for 20 min and the enzyme was removed by ultrafiltration though a Microcon (Amicon) YM100 centrifugal filter device. The cleaned, doublestranded cDNA was then digested with SfiI at 50°C for 2 hrs, followed by size fractionation on a ChromaSpin-400 column (Clontech, Palo Alto, CA). The profile of the fractions was checked on a 1.1% agarose gel and fractions containing cDNAs of more than 400 bp were pooled and concentrated using a Microcon YM100.

The cDNA mixture was ligated into the λ TriplEx2 vector (Clontech, Palo Alto, CA) and the resulting ligation mixture was packaged using the GigaPack^® III Plus packaging extract (Stratagene, La Jolla, CA) according to the manufacturer's instructions. The packaged library was plated by infecting log phase XL1 Blue E. coli cells (Clontech, Palo Alto, CA). The percentage of recombinant clones was determined by performing a blue-white selection screening on LB/MgSO₄ plates containing X-gal/IPTG. Recombinants were also determined by PCR, using vector primers (5' λ TriplEx2 Sequencing Primer and 3' λ TriplEx2 Sequencing) flanking the inserted cDNA and visualising the products on a 1.1% agarose/EtBr gel.

The X. cheopis salivary gland cDNA library was plated on LB/MgSO₄ plates containing X-gal/IPTG, to an average of 250 plaques per 150 mm Petri plate. Recombinant (white) plaques were randomly selected and transferred to 96-well MICROTEST™ U Bottom plates (BD BioSciences, Franklin Lakes, NJ), containing 100 μl of SM buffer [0.1 M NaCl; 0.01 M MgSO_4; 7H₂ O; 0.035 M TrisHCl (pH 7.5); 0.01% gelatin] per well. The plates were covered and placed on a gyrating shaker for 30 min at room temperature. The phage suspension was either immediately used for PCR or stored at 4°C for future use.

To amplify the cDNA using a PCR reaction, four microliters of the phage sample was used as a template. The primers were sequences from the λ TriplEx2 vector and named pTEx2 5 seq (5'TCC GAG ATC TGG ACG AGC 3') and pTEx2 3 LD (5' ATA CGA CTC ACT ATA GGG CGA ATT GGC 3'), positioned at the 5' end and the 3' end of the cDNA insert, respectively. The reaction was carried out in 96 well flexible PCR plates (Fisher Scientific, Pittsburgh, PA) using the TaKaRa EX Taq polymerase (TAKARA Mirus Bio, Madison, WI), on a Perkin Elmer GeneAmp^® PCR system 9700 (Perkin Elmer Corp., Foster City, CA). The PCR conditions were: one hold of 95°C for 3 min; 25 cycles of 95°C for 1 min, 61°C for 30 sec; 72°C for 2 min. The amplified products were analysed on a 1.5% Agarose/EtBr gel. 1100 cDNA library clones were PCR amplified and the ones showing single band were selected for sequencing. Approximately 200–250 ng of each PCR product was transferred to Thermo-Fast 96-well PCR plates (ABgene Corp., Epsom, Surray, UK) and frozen at 20°C, before cycle sequencing using an ABI3730 XL machine.

Expressed sequence tags (EST) were trimmed of primer and vector sequences, clusterized, and compared with other databases as described 44. The BLAST tool 77, CAP3 assembler 78, ClustalW 79, and Treeview software 80 were used to compare, assemble, and align sequences and to visualise alignments. For functional annotation of the transcripts we used the tool BlastX 33 to compare the nucleotide sequences to the NR protein database of the National Center for Biotechnology Information (NCBI) and to the Gene Ontology (GO) database34. The tool RPSBlast 33 was used to search for conserved protein domains in the Pfam 81, SMART 82, Kog 83 and Conserved Domains Databases (CDD) 35. We have also compared the transcripts with other subsets of mitochondrial and rRNA nucleotide sequences downloaded from NCBI, and to several organism proteomes downloaded from NCBI (yeast), Flybase (Drosophila melanogaster), or ENSEMBL (An. gambiae). Segments of the three-frame translations of the EST (because the libraries were unidirectional we did not use six-frame translations), starting with a methionine found in the first 100 predicted AA, or to the predicted protein translation in the case of complete coding sequences, were submitted to the SignalP server 36 to help identify translation products that could be secreted. O-glycosylation sites on the proteins were predicted with the program NetOGlyc (84. Functional annotation of the transcripts was based on all the comparisons above. Following inspection of all these results, transcripts were classified as either Secretory (S), Housekeeping (H) or of Unknown (U) function, with further subdivisions based on function and/or protein families. Phylogenetic analysis and statistical Neighbor Joining (NJ) bootstrap tests of the phylogenies were done with the Mega package 85.

Flea salivary proteins representing approximately 100 gland pairs were resolved by one-dimensional (1D) sodium dodecylsulfate polyacrylamide gel electrophoresis (4–12% gradient gels) and visualised with Coomassie blue staining (Pierce, Rockford, IL). Excised gel bands were destained using 50% acetonitrile in 25 mM NH₄HCO₃, pH 8.4 and vacuum dried. Trypsin (20 μg/mL in 25 mM NH₄HCO₃, pH 8.4) was added and the mixture was incubated on ice for one hr. The supernatant was removed and the gel bands were covered with 25 mM NH₄HCO₃, pH 8.4. After overnight incubation at 37°C, the tryptic peptides were extracted using 70% acetonitrile, 5% formic acid, and the peptide solution was lyophilised and desalted using ZipTips (Millipore, Bedford, MA).

A low molecular protein sample was prepared by resuspending 51 μg of total flea protein salivary homogenate into 2 mL of 100 mM NH₄HCO₃, pH 8.4, containing 10% acetonitrile. Low MW proteins were obtained by centrifugal ultrafiltration using Centriplus 30 kDa ultrafilters (Millipore, Billerica, MA) spun at 750 × g at 4°C. The low MW filtrate was lyophilised and resuspended in 50 μL of 25 mM NH₄HCO₃, pH 8.4, and half of the solution was digested with trypsin (enzyme:protein ratio of 1:50) for 16 h at 37°C. The undigested and digested samples were desalted using C18 ZipTips (Millipore, Bedford, MA), lyophilised to dryness and resuspended in 14 μL 0.1% TFA for subsequent nanoRPLCMS/MS analysis.

The tryptic peptides were analyzed using nanoRPLCMS/MS. A 75 μm i.d. × 360 μm o.d. × 10 cm long fused silica capillary column (Polymicro Technologies, Phoenix, AZ) was packed with 3 μm, 300 Å pore size C-18 silica bonded stationary RP particles (Vydac, Hysperia, CA). The column was connected to an Agilent 1100 nanoLC system (Agilent Technologies, Palo Alto, CA) that was coupled online with a linear iontrap (LIT) mass spectrometer (LTQ, ThermoElectron, San Jose, CA). The peptides were separated using a gradient consisting of mobile phase A (0.1% formic acid in water) and B was (0.1% formic acid in acetonitrile). The peptide samples were injected and gradient elution was performed under the following conditions: 2% B at 500 nL/min in 30 min; a linear increase of 2–42% B at 250 nL/min in 110 min; 42–98% in 30 min including the first 15 min at 250 nL/min and then 15 min at 500 nL/min; 98% at 500 nL/min for 10 min. The LIT-MS was operated in a datadependent tandem MS (MS/MS) mode in which the five most abundant peptide molecular ions in every MS scan were selected for collision induced dissociation (CID) using a normalized collision energy of 35%. Dynamic exclusion was applied to minimize repeated selection of previously analyzed peptides. The capillary temperature and electrospray voltage were set to 160°C and 1.5 kV, respectively. Tandem MS spectra from the nanoRPLCMS/MS analyses were searched against a protein fasta database derived from the flea salivary gland, using SEQUEST operating on an 18 node Beowulf cluster. For a peptide to be considered legitimately identified, it had to achieve stringent charge state and proteolytic cleavage-dependent cross correlation (X_corr) and a minimum correlation (ΔC_n) score of 0.08.

Apyrase activity was measured as described previously 86. Specific conditions are given in the legend accompanying Figure 4.