Of the most abundant transcripts found in the combined analysis of all three libraries were transcripts coding for proteins with similarities to microvilli membrane proteins from A. aegypti and A. gambiae. These transcripts are also homologous to major allergens identified in the cockroaches Blatella germanica and Periplaneta Americana 18 and to a nitrile-specifier protein (PrNSP) from the midgut of Pieris rapae. PrNSP has a role of converting toxic compounds, such as isothiocyanate, into less toxic compounds, such as nitriles, that are excreted in the feces of larval stages of this lepidopteran 19. Four different putative microvilli-associated proteins were identified in the three P. papatasi midgut cDNA libraries (Figure 2). Clusters 1, 2, and 3 represent likely polymorphisms of the same transcript named here "microvilli protein 1" (PpMVP1), which has a predicted molecular weight of 23.7 kDa. Another three transcripts coding for microvilli proteins and derived from clusters 94, 96 and 98 were named PpMVP2, PpMVP3, and PpMVP4, respectively. The predicted molecular weight for these microvilli-associated like proteins is 24.0, 25.6, and 25.6 kDa, respectively. Additionally, each of these microvilli proteins has a potential signal peptide as predicted by SignalP 3.0 and no evidence of transmembrane helices as predicted using the TMHMM 2.0 server. Identity between the amino acid sequences of these microvilli proteins ranges from 21 to 36 percent (Figure 2, black-shaded amino acids) and similarity from 45 to 57 percent (Figure 2, grey-shaded amino acids). The degree of conservation may indicate that these are biochemically distinct from one another and only commonly named based on the previous annotation of other organisms with similar sequences. Searching the translated assembled sequences from an EST database of L. longipalpis identified NSFM-139c08, NSFM-18h11, NSFM-68e08, and NSFM-47h07 as having high sequence homology to the microvilli-associated like proteins PpMVP1, PpMVP2, PpMVP3, and PpMVP4, respectively 17.

Transcripts coding for three different putative peritrophin-like molecules were identified in the midgut of P. papatasi. PpPer1 (cluster 9) and PpPer2 (clusters 12 and 13) transcripts code for secreted proteins with predicted molecular masses of 29.8 and 9.6 kDa, respectively. PpPer1 is comprised of four potential chitin-binding peritrophin-A domains (Figure 3A). PpPer2 is a much smaller predicted protein and has only one potential chitin-binding domain (Figure 3A). A third putative peritrophin, PpPer3, was identified from cluster 26 with an apparent molecular mass of approximately 32 kDa (Figure 3A) and contains two distant putative chitin-binding domains. Phylogenetic analysis using the chitin binding domains of PpPer1, Pper2, PpPer3 and those of peritrophins from several insects (Figure 3B) suggests a low level of conservation between the domains. Insect peritrophins have been reported to bind to chitin fibers via multiple chitin-binding domains, forming the scaffold that maintains the molecular structure of the peritrophic matrix (PM) in the insect gut 20. In addition to their role in the formation of the PM, peritrophins may also play a role in preventing the toxic effects of heme, a bi-product of blood meal digestion. In A. aegypti, AeIMUC1, a mucin that encodes putative chitin-binding domains was recently shown to bind heme 21. Although peritrophins have been characterised from several insects, including A. aegypti and A. gambiae 2022, no information exists related to sandfly midgut-specific peritrophins. PpPer1 and PpPer3 have high sequence similarity, at the protein level, to the translated sequences SFM-03c06 and SFM-02h07 from the L. longipalpis EST database. However, PpPer2 has lower sequence similarity to any of the assembled and translated sequences from the L. longipalpis EST database, suggesting a more divergent or novel molecule.

Among the most abundant transcripts in the cDNA libraries were the previously characterised P. papatasi trypsin-like, PpTryp1, (Cluster 18 with 158 sequences), and PpTryp4 (Cluster 89 with 114 sequences) 13. PpTryp2 (Cluster 23)and PpTryp3 (Cluster 135) were less abundant with 12 and 8 sequences, respectively. Phylogenetic analysis of trypsins from P. papatasi and from other organisms resulted in the formation of two major clades (Figure 4). P. papatasi trypsins co-localised in clade I containing other insect trypsins, while their mammalian counterparts were found in clade II (Figure 4). As detected previously, 13 PpTryp1 and PpTryp2 form a different clade apart from a clade formed by PpTryp3 and PpTryp4 (Figure 4). The P. papatasi trypsins PpTryp1, PpTryp2, PpTryp3, PpTryp4, show high protein sequence similarity to L. longipalpis ESTs NSFM-02a01, NSFM-113h06, NSFM-94b08, and NSFM-165c07, respectively.

Two previously characterised P. papatasi chymotrypsin-like cDNA, PpChym1 and PpChym2 13, as well as a novel chymotrypsin-like, PpChym3 (Cluster 113) were also found in the transcriptome database. This newly identified novel chymotrypsin-like molecule was found in low abundance in the blood-fed midgut library. The predicted Ppchym3 has 36% amino acid identity to Ppchym1 and 30% amino acid identity to Ppchym2. Furthermore, Ppchym3 has a signal secretory peptide (Figure 5A) and has the required His/Asp/Ser amino acid triad necessary for catalytic activity (Figure 5B). Ppchym1 and Ppchym2 both share sequence homology from the assembled sequence NSFM-01d03 from the L. longipalpis EST database, while Ppchym3 is most similar to sequence SFM-01b03.

A number of sequences were identified with homology to carboxypeptidases. The full-length transcript of a putative carboxypeptidase B,PpCpepB, was found from 37 sequences in cluster 16 and has high homology to a carboxypeptidase B identified in A. aegypti (GenBank accession# AAT36733). The predicted amino acid sequence of PpCpepB contains a signal peptide, a propeptide domain, and a carboxypeptidase domain. A putative carboxypeptidase A, PpCpepA, was also identified from cluster 113 based on amino acid sequence homology. Phylogenetic analysis shows that the identified P. papatasi putative carboxypeptidases are separated into distinct clades (Figure 6A). Comparison of sequence homology indicates the potential for these molecules to have substrate specificities of either carboxypeptidases A or B (Figure 6B). Sequence alignment of the two carboxypeptidases depicts the difference in amino acid composition; however, both sequences contain the zinc ion binding motifs of metallocarboxypeptidases (Figure 6B). Additionally, the presence of a putative signal peptide alludes that these molecules are midgut digestive enzymes. Similarity between these carboxypeptidases and those present in L. longipalpis EST database is evident by the high homology between PpCpepA and SFM-05c11 and between PpCpepB and NSFM-32d09.

A putative astacin-like zinc metalloprotease (PpAstacin) was identified from cluster 37, a product of five sequences. This putative astacin-like protein displays a predicted signal peptide and a slightly modified form of the signature zinc binding catalytic domain for proteins in the astacin family (HEXXHXXGFXHEXXRXDR). In PpAstacin, changes in two residues (E to M and R to A) resulted in the motif HEFLHALGFFHMQSASDR (Figure 7A). Although the altered residues may be involved in target specificity the zinc-binding catalytic domain remains conserved. The likely role of this putative protein is blood meal digestion, as astacins molecules have not been implicated in immune functions and a considerable number of transcripts constituting this cluster were derived from the blood-fed midgut cDNA library. This is the first report of this type of protease from the gut of a sandfly, though NSFM-127b08 of the L. longipalpis EST database was identified based on sequence homology.

Two Kazal-type serine protease inhibitors were identified from cluster 111 (PpKZL1) and 859 (PpKZL2) in the cDNA midgut libraries. PpKZL1 codes for a small peptide of 78 amino acids while PpKZL2 codes for a peptide of 89 amino acids. Both proteins are predicted to be secreted based on the presence of signal peptides (Figure 8). PpKZL1 is similar to various small Kazal-type inhibitors found in Drosophila pseudoobscura (gi: 125986397), C. sonorensis (gi:56199538) and the mosquitoes A. aegypti and A. gambiae, and to larger Kazal-type molecules such as infestin 23 from Triatoma infestans (Figure 8A). There is only 28% identity and 42 % similarity between PpKZL1 and PpKZL2 (Figure 8B) suggesting these may have different functions. Additionally, these two Kazal-type cDNAs are similar to the previously characterised thrombin inhibitor, rhodniin,, from the triatomine Rhodnius prolixus 24 (data not shown). Due to their anti-hemostatic effect, rhodniin and infestin are believed to play a role in the fluidity of the blood within the midgut of these vectors. It is conceivable that one or both transcripts coding for Kazal-type thrombin inhibitors identified in P. papatasi may play a role in blood fluidity within the sandfly midgut, allowing it to be fully digested by the various proteases secreted within the midgut following the blood meal. These represent the first Kazal-type serine protease inhibitors identified from sandflies. PpKZL2 shares low sequence similarity with SFM-0406 from the L. longipalpis EST database and no significant similarities were identified for PpKZL1.

Two transcripts encoding putative ferritin light (PpFLC) and heavy (PpFHC) chain subunits were identified in clusters 103 and 122, respectively (Figure 9). After the ingestion of a blood meal the fly encounters a tremendous dose of iron and heme which would be fatal to most organisms. Ferritin is one of the important factors in controlling the high iron load in hematophagous insects. The midgut of blood-feeding insects envelopes the blood meal and consequently makes the midgut tissue the most likely site of iron regulatory molecules. However, ferritin may also be important for oxidative stress not related to the presence of iron or heme, as it is induced by the presence of H₂O₂ in A. aegypti 25. PpFLC and PpFHC are similar to NSFM-144g07 and NSFM-146d09, respectively; molecules identified by searching the L. longipalpis EST database.

From clusters 125 and 232, two transcripts were identified to encode putative GSTs with homology to other dipteran GSTs in the Sigma and Delta/Epsilon classes, respectively. The predicted molecular weights of the two putative proteins are similar at 23.2 kDa for cluster 125 and 24.5 kDa for cluster 232. Within the midgut, these proteins may play an important role in the regulation of reactive oxygen species which occur as a by-product of hemoglobin digestion. Cluster 125 and 232 share high protein sequence similarity with L. longipalpis ESTs NSFM-105e10 and NSFM-74c11, respectively.

A large number of clusters produced by the three cDNA libraries have no sequence similarity to other known proteins. This has also been observed in the analysis of the Chironomus tentans midgut with good evidence that the unknown transcripts contained coding sequences 26. It is also possible that the abundance of unidentifiable sequences may be caused by the sequence quality of the transcripts or that the captured sequences are 3' untranslated regions, non-coding small nuclear RNA, or sequences of uncharacterised organisms such as bacteria and yeast present in the sandfly midgut. A number of clusters with unknown functions were identified as coding sequences which exhibited signal peptides, such as clusters 11 and 126.

From the three cDNA libraries, we identified chitinase transcripts which were then expressed as recombinant proteins for the demonstration of activity in the midgut of P. papatasi sandflies 15. Another product of the cDNA libraries was the identification and characterisation of a galectin protein as the first arthropod receptor for a parasite; specifically, L. major within the P. papatasi sandfly midgut 1.

To investigate the effects of blood feeding on the midgut expression profile in P. papatasi, we compared the abundance of transcripts in sugar and blood-fed cDNA libraries. We hypothesized that a blood meal will have an effect on the expression of sandfly midgut transcripts that will be reflected in the relative abundance of sequences forming a cluster in the two libraries. Chi-square statistical analysis was used to evaluate the significance of the differences in the abundance of midgut transcripts from unfed and blood-fed cDNA libraries thereby identifying different expression profiles of selected midgut transcripts in each cDNA library.

We observed a significant difference (P value ≤ 0.05) in the abundance of a number of midgut transcripts when we compared the sugar-fed and blood-fed sandfly midgut cDNA library. Table 3 shows a list of selected transcripts that were either more abundantly or less abundantly expressed in these two cDNA libraries.

As expected, transcripts coding for proteolytic enzymes such as trypsin (PpTryp4), and chymotrypsin (PpChym2) were more abundantly represented in the blood-fed cDNA library than in the sugar-fed cDNA library (Table 3). Other transcripts coding for peritrophin and microvilli-like proteins and ferritin were also more abundantly represented in the blood-fed cDNA library. Also, we observed a number of transcripts that were less abundantly represented in the blood-fed cDNA library, such as tryspin 1 (PpTryp1), and peritrophin (PpPer2).

In order to validate the results observed by the chi-square analysis, we further characterised several transcripts by semi-quantitative end-point reverse-transcriptase PCR as well as by real-time PCR. These were utilised to assess the relative abundance of transcripts in the midgut tissue under sugar-fed and blood-fed conditions. The investigated transcripts included peritrophins PpPer1 and PpPer2, as well as microvilli proteins PpMVP1, PpMVP2, and PpMVP4.

The results of semi-quantitative PCR can be seen in Figures 10B and 10D where the induction of PpPer1 is clearly evident. The differences in PpPer2 expression between the two midguts conditions is less clear using this technique (Figure 10D). Figure 10A shows the transcript abundance of PpPer1 as fold change over the control gene in non blood-fed and post blood-meal ingestion as measured by real-time PCR. Figure 10C shows the same real-time PCR analysis of the PpPer2 transcript. The profile of the peritrophin transcripts by real-time PCR strongly correlates with the profile found in the libraries based on the number of sequences.

Based on real-time PCR, PpPer1 expression is induced by blood digestion and it is not detected in sugar fed midguts, corresponding with the lack of any sequences produced in the sugar-fed midgut cDNA library, compared to 54 sequences found in the blood fed library. As predicted by the high sequence abundance of PpPer2 in the sugar-fed cDNA library the expression of this transcript is highest in unfed sand flies and seems to be down-regulated by the ingestion of a blood meal (Figure 10C).

Transcription levels of mRNAs coding for microvilli-like proteins (PpMVP1, PpMVP2, and PpMVP4) tested by semi-quantitative PCR and real-time PCR are shown in Figure 11 and illustrate the induction of transcription by the ingestion of a blood meal. This mirrors what is seen by the sequence abundance of the cDNA library, in which only one sequence of PpMVP2 was observed in the sugar-fed cDNA library. The remaining sequences were contributed by the cDNA library produced from blood-fed sandflies.

Pptryp1 low and Pptryp4 high transcript abundance, were in accordance with the results of previously published endpoint reverse-transcriptase PCR 13. Additionally, the previously characterised chitinase molecule, PpChit1, was identified in cluster 243 and produced by three sequences contributed by the blood-fed cDNA library with none present in the sugar-fed cDNA library. The mRNA expression levels of PpChit1 peak at 72 hours post blood-meal ingestion 15.

During its development within the midgut of the sandfly, Leishmania is faced with various potential barriers that may prevent the establishment of the infection. Among such potential barriers are digestive proteases (trypsins and chymotrypsins), the peritrophic matrix and the requirement for parasite attachment to the midgut epithelia to prevent excretion of parasites with remnants of the digested blood. Previous data suggested that Leishmania is able to downregulate proteolytic activity in the sandfly midgut 4. Also, chitinases produced either by the sandfly 15 or by the Leishmania 27 facilitates parasites in the escape from the peritrophic matrix. Attachment to the midgut epithelia occurs via the presence of L. major lipophosphoglycan receptors, such as PpGalec 1 or, in the case of permissive sandflies, via the presence of midgut glycoproteins bearing terminal N-acetyl-galactosamine 28.

In sandflies, only a handful of midgut proteins have been clearly implicated in Leishmania development. Previous data indicated that Leishmania is able to manipulate the activity of certain digestive proteases, inhibiting or delaying their peak activity, possibly in order to survive the proteolytic attack it faces in the midgut of the vector 327. We hypothesized that a blood meal containing L. major will affect the expression profile of midgut transcripts altering the abundance of the different transcripts in each of these cDNA libraries. Table 4 shows the results of the chi-square analysis when transcripts from the blood-fed and L. major-infected blood-fed cDNA libraries were compared. Of interest, the abundance of transcripts coding for proteolytic enzymes were dramatically decreased in the midgut cDNA library of sandflies fed on L. major-infected blood. Additionally, other transcripts that also appear to have their number reduced included those coding for microvilli-associated like proteins and peritrophins. Transcripts such as the one corresponding to PpTryp1 (trypsin 1) and one corresponding to PpPer2 (peritrophin 2) were more abundant. Other transcripts coding for unknown proteins were also less abundant in the L. major-infected cDNA library than in the blood-fed cDNA library. These data suggest that the parasite may be affecting the expression profile of these transcripts and this inhibition, particularly of proteolytic enzymes, may be advantageous for the survival and establishment of the parasite in the midgut of the sandfly.

Phlebotomus papatasi sandflies (Saudi Arabia strain) were obtained from colonies maintained at Walter Reed Army Institute for Research (WRAIR) and at NIAID-NIH. Three to 5-day old female sandflies were fed either on 20% sucrose solution (sugar fed) or on BALB/c mouse whole blood, via artificial meals 1, with or without the addition of 2 × 10 ⁶L. major (V1 strain) amastigotes per ml.

Phlebotomus papatasi female midguts (10 midguts) were dissected from sugar fed only, from blood fed at 6 h (6 midguts), 24 h, 48 h and 72 h post blood meal PBM (5 midguts each) and from L. major-infected at 16 h (3 midguts), 22 h and 96 h (5 midguts each) post infection (p.i.). For blood-fed and for L. major-infected, groups of midguts were pooled for RNA extraction. Pooling was done for the sugar-fed group as well. Messenger RNA was purified with the Micro-FastTrack mRNA isolation kit (Invitrogen-Life Technologies, Carlsbad, CA) and 100 ng of mRNA was used to produce a first strand cDNA. A cDNA library, enriched for full-length cDNA, was synthesized using the SMART cDNA library construction kit (Clontech Laboratories, Mountain View, CA). One microgram of double stranded DNA for each original library (sugar-fed, blood-fed, L. major-infected) was fractionated using a Chromaspin 1000 column (Clontech Laboratories, Mountain View, CA) into small (S), medium (M) and large (L) transcripts based upon their electrophoresis profile on a 1.1% agarose gel. Pooled fractions were ligated into Lambda TriplEx2 vector (Clontech, Mountain View, CA) and packaged into lambda phage (Stratagene, La Jolla, CA). Individual libraries were plated on LB agar plates in order to achieve roughly 200–300 plaques per 182 mm plates.

Unidirectional sequencing of randomly selected clones was completed as previously described 10. Single, isolated plaques were picked from the plate using sterile wooden sticks and placed into 70 μl of water. Amplification of the cDNA was performed using Platinum PCR SuperMix (Invitrogen), 4 μl template, and primers PT2F1 (AAG TAC TCT AGC AAT TGT GAG C) and PT2R1 (CTC TTC GCT ATT ACG CCA GCT G). PCR amplification products were cleaned using either. Multiscreen PCR cleaning plates (Millipore) or Edge Biosystems PCR cleaning plates and three washes with ultra pure water. The cleaned PCR product was resuspended in 25 μl of water of which 4 μl were used for cycle sequencing with PT2F3 primer (TCT CGG GAA GCG CGC CAT TGT) and either DTCS reaction kit (Beckman) or Big Dye 3.1 (Applied Biosystems). Sequencing reaction products were cleaned using Sephadex G-50 (GE Healthcare) in a multiscreen cleaning plate (Millipore) and analysed using either CEQ8000 (Beckman Coulter) or ABI3700 (Applied Biosystems) DNA sequencing instrument.

Detailed description of the bioinformatic analysis of the data appear in 1029. Briefly, prior to analysis the vector sequence was removed from the cDNA nucleotide sequences. Sequence data from the three libraries were grouped together and aligned to generate clusters of contiguous sequences or contigs based on 90% homology over 90 nucleotides, after sequences with more than 5% Ns were discarded. Three frame translations of the consensus sequence of each contig were subjected to comparison using the appropriate BLAST algorithm to the NCBI non-redundant protein database, conserved domain database 30 which contains the eukaryotic clusters of orthologous groups (COG), Simple Modular Architecture Tool (SMART) and Protein Family Database (Pfam), and the Gene Ontology database 31. Nucleotide sequences were directly compared with two customised databases, mitochondrial and ribosomal RNA (rRNA) nucleotide databases using BlastN. Determination of the presence of a signal secretion peptide or transmembrane helices was accomplished by the submission of sequence peptides to the SignalP server 32 or TMHMM server 33, respectively. The L. longipalpis BLAST server was utilized to determine homology between the P. papatasi clusters and L. longipalpis ESTs 34. The number of transcripts each library contributed to a particular contig was derived using a custom program, Count Libraries (JMC Ribeiro, personal communication). Comparisons between the sugar-fed and blood-fed midgut cDNA sequences and comparisons between blood-fed and L. major-infected midgut cDNA sequences were based on separate Chi-square analysis 35. The grouped and assembled sequences, BLAST results and signal peptide results were combined in an Excel spreadsheet and the putative function, if any was manually verified and annotated. Sequences were aligned using Clustal X, version 1.83, and converted to graphical aligned sequences using BioEdit, version 7.0.5.3 36. Phylogenetic analysis was conducted on amino acid alignments using TREE-PUZZLE, version 5.2, generating trees by maximum likelihood using quartet puzzling to calculate node support 37.

Quantitative PCR (qPCR) was performed in selected clones using the first-strand cDNA, obtained from 100 ng total RNA isolated from midguts dissected from P. papatasi females fed on sugar (unfed) or dissected after a blood meal (24–72 h post blood meal or PBM). cDNAs were synthesized using the 1^st Strand cDNA Synthesis kit (Invitrogen, San Diego CA). Transcript levels were measured with SYBR green dye using a LightCycler 2.0 (Roche Diagnostics, Manheim, Germany). For qPCR reactions, samples were subjected to an initial holding step at 95°C for 15 minutes, followed by an amplification step consisting of 35 cycles of 95°C for 10 seconds, 54°C for 20 seconds and 72°C for 20 seconds with a single acquisition. The reaction continued with a single-cycle melting step of 95°C for 10 seconds, 67°C for 30 seconds and 95°C for 10 seconds, prior to cooling for 1 minute. Equal amounts of cDNA were amplified using gene-specific primer sets targeting individual transcripts as well as a P. papatasi alpha tubulin, as control or reference transcript. Reactions were routinely done in duplicate. The relative expression ratio of the target transcript and control or reference transcript (fold over control) was calculated using the LightCycler relative quantification software (Roche).

Semi quantitative RT-PCR reactions were performed with selected transcripts to further demonstrate the differential expression of these genes in P. papatasi midgut. In this case, 100 ng of total RNA isolated from midguts dissected from P. papatasi females fed on sugar (unfed) or dissected after a blood meal (48 h PBM) were used to synthesize a cDNA using the 1^st Strand cDNA Synthesis kit (Invitrogen). PCR reactions were carried out by an initial hot start at 95°C for 5 minutes followed by 25 cycles of 95°C for 30 seconds, 54°C for 1 minute and 72°C for 1.5 minutes and a final extension cycle of 72°C for 5 minutes. PCR products were separated on 1.5% agarose.