Seventeen of the 27 An. stephensi sequences shown in Table 2 match predicted An. gambiae miRNA hairpins (Table 2, category I). Fifteen of the 17 matches coincide with the predicted mature miRNAs described either at miRBase 25 (Table 2, category Ia) or in Chatterjee and Chaudhuri (2006) 23 (Table 2, category Ib). Two sequences appear to be miRNAs*, the passenger strand of the miRNA:miRNA* duplex. The copy numbers of ast-miR-8*, and ast-mir-277* are less than those of ast-miR-8 and ast-miR-277, respectively. Intriguingly, miR-14, which is involved in the regulation of apoptosis and longevity in D. melanogaster 26, represents 25% of all the identified miRNAs. Northern analysis using total RNA from 17-day old females with antisense Locked Nucleic Acid (LNA) probes against 4 selected miRNAs (miR-9a, -14, -210, and let-7) all showed bands of the correct size, confirming cloning results (Figure 1, the last lane). The LNA oligos contain a mixture of DNA nucleotides and LNA nucleotides with 2'-4' methylene linkage providing high binding affinity and enhanced specificity to targets as compared to ordinary DNA oligos 27.

Three An. stephensi small RNAs display perfect or nearly perfect (only 1 mismatch) match to published miRNAs from organisms other than mosquitoes (Table 2, category IIa) and three more show high similarity (> 84% identity over 19 or more nt) to miRNAs from other organisms (Table 2, category IIb). These small RNAs are named according to their closest miRBase matches, which are listed in Additional file 1. All precursor sequences for each of the six miRNAs obtained from the An. gambiae genome assembly formed good hairpins (see Figure 2B for an example). Four of the six Anopheline miRNAs (miR-12, miR-375, miR-2a, and miR-76) have conserved sequences in Ae. aegypti and comparisons between the Anopheles and Aedes hairpins produced miRscan scores ranging from 8.46 to 15.99, supporting their miRNA status (see Table 2 for the range of expected miRscan scores). miRscan looks for hallmarks of miRNAs within a pair of conserved precursor stem-loop sequences by calculating a score based on seven criteria, the most important of which is the conservation of the base pairing between a miRNA and its antisense 282930. Two of the miRNAs, ast-miR-304 and ast-miR-306, do not have conserved sequences (either mature miRNA or hairpin) in Ae. aegypti and are described in the next section. We selected ast-miR-76, a miRNA that showed the lowest similarity to known miRNAs in category IIa and IIb, for further verification using Ribonuclease Protection Assay (Figure 2), which is theoretically more sensitive than northern blot (Figure 1). A product of expected size was detected thus supporting the expression of ast-miR-76.

Two of the above miRNAs are worth noting. ast-miR-2a matches dme-miR-2a perfectly thus is named as miR-2a. There is an aga-miR-2 in miRBase (derived from two precursors aga-miR-2-1 and aga-miR-2-2), which is reverse complementary to ast-miR-2a. Sequence comparison showed that ast-miR-2a is not derived from aga-miR-2* because of the existence of multiple indels/mismatches between the alignment of aga-miR-2 and aga-miR-2*. The orientation of our cloned ast-mir-2a is consistent with the miRscan prediction based on An. gambiae and Ae. aegypti hairpins (Table 2, score 15.99). In addition, ast-mir-304 is reverse complementary to its top match (dme-miR-304) with 86% identity, which is described in the next section.

As mentioned above, Anopheles miR-304 and miR-306 do not have conserved sequences in Ae. aegypti. However, they both have homologs in D. melanogaster (Figure 3). Comparisons between miR-306 sequences in An. gambiae and D. melanogaster produced a miRscan score of 7.53, which is consistent with its miRNA status. Comparisons between miR-304 sequences in An. gambiae and D. melanogaster produced no score (NS) during miRscan analysis. The failure to produce a positive score by miRscan does not automatically indicate that miR-304 is not a true miRNA because nine out of the 88 known C. elegans/C. briggsae miRNAs produced no scores and two even gave negative scores 29. A closer examination of the miR-304 hairpin from An. gambiae suggested that it met all of the previously described criteria for miRNA structures 3132.

Interestingly, miR-304 is closely flanked by miR-12 and miR-283 while miR-306 is in a different cluster with miR-9b and miR-79 (Figure 3). Clustering of An. gambiae miR-9b and miR-79 is noted on miRBase, but not miR-306; clustering of An. gambiae miR-304, -12, and -283 is not predicted in miRBase 3334. Both clusters are within introns of protein coding genes. The miR-12, -304, -283 cluster occurs within a conserved gene of unknown function, while the miR-9b, -79, -306 cluster occurs within an ortholog of a gene coding for a Drosophila serine-threonine kinase group protein. The exons flanking each of the miRNA clusters are conserved between An. gambiae, Ae. aegypti, and D. melanogaster, which indicates that the clusters are orthologous. The order of miRNAs within the introns is conserved between An. gambiae and D. melanogaster, again supporting the miRNA status of the Anopheles miR-304 and miR-306. All miRNAs in the two clusters are in the same orientation as the flanking genes, indicating that these miRNAs may be transcribed from the promoters of their respective flanking genes, which is consistent with previous reports 3536. In this regard, it is suggestive that, albeit reverse complementary to each other, the mature miR-304 in Anopheles and Drosophila are both correctly annotated because the orientation of transcription of miR-304 is consistent with the flanking gene in both species.

As shown in category IIc of Table 2, there are four An. stephensi small RNAs that produced no hits to any known miRNAs from any species based upon searches of the miRBase with an e-value cutoff of 10. These sequences are temporarily named ast-miR-x1 through ast-miR-x4. All four sequences match perfectly to unique locations in the An. gambiae genome assembly. Putative precursor sequences flanking the four miRNAs in An. gambiae showed strong hairpin structures (Figure 4). The precursors of all four miRNAs showed high similarity to Ae. aegypti genomic sequences and these hairpin pairs gave miRscan scores between 14.4–17.45, indicating that they strongly resemble the structure and conservation pattern of known miRNAs. In particular, the mature miRNA sequences were 100% conserved between An. gambiae, An. stephensi, and Ae. aegypti. Northern blot analysis using 17-day old female samples provided further confirmation for these four miRNAs (Figure 1, last lane).

We decided to expand the expression analysis beyond the adult stage and investigate the expression profiles of eight miRNAs across different developmental stages, from early embryo to female adult. Shown in Figure 1 are northern blot results for four known miRNAs as well as all four new miRNAs. Each miRNA showed a distinct pattern. miR-9a is expressed in all life stages examined and its expression is reduced in adults, which is consistent with what was observed in D. melanogaster 21. The level of miR-210 appears to be higher in late embryo and adult females than in other stages. Let-7 expression begins in late larvae in mosquitoes and continues into adult, again similar to what was observed in D. melanogaster. We also determined the expression profile of miR-14, which represents 25% of the miRNA sequences during our cloning experiment (Table 2). As shown in Figure 1, miR-14 displays a strong signal starting from late embryonic to adult stages. miR-14 is observed in all life stages in D. melanogaster as well 37. Further examination of miR-14 expression across adult lifespan showed a relatively consistent expression regardless of age, sex, and hematophagy (Figure 5). The four novel miRNAs (miR-x1, -2, -3, -4) have unique expression patterns as well, which will likely provide useful clues to their function for future research. For example, miR-x2 showed adult-specific expression while miR-x3 showed predominantly pre-adult expression (Figure 1). miR-x2 was not detected in adult males (Figure 6). miR-x3 was at best weakly expressed in adult males while miR-x1 and x4 were clearly expressed in adult males (data not shown).

We decided to carry out a detailed expression analysis of miR-x2 in both An. stephensi and Ae. aegypti. When different An. stephensi tissues were analyzed (Figure 6), miR-x2 showed strong signals in ovaries while no expression was detected in the midgut samples. The expression of miR-x2 was weak in the heads and the "remainders" (thorax plus abdomen without midgut and ovary). Following blood feeding, miR-x2 expression in the ovaries remained high at 24 hours post bloodmeal, but declined sharply by 72 hours post bloodmeal. miR-x2 was hardly detectable among the other tissues in either of the post bloodmeal time points. The same pattern of expression was observed in the distantly related Ae. aegypti (Figure 6). We also confirmed the lack of miR-x2 expression in adult males in both An. stephensi and Ae. aegypti (Figure 6). Thus miR-x2 expression profile is conserved between the two divergent mosquitoes in all samples tested.

An. stephensi (Indian wild-type strain) and Ae. aegypti (Liverpool strain) mosquitoes were maintained in humidified incubators at 27°C on a 12 hour light:dark cycle.

For small RNA cloning, approximately 1000 female An. stephensi adults were collected at 17 days post-emergence. Mosquitoes were fed blood and allowed to oviposit prior to collection. Aged females are the most relevant mosquito life stage for malaria transmission as it takes approximately two weeks for Plasmodium parasites to mature within the mosquito to an infective stage 24. RNA was isolated using a mirVana small RNA Isolation Kit (Ambion, Austin, TX). Subsequent steps were performed following published protocols 6 from the David Bartel laboratory (Whitehead Institute, MIT, Cambridge, MA). RNA oligo markers (18-mer, 24-mer; sequence as detailed from 6) were purchased from Dharmacon (Lafayette, CO) and 2' ends were deprotected prior to use as indicated by Dharmacon. These oligos were end-labeled with ³²P (³²P-γ ATP, Perkin-Elmer, Waltham, MA) by T4 polynucleotide kinase (Promega, Madison, WI) and added to the sample to isolate RNAs between 18 and 24 nt in length. Next, 5' and 3' linkers were sequentially ligated to the isolated small RNA as well as the RNA markers. cDNA were generated by RT-PCR using primers derived from the two linker sequences. cDNA were ligated into a 2.1 TOPO TA vector (Invitrogen, Carlsbad, CA). Ligated plasmids were transformed in One-Shot Mach 1-T1^R Competent cells (Invitrogen). We did not concatemerize cDNA prior to cloning. Sequencing was performed by Virginia Bioinformatics Institute core facilities and the University of Washington High Throughput Genomics Unit.

Sequences obtained from cloning were analyzed with ClustalW 43 to identify inserts and orientation. After removing low quality sequences determined by inspection of chromatographs, insert sequences were analyzed using BLAST 44 against different databases as described below to identify matches to known miRNAs, mRNAs, tRNAs, rRNA, and other small RNAs (Table 1). Sequences that were identical to previously predicted An. gambiae miRNAs were identified by comparing with the miRBase miRNA registry 25 and An. gambiae miRNA predictions listed in 23 using BLAST 44. Comparisons to miRBase also revealed sequences that match miRNAs from other organisms but were not reported in mosquitoes. The remaining An. stephensi small RNA sequences that matched the An. gambiae genome assembly were further analyzed to uncover new miRNAs from mosquitoes (Table 2). An. gambiae genome assembly was used to retrieve miRNA precursors for secondary structural analysis because there was little sequence information available for the An. stephensi genome and the two species are in the same subgenus Celia. The matching An. gambiae sequences plus 100 nt flanking sequences were obtained through Ensembl 45 and lowest energy conformations were generated by Vienna RNAfold 46. The above mentioned An. gambiae precursor sequences were used to search for conserved sequences in Ae. aegypti, and in some cases, in D. melanogaster. Conserved sequence pairs were then examined via miRscan 28. One An. stephensi miRNA (ast-miR-304) was unscorable by miRscan (Table 2) and the precursor of this miRNA was examined using the biogenesis criteria described in the literature: 1) the miRNA appears in the stem of a hairpin structure; 2) there is considerable similarity between the miRNA and miRNA*; and 3) few bulges if any are present within the miRNA:miRNA* pairing in the stem 3132.

Northern hybridizations shown in Figure 1 were conducted using miRCURY-LNA probes from Exiqon (Vedbaek, Denmark) for eight different miRNAs. Total RNA was isolated from An. stephensi of the same age or developmental stage and used for each blot. Ten micrograms of total RNA was loaded for each sample in a 15% denaturing polyacrylamide gel. After initial size confirmation using let-7 RNA, 19 nt and 23 nt single-stranded DNA oligos were used as size markers in subsequent experiments. Gels were stained with ethidium bromide for verification of equal loading and RNA quality before transferring to a BrightStar-Plus membrane (Ambion). Subsequent steps were based upon Wienholds et al. 47. Following UV crosslinking, membranes were prehybridized in a rotating oven for 30 minutes at 42°C using ULTRAhyb-Oligo Hybridization Buffer (Ambion), followed by overnight hybridization in the same buffer at 42°C with a final concentration of 0.1 nM digoxigenin (DIG)-labeled antisense miRCURY LNA probe (Exiqon). All probes used in northern blots were designed to hybridize with An. stephensi miRNAs. The membranes were washed twice for 35 minutes in an Ambion-recommended wash buffer (2 × SSC, 0.5% SDS) at 42°C, and then once in low stringency buffer for 5 minutes at room temperature. The membranes were incubated for 30 minutes in blocking buffer followed by a 1 hr incubation with Anti-DIG-alkaline phosphatase fAb (Roche, Basel, Switzerland) in blocking buffer. The membranes were washed three times for 15 minutes in low stringency wash buffer followed by twice with alkaline phosphatase buffer. The membranes were then immersed in CDP-Star solution (Roche) for 5 minutes, and placed inside saran wrap for exposure to X-ray film for 30 minutes.

For the northern blots shown in Figure 5, a similar procedure was followed. The samples were 3, 5, 10, 17, and 24 day old An. stephensi adult females as well as 3, 5, 10, and 17 day old An. stephensi adult males that were maintained on sugar water. The same cohort of adult females, which were fed on blood on day 5 after emergence and allowed to oviposit two days later, were collected at day 10, 17, and 24. RNA isolation and northerns were carried out as detailed above. For tissue sample northern blots (Figure 6), five day old adult An. stephensi females were split into two groups, one maintained on sugar water, the other provided a blood meal then maintained on sugar water. Approximately twenty-four hrs post bloodfeeding, heads, ovaries, midguts and remainders (everything left behind) were collected from 60 bloodfed and 60 non-bloodfed six day old mosquitoes. This procedure was repeated again at 72 hrs after blood feeding with eight day old mosquitoes. The mosquitoes were not permitted to oviposit. All tissues were stored in RNA later (Ambion) during collection, then vortexed and stored in -80°C until RNA isolation as above using a mirVana miRNA Isolation kit (Ambion). Northern hybridization was performed as describe above except that 5 ug of total RNA were used. The same procedure was followed for Ae. aegypti mosquito tissue analysis. In separate experiments, 10 ug of total RNA isolated from whole body were used to compare the expression of miR-x2 in males and non-bloodfed females of both species. We have performed replicates for all expression profile analyses and obtained consistent expression patterns between replicates.

RPA was used to examine ast-miR-76. Double-stranded oligos with a T7 promoter 5' of an antisense sequence of the miRNA were produced by annealing two single-stranded oligos. The annealed oligos were used to synthesize an RNA probe that was 9 nt longer than the miRNA, as seen below:

ast-miR-76

5'TTCGTTGTCGACGAAACCTGTTTTCTCCCTATAGTGAGTCGTATTA 3'

3'AAGCAACAGCTGCTTTGGACAAAAGAGGGATATCACTCAGCATAAT 5'

Our design also considered the possibility for future multiplexing by adding extra 4 adenosines immediately following +1 which were indigestible according to the instruction manual of the "mirVana miRNA Probe Construction Kit" (Ambion). This permitted the undigested probe to run at ~29 nt, and digested/protected probe to run at ~24 nt. We utilized a "MEGAscript RNAi kit" (Ambion) to synthesize radiolabeled (³²P-UTP, Perkin-Elmer) antisense transcripts. Templates were synthesized for 4 hours, and purified from acrylamide gels by elution and isopropanol precipitation (1× volume) with 1/10 volume 3 M NaOAc and glycogen. 50,000 cpm of probe was added to 5 ug total RNA and denatured at 96°C for 3 minutes followed by 42°C hybridization overnight. Next, the hybridized samples were digested with RNase A/T1 for 45 minutes at 37°C to remove single-stranded RNA, and to trim unhybridized regions of the probes. Afterwards, the samples were ethanol precipitated and resuspended in 5 ul 2 × loading buffer. Resuspended samples were denatured at 96°C for 5 minutes followed by a quick chill on ice before loading. RNA were separated using a 15% denaturing polyacrylamide gel at 150 V. Locations of bands were examined by X-ray film exposure, using an intensifying screen. Size markers were as described for northern blot analysis.