Identification and characterization of microRNAs expressed in the African malaria vector Anopheles funestus life stages using high throughput sequencing

The experimental work was performed using a colony of An. funestus (FUMOZ) that originates from southern Mozambique [59]. The mosquitoes were reared in the insectary of the Vector Control Reference Laborator at the National Institute for Communicable Diseases (NICD), Johannesburg, South Africa since 2000. The insectary is kept at 25 °C, 80% relative humidity with a 12-h day/night lighting regime and 45-min dusk/dawn cycles.

Small RNAs were sequenced in quadruplicate using Illumina sequencing from eggs, larvae, pupae and unfed adult females batches to identify An. funestus miRNAs expressed during development. More than 150 million raw reads were produced from each life stage (Table 1). The length distribution of the reads was between 18 and 30 nucleotides (data not shown). After filtering the impurities and reads of length smaller than 18 nucleotides, 75.92, 71.52, 80.85 and 75.39%, high-quality reads had Phred quality values (PQV) of 20 obtained from the eggs, larvae, pupae, and unfed adult females libraries, respectively (Table 1). The PQV has previously been reported to be an indicator of base call accuracy and therefore sequence quality [65].

Mapping reads over the unmasked genome represents an unbiased option, allowing the detection of known and still undiscovered miRNAs. The total number of the reads mapped to An. funestus genome constitutes only 62.61, 63.90, 73.85 and 59.42% of the total high-quality reads from eggs, larvae, pupae, and unfed adult females libraries, respectively (Table 1).

In order to annotate other small non-coding RNAs in the four libraries, all clean reads were aligned against Rfam version 11. As shown in Fig. 1, the most abundant class of small non-coding RNAs in the eggs library were miRNAs. However, rRNAs were most abundant in the larvae and pupae, and tRNAs in the unfed adult females library.

Of the 65 previously known An. gambiae mature miRNAs in miRBase release 21 (June 2014), 64 mature miRNAs were detected in the sequenced short reads of An. funestus (Table 2). The two miR-276 precursors in An. gambiae generate two different mature miRNAs whereas in An. funestus they produce the same mature miRNA. These include all the 46 miRNA sequences computationally predicted in the An. funestus strain FUMOZ genome hosted by the invertebrate vectors of human pathogens database (VectorBase) [66]. The full precursor structure (mature, loop and star sequence) or parts of it were detected. Only star sequences were found for miR-12, miR-306, miR-965, miR-989, miR-1174 and miR-1889 and the precursor locations for these miRNAs were not determined. Reads that do not match any known An. gambiae miRNA sequences from miRBase and mapped to the An. funestus genome were screened for precursor or hairpin structures. The miRDeep2 algorithm was employed to investigate if non-annotated sequences mapping to the An. funestus genome demonstrated folding properties of precursors or hairpins. A total of 43 mature miRNA candidates expressed from 42 precursor candidates were identified (Table 3). Among these miRNAs, 15 miRNAs have been described in the Anopheles genus (An. gambiae and/or An. anthropophagus) but not registered in miRBase database, eight have not been previously reported in the Anopheles genus but are known in other mosquitoes such as Aedes aegypti and/or Culex quinquefasciatus, and the remaining 20 miRNAs have not been described before in any species. For all the mature miRNA candidates, the full or parts of the precursor were detected in the four life stage libraries except for miR-2943 that was detected only in the eggs library. MiRNAs frequently exhibit sequence differences when compared to a reference mature sequence, generating multiple variations that are known as isoforms. New isoforms were identified for miR-927 (miR-927-3p) and miR-980 (miR-980-5p). The new isoforms of miR-980 were expressed only in the eggs library. Our analysis uncovered a third stem-loop precursor for miR-2.

To obtain insight into possible stage dependent roles of miRNAs in An. funestus, the expression patterns of miRNAs in different life stages were examined based on the number of reads obtained. The heatmaps (Fig. 2) summarize the expression of the miRNAs identified in the four life stage libraries. The majority of miRNAs were sequenced between 1–6⁶ times. The expression profiles of miRNAs varied from highly specific to ubiquitous during the four stages. Among the known miRNAs, miR-263a was the most frequently expressed miRNA in the eggs, the larvae, and the pupae libraries (>1 million reads), whereas miR-8 was the dominant miRNA in the unfed adult females library (>300 thousand reads). Nevertheless, miR-87 had the lowest expression level in the egg and the pupa, similarly miR-189 in the larva, and miR-309 and miR-929 in the unfed adult females library. The highly expressed miRNA amongst the new miRNAs in the eggs library was miR-2944a (>1.5 million reads), miR-2765 (>20 thousand reads) in the larvae and pupae libraries and miR-999 (>30 thousand reads) in the unfed adult females library. To analyse the changes in the expression of miRNAs during the mosquito life stages, all read counts within a dataset were normalized, then log2 (fold-changes) were calculated between each two stages. The results of the comparisons (egg-larva, larva-pupa and pupa-unfed adult female) of the expression level of all the miRNAs in the four libraries are shown in Figs. 3 and 4. Between the egg to the larva, 29 miRNAs were up-regulated and 12 other miRNAs were down-regulated. Six miRNAs were up-regulated and another 13 down-regulated between the larva and pupa stage. However, seven miRNAs were up-regulated and four down-regulated between the pupa and the unfed adult female.