Identification and Characterization of Two Novel RNA Viruses from Anopheles gambiae Species Complex Mosquitoes

Mosquitoes of the Anopheles gambiae complex display strong preference for human bloodmeals and are major malaria vectors in Africa. However, their interaction with viruses or role in arbovirus transmission during epidemics has been little examined, with the exception of O’nyong-nyong virus, closely related to Chikungunya virus. Deep-sequencing has revealed different RNA viruses in natural insect viromes, but none have been previously described in the Anopheles gambiae species complex. Here, we describe two novel insect RNA viruses, a Dicistrovirus and a Cypovirus, found in laboratory colonies of An. gambiae taxa using small-RNA deep sequencing. Sequence analysis was done with Metavisitor, an open-source bioinformatic pipeline for virus discovery and de novo genome assembly. Wild-collected Anopheles from Senegal and Cambodia were positive for the Dicistrovirus and Cypovirus, displaying high sequence identity to the laboratory-derived virus. Thus, the Dicistrovirus (Anopheles C virus, AnCV) and Cypovirus (Anopheles Cypovirus, AnCPV) are components of the natural virome of at least some anopheline species. Their possible influence on mosquito immunity or transmission of other pathogens is unknown. These natural viruses could be developed as models for the study of Anopheles-RNA virus interactions in low security laboratory settings, in an analogous manner to the use of rodent malaria parasites for studies of mosquito anti-parasite immunity.


Introduction
Insect specific RNA viruses from various families (i.e.Flaviviridae, Togaviridae, Bunyaviridae, Rhabdoviridae, Mesoniviridae) and the taxon Negevirus have been described in Anopheles mosquitoes [1][2][3][4][5].These viruses have been discovered by isolation from cell cultures, by RT-PCR and manual sequencing targeting regions of known viruses, or using deep sequencing on field caught insect samples [6,7].The siRNA pathway of mosquitoes is involved in the interaction and processing of the viral double strand RNA (dsRNA) intermediates produced by RNA viruses [8][9][10][11][12][13][14].Deep sequencing of small RNA and bioinformatics has been used to reconstruct active novel viruses in plants, Drosophila or mosquitoes, using detection of viralderived small interfering RNAs (viRNAs) as a criterion for active replication [7,15,16].Small RNA deep sequencing should allow sensitive detection and discovery of viruses that produce dsRNA intermediates.Alternately, viral discovery by inoculation of extracts onto cell lines is not limited to detection of dsRNA intermediates, but can be biased due to differential efficiency of viral replication across cell lineages [17][18][19].
Among Anopheles mosquitoes, the most mature reference genome sequence and genomic tools have been developed for the An.gambiae complex, particularly the sister taxa An. gambiae and An.coluzzii, and these are currently the most powerful Anopheles models.Within the African malaria vectors of the An.gambiae complex, to our knowledge no natural RNA viruses have yet been described, with the exception of O'nyong-nyong virus (ONNV), a pathogenic arbovirus transmitted to humans [20], which is a close relative of Chikungunya virus transmitted by Aedes mosquitoes [21,22].Consequently, research on viral response and antiviral immunity of Anopheles has been limited to ONNV infections [14,[23][24][25][26][27], but research using ONNV requires sophisticated biosecurity conditions.Identification of Anopheles RNA viruses that could be used as low-biosecurity model systems would facilitate studies of Anopheles antiviral immunity and mosquito-virus interactions.
In the current study, we used small RNA deep-sequence datasets to reconstruct more than 90% of a novel Dicistrovirus, and several segments of a previously unknown Cypovirus.We verified their infection prevalence within the Anopheles colony where they were discovered.Oral infection experiments were used to evaluate the species-specificity of these newly discovered viral strains and their ability to be transmitted to the progeny.Finally, we detected both viruses in wild Anopheles mosquitoes captured in Cambodia and in Senegal, confirming that they are components of the natural virome.
The results indicate that a laboratory Anopheles colony, widely used to study vector interactions with different pathogens, carries previously undetected natural insect viral infections that could potentially influence the results of such studies.These novel insect viruses also could become valuable tools to study Anopheles antiviral immunity, as well as the underlying mechanisms of vector interactions with RNA viruses, without the requirement for high biosecurity infrastructure or procedures.

Results
Four datasets of small RNA sequence generated from pooled An. coluzzii (Ngousso colony) mosquito midguts were analyzed using the Metavisitor pipeline [28], a suite of open source bioinformatic tools designed to simplify virus diagnostic, discovery and genome reconstruction on a Galaxy framework [29].Two novel viruses were detected in the mosquitoes, a Cypovirus (genus: Cypovirus, sub-family: Spinareovirinae, family: Reoviridae), and a Dicistrovirus (genus: Cripavirus, family: Dicistroviridae, order: Picornavirales).

Detection of a Cypovirus
We detected a transcript (accession number KU169880) of 892 nt aligning to segment 10 of Culex restuans Cypovirus (CrCPV) [30] and to Uranotaenia sapphirina Cypovirus (UsCPV) [31] with 91% and 83% identity, respectively.Multiple alignments of the nucleic sequences with Clustal O [32] indicated that it was not a chimeric segment of the two known viruses.This segment possesses an intact 237 aa ORF that codes for a complete polyhedrin protein with 96% and 90% amino acid identity to the polyhedrin of CrCPV and UsCPV, respectively.Cypovirus genomes are comprised of 10 to 11 segments of double strand RNA, each coding for a different viral protein.The polyhedrin, protein, coded by segment 10, is highly conserved among Cypoviruses.
A second contig of 3889 nt was reconstructed (accession number KU169879) containing an ORF coding for a 1246 aa protein that has 34% identity with the RNA dependent RNA polymerase (RdRp) of Bombyx mori Cypovirus 1 (ACX54961.1)over 98% of its sequence.Realignment of the original small RNA reads to the contig shows a dicing profile with the read length peaking at 21nt in all four datasets (Fig 1).This is the characteristic product of viral replication intermediates processed by the siRNA pathway [33], indicating that the reconstructed sequences belong to an actively replicating virus infecting the mosquitoes.
Phylogenetic analysis of the predicted polymerase and polyhedrin peptide sequences were performed using the available sequences for the different Cypovirus species (Fig 2).Neighborjoining trees show that reconstructed RdRp and polyhedrin sequences are related but distinct from known Cypoviruses.The CPVs have been classified into 16 types by the International Committee on Taxonomy of Viruses [34], or 21 types based on the electrophoretic migration patterns of their dsRNA genome segments [35].Therefore, it is not surprising that the phylogeny of these viruses inferred from the polymerase sequence, and to a lesser extent from the polyhedrin gene, does not reflect the virus types classification.We conclude that the reconstructed sequences are evidence of a previously unknown virus, here named Anopheles Cypovirus (AnCPV).We hypothesize that AnCPV is a segmented (10 to 11 segments) double stranded RNA virus of the genus Cypovirus, sub-family Spinareovirinae, family Reoviridae.Based on the phylogenetic tree on the polyhedrin gene, this virus could be a novel member of the Cypovirus type 17 [35].
Light microscopy of An. coluzzii larval midguts displayed gross pathological signs of Cypovirus infection, detectable as localized clusters of tightly packed inclusion bodies in the midgut epithelium that appear visually as iridescent blue-white features [31] (white arrows, Fig 3).RT-PCR primers were designed based on the sequence of segments 1 and 10 and successfully used to amplify fragments from viral RNA templates in the total RNA used for the original small RNA library preparation.Sanger sequencing of the amplicons showed no variation from the bioinformatically reconstructed genomic sequences (S1 Zipped Archive).

Detection of a Dicistrovirus
Other contigs reconstructed from the same small RNA datasets displayed significant similarity using blastx to Drosophila C virus (DCV) and Cricket paralysis virus (CrPV) from the Distroviridae family.Using additional long paired-end RNAseq datasets (accession number ERS977505), we were able to reconstruct a 8919 nt viral genomic sequence that completely includes these contigs and contains two ORFs (accession number KU169878).The first ORF (5' ORF) codes for an 1802 aa (nucleotides 391 to 5799 in the reconstructed genome) polyprotein with 48% blastp identity to DCV nonstructural polyprotein.The second ORF (3' ORF) codes for a predicted 852 aa protein (nucleotides 6067 to 8625 on the reconstructed genome) and displays 54% blastp identity to DCV structural polyprotein.This dicistronic organization of the reconstructed genome is typical of Dicistroviridae with an ORF1 that encodes for nonstructural proteins and an ORF2 that encodes for virion proteins (reviewed in [36]).Dicistroviridae are related to the Picornaviridae, which encode the same proteins in a monocistronic manner as one polyprotein, and to the Comoviridae, which encode the two ORFs but on two monocistronic RNAs.
Realignment of the small RNA reads on the reconstructed genome revealed, as for AnCPV, a dicing profile of virus replication intermediates, yielding virus-derived siRNAs with a read length peaking at 21 nt in all four datasets (Fig 4).These results are evidence for a previously unknown, actively replicating virus.
Phylogenetic tree reconstruction of non-structural proteins from Dicistroviridae shows that the discovered virus is related to known Dicistoviruses and groups with the Cripavirus genus (Fig 5).Similar to the detection of AnCPV, RT-PCR primers, designed on the reconstructed AnCV sequence, were also successfully used to generate amplicons from the total RNA used for small RNA library preparation.Sanger sequencing of the amplified region showed no variation from the bioinformatically reconstructed genomic sequence (S1 Zipped Archive).Because a paralysis phenotype was not observed in the source mosquito colony, we inferred that the virus is most likely a C-type Dicistrovirus rather than a paralysis virus and named it Anopheles C virus (AnCV).
We investigated whether, in addition to the two ORFs at similar positions found in other Dicistroviruses, AnCV also possessed an Internal Ribosome Entry Site (IRES) of the same class and family as observed in other Dicistroviruses in order to translate the second ORF.Searching the RNA family database (Rfam) [37], we identified a 200nt region (position 5795 to 5994) of

Cypovirus and Dicistrovirus biological evidence
Cloned virus fragments were used to synthesize RNA probes for hybridization to a Northern blot of the total RNA used for small RNA sequencing (Fig 7).The probes with sequence homology to the AnCV genome and the segment 1 (RdRp) of AnCPV revealed the presence of intact genomic viral RNA in the total RNA from mosquito pools for both viruses.This result also rules out RNA degradation as a source of the small RNAs detected by deep sequencing, and strengthens the interpretation of the dicing patterns (Figs 1 and 4) as a product of RNA from actively replicating virus.
In order to assess the infection prevalence and pattern in the An.coluzzii colony, we examined individual mosquitoes and individual mosquito organs by RT-PCR using virus-specific primers (Table 1).The results indicated that adult mosquito midguts, carcasses and salivary Mosquitoes displayed variable intensity of the RT-PCR amplification products on agarose gels.Although not a quantitative assay, this observation suggests that individual mosquitoes harbored different levels of infection.Mosquitos were infected by one, both or neither of the two viruses (S1 Table ).Some mosquitoes were PCR-positive for AnCPV in the carcass but negative for the midgut, suggesting that infection in those mosquitoes might have been resolved in the midgut or that transmission might have been vertical.The reason for differences in the AnCV prevalence between batch 1 and 2 could be explained by multiple factors such as age of mosquitoes since emergence, or blood feeding status.Detailed studies will be needed to clarify the tropism and temporal dynamics of infection in mosquitoes.

Mosquito host range and transmission
Other colonies of An. coluzzii (Fd03, Fd05, Fd09) and An.gambiae (Mbita) tested by RT-PCR were consistently infected with both viruses, but the significance of the observation is unclear because they were all maintained in the same insectaries as the Ngousso An. coluzzii colony at Institut Pasteur Paris.Samples tested from a colony of An. dirus (Pursat) maintained at Institut Pasteur Cambodia were also positive for both viruses, but the history of other insectaries where the colony was previously maintained is unknown (although it has not been kept in the Paris facility).
Interestingly, we did not detect either virus in a colony of An. stephensi (SDA500) maintained in the same Institut Pasteur Paris insectary as the An.coluzzii Ngousso colony, nor in two Aedes aegypti isofemale lines.Since oral transmission is the natural route of DCV infection [38] as well as of the two known mosquito Cypoviruses [30,31], we tested whether the novel viruses are transmissible orally to these uninfected mosquito species.We reared eggs from the virus-negative An. stephensi and the two Ae.aegypti lines and divided first instar larvae of each into two trays.One tray of larvae was supplemented with homogenized An. coluzzii Ngousso fourth instar larvae, carrying AnCV and AnCPV viruses, while the other tray of larvae was fed normally and maintained in the same insectary as a control.An. stephensi and Ae.aegypti fourth instar larvae and adult females were then collected and assayed for AnCV and AnCPV as pools or individually.Larvae were washed in multiple changes of clean water and starved over several hours before processing in order to minimize the possibility of external contamination or midgut content contamination.This per os exposure did not infect the Ae.aegypti mosquitoes, but successfully infected An. stephensi mosquitoes with both AnCV and AnCPV at a prevalence of 59% and 100% respectively, while mosquitoes issued from the control trays remained negative (Table 2).
To assess whether the viruses can be transmitted vertically, An. stephensi adult females developed from larvae experimentally infected as above (the F0 generation) were allowed to blood feed and lay eggs (F1 generation).Both F1 larvae as well as resulting adult females displayed AnCPV and AnCV infections (Table 3).These results suggest that both of the viruses can be transmitted vertically to the progeny by oviposition.An. stephensi mosquitoes maintained until the F4 generation after virus exposure displayed persistent infection with AnCPV, but AnCV prevalence decreased each generation and was absent in F4 adults.

Virus presence in wild Anopheles mosquitoes
We captured wild Anopheles spp.mosquitoes in villages in Senegal and Cambodia, and tested individual mosquito RNAs for the viruses using specific RT-PCR.Bands of the correct predicted sizes were detected in some of the samples: 494nt for AnCV and 524nt for AnCPV (Fig 8).Samples were processed along with negative controls in order to control for cross-contamination.Individual wild mosquitoes were positive for one, both, or neither virus.One wild Anopheles spp.from Senegal was positive for AnCV, and six wild Anopheles spp.from Senegal and Cambodia were positive for AnCPV.The resulting amplicons were sequenced in both orientations.In addition, AnCV amplicons were detected in two An.dirus mosquitoes from the Cambodia insectary colony, and were also sequenced.Alignment of these

Discussion
By deep sequencing of small RNAs we have identified two novel insect viruses, a Cypovirus and a Dicistrovirus, in a laboratory colony of An. coluzzii mosquitoes.Reconstructing viral genomes from viral-derived small RNAs enriches for detection of actively replicating viruses.We present biological evidence for the presence of these viruses as a persistent infection in colonies of three species of Anopheles (An.coluzzii, An. gambiae, An. dirus) and their exclusion from a colony of one other species (An.stephensi).Because of the possibility of horizontal transmission in the insectary, data from colonies are not informative for infection in nature, but it does at least indicate the species range of permissive hosts.Importantly, the viruses were present in wild Anopheles mosquitoes captured in Senegal and Cambodia, thus establishing them as components of the natural virome.
We identified Anopheles Cypovirus (AnCPV), which contains a genomic segment coding for the polyhedrin protein characteristic of this viral family and a genomic segment coding for an RNA-dependent RNA polymerase that shows conservation with the genomes of known Cypoviruses.This is, to our knowledge, the first reported member of the Cypoviridae family infecting Anopheles mosquitoes, complementing previous studies identifying Cypoviruses in other mosquito genera [30,31].We also identified an Anopheles C virus (AnCV) that displays genetic architecture consistent with other members of the family Dicistroviridae, with two open reading frames, 5' non-structural polyprotein and 3' structural polyprotein, as well as conserved RNA secondary structure and sequence of the central IRES.This virus was further confirmed to belong to the Dicistroviridae family based on the amino-acid sequence Table 1.Viral prevalence in An. coluzzii Ngousso mosquitoes.Total RNA from midgut and carcass of individual mosquitoes was tested by RT-PCR for infection with AnCV and AnCPV.Salivary gland RNA was extracted from individual mosquitoes different from those used for the carcass and midgut dissection.Percent infected represents the total number of mosquitoes with infection in either carcass or midgut.ND, not done.Data available in S2 File.conservation of the AnCV RNA-dependent RNA polymerase with its orthologs of the Drosophila C virus (DCV) and Cricket paralysis virus (CrPV).
The novel Dicistrovirus AnCV can infect some Anopheles but not Aedes mosquitoes, and appears not to induce high mortality nor detectable paralysis phenotypes in mosquitoes infected per os.These data suggest that infection phenotype and range of this virus resemble that of DCV and not CrPV.We cannot exclude that in our model, the virus and the laboratory colony could have co-evolved to minimize pathogenic symptoms.The virus-negative An. stephensi colony, when infected with AnCV per os, also did not display detectable pathogenic symptoms.This scenario could be tested with isolated viruses by infection experiments using different concentrations of virus to study symptoms and pathogenicity.Isolated and culturable viruses would also allow studies of virion morphology, mechanisms of entry, host range, obtaining complete genomic sequences, and generating infectious clones.The main mosquito organs were found to be positive for both viruses.Development of a quantitative real-time PCR assay would be useful in order to measure genome copies in different organs.
DCV, first isolated in 1972 [39], and CrPV are models for the study of insect antiviral immunity.These viruses as well as other members of the family are ecologically and agriculturally relevant [40].The discovery of a virus from this family in laboratory and wild anopheline mosquitoes is an important first step to understanding the natural viral population in these major human malaria vectors, and potentially establishing a viral infection model in those mosquitoes.
It will be interesting to determine whether AnCV has similar phenotypes for Anopheles mosquitoes as the model insect virus DCV has for Drosophila.For example, DCV leads to a strong immune response and to fly death within several days when intra-thoracically injected [41], but produces only mild effects after infection by the oral route, with transcriptional induction of only 11 genes [42].Although the small transcriptional response may suggest a coevolved state, DCV has also been associated with population crashes and shorter longevity of some laboratory colonies, even in persistently infected fly stocks [40].
The first infectious clone of a Dicistrovirus, CrPV [43], was recently generated, which suggests that an AnCV infectious clone as a tool is within reach.This would open an array of possibilities to study RNA virus antiviral mechanisms in the malaria vector mosquito under lowbiosecurity conditions.The discovery of the first RNA viruses in the An.gambiae species complex is an initial step that will facilitate work on mosquito-virus interactions and anti-viral immunity, including the influence of the microbiota as well as potential reciprocal interactions with other pathogens such as malaria parasites.

Mosquito colonies
Deep sequencing of small RNAs used mosquitoes from the An.coluzzii Ngousso colony previously established from Yaoundé, Cameroon [44], and formerly described as An.gambiae M molecular form.Other colonies used and raised in the Institut Pasteur Paris insectaries were An.gambiae Mbita from Kenya (obtained from the ICIPE Field Station, Mbita Point, Kenya), Fd03 initiated from Mali, Fd05 and Fd09 from Burkina Faso, and An.stephensi colony SDA500.In addition, two Aedes aegypti isofemale lines from Thailand were used [45].All mosquitoes were reared under standard conditions at 26°C and 80% relative humidity, with a 12 h light/dark cycle as previously described [44].Finally, mosquito samples from the An.dirus Pursat colony reared in the insectaries of Institut Pasteur Cambodia, and never grown at Institut Pasteur Paris, were also used.

Wild mosquitoes
Wild Anopheles mosquitoes were used for testing of virus presence in nature.Mosquitoes were captured in Ratanakiri and Kampong Chnang, Cambodia and Kedougou, Senegal were visually identified as Anopheles spp., and immediately transferred into RNAlater at 4°C, and then stored at -80°C.Permission to collect mosquitoes was obtained by Institut Pasteur Cambodia from authorities of Ratanakiri and Kampong Chnang, and by Institut Pasteur Dakar from authorities of Kedougou.RNA was extracted from individual mosquitoes using the Nucleospin miRNA kit (Macherey-Nagel) following the supplied protocol.RT-PCR assays for AnCV and AnCPV amplicons were performed as described below (using 45 PCR cycles), with blank reactions and negative control RNA samples.Amplicons from positive samples were Sanger sequenced and aligned with virus sequences assembled from deep sequencing of An. coluzzii Ngousso mosquito RNA.SNPs were verified using the trace files.

RNA preparation and deep sequencing
Pools of 30 An. coluzzii Ngousso colony mosquito midguts were homogenized in TRIzol (Ambion) and total RNA was extracted following the supplied protocol.Small-RNA libraries were prepared from the small RNA fraction using TruSeq Small RNA Library reagents (Illumina).Libraries were sequenced on an Illumina Hiseq 2500 in a multiplexed 51 +7 bases single read using a TruSeq SR cluster kit v3 cBot HS, and a TruSeq SBS kit v3 HS 50 cycles (Illumina).Primary analysis of the sequences was performed with Casava software (v1.8 Illumina).Library preparation and sequencing was performed by ARK Genomics at the Roslin Institute (University of Edinburgh).Data are available in the EBI European Nucleotide Archive (http://www.ebi.ac.uk/ena/) under study accession number ERP012577 and sample accession numbers: ERX1180340, ERX1180341, ERX1180342 and ERX1180343.
Long paired-end RNA libraries were prepared and sequenced by the Institut Pasteur Transcriptomics and Epigenomics core facility using standard long paired-end rapid run protocol on an Illumina Hiseq 2500.Primary analysis of the sequences was performed with Casava software (v1.8 Illumina).Long paired-end sequencing data are available in the EBI European Nucleotide Archive (http://www.ebi.ac.uk/ena/) under accession numbers: ERX1220645 and ERX1220646.

Bioinformatic virus discovery
Bioinformatic analysis was performed with the Galaxy [29,46,47] Mississippi server instance hosted by the ARTbio facility using the Metavisitor workflows (v1.0) [28].Briefly, small RNA sequencing reads were clipped of the adapter sequence, converted to fasta format and concatenated.Then reads were converted to a weighted fasta format to reduce computing load.Reads were then depleted by successive bowtie alignments on the genomes of the known organisms present in the dataset.Unaligned unique reads were then used for the subsequent contig assembly.Velvet and Oases were used to reconstruct longer contigs with k-mers ranging from 15 to 29 [48,49].The NCBI BLAST tool suite was used to search a BLAST database created from the nucleotide sequences deposited in the NCBI repository with the reconstructed virus contigs, for related homologies.To reconstruct a more complete AnCV genome, an additional dataset from long paired-end sequencing was used (ERS977505).Similarly, Metavisitor was used but with longer k-mers (ranging from 25 to 69).These analyses are accessible in the Mississippi Galaxy server using the link to a Galaxy page https://mississippi.snv.jussieu.fr/u/drosofff/p/anopheles-viruses.

IRES structure analysis
Using the structure of the IRES RF0458 family as a scaffold and the multiple alignment between the seeds of this IRES family and the potential IRES sequence of AnCV (S2 File, we used webserver based versions of TRANSAT to plot the most probable conserved helix and secondary structures of the IRES [50](Fig 6A) (http://www.e-rna.org/transat),and R-chie (http:// www.e-rna.org/r-chie) in "single covariance" mode to plot the covariance between the nucleotides in the structure [51](Fig 6B).Both programs were used with the defaults parameters and with the structure and multiple alignments given in S2 File as inputs.

RT-PCR assays for detection of virus presence
RNA from individual mosquitoes, individual organs, saliva or pools of either was extracted using either TRIzol (Ambion), NucleoSpin RNA II or NucleoSpin 96 RNA kits (Macherey-Nagel) according to the supplied protocol.RT-PCR was performed using Superscript One-Step RT-PCR kit (Invitrogen), primer (S2 Table ) annealing was performed at 56°C.For the sequencing of PCR amplicons from wild or colony mosquitoes, amplicons were purified using the QIAquick PCR Purification kit following manufacturer instructions.Sanger sequencing was performed using T7 primers by SEGENIC Eurofins, Cochin Institute, Paris.

Northern blotting
RNA migration and Northern blotting used standard procedures.Northern blot sense and antisense probes were synthesized by SP6 transcription of the PCR fragments incorporated in the PGEM-t easy vector.Multiple vector clones were verified by classical Sanger sequencing and no variation from the in silico reconstructed contigs was observed.RNA probe hybridization to the Northern blot was performed at 68°C.
MUSCLE multiple alignment and evolutionary analyses were conducted in MEGA (Tamura et al., 2011) with 1000 bootstrap replicates.Because non-conserved regions could possibly influence the phylogenetic tree reconstructions, gaps were removed using the 'Complete Deletion' option in MEGA.Phylogenetic reconstructions using neighbor joining (NJ), maximum parsimony (MP) or maximum likelihood (ML) showed the same topology for each dataset.

Fig 1 .
Fig 1. Readmaps and size distributions of small RNAs that align on the Anopheles Cypovirus reconstructed RdRp genomic sequence.The four small RNA datasets were independently aligned to the sequence of Anopheles Cypovirus segment 1 (accession number KU169879) with bowtie allowing no mismatches.From the bowtie output, lattice maps and read size histograms were generated representing the positions and size of the small RNAs that align to the viral genomic segment.doi:10.1371/journal.pone.0153881.g001 Fig 3. Microscopic observation of Anopheles coluzzii larvae midguts.Signs of Cypoviral infection are white-blue iridescent areas as indicated by the white arrows in each panel.Each panel corresponds to a different mosquito larva.doi:10.1371/journal.pone.0153881.g003

Fig 4 .
Fig 4. Readmaps and size distributions of small RNAs that align on the Anopheles C virus reconstructed genome.The four small RNA datasets were independently aligned to the sequence of the Anopheles C virus (AnCV) genome (accession number KU169878) with bowtie allowing no mismatches.From the bowtie output, lattice maps and read size histograms were generated representing the positions and size of the small RNAs that align to the viral genome.doi:10.1371/journal.pone.0153881.g004

Fig 5 .
Fig 5. AnCV non-structural protein phylogeny.Neighbor-joining tree of non-structural polyprotein sequences of Dicistroviruses.NJ bootstrap values (above 0.75) are indicated above branches.The two genera of the Dicistroviridae family are indicated on the tree.Pairwise genetic distances are indicated in S6 File.doi:10.1371/journal.pone.0153881.g005

Fig 6 .
Fig 6.AnCV IRES show conservation of structure with the Cripavirus IRES family (RF0458).Multiple sequence alignment of AnCV potential IRES sequence (nucleotides 5800 to 6001) with the seven seed IRES of the RF0458 family.Panel (A) represents the most probable conserved helix and secondary structures of the IRES using the known RF0458 structure as a scaffold, and panel (B) depicts the covariance between the nucleotides in the known RF0458 structure, showing covariation of the nucleotides to conserve IRES structural integrity.doi:10.1371/journal.pone.0153881.g006

Fig 7 .
Fig 7. Northern blot of total RNA purified from pooled midguts of An. coluzzii adult mosquitoes shows the presence of the positive strand of AnCV genome and AnCPV segment 1. doi:10.1371/journal.pone.0153881.g007

S1
Zipped Archive.Compressed file containing raw Sanger sequencing results for the RT-PCR viral fragments.(ZIP)

Table 2 .
Prevalence of infection with AnCV and AnCPV in An. stephensi adult F0 mosquitoes exposed to viruses in the larval stage.Individual adult mosquitoes were collected after feeding as larvae with extract of infected An. coluzzii, and infection with both viruses was tested by RT-PCR.Values represent the proportion of mosquitoes with infection in indicated tissues. doi:10.1371/journal.pone.0153881.t002

Table 3 .
Prevalence of infection with AnCV and AnCPV in An. stephensi F1 generation mosquitoes.Individual larvae or adult mosquitoes were collected the generation after exposure to viruses, and infection with both viruses was tested by RT-PCR.