A Systematic Review of the Natural Virome of Anopheles Mosquitoes

Anopheles mosquitoes are vectors of human malaria, but they also harbor viruses, collectively termed the virome. The Anopheles virome is relatively poorly studied, and the number and function of viruses are unknown. Only the o’nyong-nyong arbovirus (ONNV) is known to be consistently transmitted to vertebrates by Anopheles mosquitoes. A systematic literature review searched four databases: PubMed, Web of Science, Scopus, and Lissa. In addition, online and print resources were searched manually. The searches yielded 259 records. After screening for eligibility criteria, we found at least 51 viruses reported in Anopheles, including viruses with potential to cause febrile disease if transmitted to humans or other vertebrates. Studies to date have not provided evidence that Anopheles consistently transmit and maintain arboviruses other than ONNV. However, anthropophilic Anopheles vectors of malaria are constantly exposed to arboviruses in human bloodmeals. It is possible that in malaria-endemic zones, febrile symptoms may be commonly misdiagnosed. It is also possible that anophelines may be inherently less competent arbovirus vectors than culicines, but if true, the biological basis would warrant further study. This systematic review contributes a context to characterize the biology, knowledge gaps, and potential public health risk of Anopheles viruses.


Introduction
Anopheles mosquitoes are the vectors of human malaria, which causes at least 400,000 deaths and 200 million cases per year [1]. Approximately 90% of malaria deaths occur in sub-Saharan Africa, 7% in South-East Asia and 2% in the Eastern Mediterranean Region, with children under five years of age the most affected. More than 480 species of Anopheles have been described worldwide and about 70 of these are responsible for human malaria transmission, with about 40 regarded as the dominant malaria vector species [2,3].
However, the research focus on Anopheles as vectors of malaria has led to a relative lack of study about Anopheles viruses. In addition to malaria parasites, Anopheles mosquitoes also harbor viruses, collectively termed the virome. The Anopheles virome is poorly studied, and the number and function of viruses are unknown. Some of them are confirmed arthropod-borne pathogenic viruses (arboviruses), which multiply in the mosquito vector before transmission to a vertebrate host. Others are thought to be insect-specific viruses that may replicate only in the insect host [4,5].
Culicine mosquitoes such as Aedes and Culex are the main vector of arboviruses such as dengue virus (DENV; genus Flavivirus, family Flaviviridae), yellow fever virus (YFV, genus

Materials and Methods
Four databases were searched in this work: PubMed, Web of Science, Scopus, and Lissa. In addition, manual searches of online and print resources were carried out.
Three combinations of keywords were used for searching the PubMed database. These were (i) ( Searching article titles proved to be useful because the search terms were compatible across bibliographic databases. Thus, search (iii) above was easy to translate into the Web of Science, Scopus, and Lissa databases, as follows. Web of Science advanced search of all databases, including Web of Science Core Collection, KCI-Korean Journal Database, MEDLINE, Russian Science Citation Index and SciElo Citation Index: TI = Anopheles AND TI = Viruses NOT TI = (Aedes OR Culex); Scopus advanced search: (TITLE (Anopheles AND Viruses) AND NOT TITLE (Aedes OR Culex)); and the French language Lissa: 'Virus d'Anopheles.ti SAUF (Aedes OR Culex).ti".
The above searches were carried out from 15 January to 4 May 2016. Moreover, an alert with the keyword (((Anopheles AND viruses) NOT (Aedes OR Culex))) OR ((("Anopheles/virology"[Mesh]) NOT "Culex"[Mesh]) NOT "Aedes" [Mesh]) was created and followed in PubMed from 15 January 2016 to the submission date of this manuscript. Anopheles virus/virus d'anophèles were also searched in Google. In addition, the key word ''Anopheles" was used to identify arboviruses associated with Anopheles species in the online Arbocat Arbovirus Catalog resource, maintained by the Centers for Disease Control and Prevention, USA (https://wwwn.cdc.gov/arbocat/). In addition to these online searches, the books from the central library of the Institut Pasteur (Scientific Media and Information Center (CeRIS)) specialized in microbiology, virology, entomology, immunology, molecular biology, and biochemistry were searched.
Finally, eligibility criteria were applied for inclusion of a virus in the study: (i) virus species and name coherent with the standards of the International Committee on Taxonomy of Viruses (ICTV), (ii) reporting of diagnostic tools to permit independent detection, and (iii) some amount of nucleotide sequence. Reports of a putative virus were ineligible if they met none of these criteria, for example if the report was based only on observation of cytopathic effects on cultured cell lines, or pathology in mice. The Lissa database of scientific literature written in French returned a single article using the above English search terms. However, this article in French (with English keywords) was written in 1957, and was not included in any of the three other databases. In addition, when using French language search terms, the Lissa database identified 13 additional articles (for a total of 14), all in French, that were uniquely identified by Lissa and not by the other three databases. Nevertheless, some of the search terms such as "virus" and "Anopheles" are spelled the same in French and English, and further work would be required to determine whether the different search results are due to search term language, or distinct database contents.  (Table 1). Anopheles viruses have been reported on all continents except the poles (Figure 1).  The densoviruses belong to the Parvoviridae family, characterized by a non-enveloped virion containing a linear single-stranded DNA genome. AgDNV has a genome size of 4139 nt and is organized as two overlapping reading frames that encode the viral proteins (VP) of which activity is fundamental for virus infectivity and two non-structural (NS) proteins involved in the DNA replication. The NS1 portion displays 87% homology with Aedes aegypti densovirus (AeDNV) [33].

Viruses
AgDNV was discovered in the An. gambiae cell line, Sua5B, during an experiment to infect the cells with Wolbachia. It was maintained between different generations by vertical and horizontal transmission and has no detectable effect on mortality of An. gambiae larvae. Virus purification was done from crude cell lysates on a density gradient, and icosahedral, non-enveloped particles of 20 nm were observed by transmission electron microscopy [33].
The use of DNVs as expression vectors was demonstrated by the transfection of Aedes albopictus C6/36 cell line with a plasmid containing an infectious sequence of Ae. aegypti DNV. This infection yielded the same quantity of the monomeric replicative form as infection with wild type virions [64]. A recombinant AgDNV carrying the enhanced green fluorescent protein (EGFP) under control of actin5C promoter has been produced [33]. EGFP transducing virions infected 50% of adults and were able to disseminate to the fat body, midgut, hindgut, malpighian tubules, and ovaries, and were vertically transmitted to subsequent generations. These results indicate that recombinant AgDNV could be a candidate for paratransgenesis, for example by carrying an anti-Plasmodium peptide for reducing the vector competence of Anopheles to Plasmodium spp. In addition, the AgDNV titer is higher in older An. gambiae adults compare to the larvae and pupae stages, suggesting that it could have potential as an adult stage bio-insecticide [65]. Tested DNVs are innocuous to mammals [64].

Iridovirus: Anopheles minimus Virus (AMIV)
Anopheles minimus virus is an iridescent virus (IIV) belonging to Iridovirus genus (family, Iridoviridae). The genera Chloriridovirus, Lymphocystivirus, Megalocytivirus, and Ranavirus are also included in this virus family. The icosahedral virions of AMIV are roughly 130 nm in diameter and the DNA genome is 163 kb in size [66]. The genomic DNA is associated with proteins and the internal membrane is composed of phospholipids. Circularly permuted double-stranded DNA (dsDNA) genomes are characteristic of this family.
AMIV was isolated from wild adult Anopheles minimus, a major Southeast Asian malaria vector, after inoculation of C6/36 cells with mosquito extract [35]. BHK21 and Vero-E6 cells can also be infected with AMIV, with a significant cytopathic effect [66]. AMIV is the first iridovirus isolated from Anopheles species. An Ae. aegypti iridovirus (IIV-6) causes cytopathic damage leading to the reduction of body size, fecundity and longevity. Horizontal transmission by cannibalism and vertical transmission of IIV-3 were observed in Ochlerotatus taeniorhynchus [67]. Iridovirus infection leads to an apoptotic response in invertebrate and vertebrate cells [67].

Poxvirus: Myxoma Virus (MYXV)
The myxoma virus genome was detected by polymerase chain reaction (PCR) and sequencing in wild caught An. maculipennis that fed on wild rabbits [22]. The An. maculipennis group includes major historic vectors of human malaria in North African and Europe [68]. Myxoma virus is in the Leporipoxvirus genus and Poxviridae family. This family includes the subfamilies Chordopoxvirinae and Entomopoxvirinae. The latter infects insects and comprises Alphaentomopoxvirus, Betaentomopoxvirus, and Gammaentomopoxvirus genera. Myxoma virus can be mechanically transmitted by mosquitoes and fleas but does not replicate in them, and thus is not an arbovirus. The Poxviridae genome is comprised of linear dsDNA, and is enveloped. Poxviruses share features with other DNA viruses such as the asfarviruses, iridoviruses, and phycodnaviruses [69].

Alphavirus: O'nyong Nyong Virus (ONNV), Venezuelan Equine Encephalitis Virus (VEEV), Western Equine Encephalitis Virus (WEEV), Sindbis Virus (SINV), Semliki Forest Virus (SFV), and Eilat Virus (EILV)
The Alphavirus genus in the Togaviridae family is frequently associated with Anopheles mosquitoes. O'nyong nyong virus (ONNV) is generally regarded as the only arbovirus transmitted by Anopheles mosquitoes. Alphaviruses are enveloped viruses with a single-stranded RNA of positive sense and a genome size of around 10,000 nucleotides. The 5 end is capped and the 3 end is polyadenylated. The linear RNA genome encodes nonstructural proteins (nsP1 to nsP-4), and structural proteins, although only the nsPs are translated from the genome, while the structural proteins are translated from a subgenomic RNA transcribed after infection. The nonstructural protein nsP1 is necessary for infectivity, nsP2 is necessary for replication and transcription of viral RNAs, nsP3 forms cytoplasmic complexes with different host factors, and nsP4 is a RNA-dependent RNA polymerase. A domain-swap experiment replacing part of the CHIKV nsP3 with the ONNV sequence allowed the chimeric virus to infect An. gambiae [70], highlighting a determinant of mosquito host specificity that requires further investigation. The capsid protein, the glycoproteins (E1 and E2) and the small peptides, E3 and 6K are structural proteins. A single mutation of E1 glycoprotein (E1-A226V) increases CHIKV transmission by the mosquito Aedes albopictus [71]. E2 is involved in antigenicity and viral pathogenesis [72]. The E3 peptide is necessary for protein heterodimerization, and the deletion of the 6K peptide in Ross River Virus genome reduces pathogenicity and viral titer in mice [73]. The interactions between the structural proteins are also indispensable for virion integrity and virus assembly [32,74].
The common symptoms of ONNV in humans are fever, rash, headache, polyarthritis-like illness, and back pains, but the infection is often asymptomatic. An ONNV epidemic in Uganda, Tanzania, Kenya, Malawi, Senegal, Democratic Republic of Congo infected approximately 2 million people, but no fatal cases were reported [7,18]. Another outbreak of ONNV was reported in Uganda in 1996, with morbidity rates of 45-65% in some villages. ONNV was detected by serologic test and quantitative RT-PCR in 26% of Liberian refugees tested in 2003 [8,9]. The virus was detected in a pool of wild Anopheles funestus and An. gambiae collected in Uganda and Kisumu, Kenya by inoculation of a suspension into albino Swiss mice, and independently also from patient sera [8,18,50]. Antibodies against ONNV found in the sera of inoculated mice suggested that the virus was pathogenic.
The important African malaria vectors, An. funestus and An. gambiae, are also able to transmit ONNV to mice [75,76]. The RNA interference (RNAi) pathway is necessary for protection of An. gambiae against ONNV infection [27]. However, the RNAi pathway displays ONNV antiviral activity in An. gambiae only during the disseminated systemic infection, and not in the primary midgut infection by bloodmeal [26]. The immune pathways Janus kinases/Signal Transducer and Activator of Transcription (JAK/STAT) and Immune Deficiency (Imd) display a reciprocal effect, as ONNV antiviral mechanisms in the primary midgut infection but not against the disseminated infection in the hemocoel.
Venezuelan equine encephalitis virus (VEEV) is an arbovirus in the Americas, where numerous outbreaks have been reported with equine and human deaths [77]. A large epidemic of VEEV in Colombia in 1995 caused more than 70,000 human cases and 300 deaths [78]. VEEV was isolated from two pools of Anopheles p. pseudopunctipennis in Mexico in 1972, was cultured in C6/36 cells, and an infectious clone was generated [60]. Western equine encephalitis virus (WEEV) is a recombinant virus that can infect humans and other vertebrates, with a capsid protein related to Eastern equine encephalitis virus (EEEV) whereas the glycoprotein sequences are closer to Sindbis virus [79]. An epizootic occurred in Argentina in 1982, and more than 150,000 mosquitoes of different genera were collected by Centers for Disease Control (CDC) light traps between 1982 and 1983 [63]. From these mosquitoes, WEEV was isolated from a pool of An. albitarsis inoculated on Vero cells [63].
An. albimanus mosquitoes were competent for infection with Sindbis virus (SINV) by feeding on infected rabbit blood, displaying an infection prevalence of 64%, as well as the ability to transmit the virus to baby chicks [80]. SINV was also isolated from a pool of Anopheles spp. collected in China in 1990 and was used to generate an infectious clone [53]. SINV was first isolated in Sindbis village in Egypt in 1952 in Culex spp., but Anopheles mosquitoes seem also able to transmit this virus [80].
Eilat virus (EILV) was isolated from a pool of An. coustani mosquitoes and displays inability to replicate in the vertebrate cellular environment [13,43]. EILV was transmitted by bloodfeeding to An. gambiae, C. quinquefasciatus, and Ae. aegypti but not to Ae. albopictus, and may be vertically transmitted in the infected species, but the virus was not detected in the ovaries [13]. EILV did not display cytopathic effect in Aedes C6/36 and or C7/10 cells despite high replication at 12 h post infection. One proposed hypothesis is that EILV may have secondarily lost the ability to infect vertebrate cells, rather than being insect-specific as an ancestral character. Members of the Flavivirus genus in the family Flaviviridae are generally thought to be transmitted by culicine mosquitoes. However, a number studies have detected flaviviruses in Anopheles species, suggesting that anophelines could also be involved in transmission [23]. Flaviviruses are characterized by an enveloped virion carrying a single-stranded RNA genome of positive polarity. The linear genome of 10 to 11 kb is flanked by 5 and 3 untranslated regions (UTR) that encode for a single open-reading frame. Translation produces a single polyprotein cleaved in 10 proteins: three structural (C, prM, E), and seven non-structural (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) proteins [83].
The main reported vectors of West Nile virus (WNV) are Culex spp. Transmission is mainly zoonotic but with significant levels of human infection, including occasional mortality [84]. A pool of An. pauliani wild unfed females was found to be positive for WNV RNA by RT-PCR in Madagascar [23]. A pool of An. maculipennis collected in Italy between 2008 and 2012 was also found positive for WNV RNA by RT-PCR [61]. A pool of An. maculipennis collected in Serbia after the 2012 WNV outbreak was also positive for WNV RNA by RT-PCR, which suggested a potential important role for this species in transmission [62].
Japanese encephalitis virus (JEV) is an endemic causative agent of encephalitis in Asia and India. JEV infects humans and other vertebrates such as horses, dogs, and reptiles. Based on envelope protein gene sequences, JEV strains can be classified into five different genotypes, I-V, with specific geographic distributions [85]. Due to the availability of a vaccine, outbreaks of JEV are rare. JEV was detected and isolated from a pool of Anopheles peditaeniatus females collected from 1985-1987 in India [45]. JEV was isolated from C6/36 cells inoculated with extract from pools of Anopheles sinensis captured by carbon dioxide traps or sweep nets in Taiwan from 2005 to 2012, for a calculated infection rate of 2.3 per 1000 mosquitoes [21]. JEV genotype I was isolated from host-seeking An. sinensis collected by aspirator in Japan [46]. An. sinensis is a dominant human malaria vector in Asia [2,86].
Wesselsbron virus (WSLV) shares ecological niches and similar livestock symptoms with Rift Valley Fever virus and misdiagnosis is common [34,87]. WSLV RNA was detected using molecular diagnostic tools in Kenya from An. coustani and in humans in Senegal [34,87]. Aedes mosquitoes are the presumed main vector of WSLV but the involvement of Anopheles mosquitoes remains to be elucidated.
Anopheles flavivirus (AnFV), the first flavivirus discovered in Anopheles, was identified in a population virome survey in Liberia and Senegal [32]. An RT-PCR diagnostic assay confirmed the presence of AnFV RNA in wild Anopheles, with a prevalence of 12%. Anopheles gambiae flavivirus (AngFV) and An. squamosus flavivirus (AnsFV) were detected in An. gambiae and An. squamosus respectively by a nucleic acid melting-curve analysis from mosquitoes collected in Kenya [34]. AngFVs and AnsFV share 77% of nucleotide identity.
Stratford virus (STRV) RNA was detected in an isolate of Anopheles annulipes collected in Australia from 1995-2013 [54]. Sequence analysis based on part of the NS5 gene sequence displayed 95-99% homology among the different isolates of STRV, and 2% divergence was detected between the isolates collected in 1995-2013 and the first isolates of 1961. The main vector of STRV is thought to be Aedes spp., without important contribution by Culex spp. [54].
A study of Anopheles samples collected in Australia identified multiple insect-specific viruses that do not infect vertebrate cells, and display fine species-specific host restriction for the Anopheles host in which they were identified [42]. The genome of Karumba virus (KRBV) was obtained from single and pooled mosquitoes collected in two different sites in Australia. In addition, the same authors also discovered Haslams Creek virus (HaCV), Dairy Swamp virus (DSwV), and Mac Peak virus (McPV) in pools of An. annulipes, An. bancrofti, and An. farauti, respectively [42]. HaCV, DSwV, and McPV failed to replicate in vitro on cell lines C6/36 Aedes albopictus, MOS55 An. gambiae, ISE6 tick, or S2 Drosophila melanogaster. No replication of KRBV was observed on the following vertebrate cells: BSR Mesocricetus auratus, Vero Cercopithecus aethiops, or DF-1 Gallus gallus [42]. The complete genome sequences were assembled by deep sequencing of mosquito RNA. Viral sequences were detected in the 21-nucleotide small RNA fraction, which suggested that double stranded RNA (dsRNA) intermediates produced during viral replication were cleaved by the siRNA pathway into viral RNAs (viRNAs). Presence of viRNAs is evidence of active virus replication. The prevalence of KRBV found in Anopheles meraukensis was 91.7% and 100% in Wyndham and Karumba, respectively. The prevalence of KRBV in wild type mosquitoes is quite high and highlights the need to study its impact on Anopheles species. Specific RT-PCR assays detected HaCV and McPV RNA in wild An. annulipes, An. bancrofti, and An. farauti.
Long Pine key virus (LPKV) was isolated in the United States from a pool of 50 An. crucians, and Kampung Karu virus (KPKV) was isolated from a single An. tesselatus in Malaysia [47]. Both viruses were cultured on C6/36 cells and cytopathic effects were observed at 7 d post-inoculation. Inoculation of BHK-21 and Vero cells with LPKV and KPKV produced neither replication nor cytopathic effects, and these viruses did not display pathology in mice [47]. Hemagglutination-inhibition tests for KPKV were not possible due to absence of reactive hemagglutinin. Despite being insect-specific, LPKV and KPKV reacted serologically with antibodies directed against some dual-host flaviviruses such as WNV, JEV, and Dengue virus [47].

Phlebovirus: Rift Valley Fever Virus (RVFV)
Rift Valley fever virus (RVFV) belongs to the Phlebovirus genus in the family Phenuiviridae. RVFV is present throughout Africa and the Middle East and causes important economic losses in livestock [6,25,88]. During an epidemic in 2012, human cases and deaths occurred in Mauritania [6]. Culex and Aedes spp. are proven vectors of this virus, but An. arabiensis, An. coustani, An. rufipes, An. pharoensis, An. rhodesiensis, and An. christyi have also been implicated in transmission during epizootics and epidemics, as well as in maintenance by vertical transmission [25]. At least 14 species of the genus Orthobunyavirus in the family Peribunyaviridae have been detected in Anopheles species. The orthobunyaviruses are characterized by a single-stranded RNA genome of negative polarity composed of three segments that are large (L), medium (M), and small (S) encoding the RNA dependent RNA polymerase, the glycoproteins (Gn and Gc) and the nucleoprotein, respectively [49].
Leanyer virus (LEAV) was isolated from An. meraukensis pools in a suburb of Darwin, Australia in 1974, and was cultivable on BHK-2I and Vero cells [48]. Peptide sequence analysis indicated that LEAV is related to Oropouche virus with 59% similarity in the polymerase sequence. LEAV does not cross-react serologically with other orthobunyaviruses and the L and S segment peptide sequences display divergence from other orthobunyaviruses, and therefore LEAV could be considered as a new antigenic complex [89].
Batai virus (BATV) was first isolated from Culex gelidus in Malaysia [90]. More recently, two different entomological surveys in Germany and Italy identified and isolated BATV from Anopheles maculipennis complex mosquitoes [20,37]. The Italian and German strains were related to strains isolated in Slovakia [37]. BATV causes hemorrhagic disease, with fever, headache, nausea, and vomiting. BATV was also identified in Sudan from sera of febrile patients [90]. Cattle are a potential host of BATV, and in a survey in Germany, the serological prevalence in cattle was 0.55%, while a 2.1% positivity rate was detected by RT-PCR in cattle from Mongolia [91,92]. The An. maculipennis complex is an important historical European vector of P. vivax, and a current vector in Europe and Asia [93].
Recombination events between BATV and Bunyamwera virus generated Ngari virus (NRIV), which has mainly been reported in Africa. The first isolation of NRIV was done from Ae. simpsoni in Senegal in 1979. It was also isolated from An. gambiae and An. pharoensis, and cytopathic effects were observed in Vero cells [36]. Pools of An. funestus collected in Kenya between 2007 and 2012 were positive for NRIV RNA by RT-PCR and sequencing. The full genome sequence of NRIV was obtained from mosquito and human samples [94].
Bangui virus (BGIV), a probable arbovirus of this family, was first isolated from humans in 1973 in the Central African Republic, where it was also detected in An. pharoensis [36]. BGIV produces cytopathic effects on Vero and amphibian Xenopus cells, and is sensitive to ether and acid pH [95]. Serological tests are used for the detection of BGIV, but molecular tools are lacking.
Cache Valley virus (CVV), in vector competence assays, was more infectious to An. quadrimaculatus than to Coquillettidia perturbans. At 16 d post-infection, infection prevalence was greater than 90% for both species [39]. Human and other vertebrate cases of CVV have been reported, including from a woman diagnosed with aseptic meningitis [96]. CVV can be cultured on many vertebrate cells such as Buffalo green monkey kidney (BGMK), human colon adenocarcinoma (CaCo2), human lung carcinoma (A549), and Vero cells.
Mapputta virus (MAPV) was isolated in 1960 from An. meraukensis in Australia. MAPV antibodies react with other viruses of Mapputta group such as Maprik virus (MPKV), Trubanaman virus (TRUV), and Gan Gan virus (GGV), indicating that a serological test is not sufficient to distinguish them [49]. There is at least 60% nucleotide identity of S and L segments between MAPV and MPKV [49]. An. meraukensis bites humans and other vertebrates, but it has not been incriminated as a malaria vector [97,98]. Mapputta virus can be cultured on hamster kidney BHK-21 cells, which display cytopathic effects 4 d post-inoculation [49].
Tahyna virus (TAHV) was isolated from pools of Anopheles hyrcanus females collected in South Moravia [55], although it has been more often found in Aedes spp. An. hyrcanus extract was inoculated intracerebrally into mice and the virus was identified by neutralization tests on Vero E6 cells and confirmation by RT-PCR. The virus was generally fatal to inoculated mice. TAHV infection causes human fever, conjunctivitis, pharyngitis, malaise, arthralgia, headache, and drowsiness, and anti-TAHV IgM antibodies were detected by IFA in asymptomatic patients in China, but no human deaths have been attributed to this virus [55,99,100]. The An. hyrcanus group is a widespread species group involved in the transmission of P. vivax and P. falciparum in Europe and Asia [101,102].
Tataguine virus (TATV) takes its name from the village in Senegal where it was first isolated in 1962 from Anopheles and Culex species [56]. In 1966, it was also isolated in Cameroon from An. gambiae and from serum of a 14-year-old boy. The patient presented with fever, exanthema, asthenia, muscle aches, and neutropenia, and TATV was confirmed by inoculation in mice [56]. TATV was widespread in African countries surveyed from 1960 to 1970, including Nigeria, South Africa, and Ethiopia [103,104].
Nyando Virus (NDV) was isolated by inoculation of mice with extract of An. funestus collected in Kenya during the ONNV outbreak of 1959-1960. [18,50]. Cytopathic effects of NDV were observed on Vero E6 and RE05 cells but not on C6/36. NDV displays at least 90% nucleotide and peptide identity with Bwamba virus (BWAV) and Pongola virus (PGAV) [105]. NDV causes moderate to severe febrile disease in humans, and human exposure was detected in serological surveys in Kenya, Central African Republic, and Uganda [18,19].
Ilesha virus (ILEV) was isolated from a pool of An. gambiae collected in the Central African Republic [44]. ILEV was recovered in 1990 from the blood of a woman who died with fever, anemia, leucopenia, and coagulative disorders. The virus was isolated after inoculation of mosquito extract into suckling mice and was cultured on Vero E6 and AP61 cells [106].
Bwamba virus (BWAV) was isolated from extract of An. funestus in Uganda and from a human blood sample from a refugee camp in Tanzania [38]. BWAV appears to be a widespread human infection in Africa, with short duration symptoms including fever, headache, exanthema, arthralgia, body rash, and diarrhea [44,107].
Anopheles A virus (ANAV) and Anopheles B virus (ANBV) were first isolated from female An. boliviensis in Colombia. Mice injected with ANAV and ANBV displayed central nervous system pathologies [30], although these viruses have not been isolated from naturally-infected vertebrates. An. boliviensis is a minor malaria vector in Colombia [108]. ANAV and ANBV are distinct from other Bunyamwera viruses because the S segment encodes only the nucleocapsid protein N, and therefore the nonstructural protein (NSs) involved in replication and pathogenesis is absent [109][110][111].
Tensaw virus (TENV) from Tensaw River in the southeastern United States was isolated from An. crucians and An. quadrimaculatus [57]. Humans, dogs, raccoons and cows were positive for TENV by serological tests, indicating that TENV was transmitted from mosquitoes to vertebrates. An. quadrimaculatus and An. albimanus remained infective from 2-14 d post-infection [57].

Dicistrovirus: Anopheles C Virus (AnCV) and Anopheles Associated C Virus (AACV)
Two Anopheles dicistroviruses were identified by deep sequencing and de novo assembly. Anopheles associated C virus (AACV) [16] and Anopheles C virus (AnCV) [5] belong to the genus Cripavirus in the Dicistroviridae family, non-enveloped viruses with a single stranded RNA genome of positive polarity. The dicistrovirus genome is comprised of two open reading frames (ORF). ORF1 encodes the non-structural proteins necessary for virus replication, and ORF2 the viral proteins VPO to VP4. In the C-terminal region of ORF1, there is an RNA dependent RNA polymerase (RdRp) domain that is highly conserved among dicistroviruses [112]. Dicistroviruses are only known to infect insects.
AnCV is distinct from but related to Drosophila C virus, and was discovered in wild-caught human host-seeking Anopheles in Senegal, and in laboratory colonies of An. coluzzii in France and An. dirus in Cambodia [5]. Both horizontal and vertical modes of transmission were demonstrated in the Ngousso colony of An. coluzzii, while a colony of Anopheles stephensi maintained in the same laboratory as the An. coluzzii colony was not infected by AnCV [5].
AACV was discovered in Anopheles maculipennis collected in the Camargue region of France. AACV displays approximately 20% peptide sequence divergence from chronic bee paralysis virus (CBPV) in examined regions of the genome [16].

Cypovirus: Anopheles Cypovirus (AnCPV)
Viruses in the genus Cypovirus, family Reoviridae, are characterized by a genome of 9 to 12 segments of linear dsRNA within a single capsid shell [113]. Cypoviruses, or cytoplasmic polyhedrosis viruses, are so named because the infectious forms are occlusion bodies in crystalline polyhedral form within the cytoplasm of infected cells. The protein polyhedra protect the virus against harsh environmental conditions such as high pH. Cypoviruses are only known to infect insects. The first Anopheles cypovirus was detected in adult An. stephensi by microscopy after staining with ammonium molybdate [41]. However, culture and transmission experiments did not succeed, the virus disappeared from the colony after nine months, and was not sequenced or named. This cypovirus was observed within the cytoplasm of Plasmodium berghei or P. yoelii rodent malaria oocysts in co-infected mosquitoes, and observations suggested it might have reduced the numbers of developing oocysts.
More recently, Anopheles cypovirus (AnCPV) was discovered in wild-caught human host-seeking Anopheles in Senegal and Cambodia, and was also present in the Ngousso laboratory colony of An. coluzzii [5]. AnCPV infection was absent in an An. stephensi colony maintained in the same laboratory. Viruses of the Orbivirus genus, in the Reoviridae family, infect plants and vertebrates, and are characterized by a dsRNA genome of 10-12 segments. Tibet orbivirus (TIBOV) was isolated from Anopheles maculatus collected in China [58]. TIBOV was cultured from mosquito extract inoculated on C6/36 and BHK-21 cells. Cytopathic effects characterized by cell rounding, lysis, and floating cells were observed only in the BHK-21 cells after 3 d of infection, but viral RNA was detected by RT-PCR in both cell types. Gel electrophoresis revealed ten dsRNA genome segments, which were sequenced [58]. An. maculatus species are important vectors of human malaria in Asia [114].
Orbivirus sequences detected in Anopheles annulipes and An. hinesorum from Australia were named Anopheles annulipes orbivirus (AAOV) and Anopheles hinesorum orbivirus (AHOV) [31]. The viruses could not be cultured on C6/36 cells, but virus sequences were present in the mosquito 21-nucleotide viral RNA fraction, indicative of active replication and dicing of dsRNA replication intermediates.
Tilligerry virus (TILV) was isolated in 1971 from An. annulipes in Australia, and the complete genome sequence was determined [59]. The G + C content of the full genome of TILV is 45% and its 10 segments are visible on agarose and acrylamide gels [59,115]. TILV leads to cytopathic effects in BHK and BSR cells 2-3 d post inoculation [59]. TILV cross-reacts with bluetongue virus in complement fixation tests [115].
Orungo Virus was isolated in Uganda from An. funestus after inoculation on Vero and BHK-21 cells [51]. Viral replication was detected in the brains of inoculated mice and hamsters. Antibodies against ORUV were detected in human sera from Nigeria. The human symptoms of ORUV are fever, headache, myalgia, nausea, and vomiting, and ORUV also infects other vertebrates such as sheep, monkeys and cows [116]. On the basis of nucleotide G + C content and amino acid composition of the T2 protein, ORUV appears closer to Culicoides-borne than mosquito-borne orbiviruses [51]. Coot Bay virus in the Almendravirus genus belongs to the Rhabdoviridae family in Mononegavirales order [40]. CBV was isolated from An. quadrimaculatus mosquitoes collected in 2013 in Florida, USA [40]. CBV could be cultured in Aedes C6/36 cells, but not in mammalian BHK-21 and Vero cells [40]. CBV does not cause apparent illness or deaths in suckling mice. The virion has a diameter of about 50 nm [40].

Discussion
Little is known about the Anopheles virome, as evidenced by the relatively small number of scientific publications on the topic. Nevertheless, when the literature summarized here is taken together, it is evident that Anopheles viruses, including among them pathogens of humans and other vertebrates, are abundant in nature but understudied.
We found published evidence of at least 51 viruses associated with Anopheles. This number is likely an underestimate, because it does not include publications in journals not indexed by the databases searched. The Anopheles virome appears to be dominated by RNA viruses. RNA viruses are also dominant in Aedes (Yellow fever virus, Zika virus, dengue virus, chikungunya virus) and Culex (Eastern equine encephalitis virus, Rift Valley fever virus, West Nile virus). RNA viruses evolve rapidly because of high mutation and recombination rates [119], and can potentially adapt rapidly to new hosts. There could also be an ascertainment bias favoring the detection of RNA as compared to DNA viruses, because sequencing of small viral-derived RNAs is a powerful tool to identify actively replicating RNA viruses [5,16,17].
It is likely that the main evolutionary pressure shaping mosquito antiviral mechanisms in general is their persistent exposure in nature to members of the natural virome, rather than the probably less frequent exposure to vertebrate-pathogenic arboviruses. Despite the apparently abundant presence of viruses in Anopheles, there is debate as to whether Anopheles mosquitoes serve as merely occasional hosts of pathogenic arboviruses, or to what extent they help mediate transmission as vectors. The distinction between host and vector will require evaluation of Anopheles vector competence in the laboratory and Anopheles vectorial capacity in the field. Vector competence is the ability to acquire, maintain, disseminate, and transmit a virus, whereas vectorial capacity or vector efficiency is the rate at which a putative vector population generates new inoculations from an infectious case.
The potential of the majority of Anopheles-associated viruses for transmission to humans or other vertebrates is currently unknown, because few studies of host range and transmission have been done. Some viruses may have a host range restricted to only Anopheles and other insects. For example, Anopheles cypovirus and Anopheles C virus were found to replicate and be maintained by vertical transmission in An. coluzzii, but were not able to infect Ae. aegypti in exposure experiments [5]. Both of these viruses were able to replicate in An. stephensi after exposure, but Anopheles C virus was not stably maintained and disappeared after several generations. Thus, these two viruses may be Anopheles-specific, but possibly not adapted to all Anopheles species.
A first group of viruses display either known or potential restriction of infection to Anopheles or insect cells. Myxoma virus can be mechanically transmitted by mosquitoes and fleas but does not replicate in them, and thus is not an arbovirus. Of these, transmission by Anopheles has been demonstrated for ONNV, but additional work is required to determine the vector competence and capacity of Anopheles for the other arboviruses.
The above grouping of viruses is likely to be porous and is expected to change with the addition of data from new studies. The interesting question remains, nevertheless, whether Anopheles are less efficient arbovirus vectors than Aedes and Culex, or are simply under-recognized as virus vectors. If the first case were true, that is, lower vector competence of Anopheles for arboviruses, then understanding the biological mechanisms leading to their general resistance to virus transmission would be important and could lead to novel tools to control arbovirus transmission by Aedes and Culex vectors. As the current systematic review indicates, Anopheles are not inherently resistant to virus replication. Although the natural virome data do not yet exist to make a numerical comparison with Aedes and Culex, there is no evidence that rates of natural carriage of viruses are substantially different between these mosquito genera.
Genetically encoded differences between mosquito species can interact with viral factors to influence host permissiveness and restriction. The protein nsP3 of ONNV influences host specificity of this virus to Anopheles as compared to Aedes, because substitution of chikungunya virus nsP3 by ONNV nsP3 in the chikungunya backbone allows chikungunya infection of An. gambiae [70]. In addition to interaction between viral factors and host cell proteins, differences in the small RNA regulatory pathways such as microRNAs and piwi-RNAs between Culex, Aedes, and Anopheles may also play a role in restricting host range [120].
The incidence of co-infection of malaria and arboviruses is probably underestimated in endemic areas. In a recent survey in Senegal, the frequency of human co-infection by P. falciparum malaria and arboviruses (dengue, yellow fever, Zika, chikungunya, Rift Valley fever) was close to 50% [121]. It is unknown whether members of the natural virome can influence malaria parasite or arbovirus infection and transmission by the vector, for example by stimulating or diverting mosquito basal immunity, or otherwise either promoting or diminishing superinfection.
One conclusion of this study is that the databases PubMed, Web of Science, Scopus, and Lissa can yield different results for the same specific area of research. The four bibliographic databases are complementary. PubMed and Lissa are free to access without registration, which is not the case for Scopus and Web of Science. Web of Science yielded the most results (58 articles), because it includes articles from many countries. However, PubMed contains the most recent articles. In contrast, the articles found in Lissa are in French and include older works, which can fill gaps not covered by other databases. The Arbocat Arbovirus Catalog indexes arboviruses or potential arboviruses harbored by arthropods, but for some of these viruses, the support is unpublished or no longer available.
Finally, it should be noted that a systematic automated search profile can only reveal reports that are indexed using informative combinations of terms. A limiting factor is thus the quality of index terms in the records. The search parameters presented here could perhaps be modified with additional systematic terms. However, search terms that might appear to be more precise may in fact generate less informative results and lead to diminishing returns. For example, searching of databases with [virus name + Anopheles], can generate large numbers of often low-quality results (e.g., [West Nile virus] + Anopheles generates 140 records). Such searches require increased levels of labor-intensive manual curation to identify on-target records. Systematic automated searches can be easily re-run. Moreover, the most important biological observation is made when a virus is first reliably reported in at least one Anopheles species. After that, finding the virus in other Anopheles is useful, but hardly surprising, while studies of the biology of host range restriction and mode of transmission of Anopheles viruses are sorely lacking.

Conclusions
This study is the first, to our knowledge, to present an overview of the published literature on the Anopheles virome. At least 51 viruses have been reported in Anopheles in almost all continents. The quantification and identification of the Anopheles virome is important for general understanding of microbiome diversity, for surveillance and prevention of emergence of unknown viruses, to understand the phenomenon of human malaria and arbovirus co-infection, and to study the antiviral and immune responses of Anopheles mosquitoes. The availability of next-generation sequencing and de novo assembly will likely continue to augment knowledge of Anopheles viruses, and more effort will be required to characterize their biology and public health risk.