An elusive endosymbiont: Does Wolbachia occur naturally in Aedes aegypti?

Abstract Wolbachia are maternally inherited endosymbiotic bacteria found within many insect species. Aedes mosquitoes experimentally infected with Wolbachia are being released into the field for Aedes‐borne disease control. These Wolbachia infections induce cytoplasmic incompatibility which is used to suppress populations through incompatible matings or replace populations through the reproductive advantage provided by this mechanism. However, the presence of naturally occurring Wolbachia in target populations could interfere with both population replacement and suppression programs depending on the compatibility patterns between strains. Aedes aegypti were thought to not harbor Wolbachia naturally but several recent studies have detected Wolbachia in natural populations of this mosquito. We therefore review the evidence for natural Wolbachia infections in A. aegypti to date and discuss limitations of these studies. We draw on research from other mosquito species to outline the potential implications of natural Wolbachia infections in A. aegypti for disease control. To validate previous reports, we obtained a laboratory population of A. aegypti from New Mexico, USA, that harbors a natural Wolbachia infection, and we conducted field surveys in Kuala Lumpur, Malaysia, where a natural Wolbachia infection has also been reported. However, we were unable to detect Wolbachia in both the laboratory and field populations. Because the presence of naturally occurring Wolbachia in A. aegypti could have profound implications for Wolbachia‐based disease control programs, it is important to continue to accurately assess the Wolbachia status of target Aedes populations.

Insects that are not naturally infected with Wolbachia may be amenable to infection experimentally. Novel Wolbachia infections have been generated through microinjection, where cytoplasm or purified Wolbachia from an infected donor is transferred to an uninfected embryo (Hughes & Rasgon, 2014). Deliberate transfers of Wolbachia between species are challenging and can take thousands of attempts to generate a stable line (McMeniman et al., 2009;Walker et al., 2011). But once an infection is introduced, Wolbachia infections have applications for pest and disease vector control since they can alter host reproduction and block virus replication and transmission (Hoffmann, Ross, & Rašić, 2015).

| RELE A S E S OF NOVEL WOLBACH IA INFEC TI ON S FOR VEC TOR AND D IS E A S E CONTROL
There is increasing interest in deploying mosquitoes with experimentally generated Wolbachia infections into the field for disease control. Over 25 novel Wolbachia infection types have been generated in mosquitoes through embryonic microinjection, mainly in the principal dengue vectors Aedes aegypti and Aedes albopictus (Ross, Turelli, & Hoffmann, 2019). Most of these infections induce cytoplasmic incompatibility, and many also reduce the ability of their hosts to transmit viruses, making them desirable for field release. For mosquito species that are naturally Wolbachia-infected such as A. albopictus, novel infections can be generated either by first removing the natural infections with antibiotics (Calvitti, Moretti, Lampazzi, Bellini, & Dobson, 2010;Suh, Mercer, Fu, & Dobson, 2009) or by introducing the novel infection into an infected mosquito, resulting in a superinfection (Suh, Fu, Mercer, & Dobson, 2016;Zhang, Zheng, Xi, Bourtzis, & Gilles, 2015). Different novel Wolbachia infections may be incompatible with each other (Ant, Herd, Geoghegan, Hoffmann, & Sinkins, 2018) and the addition of Wolbachia strains to create superinfections can lead to unidirectional incompatibility, where females of the superinfected strain produce viable offspring following matings with males with any infection type, but superinfected males induce cytoplasmic incompatibility when mated with singly infected and uninfected females (Joubert et al., 2016).
Mosquitoes with novel Wolbachia infections are being released into the field for two main purposes: population replacement and population suppression. The objective of the former approach is to replace natural populations with mosquitoes possessing Wolbachia infections that interfere with virus transmission. This is achieved through the release of males that induce cytoplasmic incompatibility and females that transmit the Wolbachia infection and have reduced vector competence (Walker et al., 2011). Successful population replacement of A. aegypti with novel Wolbachia infections has been achieved in several countries (Garcia et al., 2019;Hoffmann et al., 2011;Nazni et al., 2019). Following releases in Australia and Malaysia, Wolbachia infections have maintained a stable, high frequency in most locations, coinciding with reduced local dengue transmission (Nazni et al., 2019;O'Neill et al., 2018;Ryan et al., 2019).
Population suppression can be achieved through male-only releases of Wolbachia-infected males, resulting in cytoplasmic incompatibility with wild females. This was first demonstrated in 1967 in Cx. pipiens (Laven, 1967) by exploiting the natural variation in Wolbachia infection types between mosquitoes from different locations (Atyame et al., 2014). Other releases have used Wolbachia from a closely related species through introgression (O'Connor et al., 2012) and novel Wolbachia transinfections generated through microinjection (Mains, Brelsfoard, Rose, & Dobson, 2016;Zheng et al., 2019).
Both population replacement and suppression approaches rely on the novel Wolbachia infection types inducing cytoplasmic incompatibility with the resident mosquito population. Thus, the presence of natural Wolbachia infections in mosquitoes may interfere with disease control programs, making population replacement or suppression challenging or even impossible.  Hoffmann et al. (2011)). Until recently, A. aegypti was not thought to harbor Wolbachia naturally (Kittayapong, Baisley, Baimai, & O'Neill, 2000), though it is clearly amenable to infection given the number of stable experimental infections generated in this species (Ross, Turelli, et al., 2019). Evidence for horizontal gene transfer between Wolbachia and A. aegypti may reflect a historical infection (Klasson, Kambris, Cook, Walker, & Sinkins, 2009).

| DE TEC TI ON S OF WOLBACH IA IN A ED E S AEGYPTI
The most comprehensive survey to date found no evidence for  (Table 1). These studies report variable infection frequencies in populations and identify infections from several Wolbachia supergroups. Most studies found that the infections detected were closely related to or identical to the wAlbB infection that occurs natively in Aedes albopictus (Balaji et al., 2019;Carvajal et al., 2019;Coon et al., 2016;Kulkarni et al., 2019), while other studies also detected Wolbachia from supergroups that do not normally occur within Diptera (Carvajal et al., 2019;Thongsripong et al., 2018). Most evidence is limited to molecular detection, and not all studies claim to have discovered an active infection. However, some studies have established laboratory colonies and reported maternal transmission of Wolbachia (Kulkarni et al., 2019) or the loss of infection through antibiotic treatment (Balaji et al., 2019).

Similar to A. aegypti, Anopheles mosquitoes (which transmit
Plasmodium parasites that cause malaria) were also thought to be uninfected by Wolbachia, though several recent studies have detected Wolbachia in this genus (Ayala et al., 2019;Baldini et al., 2014;Jeffries et al., 2018). In a critical analysis of studies in Anopheles gambiae, Chrostek and Gerth (2019) assert that the evidence is currently insufficient to diagnose natural infections in this species. We highlight similar issues with detections of Wolbachia in A. aegypti but also discuss the potential implications for disease control if Wolbachia do occur naturally in this species.

| P OTENTIAL IMPLI C ATI ON S OF NATUR AL WOLBACH IA INFEC TI ON S FOR RELE A S E S OF NOVEL INFEC TI ON S
The presence of natural Wolbachia infections may influence compatibility patterns between mosquitoes with the novel Wolbachia infection and the natural population. These patterns are summarized in Figure 1, although crossing patterns in nature are likely to be more complex. Natural Wolbachia infections can have heterogeneous densities and frequencies in populations (Calvitti, Marini, Desiderio, Puggioli, & Moretti, 2015), making compatibility patterns hard to predict. Crosses may differ in the strength of incompatibility in different directions Paterson, 1992, Joubert et al., 2016;Sinkins et al., 1995), and there are also environment-dependent effects on cytoplasmic incompatibility including adult age (Kittayapong, Mongkalangoon, Baimai, & O'Neill, 2002) and temperature (Ross, Ritchie, Axford, & Hoffmann, 2019). The presence of Wolbachia superinfections also increases the number of potential compatibility patterns (Dobson, Rattanadechakul, & Marsland, 2004). The presence of natural Wolbachia infections in a population may result in crossing patterns that make population replacement or suppression more challenging (Figure 1b-e). The following scenarios assume that the natural infection is at fixation in the population, though infections may be at intermediate frequencies (Table 1) so any impacts on Wolbachia release programs will be weaker. When novel and natural infections are compatible with each other (no reduction in egg hatch in any combination), invasion will depend on the relative fitness of each infection type due to a lack of cytoplasmic incompatibility ( Figure 1b). Since transinfections in mosquitoes typically impose fitness costs while natural infections tend to be beneficial (Ross, Turelli, et al., 2019), population replacement may be unachievable even if high frequencies are reached. In this situation, population suppression is impossible due to the lack of cytoplasmic incompatibility in any direction.

With most novel infections generated in
Such patterns occur in Cx. pipiens, with multiple compatible strains coexisting within natural populations (Atyame et al., 2014;Duron, Raymond, & Weill, 2011).

Incompatibility between males of novel and natural infections
and females of the opposite infection type in both directions, or bidirectional cytoplasmic incompatibility, may occur ( Figure 1c).
Bidirectional incompatibility is desirable for population suppression programs because it reduces the risk that inadvertently released females will replace natural populations (Moretti, Marzo, Lampazzi, & Calvitti, 2018). Novel Wolbachia infections that are bidirectionally incompatible with natural populations have been generated in A.
When bidirectional incompatibility occurs, population replacement will be difficult to achieve unless high frequencies of the novel infection are reached. Where population replacement is successful, spread beyond the release area is unlikely since the frequency required for invasion is 50% when fitness is equal (O'Neill et al., 1997). Novel infections may instead persist with natural infections (Telschow, Yamamura, & Werren, 2005), particularly in fragmented populations (Keeling, Jiggins, & Read, 2003).
Unidirectional incompatibility may also occur between natural and novel infections (Figure 1d (Figure 1d). In this situation, population suppression is impossible and population replacement will be challenging; therefore, such infections are not being considered for release.
Natural populations of A. albopictus are superinfected with the wAlbA and wAlbB strains at a high frequency although either strain may occasionally be lost (Joanne et al., 2015;Kittayapong, Baisley, et al., 2002), resulting in unidirectional cytoplasmic incompatibility (Dobson et al., 2004). Unidirectional cytoplasmic incompatibility can also occur in crosses between two single Wolbachia infections (Figure 1d,e) as demonstrated in Cx. pipiens (Atyame et al., 2014;Bonneau et al., 2018). In this situation, both strains induce cytoplasmic incompatibility, but one lacks the ability to restore compatibility with males of the other infection.
Cytoplasmic incompatibility induction by males is governed by two genes while the ability to restore compatibility by females is governed by a single gene (Shropshire, On, Layton, Zhou, & Bordenstein, 2018); the two phenotypes are therefore not always linked.
Although natural infections may interfere with releases of novel infections, their presence may also provide opportunities for disease control. Wolbachia infections that cause cytoplasmic incompatibility can be released in other locations for population suppression without the need for novel infections (Laven, 1967). Natural infections may also be useful for population replacement if they can block virus transmission (Glaser & Meola, 2010;Mousson et al., 2012), but like releases of novel infections, it will be important to match the genetics of the release strain to the target population (Garcia et al., 2019).

| TE S TING A PUTATIVELY WOLBACH IA-INFEC TED L ABOR ATORY P OPUL ATION OF A . AEGYPTI
Of the eight studies reporting natural Wolbachia infections in A.
aegypti, only two established laboratory populations (Table 1). We aegypti (Axford et al., 2016) were included as negative and positive controls, respectively, in each assay. Through these two approaches, we did not detect any wAlbB infection in 120 mosquitoes (including larvae and adults from both generations) from the LC population (Supporting Information Appendix S2), demonstrating that the LC laboratory population is not infected with wAlbB.
To test whether the LC population harbors any Wolbachia infection, we performed additional assays with general Wolbachia primers. TaqMan probe assays were performed as described previously  (Mee et al., 2015), targeting the 16S rDNA (Supporting Information Appendix S1). We also performed conventional PCR with 16S rDNA and gatB primers following methods described by the authors of the original study (Kulkarni et al., 2019). Finally, LAMP assays were performed using our protocol (Jasper et al., 2019) but with primers used to diagnose Wolbachia infections by the original study (Kulkarni et al., 2019). From analyses of 72 individuals from both generations with the three molecular assays, zero were infected (Supporting Information Appendix S2). Negative and positive controls were confirmed in all assays. Through these analyses, we demonstrate conclusively that the LC population does not harbor Wolbachia. These results conflict with those from the original study (Kulkarni et al., 2019)

and more recent tests by the authors
where Wolbachia is at a high frequency (28/32, 87.5%) in the fourth laboratory generation (Jiannong Xu, personal communication).
Although the reason for this conflicting result is unclear, our study emphasizes the need for independent evaluation of Wolbachia infections in A. aegypti. Teo et al. (2017)

| LIMITATI ON S OF S TUD IE S TO DATE
Detections of Wolbachia in A. aegypti are accumulating (Table 1) (Dobson et al., 1999). Electron microscopy images show apparent localization of Wolbachia to the ovaries, but images are low resolution and there is no clear distinction between Wolbachia and organelles as in other recent studies (Leclercq et al., 2016;Li et al., 2017).

| E VIDEN CE REQU IRED TO CONFIRM NATUR AL WOLBACH IA INFEC TIONS
From the studies discussed above, we believe the evidence is currently insufficient to indicate that A. aegypti mosquitoes harbor a natural Wolbachia infection. We propose three lines of evidence as a minimum requirement for confirming a Wolbachia infection in this species: intracellular localization, maternal transmission, and removal of Wolbachia. Following molecular detection (traditional PCR, qPCR, or LAMP assays targeting the wsp gene should suffice), laboratory populations can be established from larvae, pupae, or adults from Wolbachia-positive locations to enable further characterization. Figure 2 shows a suggested approach for confirming natural Wolbachia infections in insects, following an initial field survey.

Intracellular localization is an important step in confirming a
Wolbachia infection because it will help to distinguish between an active infection and merely the detection of Wolbachia sequences from horizontal gene transfer or the environment. It can be demonstrated by visualizing Wolbachia within host tissues, such as through fluorescence in situ hybridization (FISH) (Moreira et al., 2009). These observations require appropriate controls including separate probes for Wolbachia and host, and visualization of tissues with the Wolbachia infection removed (see below). If FISH is not available, transmission electron microscopy (TEM) can also be used (Binnington & Hoffmann, 1989;Li et al., 2017;Yen & Barr, 1973) though Wolbachia needs to be carefully distinguished from host organelles. Quantitative PCR of separate host F I G U R E 2 Suggested procedure for confirming natural Wolbachia infections tissues can also help to confirm Wolbachia infections because they often exhibit tissue-specific localization.
Maternal inheritance of Wolbachia can be demonstrated through reciprocal crosses between Wolbachia-infected and uninfected mosquitoes, ideally with all four combinations (see Figure 1). The uninfected mosquitoes may be from a laboratory colony or collected from the field. In a true natural infection, only offspring from infected mothers are expected to test positive for Wolbachia. Paternal transmission may occur very rarely (Hoffmann & Turelli, 1988), and maternal transmission may be imperfect, particularly if the infection has a low density in the ovaries (Narita, Nomura, & Kageyama, 2007), so we recommend testing at least ten offspring from ten individual females from each cross to account for this. Other patterns of inheritance point against a Genome sequencing may provide insights into their origin. Finally, natural infections could be transferred to other species through microinjection to study their effects in novel hosts and provide further opportunities for disease control.
Although several studies have now claimed to detect Wolbachia in natural A. aegypti populations, the evidence is not compelling.
Studies to date have relied mostly on molecular approaches that may be prone to contamination. These results conflict with a growing body of evidence for a lack of infection in this species which includes a comprehensive global survey (Gloria-Soria et al., 2018), monitoring undertaken before releases of novel infections  and the data presented here. Our inability to detect Wolbachia in a putatively infected laboratory population demonstrates the need for more robust evidence when reporting natural Wolbachia infections. Although natural Wolbachia infections in A. aegypti may not exist, releases of novel Wolbachia infections are continuing to expand, and new target populations should therefore continue to be monitored prior to releases taking place.

ACK N OWLED G M ENTS
The authors thank Kelly Richardson for assistance with molecular diagnostics, Jiannong Xu and Aditi Kulkarni for providing the Aedes aegypti LC strain, and Nancy Endersby-Harshman for coordinating the import of the strain into our quarantine facility. We also thank the anonymous reviewers for their constructive feedback on the manuscript. AAH was supported by the National Health and Medical Research Council (1132412, 1118640, www.nhmrc.gov.au), the Australian Research Council (LE150100083, www.arc.gov.au), and the Wellcome Trust (108508, wellcome.ac.uk). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
PAR conceived the study, performed the live mosquito work, made the figures and tables, and drafted the manuscript; AGC, QY, and MJ performed the molecular diagnostics on the laboratory population; MAKA, ANA, and WAN conducted the field survey; AAH supervised and coordinated the project; and all authors contributed to writing and editing the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are contained within the manuscript and its appendices.