Identifying potential candidate Culicoides spp. for the study of interactions with Candidatus Cardinium hertigii

Culicoides biting midges (Diptera: Ceratopogonidae) are vectors responsible for the transmission of several viruses of veterinary importance. Previous screens of Culicoides have described the presence of the endosymbiont Candidatus Cardinium hertigii (Bacteroidetes). However, any impacts of this microbe on vectorial capacity, akin to those conferred by Wolbachia in mosquitoes, are yet to be uncovered and await a suitable system to study Cardinium‐midge interactions. To identify potential candidate species to investigate these interactions, accurate knowledge of the distribution of the endosymbiont within Culicoides populations is needed. We used conventional and nested PCR assays to screen Cardinium infection in 337 individuals of 25 Culicoides species from both Palearctic and Afrotropical regions. Infections were observed in several vector species including C. imicola and the Pulicaris complex (C. pulicaris, C. bysta, C. newsteadi and C. punctatus) with varying prevalence. Phylogenetic analysis based on the Gyrase B gene grouped all new isolates within ‘group C’ of the genus, a clade that has to date been exclusively described in Culicoides. Through a comparison of our results with previous screens, we suggest C. imicola and C. sonorensis represent good candidates for onward study of Cardinium‐midge interactions.

The microbiome of arthropods is known to modify host biology in a number of ways, ranging from nutritional provisioning to parasite protection. These effects are particularly pronounced in the case of endosymbionts (Su et al., 2013). For example, the endosymbiont Wolbachia (Rickettsiales: Anaplasmataceae) leads to an arbovirus blocking effect in Aedes aegypti mosquitoes, which has led to their successful deployment as a dengue fever control strategy in field trials (Nazni et al., 2019). The potential to harness viral blocking effects in wild populations is enabled by Wolbachia-induced cytoplasmic incompatibility (CI)-embryo death in mating between infected males and uninfected females. CI is exploited as a mechanism to drive the Wolbachia into a population to distribute this virus blocking phenotype into a naïve population; as Wolbachia pres-ence in a population reduces the fitness of uninfected females, the bacterium is able to reach high frequencies of infection. Importantly, this protective phenotype appears to be effective against a broad range of RNA viruses, supporting the potential use of endosymbiont-based biocontrol in major midge-borne pathogens such as BTV, SBV and AHSV.
The endosymbiont Candidatus Cardinium hertigii (Bacteroidetes) has been found to be widely associated with Culicoides (Morag et al., 2012;Mee et al., 2015;Pagès et al., 2017). Although the biological role of Cardinium in biting midges is undetermined, data from different host species demonstrate various Cardinium-dependent reproductive alterations including parthenogenesis-induction and CI (Zchori-Fein et al., 2001;Hunter et al., 2003), with the latter providing a potential drive mechanism for future introductions into Culicoides populations. Importantly, potential effects relating to pathogen-blocking and fitness (e.g., fecundity) remain underexplored.
A key requirement for studying Cardinium impacts in midges stems from the need for a model system. The main barrier to such a model is the difficult laboratory mass-rearing of midges, with only C. nubeculosus and C. sonorensis colonies currently in existence. Despite this, vector competence studies of field-caught Culicoides are still possible, such that identifying populations containing both Cardinium infected and uninfected individuals (for negative controls) is also of interest. In light of this, we investigated the distribution and prevalence of Cardinium infections in Culicoides populations from Palearctic and Afrotropical regions, using a conventional and nested PCR screen approach. These data are then added and compared to previous screen data to assess what we know about the distribution of Cardinium in field and laboratory populations and the suitability of specific midge species for further investigation pertaining to the endosymbiont.
Culicoides specimens were collected between 2007 and 2016 from sites spanning France, South Africa and the United Kingdom (Table 1). Details of Culicoides collection methods, storage and species identification can be found in Pilgrim et al. (2017). DNA extracts from a previous study by Lewis et al. (2014), describing positive Cardinium infection in C. pulicaris and C. punctatus from the United Kingdom, were used to validate screening assays.
Amplification of the host COI gene was initially assessed as a means of quality control by conventional PCR assay. COI amplicons from the Pulicaris complex were Sanger sequenced to determine cryptic species, which are difficult to distinguish morphologically. To this end, individuals were designated as belonging to a particular species if their COI haplotype had >99% sequence identity to previously designated DNA barcodes. DNA extracts, which passed quality control, were then screened for Cardinium using conventional and nested primers amplifying partial sequences of the Gyrase B (GyrB) gene (Primer sequences and PCR cycling conditions in File S1). Prevalence estimates were defined by the nested PCR data and categorised into no infection (0%), polymorphic infection (>0.1% and < 100%) and fixation (100%). Collection sites of Culicoides populations were then plotted against Cardinium infection prevalence based on the nested PCR assay and mapped geographically. For populations containing both males and females, Fisher's exact test (two-tailed) was used to investigate possible associations between endosymbiont and sex with a significance cut-off of P < 0.01.
The relatedness of Cardinium strains from different host species was analysed using nucleotide sequences of amplicons derived from conventional PCR. The Gyrase B gene (GyrB) was chosen for phylogenetic analysis because it has a higher level of divergence when compared to the conserved 16S rRNA gene, another gene used in Cardinium phylogeny reconstruction. Amplicons were purified enzymatically (ExoSAP) before being sequenced using a BigDye ® Terminator v3.1 kit (Thermo Scientific, Waltham, MA, U.S.A.), and capillary sequenced through both strands on a 3500 xL Genetic Analyser (Applied Biosystems, Austin, TX, U.S.A.). Sequences were aligned using the LINSI algorithm in MAFFT v7 (Katoh & Standley, 2013). A maximum likelihood (ML) phylogeny was then inferred with RAxML v8 (Stamatakis, 2014) using 1000 rapid bootstrap replicates and using the GTR + I + G model, which was selected with jModelTest 2 (Darriba et al., 2012) using the Akaike information criterion, with the topology search taking the best of Subtree Pruning and Nearest Neighbour Interchange rearranging.
Collection and screening of 337 specimens of 35 midge populations consisting of 25 species from both Palearctic and Afrotropical regions indicated varying prevalence of Cardinium between species (Table 1). All specimens were female apart from some males collected for C. imicola, C. punctatus, C. impunctatus and C. bolitinos (File S1). Most populations (23/35) did not show any signs of Cardinium infection, although low collection size in some cases tempers against a hard conclusion that infection is absent from the population. Of the remaining 12 populations, seven were at fixation for Cardinium (although a few of these had low sample sizes) with the other five showing signs of polymorphism, with a mix of infected and uninfected individuals. This prevalence range is similar to the study of Australian Culicoides by Mee et al. (2015), in which 3/26 positive species had a fixed infection, with the remainder being of intermediate prevalence. Notably, Mee's study (Mee et al., 2015) detected at least one Cardinium positive individual in each population screened, which is at odds to our findings and another study by Pagès et al. (2017), investigating Cardinium distribution in Culicoides from Spain.
It is possible that the apparent 'hotspot' in Australia is as a result of a more sensitive assay, as the qPCR method used by Mee et al. (2015) is sometimes preferred to nested PCR in screening. Alternatively, variation in thermal environment could explain the prevalence discrepancies observed between these bioclimatic zones. Morag et al. (2012) found a positive correlation between land surface temperature (LST) and Cardinium prevalence in Culicoides imicola in Israel, with higher prevalence at higher mean LST. Furthermore, there is a known relationship between Cardinium presence and latitude more globally, with Cardinium infection being more common for hosts near the equator (Charlesworth et al., 2019).
In order to investigate Cardinium-Culicoides interactions, candidate model midge species must be identified, which combine the ability to vector etiological agents with either lab culturability or a natural polymorphism in endosymbiont presence that permits comparison between infected and uninfected   individuals. Conventional PCR detected Cardinium in four putative vector species of bluetongue virus (BTV): C. imicola, C. newsteadi, C. bysta and C. punctatus (Table 1). Additionally, evidence of a low-level Cardinium infection was detected in one individual of the vector species C. pulicaris when screened with the nested assay. The infection patterns within the Pulicaris complex species (C. bysta, C. newsteadi, C. pulicaris and C. punctatus) suggest field populations could be suitable for future Cardinium work. However, the lack of polymorphic infection observed in some populations from this study, alongside the difficulties in species differentiation, could prove problematic. Furthermore, the utility of C. pulicaris as a candidate should be met with caution due to the conflicting reports of Cardinium prevalence within this species in past study (Lewis et al., 2014;Pagès et al., 2017). Notably, our more sensitive nested PCR assay detected a very low Cardinium frequency of 1/27 for C. pulicaris, compared to Lewis et al. (2014)   Phylogenetic analysis based on a 1300 bp region of the GyrB gene confirmed a monophyletic clade of Cardinium, grouping all Culicoides Cardinium isolates from this study in group C of the genus (Nakamura et al., 2009), in which all previously described sequences of Culicoides clustered (Fig. 1) (Genbank accession numbers [https://www.ncbi.nlm.nih.gov/genbank/]: LR877462-LR877465). The C. imicola GyrB sequence obtained from South Africa in this study (LR877465) was identical to both of those obtained from Kenya (KR026927) and Israel (JN166963), indicating a global infection. Likewise, the Cardinium sequences obtained from C. punctatus from Wolverhampton, U.K. (LR877464) had 100% identity to the sequence reported in the same species by Lewis et al. (2014;HG380244). Mee et al. (2015) have previously suggested that Cardinium strains group by geography; by contrast, we find a sporadic distribution of strains with respect to location. For example, the C. bysta and C. newsteadi strains (LR877462 and LR877463) identified from Corsica, group with strains from Israel and Japan, respectively (JN166964 and AB506792). Furthermore, no clear pattern of Cardinium distribution was observed between the three geographical regions of this study, though the small sample size for some species forbids any formal test of the presence or absence of heterogeneity (File S2).
By contrast with other endosymbiont-insect associations, the biological significance of Cardinium in Culicoides remains unknown and requires further research. Despite a likely vertical transmission route, our observed lack of sex-bias in infection observed in this study (C. punctatus; Fisher's exact, P = 0.31 and C. imicola; Fisher's exact, P = 0.68) corroborates previous studies (Morag et al., 2012;Mee et al., 2015;Pagès et al., 2017) and suggests the induction of parthenogenesis, feminisation or male-killing is unlikely to be associated with Cardinium. Other Cardinium strains have been implicated in cytoplasmic incompatibility (CI), including in the wasp Encarsia pergandiella (Hunter et al., 2003). A recent publication of the C. punctatus Cardinium genome by Siozios et al. (2019) has suggested possible unique genes related to CI. Overall, these observations suggest potential Cardinium-induced CI should be a priority of investigation in the future due to its possible role in driving the endosymbiont (and any desired effects) into midge populations.
The major Palearctic vector of BTV, C. obsoletus, has previously been shown to contain Cardinium at 0.4% prevalence (Pagès et al., 2017). We did not observe the endosymbiont in the one C. obsoletus population screened in this study, although the sample size was too small to conclude the absence of infection, and lack of DNA barcodes in this case prevents matched comparison. Nevertheless, a low prevalence indicates Cardinium is likely not relevant to vector biology at a population level and thus this species complex does not appear to be a suitable candidate for investigation.
By contrast, two other important vector species, C. imicola and C. sonorensis, have been reported to carry Cardinium infection (Morag et al., 2012;Möhlmann, 2019). Our results demonstrated the presence of Cardinium in one population of C. imicola from South Africa (Onderstepoort), whereas specimens of two populations from Corsica were negative for the presence of Cardinium, a pattern likely associated with properties of founder individuals (Pilgrim et al., 2021). Thus, these species appear to be the most promising candidate species for investigating Cardinium effects on vectorial capacity. As certain C. imicola and C. sonorensis populations are polymorphically infected (Morag et al., 2012;Möhlmann, 2019), the two species both present naturally occurring negative controls without the need for the confounding effects of curing through antimicrobials. In favour of C. imicola as a model is the ease of obtaining large field catches, but barriers in laboratory cultivation are still to be overcome. However, C. sonorensis laboratory colonies already exist, suggesting this species is the most promising candidate for investigating symbiont-virus interactions and symbiont-mediated reproductive effects. It will be worthwhile, additionally, to screen other Culicoides species, which may become colonised in the near future, such as C. stellifer (Erram & Burkett-Cadena, 2020).

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.
File S1. Primer attributes, PCR conditions and statistical test results of sex-based infection patterns.