Bacterial communities of the reproductive organs of virgin and mated common bedbugs, Cimex lectularius

1. Microbes associated with reproductive organs of animals are either sexually transmitted or opportunistic. Both can affect host defence, immunity, and future colonisation with other microbes. There are only few studies on the microbiota of reproductive organs in insects and how they are affected by copulation.

In vertebrates, microbes colonise various reproductive organs in a wealth of species (Hirsh, 1999;Lombardo & Thorpe, 2000;Hupton et al., 2003;Virecoulon et al., 2005;González-Marín et al., 2011;White et al., 2011). The few studies in insects found bacteria associated with the reproductive organs of female Formosan subterranean termites (Raina et al., 2007), male wood-boring beetles (Rizzi et al., 2013), and male and female common bedbugs (Reinhardt et al., 2005, Otti et al., 2017; see Table 1). Most of the bacteria in and on the copulatory organs of insects belong to the classes of Actinobacteria, Bacilli, or Gammaproteobacteria (Otti, 2015). Table 1. Presence of previously found bacterial genera in the reproductive organs of non-mated and mated bedbugs.

Genus
Present study Reinhardt et al. (2005) Otti et al. Given are the results from Reinhardt et al. (2005) and Otti et al. (2017) and the present study for each organ (S, male sperm container; I, male intromittent organ; C, female copulatory organ) and mating status (NM, non-mated; M, mated). In contrast to our study, both previous studies are based on culture-dependent methods.
The effect of these microbes has been less considered even though OMs are ubiquitous and copulatory wounding is both common and widespread across taxa (Lange et al., 2013;Reinhardt et al., 2015). Despite causing infections (Klainer & Beisel, 1969), bacteria affect sperm function directly (Otti et al., 2013;Reinhardt et al., 2015), lower the proportion of viable sperm in the female storage organs (McNamara et al., 2014), and cause fitness costs due to resource allocation to the immune system (Sheldon & Verhulst, 1996;Zuk & Stoehr, 2002). Such fitness costs due to immune challenge include reduced probability of reproduction (Rigby & Jokela, 2000) and reduced offspring quality (Ilmonen et al., 2000).
Opportunistic microbes have the potential to become pathogenic when the immune system of the host is disturbed (Klainer & Beisel, 1969). They interact with the immune system of the host that regulates the growth of OMs, and even endosymbionts (Login et al., 2011). If such an immune response does not target specific OMs, the immune response might affect symbionts at the same time. Mating therefore has the potential to change the microbial communities in the reproductive organs not only by transmitting microbes but also by eliciting immune responses that shape the resident microbial community. Even more complicated, symbionts probably compete with OMs or STMs because invasion into a bacterial community is limited by available resources (Li & Stevens, 2012;Mallon et al., 2015). Symbionts have actually been shown to provide protection against invading microbes (Reid et al., 1987;Boris et al., 1998;Oh et al., 2009;Koch & Schmid-Hempel, 2012;Mattoso et al., 2012;Kamada et al., 2013;Kaltenpoth & Engl, 2014;Braquart-Varnier et al., 2015;King et al., 2016). We suggest that this protection might also be directed against sexually transmitted microbes.
Some of the interactions among OMs, STMs, and symbionts and some of the fitness effects have been established, but it seems important to consider two basic but fundamental insights from community ecology -that species are not necessarily redundant and that ecosystem effects vary with species composition of the community (Allison & Martiny, 2008;Rillig et al., 2015). In terms of species redundancy, it is clear that different microbe species, and even populations, have different effects on the host. For example, in humans, Chlamydia trachomatis (Eley et al., 2005), Escherichia coli (Diemer et al., 1996;Diemer et al., 2003;Prabha et al., 2010), Pseudomonas aeruginosa (Huwe et al., 1998), and Staphylococcus aureus  were shown to cause agglutination and apoptosis of spermatozoa, whereas an Enterococcus species, Staphylococcus saprophyticus (Huwe et al., 1998) and Chlamydia trachomatis (Puerta Suarez et al., 2017) did not have such an effect. In terms of community effects, the interaction of invading bacteria with each other or with resident bacteria (Otti et al., 2017) could affect the outcome of an infection. If genital-associated bacteria behave as a community, then changes in the species composition, and even in the abundance of individual species, will decisively affect the impact of this community on the host. As a minimal precaution and a first step towards this somewhat visionary notion of insect reproductive ecology, it will be important for descriptive and functional studies on the microbes' role in reproduction to consider the mixed-species nature of natural microbe associations with reproductive organs.
Common bedbugs, Cimex lectularius, have previously been used as a model to study the effects of microbes associated with reproduction. Male bedbugs traumatically inseminate females by piercing the female's abdominal wall with an intromittent organ (Carayon, 1966), called paramere. Sperm are injected from the sperm container (sperm vesicles) into a paragenital female copulatory organ, called the mesospermalege. The microbial species situated on the male intromittent organ consist of environmental bacteria and fungi (i.e. OMs) (Reinhardt et al., 2005;Otti et al., 2013) and can kill females (Reinhardt et al., 2003). The female copulatory organ has evolved to reduce the mortality after infections derived from microbes on the male intromittent organ (Reinhardt et al., 2003; but see Morrow & Arnqvist, 2003). Bacteria found on the male intromittent organ kill sperm in vitro (Otti et al., 2013). Male responses include the presence of a constitutive immune effector (lysozyme-like activity) in the seminal fluid (Otti et al., 2009) which can reduce sperm mortality (Otti et al., 2013). A notable omission by many, if not most, of these studies is that the transmission is rarely examined, either directly by observation or indirectly by inference from comparing the microbiota of virgin and mated individuals.
Here we use a community ecology approach to describe the bacterial communities of different reproductive tissues in both sexes of the common bedbug and to examine the potential for sexual transmission of bacteria. We expected the communities to be shaped by the location of the given organ or its function. We assumed that the external intromittent organ of males harbours different communities, mostly consisting of environmental bacteria, as compared with internal organs of females and males which might harbour a core microbiome. We expected the communities in the female copulatory organ to be different from the communities in the male sperm container, because the sperm-receiving female copulatory organ is more likely to be invaded by bacteria during mating than is the male sperm container.
We mainly focused on analysing four assumptions of sexual transmission of bacterial communities. For this we analysed differences in bacterial diversity and abundance, represented by the number of reads, between the organ communities of non-mated and mated bedbugs, and the prevalence of bacteria introduced during mating. For male-to-female transmission we expected that the copulatory organs of mated female bedbugs show increased diversity and abundance, compared with virgins. Furthermore, we expected the copulatory organ of mated females to harbour bacteria that are found in non-mated males but not in non-mated females. If transmission is quantitatively significant, we expected that bacterial diversity or abundance would decrease in the sperm container and/or intromittent organ of males after versus before mating. Although hardly considered in the literature, it may be that female-to-male transmission is significant, in which case we would expect that the sperm containers and/or intromittent organs of mated males show increased diversity and/or abundance compared with non-mated ones. We would also expect the copulatory organ of mated males to harbour bacteria that are found in non-mated females but not in non-mated males, and possibly that bacterial diversity and abundance would be lower in mated than in virgin females.

Bedbug culture and reproductive biology
All bedbugs were maintained in an incubator at 26 ± 1 ∘ C, at 70% RH with an LD 12:12 h photoperiod. After eclosion, we divided virgin males and females into sex-specific groups and fed them twice with an interval of 1 week. The feeding and maintenance protocol was as described by Reinhardt et al. (2003). We used individuals from one large stock population (> 1000 individuals) which had been collected from an infestation in London and started as laboratory culture in the laboratory in Sheffield in 2006. This population was transferred to the laboratory of Animal Population Ecology at the University of Bayreuth in 2011 and maintained under identical culture conditions.

Mating and sample preparation
We collected the reproductive organs of 35 individual bedbugs in May 2012 to analyse the bacterial communities of the bedbug reproductive system. Ten 3-week-old virgin females were mated for 60 s to the same number of 3-week-old virgin males. Within 1-2 h after mating, we dissected the mated bedbugs. We collected the copulatory organ (mesospermalege) from the mated females and both the intromittent organ (paramere) and sperm containers (sperm vesicles) from mated males by sampling both organs from the same male. Spermatozoa leave the female copulatory organ after 4 h to travel through the haemolymph to the sperm storage organ (Carayon, 1966). This means that the sperm were still inside the copulatory organ at the time of dissection. We also collected the reproductive organs of five virgin females and 10 males randomly drawn from the stock populations. These males of unknown age and unknown mating status were isolated for 2 weeks prior to dissection. Not allowing them to copulate with a female ensured that they were at their full reproductive potential. Hereafter, we refer to virgin females and these males collectively as 'non-mated'. Except for the copulatory organ of non-mated females (n = 5), we collected the organs of 10 individuals for each reproductive organ and mating status.
We used standard dissection techniques under sterile conditions using a laboratory butane burner (Labogaz 206; Campingaz, Hattersheim, Germany) to minimise the potential of aerial bacteria contamination. We checked for contamination by placing LB agar plates next to the dissection microscope. No colonies were observed on these plates. Prior to any dissections, we autoclaved the dissection kit and, after each dissection, forceps and surgical scissors were dipped in ethanol (70%) and flame-sterilised. To prevent contamination with bacteria from the integument, we rinsed the integument of females with 70% ethanol prior to dissection. To further reduce the risk of contamination, we used different forceps to hold the male bedbug and to collect the internal organs. Dissected organs were transferred directly into the MicroBead solution (MO BIO Ultra Clean Microbial DNA Isolation Kit, catalogue no. 12224-250, dianova GmbH, Hamburg, Germany) for DNA extraction.

DNA extraction, library preparation, and sequencing
The bacterial community in and on the reproductive organs of bedbugs was described by the sequences of the 16S V4 region of bacteria obtained from three organs. We followed the protocol from the MO BIO UltraClean Microbial DNA Isolation kit with some additional steps. Instead of the MO BIO Vortex Genie, we used the Vortex Disruptor Genie (vertical 12-sample vortex). Before vortexing the samples in MicroBead tubes, we homogenised the samples with sterile pipette tips (200 μl) melted at the tip to form a pestle. These samples were then incubated and shaken at 65 ∘ C for 10 min. The kit uses microbeads and a lysing solution in combination to homogenise the tissue and extract the bacteria. We subjected the samples to PCR with barcoded versions of the universal primers 27f and 519r. Roche multiplex identifiers were incorporated between the sequences of adaptor A and 519r to give the structure: 5 ′ -Adaptor_A-sequencing_key-multiplex_identifier-519r-3 ′ . PCR consisted of an initial denaturation step of 2 min at 94 ∘ C and 25 cycles of 30 s at 94 ∘ C, 20 s at 52 ∘ C, and 60 s at 65 ∘ C. We checked the PCR products by gel electrophoresis, purified them with AMPure XP beads (catalogue no. A63881, Beckman Coulter GmbH, Krefeld, Germany), and sequenced them at the Earlham Institute (Norwich, U.K.) on a 454 titanium GS FLX (Roche, Basel, Switzerland) at 24-plex per quarter pico-titre plate.

Bioinformatic analysis
The data were demultiplexed with qiime (Caporaso et al., 2010). We removed sequences that did not match the default parameters of the 'split_libraries.py' script regarding quality score, sequence length and ambiguous bases. After this step, 68 513 out of 226 789 raw sequences remained in the dataset.
We subjected the remaining sequences to the dada2 pipeline (Callahan et al., 2016) in r (R Core Team, 2013). The sequences were filtered and trimmed with the default parameters of the 'fastqFilter' function. The first 15 bp were removed and the sequences were truncated after 300 bp to remove low-quality tails. Sequences with expected errors > 2 and a quality score < 2 were discarded. The remaining 23 695 sequences were dereplicated with the default parameters of the 'derepFastq' function and denoised with the 'dada' function with the 'selfConsist' option enabled, a homopolymer gap penalty of −1 and a band size of 32. We then constructed a sequence variant (SV) table with the 'makeSequenceTable' function. The remaining 22 566 SVs were checked for chimeras with the 'removeBimer-aDenovo' function and default parameters, resulting in 22 149 chimera-free sequences. The taxonomy of the SVs was assigned with the Greengenes database (De Santis et al., 2006). We used NCBI's BLASTn with the default options to verify the taxonomical assignments. We excluded uncultured and environmental sample sequences. The taxonomy assignments of Greengenes and BLASTn were in accordance for kingdom, phylum, class, order, family, and genus level in 22 out of 31 SVs. In three of the cases that were not in accordance, the BLAST hit with the highest e-value and coverage belonged to an endosymbiont of C. lectularius, the unclassified gammaproteobacterium mentioned by Hosokawa et al. (2010). We therefore changed the taxonomy assignment of these SVs. Two out of the misassigned SVs had BLAST hits that all agreed on one genus and we therefore changed the assignment. In four other cases there was no clear BLAST result. Hence, we kept the Greengenes assignment for the levels that were congruent with the BLAST results and changed the assignment of the other levels to 'unclassified'. We compiled all sample descriptions, read numbers and assigned taxonomy for the SVs in the Supporting Information (Tables S1-S3). Sequences were deposited in NCBI's Sequence Read Archive with the accession number PRJNA534453. Rarefaction curves drawn with the 'rarecurve' function in the vegan package (Oksanen et al., 2018) showed that our sampling captured most of the communities, as almost all curves reached a plateau (Fig. S1). We filtered out all SVs that belonged to chloroplasts or hosts and all SVs that occurred in less than two samples. The final SV table contained 31 SVs. There was one sample from the intromittent organ of a mated male which did not yield any SVs after the mentioned filtering. It was therefore excluded from the statistical analysis.

Statistical analysis of the bacterial communities in non-mated bedbugs
We focused on differences in bacterial diversity, prevalence, and abundance between the reproductive organs of non-mated bedbugs to describe the primary communities. We analysed the dissimilarity of bacterial communities between organs of non-mated bedbugs with a permanova ('adonis', 999 permutations, vegan package; Oksanen et al., 2018). Distances were estimated with the Jaccard index with the 'distance' function in the phyloseq package (McMurdie & Holmes, 2013). We calculated pairwise contrasts between the organs with the function 'pairwise.adonis' from the pairwiseadonis package (Martinez Arbizu, 2017) and corrected the P-values with the inbuilt Benjamini-Hochberg procedure. We used the function 'betadisper' (vegan package; Oksanen et al., 2018) followed by an anova to assess between-individual variation of bacterial communities across organs. To compare organs pairwise, we applied the 'TukeyHSD.betadisper' function in the vegan package (Oksanen et al., 2018). We estimated alpha diversity (Simpson index, 1 -D) with the 'estimate_richness' function in the phyloseq package (McMurdie & Holmes, 2013) and compared alpha diversity between organs of non-mated bedbugs with a generalised linear model followed by an anova. We visually inspected residual versus fitted plots to verify that residuals followed a normal distribution. To calculate the relative abundances of classes and genera, we divided the number of reads for the specific class or genus within a given sample by dividing them by the total number of reads within that sample.

Statistical analysis of mating-induced changes in bacterial communities
We analysed the effect of mating status regarding a possible sexual transmission of bacteria, including samples from non-mated and mated bedbugs. To compare alpha diversity (Simpson index 1 -D) between organs from non-mated and mated individuals, we fitted a generalised linear model followed by an anova. Included as fixed effects were organ and mating status and their interaction term. We visually inspected residual versus fitted plots to verify that the residuals followed a normal distribution. We applied the function 'betadisper' (vegan package; Oksanen et al., 2018) followed by an anova to assess between-individual variation of bacterial communities between mating status. Distances were estimated based on the Jaccard index with the 'distance' function in the phyloseq package (McMurdie & Holmes, 2013). We compared the number of reads in samples from non-mated and mated bedbugs with exact tests in the edger package (Robinson et al., 2010;McCarthy et al., 2012) after normalising read numbers based on the median ratio of each sample to the median library as a scale factor (Anders & Huber, 2010). To evaluate the effect of mating on the normalised number of reads of each bacterial genus, organs were analysed separately. P-values were adjusted with the inbuilt Benjamini-Hochberg procedure and a false discovery rate of 1%, and SVs that occurred only in non-mated or mated individuals were discarded. We used a Principal Coordinates Analysis to analyse whether mating increases the similarity of the bacterial communities in the reproductive organs. This analysis was based on an ordination calculated with the 'ordinate' function in the phyloseq package (McMurdie & Holmes, 2013) and the Jaccard index. We then analysed the dissimilarity of bacterial communities between non-mated and mated individuals with a permanova ('adonis', 999 permutations, vegan package; Oksanen et al., 2018), including the interaction of organ and mating status. To analyse which bacteria might be sexually transmitted, we extracted the SVs that are found in mated but not in non-mated individuals of one sex and in the organs of non-mated individuals of the opposite sex. Partitioning beta diversity (Sørensen index) into turnover and nestedness with the function 'nestedbetasor' in the vegan package (Oksanen et al., 2018), we investigated the mechanism of the mating-induced change in bacterial communities. We therefore produced a presence-absence matrix for each organ and mating status, which included all SVs present in the particular group of samples. We then calculated the proportion of beta diversity that was explained by turnover, i.e. a replacement of resident SVs with newly introduced SVs, and the proportion explained by nestedness, i.e. a loss or introduction of SVs.

Results
We sequenced the bacterial communities of three reproductive organs of the common bedbug, (i) the female copulatory organ, (ii) the male intromittent organ, and (iii) the male sperm container. Except for the copulatory organ of non-mated females (n = 5), we sequenced the communities of 10 individuals for each organ and mating status, resulting in a total of 55 samples. After filtering out chimeric sequences, chloroplast sequences, host sequences, and SVs that occurred in only one sample, we identified a total of 31 SVs. On average, each sample contained 3 ± 1 SVs (mean ± SD) and 340 ± 224 reads. Average alpha diversity was 0.38 ± 0.28 (Simpson index, 1 -D), or 0.68 ± 0.52 (Shannon index). There was one sample that harboured only SVs that were filtered out. It was therefore excluded from the statistical analysis.
In total, we detected 20 bacterial genera from six different classes in the reproductive organs of bedbugs. Most SVs belonged to the classes of Actinobacteria, Alphaproteobacteria, and Gammaproteobacteria. The relative abundance of these classes was highly variable between individual bedbugs (Fig. 2), even though they originated from the same population and environment. The female copulatory organ harboured almost only Alpha-and Gammaproteobacteria. In addition to Alpha-and Gammaproteobacteria, males harboured large proportions of Actinobacteria and a few males had Bacilli and Clostridia. Whereas females seem to have a core microbiome with two SVs shared by all females (Table 2), male reproductive organs did not consistently share the same SVs (Table 2). Also, the relative abundances of the genera in male organs varied tremendously across individuals (Fig. 3). Compared with the other SVs, both shared SVs in females, Rickettsia and the gammaproteobacterial endosymbiont discovered by Hosokawa et al. (2010), had high relative abundances of 43-95% and 2-28%, respectively (Fig. 3).
As a pattern of sexual transmission of bacteria, we expected changes in diversity and abundance of bacteria in the reproductive organs. Newly introduced bacteria in mated individuals of one sex that can also be found in non-mated individuals of the opposite sex would further indicate sexual transmission. We also expected that mating would homogenise the communities. In contrast to our predictions, there were no mating-induced changes in alpha diversity (F 1,50 = 2.324, P = 0.13) (Fig. 1). Alpha diversity was not affected by an organ-specific effect of mating status (F 2,48 = 0.448, P = 0.64). Mating did not change between-individual variation (F 1,52 = 0.141, P = 0.70) or the abundance of specific SVs in any of the organs (−2 < log 2 -fold change < 2, Q > 0.01).
However, as predicted for sexual transmission of bacteria, mating led to a larger overlap between the bacterial communities of females and males (Fig. 4). The community structure within organs (Jaccard index) was changed by mating status (F 1,48 = 2.793, R 2 = 0.044, P = 0.003) and the effect was dependent on organs (F 2,48 = 3.515, R 2 = 0.111, P = 0.001). As predicted by a transmission from the male to the female, mating introduced four new SVs to the copulatory organ (Table 3a), out of which one SV was harboured by both organs of non-mated males. Out of the two remaining SVs, three were found in the sperm container of non-mated males, and one on the intromittent organ of non-mated males. Out of the SVs that appeared on the intromittent organ after mating, two were found in the copulatory organ of non-mated females (Table 3b), suggesting a transmission from the female to the male. Mating introduced three SVs to the male sperm container that were harboured by non-mated females (Table 3c).

SV ID Class Genus
Male sperm container (n = 10) Male intromittent organ (n = 10) Female copulatory organ (n = 5) Shown are the proportions of samples from non-mated bedbugs harbouring the given SV.
Moreover, there was a large SV turnover of non-mated and mated bedbugs (male sperm container, 93%; male intromittent organ, 98%, female copulatory organ, 100% of the Sørensen index), suggesting a replacement of resident with newly introduced SVs in all organs.

Discussion
Using an ecological community approach, we found 31 SVs associated with the reproductive organs of the common bedbug, C. lectularius. The size of this bacterial community might be underestimated given that microbiomes of laboratory animals often seem to exhibit lower diversity compared with wild-caught individuals. For example, the gut microbiome of laboratory-reared Drosophila melanogaster consists of fewer taxa than the gut microbiome of wild-caught individuals (Broderick & Lemaitre, 2012).
We observed differences in bacterial community composition between individuals. These differences suggest either a strong host genotypic contribution or other properties of individual bedbugs that were not measured. Although large variation between closely related species (Chaston et al., 2015) or individuals of the same species are known (Costello et al., 2009;Nasidze et al., 2009;Ravel et al., 2011;Siddiqui et al., 2011;Human Microbiome Project Consortium, 2012;Moran et al., 2012;Osei-Poku et al., 2012;Hou et al., 2013;Zhou et al., 2013;Liu et al., 2014), these differences at the same time were likely to have hampered our organ-and mating-status-specific approach. Nevertheless, we found potential evidence of sexual transmission of bacteria because bacterial communities were more similar after mating, there was a high turnover of SVs in both sexes, and mating introduced new bacteria to the organs of mated females and males. Our results confirm that a community ecology approach can actually be used to analyse microbial transmission in insects.
Previous studies revealed that most of the bacteria associated with the reproductive system of insects belonged to the classes of Actinobacteria, Bacilli, or Gammaproteobacteria (Otti, 2015). In our study, a large proportion of individuals harboured Actinobacteria and Gammaproteobacteria. Bacilli were only present in a few individuals, whereas many individuals harboured Alphaproteobacteria. In the reproductive organs we found two genera that have been reported as endosymbionts of the common bedbug, Wolbachia and an unclassified gammaproteobacterium (Hosokawa et al., 2010). Both symbionts can be Fig. 3. Relative abundance of the 10 most abundant out of the 20 genera that were harboured by the reproductive organs of non-mated (NM) and mated (M) bedbugs. Relative abundances were calculated based on the number of reads of the same genus within a given sample divided by the total number of reads within that sample. Each bar represents one individual bedbug. Bars across male organs correspond to the same individual and bars across organs of mated bedbugs are ordered by mating pair. If the sequence variant was not assigned any genus, we report the lowest assigned taxon instead.
found in the bacteriomes that are attached or in close proximity to the reproductive organs. Wolbachia provides the host with B vitamins (Hosokawa et al., 2010), whereas to date the function of the gammaproteobacterium is unknown.
Our results are in line with the findings of culture-dependent studies that common bedbugs (C. lectularius) carry opportunistic bacteria (Reinhardt et al., 2005;Otti et al., 2013;Otti et al., 2017). Three genera previously found on the intromittent organ were not present in our data (Table 1). Staphylococcus was the only genus that was repeatedly found and that was associated with the same organ in more than one study (present study;Otti et al., 2017). It is not clear whether this is due to differences in the study design or the consequence of the large differences in bacterial composition across individuals that our study reports. Such between-individual differences in the microbiome have been reported in a variety of human tissues, including saliva, urine, skin, nares and stool (e.g. Costello et al., 2009;Nasidze et al., 2009;Siddiqui et al., 2011;Human Microbiome Project Consortium, 2012), and even in reproductive organs (Ravel et al., 2011;Hou et al., 2013;Zhou et al., 2013;Liu et al., 2014). Studies reporting on compositional between-individual differences in the microbiome of insects comprise, for instance, the gut of mosquitoes (Osei-Poku et al., 2012) and honey bees (Moran et al., 2012). To our knowledge, this is the first study to report on compositional between-individual differences in the bacterial community of the insect reproductive tract. Such microbial differences between individuals in a similar environment may be important when interpreting aspects of sexual behaviour in the context of animal 'personalities'.
The male intromittent organ and the female copulatory organ of non-mated bedbugs harboured distinct bacterial communities. In addition, the copulatory organ of non-mated females had a lower between-individual variation compared with the organs of non-mated males. Differences in community composition are unlikely to be caused by organ location as the intromittent organ and the sperm containers of bedbug males harboured Each circle represents a sample with its bacterial community. Ellipses give the 95% confidence interval. There were 10 samples for each organ and mating status, with the exception of the intromittent organ of mated males (n=9) and the copulatory organ of non-mated females (n = 5).
bacterial communities with similar composition. Sex differences in diversity or composition of bacterial communities extracted from whole-body homogenates or from the gut are common across animals (Markle et al., 2013;Valiente Moro et al., 2013;Haro et al., 2016). Some of these differences may arise from the different niche that the sexes and even organs may occupy; others, however, are likely to be closely linked to the most pronounced difference found between the sexes -reproduction. Hupton et al. (2003) and Otti et al. (2017) found pronounced differences in the microbiome of reproductive organs of female and male red-winged blackbirds and bedbugs, respectively.
For the first time, we showed that mating changes the bacterial communities of the reproductive organs in insects. We even found indications of a sexual transmission of bacteria in insects. We predicted that, in the case of a transmission, some of the bacteria found in one sex might be transmitted to the opposite sex. This should be reflected in a decrease in diversity or abundance in the organs of one sex and an increase in the other. Furthermore, mating should homogenise the communities of both sexes. We showed that copulation increased the similarity of the bacterial communities of male and female organs. A high turnover of SVs between the organs of non-mated and mated males was found, suggesting a replacement of resident with newly introduced bacteria.
Indeed, there were newly introduced bacteria in the copulatory organs after mating that were also present in the organs of non-mated individuals of the opposite sex. Taken together, these observations suggest sexual transmission of bacteria in the common bedbug. Even more interesting, the transmission seems to be two-sided. We found shared bacteria between non-mated females and mated males, indicating a transmission from the female to the male.
Contradicting our expectations about sexual transmission, mating did not induce changes in bacterial abundance, as shown by similar read numbers in organs from non-mated and mated individuals. OMs might cause an infection, or lower reproductive success by increasing sperm mortality (Otti et al., 2013). Therefore, females, and potentially even males, should have evolved mechanisms to protect themselves from these effects of OMs. Haemocytes are constantly present in the copulatory organ of female bedbugs (Carayon, 1966) and they can readily phagocytose bacteria as part of insect immune defence (Lavine & Strand, 2002). Even if bacteria are transmitted to the female, these could be eliminated by haemocytes without a costly systemic immune response. Physical barriers may also reduce the receipt of bacteria by females and therefore bacterial abundance. In the case of the bedbug, one may speculate that the highly elastic membrane in the bedbug copulatory organ of females (Michels et al., 2015), which the male penetrates during copulation, may function like a boot scraper.
The microbiome of the reproductive tract might protect its host from invading microbes during copulation. In humans, the female genital tract is inhabited by high proportions of lactobacilli (Ravel et al., 2011; but see Anahtar et al., 2015), which were reported to inhibit the growth of uropathogenic bacteria and their adhesion to epithelial vaginal cells (Reid et al., 1987;Boris et al., 1998). Rickettsia and the gammaproteobacterial endosymbiont reported by Hosokawa et al. (2010) were the only genera that were commonly found in all non-mated females and had high relative abundances. The relationship Table 3. Sequence variants (SVs) that were introduced to the reproductive organs of mated bedbugs: (a) the female copulatory organ; (b) the male intromittent organ; and (c) the male sperm container. It is indicated whether these SVs were present in the organs of non-mated bedbugs from the opposite sex (S, male sperm container: n = 10; I, male intromittent organ: n = 10; female copulatory organ: n = 5). A sexual transmission would be indicated by shared SVs between mated individuals of one sex and non-mated individuals of the opposite sex.
between Rickettsia and its host is not essentially mutualistic as it has the ability to manipulate the reproduction of ladybird beetles (Werren et al., 1994;Hurst et al., 1999;von der Schulenburg et al., 2001) and parasitoid wasps (Hagimori et al., 2006;Giorgini et al., 2010). However, Rickettsia has been shown to protect its whitefly host against a challenge with Pseudomonas syringae (Hendry et al., 2014). Unfortunately, the function of the gammaproteobacterial endosymbiont of C. lectularius is still unknown. If symbionts provide protection against transmitted bacteria in C. lectularius, Rickettsia and the gammaproteobacterial endosymbiont might be the genera involved in a protection against invading bacteria. To date, nothing is known about potential protection mechanisms of symbionts in the reproductive organs of insects.
We found distinct bacterial communities in and on the mating-associated organs of C. lectularius. These communities were composed of species that are known to be endosymbiont of the common bedbug but also species that are thought to be OMs. Future research should investigate their role in reproduction in more detail and whether they can provide protection against bacteria invading the reproductive organs. Taken together, our results suggest that mating has an effect on the bacterial flora of organs involved in mating and that there might be sexual transmission of bacteria. The identification of the bacteria of reproductive organs using a community approach is an important first step to study the transmission of bacteria between the sexes. Our study highlights the need to consider the role of the entire microbial community when examining the impact of sexually transmitted bacteria on reproduction, both generally and, in insects, specifically. This notion includes traditional single-species transmission assays that quantify how often bacteria are actually transmitted during copulation because the results of these assays may depend on the microbial community present in the organ(s) considered, and whether this transfer is one-sided or reciprocal. Assuming that transmitted microbes also perturb the microbiome of the reproductive organs in species other than bedbugs, it would be interesting, too, to consider whether the immune responses in females and males also differ with respect to which microbe species enter which particular microbial communities of reproductive organs. PRJ performed the bioinformatics and statistical analysis. SB, OO, and KR interpreted the results and wrote the manuscript. All authors read and approved of the final manuscript.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.