Using axenic and gnotobiotic insects to examine the role of different microbes on the development and reproduction of the kissing bug Rhodnius prolixus (Hemiptera: Reduviidae)

Abstract

Kissing bugs (Hempitera: Reduviidae) are obligately and exclusively blood feeding insects. Vertebrate blood is thought to provide insufficient B vitamins to insects, which rely on symbiotic relationships with bacteria that provision these nutrients. Kissing bugs harbour environmentally acquired bacteria in their gut lumen, without which they are unable to develop to adulthood. Rhodococcus rhodnii was initially identified as the sole symbiont of Rhodnius prolixus, but modern studies of the kissing bug microbiome suggest that R. rhodnii is not always present or abundant in wild‐caught individuals. We asked whether R. rhodnii or other bacteria alone could function as symbionts of R. prolixus. We produced insects with no bacteria (axenic) or with known microbiomes (gnotobiotic). Gnotobiotic insects harbouring R. rhodnii alone developed faster, had higher survival, and laid more eggs than those harbouring other bacterial monocultures, including other described symbionts of kissing bugs. R. rhodnii grew to high titre in the guts of R. prolixus while other tested species were found at much lower abundance. Rhodococcus species tested had nearly identical B vitamin biosynthesis genes, and dietary supplementation of B vitamins had a relatively minor effect on development and survival of gnotobiotic R. prolixus. Our results indicate that R. prolixus have a higher fitness when harbouring R. rhodnii than other bacteria tested, that this may be due to R. rhodnii existing at higher titres and providing more B vitamins to the host, and that symbiont B vitamin synthesis is probably a necessary but not sufficient function of gut bacteria in kissing bugs.

1. INTRODUCTION

It is now widely appreciated that most, if not all, metazoans have intimate, complex relationships with beneficial microorganisms. These relationships are often obligate from the perspective of the host as absence of the microbes is highly deleterious to host fitness. In some systems, the microbial partners are variable between hosts, suggesting low specificity between host and symbiont. In others, the relationship is highly specific, with host and symbiont coevolving over millions of years. In both systems it is essential that appropriate symbionts be acquired and maintained. Highly specific symbioses have often evolved mechanisms or behaviours to ensure transmission of appropriate symbionts between generations (Braendle et al., 2003; Martinson et al., 2012; Salem et al., 2015). In less specific relationships, multiple microbes may be capable of fulfilling the symbiotic role for the host, allowing for lower fidelity of transgenerational transmission (Coon et al., 2016). These microbes may be environmentally acquired or transmitted through contact with individuals harbouring appropriate microbes. In these cases, communities of microbes are often present, comprised of both beneficial microbes as well as commensals (Coon et al., 2014). Our understanding of the ubiquity of these more complex systems has been revolutionized by culture‐independent analyses of microbes which often defy cultivation (Tinker & Ottesen, 2021). While sequence‐based approaches can inform which microbes are present and what role they may be playing in host biology, they are not always suitable for determining what microbes in a community are serving as symbionts and which are commensals. To explore these interactions ideally requires a system in which host acquisition of beneficial microbes from the environment can be easily manipulated.

All known obligately and exclusively blood feeding arthropods require bacteria to successfully complete development. In most of these cases, the arthropods harbour intracellular symbionts which are passed vertically from mother to offspring with extremely high fidelity, usually via transovarial transmission (Allen et al., 2007; Hosokawa et al., 2010; Pais et al., 2008). Kissing bugs (Hemiptera: Reduviidae: Triatominae) are a group of ~130 species of insects that obligately and exclusively feed on vertebrate blood, and also require bacteria for successful development (Brecher & Wigglesworth, 1944; Durvasula et al., 1997; Lake & Friend, 1967; Nyirady, 1973). In contrast to the symbionts of other exclusively blood feeding arthropods, the symbionts of kissing bugs are not intracellular and are acquired from their environment, often through coprophagy of faeces from cohabiting kissing bugs (Brown et al., 2020). Early studies of kissing bug microbiomes suggested that a single microbial species dominated the microbiome (Brecher & Wigglesworth, 1944; Lake & Friend, 1967; Wigglesworth, 1936). In the model species Rhodnius prolixus, an Actinobacteria, Rhodococcus rhodnii, was initially identified as the single symbiont inhabiting the anterior midgut, with other kissing bug species harbouring distinct microbes (Brecher & Wigglesworth, 1944; Cavanagh & Marsden, 1969; Duncan, 1926; Goodchild, 1955; Gumpert, 1962; Marchette & Hatie, 1965; Weurman, 1946; Wigglesworth, 1936). These studies relied on culture‐dependent methods, which we now know underestimated the microbiome.

Explorations of the kissing bug microbiome employing modern 16S rDNA amplicon sequencing techniques have shown that the microbiome of most kissing bug species is composed of dozens to hundreds of members (Brown et al., 2020; Carels et al., 2017; da Mota et al., 2012; Díaz‐Sánchez et al., 2018; Dumonteil et al., 2018; Gumiel et al., 2015; Kieran et al., 2019; Mann et al., 2020; Montoya‐Porras et al., 2018; Oliveira et al., 2018; Rodríguez‐Ruano et al., 2018; Waltmann et al., 2019). Differences in species tested, sample size, sampling methodology, and analysis have made comparisons across studies difficult, though several trends are apparent. Studies that have repeatedly sampled the same population and carefully controlled for insect developmental stage have shown that the microbiome shifts through development, becoming progressively less diverse over the life of the insect (Brown et al., 2020; Rodríguez‐Ruano et al., 2018). Secondly, Actinobacteria are often found as dominant members of the microbiome (Brown et al., 2020; Carels et al., 2017; Dumonteil et al., 2018; Gumiel et al., 2015; Montoya‐Porras et al., 2018; Oliveira et al., 2018). These microbiome surveys suggest that triatomine symbionts lie between the seemingly random acquisition of environmental bacteria as seen in larval mosquitoes (Coon et al., 2014, 2016) and the establishment of specific monoxenic Burkholderia symbionts of Riptoris sp. (Acevedo et al., 2021; Kikuchi et al., 2007).

In other obligately hematophagous insects, a single species of symbiont is sufficient to support host development. Early studies indicate that R. rhodnii is a sufficient symbiont for R. prolixus on its own (Brecher & Wigglesworth, 1944; Lake & Friend, 1967), but many of these studies lacked modern tools to assure that their insects were indeed gnotobiotic or axenic. When considering the culture‐independent surveys, the lack of a consistent microbiome among individuals within a species suggests that a number of different bacteria can function as symbionts of kissing bugs. Thus, the identity of bacteria that can function as symbionts in this system is unknown.

Another unresolved question in this system is the function of the symbionts. Initial studies implied that the symbionts were necessary for B vitamin synthesis, similar to other obligately hematophagous insects (Baines, 1956; Duncan, 1926). The ability of R. rhodnii to synthesize thiamine (B₁) and folic acid (B₉) was established by 1960 (Harington, 1960), while the draft genome of R. rhodnii identified genes involved in synthesis of B vitamins (Pachebat et al., 2013). However, experimental studies on the role of B vitamins provided mixed evidence for their role in the symbiosis. Lake and Friend (1968) used artificial diets with varying B vitamin composition to show that loss of any B vitamin had a negative effect on moulting success, but that even insects reared on a diet containing all the B vitamins did not moult as successfully as bugs harbouring symbionts. Hill et al. (1976) developed B vitamin auxotrophic strains of R. rhodnii and tested their effect as sole symbionts of R. prolixus. They found that for all auxotrophic strains, moulting of insects was higher than for aposymbiotic insects. From these data, they concluded that while B vitamin biosynthesis is potentially a necessary feature of a successful symbiont of R. prolixus, it may not be sufficient to fully support development of R. prolixus.

In this study we focused on members of the genus Rhodococcus to assess whether other species in this genus could function as symbionts of R. prolixus. These Gram‐positive bacteria inhabit diverse environments and ecological niches. Many species in this genus are of interest due to their abilities to break down complex organic compounds including industrial wastes and pollutants (Bell et al., 1999) and have been isolated from diverse environments such as waste water sludge, soil, and aquatic sediments (Garrido‐Sanz et al., 2020). One of the most well‐studied species of the genus is R. hoagii/equii, a pathogen of foals, other livestock, and an important pathogen of immunocompromised humans (Mosser & Hondalus, 1996). In addition to R. rhodnii, other species within the genus have been identified as symbionts. R. triatomae was isolated from the gut of a kissing bug (Yassin, 2005), while Rhodococcus sp. UM008 was isolated from the renal tissue of the skate Leucoraja ocellata and may be involved in steroid biosynthesis (Wiens et al., 2016), and Rhodococcus WMMA185 was isolated from a sponge, though whether it is strictly host‐associated has not been determined (Adnani et al., 2016).

The repeated evolution of symbiosis and host association in this genus led us to ask whether other species could function as symbionts of R. prolixus. We examined the effects of five Rhodococcus species on R. prolixus development and reproduction, and compared this to another Gram‐positive bacterium (Micrococcus luteus) and Escherichia coli, which has previously been determined to rescue development of axenic mosquito larvae (Coon et al., 2014; Wang et al., 2021). We also examined available genome sequences of Rhodococcus spp. to assess whether each tested strain was able to synthesize B vitamins.

2. MATERIALS AND METHODS

2.1. Insect colonies

Rhodnius prolixus were obtained from the laboratory of Dr Ellen Dotson at the Centres for Disease Control and Prevention through BEI Resources. Insects were housed in an environmental chamber maintained at 28°C with a 12 h L:12 h D photoperiod and 80% RH. Conventionally reared insects were maintained in 1 L Nalgene containers with a mesh lid, cardstock folded to allow insects to reach the top of the container, and a 10 cm diameter filter paper on the bottom of the container. Insects were kept in cohorts of several hundred insects per container (first to third instar), ~100 insects per container (fourth to fifth instar) or ~50 insects/container (adults). Colony insects were fed defibrinated rabbit blood (Hemostat Laboratories) inoculated with 10⁶ CFU/ml of R. rhodnii through an artificial membrane feeder. The feeder consisted of a water‐jacketed glass bell with a latex membrane spread over the opening which was attached to a recirculating water bath heated to 37°C. Nymphs were fed biweekly for 3 h in the dark at 25°C. Following the blood meal, insects were sorted based on their feeding. Adults were blood fed biweekly and eggs were collected following each feed.

2.2. Bacteria

To select bacteria for experiments, 116 Rhodococcus genomes classified as having assemblies at either “complete” or “scaffold” level were downloaded from NCBI (Table S1). Four genomes from species in the sister genus Nocardia were selected as outgroups. Protein sequences from 30 housekeeping genes were extracted from the genomes (Table S1) and aligned individually using MAFFT (Katoh et al., 2002) with the –L‐INS‐i option. Aligned sequences were then concatenated and low‐quality regions removed from the concatenated alignment (see Supporting Information S1). The alignment was used to build a phylogeny of the chosen genomes using PhyloBayes‐MPI (Lartillot et al., 2009) using the default options (“‐cat ‐gtr”). Two chains were run and used as input to bpcomp with a burnin of 100 generations and sampling every 10 trees to generate a consensus tree. The consensus tree (Figure S1) was visualized with Figtree (https://github.com/rambaut/figtree/releases).

Rhodococcus rhodnii and R. triatomae were selected based on previous reports of symbiotic associations with kissing bugs (Brecher & Wigglesworth, 1944; Pachebat et al., 2013; Yassin, 2005). Other species were selected based on their representation of major clades of Rhodococcus (Figure S1) and availability in public repositories. Cultures of R. rhodnii were obtained through the American Type Culture Collection (ATCC), R. triatomae DSM‐44892 and DSM‐44893 were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ), R. rhodochrous and R. erythropolis and R. koreensis were obtained from NITE Biological Resource Centre (NBRC), Escherichia coli MG1655 and Micrococcus luteus were gifts of Eric Stabb and Michael Strand, respectively (Table S2). Bacteria were cultured on LB plates or liquid media at 28°C. Growth curves were constructed for each bacteria by inoculating 10 ml of sterile LB with a single bacterial colony and measuring optical density (OD) at 600 nm on a Beckman Coulter DU 640 spectrophotometer until the culture reached stationary phase. At each timepoint, bacteria were serially diluted and plated on LB agar to determine the number of colony forming units (CFU) corresponding to a given OD. To assess growth of bacteria in the absence of B vitamins, all strains were grown in standard M9 minimal media without addition of biotin or thiamine. Strains were grown in triplicate at 28°C for 72 h and OD600 was measured every 24 h as described above. To determine whether incubation with whole blood compromised bacteria, we incubated bacteria at 10⁶ CFU/ml for 24 h at 28°C in sterile blood. Inoculated blood was then serially diluted and plated on LB media to determine the number of surviving CFUs for each strain.

2.3. Generation of axenic and gnotobiotic insects

To generate axenic insects, eggs from the colony were collected 4 days after being laid. In a microbiological safety cabinet, eggs were first immersed in 70% ethanol for 5 min in a cell strainer, followed by 3 min in 10% povidone‐iodine solution, then another 5 min wash in 70% ethanol, followed by three rinses in sterilized deionized water. Washed eggs were left to dry for at least 1 h, then placed in sterile 550 ml glass mason jars with a 25 mm filter paper at the bottom and covered with mesh fabric. These containers were then placed in sterile 1 L screw‐top Nalgene containers with a 2.5 cm hole drilled in the lid which was covered with sterile gas‐exchange tape. These containers were returned to the general environmental chamber and allowed to hatch for 5 days. Axenic nymphs were separated in the microbiological safety cabinet with sterile forceps into individual sterile mason jars and starved for 7 days prior to feeding. All feedings occurred in a closed microbial safety cabinet that had been UV‐irradiated for 10 min prior to feeding. Nymphs were fed as described above, except feeding bells were autoclaved prior to use, gamma‐irradiated latex was used as a membrane, and all feedings were conducted in the microbial safety cabinet with the sash closed. The axenic status of nymphs was tested for each cohort by sacrificing 3–4 nymphs, extracting total DNA using the methods described below, and performing PCR on the DNA using primers targeting highly conserved regions of the bacterial 16S rRNA gene (Table S3). Only cohorts with no amplification of 16S rDNA were used for further experiments.

Gnotobiotic insects were generated by adding 10⁶ CFU/ml of log‐phase bacterial cultures to sterile blood immediately prior to feeding. Bacterial concentrations were chosen based on standard culturing procedures of R. prolixus provided by the laboratory of Ellen Dotson via BEI Resources and the Centres for Disease Control and Prevention and previous studies (Durvasula et al., 1997). Fully engorged insects were then sorted into the wells of a 48‐, 24, or 12‐well sterile polystyrene cell culture plate for observation. Additional nymphs for bacterial quantification were kept in a sterile mason jar. In three cases, insects from these mason jars were added to the well plates and counted for developmental data (asterisks, Table S6). Insect development was monitored daily, and insects were refed 2 weeks after all insects had moulted or died. Subsequent blood meals were either sterile blood (single inoculation experiments) or inoculated with bacteria as above (repeated inoculation experiments). To examine the role of B vitamin provisioning by symbionts, axenic, fourth instar R. prolixus were fed blood meals containing a solution of B vitamins (Lake & Friend, 1968; Nikoh et al., 2014; Table S4) along with 10⁶ CFU/ml of bacteria. Insects were then monitored for development to adulthood as before.

2.4. Quantification of gut bacteria

Gnotobiotic or axenic nymphs were sampled at 1 and 5 days post‐blood meal in each instar. Whole insects were used to capture all bacteria within an insect. Insects were washed with ethanol and rinsed with autoclaved water, then ground with a pestle in a microcentrifuge tube. Tissues were resuspended in lysis buffer (10 mM EDTA, 50 mM Tris‐Cl pH 7.6, 2% Sarkosyl, 100 U Mutanolysin [Sigma Aldrich]) and incubated at 37°C for 1 h, then heated to 62°C for 10 min. Proteinase K (25 μg) and lysozyme (5 mg) were then added to the solution and incubated for 1 h at 37°C. Lysate was then washed twice with phenol:chloroform:isoamyl alcohol (25:24:1) and twice with chloroform. DNA was precipitated from the aqueous phase with 1/10th volume of 5 M sodium acetate and two volumes of 100% ethanol. DNA pellets were resuspended in 50 μl of 10 mM Tris pH 7.4 and frozen at −20°C.

Total genomic DNA was used as template for qPCR targeting single‐copy genes of each bacteria using optimized species‐specific primers (Table S3). Qiagen QuantiFast SYBR green qPCR mix was used in a Qiagen Rotogene qPCR machine with a three‐step program: an initial 10 min denaturation at 95°C, then 30 cycles of 95°C for 20 s, 55°C for 45 s, 68°C for 30 s, and a final extension at 68°C for 5 min, followed by a melt curve analysis. qPCR products were cloned into the pSCA vector of the Agilent StrataClone PCR cloning kit and used to generate absolute copy number standard curves of the target genes, which were used to quantify the copy number of bacterial genomes in the DNA extraction from a whole, individual insect. To avoid biasing our measurements of bacterial genomes, we did not normalize our bacterial qPCR data to a host gene, as differences in host size may have confounded estimation of bacterial titre.

Acquisition of bacteria via coprophagy was also tested. Immediately post‐blood meal, gnotobiotic third instar nymphs were placed vertically in a sterile 0.6 ml microcentrifuge tube oriented vertically. Faeces was collected for 48 h after which insects were removed from the tubes. Faeces was resuspended in 100 μl of sterile PBS and then pipetted on to autoclaved filter paper. The filter paper was provided to five axenic first instar nymphs for 3 days, after which the nymphs were fed a sterile blood meal then collected 24 h PBM and used for bacterial quantification as described above.

2.5. Egg laying assays

Adult R. prolixus from gnotobiotic colonies were fed 2 weeks post‐eclosion on either sterile (single inoculation) or inoculated blood (repeated inoculation) as before and allowed to mate for 1 week. Females that fed were then isolated in individual wells of sterile six‐well polystyrene plates and provided with sterilized cardstock to deposit eggs. Females were returned to the environmental chamber and allowed to deposit eggs for 10 days, after which egg numbers were counted.

2.6. Genome sequencing, assembly, annotation, and comparisons

Cultures of R. rhodnii and R. triatomae were grown in LB media to log phase, then genomic DNA was extracted using the method described above. Five micrograms of genomic DNA was used as input for SMRT library preparation using the Express TPK 2.0 kit and version 4 sequencing primer. Multiplexed libraries were run on a PacBio Sequel II system at the Georgia Genomics and Bioinformatics Core. Reads were assembled using CANU (Koren et al., 2017), FLYE (Kolmogorov et al., 2019), and ARROW (Korlach et al., 2017). Assembled contigs were annotated using the NCBI RefSeq annotation pipeline. Rhodococcus species used in the growth assays were selected for the availability of high‐quality genome sequences (“complete” according to NCBI Microbial Genome tables). Accessions are listed in Table S2. We used KEGG (https://www.genome.jp/kegg/) and MetaCyc (https://metacyc.org/) databases to reconstruct B vitamin synthesis pathways from genomes of microbes tested. Orthologous protein groups (“orthogroups”) were determined using Orthofinder (Emms & Kelly, 2019). We performed gene enrichment analysis using Fisher's exact test implemented in the Blast2GO package. We first selected orthogroups unique to R. rhodnii relative to the other species and compared them to all protein coding genes in R. rhodnii. We then performed a second enrichment analysis on the orthogroups that are present in all Rhodococcus species tested except R. rhodnii using the same approach. In this analysis, genes from R. triatomae that had orthologues present in all species tested except R. rhodnii were compared for functional enrichment against the complete R. triatomae genome.

2.7. Data analysis

Development time and survival were analysed using a Cox proportional hazards model via the R packages survminer (Kassambara et al., 2021) and survival (Therneau & Grambsch, 2000). Due to a large number of zero counts, egg laying data were analysed with a hurdle model implemented in the package pscl (Zeileis et al., 2008), with Tukey's post‐hoc tests performed with the package emmeans (Lenth, 2022).

3. RESULTS

3.1. Bacteria vary in their symbiotic potential

We first tested whether different bacterial species in the gut of R. prolixus influenced the development time and moulting success of the insects. As insects which failed to moult ultimately died, moulting success is equivalent to survival in our experiments. Instar length varied significantly across development irrespective of the bacterium used for inoculation or inoculation regime (single or repeated inoculation), with 5th instars generally taking substantially longer than the first and second instars (Figure 1; Tables S5 and S6; p < .001, Cox proportional hazards, Table S7). Regardless of treatment, developmental times of R. prolixus were similar in the first and second instars (Figure 1; Tables S5 and S6; p = .76, Cox proportional hazards, Table S7) suggesting that symbionts play a minimal role in early development. The length of subsequent instars was highly significantly different among treatments and feeding regimes (p < .001, Cox proportional hazards test, Tables S6 and S7). Gnotobiotic insects in the single inoculation group developed slower than insects that were repeatedly inoculated during each blood meal, though this was significant only in the third to fifth instars (Table S7). Conventional nymphs and gnotobiotic nymphs harbouring other Rhodococcus species exhibited slower moulting than R. rhodnii nymphs, especially in the later instars, taking roughly 5 days longer to moult in their final instar regardless of inoculation regime.

FIGURE 1.

Cumulative proportion of R. prolixus nymphs moulting presented as inverted survival curves. Nymphs were inoculated either once (single inoculation) or with each blood meal (repeated inoculation). Moulting success is equivalent to survival as insects that failed to moult ultimately died. Insects inoculated with R. rhodnii performed significantly better than other treatments (p < .05, cox proportional hazards). Nymphs inoculated with other Rhodococcus species and E. coli could successfully develop to adulthood, though their development time was longer, and more nymphs failed to moult. No nymphs inoculated with M. luteus successfully moulted to adulthood, and all died prior to the fifth instar. Nymphs were blood fed, and only nymphs that successfully fed were retained for the experiment. Nymphs were observed daily to determine the number of days to moult. Nymphs were given 30 days post‐blood meal to complete a moult, after which they were classified as having failed to moult. Different coloured lines correspond to different instars. A cladogram next to the bacterial species names indicates relatedness inferred from the phylogeny in Figure S1. Samples sizes are given in Table S5, summarized data are given in Table S6, and statistical details are given in Table S7.

Nymphs that did not moult ultimately died, thus moulting success is equivalent to survival. The proportion of nymphs moulting varied with the bacteria used to generate gnotobiotic R. prolixus and feeding regime (Figure 1; Table S6). As with development time, R. rhodnii gnotobiotic nymphs had higher moulting success and survival regardless of inoculation regime or instar, with nearly 100% of all nymphs moulting at each instar. Similarly, almost all R. erythropolis gnotobiotic nymphs successfully moulted in each instar. Moulting success was high for the other Rhodococcus sp. and E. coli treatments throughout early instars but declined in the fourth and fifth instars in the single inoculation regime. Rhodococcus and E. coli nymphs in the repeated inoculation regime had improved moulting for almost every treatment and instar.

As expected, axenic insects nearly all died prior to the final moult. Insects inoculated with M. luteus, regardless of inoculation regime, died prior to reaching adulthood, suggesting that it cannot function as a symbiont of R. prolixus. In the repeated inoculation, insects with M. luteus died at a higher rate, suggesting that its presence in the gut is detrimental to the host. Together, these data suggest that R. rhodnii is a superior symbiont to other Rhodococcus species, but that the other Rhodococcus species tested, along with E. coli, can support development of R. prolixus nymphs to adulthood, though at a substantial cost if the bug is not repeatedly inoculated with the bacteria. In contrast, M. luteus does not seem capable of functioning as a symbiont.

3.2. R. Rhodnii has a consistently high titre

We hypothesized that symbiotic potential may be related to bacterial titre in the gut. To assess this, we generated cohorts of gnotobiotic or axenic nymphs similar to the nymphs used in the developmental experiments. We fed axenic first instar nymphs with blood inoculated with 10⁶ CFU/ml while all remaining blood meals were sterile. For each instar, five nymphs from each cohort were sacrificed 1 and 5 days post‐blood meal and used for total DNA extraction. We performed qPCR on the extracted DNA to measure the copy number of a single‐copy bacterial gene using primers unique to each bacterial species tested.

Bacterial abundance varied greatly among the species tested and developmental stage (Figure 2). R. rhodnii titre was consistently higher than other species, quickly reaching ~10⁶ copies and remaining high throughout development (10⁴–10⁷ copies/insect). The titre decreased after each moult, presumably as many bacteria are shed with the contents of the gut prior to moulting, then recovering towards the end of each instar. There was a decrease in R. rhodnii titre between the second and third instar, though its abundance recovered throughout the remaining instars. We also tested whether bacteria could survive in the blood meal, inoculating sterile blood with a known concentration of CFUs of each strain, incubating the blood with the bacteria for 24 h, the plating dilutions of inoculated blood on LB plates. We found that bacteria responded differently to incubation in blood, but most species exhibiting an approximately 50% difference in CFU counts between the blood‐incubated cultures and control cultures in LB media. M. luteus titre dropped dramatically after incubation in blood, which may explain its inability to function as a symbiont in R. prolixus (Table S11).

FIGURE 2.

The titre of bacteria in insects from the single inoculation experiments. Points represent titre of individual insects, while lines represent the average titre (left axis). Grey bars indicate the percentage of nymphs successfully moulting (surviving) to the next instar (right axis). R. Rhodnii was found at a higher titre than other bacteria at every time point. Bacterial abundance fluctuated over time, decreasing after a moult and subsequently recovering by 5 days post blood meal (PBM). The titre of most bacteria declined after the second instar and failed to recover, though R. erythropolis titre remained low but stable throughout development. No insects inoculated with M. luteus successfully moulted beyond the fourth instar. The titre was measured in 3–5 insects using qPCR of single copy genes from each genome using species‐specific primers (Table S3). In the conventional treatment, only the titre of R. rhodnii was measured. Bacteria were fed at 10⁶ CFU/ml in the first blood meal and all subsequent blood meals were sterile. The titre was measured at 1 and 5 days PBM by homogenizing whole insects (first to third instars) or guts only (fourth and fifth instars).

Other bacteria displayed different patterns of abundance. R. triatomae, R. rhodochrous, and R. koreensis had steady titre for the first two instars but declined rapidly following the second to third instar moult and never recovered (Figure 2). The decrease in titre roughly corresponded to lower moulting success and delayed development of these gnotobiotic groups of R. prolixus (Figure 1). The titre of E. coli in gnotobiotic R. prolixus peaked during the first instar then dropped to a low level around 10² copies/insect and remained low throughout the rest of insect development. Similarly, titre of R. erythropolis remained near 10² copies/insect, yet a high proportion of these gnotobiotic insects moult to adulthood, whereas other gnotobiotic insects in the single inoculation regime experiments had low success reaching adulthood (Figure 1). The titre of M. luteus dropped precipitously after inoculation and remained at very low levels until all insects ultimately died. These data suggest that the low survival of R. prolixus inoculated with M. luteus is unlikely to be due to proliferation or pathogenicity of the bacteria to the insect and may instead be due to failure of the bacteria to grow in the host. The decline in bacterial abundance in most species tested along with the increased moulting success and shorter development times of insects that are repeatedly inoculated with bacteria suggest that the fitness benefit of most bacteria is related to higher abundance in the insect, perhaps due to increased B vitamin availability, or direct consumption of the bacteria.

Our inoculation regime utilized an artificial method to produce gnotobiotic insects, providing bacteria in the blood meal at a high concentration. We chose this inoculation route to reduce variation in our experiments and due to its previous use in other studies and to minimize variation in initial inoculum size (Brecher & Wigglesworth, 1944; Durvasula et al., 1997). In natural conditions, nymphs acquire bacteria from their environment, primarily through coprophagy (Baines, 1956). To test whether each of the bacteria tested could be acquired through coprophagy, we exposed axenic nymphs to faeces of gnotobiotic insects then measured the titre of bacteria acquired by the axenic nymphs. After 3 days of exposure to inoculated faeces and following a blood meal, we detected all bacteria in the guts of nymphs though at titres lower than through blood meal inoculation (Figure S2).

3.3. Gut bacteria influence egg laying

Both single and constant inoculation gnotobiotic R. prolixus were assayed for egg production. Following the final moult to adulthood, male and female R. prolixus were fed a blood meal then allowed to mate for 1 week. Adults were fed according to their feeding regime, either sterile (single inoculation) or inoculated with bacteria (repeated inoculation). Females that fed were separated into wells of a sterile six‐well cell culture plate and provided with autoclaved cardstock to deposit eggs for 10 days. Egg laying was analysed with a two‐step hurdle general linearized model to account for the large number of insects that laid no eggs across treatments. There was a highly significant effect of bacterial treatment on egg number (p < .002 for all treatments, Figure 3; Table S8) and inoculation regime (p = .005, Table S8). Insects harbouring R. rhodnii laid the most eggs regardless of inoculation regime, with conventionally reared and R. erythropolis‐inoculated bugs laying the next most eggs. The R. prolixus inoculated with E. coli or repeatedly inoculated with R. rhodochrous laid the fewest eggs. M. luteus gnotobiotic nymphs did not successfully moult to adulthood, so no egg laying data were available.

FIGURE 3.

Egg laying of adult female R. prolixus by gut bacteria. Points indicate eggs laid by individual females and grey bars represent mean eggs laid for a given treatment. Insects harbouring R. rhodnii laid more eggs than all other treatments. No insects inoculated with M. luteus ever reached adulthood. Females were blood fed following their final moult and allowed to lay eggs for 10 days post‐blood meal. Significant differences were determined using a two‐step hurdle generalized linear model followed by a Tukey's post‐hoc test. Significant differences between the R. rhodnii single inoculation treatment and other treatments are indicated by asterisks (*p < .05, ***p < .001, ns, p > .05). Number of individuals tested is given below the x‐axis. Full pairwise comparisons are given in Table S9.

3.4. Genomic comparisons between symbiotic and nonsymbiotic Rhodococcus species

The bacteria tested varied in their ability to fulfil a symbiotic role in R. prolixus. R. rhodnii‐inoculated nymphs performed better in every metric tested (survival, developmental time, and egg production), while M. luteus gnotobiotic insects all died prior to adulthood. However, other strains of Rhodococcus had mixed success in their ability to function as symbionts of R. prolixus, and this did not appear to reflect the phylogenetic distance between bacterial species. To identify potential genes associated with successful symbionts, we explored the genome sequences of the Rhodococcus species used in the study. We sequenced the genomes of R. rhodnii and R. triatomae using the PacBio Sequel system. We successfully produced near‐chromosome level assemblies, producing four contigs in our assembly of R. rhodnii and 1 contig each for two R. triatomae strains. RefSeq annotations were produced from these assemblies by the NCBI Prokaryotic Genome Annotation Pipeline (PGAP). Assemblies and annotations can be found under the NCBI accessions PRJNA483464 (R. rhodnii) and PRJNA606336 (R. triatomae).

We first compared the presence of orthologues among the bacteria using OrthoFinder (Emms & Kelly, 2019). Protein coding genes were assigned to phylogenetically determined clusters of orthologues among the genomes, referred to subsequently as orthogroups. Using these data, we assessed the presence or absence of orthogroups between different sets of species. We found a core genome of 2231 orthogroups shared among all species. The genome of R. rhodnii encoded 304 unique orthogroups not found in the other genomes examined, while there were 316 proteins found in the non‐R. rhodnii species but absent from R. rhodnii. We examined functional enrichment of these unique gene sets using Blast2GO and found two clusters that were overrepresented in the unique R. rhodnii gene set (Table 1). In the genes unique to R. rhodnii, two functional categories were overrepresented: IPR010982 (12 genes, 4.4% of R. rhodnii unique sequences versus 0.55% of genomic sequences, p = 1.87 × 10⁻⁶, Fisher's exact test) and IPR001387 (11 genes, 3.8% of R. rhodnii unique sequences vs. 0.47% of genomic sequences, p = 3.15 × 10⁻⁶, Fisher's exact test). IPR001387 and IPRO010982 are the two components of the Cro/C1 bacteriophage repression system that has been shown to suppress transition of bacteriophage λ from the lysogenic to lytic pathways (Schubert et al., 2007). These results prompted us to examine the genome of R. rhodnii for phage using the PHASTER server (Arndt et al., 2016; Zhou et al., 2011), which identified three complete phage genomes present in the bacterium. Of the other Rhodococcus species, only R. erythropolis had a complete phage. R. triatomae, R. koreensis, and R. rhodochrous had none.

TABLE 1.

Gene set enrichment analysis (GSEA) of orthogroups from R. rhodnii

Over or underrepresented in GSEA	GO or Interpro ID	GO name	GO category	FDR	p‐value	Nr test	Nr reference	Nonannot test	Nonannot reference
Orthogroups unique to R. rhodnii a
OVER	IPR010982 (G3DSA:1.10.260.GENE3D)	Lambda repressor‐like, DNA‐binding domain superfamily		0.0104	1.88 × 10−6	12	21	292	3783
OVER	IPR001387 (CDD)	Cro/C1‐type helix‐turn‐helix domain		0.0117	3.15 × 10−6	11	18	293	3786
Orthogroups with higher numbers of paralogues in R. rhodnii relative to other Rhodococcus
OVER	GO:0006812	Cation transport	Biological process	0.0052	2.75 × 10−5	9	24	202	3873
OVER	GO:0006811	Ion transport	Biological process	0.0384	4.59 × 10−4	10	46	201	3851
OVER	GO:0071702	Organic substance transport	Biological process	0.0239	2.44 × 10−4	10	42	201	3855
OVER	GO:0072348	Sulphur compound transport	Biological process	0.0239	× 10−4	6	13	205	3884
Orthogroups with lower numbers of paralogues in R. rhodnii relative to other Rhodococcus
OVER	GO:0016491	Oxidoreductase activity	Molecular function	0.0001	8.48 × 10−8	11	119	34	3944
OVER	GO:0016620	Oxidoreductase activity, acting on the aldehyde or oxo group of donors, NAD or NADP as acceptor	Molecular function	0.0226	4.17 × 10−5	3	4	42	4059
OVER	GO:0016903	Oxidoreductase activity, acting on the aldehyde or oxo group of donors	Molecular function	0.0240	6.62 × 10−5	3	5	42	4058
Orthogroups absent in R. rhodnii but present in all other Rhodococcus examined
OVER	GO:0009236	Cobalamin biosynthetic process	Biological process	0.0001	7.63 × 10−8	9	5	306	3994
OVER	GO:0009235	Cobalamin metabolic process	Biological process	0.0001	7.63 × 10−8	9	5	306	3994
OVER	GO:0006766	Vitamin metabolic process	Biological process	0.0032	8.94 × 10−6	12	26	303	3973
OVER	GO:0006767	Water‐soluble vitamin metabolic process	Biological process	0.0032	8.94 × 10−6	12	26	303	3973
OVER	GO:0042364	Water‐soluble vitamin biosynthetic process	Biological process	0.0032	6.54 × 10−6	12	25	303	3974
OVER	GO:0009110	Vitamin biosynthetic process	Biological process	0.0032	6.54 × 10−6	12	25	303	3974
OVER	GO:0033014	Tetrapyrrole biosynthetic process	Biological process	0.0146	5.41 × 10−5	9	17	306	3982
OVER	GO:0033013	Tetrapyrrole metabolic process	Biological process	0.0146	5.41 × 10−5	9	17	306	3982
UNDER	GO:0044238	Primary metabolic process	Biological process	0.0363	1.51 × 10−4	42	888	273	3111
UNDER	GO:0019538	Protein metabolic process	Biological process	0.0389	1.98 × 10−4	2	180	313	3819
OVER	GO:0016701	Oxidoreductase activity, acting on single donors with incorporation of molecular oxygen	Molecular function	0.0389	1.98 E‐04	5	4	310	3995

Abbreviations: FDR, false discovery rate; GO, gene ontology; Nonannot test, genes in the test set that did not map to the GO or InterproID; Nonannot reference, genes in the reference set that did not map to the GO or InterproID; Nr test, genes that mapped to a given GO or InterproID in the test gene set; Nr reference, genes that mapped to a given GO or InterproID in the reference gene set.

^a No GO categories were overrepresented in the orthogroups unique to R. rhodnii, though 2 InterProScan groups were.

We next investigated orthogroups that had more paralogues in R. rhodnii than in other species examined, which we refer to as expanded. A given orthogroup was considered to be expanded in R. rhodnii if the R. rhodnii genome encoded a number of paralogues greater than 2 standard deviations above the mean number of paralogues across all species tested. Gene set enrichment analysis using gene ontology (GO) terms identified four biological processes that were overrepresented in the expanded orthogroups in R. rhodnii and all were related to transport (Table 1). They include cation and ion transport, organic substance transport, and sulphur compound transport. These results suggest that ability to either import or export compounds may underly the success of R. rhodnii as a symbiont and provide targets for future studies of symbiosis enabling genes.

The loss of specific genes may also define an effective symbiont and so we examined genes that were present in all the genomes of the tested Rhodococcus species but absent in R. rhodnii. Surprisingly, we found genes associated with cobalamin (vitamin B12) metabolism were overrepresented in the genes absent in R. rhodnii but present in other Rhodococcus species (Table 1). Cobalamin is not an essential B vitamin for insects (Dadd, 1973; Degnan et al., 2014), but is required by some bacterial species (Roth et al., 1996). We also examined genes with fewer paralogues in R. rhodnii than in the other Rhodococcus species, using a similar cutoff of orthogroups in which the number of R. rhodnii genes was less than two standard deviations below the mean number of genes per species in the other Rhodococcus species. Three GO categories, all corresponding to oxidoreductase activity were significantly enriched in genes with more paralogues in Rhodococcus species other than R. rhodnii (Table 1).

Next, we examined the presence of B vitamin biosynthesis genes among the different species tested in our bioassays (Figure 4a; Table S10). While several authors have provided evidence that R. rhodnii encodes some genes necessary for B vitamin biosynthesis (Pachebat et al., 2013; Tobias et al., 2020), to date, the presence or absence of complete pathways in the genome has not been examined. All tested Rhodococcus strains and E. coli encode the complete suite of genes necessary for the de novo synthesis of the B vitamins thiamine (B₁), riboflavin (B₂), nicotinate (B₃), pantothenate (B₅), pyridoxine (B₆), and folate (B₉), though in some cases the synthesis pathways differ between the Rhodococcus species and E. coli. This is most obvious in pyridoxine (B₆) synthesis, where the genes pdxABHJ and epd are missing from the Rhodococcus species and M. luteus. In these species, pyridoxine is synthesized via proteins encoded by pdxST. R. koreensis lacks an orthologue to the alkaline phosphatase phoA/B involved in folate biosynthesis but encodes numerous other alkaline phosphatases that may perform this reaction. The Rhodococcus genomes encode several highly conserved genes for the biosynthesis of biotin (B₇) via the 8‐amino‐7‐oxononanoate I pathway (bioABCDF) but lack the genes necessary to synthesize the pimeloyl coenzyme A moiety (fabBFZI/bioH), suggesting these bacteria cannot synthesize biotin unless they acquire this precursor. In contrast, M. luteus lacks nearly all genes necessary to synthesize biotin (bioADCFH and fabBFIZ). All the Rhodococcus species and M. luteus encode bioYNM which encode a biotin transporter, suggesting that the bacteria can acquire biotin from their environment. These results further suggest that biotin is probably only partially synthesized by these bacteria, relying on external pimeloyl‐CoA for de novo synthesis or importation of biotin via bioYNM. We did not fully consider biosynthesis of cobamides (B12), as these are thought to not be essential to insect development (Dadd, 1973). The invariable presence of the complete synthesis pathways for thiamine, riboflavin, nicotinate, pantothenate, pyridoxine and folate in the Rhodococcus species tested suggests that while provisioning one or more of these B vitamins is probably a necessary feature of a successful symbiont of R. prolixus, other factors probably underly the variation observed in development, survival, and reproduction among the various gnotobiotic treatments.

FIGURE 4.

B vitamin biosynthesis potential and effects of B vitamin supplementation on the development of gnotobiotic and axenic R. prolixus. (a) Presence and absence of core B vitamin biosynthesis genes in the genomes tested. Orthologues were identified through BLAST searches of genomes and by querying the KEGG and MetaCyc databases. Open circles indicate no orthologue of the enzyme was found in the annotated proteins of the genome. ^A ThiCD is a single bifunctional enzyme in Rhodococcus species and M. luteus, but is encoded by two separate enzymes in E. coli. ^B ThiO is necessary for B₁ synthesis using glycine as a precursor while ThiH is required when tyrosine is the precursor. ^C Rhodococcus sp. encode thiI, but the gene lacks a rhodanase domain which is necessary for sulphurtransferase activity (Martinez‐Gomez et al., 2011). In Salmonenlla enterica, YdjN and CdsH have been shown to fulfil the sulphurtransferase role of ThiI (Palmer et al., 2014). ^D YigB and YbjI are not required for riboflavin synthesis in Rhodococcus or M. luteus but are essential for E. coli riboflavin biosynthesis. ^E FolM is probably not essential for folate synthesis, but rather involved in the production or reduction of 7–8‐dihydropterins (de Crécy‐Lagard et al., 2007). ^F Pyridoxine is synthesized via different pathways in E. coli relative to the Rhodococcus species and M. luteus. In the non‐E. coli species, pyridoxal 5′‐phospate synthase (PdxST) synthesizes B pyridoxal 5′‐phosphate from glutamine, D‐ribose 5′‐phosphate, and glyceraldehyde 3‐phosphate. ^G While all the species tested except M. luteus can synthesize biotin, only E. coli can synthesize the pimeloyl‐ACP. (b) Cumulative moulting of fourth instar gnotobiotic and axenic R. prolixus nymphs supplemented with B vitamins in their blood meal. Lines show moulting success of nymphs supplemented with B vitamins; dashed lines indicate moulting and survival of fourth instars to fifth instars, solid lines indicate moulting of fifth instars to adults. Shaded areas represent moulting success of single inoculation nymphs with no B vitamin supplementation for comparison. Sample sizes are given in Table S5.

While our genomic data indicate that the tested Rhodococcus species cannot make biotin de novo, we directly tested whether the bacteria could be reared without any B vitamins in the media. We inoculated M9 minimal media lacking any B vitamins with single colony isolates of bacterial cultures and then measured bacterial growth over 72 h. We detected growth of all the Rhodococcus species tested (Figure S3), suggesting that they are capable of synthesizing biotin without the canonical genes fabBFZI/bioH. As expected, E. coli also grew in M9 while M. luteus did not (Figure S3).

Given the lack variation in B vitamin biosynthesis pathways among the bacteria and the variation in bacterial abundance in the insects (Figure 2), it is possible that B vitamin abundance is directly related to titre of bacteria in the gut, and that this variation is ultimately responsible for the phenotypes of bugs inoculated with different bacteria. If B vitamin provisioning were the only necessary function of symbionts, then supplementing blood meals with B vitamins should lead to similar development times of all gnotobiotic insects. We therefore tested whether supplementation with B vitamins via a blood meal could rescue R. prolixus development. Axenic or gnotobiotic fourth instar nymphs were tested for the effects of B vitamin supplementation as developmental differences among gnotobiotic treatments were greatest in the final two moults (Figure 1). Nymphs were provided with a blood meal containing all standard B vitamins (Table S4, Hosokawa et al., 2010), along with 10⁶ CFU/ml of each bacteria tested. Axenic insects given blood meals supplemented with the B vitamins all successfully moulted to fifth instars, while nearly 40% of axenic fifth instars successfully moulted to adulthood (Figure 4b), suggesting that B vitamins are partially responsible for the observed failures of axenic nymphs to reach adulthood. The effect of B vitamin supplementation on the growth and moulting success of gnotobiotic nymphs was limited, with only slight increases in moulting success or reductions in development time, though several of these were significant (Table S12). Together with the data from axenic nymphs, these data suggest that B vitamin provisioning by gut bacteria is probably necessary but not sufficient to support optimal insect growth and development.

4. DISCUSSION

Despite requiring microbes to enhance their fitness, multiple animals rely on acquisition of these partners from their environment. This has led to the evolution of mechanisms to ensure the appropriate microbe colonizes the host such as in the squid‐Vibrio fisheri symbiosis (Wang et al., 2010), while in other relationships the role of the microbiota is general enough that there is a high probability that environmentally‐acquired bacteria can fulfil the symbiotic role in the host (Coon et al., 2014). In other systems, the specificity of the relationship is less clear, and intermediate states where a limited set of microbes can function as symbionts may offer clues to whether specific host‐microbe relationships can evolve from less specific ones.

Initial studies of the kissing bug microbiome implemented culture‐dependent manipulations of kissing bugs and specific microbiota (Gumpert, 1962; Lake & Friend, 1967, 1968), but lacked molecular tools to verify axenic or gnotobiotic states, which may explain their conflicting results (Nyirady, 1973). More recent investigations of the kissing bug symbionts have utilized a culture‐independent 16S sequencing approach and have focused on characterizing the microbial community of various kissing bug species (Brown et al., 2020; Carels et al., 2017; da Mota et al., 2012; Díaz‐Sánchez et al., 2018; Dumonteil et al., 2018; Gumiel et al., 2015; Kieran et al., 2019; Mann et al., 2020; Montoya‐Porras et al., 2018; Oliveira et al., 2018; Rodríguez‐Ruano et al., 2018; Waltmann et al., 2019). While these studies have broadened our understanding of kissing bug microbiomes, comparisons between studies have been difficult due to different sampling techniques, species examined, use of laboratory versus wild‐caught insects, and other confounding variables. These studies are by design, descriptive, and provide limited inference as to which members are symbiotic or commensal, and what roles specific species play in the biology of the host. As a result, our understanding of the functional roles of microbes in the kissing bug gut remain poorly understood.

We produced bacteria‐free, axenic R. prolixus via egg surface sterilization and verified their axenic state using culture‐independent methods, then generated gnotobiotic insects by providing bacterial inocula through a blood meal. Our methods allowed a high degree of control over the kissing bug microbiome as well as a platform to conduct experiments on the roles of specific microbes in kissing bugs. Consistent with previous studies, axenic insects had high mortality and rarely reached adulthood. R. rhodnii‐gnotobiotic R. prolixus developed faster and with higher success than other gnotobiotic insects. Gnotobiotic insects inoculated with other Rhodococcus species and E. coli had increased developmental times and decreased survival, which was particularly pronounced in the final nymphal instar. However, many of the insects from these gnotobiotic colonies did eventually reach adulthood, though most laid significantly fewer eggs than insects harbouring R. rhodnii. In contrast, M. luteus proved to be a poor symbiont with no gnotobiotic insects surviving beyond the fourth instar. While R. rhodnii is not consistently recovered from 16S surveys of triatomine microbiomes, our data suggest that it confers a significant fitness advantage to R. prolixus.

A number of different microbes have been identified in the guts of kissing bugs in addition to R. rhodnii, largely from culture‐independent studies (Brown et al., 2020; Eberhard et al., 2022; Gumiel et al., 2015; Mann et al., 2020; Rodríguez‐Ruano et al., 2018; Tobias et al., 2020). Several groups have been consistently recovered from various species and disparate locations, suggesting that other bacteria can also function as symbionts of kissing bugs beyond R. rhodnii. Often, non‐Rhodococcus members of the closely‐related Corynebacteriales have been recovered from kissing bug guts, suggesting that other members of this family may also be able to function as symbionts of kissing bugs (Brown et al., 2020; Durvasula et al., 2008/5; Eberhard et al., 2022; Gumiel et al., 2015; Oliveira et al., 2018). At least one study has demonstrated that a Corynebacterium species can function as a symbiont of the kissing bug Triatoma infestans, though it was not compared to other bacteria (Durvasula et al., 2008). Our experiments did not test other identified symbionts as they were not available to us through microbial repositories, and so it is possible that other bacteria, especially in the Corynebacteriales, may have similar symbiotic potential as R. rhodnii. We are currently examining whether the fitness benefits of R. rhodnii to R. prolixus persist when additional microbes are present in the gut, and whether this fitness benefit extends to other species of kissing bugs.

Culture‐independent 16 S surveys of kissing bug microbiomes suggest that nymphs acquire bacteria from their environment which then persist in the insect midgut throughout development (Brown et al., 2020; Rodríguez‐Ruano et al., 2018). The ability of the bacterial species tested to persist in the kissing bug gut may reflect the fitness differences observed between gnotobiotic insects harbouring different bacteria. We found that many of the bacteria tested persist in the gut throughout development, but at significantly lower levels than R. rhodnii. Repeated inoculation of insects with a high titre of bacteria in every blood meal did result in increased survival and faster development, but ultimately did not rescue egg production. We suspected that lower titre of bacteria may have led to lower amounts of B vitamin provisioning by symbionts, as we found that supplementation of blood meals with B vitamins did increase survival and shorten developmental times. However, supplementation of B vitamins did not restore host fitness to the same degree that harbouring R. rhodnii without supplementation did. Our experiments did not examine whether dead bacteria are capable of rescuing development of axenic R. prolixus. If B vitamin or other nutrient supplementation is the predominant function of symbionts of kissing bugs, dead R. rhodnii or other symbionts may rescue axenic R. prolixus development, a possibility that will be tested in future studies.

Given that R. triatomae was isolated from another kissing bug species and that it was the closest known relative of R. rhodnii, we expected R. triatomae‐inoculated insects to have a fitness most like those inoculated with R. rhodnii. We found instead that gnotobiotic insects inoculated with R. triatomae were more similar to those harbouring other Rhodococcus species and not to R. rhodnii‐inoculated insects. R. rhodnii and R. triatomae have identical B vitamin synthesis pathways and share nearly 80% of their orthogroups. One possible explanation is the dramatic decrease in R. triatomae titre in R. prolixus, as fewer than 100 R. triatomae were recovered from late‐instar insects compared with >10⁶  R. rhodnii. We are currently investigating whether this is due to an inability of the bacteria to survive in the gut environment or factors such as immune activation by R. triatomae in R. prolixus.

We identified two gene clusters that were overrepresented in the unique genes of R. rhodnii, and both were associated with regulation of the lytic‐lysogenic cycle of bacteriophages. This led us to examine the presence of phage in the Rhodococcus genomes tested, and we discovered that R. rhodnii had three intact phages integrated into the genome. In contrast, R. triatomae, R. koreensis, and R. rhodochrous did not encode any complete phage while R. erythropolis shared the Rhodococcus “sleepyhead” phage (NC_048782.1) with R. rhodnii. Whether the phages of R. rhodnii contribute to its symbiotic abilities in R. prolixus is unknown, but in other symbiotic systems phage play a central role in symbiotic function. The phages of Hamiltonella defensa, a secondary symbiont of aphids, encode toxins that explain variation in the ability of different strains of the bacteria to kill attacking parasitoid wasps (Boyd et al., 2021; Brandt et al., 2017; Lynn‐Bell et al., 2019; Oliver et al., 2009).

Historically, the role of bacterial symbionts in the biology of kissing bugs has been debated, with some researchers suggesting that the necessity of symbionts for kissing bug development was not directly related to B vitamin provisioning (Hill et al., 1976), while others argued that nutrient provisioning was the primary function of symbionts in this system (Lake & Friend, 1968). Our results indicate that B vitamin provisioning is probably necessary when R. prolixus is fed on rabbits, though we did not test other sources of blood. However, there was substantial variation in symbiotic potential among the other bacteria tested, despite an apparently uniform ability to synthesize B vitamins. M. luteus cannot synthesize biotin, and this may explain why it failed to support R. prolixus development. Addition of B vitamins to blood meals of axenic insects did not fully rescue their development. One potential explanation for the variation in symbiotic potential despite similar B vitamin synthesis pathways is that the bacterial species tested varied dramatically in their abundance, leading to differences in the amount of B vitamins provisioned to the host. We intend to quantify specific B vitamin contributions in a future study. From these experiments we conclude that B vitamin provisioning is probably a necessary function of symbiotic bacteria in R. prolixus but is not alone sufficient. Studies using metagenomics approaches in R. prolixus with unmanipulated microbiomes suggest that a number of bacterial species may cooperate to provide the host with B vitamins (Tobias et al., 2020). Tobias and colleagues found that among the reads sequenced from a laboratory colony of R. prolixus, many of the B vitamin biosynthesis genes mapped to a species within the genus Dickeya, suggesting that in their colony, R. rhodnii was not the only bacterium synthesizing B vitamins for the host. Their results and the variable presence of R. rhodnii in field‐collected R. prolixus hint at potential cooperation or perhaps competition between symbionts in the R. prolixus gut.

Kissing bugs are also notable vectors of Trypanosoma cruzi, the causative agent of Chagas' disease. This neglected tropical disease affects millions of people, primarily in the Neotropics, though it is increasingly recognized as a public health threat in the Nearctic (Lynn et al., 2022). T. cruzi resides exclusively in the alimentary tract of kissing bugs, and therefore co‐occurs in the gut along with the microbiome. T. cruzi can establish during any phase of the insect's life and persist until the insects death (Vallejo et al., 2009). R. prolixus' lifespan can be several months long in the laboratory and potentially longer in the wild, giving a long window of interaction between gut bacteria and T. cruzi. Efforts were made to develop R. rhodnii into a biocontrol agent of T. cruzi by engineering the bacterium to express antitrypanosomal proteins that killed the parasite and demonstrated success in both laboratory and simulated field conditions (Durvasula et al., 1997, 1999). This work largely predated culture‐independent surveys of the kissing bug microbiome, though the researchers did demonstrate that R. rhodnii could establish in the gut of kissing bugs under semi‐natural conditions. Our results suggest this may be due to the ability of R. rhodnii to reach high titres in the guts of R. prolixus, but additional experiments are necessary to determine if this is true for other kissing bug species. Recent work has shown that various microbes in the gut of kissing bugs have antagonistic interactions with T. cruzi, though whether those bacteria are otherwise beneficial to the insect is unclear (Azambuja et al., 2004; da Mota et al., 2012; Gumiel et al., 2015). Others have demonstrated that the presence of T. cruzi alters the host gut microbiome, possibly through activation of the host immune system (Eberhard et al., 2022; Mann et al., 2020; Vieira et al., 2018). Though the precise interaction of the gut microbiota and T. cruzi is not fully understood, ongoing studies in our laboratory and others will continue to delineate the nature of this tripartite interaction.

Kissing bugs and their gut microbes represent an important intermediate between the obligate, intracellular symbionts and the environmentally acquired gut communities. The extremely long association of intracellular symbionts and their hosts is in part due to strict vertical transmission, often via the egg or direct provisioning to developing insect. Although comparative genomics between symbiont lineages can inform how the symbiont has evolved in different lineages, the long association has obscured the initial stages of symbiosis and the events leading to these relationships are unknown (Bennett et al., 2015). On the other extreme, the gut communities of mosquito larvae are essential to development of the host but appear to be largely nonspecific (Coon et al., 2014; Coon & Strand, 2020).

Few systems between these extremes have been intensively studied and as a result, our understanding of how specific associations evolve from less specific ones has been neglected. Bugs in the genus Riptoris and Anasa acquire highly specific Burkholderia symbionts from their environment, which in turn influence the development of the insect gut and are essential for proper host development (Acevedo et al., 2021; Itoh et al., 2019; Kikuchi et al., 2020). Honey bees have a relatively simple core microbiota of five bacterial species that strongly influence host fitness (Martinson et al., 2011; Raymann & Moran, 2018). In both cases, symbionts can be cultured outside of the host and genetically manipulated to examine the role of symbiont genes in supporting host fitness. Along with bees, bean bugs, and squash bugs, kissing bugs present a system in which a symbiont can be easily separated from the host and various bacteria reintroduced to the insect. Our results highlight that R. rhodnii is a superior symbiont for R. prolixus, while other bacteria exhibit some ability to function as symbionts in this system, albeit with a significant fitness cost. This study provides a foundation from which we can explore the symbiont and host genes which are necessary to support symbiosis, and whether these mechanisms evolved convergently in other systems.

The role of symbiotic bacteria in kissing bugs is probably complex and multifaceted, with symbionts fulfilling nutritional as well as other roles in host physiology. Here, we attempted to identify individual bacterial strains that are capable of functioning as symbionts and determined that R. rhodnii was a superior symbiont to other species tested, despite conservation of B vitamin biosynthesis genes among tested species. While the association of kissing bugs and microbes has been known for nearly a century, the mechanisms which underpin this relationship remain understudied. With the application of modern molecular tools, the kissing bug symbiosis may reveal key transitions in the evolution of insect‐microbe associations.

AUTHOR CONTRIBUTIONS

Carissa A. Gilliland, Vilas Patel, and Kevin J. Vogel designed the research, Carissa A. Gilliland, Vilas Patel, Ashley C. McCormick, Bradley M. Mackett, and Kevin J. Vogel performed the research. Carissa A. Gilliland, Vilas Patel, and Kevin J. Vogel analysed the data. Kevin J. Vogel wrote the manuscript.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

Figure S1. Phylogeny of Rhodococcus species.

Figure S2. Acquisition experiments of bacterial strains tested.

Figure S3. Growth of bacterial strains in minimal media.

Table S1. Accession numbers of genomes and genes for phylogeny.

Table S2. Accessions of bacterial strains used.

Table S3. Primers used for quantification of bacteria.

Table S4. B vitamin concentrations for feeding experiments.

Table S5. Sample sizes of experiments.

Table S6. Moulting/Survival and developmental rate data.

Table S7. Cox proportional hazard model statistics for moult success and survival.

Table S8. Details of the two‐step hurdle model for egg laying.

Table S9. Pairwise statistics of egg laying data.

Table S10. B vitamin accession numbers.

Table S11. Survival of bacterial species after incubation in whole blood.

Table S12. Cox proportional hazard model statistics for B vitamin supplementation experiments.

Data S1. Alignment of genes for phylogeny.

Gilliland, C. A. , Patel, V. , McCormick, A. C. , Mackett, B. M. , & Vogel, K. J. (2023). Using axenic and gnotobiotic insects to examine the role of different microbes on the development and reproduction of the kissing bug Rhodnius prolixus (Hemiptera: Reduviidae). Molecular Ecology, 32, 920–935. 10.1111/mec.16800

DATA AVAILABILITY STATEMENT

Data Accessibility and Benefit‐Sharing: Genome sequence data from this experiment is available through NCBI (accession numbers PRJNA483464 and PRJNA606336). All data and scripts for analysis are available through the Dryad Repository at https://doi.org/10.5061/dryad.0zpc86714.