Plant-mediated interspecific horizontal transmission of an intracellular symbiont in insects

Intracellular reproductive manipulators, such as Candidatus Cardinium and Wolbachia are vertically transmitted to progeny but rarely show co-speciation with the host. In sap-feeding insects, plant tissues have been proposed as alternative horizontal routes of interspecific transmission, but experimental evidence is limited. Here we report results from experiments that show that Cardinium is horizontally transmitted between different phloem sap-feeding insect species through plants. Quantitative PCR and in situ hybridization experiments indicated that the leafhopper Scaphoideus titanus releases Cardinium from its salivary glands during feeding on both artificial media and grapevine leaves. Successional time-course feeding experiments with S. titanus initially fed sugar solutions or small areas of grapevine leaves followed by feeding by the phytoplasma vector Macrosteles quadripunctulatus or the grapevine feeder Empoasca vitis revealed that the symbionts were transmitted to both species. Explaining interspecific horizontal transmission through plants improves our understanding of how symbionts spread, their lifestyle and the symbiont-host intermixed evolutionary pattern.

Scientific RepoRts | 5:15811 | DOi: 10.1038/srep15811 symbiont Coriobacterium glomerans of the red firebug Pyrrhocoris apterus and the symbionts of the firebrat Thermobia domestica are acquired horizontally by juveniles through symbiont-containing eggshells, faeces, or adult bugs 21,22 . Rickettsia symbionts of the whitefly Bemisia tabaci can be acquired by its parasitoids 23 or by other whiteflies feeding on the same plant 24 .
Phylogenetic analyses of Wolbachia, a reproductive manipulator, and its associated hosts showed horizontal transmission between prey and predator mites 25 and between hosts and parasitoids 26 , as well as among insects sharing the same food source or parasites [27][28] . Direct demonstrations of intra-or inter-specific horizontal transmission of Wolbachia from the host to a parasitoid 29 , or in parasitoids sharing the same host eggs 30,31 , have been reported in few host models. Horizontal transmission has also been directly or indirectly observed in Arsenophonus nasoniae, which is transferred among parasitic wasps sharing the same hosts 32 . On the other hand, in the case of Cardinium, there has only been speculation on possible horizontal transmission 9,10,13,16,[33][34][35] .
Here we demonstrate that an inter-specific horizontal transmission of Cardinium can take place in leafhoppers and that such a transmission occurs through the plant tissue pierced by the insect host. By using the cicadellid Scaphoideus titanus, a Cardinium-holder strictly feeding on grapevine leaves, as a model, we show that Cardinium is released from the insect's salivary glands to the plant tissues and then horizontally acquired by other grapevine feeder and non-grapevine feeder insects.

Results
Cardinium release in artificial feeding media and plant tissue. The first set of experiments was conducted to assess the capability of S. titanus to release Cardinium during feeding on an artificial substrate (a feeding medium consisting of a sterile sugar solution) or on grapevine leaves. S. titanus individuals used in the inoculation experiments without antibiotic treatment were found to be highly colonized by Cardinium. The infection rate and the concentration of Cardinium cells were very similar in insects fed on the artificial media and the grapevine leaves (Table 1). In rifampicin-treated insects fed on both substrates, Cardinium was found as well, and the average infection rate and density of symbiont cells per leafhopper were not significantly lower than those in untreated specimens. However, according to the Cardinium-to-bacteria ratio (CBR), the percentage of Cardinium in the whole leafhopper bacterial community after the rifampicin treatment was drastically decreased in both experimental settings (Table 1).
Cardinium was consistently found to colonize S. titanus individuals, which were found to be able to inoculate symbiont cells in either feeding substrate and in either the absence or the presence of antibiotic treatment. However, both inoculation/infection rates and titres decreased in the insects treated with rifampicin compared with those that were untreated regardless of the feeding substrate. These values consistently declined after antibiotic treatment in the sugar diet and the grapevine diet, and a statistical significance was detected between samples with or without rifampicin. The CBR calculated for the diet samples followed a similar decrease (Table 1).
While grapevine leaves collected from S. titanus-free plants were always negative for the bacterium, 34.3% of grapevine leaves collected from plants in the field infested by the leafhopper were Cardinium-positive (Table 1). Even so, the average infection rate and titre were not significantly higher than the average infection rate and titre of grapevine leaves exposed to antibiotic-treated S. titanus.
We tested some of the salivary glands of S. titanus and the artificial feeding media and grape leaves using in situ hybridization (ISH) and fluorescence in situ hybridization (FISH). Hybridization signals related to the Cardinium probes were detected (Figs 1E-H, 2). TEM observations of the salivary glands of S. titanus confirmed the presence of bacterial cells with microtubular-like complexes typical of Cardinium 36 (Fig. 1D). Moreover, FISH experiments on grapevine leaves collected in S. titanus-infested vineyards in the field also indicated the presence of Cardinium bacterial cells (Fig. 2B). On the other hand, when antibiotic-treated leafhoppers and their respective feeding substrates were tested by FISH, no hybridization signal was detected for Cardinium. Negative controls exhibited a lack of hybridization as well (Fig. S3).
Cardinium transmission experiments. Following the demonstration of the capability of S. titanus to inoculate Cardinium cells in its feeding medium, we tested if other insects could then acquire Cardinium from the feeding medium. We utilized the leafhoppers Macrosteles quadripunctulatus and Empoasca vitis as receiver hosts. Prior to the experiment, we assessed the Cardinium infection status of the recipient species. We did not detect the bacterium neither in M. quadripunctulatus nor in E. vitis individuals that had never been exposed to S. titanus's feeding media. Transmission experiments via artificial feeding media (M. quadripunctulatus) or grapevine leaves (E. vitis) were then conducted. Cardinium was initially detected by qPCR in all S. titanus donor individuals. Moreover, in both cases, a transfer of Cardinium cells to M. quadripunctulatus and E. vitis was detected either by qPCR or by FISH experiments (Table 2; Fig. 3).
In the M. quadripunctulatus experiments, the Cardinium infection rate in recipients was initially about 50% (after one and three days of co-feeding) and then increased after one week. However, the average titres of Cardinium in positive individuals were modest with respect to donors in all experiments (with a peak after three days of co-feeding). The CBR in the recipient M. quadripunctulatus was low as well, following a similar trend of the titres ( In the E. vitis experiments, both infection rates and concentrations were quite variable, with a peak after one day of co-feeding, a decrease after three days and a new increment after seven days. Interestingly, although the Cardinium concentration decreased in the seven-day trial, the CBR was higher for the longer co-feeding time; furthermore, Cardinium remained above 1% of total bacteria and always represented a higher percentage of the microbial community in E. vitis compared with M. quadripunctulatus ( Table 2).
The transmission was confirmed by FISH and sequencing. Indeed, a massive hybridization signal correlated with Cardinium was detected in the midguts of M. quadripunctulatus and E. vitis after three days of co-feeding (Fig. 3A,C), whereas the negative controls were free of fluorescence signals related to probe hybridization (Fig. S4). Moreover, 835 bp amplicons were sequenced after the Cardinium-specific PCR was performed on both M. quadripunctulatus and E. vitis used in the co-feeding experiments. The sequences were identical to the deposited sequences of the Cardinium symbiont of S. titanus (AM042540), clustering in the A-group 1 (Fig. S5).

Discussion
This study sought to verify if S. titanus is able to inoculate Cardinium into the surrounding feeding substrate during its feeding activity and then if Cardinium in the feeding substrate can be taken up by other insects. By means of qPCR and FISH results, we firstly demonstrated that S. titanus injects the symbiont into its feeding substrates, both under controlled conditions (artificial diet) and in a semi-natural environment (grapevine leaves) ( Table 1, Fig. 2). High concentration values of Cardinium were observed in the leafhoppers and in their artificial diet, in agreement with previous data 16 . Additionally, the detection of Cardinium in grapevine leaves, used for feeding by insects infected by the symbiont, suggests that Cardinium can spend part of its life cycle in grapevine leaf tissues, exploiting the plant as a means to be transmitted among S. titanus populations, as previously observed in other plant sap-feeders 17 suggests that the leafhopper regularly releases Cardinium into the plant tissues, although infection rates and symbiont densities observed in the field were lower than those detected in grapevine leaves experimentally exposed to the leafhopper. This finding opens questions on the nature of the interaction between Cardinium and grapevine leaves, as well as the occurrence of the symbiont in other plants hosting sap-feeding insects, in light of the wide diffusion of this bacterial symbiont among arthropods 9,37 . However, when S. titanus individuals were treated with rifampicin, we were not able to obtain completely Cardinium-free specimens. Moreover, these antibiotic-treated insects were still able to inoculate Cardinium into their feeding substrates ( Table 1). The use of antibiotics was demonstrated to inhibit the occurrence of many symbionts completely [38][39][40] even after a few days of administration 41,42 . On the contrary, symbiotic bacteria in some cases could still be found in their hosts after antibiotic treatment 42,43 . A reduction of the Cardinium titre by about one order of magnitude after rifampicin treatment was not as high as previously reported by Pajoro et al. 16 , while it was in the same range of the results obtained by Boucias and colleagues 43 for Burkholderia in the heteropteran Blissus insularis Barber. Due to the impossibility of obtaining Cardinium-cured S. titanus, we employed different Cardinium-free leafhopper species as recipients to demonstrate the symbiont acquisition step. Both M. quadripunctulatus and E. vitis were able to acquire Cardinium that then persisted in the two recipient species (Table 2, Fig. 3).
M. quadripunctultus, which shared the artificial sugar diet with S. titanus, was stably colonized by Cardinium after feeding. Indeed, the symbiont infection rates observed during these transmission experiments increased at the longest time of exposure, indicating that longer co-feeding periods result in greater acquisition by the recipient. On the other hand, the Cardinium concentration in M. quadripunctulatus was always low and never reached values comparable with those detected in S. titanus, indicating that longer periods may be necessary to achieve stable colonization. It would be of great interest to elucidate whether Cardinium is able to persist in the new host and finally to be maternally transmitted. Moreover, because M. quadripunctulatus is known to be a very efficient phytoplasma vector 44,45 , this species may be particularly suitable for bacterial invasion of the salivary glands, leading to high transmission competence. The observation of the acquisition of Cardinium by M. quadripunctulatus raises a question about  the capability of newly colonized insect individuals to transmit the acquired symbiont subsequently in the feeding medium and possibly to other insects. Varying results were obtained both in terms of infection rates and density in the E. vitis experiments. After one day of co-feeding, a high percentage of the recipient leafhoppers exhibited considerable amounts of Cardinium uptake. However, after three days, the infection rates and concentrations declined and then increased again after seven days ( Table 2). This variability suggests that the success of E. vitis colonization by Cardinium may be due to chance more than to a solid capability of the symbiont to invade the tissues of this insect. Hence, even though we demonstrated that at least in some cases the ingestion of Cardinium can be followed by successful colonization, the symbiont is likely to be infrequently established in E. vitis. Nonetheless, the symbiont concentration in Cardinium-positive E. vitis was always higher than that observed in M. quadripunctulatus, and it even reached values as high as those detected for S. titanus (Table 2). This evidence indicates that the tissues of grapevine leaves are suitable media for the acquisition of high titres of this bacterium, possibly because Cardinium is concentrated in the phloem of leaves and is more likely to be ingested than Cardinium from the artificial diet in which the bacteria are dispersed throughout the sugar solution. Since we never found Cardinium-positive field-collected E. vitis, the transmission from S. titanus to this leafhopper via grapevine leaves is probably not persistent, and the bacterium may not find optimal conditions in the new host to be then vertically transferred to its progeny. However, given that Cardinium was detected in grapevine leaves in the field, it is apparent that this plant may be a continuous reservoir of the symbiont, at least for S. titanus, and acquisition of this bacterium by other phloem-feeding insects residing on grapevine leaves cannot be excluded.
The overall results provide direct demonstration of the horizontal transmission of Cardinium in phloem-feeding leafhoppers via co-feeding, previously widely hypothesized 9,34,35 but never demonstrated for this symbiont. The data presented demonstrate and support previous molecular evidence that horizontal transfer of symbionts does actually occur and may contribute to symbiont spread in natural insect lineages 9,28,35 . Even though maternal transmission is the main diffusion system for reproductive manipulators such as Cardinium, we showed that horizontal transmission through plants provides an alternative route of spread of these symbionts. This finding has major implications for our understanding of the evolutionary history of Cardinium symbionts and suggests that these bacteria may have adapted to living in plant tissues outside the insect host. The genome of the Cardinium endosymbiont of the whitefly Bemisia tabaci (a phloem feeder like S. titanus) showed peculiar traits like genes associated with gliding motility, which may be responsible for the colonization of new hosts 46  All experiments were conducted in accordance with the legislation and guidelines of the European Union for the protection of animals used for scientific purposes (http://ec.europa.eu/environment/chemicals/lab_animals/legislation_en.htm). All experimental protocols using animals were approved by the ad-hoc Committee of DISAFA of the University of Turin. In addition, all necessary permits were in hand when the research was conducted.

Cardinium inoculation experiments through artificial and plant diets.
To observe-under controlled conditions-the injection of bacterial cells by S. titanus while feeding, artificial feeding systems were set up (Fig. S2A). A total of 100 newly emerged adults were maintained on the artificial diet for three days according to 16 (see also SI material and methods). Half of the specimens (50 insects) were reared on the artificial diet supplemented with 300 μ g ml −1 rifampicin, known to be active against Cardinium 9,16 .
To confirm bacterial release under semi-natural conditions, a cage system on the grapevine leaves was set up 16,47 with small plastic insect chambers placed on the surfaces of the leaves (Fig. S2B) to force a total of 90 S. titanus to feed on small areas of the plants for three days (see also SI material and methods). Half of the leaves were supplied with 300 μ g ml −1 rifampicin together with a nutritive solution, and the leafhoppers were exposed to these treated leaves for three days.
A total of 35 leaves from grapevine seedlings never exposed to S. titanus were used as the control in the molecular analyses. Furthermore, to check for grapevine infection by Cardinium in the field, 45 leaves (35 for qPCR and 10 leaf midribs for FISH) from different Barbera plants in Piedmont vineyards with high S. titanus incidence were also tested.
Cardinium transmission tests. Two cicadellid species, the phytoplasma vector M. quadripunctulatus and the grapevine feeder E. vitis, were used to evaluate the actual transfer of the symbiont to recipient insects after the inoculation by S. titanus during feeding events (see also SI material and methods). Twenty-five M. quadripunctulatus and 25 E. vitis were firstly checked by PCR and qPCR (with Card192F/1069R and EndoF1/R3 primer pairs, respectively, as described in SI material and methods) for the actual absence of Cardinium in the used populations to verify whether these species were suitable for transmission experiments.
The horizontal transfer of Cardinium to M. quadripunctulatus was assessed via feeding experiments in artificial media chambers (Fig. S2C). Adult donor individuals of S. titanus were maintained for three days in groups of 15 insects on sugar solutions in suitable feeding chambers to allow the symbiont to be released in the supplied substrate. They were then replaced by groups of 15 M. quadripunctulatus adults maintained for different acquisition periods: one day, three days, and seven days.
The transmission of Cardinium to E. vitis occurred via the grapevine leaves. Single donor S. titanus specimens were forced to feed for three days on portions of Barbera grapevine leaves for the release of bacterial cells into the plant tissues and then removed and preserved for subsequent analyses. The same leaf portions were then exposed to E. vitis individuals to allow bacterium acquisition for one day, three days, and seven days.
DNA extraction and PCR-based analyses. Subsequent to the inoculation experiments, total DNA was extracted from the leafhoppers and the respective sugar diets and the leaf portions for molecular analyses. Nucleic acid extraction from the insects and artificial diet was performed as previously Scientific RepoRts | 5:15811 | DOi: 10.1038/srep15811 reported 47,48 , whereas plant DNA was extracted from the leaf portions previously ground with liquid nitrogen in a sterile mortar, according to DNeasy Plant Mini Kit protocol (Qiagen, Italy) instructions.
Quantitative real-time PCR was performed to measure the presence and concentration of Cardinium cells in the insects, artificial diet and leaves. In all samples, reactions targeting the 16S rRNA gene of Cardinium were carried out. In addition, insect DNA was submitted to qPCR targeting the insect's 18S rRNA gene to normalize the absolute Cardinium density. Furthermore, insect and artificial diet DNA (but not the grapevine DNA due to a non-specific reaction with the plastids of bacterial universal primers) was used in reactions with universal bacterial primers to define the overall bacterial concentration in each sample. The CBR was then calculated to estimate the relative abundances of Cardinium in the bacterial community associated with the leafhoppers or with colonizing the diet. qPCR conditions are reported in the SI material and methods.
Subsequent to qPCR screening, 10 Cardinium-positive M. quadripunctulatus and E. vitis individuals were used for PCR screening (SI materials and methods). The obtained PCR products were purified and sequenced (Genechron, Rome, Italy), and the resulting sequences were compared with those in the National Center for Biotechnology Information (NCBI) sequence database by using BLAST (http:// www.ncbi.nlm.nih.gov/blast). A phylogenetic analysis was performed using the MEGA 6 software by the neighbor joining method.
Transmission electron microscopy and in situ hybridization. Transmission electron microscopy was carried out on S. titanus individuals after dissection in saline solution, as previously described 49 . ISH and FISH analyses were performed on the salivary glands of S. titanus, whereas the artificial feeding media, grapevine leaves, and Cardinum-recipient M. quadripunctulatus and E. vitis individuals were analyzed by FISH only.
ISH experiments were carried out by using specific oligonucleotide probes, Card172 and Card1069 50 , 5′ labelled with digoxigenin (DIG). FISH experiments performed on insect tissues and feeding media were carried out with Cardinium-specific probes along with a universal bacterial probe, which was employed as a positive control for the hybridization experiment. Conversely, for experiments on plant tissues only, Cardinium-specific probes were employed, due to non-specific hybridization of the eubacterial probe with the plastids. For the complete procedures, see also SI materials and methods.