Wolbachia affect behavior and possibly reproductive compatibility but not thermoresistance, fecundity, and morphology in a novel transinfected host, Drosophila nigrosparsa

Abstract Wolbachia, intracellular endosymbionts, are estimated to infect about half of all arthropod species. These bacteria manipulate their hosts in various ways for their maximum benefits. The rising global temperature may accelerate species migration, and thus, horizontal transfer of Wolbachia may occur across species previously not in contact. We transinfected and then cured the alpine fly Drosophila nigrosparsa with Wolbachia strain wMel to study its effects on this species. We found low Wolbachia titer, possibly cytoplasmic incompatibility, and an increase in locomotion of both infected larvae and adults compared with cured ones. However, no change in fecundity, no impact on heat and cold tolerance, and no change in wing morphology were observed. Although Wolbachia increased locomotor activities in this species, we conclude that D. nigrosparsa may not benefit from the infection. Still, D. nigrosparsa can serve as a host for Wolbachia because vertical transmission is possible but may not be as high as in the native host of wMel, Drosophila melanogaster.

embryonic death. It has been proposed that cytoplasmic incompatibility can promote host speciation by inducing reproductive barriers when the same host species hosts multiple, incompatible strains (Sinkins et al., 2005).
Wolbachia also have been shown to affect the morphology of their arthropod hosts, for example, in wing size and shape (Dutra et al., 2016;Kriesner, Conner, Weeks, Turelli, & Hoffmann, 2016) and larva size (Dutra et al., 2016). Depending on the particular hoststrain interaction, host animals can also benefit from Wolbachia infection. wMel-infected Drosophila melanogaster were reported to have higher fecundity, higher mating rate, and longer wings (Table 1) than uninfected individuals. Laodelphax striatellus planthoppers infected with wStri also had higher fecundity than uninfected ones (Guo et al., 2018). Bigger body size and longer life span were reported in Callosobruchus chinensis beetles infected with wBruCon, wBruOri, and wBruAus (Okayama, Katsuki, Sumida, & Okada, 2016).

| MATERIAL S AND ME THODS
For flies of all lines used in this study, around 50 adult males and 50 adult females were put in a mating cage modified from Kinzner et al. (2018) and supplied with grape juice agar, malt food, and fresh yeast for embryo collection. Food was changed every 5 days.
Embryos and larvae were collected and transferred to glass vials filled with malt food at a density of around 80 eggs per vial. All flies were reared at 19°C, 70% humidity, and a 16 hr:8 hr light:dark cycle.

| Curing from Wolbachia
In Generation 14 after transinfection, two subpopulations of each stably Wolbachia-infected fly line were treated with tetracycline hydrochloride (lot number SLBQ2368V, Sigma-Aldrich, Germany) mixed in the malt food at final concentrations of 0.01% (Miller, Ehrman, & Schneider, 2010) or 0.05% (Schneider et al., 2013). After three generations of treatment, flies were transferred to normal malt food for another two generations to eliminate effects of tetracycline (Ballard & Melvin, 2007;Chatzispyrou, Held, Mouchiroud, Auwerx, & Houtkooper, 2015). Three cured lines, namely nc_3, nc_6, and nc_8, were generated. Five female flies of every cured generation were randomly collected and checked for Wolbachia infection by PCR using the primers wsp81F and 691R.

| Cytoplasmic incompatibility test and fecundity
The cytoplasmic incompatibility level was assessed at Generation 19 by crossing infected, cured, and uninfected flies in all possible combinations except crosses between infected and cured flies. Five one-day-old virgin females were allowed to mate with five males of the same age from a different line in a mating cage, three cages per cross. Drosophila nigrosparsa females start laying eggs 7 days after their first mating, and the larvae hatch 2 days after egg laying (data not shown). Thus, flies were allowed to mate for 7 days. Males were removed on the eighth day, and each female was individualized into a

| Critical maximum and minimum and heat knockdown temperatures
Critical temperature experiments were modified from Kinzner et al. (2018). In Generation 19, seven-day-old female flies of infected (ni_3, ni_6, and ni_8), cured (nc_3, nc_6, and nc_8), and uninfected (nu_0) lines were used. Flies were separated under carbon dioxide anesthesia 2 days before the experiments. On the days of experiments, flies were placed at room temperature for 1 hr before the experiments started and were transferred to 5-ml vials without anesthesia immediately before the experiments.
For the critical maximum and minimum temperature assays (CTmax and CTmin, respectively), three females from the same line were transferred into a 5-ml vial, four vials per temperature. The fly-containing vials were sealed and exposed in a water bath for 5 min to six different temperatures from 37 to 39°C for CTmax and 0.5 to 3.5°C for CTmin with 0.5°C intervals. Temperatures from the thermostat reservoir (VWR, USA) and from a thermometer (Ebro TFX430, Xylem Analytics) inside the control vial were recorded with an accuracy of 0.05°C. After 5 min, the vials were removed from the water bath, and the flies were checked quickly for coma by tapping the vials. Flies were discarded after each run.
For the heat knockdown assay, three females were transferred into a 5-ml vial, four replicates per line. The vials were sealed and submerged in a transparent water bath with continuously increasing temperature from 25°C to 39°C at a rate of 0.47°C/min. Temperature was measured as described above. The number of flies in coma and the temperature inside the vials were recorded throughout the assay every 30 s.

The percentages of flies in coma in each vial of the CTmax and
CTmin experiments were used to calculate generalized linear mixed models by maximum likelihood with a binomial error structure and logit link function of flies in coma against temperature. For heat knockdown, the temperature of each fly that was in coma was used.

Analysis of covariance (ANCOVA) between infected and cured lines
and t test between infected and their cured lines were performed.
Bonferroni correction for multiple comparisons was used.

| Locomotion
In Generation 19, 20 larvae at the age of 5 days from each infected and cured line and 31 larvae from the uninfected line were randomly collected. The experimental setup for assessing larval mobility was modified from Brooks, Vishal, Kawakami, Bouyain, and Geisbrecht (2016). Briefly, each larva was put on 2% agarose in a 55 mm petri dish over a light pad (A4 Light Box, M.Way, China). The order of lines scored was randomized, and all larvae were recorded at the same time of the day (9-12 hr). The crawling path of each larva was recorded for 3 min using a video camera (XR155 Full HD, Sony, Japan). Total crawling distance (mm) and mean speed (mm/s) were analyzed using wrMTrck plugin (Nussbaum-Krammer, Neto, Brielmann, Pedersen, & Morimoto, 2015) implemented in Fiji (Schindelin et al., 2012), a version of ImageJ (Schneider, Rasband, & Eliceiri, 2012) with slight modifications as described by Brooks et al., (2016).

The adult locomotion experiment Rapid Iterative Negative
Geotaxis (RING) was modified from (Gargano, Martin, Bhandari, & Grotewiel, 2005). In Generation 19, 14-day-old females from infected, cured, and uninfected lines were anesthetized with carbon dioxide for sexing and separated 2 days before the experiment. Ten female adults from each infected and cured line and 28 female adults from the uninfected line were used. Each female was transferred into a vial (100 × 24 × 1 mm, Scherf-Präzision Europa) and placed at room temperature an hour before the experiment. Fly-containing vials were tapped quickly so that all flies fell to the bottom, and locomotion activities (jumping and walking) were video recorded using a video camera (XR155 Full HD) for 3 min. All lines were included in each run, and the fly-containing vials were randomly placed inside the RING apparatus. All glass vials used were cleaned with heptane 2 days before the experiment. Stack images of the recorded videos were used for analysis using Fiji (Schindelin et al., 2012). The numbers of jumps and walks of each fly were counted manually.
For both larval and adult locomotion, nested ANOVAs were used to test for differences among lines within infection status. F tests and t tests were used to test between infected and their corresponding cured lines. MorphoJ version 1.06d (Klingenberg, 2011) was used to process the tps file. Images were aligned by the principal axis. Outliers were performed. Asymmetry on size and shape between left and right wings between infected and cured lines was calculated as previously described (Padró, Carreira, Corio, Hasson, & Soto, 2014). In brief, Pearson correlations were used between mean individual wing size and the difference between left and right wings for size asymmetry,

| Wing geometric morphometrics
and Procrustes ANOVA of wing shape was used for shape asymmetry.

| RE SULTS
Drosophila nigrosparsa line nu_0 was found to be not infected with Wolbachia, Cardinium, Spiroplasma, and Rickettsia before the start of our experiments.

| Quantification of Wolbachia
The Wolbachia titer of all infected lines of Generation 12 was generally low. We observed, on average (mean ± standard error), 0.04 ± 0.01, 0.06 ± 0.01, and 0.06 ± 0.01 Wolbachia genomes per fly genome in the first 13 days for ni_3, ni_6, and ni_8, respectively (n = 21 per line). Wolbachia titer increased and reached the highest density after the second week ( Figure 2). In general, line ni_8 had lower Wolbachia titer than lines ni_3 and ni_6. Because of this high variation in Wolbachia titer, we refrained from statistical tests for differences across lines and treatments.

| Curing from Wolbachia
We did not detect Wolbachia with PCR during the treatment with 0.01 or 0.05% concentrations of tetracycline. However, we detected Wolbachia in all lines in the first generation after having stopped treating the flies with 0.01% tetracycline. Wolbachia were successfully removed with 0.05% tetracycline. The third generation of flies after treatment with 0.05% tetracycline was used for further experiments.

| Cytoplasmic incompatibility and fecundity
Each female laid on average (mean ± standard error) between 9.3 ± 3.5 and 15.7 ± 4.6 eggs for crosses between infected lines, 8.2 ± 2.0 and 13.0 ± 2.7 eggs between cured lines, and 12.1 ± 1.6 eggs for unin- Crosses between infected males and females yielded similar percent hatch per cross to those between uninfected flies (mean ± standard error: 73.6 ± 6.1% and 84.7 ± 8.9%, respectively). Hatch rate dropped from 60.7 ± 5.4% in crosses of uninfected males with infected females (expected compatible cross) to just 37.7 ± 3.8% in crosses of infected males with uninfected females (expected incompatible cross) (Figure 3), but these two groups were not significantly different (generalized linear models; z = −1.35, p = .18).

| Critical maximum and minimum and heat knockdown temperatures
For CTmax and CTmin, generalized linear models of the numbers of flies in coma against temperatures were significant in many lines (

| Locomotion
For larval locomotion, infected line ni_8 had the highest mean crawling speed and the longest mean distance (Figure 5a,b, respectively).
In addition, we found no difference between cured and uninfected lines in any of the two larval activities (nested ANOVA; F 1,2 = 0.44, p = .58 and F 1,2 = 1.40, p = .36 for average speed and total length, respectively).
In adults, infected lines had higher activities than cured lines (Figure 5c,

| Wing geometric morphometrics
The imaging of wings can be assessed as done accurately, and in that the mean squares of imaging error were very low for both centroid size and shape (2.75 and 4.54 times lower than individual by side interactions for centroid size and shape, respectively). We

F I G U R E 3
Number of eggs laid (a) and percent egg hatch of crossing between male and female of each group (b). Flies were allowed to mate for 8 days, and the numbers of eggs and of hatched larvae were counted on Day 9 and Day 14, respectively. The numbers of eggs laid were not significantly different among crosses. The hatch rates were reduced in crosses of uninfected males with infected females compared with crosses of infected males with uninfected females to each other (Figure 6a).
There was no difference in average shape of all cured and infected lines (Figure 6b). When comparing infected and its corresponding cured lines, we observed significant changes in centroid size and shape between ni_3 and nc_3 (size, F 1,1 = 509.87, p = .03; shape, F 22,22 = 104.27, p < .01) and ni_6 and nc_6 (size, F 1,1 = 4,815.15, p = .01; shape, F 22,22 = 26.99, p < .01), and significant difference in shape between ni_8 and nc_8 (shape, F 22,22 = 2.78, p = .01). Centroid size and shape of infected and cured lines differed significantly from naturally uninfected line nu_0 (p < .05). However, there was a small distance between groups relative to within-group variation (Mahalanobis distance = 1.20), and most flies were assigned into wrong groups. There was no significant difference in size and shape asymmetry between left and right wings (p > .05).

| D ISCUSS I ON
Transinfection is a useful tool to investigate effects of Wolbachia on new host species (Hughes & Rasgon, 2014), and in that it provides possibility to study a broad range of phenotypic effects on the host. We used transinfection to study effects of Wolbachia on  (Table 1), and, within a host, titers vary among tissues such that, for example, higher titers were observed in reproductive than in somatic tissues (Martinez et al., 2015;Osborne, Iturbe-Ormaetxe, Brownlie, O'Neill, & Johnson, 2012). In addition, Wolbachia titer might be higher if D. nigrosparsa was raised at a temperature cooler than 19°C, as in our experiment, because

D. nigrosparsa because this fly species may become infected by
higher Wolbachia density was detected in D. melanogaster developed at cool temperatures than those developed at warm temperatures (Moghadam et al., 2018).
Titer can also change with host age as observed in many arthropods including Drosophila spp. (Chrostek et al., 2013;McGraw et al., 2002;Tortosa et al., 2010;Unckless, Boelio, Herren, & Jaenike, 2009). The Wolbachia titer we observed (Figure 2) is likely to correlate with egg-laying activity in D. nigrosparsa, which was reported to peak between the second and the fourth week (Kinzner et al., 2018). As Wolbachia are mainly found within host's reproductive tissues (Frydman, Li, Robson, & Wieschaus, 2006;Werren, 1997), the declining of Wolbachia titer when the flies neared completion of their fourth week could be explained by the declining of germline stem cell division with increasing individual age (Zhao, Xuan, Li, & Xi, 2008).
To cure D. nigrosparsa from Wolbachia, we tried two tetracycline concentrations, 0.01 and 0.05%. High tetracycline concentration has been reported to have negative fitness effects on hosts during the process of curing (Miller et al., 2010), and lower concentrations should therefore be preferred. However, the 0.01% concentration was too low to eliminate Wolbachia, in line with observations made on Wolbachia-infected Drosophila paulistorum (Miller et al., 2010). In addition, both D. nigrosparsa treated with 0.01% and 0.05% tetracycline suffered from low fecundity and low hatch rates (data not shown). We waited for another two generations before using them for our remaining experiments to recover flies from tetracycline because effects of tetracycline on mitochondrial density and metabolism can last up to two generations after treatment (Ballard & Melvin, 2007).
We note that the recovering time of hosts after antibiotic treatment is important. Effects of antibiotics on the fly hosts were eliminated entirely within a few generations after treatment (Chaplinska, Gerritsma, Dini-Andreote, Falcao Salles, & Wertheim, 2016;Fry, Palmer, & Rand, 2004). However, after five generations, the effects of antibiotics were not fully eliminated in D. simulans (Poinsot & Mercot, 1997). To better evaluate potential effects of antibiotics on D. nigrosparsa, comparisons over multiple generations between uninfected flies never treated with tetracycline and uninfected flies after tetracycline treatment should be done.
Cytoplasmic incompatibility is the most commonly observed phenotype of Wolbachia on their hosts (Werren et al., 2008). Despite No significant differences in average shape of all cured (green) and infected lines (pink) using discriminant analysis. The differences were magnified ten times, and all thirteen landmarks are shown were reduced, although there was no difference in the number of eggs laid. Increasing the number of compatible and incompatible crosses would be needed to decide whether the lack of statistical significance in the data presented here is due to the lack of a biological effect or due to the effect being just weak; for technical reasons, additional crosses are impractical at the point of writing this manuscript. In contrast to our results, Wolbachia wMel, once transinfected into other hosts, induced a high level of incompatibility, such as in Drosophila simulans (Poinsot, Bourtzis, Markakis, Savakis, & Merçot, 1998), in the whitefly Bemisia tabaci (Zhou & Li, 2016) and in the mosquito Aedes aegypti (Hoffmann, Iturbe-Ormaetxe, et al., 2014;Hoffmann, Coy, Gibbard, & Pelz-Stelinski, 2014;Walker et al., 2011) ( Table 1).
The levels of cytoplasmic incompatibility depend on many factors. A high level of cytoplasmic incompatibility has been reported to positively correlate with high Wolbachia titer (Bourtzis, Nirgianaki, Markakis, & Savakis, 1996;Noda, Koizumi, Zhang, & Deng, 2001;Noda, Miyoshi, et al., 2001). Young males and a high number of infected sperms also caused high level of cytoplasmic incompatibility Reynolds & Hoffmann, 2002;Veneti et al., 2003). For example, Reynolds and Hoffmann, (2002) found a lower level of incompatibility when using 5-day-old males for crossing than using 1-day-old males.
The ability to adapt to elevated temperatures is an important criterion for species distribution in Drosophila (Kellermann et al., 2012).
A previous study found no effect on heat knockdown temperature in wMel-infected Drosophila melanogaster (Harcombe & Hoffmann, 2004). This finding for Wolbachia contrasts one for Rickettsia, which were reported to increase heat shock tolerance in Bemisia tabaci to up to 40°C (Brumin et al., 2011). In D. nigrosparsa, a recent selection experiment on naturally uninfected flies reported that this species is unlikely to adapt to increasing temperature .
Here, we conclude that infection with Wolbachia wMel did not increase heat and cold tolerance in D. nigrosparsa. Wolbachia-infected and Wolbachia-free D. nigrosparsa responded to knockdown temperature at around 38°C like in an earlier study of this fly species (Kinzner et al., 2018). Thus, we cannot expect a rescue from heat stress due to infection by the Wolbachia strain used here in D. nigrosparsa. We note that the absolute value of knockdown depends on ramping speed and that it has been a topic of debate what ramping speed to use (Santos, Castañeda, & Rezende, 2011) but that in the frame of this study not absolute knockdown but the performance of infected flies relative to that of uninfected and cured flies was important.
Thermal tolerance is one of the many aspects in thermal biol- Beetles Callosobruchus chinensis infected with Wolbachia wBruCon and wBruOri walked significantly more than uninfected ones, which might help increase their chance for mating (Okayama et al., 2016).
Mosquitoes Aedes aegypti infected with wMelPop had up to 2.5-fold increase in activity compared with uninfected ones (Evans et al., 2009).
We found significant differences in wing size and shape of D. nigrosparsa between infected and cured lines, but these differences were more likely due to genetic drift and not due to Wolbachia as the cured lines were subpopulations of infected lines and had been separated from their parent populations for five generations before the wing measurement. Although genetic variation in isofemale lines is highly reduced, morphological variation still was observed, for example, in Drosophila buzzatii and in Drosophila koepferae (Carreira, Soto, Hasson, & Fanara, 2006) and in D. melanogaster (Bubliy, Loeschcke, & Imasheva, 2001;Imasheva, Bosenko, & Bubli, 1999). Effects of genetic drift in Drosophila can occur within a few generations, for example, in Drosophila subobscura (Santos et al., 2013). In addition, if Wolbachia affect wing morphology, we would observe similar changes in those cured lines once Wolbachia were removed. Nevertheless, differences in the microbiome may have contributed to the changes in morphology we observed like was shown in D. melanogaster (Broderick, Buchon, & Lemaitre, 2014).
Our study indicated that D. nigrosparsa could be a host for Wolbachia like Drosophila melanogaster, the native host of Wolbachia wMel, because vertical transmission is possible in this species. On the long term, the transmission of Wolbachia in D. melanogaster may be better than in D. nigrosparsa because D. melanogaster has a higher oviposition rate and a better tolerance of warm temperatures than D. nigrosparsa (Kinzner et al., 2018), both of which could increase the chance for horizontal transfer. This is because horizontal transfer is a stochastic event, and an infected host is therefore more likely to transfer Wolbachia to a new host species if there are more infected hosts available and if the number of Wolbachia cells is higher per host.
Here, we report effects of Wolbachia wMel on D. nigrosparsa as a novel host. We observed low Wolbachia titer, possible cytoplasmic incompatibility, and increased locomotion in both larvae and adults.
Drosophila nigrosparsa is likely to suffer from an increasing temperature independently of whether uninfected  or infected, as Wolbachia had no impact on heat tolerance (this paper).
However, Wolbachia wMel might provide some benefits to this fly such as concerning life history traits not assayed here (e.g., longevity) or concerning resistance to viruses or nutrition supplements, which both will be interesting to analyze in the future. Finally, infection by Wolbachia strains other than wMel may trigger different effects in this alpine vinegar fly.

ACK N OWLED G M ENTS
We thank Yuk-Sang Chan for teaching and partly performing the mi- Reiter for help with fly maintenance and laboratory work; Veronika Hierlmeier and Clemens P. Maylandt and two anonymous reviewers for comments on an earlier version of the manuscript; the University of Innsbruck for providing a doctoral grant to MD.

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