Nighttime activities and peripheral clock oscillations depend on Wolbachia endosymbionts in flies

Wolbachia are ubiquitous bacterial endosymbionts of arthropods and affect host gene expression. Although Wolbachia infections were suggested to modulate sleep in flies, their influence on the circadian clock remained obscure. Here, we screened bacterial symbionts in a laboratory Drosophila melanogaster colony, and observed widespread infections of wMel strain Wolbachia. We established a Wolbachia-free strain from a clock gene reporter strain, period-luciferase (per-luc). Temperature (19–29 °C)-compensated free-running periods were detected regardless of infections which may reflect the lack of wMel infections in central circadian pacemaker neurons. However, locomotor activity levels during the night or subjective night were significantly amplified in uninfected flies. Moreover, the behavioral phenotype of F1 offspring of an uninfected female and infected male resembled that of uninfected flies. This trait is consistent with maternal transmission of Wolbachia infection. Interestingly, per-luc activities in headless bodies, as an index of peripheral circadian oscillators, were severely damped in uninfected flies. Additionally, circadian amplitudes of PER immunoreactivities in Malpighian tubules were reduced in uninfected flies. These results demonstrate that Wolbachia boost fly peripheral clock oscillations and diurnal behavioral patterns. Genetic mechanisms underlying behavioral rhythms have been widely analyzed using mutant flies whereas screening of Wolbachia will be necessary for future studies.

Intracellular alpha proteobacterium Wolbachia are ubiquitous endosymbionts of arthropods [1][2][3] . They are maternally transmitted to offspring and are involved in sex determination 4 , sperm-egg cytoplasmic incompatibility 5 , and pathogen defense 6,7 . Wolbachia infections were also recently suggested to increase nighttime sleep in flies by transcriptionally regulating dopamine synthetic enzymes 8 . However, despite their wide range of influences on host activities, the cellular mechanisms underlying Wolbachia-mediated sleep regulation are not fully understood.
The involvement of symbiosis in host circadian activities has been demonstrated in several species. For example, symbiosis between the squid Euprymna scolopes and the luminous bacterium Vibrio fischeri enables the daytime switch-off of light-emitting organs 9 . Interestingly, synchronous gene expression patterns were found for these symbiotic partners over the course of a day-night cycle, and metabolic modes were switched between day and night in the organ. Additionally, the notable influence of endosymbionts on host circadian rhythms has been reported in the unicellular ciliate Paramecium bursaria 10,11 . Its symbiotic algae, Chlorella, enable the host's mating reactivity rhythm to persist under constant light conditions, as shown by the fact that Chlorella-free cells derived from the same strain fail to display such a rhythm. However, whether such a dynamic influence of symbiosis on host circadian activities could be the general case in many animal species is currently unknown.
Daily temporal patterns from cellular activities to animal behaviors are largely governed by the endogenous circadian clock system, which has been proposed to be driven by gene transcription-translation feedback loops. Circadian oscillations in "clock genes" have been observed in a wide range of systems, from unicellular organisms to mammalian cells, and from peripheral organs to the central nervous system [12][13][14] . It is widely accepted that the circadian clock system in animals consists of central pacemaker neurons and various peripheral oscillator cells. For example, in mammals, circadian rhythms in PER-immunoreactivity (ir) and/or per promoter-driven luciferase (per-luc) levels were observed not only in central pacemakers (i.e., hypothalamic suprachiasmatic nucleus neurons) but also in fibroblasts 15 , livers, lungs, and muscles 13 . Additionally, circadian rhythms in fly PER-ir and/ or per-luc levels were reported not only in central pacemaker neurons [16][17][18] but also in Malphigian tubules (MTs) 19  prothoracic gland cells [20][21][22] , antenna, legs, and wings 23,24 . Although Wolbachia affect various host gene expression 8,25 , its effect on clock gene transcriptional rhythms are currently unknown.
We have found frequent infection by wMel strain Wolbachia in fly strains used for circadian rhythm assays and kept within enclosed laboratory conditions. Thus, we established a Wolbachia-free fly strain from per-luc strain flies using tetracycline to ensure freedom from infection. It is generally accepted that the circadian clock provides three major functions: (i) self-sustaining oscillations, (ii) temperature-compensated oscillations (i.e., stability against temperature perturbations), and (iii) ability to entrain to environmental time cues (i.e., temporal adjustment). Therefore, in the present study, we analyzed free-running locomotor activity rhythms under constant darkness (DD) at different ambient temperature levels and entrainment of rhythms to 8-h shifts of 12:12-h light/dark (LD) cycles in infected and uninfected per-luc flies. We also analyzed effects of Wolbachia symbiosis on PER-ir and per-luc levels in these flies to further understand the involvement of Wolbachia in molecular clock movements.

Results
Detection of endosymbionts and their elimination. We examined symbiotic bacteria in the per-luc line of D. melanogaster by analyzing 16S ribosomal RNA (16S rRNA) gene sequences amplified with PCR using the universal eubacterial primers. The restriction fragment length polymorphism (RFLP) analysis detected only one sequence type, which indicated the presence of one bacterial symbiont in the per-luc line. The sequence of the bacterial gene showed high similarity to 16S rRNA gene sequences of Wolbachia from various insects (Fig. S1A). Detailed analysis using the Wolbachia surface protein (wsp) gene was conducted to clarify the phylogenetic position of the Wolbachia species. RFLP and sequence analysis of the wsp gene showed that the per-luc line contained only one wsp sequence. Molecular phylogenetic analysis of the wsp sequence revealed that the symbiont of the per-luc line belonged to the wMel group of Wolbachia (Fig. S1B). All individuals of the per-luc line were infected with wMel (24/24 individuals).
To examine the effects of Wolbachia infection on host circadian rhythms, we generated Wolbachia-free per-luc flies. Following antibiotic treatment (50 µg/mL tetracycline in fly foods) for three generations, a single female was transferred to an isolation vial to establish isofemale lines. Uninfected offspring were selected and supplied with tetracycline-free regular food. Elimination of Wolbachia was confirmed by PCR ( Fig. S2) or by immunohistochemistry against a heat shock protein (hsp60) [26][27][28] .
Body weight was compared among infected and uninfected adult male (0.67 ± 0.01 mg for infected flies and 0.66 ± 0.01 mg for uninfected flies, N = 10 for both groups) and female flies (1.12 ± 0.02 mg for infected flies and 1.10 ± 0.01 mg for uninfected flies, N = 10 for both groups). No significant difference in body weight was found between infected and uninfected groups (n.s. by Student's t-test irrespective of sex).
The effects of Wolbachia endosymbionts on host behavioral rhythms. We analyzed locomotor activity in infected and uninfected flies at different ambient temperatures (19,24, and 29 °C). Under 12:12-h LD cycles, both infected and uninfected flies showed typical bimodal activity patterns consisting of morning and evening peaks regardless of temperature levels (Fig. 1A,B). Interestingly, total activity counts during the nighttime in uninfected strain flies were significantly higher than in infected flies at any temperature examined (Fig. 1B). To exclude the possible involvement of other microorganisms in the regulation of nighttime activities, we further analyzed locomotor activities in F1 offspring of uninfected females and infected males. Consistent with uninfected strain flies, crossed F1 flies showed a significant increase in total activity counts during the night (Fig. 1A,B).
Under DD, infected-, uninfected-and crossed F1 flies displayed temperature-compensated free-running rhythms; the average τs converged between 23.65 h and 23.94 h ( Table 1; Q 10 = 1.001 for infected group, 1.006 for uninfected group, and 1.008 for F1 group). No differences were found in the average τs in DD conditions at any temperature examined (Table 1). At 19 °C, a lower percentage of rhythmic flies occurred because of low temperature immobilization ( Table 1, Fig. 2), yet total activity counts during the subjective nighttime were significantly higher in uninfected flies and crossed F1 flies (Fig. 2).
To investigate the effects of Wolbachia infection on photic entrainment of locomotor activity rhythms, infected-and uninfected flies were exposed to an 8-h advance or delay of LD cycles. Both groups of flies adjusted their circadian activity pattern immediately to the new LD regime (Figs 3, S3). Thus, Wolbachia infection seems to have little effect on photic entrainability of the circadian clock.

Localization of Wolbachia endosymbiont.
To determine the localization of Wolbachia endosymbionts in flies, we stained oocytes, MTs, and brains with an anti-hsp60 antibody which labels Wolbachia [26][27][28] . In infected flies, anti-hsp60 signals were extensively present in oocytes (Fig. 4A) and various peripheral tissues of the adult body, including MTs, whereas signals were undetectable in the brain (Fig. 4B). Circadian pacemaker neurons (ventral lateral neurons; LN v s) in flies were labeled by an antibody against the pigment-dispersing factor (PDF) 29 , whereas hsp60 signals were lacking in these neurons (Fig. 4B). Relatively strong green fluorescent background levels were detected in all MTs but strong punctate signals were detected only in infected flies (Fig. 4B).
Absence of Wolbachia was further confirmed in physically isolated brains using real-time RT-PCR with the wsp promotor, where wsp transcriptions were undetectable in the brain homogenates regardless of Wolbachia infection status (Fig. S2B).
The effects of Wolbachia endosymbionts on peripheral circadian clocks. Based on the histological localization of Wolbachia endosymbionts, per-luc bioluminescence was compared in headless fly bodies. In infected flies, the bioluminescent intensity displayed circadian oscillations over five cycles (Fig. 5A, mean

Discussion
Wolbachia are intracellular alpha proteobacteria that are maternally transmitted to offspring and are widespread in arthropods 1-3 . Our laboratory Drosophila colony is composed of multiple fly strains from all over the world, including those from Bloomington Drosophila Stock Center (Indiana University, USA), and the Vienna Drosophila Resource Center (Austria). Approximately half of the clock mutant fly stains randomly tested in our laboratory colony were infected with Wolbachia (Fig. S2A). The presence of Wolbachia in laboratory fly colonies has also been reported by other groups [30][31][32] . These frequent infections are not surprising because more than 60% of insect species, including D. melanogaster, are infected by Wolbachia 2 . Thus, we speculate that many D. melanogaster studies, even those performed using in a closed laboratory environment, are under the influence of Wolbachia symbionts.
The most dramatic behavioral phenotype observed in this study was a significant increase in nighttime activities in uninfected flies regardless of ambient temperature levels. At 29 °C under LD cycles, average nighttime activities were beyond the daytime levels, resulting in a loss of diurnal locomotor rhythms in these flies. To determine the specific influence of Wolbachia symbionts, we analyzed locomotor activities in F1 offspring of uninfected females and infected males. Because most of the other microorganisms including the gut-associated microenvironment are restored in the F1 offspring 33 , we compared the phenotypes with those of fully sterilized flies. The observed increased nighttime activities in F1 offspring were similar to those seen in uninfected strain flies, suggesting the importance of Wolbachia symbionts for the manifestation of apparent diurnal rhythms in flies. To further explore these mechanisms, we analyzed the temporal distribution of locomotor activities under DD, revealing increases in locomotor activities during subjective nighttime in uninfected strain flies and F1 offspring. Thus, the influence of Wolbachia on nighttime activities was not the result of "masking effects" which are not mediated by circadian clock oscillations.
Heterogeneity of peripheral circadian clocks has been reported in D. melanogaster 34 : endocrine oscillators in the prothoracic gland are strongly regulated by central pacemakers 22 whereas MTs contain independent oscillators 19 . The present study demonstrated damped per transcriptional oscillations in the peripheral circadian clocks of uninfected flies. Moreover, hsp60 and PER immunostaining assays revealed that Wolbachia symbionts have little effect on central circadian pacemaker neurons, making unique unbalanced clock oscillations within uninfected fly bodies. Because free-running rhythms and photic entrainment of circadian rhythms are indistinguishable between infected-and uninfected flies, a reduction in peripheral clock oscillations may have limited effect on basic clock movements. However, we suggest that daily activity patterns in flies are under the strong influence of peripheral clock movements and Wolbachia symbionts.
Wolbachia infection has previously been documented in the Drosophila brain, but the extent of infection was dependent on the Wolbachia strain, host Drosophila strain, and brain region 35 . Indeed, the Wolbachia strains wRiv and wPop showed higher titers in the brain than wMel, while higher Wolbachia titers were seen in D. simulans than in D. melanogaster 35 . Correlation analyses suggested that Wolbachia infection influenced sleep time in flies 35 . Furthermore, Bi et al. 8 recently reported that wMel infection increased nighttime sleep quantity under LD cycles in D. melanogaster. Thus, the present results of an increase in nocturnal activities in uninfected flies may be caused by the reflective effects of sleep disturbance. Bi et al. 8 found that Wolbachia may affect dopaminergic activities because two essential genes involved in dopamine synthesis, Pale and Ddc, were significantly upregulated in Wolbachia-infected flies. In the present study, we also observed partial enhancement of PER-ir rhythms in s-LN v s in uninfected flies. However, wMel infections were barely detectable in the brain of D. melanogaster, including in PDF-positive pacemaker neurons. This result is consistent with the findings of Albertson et al. 35 . Thus, the actions of wMel on central neurons could be the result of indirect influences on peripheral systems.  Table 1. The effect of Wolbachia symbionts may influence diverse host cellular activities. Of these, one of the most common phenotype induced by Wolbachia is sperm-egg cytoplasmic incompatibility (CI), a form of male-derived zygotic lethality 1,5,36 . Additionally, a series of studies demonstrated that Wolbachia infections elevated reactive oxygen species and caused changes in the redox homeostasis of host cells 37,38 . Furthermore, a microarray assay for filarial nematodes demonstrated that diverse genes regulating translation, transcription, protein folding/sorting, metabolic control, and intracellular signaling are under the control of Wolbachia symbionts 25 . The synthesis of heme in filarial nematodes is dependent on Wolbachia, so heme-dependent energy metabolisms, which can be compensated for by mitochondrial gene expression, are under the strong influence of Wolbachia symbionts 25 . This indicates that metabolic oscillators, as supportive components of clock gene transcription rhythms 39 , could also be under the influence of Wolbachia symbionts.
Oxidation and reduction cycles of peroxiredoxin represent a circadian periodicity in a diverse range of cells from human red blood cells to bacteria, so the redox cycle may be an evolutionarily conserved, nuclear-independent circadian oscillator [40][41][42] . Thus, we speculated that oxidation and reduction cycles may be primary candidates for Wolbachia-dependent circadian oscillators. Based on this hypothesis, we examined the effects of Wolbachia symbionts on the levels of oxidized peroxiredoxin in headless fly bodies. However, oxidation  . For brain staining, pigment-dispersing factor (PDF) was also stained as a marker for central circadian pacemakers (magenta). The approximate brain shape (dotted line) is determined by blue autofluorescence. CB, central brain; OL, optic lobe. Note that large LN v s were visualized in these frames. rhythms were observed both in infected and uninfected flies with shifts in their circadian phases (Fig. S4). Therefore, further studies are required to uncover the intracellular mechanisms regulating Wolbachia-dependent per oscillations in peripheral circadian clocks. Here, we demonstrated that a ubiquitous Wolbachia endosymbiont could influence peripheral clock oscillations and circadian activity patterns in Drosophila models. Notably, in 2017, the US Environmental Protection Agency approved the utilization of Wolbachia to kill wild mosquitoes that transmit viruses such as dengue, yellow fever, and Zika 43 . Based on our findings, we suggest that future work with Wolbachia, including other Wolbachia strains, and other native endosymbionts should consider whether they modulate the circadian behavioral patterns of insect species. Such a question is beyond the scope of the current study but is an important concern.
Methods Fly stocks. The per-luc (second chromosome) line of D. melanogaster, which was originally obtained from Dr. Jeffery C. Hall (Brandeis University, Waltham, MA) via Dr. Akira Matsumoto's laboratory (Juntendo University, Tokyo, Japan), was predominantly used in this study. Three additional lines of D. melanogaster (per 01 , Clk Jrk , and cry 01 ) were used to examine infection frequencies of Wolbachia. The per 01 line was kindly provided by Dr. Akira Matsumoto's laboratory. The Clk Jrk and cry 01 lines were kindly provided by Dr. Taishi Yoshii's laboratory (Okayama University, Okayama, Japan). Flies were raised on cornmeal, molasses, and yeast food at 25 ± 1 °C under 12:12-h LD (light on at 07:00) cycle conditions. DNA extraction. DNA was extracted using a standard phenol: chloroform protocol. For symbiont screening, Detection of endosymbionts. 16S rRNA genes of whole eubacteria from the whole-insect DNA extract were amplified by PCR with Ex Taq DNA polymerase (Takara Bio, Tokyo, Japan) using the forward primer 16SA1 (5′-AGAGTTTGATCMTGGCTCAG) in combination with the reverse primer 16SB1 (5′-TACGGYTACCTTGTTACGACTT) as described previously 44 . The protocol utilized a temperature profile of 94 °C for 2 min followed by 35 cycles of 94 °C for 30 sec, 55 °C for 30 sec and 72 °C for 1 min. The wsp gene was also amplified using the forward primer wspF (5′-GGGTCCAATAAGTGATGAAGAAAC) in combination with the reverse primer wspR (5′-TTAAAACGCTACTCCAGCTTCTGC) 45 under the same temperature profiles. The PCR products were cloned, and 24 clones of each gene were subjected to RFLP genotyping using RsaI and HaeIII, before being sequenced as previously described 44 . The sequence similarity of a detected symbiont was analyzed by BLAST 46 on nucleotide sequences deposited with DDBJ/NCBI/GenBank databases. Candidate taxa for phylogenetic analysis were selected from a variety of bacterial 16S rRNA gene or wsp gene sequences including those known for bacteria isolated from insects and those showing high similarity scores in a BLAST search. Phylogenetic analysis of the symbiont was conducted using the Neighbor-joining method in the program MEGA 5 47 . Bootstrap tests were conducted with 1,000 resamplings. Specific PCR was conducted to examine the infection frequency of Wolbachia in the per-luc line using the primers wspF and wspR.
Nucleotide sequence accession numbers. The nucleotide sequences determined in this study were deposited in the DDBJ/EMBL/GenBank nucleotide sequence database under accession numbers LC108848 (Wolbachia 16S rRNA gene) and LC108849 (wsp gene).
Wolbachia elimination. We generated a Wolbachia uninfected line using antibiotics. Freshly-laid eggs were surface-sterilized in 2.7% sodium hypochlorite (Sigma-Aldrich, St. Louis, MO) for 2 min and twice in 70% ethanol for 2 min. Following two 1 min rinses in distilled water, they were raised on a normal cornmeal, molasses, and yeast food containing 50 µg/mL tetracycline hydrochloride (Wako Pure Chemical Industries, Osaka, Japan). After tetracycline treatment for three generations, a single female was transferred to a vial to establish isofemale lines. After laying eggs, each female was assayed for the presence or absence of Wolbachia infections by PCR (see Supplementary Methods). After selection of the vial derived from an uninfected female, flies were maintained on tetracycline-free normal food. Because antibiotic treatment can eliminate other fly microorganisms besides Wolbachia such as beneficial gut-associated microbes, these were restored by backcrossing Wolbachia uninfected tetracycline-treated females with control infected males as described previously 33 . Behavioral analysis. Locomotor activity monitoring was carried out with 3-7-day-old male flies in Drosophila activity monitors (DAM2, Trikinetics, Waltham, MA) set in a temperature-controlling incubator (MR-553, Sanyo Electric, Osaka, Japan). Ambient temperature levels (19 °C, 24 °C, or 29 ± 1 °C) and lighting (250 lux, day light fluorescent bulb) were controlled by the incubator. Double-plotting actograms and chi-square periodograms of circadian locomotor rhythms were analyzed using originally developed software written by M.I. (PACR ver. 2.0). To analyze free-running locomotor rhythms, flies were first entrained to 12:12-h LD cycles for 4 days, and then the lighting schedule was switched to DD. For the LD shift experiment, the light phase was lengthened (i.e., phase-delayed) or shortened (i.e., phase-advanced) by 8 h on the 7th day of locomotor activity monitoring under standard LD cycles.
Statistical analysis. Means were calculated with standard errors. Student's t-test was used for pairwise comparisons. One-way analysis of variance (ANOVA) followed by Duncan's multiple range test was used for statistical comparisons among multiple groups. Nuclear PER immunostaining levels from several samples were analyzed using a regression sine curve fitted with 95% confidence levels using SigmaPlot ver 7.0 software (IBM SPSS Statistics, Armonk, NY).

Data Availability
DNA and RNA sequences: The nucleotide sequences determined in this study were deposited in the DDBJ/ EMBL/GenBank nucleotide sequence database under accession numbers LC108848 to LC108849.