1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The maternally inherited bacterium Wolbachia infects the germline of most arthropod species. Using Drosophila simulans and D. melanogaster, we demonstrate that localization of Wolbachia to the fat bodies and adult brain is likely also a conserved feature of Wolbachia infection. Examination of three Wolbachia strains (WMel, WRiv, WPop) revealed that the bacteria preferentially concentrate in the central brain with low titres in the optic lobes. Distribution within regions of the central brain is largely determined by the Wolbachia strain, while the titre is influenced by both, the host species and the bacteria strain. In neurons of the central brain and ventral nerve cord, Wolbachia preferentially localizes to the neuronal cell bodies but not to axons. All examined Wolbachia strains are present intracellularly or in extracellular clusters, with the pathogenic WPop strain exhibiting the largest and most abundant clusters. We also discovered that 16 of 40 lines from the Drosophila Genetic Reference Panel are Wolbachia infected. Direct comparison of Wolbachia infected and cured lines from this panel reveals that differences in physiological traits (chill coma recovery, starvation, longevity) are partially due to host line influences. In addition, a tetracycline-induced increase in Drosophila longevity was detected many generations after treatment.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

An estimated 66% of all arthropods are infected with Wolbachia (Hilgenboecker et al., 2008), a Gram-negative, intracellular bacterium that is transmitted through the maternal germline (Serbus et al., 2008; Werren et al., 2008). Wolbachia infection often affects the host's reproduction to promote its own transmission (Werren et al., 2008). Therefore, much research has focused on Wolbachia–host interactions in the germline. However, in some species, including Drosophila melanogaster, Wolbachia's self-promoting effects are weak and it is unclear how Wolbachia infection is maintained (Hoffmann et al., 1998; Yamada et al., 2007). Reports demonstrating Wolbachia localization to somatic tissues raise the possibility that Wolbachia may influence somatic processes to the benefit of the host. For example, Wolbachia has been shown to decrease Drosophila's susceptibility to viral infection (Hedges et al., 2008; Teixeira L and Ashburner, 2008; Osborne et al., 2012). Other studies point towards a less general Wolbachia effect and highlight a host-species dependence of the Wolbachia influence. For example, in D. melanogaster, Wolbachia influences host size (Hoffmann et al., 1998) and longevity (Driver et al., 2004; Fry et al., 2004; Toivonen et al., 2007), but the extent of the Wolbachia effect and even the direction of the effects vary widely among host strains and species. In some D. melanogaster lines, altered behaviour has also been associated with Wolbachia infection, such as olfactory-cued locomotion (Peng et al., 2008), mating rate (de Crespigny et al., 2006; Gazla and Carracedo, 2009) and fertility (Gazla and Carracedo, 2009), but these results also vary with host and Wolbachia strains. In laboratory Drosophila lines that have been evolved separately towards tolerance for various toxins over a 30-year time period, Wolbachia has been shown to contribute significantly to mating discrimination between these populations (Koukou et al., 2006). However, no Wolbachia effect has been detected within populations, indicating that Wolbachia can enhance a mating bias that has evolved independently in these populations (Koukou et al., 2006).

Previous studies demonstrated the presence of Wolbachia in the brains of Drosophila (Min and Benzer, 1997; Albertson et al., 2009) and Collembola (springtails) (Czarnetzki and Tebbe, 2004), and deduced a Wolbachia infection by qPCR in Eurema hecabe (Butterfly) (Narita et al., 2007) and Drosophila (Dobson et al., 1999; McGraw et al., 2002). However, it remains unclear whether this is a sporadic event or a conserved feature of Wolbachia infection. Here we directly address this issue by taking advantage of Drosophila simulans and D. melanogaster lines established from wild populations, including lines established in 2003 from a farmer's market collection in North Carolina (Edwards et al., 2009). Known as the ‘Drosophila Genetic Reference Panel’, these lines have been inbred for at least 20 generations and are widely used for behaviour studies and expression profiles (Ayroles et al., 2009; Edwards et al., 2009; Morozova et al., 2009; Mackay, 2010). The lines were found to have a range of differences in starvation resistance, lifespan, chill coma recovery time, copulation latency and other traits (Ayroles et al., 2009). Of the 192 RAL (Raleigh) D. melanogaster lines deposited at the Bloomington Stock centre, we used the core group of 40 lines to analyse them with regard to Wolbachia infection and related effects on physiological parameters. Significantly, 16 of the lines are infected with Wolbachia and were analysed for a bacterial presence in the brain as well as behavioural and physiological responses.

Wolbachia has been shown to localize to the Drosophila brain during larval and adult stages (Albertson et al., 2009). The Drosophila brain arises from divisions of neuronal stem cells. During embryogenesis, neuroblasts continuously divide asymmetrically to produce a self-renewing neuroblast and a primary neuron of the embryonic and larval central nervous system (CNS) (Doe, 2008; Egger et al., 2008). Neuroblast divisions continue into larval stages, producing secondary neurons that give rise to the adult central nervous system (Spindler and Hartenstein, 2010). In dividing neuroblasts of infected embryos and larvae, Wolbachia has been found to preferentially localize to the self-renewing neuroblast rather than to the primary neuron (Albertson et al., 2009). This asymmetric distribution may influence the final Wolbachia distribution in the adult Drosophila brain.

The adult Drosophila brain is composed of approximately 100 000 neurons. The soma of neurons coalesce in certain regions and project their neurites into densely interwoven neuropils, which form distinct lobes (Spindler and Hartenstein, 2010; Yu et al., 2010). Functional studies of neuropil have defined the major brain centres (Vosshall and Stocker, 2007; Olsen and Wilson, 2008; Tanaka et al., 2012). These include (i) the protocerebrum (several distinct interlinked neuropils), (ii) mushroom body (learning and memory), (iii) antennal lobes (olfactory chemosensory pathways), (iv) the subesophageal ganglion (gustatory neurons and taste behaviour), (v) antennal nerves (convergence of olfactory receptor neurons), (vi) the ventrolateral protocerebrum (visual projection neurons connecting the central brain and optic lobe) and (vii) the optic lobes (comprising the compound eye) (Hanesch et al., 1989; Pereanu et al., 2010). As described below, we have generated detailed Wolbachia distribution maps of the bacterial infection in the brain of four host/Wolbachia combinations: WRiv in D. simulans, WMel in D. simulans, WMel in D. melanogaster and WPop in D. melanogaster. These studies demonstrate that bacteria localization to the adult brain is a conserved feature of Wolbachia infection, yet the specific distribution within the brain differs among different Drosophila species and Wolbachia strains.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Localization to adult brains is a conserved feature of Wolbachia infections

This study examines whether brain infection is always present in Wolbachia-infected D. melanogaster and D. simulans lines. To survey D. melanogaster, we assayed the ‘core’ group of 40 lines from the ‘Drosophila Genetic Reference Panel’ (Ayroles et al., 2009) for the presence of Wolbachia. Infection was determined by PCR using entire flies and by cytology of the ovarioles (Fig. S2). In all 40 lines examined, the PCR and cytological analysis were in accord: 24 lines were uninfected and 16 lines were stably Wolbachia infected (Table 1). Sequencing the wsp gene from four of the infected lines (304, 360, 712, 820) produced sequences identical to the published wsp sequence from wMel (Tigr, wsp is one of the fastest evolving Wolbachia genes (Baldo et al., 2010) and identical wsp sequences from flies in a limited geographical may indicate that the lines are infected with the same bacterial strain, although examples of divergent Drosophila lines with identical wsp sequences exist (Riegler et al., 2005). Adult brains from each of the infected D. melanogaster lines were dissected from at least two male and female flies and stained with Syto-11, a DNA dye that preferentially stains Wolbachia (Casper-Lindley et al., 2011). Wolbachia was clearly present in all D. melanogaster brains examined (Fig. S1). Wolbachia infections were also determined by PCR for a field population of D. melanogaster, captured in Albion, MI. Six infected lines were stained with Syto-11 and showed Wolbachia in the adult brain (data not shown).

Table 1. Infection status of flies from the ‘Drosophila Reference Panel’
Bloomington strain numberRAL numberInfection status
  1. Flies were analysed by PCR and oocyte cytology to determine the Wolbachia infection status of the inbred lines.


To survey D. simulans, isofemale lines were established from D. simulans flies captured near Davis, California, and tested for Wolbachia infection by PCR (Michael Turelli, UC Davis). Fourteen infected lines were analysed for bacteria distribution in somatic and germline tissues. The adult ovaries and brains of four adult females per line were dissected and stained; both tissues showed robust Wolbachia titre with a 100% infection frequency (n = 56, Fig. S1). Wolbachia infections were also determined by PCR for a field population of D. simulans captured in the Landels-Hill Big Creek Reserve, CA. Several infected lines were stained with Syto-11 and showed Wolbachia brain localization similar to the field populations described above (data not shown). Taken together, these data indicate that localization of Wolbachia to the Drosophila adult brain is likely a conserved feature of Wolbachia infection.

Wolbachia exhibit distinct intra and extracellular distributions in the adult Drosophila brain

Wolbachia localization patterns were further examined at the cellular level within the adult central brain of the D. melanogaster laboratory strain that was infected either with the native WMel Wolbachia or with the pathogenic WPop variant (Min and Benzer, 1997), and within brains of the D. simulans laboratory strain infected with WRiv or WMel. To quantify Wolbachia, dissected brain tissue was fixed and stained with propidium iodide (PI), anti-CG9850 antibody, and fluorophore-conjugated phalloidin (Fig. 1). Propidium Iodide stains both host and bacterial DNA, anti-CG9850 fortuitously cross-reacts with Wolbachia (Cho, 2004), and phalloidin labels the actin-rich host cell cortex. In uninfected flies, low-intensity anti-CG9850 staining is observed, with occasional small background puncta (Fig. 1). In contrast, infected flies stained with anti-CG9850 show intense puncta that tightly colocalize with propidium iodide puncta. All Wolbachia strains that are analysed in this study show this colocalization. In quantitative analyses, Wolbachia were scored as both CG9850- and PI-positive staining puncta unless otherwise noted. The stainings also reveal that Wolbachia does not reside in axons (Fig. 2A–C), but in or next to the host cell bodies of neurons in the central brain. All examined Wolbachia strains either occurred as few, individually discernible dots within the host cell bodies (Fig. 2D–F), or as larger clusters (Fig. 2G–I). Wolbachia clusters were especially large and abundant in the WPop strain and many clusters appeared to be extracellular. Figure 2J shows Z-slices through an area with a large WPop cluster in D. melanogaster. The actin stain (green in the merged image and the top row of individual channel images) shows that the cluster is not surrounded by host cortical actin, indicating that it is extracellular. Furthermore, the identical stain of PI (DNA) and the anti-CG9850 antibody (purple in the merged image and the second and third row of individual channel images) indicates that there is no host DNA in this Wolbachia cluster.


Figure 1. Propidium Iodide and anti-CG9850 stain Wolbachia in the adult brain. The DNA stain propidium iodide (PI) highlights host nuclei. In infected fly lines, DNA staining of Wolbachia stands out as small, brighter puncta in the cytoplasm. The anti-CG9850 antibody also stains Wolbachia in the infected fly lines and the staining is tightly overlapping (right panels). Phalloidin marks the actin-rich cell cortex. The top row shows tetracycline-cured D. melanogaster and the other rows show infected fly lines as indicated. Scale bars, 5 μM.

Download figure to PowerPoint


Figure 2. In the adult central brain Wolbachia do not reside in axons, but are either in the cell body or as aggregates between host cells.

A–C. Overviews of the central brain infected with WRiv, WMel or WPop respectively. Wolbachia does not reside in the axon-rich neuropil (asterisks).

D–F. Wolbachia bacteria localize to the soma of neurons either as small aggregates of single bacteria (arrows) or as larger clusters (arrowheads).

G–I. Large bacteria clusters can occur with or without host cell cortical actin.

J. Images of a Z-series through a Wolbachia (WPop) cluster in WMel show that it is not encased in an actin-rich host membrane and lacks a host nucleus. Cortical actin (top row of the individual channels and green in the merged image) does not envelop this bacterial cluster and no host nuclei are visible within the cluster or in the nearby area.

Scale bars, 15 μM (A–C) or 5 μM (D–J). A–J (except E′) staining: cell cortex is green (phalloidin-488), host nuclei are red (PI) and Wolbachia are purple (PI and anti-CG9850). E′ staining: host nuclei are marked blue (anti-Histone H1).

Download figure to PowerPoint

Wolbachia distribution in the brain depends on both host line and Wolbachia strain

The bacteria distribution within the central brain varied among the two Drosophila species and the different Wolbachia strains. Intracellular as well as cluster infection were quantified in eight brain regions. From a dorsal view of the brain (Fig. 3A and A′), the quantified regions include: (1) the soma between the superior medial protocerebra, (2) the soma posterior to mushroom body and anterior to the antennal lobe, (3) soma posterior to antennal lobes and anterior to suboesophageal ganglion, (4) soma anterior to the superior protocerebrum lobe, (5) soma lateral to antennal lobes and medial to the ventrolateral protocerebrum lobe, (6) soma posterior antennal nerve neuropil and the ventrolateral protocerebrum lobe, (7) soma lateral to the lateral superior protocerebrum, and (8) soma surrounding the optic lobe. Brain cells were quantified as having 0 bacteria, 1 bacterium, 2–5 bacteria, or > 5 bacteria (Table S1). In D. simulans, WRiv had a very similar titre in regions 1 through 4, and in regions 5 through 7 (Table S1). Therefore, we averaged these regions in two groups: regions 1 to 4 and regions 5 to 7 (Fig. 3K). Representative images of these regions and of region 8 are shown in Fig. 3B–D. In contrast, the Wolbachia distribution in D. melanogaster was similar in all regions from 1 to 7, for either WMel or WPop (Table S1) (Fig. 3G and I). The average infection quantities of regions 1 to 4 and regions 5 to 7 for all strains are shown in Fig. 3K. Because titre level and distribution differed between WRiv in D. simulans and WMel in D. melanogaster, we investigated whether the microbe or host control these differences by examining WMel in D. simulans hosts. WMel Wolbachia in D. simulans hosts were scored in three groups (area 1 through 4, area 5 through 7, and area 8), with the first two groups having the same titre distribution, similar to the WMel infection in D. melanogaster (Table S1 and Fig. 3K). However, in regions 1 through 7, WMel in D. simulans showed an intermediate titre between, WRiv in D.simulans and WMel in D. melanogaster, as described below.


Figure 3. Wolbachia distribution in the adult brain.

A and A′. Overview over the adult brain: panel (A) shows PI staining highlighting host cell bodies, and indicates the regions of Wolbachia quantification. (A′) is a merged image of the DNA staining (PI-red) and actin staining (green), highlighting axons and the major brain lobe centres. The regions are indicated as: MB, mushroom bodies; AL, antennal lobes, SEG, subesophageal ganglion, AN, antennal nerve; VLPR, ventrolateral protocerebrum; OL, optic lobes.

B–D. Representative images of WRiv in D. simulans as they are found in regions 1–4, 5–7 and 8 respectively.

E and F. Representative images of WMel in D. simulans in areas of regions 1–7 and 8 respectively.

G and H. Representative images of WMel in D. melanogaster in areas of regions 1–7 and 8 respectively.

I and J. Representative images of WPop in D. melanogaster in areas of regions 1–7 and 8 respectively.

Panels (B)–(J) show only the anti-CG9850 channel, (B′)–(J′) show merged images of the PI (red), the anti-CG9850 (blue) and the phalloidin (green) channels. Scale bars, 100 μM (A, A′) and 10 μM (B–J).

K. Quantification of Wolbachia-containing cells in the respective strains for each region average. Pie charts in shades of grey indicate the per cent of cells without Wolbachia (white), with one Wolbachia (light grey), with two to five Wolbachia (medium grey), or more than five Wolbachia (dark grey). Data for each individual region are shown in supplementary Table S1.

Download figure to PowerPoint

Overall, WRiv in D. simulans had the highest number of infected cells: more than half of the cells were infected in regions 1 through 4, and over 70% were infected in regions 5 through 7 (Fig. 3K). In contrast, only about 40% of all brain cells were infected in regions 1 through 7 in D. simulans with WMel, and less than 40% of cells where infected in D. melanogaster with either WMel or WPop (regions 1 through 7). Among the cells that contained bacteria, D. simulans with WRiv exhibited the most cells with more than 5 Wolbachia per cell. These cells comprised nearly 20% in region 1 to 4, and about 35% in regions 5 to 7. In contrast, WMel in either D. simulans or D. melanogaster had fewer than 12% (region 1 to 4) or 14% (region 5 to 7) of cells with more than five bacteria. Similarly, only 12% of WPop-infected cells showed more than 5 bacteria. Region 8, the optical lobe, showed very little infection in all lines with 98% and 92% host cells without bacteria in D. simulans (WRiv and WMel) and 99%, and 93% cells uninfected in D. melanogaster (WMel and WPop) (Fig. 3K).

Wolbachia cluster size and frequency were also quantified (Fig. 4). Extracellular Wolbachia clusters were defined as aggregates of bacteria with no host nuclei and an area greater than 11 μm2 (slightly larger than an average neuronal cell body). The average cluster size is largest in WPop in D. melanogaster (54 μM) and significantly smaller in WMel in D. melanogaster and in WRiv in D. simulans (21 and 22 μM) (Fig. 4A). WMel in D. simulans forms intermediate size clusters (32 μM). Cluster size frequency was quantified by counting clusters of three size categories (11–40 μm2, 40–90 μm2, and greater than 90 μm2). Figure 4B–D shows the frequency of bacterial clusters for each size category per 100 host cell bodies. Clusters of WMel in D. melanogaster and of WRiv in D. simulans occur mostly in the smallest category and were not observed at all in the optical lobe (area 8, Fig. 4B). WPop in D. melanogaster showed dramatically higher frequency of the small cluster size, especially in areas 2 through 6, and also showed clusters in region 8, the optical lobe. The larger cluster sizes (Fig. 4C and D) were very rare in WMel in D. melanogaster and WRiv in D. simulans, yet were commonly formed by the pathogenic WPop strain, which is known to cause premature death of its D. melanogaster host. Also, the largest observed cluster size developed by WPop is dramatically larger (858 μm) than those of WMel in D. melanogaster (43 μm), in D. simulans (87 μm), or of WRiv in D. simulans (71 μm).


Figure 4. Wolbachia cluster size and distribution in the adult brain.

A. Average cluster size in the indicated strains.

B–D. Cluster frequency per 100 cells according to cluster size (11–40 μm2 cluster size: B; 40–90 μm2 cluster size: C; over 90 μm2 cluster size: D). Cluster frequency was counted for WRiv in D. sim (blue), WMel in D. mel (red) and WPop in D. mel (green). WPop forms more numerous and larger clusters.

Download figure to PowerPoint

In the ventral nerve cord, Wolbachia preferentially localize to the neuronal cell bodies but not to axons

The ventral nerve cord (VNC) is an integral component of the adult central nervous system and functions as a major neural circuit centre for motor activities such as walking (Burrows et al., 1988; Laurent and Burrows, 1988; Yellman et al., 1997) and flying (Burrows, 1975; Peters et al., 1985; Reye and Pearson, 1987). An overview of the D. simulans VNC is shown in Fig. 5A. Intracellular Wolbachia are visualized in a larger magnification (Fig. 5B). Wolbachia localize to the VNC in both Drosophila species and all three examined Wolbachia lines (Fig. 5C–F). Similar to the central brain, we observed a higher cellular infection frequency and titre in D. simulans compared with D. melanogaster with either Wolbachia line (Fig. 5C–D and data not shown). Wolbachia clusters are found in the VNC of WPop-infected hosts (Fig. 5F).


Figure 5. Wolbachia localization in the ventral nerve chord (VNC).

A. Overview of the central brain and VNC (bracket) from D. simulans infected with WRiv.

B–B″. Higher magnification of (A), showing that Wolbachia is contained within the cortical actin cortex of host cell bodies in the VNC (arrows). (B) Merged image of PI (red) and actin (green), (B′) PI staining DNA, (B″) Phalloidin-488 staining cortical actin.

C–F. Bacteria in host cells: WRiv in D. simulans, WMel in D. melanogaster and WPop in D. melanogaster respectively.

G–I. WRiv in D. simulans and WMel in D. melanogaster do not occur in regions of axonal bundles (G, H asterisks), but WPop clusters are detected in regions of D. melanogaster axonal bundles (I, arrows). Scale bars, 100 μM (A) or 10 μM (B, C–I).

Download figure to PowerPoint

The central brain and the VNC are connected through a dense network of axonal tracks (Fig. 5A, upper part of the bracket). Similar to the axon-dense central brain neuropil, Wolbachia is generally not apparent within the axon tracks (Fig. 5G–I, asterisks) except for occasional WPop clusters in D. melanogaster (Fig. 5I, arrows).

Localization to the fat bodies is likely a conserved feature of Wolbachia infections

The Drosophila fat body senses the nutritional status of the flies and consequently regulates global growth (Colombani et al., 2003). In addition, the fat body plays a role in mating behaviour (Lazareva et al., 2007). Wolbachia has been observed in larval fat bodies in of D. melanogaster (Clark et al., 2005). To analyse if this Wolbachia localization in fat body tissue is conserved, we tested all infected lines form the ‘Drosophila Genetic Reference Panel’. For fat body analysis, third instar larvae were dissected and stained for Wolbachia visualization with Syto-11. All infected strains showed bacteria in the fat bodies and examples are shown in Fig. 6.


Figure 6. Wolbachia in larval D. melanogaster fat bodies. Syto-11 staining shows that Wolbachia reside in fat bodies of the infected laboratory line and lines 639 and 786 (arrows). No puncta are seen in the uninfected laboratory line.

Download figure to PowerPoint

Wolbachia influence on physiology and behaviour

It is intriguing that Wolbachia localization to brain and fat bodies is a conserved feature of Wolbachia infections in Drosophila. The behavioural manipulation hypothesis proposes that a microbial endosymbiont can alter host behaviour specifically to increase its own transmission (Thomas et al., 2005). To analyse if Wolbachia influences behaviour and physiology, previously published assays on strains of the Drosophila Genetic Reference Panel (Ayroles et al., 2009; Edwards et al., 2009; Harbison et al., 2009; Morozova et al., 2009) were reanalysed with respect to the lines' infection status. anova analysis showed that infection status of a line is correlated with differences in chill coma recovery, sleep time during the day, and some of the ethanol sensitivity, locomotor startle response and olfaction responses (Table 2 column W, asterisks). There was no significant correlation between Wolbachia infection and quantification of the other examined traits, which include aggressive behaviour (Edwards et al., 2009), competitive fitness, copulation latency (Ayroles et al., 2009), starvation resistance and longevity (Morozova et al., 2009), sleep time (night) and sleep bout number (day and night) (Harbison et al., 2009).

Table 2. anova analysis of the relationship between infection status and published physiological and behavioural traits
  1. Column W = Wolbachia-related effect, column S = strain identity-related effect, W*S = interaction between infection status and strain identity. Asterisks indicate a significant Wolbachia-related effect (P < 0.05).

Aggressive behaviour (males only)0.9412  
Chill coma recovery time0.0014*0.55630.8274
Competitive Fitness0.3076  
Copulation Latency0.7874  
Ethanol sensitivity 1 (alcohol medium)0.36730.37250.5527
Ethanol sensitivity 1 (standard medium)0.0112*0.54010.6135
Ethanol sensitivity 2 (alcohol medium)0.15050.12180.7537
Ethanol sensitivity 2 (standard medium)0.110.5120.4851
Locomotion (activity per waking minute)0.29150.0006*0.8963
Locomotion (distance moved in 12 h)0.0127*0.0127*0.9064
Locomotion (time spent moving in 12 h)0.0181*0.0171*0.827
Locomotor startle response (2006 data, standard medium)0.06060.97060.9737
Locomotor startle response (2007/8 data, dopamine medium)0.0485*0.49680.5821
Locomotor startle response (2007/8 data, ethanol medium)0.0384*0.75430.445
Locomotor startle response (2007/8 data, serotonin medium)0.0222*0.74330.665
Locomotor startle response (2007/8 data, standard medium)0.11720.8370.8344
Olfaction – 0.1% benzaldehyde (ethanol medium)0.42850.50630.9754
Olfaction – 0.1% benzaldehyde (standard medium)0.0429*0.61720.6003
Olfaction – 0.1% benzaldehyde (tomato medium)0.55130.65430.846
Olfaction – 0.3% benzaldehyde (ethanol medium)0.0235*0.46240.3814
Olfaction – 0.3% benzaldehyde (standard medium)0.39340.32490.2553
Olfaction – 0.3% benzaldehyde (tomato medium)0.05280.63890.6352
Olfaction – Acetephenone0.60160.47970.7252
Olfaction – hexanol0.72390.49460.8374
Sensory bristle number (abdominal bristles)0.56550.08410.937
Sensory bristle number (sternopleural bristles)0.17740.11980.9497
Sleep bout number (day)0.16210.19240.2559
Sleep bout number (night)0.10380.83420.978
Sleep time (day)0.013*0.0001*0.5113
Sleep time (night)0.98110.0231*0.9529
Starvation stress resistance0.8951< 0.0001*0.7204

To test if Wolbachia-induced effects existed within individual lines, we repeated the assays after curing the lines of Wolbachia. Nine of the 16 infected lines were used to evaluate the physiological effects of Wolbachia and four of the 24 uninfected lines were also treated and used to control for general tetracycline-induced effects. The assays described below directly compare cured and uncured isolines from the Drosophila Genetic Reference Panel.

Chill coma

Flies were kept on ice for 30 min and the time of their first movement after shifting to 25°C was recorded. Of the nine infected lines tested, females in one line and males in three lines had a significant response to tetracycline, and the response was either a longer or a shorter recovery time, depending on the line (Fig. 7A). A slower chill coma recovery was observed in uninfected, treated lines 765 (males and females) and 379 (females). Using a nested anova analysis, line identity was the only significant factor in explaining the response variation (Table 3). This result indicates that neither Wolbachia infection nor tetracycline treatment have a systematic effect. The average chill coma recovery times for the individual lines were similar to the ones published by Ayroles (Ayroles et al., 2009). However, the anova analysis of those untreated lines had indicated a trend for slower recovery times in infected fly lines (Table 2). We have used fewer uninfected fly strains in the second analysis, which may explain why a slower recovery time was not observed. It is noteworthy that none of the four uninfected lines had a faster recovery time after tetracycline treatment.


Figure 7. Physiological traits of infected and uninfected D. melanogaster lines.

A. Recovery time of different fly lines after chill-induced coma.

B. Survival time without food supply.

C. Fly longevity on regular food supply.

Results of female flies are in the left graphs and results from male flies are in the right graphs. Infected lines are represented by light grey bars, uninfected lines by white bars and tetracycline-treated flies by dark grey bars. Significant differences between lines before and after Wolbachia-curing with tetracycline are marked by asterisks (T-test, P > 0.05). (n for each experiment are listed in supplementary Table S1).

Download figure to PowerPoint

Table 3. Nested anova analysis of the effect of sex, infection status, and fly line identity and tetracycline treatment on chill coma recovery, starvation survival time and longevity
Chill coma recovery
Source of variationFSignificance
Infection status0.0610.810
Tetracycline treatment1.8730.198
Sex*tet treatment0.1790.676
Infection*tet treatment1.6230.229
Tet treatment*line1.0750.419
Starvation survival time
Source of variationFSignificance
Infection status0.3370.575
Tetracycline treatment3.0740.113
Sex*tet treatment0.0530.820
Infection*tet treatment0.0380.849
Tet treatment*line1.0010.471
Source of variationFSignificance
  1. Asterisks indicate a significant Wolbachia-related effect (P < 0.05).

Infection status1.7370.203
Tetracycline treatment5.3490.027*
Sex*tet treatment
Infection*tet treatment1.9920.186
Tet treatment*line0.2930.982


The response to starvation was significantly influenced by tetracycline treatment in six of the eight infected lines (Fig. 7B). However, the direction of the response differed according the line and/or sex. Of the three uninfected lines, two (and males of the third) were significantly affected by tetracycline, also in differing directions. Similar to the outcome of the chill coma experiment, variation in starvation survival was dependent on the line identity and there was no general significant influence of either tetracycline or Wolbachia infection. However, in agreement with earlier studies (Table 2) (Ayroles et al., 2009; Goenaga et al., 2010), the duration of starvation survival was significantly dependent on sex, with females surviving for a longer duration than males.


Of the nine infected lines tested, male longevity was extended after tetracycline treatment in all lines except for 639 (Fig. 7C). In females, four of nine lines had increased longevity after tetracycline treatment. Of the four uninfected lines, two also had significantly increased longevity after tetracycline treatment in both sexes. Data analysis using nested anova indicates that tetracycline has a significant effect on longevity (Table 3), independent of Wolbachia infection. Line identity was also a significant factor in determining longevity.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Localization to the adult brain and fat bodies is likely a conserved feature of Wolbachia infection

The success of Wolbachia dispersal has been attributed to its efficient localization to the male and female germ lines and successful transmission through the latter (Serbus et al., 2008). However, there are a number of reports documenting the presence of Wolbachia in somatic tissues of larval and adult insects (Min and Benzer, 1997; Dobson et al., 1999; McGraw et al., 2002; Czarnetzki and Tebbe, 2004; Frydman et al., 2006; Narita et al., 2007; Albertson et al., 2009). To determine whether this was a sporadic occurrence or a conserved feature of Wolbachia-host interactions, we examined two field populations of Wolbachia-infected D. melanogaster and two field populations of Wolbachia-infected D. simulans strains for the presence of Wolbachia in the adult brain. In all examined strains, there was a robust presence of Wolbachia in the adult brain. Similar studies examining the fat bodies in 14 Wolbachia-infected D. melanogaster strains also revealed the presence of Wolbachia in all cases. Thus, while Wolbachia has been traditionally viewed as a germline endosymbiont, these studies, together with the previous work described above, demonstrate that Wolbachia has a rich somatic life as well.

The presence of Wolbachia in these somatic tissues raises intriguing issues regarding their route and mechanism of localization. In the developing Drosophila brain, Wolbachia exhibits a microtubule dependent, asymmetric segregation pattern during neuroblast divisions, indicating that they rely on intracellular mitotic cues for their ultimate somatic localization (Albertson et al., 2009). Alternatively, other work demonstrates that Wolbachia injected into the adult abdomen is capable of an extraordinary migration to the specific somatic niche cells of the female germline (Frydman et al., 2006; Fast et al., 2011). Thus, Wolbachia possesses the ability for both, intracellular mitotic-based cell-to-cell transmission, and extracellular migration. This conclusion is in accord with observations of Wolbachia in the nematode Brugia malayi, where it has been demonstrated that Wolbachia relies on both cell-to-cell transmission and internal mitotic mechanisms for their germline localization (Landmann et al., 2012).

Because we did not analyse all tissues in the adults, it is possible that Wolbachia are equally abundant throughout many tissues in the adult. The presence of Wolbachia in many host tissues has been suggested by PCR-based studies (Dobson et al., 1999; McGraw et al., 2002). Whether Wolbachia is specifically targeted to the brain and fat bodies, and possibly to other areas, remains unclear. However, the discovery of asymmetric Wolbachia segregation in the Drosophila neuroblasts and migration to the germline niche cells indicate that specific targeting mechanisms are involved (Frydman et al., 2006; Albertson et al., 2009; Fast et al., 2011). Resolving this issue will require careful cellular analysis of Wolbachia localization patterns in a variety of tissues. Mechanisms of targeted Wolbachia localization within the arthropod CNS may shed insight into strategies utilized by microbes that target specific regions in vertebrate host brains. For example, Rickettsia, bacteria closely related to Wolbachia, preferentially infect brain cells compared with endothelial cells in the mammal host (Joshi and Kovacs, 2007). Toxoplasma gondii is a protozoan parasite widely prevalent in animals causing a variety of neuropathologies. T. gondii infections target neural and glial cells in the intermediate rodent hosts, causing diverse alterations in cellular activity (Kamerkar and Davis, 2012). In the rodent, T. gondii preferentially localizes to limbic system of the adult brain, in particular to the medial and basolateral amygdala (Vyas et al., 2007). For both Rickettsia and T. gondii, little is known concerning the molecular and cellular mechanisms involved in nervous system targeting. Our finding that Wolbachia targets specific regions of the adult Drosophila central nervous system provides an excellent opportunity to apply powerful Drosophila genetic approaches to this issue.

Factors intrinsic to Wolbachia influence its distribution in the brain

Previous studies demonstrated that factors intrinsic to the Wolbachia strain determine its localization in the insect oocyte, while host factors played a major role in influencing titre (Veneti et al., 2004; Serbus and Sullivan, 2007). To determine whether Wolbachia or host factors influence Wolbachia titre and distribution in the adult brain, we examined WMel and WPop in D. melanogaster, WRiv in D. simulans and WMel in D. simulans. All examined Wolbachia strains infect the optic lobes at low frequencies and predominantly reside in the central brain. Within the central brain, regional Wolbachia distribution differs among Wolbachia strains. WMel and WPop in D. melanogaster showed a relatively even distribution, while WRiv in D. simulans showed significantly higher titres in specific regions (Table 4). WRiv bacteria in D. simulans are present in much higher titres in regions 5 to 7 compared with regions 1 to 4 (regions 5 to 7 include soma medial and posterior to the ventrolateral protocerebrum lobe, and soma lateral to the lateral superior protocerebrum; regions 1 to 4 include soma between the superior medial protocerebra, soma posterior to the mushroom body, soma posterior to antennal lobes and soma anterior to the superior protocerebrum lobe). In contrast to WRiv, WMel in D. simulans exhibited an even distribution among the regions, similar to WMel in D. melanogaster (Table 4). This result leads to the conclusion that – similar to Wolbachia localization in the oocyte – the even versus uneven pattern of Wolbachia localization in the brain is intrinsic to the Wolbachia strain rather than to the host. Also similar to the oocyte, both the host species and bacteria strain influence Wolbachia titre. For example, the titre in regions 1 through 4 of WMel in D. simulans is intermediate between that of WMel in D. melanogaster and WRiv in D. simulans. In contrast, the cluster size of WMel in D. simulans is larger than those of either WMel in D. melanogaster and that of WRiv in D. simulans and appears to result from specific interaction between the D. melanogaster host with the WMel strain (Table 4). How the host impacts titre and how factors intrinsic to the Wolbachia strain influence tissue and cellular distribution is unclear. It may be that nutrient levels in the host cells play a key role in influencing titre while Wolbachia surface proteins that interact with specific host factors determine its localization. Support of this notion comes from the close association between Wolbachia and polarity determinants in the Drosophila oocyte (Serbus and Sullivan, 2007; Serbus et al., 2011).

Table 4. Summary of Wolbachia distribution characteristics and Wolbachia titre in brain and oocyte tissues
 WRiv/D. simWMel/D. simWPop/D. melWMel/D. mel
  1. From (*) Serbus and Sullivan (2007) and (**) L.R. Serbus (unpubl. obs.).

Region-specific distribution in CNSYesNoNoNo
Titre in brain (intracellular)+++++++
Titre in brain (cluster, extracellular)+N/A++++
Cluster size+++++++
Posterior oocyte localization (intracellular)No(*)Yes(*)Yes(**)Yes(*)
Oocyte titre++(*)+++(*)+(**)+(*)

Wolbachia is present both intracellularly and extracellularly in the adult brain

All of the examined Wolbachia strains formed clusters, with WPop having the largest and most numerous clusters. WPop clusters in D. melanogaster were first identified by EM imaging and were proposed to arise by fusion of several WPop-infected cells (Min and Benzer, 1997). However, our confocal imaging did not reveal any multinucleated cells. Furthermore, large WPop clusters lacked a host nucleus and were generally not encased within a clearly defined host membrane, indicating that bacterial clusters may also form through a different mechanism. Wolbachia is an obligate intracellular endosymbiont and previous work has shown that Wolbachia does not divide extracellularly (Rasgon et al., 2006). Wolbachia bacteria that are experimentally injected into the fly abdomen survive within the haemolymph for a few days before invading host cells, but become established only within host cells (Fast et al., 2011). Therefore, one possibility is that Wolbachia overproliferates in a small number of host cells, causing the cell to lyse. The data presented in this work raise the possibility that Wolbachia reside extracellularly within the adult brain. However, if these were transient aggregates that invaded nearby host cells, we would expect a high frequency of cellular infection near clusters, but the WPop cellular infection frequency was similar to WMel, indicating that bacteria originating from large WPop clusters are not invading nearby cells. It is also possible that these clusters represent the dense Wolbachia form that was found in EM analysis by Min and Benzer (1997) and might be a quiescent state. An alternative model to explain the presence of large bacteria clusters is aligned with the Min and Benzer cell fusion model, in that Wolbachia overproliferates in host cells leading to an expansion of host cell membrane. Our data revealed that some small clusters were encased by actin-rich staining (presumably remnants of a host plasma membrane). However, in this model, the host cells are severely aberrant and non-functional since they do not have an intact nucleus or host DNA. Previous studies have indicated that intracellular bacteria induce and/or block apoptosis (Gao and Abu Kwaik, 2000), as it has been shown for Wolbachia (Landmann et al., 2011; Zhukova and Kiseleva, 2012). Subcellular studies assaying factors such as organelle integrity, apoptosis and necrosis will further advance our knowledge into bacterial cluster formation and the consequences on host cell function. Surprisingly, our data indicate that, apart from the bacteria clusters, WPop has a relatively low cellular infection frequency and a low number of Wolbachia per cell. In this regard, the pathogenic WPop resembled the low titre WMel strain more than the high titre WRiv strain.

Host line identity rather than a general effect of Wolbachia infection determines response to chill coma, starvation and longevity

Our observation of WRiv in D. simulans is the first report of a microbe preferentially localizing to specific regions of the Drosophila central brain. At this point it is unclear if the preferential localization might translate into functional significance. Our experiments suggest that the host line identity determines how tetracycline treatment influences the flies' longevity and response to cold stress or starvation. In light of this result, it is not surprising that previous publications that analyse Wolbachia effects of physiological traits have come to varying conclusions. For example, removing Wolbachia by tetracycline decreased lifespan drastically (Toivonen et al., 2007), or had mixed effects (Driver et al., 2004; Fry et al., 2004). Our results confirm and extend those of Fry et al. who examined fitness effects (survival and fecundity) in inbred D. melanogaster lines. The authors concluded that beneficial or harmful Wolbachia effects depend on the host genetic background and sex (Fry et al., 2004). We expand their observations by additional physiological parameters and also by analysing the tetracycline effect on uninfected lines. We found that tetracycline also has a line-dependent effect, confounding the analysis of Wolbachia infection.

Previous analyses of host behaviour in response to Wolbachia infection have led to results indicating that host species and Wolbachia strain influence the observations, for example in olfactory-cued locomotion (Peng et al., 2008). It would be interesting to examine whether a similar variability can be found even among lines of the same Drosophila species. In this study we have not examined the variation of Wolbachia distribution and titre in the brain within lines of one species. However, it is interesting to speculate that the line-specific variations that are observed in physiological and behaviour parameters might be linked to a different, line-specific Wolbachia infection pattern in the brain.

In Drosophila, neural circuits implicated in sexual and defensive behaviours overlap considerably, yet recent reports have identified that differential gene expression in the central brain plays a critical role in regulating behaviour. For example, the male isoform of fruitless (fruM), a key regulator of male courtship behaviour, is expressed in only 2% of neurons, which are distributed into 21 clusters in specific regions of the adult brain (Stockinger et al., 2005). Gene expression profiles have indicated that Wolbachia can induce differential gene expression in a variety of hosts, including wasp (Kremer et al., 2012) and silkworm (Nakamura et al., 2011). Specifically, WMel has been shown to alter gene expression in D. melanogaster (Zheng et al., 2011b, 2011a) while WPop and WRiv have been reported to alter host gene expression in the mosquito (Hughes et al., 2011). To date, no genomic profiles have been performed specifically on adult brain tissue. Among the Wolbachia strain/host combinations examined in our studies, the data clearly indicate distinct differences in adult brains regarding Wolbachia distribution, cellular infection frequencies, and cluster size and frequency. Specific Wolbachia localization patterns in the brain may influence host-specific physiological and behavioural responses to Wolbachia.

Tetracycline treatment extends longevity

We observed that tetracycline increased longevity in most lines, including Wolbachia-free lines. In addition, increased longevity is strain dependent. This observation was surprising given that the lines were treated many generations before assaying for longevity. Because the effect was independent of a previous Wolbachia infection, this result could imply that the lines harbour additional bacteria that shorten their lifespan. A tetracycline effect on mitochondria has been reported for two generations after tetracycline treatment (Ballard and Melvin, 2007), but it is unlikely that mitochondria are still affected 6 months to 2 years after treatment. In contrast to our observation, a previous publication reported that a lack of bacteria in early Drosophila development decreases the flies' lifespan (Brummel et al., 2004). However, Brummel et al. compared flies in regular and axenic conditions, whereas our lines where grown under regular conditions that probably restored gut fauna. Min and Benzer found no tetracycline-effect on D. melanogaster longevity and report that tetracycline treatment immediately restores the original lifespan to flies that have been infected with the pathogenic Wpop bacteria strain (Min and Benzer, 1997). A tetracycline-induced life-shortening effect, attributed to Wolbachia loss, was observed in indy mutants, although an additional Wolbachia-infected strain was found to be unchanged by the antibiotics in that study (Toivonen et al., 2007). This long-lasting, line-dependent tetracycline effect will need to be taken into account in future comparisons of Wolbachia-infected lines and their cured control counterparts.


Wolbachia infection of adult brains is a conserved feature in D. simulans and D. melanogaster. Bacteria distribution within different brain regions depends on the Wolbachia strain, whereas the titre in the brain is determined by both the host species and the bacteria strain. In addition, we found that the pathogenic WPop Wolbachia strain infects host cells at a similar frequency as non-pathogenic strains, but forms more numerous and larger bacteria clusters than the benign strains WMel and WRiv. It appears that some of these aggregates are not contained within host cells, which may indicate that the bacteria have lysed the host cells. In spite of Wolbachia distribution into areas that control physiology and fly behaviour, the effect of a Wolbachia infection on individual D. melanogaster lines varies with the individual host lines.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Drosophila stocks

All RAL Drosophila lines were obtained from the Bloomington stock centre and are described by Ayroles et al. (2009). Infected and uninfected Oregon-R stocks used in this study are described by Ferree et al. (2005). D. melanogaster with WPop, D. simulans with WRiv or WMel and D. melanogaster with WMel are labstocks (Serbus and Sullivan, 2007). The infected ‘Turelli’ D. simulans flies were captured in California by the Michael Turelli lab (University of California, Davis). The Drosophila lines were reared on cornmeal-molasses-yeast food at 25°C, with a 12/12 h light/dark cycle.

Tetracycline treatment

Females laid eggs on food vials containing 25 mg of tetracycline in 100 ml of regular cornmeal-molasses food. Lines were established from individual females raised from egg to adulthood in these tetracycline-spiked vials. The offspring was analysed for Wolbachia infection by PCR and cytology to ensure they were cured. Experiments were performed on paired infected and cured lines at least seven generations after tetracycline treatment and up to 2 years after treatment.

Assaying Wolbachia infection status by PCR

Wolbachia infection status of each line was analysed by PCR and cytology. PCR: flies were crushed in PCR buffer (Sambrook et al., 1989) containing proteinase K (0.8 mg ml−1), heated to 60°C for 45 min, and to 95°C to for 10 min. The wsp sequence was amplified using the following primers: aacgctactccagcttctgc (reverse) and gatcctgttggtccaataagtg (forward). When indicated, the PCR products were cloned into pCR®2.1-TOPO (Invitrogen) and sent for sequencing (UC Berkeley sequencing facility).

Fixed cytology

Ovaries from adult female Drosophila were dissected in PBS and fixed in 3.7% Paraformaldehyde and Heptane, as described previously (Ferree et al., 2005). After RNase treatment, Wolbachia and host DNA were stained with Propidium Iodide (Ferree et al., 2005). Adult brain cytology: 2- to 5-day-old adult flies were briefly anaesthetized with CO2 and placed in a watch glass containing PBS (with 0.02% Triton X-100). Dissected brains were immediately fixed in PEM (100 mM PIPES, 2 mM EGTA, 1 mM MgSO4) with 2% paraformaldehyde for 16–18 min. Primary antibodies include mouse anti-Histone H1 (1:100, Millipore), and mouse anti-FasII (1:100, DSHB). For Wolbachia detection, we used mouse anti-CG9850 (1:100, a generous gift from Kyungok Cho, Korea Advanced Institute of Science and Technology), which was found to specifically bind to Wolbachia. It is often the case that antibodies generated against an Escherichia coli-expressed protein highlight Wolbachia (Cho, 2004). We verified the specificity of this antibody with the established Propidium Iodide staining (Fig. 1). Brains were incubated in primary antibody + PBST (0.1% Triton X-100) for 4 h at room temperature (or overnight at 4°C). Secondary antibodies included anti-mouse Cy5 (1:150, Invitrogen), anti-mouse Alexa Fluor 488 (1:400, Invitrogen), and conjugated-phalloidin488 (1:100 Invitrogen). Brains were incubated in secondary antibody + PBST (0.1% Triton X-100) for 1 h at room temperature. For PI staining, fixed brains were incubated in RNase (15.4 mg ml−1 PBS) for 2 h in a 37°C water bath and mounted in mounting medium containing PI (10 μg ml−1 PI, 1× PBS, 70% glycerol in water).

Live cytology

Adult flies were dissected in PBS and ovaries, brains, or fat bodies (from male and female flies) were placed into a drop of Syto-11 (Invitrogen, 1:100 dilution of stock in PBS) on a coverslip, 20 min on ice. Samples were then overlaid with a smaller coverslip and analysed for Wolbachia by confocal microscopy (Leica TCS SP2) (Casper-Lindley et al., 2011). For brains, broken pieces of coverslips were used as spacers to avoid sample squashing.

Behavourial and physiological assays

All assays were performed as described in Ayroles et al. (2009).

Image capture, quantification and preparation

All images were collected with a TCS SP2 confocal system on a Leica DM IRB inverted microscope. For adult brains, x–y–z three-dimensional image stacks were analysed and quantified with Lecia LAF AS Lite software. The Wolbachia cluster areas were measured by multiplying the longest axis and the orthogonal axis of the cluster. For circular-shaped clusters, area was measured as π-r2. Images were assembled with Photoshop CS4.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We gratefully acknowledge funds from NSF (MCB-1122252). R.A. received a Faculty Development Grant from Albion College. M.R. received funds from the MBRS/MARCS programme at UCSC. We thank Ary Hoffmann, Julien Ayroles and Steven Lindley for help with statistical analysis. We are thankful for productive discussions with Wolfgang Miller. We thank Mark Readdie and the Landels-Hill Big Creek Reserve for assistance to collect flies. Kyungok Cho generously provided us with the anti-CG9850 antibody.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1. Egg chambers (upper panels) and brain sections (lower panels) from wild-caught, Wolbachia-infected D. simulans lines. Host DNA and Wolbachia are stained with Propidium Iodide and Wolbachia are visible as small puncta in the egg chambers or among the host neuroblast nuclei. Scale bars, 20 μM. Similar images were obtained with all examined D. melanogaster lines.


Fig. S2. A. Egg chambers from infected lines (712 and 852), tetracycline-cured lines (712 and 852) and uninfected lines (375 and 517).

B. PCR from infected and cured D. melanogaster lines as indicated. Scale bars, 10 μM.


Table S1. Quantification of individual Wolbachia bacteria in different brain regions (data of the summary shown in Figs 3K and 4)


Table S2.n of the physiological experiments shown in Fig. 7. Status indications are: I = infected, T = infected, treated, U = uninfected, UT = uninfected, treated.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.