Polar cell fate stimulates Wolbachia intracellular growth

During their life cycle, all animals host a variety of microorganisms in their bodies. These microbes preferentially localize to specific tissues or organs, a phenomenon termed tissue tropism. Understanding the mechanisms and consequences of specific tissue tropism is key to elucidating host-microbe interactions. One of the largest world-wide pandemics is caused by maternally transmitted alphaproteobacteria belonging to the genus Wolbachia, which infect a large fraction of invertebrates, including isopods, parasitic filarial worms and insect vectors of infectious diseases (Hughes et al., 2011; Kambris et al., 2010; Moreira et al., 2009; Walker et al., 2011). Wolbachia are stably maintained in host populations and have a profound effect on host biology, including their evolution, physiology, reproduction, immunity and development (Werren et al., 2008). During evolution, Wolbachia have developed tropism to specific host tissues to facilitate their efficient vertical transmission (Ferree et al., 2005; Frydman et al., 2006; Hadfield and Axton, 1999; Serbus and Sullivan, 2007; Veneti et al., 2004; Werren et al., 2008). Infection of the germline in the gonads is essential for maternal transmission. However, Wolbachia also infect several different somatic tissues of the host (Cheng et al., 2000; Clark et al., 2005; Dobson et al., 1999; Espino et al., 2009; Fischer et al., 2011; Hosokawa et al., 2010; Pietri et al., 2016).

In the Drosophila gonads, Wolbachia infect the germline and the stem cell niches at high levels (Fast et al., 2011; Frydman et al., 2006; Toomey and Frydman, 2014; Toomey et al., 2013). Stem cell niches are microenvironments that support the stem cells. In females, these encompass both the niche supporting the somatic stem cells (SSCs) and that supporting the germline stem cells (GSCs) (Fig. 1A-C), whereas in the male there is a single niche for both SSCs and GSCs, known as the hub (Fig. 1D,E). The somatic stem cell niche (SSCN) harbors the SSCs, which generate all the somatic cells that envelope the germline and secrete the egg shell (Fig. 1A). Wolbachia tropism to the SSCN has been shown to be important in their transmission to the germline and therefore to the next generation (Toomey et al., 2013). Moreover, previous work has demonstrated that upon recent infection, Wolbachia first colonize the SSCN of adult Drosophila melanogaster (Frydman et al., 2006).

These observations of niche tropism were from differentiated niches in adults. The kinetics of Wolbachia tropism to the niches during their specification and development has not been defined. This analysis is not easily accomplished because the morphogenesis of these niches occurs prior to adulthood. The SSCN is specified during pupal development in the presence of differentiated germ cells (Nystul and Spradling, 2007; Sahai-Hernandez and Nystul, 2013; Vlachos et al., 2015). Furthermore, the SSCN precursor cells are not predefined, making it difficult to study tropism during niche morphogenesis (Sahai-Hernandez and Nystul, 2013; Vlachos et al., 2015). The male stem cell niche, termed hub, also displays Wolbachia tropism. Nevertheless, the specification of the hub occurs in mid embryogenesis (Le Bras and Van Doren, 2006; Sheng et al., 2009), and its development spans multiple life stages of the insect. Therefore, to determine the kinetics of Wolbachia accumulation to these somatic tissues during development requires quantification of multiple developmental stages, including pupal stages, making stem cell niches a challenging system to study Wolbachia tropism during their development.

We probed for Wolbachia tropism to other cell types during Drosophila oogenesis, where most developmental stages of different cell types from stem cell division to egg maturation can be observed in a single adult fly (Spradling, 1993; Wu et al., 2008). Furthermore, it is a well-characterized system with a vast array of cellular and molecular tools and markers available for each cell type. Each Drosophila egg begins as a 16-cell germline cyst, from which one cell will become the oocyte and the remainder will become the supporting nurse cells. The cyst then gets encapsulated by a monolayer of somatic follicle cell precursors (blue cells in Fig. 1A). As the cyst exits the germarium, a population of follicle cells (FCs) on either pole ceases to proliferate, differentiating into a pair of polar cells (PCs) at the poles of the egg chamber. From the same precursor population, cells differentiate into a stalk between the consecutive chambers. The other FCs, known as lateral FCs, remain undifferentiated and keep dividing to encapsulate the germline (González-Reyes and St Johnston, 1998; Grammont and Irvine, 2002; Margolis and Spradling, 1995; Xi et al., 2003). In a single ovary, we can observe all events of PC morphogenesis. Here, we describe preferential Wolbachia targeting in the PCs. This makes the PCs a powerful system to study Wolbachia tropism during development. Moreover, there are genetic tools that allow easy manipulation of these cells, including the capability to induce ectopic PCs.

Wolbachia PC tropism is a novel system to study host-Wolbachia interactions in the somatic cell types. Using confocal microscopy and transgenic flies to generate ectopic PCs, we demonstrate that Wolbachia show remarkable specificity and coordination of their accumulation in PCs with specific developmental events during oogenesis. Furthermore, we show that Wolbachia PC tropism is evolutionarily conserved in the Drosophila genus, but is absent in another dipteran, the mosquito Culex pipiens.

In the fly ovary, usually Wolbachia accumulate at high levels in the germline. Upon imaging Wolbachia wMelPop strain in D. melanogaster, we noticed a consistently high Wolbachia accumulation at the polar regions of the follicular epithelium of the egg chamber, reaching levels equivalent to those in germline infection (Fig. 2B). The pattern of this accumulation was consistent with the localization of the PCs (schematic in Fig. 2A). Using an antibody against Fasciclin III (FasIII) to label PCs (Patel et al., 1987; Ruohola et al., 1991), we confirmed that Wolbachia, although present at low levels in the lateral FCs, accumulate at high levels in the PCs (Fig. 2B).

To address whether PC tropism is evolutionarily conserved, we surveyed 10 Wolbachia strains that naturally infect seven Drosophila species. Using fluorescence in situ hybridization (FISH) to label Wolbachia and FasIII to label PCs, we quantitatively assessed Wolbachia PC tropism in ovaries of all 10 Wolbachia-Drosophila pairs. In every ovary analyzed, we found the presence of Wolbachia in PCs (Fig. 2B-K). To quantify Wolbachia levels, voxel density from representative z-stacks of stage 8 egg chambers was determined using image analysis software (see Fig. S1 and Materials and Methods). By comparing relative Wolbachia levels in PCs with those in lateral FCs, we found that Wolbachia was enriched in the PCs relative to the follicular epithelium in each of the Drosophila-Wolbachia pairs except for Dsim wRi (Fig. 2L). Fitting the PC tropism phenotype to the Wolbachia phylogenetic tree (Fig. S2) (Paraskevopoulos et al., 2006) indicated that the ancestral strain of Wolbachia wRi had PC tropism and this feature was lost in wRi, most likely during the separation between wRi and its closest living relative analyzed, wSh (Fig. S2). These data indicate a strong selective pressure for an evolutionarily conserved Wolbachia tropism to PCs.

PCs are believed to be a unique feature of the Diptera order. Although PCs have been identified in all dipterans investigated, so far they have never been found in nondipterans (Jaglarz et al., 2008). Here, we identified PC tropism in most Drosophila species, a higher dipteran (Brachycera). To determine whether Wolbachia PC tropism is pervasive outside the Drosophila genus, we investigated the mosquito Culex pipiens, a lower dipteran (Nematocera), an evolutionarily distant species. C. pipiens are infected with wPip, a Wolbachia strain highly divergent from wMel (Hertig, 1936). Putative PCs have been visualized in C. pipiens by transmission electron microscopy (Soumaré and Ndiaye, 2005). We found that antibody against Drosophila N-cadherin labels two FCs at opposing poles of Culex follicles, possibly the mosquito PCs. Double staining of Wolbachia and N-cadherin shows that Wolbachia do not accumulate in the putative PCs of C. pipiens (Fig. S3). Therefore, PC tropism is either an evolutionary novelty that occurred in higher dipterans or it was lost in Culex.

To characterize this novel Wolbachia tropism, we carried out further analyses in D. melanogaster and its three Wolbachia endosymbiont strains, wMel, wMelCS and wMelPop. The egg chamber development can be divided into 14 stages (Fig. 3A) (King, 1970; Spradling, 1993). PCs differentiate early in follicular epithelium morphogenesis (Grammont and Irvine, 2002; Margolis and Spradling, 1995). Two pairs of PCs can be unequivocally distinguished from the remaining follicular epithelium by stage 4 of oogenesis by immunostaining for FasIII (Besse and Pret, 2003; Khammari et al., 2011; Ruohola et al., 1991). We quantified Wolbachia density in PCs relative to lateral FCs at various stages of oogenesis. In all three strains of Wolbachia tested, we observed an elevated Wolbachia density in PCs relative to lateral FCs, starting from stage 4 up until stage 10 (Fig. 3B-F). This observation shows that Wolbachia infect PCs early in their development and maintain a high titer throughout their development.

To further study the kinetics of Wolbachia PC tropism, we quantified Wolbachia titers in PCs relative to lateral FCs in the same egg chamber at various stages of oogenesis (4, 5, 8, 9 and 10). Normalization to lateral FCs was performed to account for variability of staining and confocal image acquisition across different experiments, and the overall Wolbachia titers across different egg chambers, ovaries and organisms (see Materials and Methods). In all three strains of Wolbachia, we found that bacterial density in the PCs relative to lateral FCs increased steadily from stage 4 to stage 10 (Fig. 3G-I). This shows that Wolbachia have a preferential tropism to the PCs and increase in density as they progress through development.

As the egg chamber progresses through development, the lateral FC number increases by ∼20-fold until stage 6, while the number of PCs remains at four (Assa-Kunik et al., 2007; Wu et al., 2008; Xi et al., 2003). After stage 6, the FCs undergo endoreplication and increase in size (González-Reyes and St Johnston, 1998; Wu et al., 2008). The increased number and size of lateral FCs could inflate the relative Wolbachia density in PCs. To address this, we compared the Wolbachia densities in lateral FCs over different stages of oogenesis. For wMel and wMelCS, we observed no significant decrease in Wolbachia density in lateral FCs over different stages of development (Fig. S4A,B). Surprisingly, we observed a steady decrease in wMelPop density in lateral FCs as the egg chambers went through oogenesis (Fig. S4C). wMelPop is a pathogenic strain of Wolbachia, which replicates at a high rate and leads to premature deaths of infected individuals (Woolfit et al., 2013). In the PCs, it reaches a high density early in oogenesis and maintains these levels throughout oogenesis (Fig. S6), even as the levels in the lateral FCs decrease (Fig. S4). Together, these results show that our measurements of Wolbachia tropism to the PCs are not augmented by the normalization to lateral FCs.

Interestingly, we observed a large increase in Wolbachia density between stages 8 and 9 in PCs (Fig. 3G-I). Wolbachia density increases by about threefold in wMel and wMelCS and twofold in wMelPop. In stage 9, the anterior PCs, along with a few surrounding cells (called border cells), migrate amidst the nurse cells to associate with the anterior of the oocyte (Fig. S5). During this migration, they are closely associated with the germline (Montell, 2003; Montell et al., 1992). Therefore, border cell migration provides an opportunity for Wolbachia to traverse from PCs to the germline. Using previously described kinetics of oogenesis (Lin and Spradling, 1993), we plotted Wolbachia density as a function of time from stage 4 to 10 (Fig. S7). This analysis shows that Wolbachia PC density increases moderately (wMel, 1.12-fold; wMelCS, 1.68-fold; wMelPop, 1.04-fold) between stages 5 and 8 over a period of ∼17 h (Fig. S7). However, between stages 8 and 9, Wolbachia PC density increases rapidly (wMel, 2.73-fold; wMelCS, 2.32-fold; wMelPop, 2.18-fold) in ∼8 h. Remarkably, this shows that Wolbachia coordinate their replication and accumulation with specific host developmental events.

The anterior PCs are required for border cell migration (Montell, 2003) that later contribute to micropyle and dorsal appendage formation in the mature egg (Ward and Berg, 2005). The posterior PCs facilitate oocyte localization and anteroposterior dorsoventral axes formation (González-Reyes et al., 1995; Roth et al., 1995). As each set of PCs have distinct functions, we tested for Wolbachia differences between them. Upon quantification, we observed that, in most stages of oogenesis, there was no significant difference in Wolbachia density between the anterior and posterior PCs (Fig. 4G). However, wMel and wMelPop accumulate slightly more in the anterior PCs at stages 8 and 9, respectively (Fig. 4I,G). Although, the anterior and posterior PCs differ in their functions, they share common developmental regimes (Besse and Pret, 2003; Grammont and Irvine, 2001; Torres et al., 2003). These observations demonstrate that the Wolbachia tropism to PCs is determined by factors common to both populations of PCs.

FasIII immunostaining identifies PCs unequivocally only at stage 4 and beyond. It is not clear at what stage of PC maturation Wolbachia accumulate in PCs. Here, we addressed this question by two methods.

First, we assessed Wolbachia levels in the stalk cells (SCs), which form a narrow stem connecting subsequent egg chambers (King, 1970; Wu et al., 2008). The SCs are specified when egg chambers exit the germarium in stage 2 of oogenesis. SCs and PCs derive from the same precursor population that is separated from the lateral FC progenitors (Fig. 8B,H) (Chang et al., 2013; Tworoger et al., 1999). From the PC/SC precursors, one population differentiates to PC progenitors and the other into SC progenitors (Fig. 8H) (Chang et al., 2013; Tworoger et al., 1999). We hypothesize that if Wolbachia infect only the PC progenitors, we would expect no or little accumulation of bacteria in SCs. Upon quantifying Wolbachia levels in SCs, we found that only a small proportion of SCs had any Wolbachia infection at all (Fig. 5). wMel and wMelCS infected only 28% (7/25) and 36% (9/25) of SCs, respectively (Fig. 5J). This contrasts with 100% of infected PCs in wMel (144/144) and wMelCS (201/201). wMelPop infected a relatively higher proportion (68%, 31/45, Fig. 5J) of SCs. However, even among the infected SCs, we observed a low Wolbachia density as compared with PCs (Fig. 5D-F). Only in rare instances, we found SCs with Wolbachia levels comparable to the PCs (Fig. 5G-I). Furthermore, the frequency of infected lateral FCs was consistently lower than the frequency of infected PCs, which was 100% (179/179). These data suggest that Wolbachia do not accumulate in the PC/SC precursor population, and that Wolbachia accumulate in PC precursors after they separate from the common PC/SC precursors (Fig. 8H).

We confirmed this possibility by quantifying Wolbachia in the common PC/SC precursors, located at the anterior part of the ovariole (Fig. 6A). The FCs between regions 2b and 3 of the germarium are thought to be the PC/SC precursors (yellow cells in Fig. 6B) (Bai and Montell, 2002; Chang et al., 2013; Larkin et al., 1996; Tworoger et al., 1999). We found no cells within this region that consistently had high Wolbachia accumulation (Fig. 6C,C′, Fig. S9A-C″). We also compared Wolbachia density in region 2b FCs (yellow dashed line in Fig. 6D″) with that in lateral FCs (green dashed line in Fig. 6D″) and observed no Wolbachia enrichment in region 2b for all three strains tested (Fig. 6I). Moreover, the proportion of Wolbachia-infected cells was equivalent between regions 2b and 3 of the germarium (Fig. 6H). In stage 3, we observed a few instances of early egg chambers that had matured posterior PCs (marked by higher FasIII staining, Fig. 6E′-E″) but not anterior PCs. In these, we observed that Wolbachia had already accumulated in the posterior differentiated PCs suggesting that PC specification triggers Wolbachia intracellular growth. By stage 4, most of the anterior PCs are completely differentiated and also present high Wolbachia accumulation (Fig. 6F-F‴).

Furthermore, we used another method to identify PC/SC precursors in the germarium: evaluation of Eyes absent (Eya) expression. Loss of Eya expression is considered one of the first markers of the PC/SC lineage in the germarium (Bai and Montell, 2002; Chang et al., 2013). We identified cells in region 2b of the germarium that had markedly reduced Eya expression compared with lateral FCs in region 3 (Fig. 6G-G″, Fig. S9D-E″). Quantification of Wolbachia density in Eya-lacking cells (yellow dashed line in Fig. 6G′) compared with Eya-expressing cells (green dashed line in Fig. 6G″) showed no enrichment of Wolbachia in these putative PC/SC precursors (Fig. 6J), corroborating that Wolbachia accumulate in the PCs only after their specification.

To further test our hypothesis, we investigated whether PC fate is sufficient to drive high Wolbachia density. Using transgenic flies, we induced ectopic PCs. Knockdown of Eya is sufficient to induce a PC fate (Bai and Montell, 2002). We expressed RNAi against eya under the control of GR1-Gal4, an FC driver (Goentoro et al., 2006; Gupta and Schupbach, 2003) that is expressed beginning in stage 3 of oogenesis (Fig. 7A) (Etchegaray et al., 2012). We observed multiple FasIII-positive ectopic PCs, primarily in stages 5 and 8. We found that every ectopic PC observed was infected with Wolbachia in stage 5 (23/23, Fig. 7B-B″) and stage 8 (33/33, Fig. 7C-C″) egg chambers. By contrast, all these egg chambers displayed large fractions of FCs with no Wolbachia infection (56/56, Fig. 7B,C). Upon quantification, we found that Wolbachia density in these ectopic PCs was comparable to Wolbachia density in the normal PCs in the respective stages (Fig. 7D). Taken together, these findings demonstrate that Wolbachia accumulate after PC specification from the common PC/SC precursors (Fig. 8H).

Next, to investigate whether Wolbachia accumulate by over-replication in PCs or by uptake from surrounding FCs, we analyzed the egg chambers with ectopic PCs. We observed large populations of ectopic PCs in each egg chamber analyzed. The density of Wolbachia in ectopic PCs was equivalent to that in normal PCs (Fig. 7C); however, the ectopic PCs had a substantially increased volume compared with the normal PCs in both stage 5 (∼13-fold) and stage 8 (∼6.4-fold) egg chambers (Fig. 7E). To account for the additional bacteria, Wolbachia either over-replicate in the PCs or migrate from the surrounding FCs. As the Wolbachia in the PCs are tightly packed in clumps containing several bacteria, it is challenging to count individuals or visualize dividing bacteria. Therefore, we investigated the second possibility by analyzing Wolbachia accumulation in the surrounding FCs. If Wolbachia growth in the PC is caused by uptake from the surrounding FCs, we would expect a reduction in Wolbachia in FCs surrounding the ectopic PCs as compared with FCs surrounding the normal PCs. Upon comparing these values, we found that the FCs surrounding the ectopic PCs have comparable Wolbachia density to that in the FCs surrounding the normal PCs in corresponding stages (Fig. 7F, Fig. S10). These findings suggest that Wolbachia accumulate in PCs owing to over-replication and not to uptake from surrounding FCs.

The influence of bacteria in host development is clearly evident for the intracellular bacteria Wolbachia. Wolbachia, one of the most common symbionts in arthropods, are maternally transmitted and affect several aspects of host development and reproduction. In cases in which Wolbachia symbiosis is obligatory, host development depends on the presence of the bacteria. For instance, the wasp Asobara tabida requires Wolbachia for completion of oogenesis and egg maturation (Dedeine et al., 2005). Several filarial nematodes also require Wolbachia for successful reproduction. In the absence of Wolbachia, embryonic and larval development are impaired (Slatko et al., 2010).

Despite extensive evidence of Wolbachia affecting host development and reproduction, the converse aspect of this interaction, that is, how host developmental pathways affect Wolbachia intracellular growth, is less well established. In the literature, there are few examples of coordination of Wolbachia growth with host development. In certain mosquito species, when adverse environmental conditions suspend embryonic development and eggs enter diapause, Wolbachia levels are reduced (Ruang-areerate et al., 2004). Another remarkable example is the tortuous path that Wolbachia undertake to infect the female germline in filarial nematodes. In several species of the Onchocercidae family, Wolbachia are excluded from precursors that will form the germline. However, later in development, they invade the germline from distal tip cells, the worm equivalent of the germline stem cell niche (GSCN) (Landmann et al., 2012). Wolbachia tropism to distal tip cells facilitates maternal transmission.

In the Drosophila genus, Wolbachia also have tropism to stem cell niches. Tropism to the ovarian SSCN (Fig. 1C) is ubiquitous in all Drosophila species tested and contributes to increasing Wolbachia germline titers (Toomey et al., 2013). However, in this system, determining the kinetics of infection from SSCN progenitor cells to ovary maturation is challenging. Unlike worms, in which the precursor cells and their lineages have been traced, the precursors of the SSCN in Drosophila ovaries are not easily identified. The SSCN forms during pupal development from a dynamic, nondedicated population of cells, originating from signaling between the germline and maturing stalk cells (Sahai-Hernandez and Nystul, 2013; Vlachos et al., 2015). Here, we identify a novel Wolbachia tropism to the PCs of Drosophila, a cell type that allows us to determine the kinetics of tropism throughout all developmental stages from differentiation of progenitor cells to the maturation of the cell type.

The PCs are a subset of FCs present at either pole of the developing egg chamber and have multiple signaling pathways in common with the stem cell niches during development (Decotto and Spradling, 2005; Le Bras and Van Doren, 2006; Wu et al., 2008; Xi et al., 2003). In a survey across various Drosophila species, PC tropism was found to be ubiquitous in most Drosophila species tested and in all stages of PC development (Fig. 2). However, one Wolbachia strain (wRi) did not have a preferential Wolbachia tropism to PCs. We fitted the PC tropism character to Wolbachia phylogeny created using multilocus sequence typing (Fig. S2) (Paraskevopoulos et al., 2006). This analysis shows that the ancestral strains of Wolbachia had PC tropism that was lost in wRi during the recent separation from its closest living relative, wSh. wRi has a highly mosaic genome with a multitude of mobile elements and breakpoints (Klasson et al., 2009). It is considered one of the most highly recombining intracellular bacterial genomes known to date (Klasson et al., 2009). This has led to wRi exhibiting highly variable phenotypes. For instance, wRi infection initially caused a fecundity deficit in Drosophila simulans upon recent infection in 1988. However, it evolved within 16 years to give a fecundity benefit by 2004 (Weeks et al., 2007). Moreover, it has evolved variable tropism compared with its closest living relative, wSh, to both the ovarian GSCN as well as the testis hub (Toomey and Frydman, 2014). Therefore, it is not surprising that wRi has lost its PC tropism phenotype, an ancestral characteristic in the Drosophila genus.

We studied three different Wolbachia strains infecting D. melanogaster in further detail. We observed that Wolbachia PC tropism occurred early in oogenesis (stage 4) and persisted up until late oogenesis. As the egg chambers progress through oogenesis, we observed that Wolbachia density in PCs relative to the lateral FCs increased constantly. As the egg chamber develops, the number of lateral FCs increase from ∼50 in stage 4 to ∼1000 in stage 6 (González-Reyes and St Johnston, 1998; Wu et al., 2008; Xi et al., 2003). However, during this time the number of PCs remains constant (two at each pole). So, the apparent increase in Wolbachia PC density could be attributed to the decrease in density in other FCs. While the number of FCs increases rapidly by ∼20-fold, it is possible that Wolbachia are unable to replicate fast enough to maintain their density. However, in PCs, they can replicate to a much higher level as the number of cells does not increase. To address this, we compared Wolbachia densities in PCs alone and lateral FCs alone across various stages of oogenesis (Fig. S4). This shows that the Wolbachia densities in FCs do not increase over time, whereas the densities in PCs do increase. Therefore, the increase in Wolbachia in the PCs is not an artifact of normalization to the lateral FCs. Furthermore, these data also suggest that Wolbachia coordinate their intracellular levels with lateral FC development. The wMel and wMelCS strains coordinate their growth with lateral FC mitosis until stage 6, and volume increase until stage 10, remaining at constant densities.

However, we found that wMelPop density in lateral FCs decreased steadily from stage 4 to stage 10 (Fig. S4C). In this case, the high density in PCs could be partially contributed by the normalization to lateral FCs. However, the absolute quantification of Wolbachia densities in the PCs shows that a high density is achieved early in oogenesis and those high levels are maintained (Fig. S6). wMelPop, a pathogenic strain of Wolbachia, over-replicates in insect tissues and leads to cell lysis in neuronal cells (Min and Benzer, 1997). Previously, we also observed cell lysis of hub cells infected with wMelPop (Toomey and Frydman, 2014). This suggests that wMelPop bacteria have lost their coordination with the host and keep growing even when the intracellular environment no longer can support their growth. In agreement, our data in the PCs also indicate that unlike the nonpathogenic strains, wMelPop bacteria lost their coordination with host cell developmental events (Fig. S6).

We observe a substantial increase in Wolbachia density between stages 8 and 9 (Fig. 3G-I). This indicates that certain signaling pathways specific to PCs are involved in the regulation of Wolbachia density. Calculating the rate of Wolbachia PC accumulation over time further shows that Wolbachia increase their intracellular accumulation rapidly between stages 8 and 9 of oogenesis over a short period of time (Fig. S7). At the end of stage 8, the anterior PCs, along with six to eight border cells, detach from the epithelium and migrate through the egg chamber to the anterior of the oocyte. During this process, the PCs are closely associated with the germline and provide ample opportunity for Wolbachia to traverse into the germline. Previously, Toomey et al. had proposed that the amplification of Wolbachia in the follicular epithelium would be a crucial step for Wolbachia to colonize the germline at high titers (Toomey et al., 2013). Moreover, in both parasitic worms Brugia malayi and Litomosoides sigmodontis, Wolbachia have been shown to preferentially replicate in the rachis, a central actin-rich structure which connects the distal ovaries, before colonizing newly formed germ cells (Landmann et al., 2012). Preferential replication of Wolbachia in the PCs could act as an amplification step between Wolbachia tropism to the SSCN and the mature germline. In certain Drosophila species, Wolbachia germline infection is inconsistent, wherein certain egg chambers lack Wolbachia in their germline (Casper-Lindley et al., 2011). However, all the eggs laid by these females have Wolbachia, suggesting that Wolbachia re-enter the germline from somatic cells in the egg chamber. In D. melanogaster, electron micrographs show Wolbachia invading the germline from FCs (Toomey et al., 2013). Our data suggest that the stage 9 migration of anterior PCs provides an opportunity for Wolbachia transfer from PCs to the germline. These findings therefore demonstrate that Wolbachia can coordinate their replication and intracellular accumulation with specific host developmental events to facilitate their efficient transmission. Wolbachia has been shown to respond to specific host developmental events. For instance, in the parasitic worm Brugia malayi, Wolbachia levels remain constant in the microfilaria and the larval stages, but increase significantly within one week of infection of a mammalian host (Fenn and Blaxter, 2004; Landmann et al., 2010; McGarry et al., 2004).

If stage 9 migration of PCs is a viable option for Wolbachia to enter the germline, it follows that the density in anterior PCs should increase at a much higher rate than in the posterior ones. Some observations do support our hypothesis. For instance, wMel has a higher density in stage 8 anterior PCs (just before the migration event) and wMelPop has a higher density in stage 9 anterior PCs (during the migration event). However, we observed no significant difference between Wolbachia levels in anterior and posterior PCs in other stages of wMel- or wMelPop-infected flies, or any stage of wMelCS-infected flies. This indicates that signaling pathways common to both sets of PCs are important for Wolbachia accumulation, including Jak-Stat signaling and DE-Cadherin/Armadillo-mediated cell adhesion (Niewiadomska et al., 1999; Silver and Montell, 2001). Further, among the migrating cells, we observe an increased Wolbachia level specifically in the PCs and not in the border cells (Fig. 4, Fig. S5A″,B″), indicating that Wolbachia tropism to the PCs is highly specific in nature. Future studies will aim at characterizing signaling pathways that would be important for this stage-specific increase in Wolbachia titers.

PCs are determined early in egg chamber morphogenesis. Somatic stem cells in the germarium divide and undergo transient amplification to give rise to all FCs surrounding the germline. Unlike other FCs, which keep dividing until stage 6 of oogenesis, the PCs and SCs cease division as soon as they are specified in the germarium (Assa-Kunik et al., 2007; González-Reyes and St Johnston, 1998; Roth, 2001; Torres et al., 2003; Wu et al., 2008). The PCs and SCs are thought to arise from the same precursor population (Larkin et al., 1996; Tworoger et al., 1999). Recently, Chang et al. corroborated the presence of PC/SC precursors by showing the expression of Castor, a zinc finger protein, in the PC/SC precursor population in region 2b of the germarium (Chang et al., 2013). Considering these findings, if Wolbachia accumulate highly in these PC/SC precursor populations, we should find similar levels of Wolbachia in the SCs. However, when investigated, we found that only a small proportion of wMel (29%, Fig. 5J) and wMelCS (35%, Fig. 5J) infect SCs (Fig. 5A-C). wMelPop unsurprisingly infected ∼70% of SCs (Fig. 5J), but the density was rarely as high as a PC infection (Fig. 5D-F). The PC/SC precursors reside in region 2b of the germarium (Bai and Montell, 2002; Chang et al., 2013; Larkin et al., 1996; Tworoger et al., 1999). Although the exact separation point of PC and SC lineages is not clear, if Wolbachia are already enriched in the PC precursors, we might observe a few cells with high bacterial density compared with the lateral FCs in region 3. However, we observed no single cell/group of cells which had a preferential accumulation of Wolbachia in this region. Moreover, the overall density of Wolbachia in these FCs was comparable to that in the lateral FCs (Fig. 6G). The putative PC/SC precursors in region 2b can also be identified by a loss of Eya expression (Bai and Montell, 2002; Chang et al., 2013). We observed no specific Wolbachia accumulation in these putative PC/SC precursor populations (Fig. 6G-G″). We also observed certain egg chambers with matured posterior anterior PCs (Fig. 6E-E‴). In these, we observed Wolbachia preferential accumulation in the matured PCs, but not in precursors including the SC precursors. These results indicate that Wolbachia preferentially accumulate in PCs after differentiation of the PC lineage from the common PC/SC lineage.

To further distinguish the lineage specificity of Wolbachia accumulation, we induced ectopic PCs by expressing eya RNAi under the control of the GR1-Gal4 driver. Knockdown of Eya, a known PC fate repressor, led to the induction of multiple FasIII-positive ectopic PCs (Bai and Montell, 2002; Grammont and Irvine, 2002). GR1-Gal4 drives gene expression beginning at stage 3 of oogenesis (Etchegaray et al., 2012), and consistent with that we found a majority of ectopic PCs in stages 5 and 8. As the ectopic PCs were induced after stage 3, we can be certain that these cells had assumed lateral FC fate prior to becoming PCs. Wolbachia infected all ectopic PCs at a high density (Fig. 7B,C). By contrast, many of the lateral FCs lack Wolbachia altogether, suggesting that induction of PC fate is sufficient to drive Wolbachia tropism to these cells which had previously acquired a lateral FC fate.

Our results also suggest that Wolbachia accumulate in PCs owing to over-replication. Upon eya RNAi, we observe large populations of ectopic PCs in both stage 5 and stage 8 egg chambers (Fig. 7A-C), with densities equivalent to those of the normal PCs (Fig. 7D). The volume occupied by these ectopic PCs is six- to 13-fold larger than that of the corresponding normal PCs (Fig. 7E). This means that there are about six- to 13-fold more Wolbachia in the ectopic PCs compared with normal PCs. If Wolbachia from the surrounding FCs migrate to the PCs to make up for this increase, we would expect the FCs surrounding ectopic PCs to have a much lower Wolbachia density compared with the FCs surrounding the normal PCs. However, we observe no difference in Wolbachia between FCs surrounding the ectopic PCs and FCs surrounding the normal PCs. Further, when we compare FC Wolbachia densities across various stages of oogenesis, we see that Wolbachia densities remain constant from stage 4 to 10 (Fig. S4), whereas PC Wolbachia densities increase significantly (Fig. 3). Taken together, these results suggest that the mechanism of Wolbachia growth in the PCs is over-replication in the PCs and not migration from surrounding lateral FCs, although uptake from neighboring cells cannot be completely ruled out.

To summarize our findings, we propose a model (Fig. 8) for Wolbachia accumulation in the PCs during development. As SSCs divide in the germarium, their progeny assume either a PC/SC precursor fate or lateral FC fate (Fig. 8B,C). At around stage 2 of oogenesis, the PC/SC precursors differentiate into PCs (red in Fig. 8D,D′) or SCs (pink in Fig. 8D,D′). Wolbachia start accumulating in PCs once they separate from the PC/SC precursors (Fig. 8H). When oogenesis progresses further, between stage 4 and 8, Wolbachia (green dots in Fig. 8D′,E′) divide rapidly to increase their intracellular levels in PCs, whereas the Wolbachia levels in lateral FCs (gray) remain mostly constant. Between stage 8 and 9 (Fig. 8E-G′), Wolbachia levels increase rapidly in the PCs. At stage 9, the anterior PCs migrate through the germline allowing for a perfect opportunity for Wolbachia to traverse into the germline (Fig. 8F,F′). Also, high Wolbachia density in the posterior PCs at this stage means that Wolbachia can traverse directly into the oocyte (Fig. 8G,G′).

Being maternally transmitted, Wolbachia need to colonize the germ cells at high densities in their hosts. However, in many hosts, Wolbachia end up colonizing other somatic cell types at high densities (Frydman et al., 2006; Landmann et al., 2012; Toomey and Frydman, 2014; Toomey et al., 2013) and then migrate to the germline during their maturation. It has been challenging to study Wolbachia accumulation to these somatic cell types for several reasons. Here, we describe a novel Wolbachia tropism to the PCs of the Drosophila ovary – a molecularly well characterized system that will be extremely beneficial to study the molecular mechanisms of Wolbachia tropism to somatic tissues of the host. It is vital for symbionts to coordinate their accumulation with host developmental events and keep their replication in check so that they do not harm the host. Wolbachia have been shown to be able to sense certain host developmental cues and coordinate their accumulation (Landmann et al., 2010). Here, we demonstrate that even in the PCs, Wolbachia are able to coordinate their replication with specific host developmental events and accumulate at very high densities. Overall, we have demonstrated a novel Wolbachia tropism to the PCs of Drosophila ovaries. Further studies of the mechanisms of this tropism will shed light on the molecular cues utilized by Wolbachia to target specific cells in the host and sense host developmental events to coordinate their high intracellular accumulation.

We analyzed eight different species of the Drosophila genus for this study. Stocks analyzed in this study and their sources are shown in Table S1. Flies were raised at 25°C and fed typical molasses, yeast, cornmeal, agar food, with the exception of Drosophila sechellia flies, which were supplemented with reconstituted Noni Fruit (Hawaiian Health Ohana) (Amlou et al., 1998). The three strains of D. melanogaster infected with different Wolbachia strains (wMel, wMelCS and wMelPop) were all backcrossed for five generations to the males from the wMelCS-infected strain. For generating ectopic PCs, GR1-Gal4 (expressed in all FCs) (Goentoro et al., 2006; Gupta and Schupbach, 2003) females infected with wMelCS were crossed to UAS-eyaRNAi (Bloomington Drosophila Stock Center 28733) males.

The protein immunofluorescence and FISH protocol was adapted from Zimmerman et al. (2013). Flies were aged for 7 days, dissected, and fixed in a 4% paraformaldehyde solution. The ovaries were first immunostained for the PCs using a mouse anti-FasIII antibody [Developmental Studies Hybridoma Bank (DSHB), 1:2000] (Patel et al., 1987) or mouse anti-Eya antibody (DSHB, 1:300) as described previously (Frydman et al., 2006; Frydman and Spradling, 2001). This was followed by FISH against Wolbachia 16S ribosomal RNA (rRNA). Specific oligonucleotide probes were designed against the 16S rRNA of Wolbachia (Integrated DNA Technologies). Two Wolbachia probes labeled with Cy3 at the 5′ end were used: Wpan16S887 5′-ATCTTGCGACCGTAGTCC-3′ and Wpan16S450 5′-CTTCTGTGAGTACCGTCATTATC-3′. Hybridization was performed at 37°C in 50% formamide, 5× saline-sodium citrate (sal-SC), 250 mg/l salmon sperm DNA, 0.5× Denhardt's solution, 20 mM Tris-HCl and 0.1% sodium dodecyl sulfate (SDS). After a 30 min preincubation period, tissue was incubated in 100 ng of each probe for 3 h. Tissue was then washed twice for 15 min at 55°C in a 1× sal-SC wash with 0.1% SDS and 20 mM Tris-HCl and then twice for 15 min in a 0.5× sal-SC wash with 0.1% SDS and 20 mM Tris-HCl. Nuclei were counterstained with Hoechst (1 μg/ml, Molecular Probes), added to the second 0.5× sal-SC wash at a concentration of 10 μg/ml. Tissue was then washed in PBS and mounted in Prolong Gold antifade solution and imaged as described below. Negative control of the FISH was performed on Wolbachia uninfected ovaries to show that there is no background staining by the oligonucleotide probes for 16S RNA (Fig. S8).

Culex ovaries were dissected and fixed in a 4% paraformaldehyde solution. The ovaries were first immunostained for the PCs using a rat anti-DN-Cad antibody (DSHB, 1:200) to label the putative PCs and mouse anti-Hsp60 (Sigma-Aldrich, H3524, 1:100) as described previously (Frydman et al., 2006; Frydman and Spradling, 2001). Nuclei were counterstained with Hoechst (Molecular Probes, 10 μg/ml).

All images were acquired on a FV1000 confocal microscope (Olympus) using the 60× objective (NA 1.42). Image acquisition parameters were kept constant for all images within each Wolbachia strain.

For representative images, four to eight z-stacks, 1 µm apart, encompassing the PCs were combined, and images were processed in Adobe Photoshop CS6 to prevent neighboring egg chambers not relevant to the image being shown. This was performed for clarity.

We thank Dr Kim McCall, Dr Trudi Schupbach, Dr Luiz Teixeira, Dr Bill Sullivan, Dr Virginie Orgogozo, Dr Ruth Lehmann and Bloomington Drosophila Stock Center at Indiana University for reagents and fly stocks; Dr Celeste Berg for help with troubleshooting of the IF/FISH double labeling protocol; and members of the H. Frydman laboratory, Dr Kim McCall and Dr Rachel Cox for assistance and suggestions in the realization of this work. The FasIII antibody (7G10), developed by N. H. Patel, P. M. Snow and C. S. Goodman, and Eya antibody (10H6), developed by S. Benzer and N. M. Bonini, were obtained from the Developmental Studies Hybridoma Bank, maintained at The University of Iowa, Iowa City, IA, USA.