Infection with wMel rescues female D. melanogaster fertility in mei-P26 deficient flies

Infection with wMel significantly rescued fertility defects induced by the loss of mei-P26 function relative to the OreR wild-type strain, as measured by offspring produced per female D. melanogaster per day (Figs 1A–1E and S3; p< = 2.2e-16 to 2.9e-2, Wilcoxon rank sum test; see S2–S6 Figs). Offspring production requires successful maintenance of the GSC, differentiation of a GSC daughter cell into a fully developed and fertilized egg, embryogenesis, and hatching into a first-instar larva. Breaking offspring production down into egg lay and egg hatch components (Fig 1B and 1C) revealed that as allelic strength increased, from nos-driven RNAi to hypomorphic and null alleles, fertility impacts shifted from those that impacted differentiation/development (egg hatch) to those that also impacted egg production (egg lay).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. wMel infection rescues the loss of essential host fertility gene mei-P26 in females (and males, see S4E–S4G Fig).

(A–C) Beeswarm boxplots of (A) overall offspring production, broken into (B) eggs laid and (C) eggs hatched, from single female RNAi, hypomorphic, and null mei-P26 knockdown female flies crossed to single OreR males of the same infection state and age (3–7 days old). The null allele mei-P26[mfs1] reduced egg lay rates in both mutant heterozygotes and homozygotes below an average of 20 for both infected and uninfected females, necessitating an analysis of females with and without offspring (see D). Wilcoxon rank sum test *p< = 0.05, **< = 0.01, ***< = 0.001, ****< = 1e-4. Fisher’s exact test [**]p< = 0.01, [****]< = 1e-4. (D–E) (D) Barplots of the proportion of female samples of the indicated genotypes with offspring and without offspring, infected and uninfected with Wolbachia. (E) Table listing the percentage of wild-type fecundity demonstrated by wMel-infected (left) or uninfected (right) mei-P26 mutants in order of increasing severity, from top to bottom (see also S4–S6 Tables). Cell values from low to high proportions are colored from light to dark blue, respectively. (F, G) Knockdown mei-P26 D. melanogaster fecundity versus female age, fit with a local polynomial regression (dark gray bounds = 95% confidence intervals). Infection with wMel increases offspring production over the first 25 days of the (F) mei-P26 RNAi and (G) mei-P26[1] hypomorphic female lifespan. (H–O) Confocal mean projections of D. melanogaster ovarioles and germaria (3–7 days old). (H–I’) Intracellular wMel FtsZ localizes to germline (yellow, Vas+, inner dotted outline) and somatic (Vas-, space between inner dotted and outer dashed outlines) cells at high titers, with low background in uninfected flies. Bacteria are continuously located in the mei-P26[1] germline, starting in the GSC (arrowhead near Hts-bound spectrosome). Somatic cell regions are indicated with empty arrowheads. (J–O) Cytology reveals that wMel-infection rescues (J, K) mei-P26 RNAi, (L, M) mei-P26[1] (see also S5 Fig), and (N, O) mei-P26[1/mfs1]-induced oogenesis defects. Fluorescence channels were sampled serially and overlaid as indicated on each set of images as follows: cyan = anti-Hu Li Tai Shao (Hts) ring canal isoform (-RC) or fusome isoform (-FUS), yellow = anti-Vasa (Vas), blue = wMel anti-Filamenting temperature-sensitive mutant Z (FtsZ), and red = PI DNA staining. Scale bars: H–I’ = 10 μm, L–O = 50 μm, J, K = 100 μm. (P) Diagram of a wild-type D. melanogaster germarium, highlighting key events and cell types by region as in [42]. Somatic cells in beige, GSC-derived cells are brightly colored gradients leading to differentiated cyst cells in yellow, with specified oocytes in blue. The black structure that originates in the GSC is the spectrosome, which becomes a branching fusome as the CB is formed, moves away from the niche, and divides into CCs. The data underlying this figure can be found in S3 Table and on Dryad at doi.org/10.7291/D1DT2C. CB, cystoblast; CC, cystocyte; GSC, germline stem cell; OreR, Oregon R; PI, propidium iodide.

https://doi.org/10.1371/journal.pbio.3002335.g001

Infection with wMel mediated a significant degree of fertility rescue for all allelic strengths and nos:GAL4>UAS-driven RNAi knockdowns, suggesting a robust bacterial rescue mechanism that compensates for loss of protein dosage and function. Rescue was more than complete for nos-driven mei-P26 RNAi depletion (from 26% deficient offspring production in uninfected flies): wMel-infected mei-P26 RNAi females produced 44% more offspring than infected OreR wild-type females (37 versus 25 larvae/female/day, p< = 1.15E-02 Wilcoxon rank sum test, S4 Table and Fig 1A and 1E) due to a higher rate of egg production (p< = 6.99E-05 Wilcoxon rank sum test, S5 Table and Fig 1B and 1E), opposed to a higher hatch rate (Fig 1C and 1E and S6 Table). This may suggest a synergistic function between wMel and low-dose mei-P26 in the GSC. In uninfected and wMel-infected mei-P26[1] hypomorphic females, offspring production decreased 94% and 73%, respectively, compared to OreR wild-type of the matching infection status (p< = 2.26E-14 and 3.45E-12 Wilcoxon rank sum test, respectively, S4 Table and Fig 1A and 1E). Comparing uninfected and wMel-infected female mei-P26[1] fecundity revealed that wMel infection increases the number of offspring produced per day (1.6 versus 7 larvae/female/day, p< = 9.37E-05 Wilcoxon rank sum test, S4 Table and Fig 1A) through increasing both the rate of egg lay (p< = 2.82E-02 Wilcoxon rank sum test, S5 Table and Fig 1B) and egg hatch (p< = 2.55E-03 Wilcoxon rank sum test, S6 Table and Fig 1C). In uninfected and wMel-infected mei-P26[1]/mei-P26[mfs1] trans-heterozygous females, offspring production decreased 99% and 82%, respectively, relative to OreR wild-type (p< = 3.40E-04 and 5.01E-10 Wilcoxon rank sum test, S4 Table and Fig 1A and 1D). Comparing uninfected and wMel-infected trans-heterozygous flies showed that infection elevates offspring production (0.2 versus 4.6 larvae/female/day, p< = 1.03E-05 Fisher’s exact test, S4 Table and Fig 1A) through increasing the rate females lay eggs (p< = 4.71E-05 Wilcoxon rank sum test; Fig 1B) and the rate those eggs hatch (p< = 4.20E-03 Wilcoxon rank sum test; S6 Table and Fig 1C). In females homozygous for the most severe allele, mei-P26[mfs1], offspring production decreased 100% and 97% in uninfected and wMel-infected flies relative to wild-type, respectively (p< = 5.14E-06 Fisher’s exact test, S4 Table and Fig 1A). Offspring production was marginally rescued in mei-P25[mfs1] flies by wMel infection (0 (uninfected) versus 0.76 (infected) larvae/female/day, p< = 5.14E-06 Fisher’s exact test, S4 Table, Fig 1A), in part due to infected flies laying eggs at a higher rate than uninfected flies (p< = 3.04E-05 Wilcoxon rank sum test, Fig 1B). However, no female laid 20 or more eggs in one day, precluding an estimation hatch rate rescue (Fig 1C). See S2–S6 Tables and S3 Fig for a full description of the fecundity assays and data.

wMel-mediated mei-P26 rescue is robust and sufficient to rescue this gene in a stable stock. Infection with wMel elevates the number of offspring produced per day across the fly lifespan in mei-P26[1] hypomorphic and nos-driven RNAi-knockdown females (p< = 1.5e-11 to 3.4e-2 Kolmogorov–Smirnov test, S7 Table and Fig 1F and 1G). However, the underlying number of eggs laid per day and the hatch rate do vary considerably with age for both infection states and genotypes (S4A–S4D Fig). The wMel rescue mechanism is not specific to females, as the weaker impacts of mei-P26 loss on male fertility are mitigated by wMel-infection (S4E–S4G Fig; p< = 6.1e-6 to 2.9e-2 Wilcoxon rank sum test, S3–S5 Tables). Thus, we were able to establish a homozygous stock of mei-P26[1] flies infected with wMel Wolbachia. In contrast, the uninfected stock only lasted a few generations without balancer chromosomes (S4H and S4I Fig). Given the severe, yet significantly rescuable nature of the mei-P26[1] allele (Figs 1A–1C, S2B and S2D), we proceeded with this genotype for many of our subsequent assays for specific immunocytological rescue phenotypes.

Female germline morphology is rescued by wMel infection in mei-P26-deficient flies

In the D. melanogaster germarium, wMel is continuously present at high titers in both germline and somatic cells (Fig 1H–1H’, compared to control in Fig 1I–1I’), consistent with previous reports [26,27,43]. The bacteria localize more strongly to the germline-derived cells than the somatic cells (co-localization of FtsZ and Vas in Fig 1H–1H’). Intracellular wMel can be identified in the GSC, the cystoblast (CB), the cystocytes (CCs), and the developed cyst. This positioning puts wMel in all of the critical cell types and stages that mei-P26 is active, enabling the bacterium to compensate for mei-P26’s developmental functions in GSC maintenance and differentiation.

Comparing the cytology of mei-P26 knockdown oocytes infected and uninfected with Wolbachia confirmed that wMel-infected ovarioles exhibit far fewer developmental defects than uninfected, more closely resembling OreR wild-type (Figs 1J–1O and S5 versus Fig 1P; Vasa (Vas) = germline [44], Hu Li Tai Shao (Hts) = cytoskeletal spectrosome/fusome [45], propidium iodide (PI) = DNA). Specifically, wMel-infected mei-P26-knockdown ovarioles exhibited more normally formed cysts, consisting of somatically derived follicle cells surrounding germline-derived nurse cells and an oocyte, than uninfected ovarioles. Aberrant phenotypes included cysts lacking the follicle cell exterior (Fig 1K and 1M), the germline-derived interior (Fig 1M and 1N), or a normal number of differentiated nurse cells (Fig 1O).

These comparative data for a range of mei-P26 alleles, RNAi-knockdowns, and host ages indicate that wMel rescues mei-P26’s developmental functions in early host oogenesis. In contrast, wMel does not rescue mei-P26’s role in meiosis. Segregation defects were not rescued by wMel infection, as indicated by elevated X-chromosome nondisjunction (NDJ) rates in both infected and uninfected mei-P26[1] homozygous hypomorphs (NDJ = 7.7% and 5.6% by experiment, respectively, S6A and S6B Fig). In the following sections, we analyze wMel’s ability to rescue mei-P26 function at each of the critical time points in early oogenesis.

Host GSCs are maintained at higher rates in wMel-infected than uninfected mei-P26-deficient flies

GSCs were quantified by immunofluorescence staining of the essential GSC markers pMad and Hts in “young” 4- to 7-day-old (Fig 2) and “aged” 10- to 13-day-old females (S7A–S7C Fig). Positive pMad staining indicates the cell is responding to quiescent signals from the surrounding somatic GSC niche cells. Positive Hts staining requires localization to a single punctate spectrosome, opposed to a branching fusome [46].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. wMel infection rescues mei-P26-induced defects in female GSC maintenance.

(A–E) Infection with wMel confers higher numbers of GSCs per germarium in mei-P26[1] germaria than in uninfected flies. (A–D) Confocal mean projections of D. melanogaster germaria stained with antibodies against Hts and pMad. (E) Violin plots of the number of GSCs per germarium. As fully functional GSCs express pMad and have Hts-labeled spectrosomes, each was weighted by half, allowing for partial scores. Wilcoxon rank sum test * = p<0.05, ** = 0.01, *** = 0.001, **** = 1e-4. (F–J) Mitotic GSCs detected by pHH3 expression revealed that wMel restored mitosis in hypomorphic mei-P26 GSCs relative to uninfected GSCs. (F–I) Confocal mean projections of D. melanogaster germaria stained with antibodies against Hts and pHH3. (J) Stacked bar chart of the proportion of GSCs found expressing pHH3 in female germaria. Fisher’s exact test * = p < 0.05. Fluorescence channels were sampled serially and overlaid as indicated on each set of figs as follows: cyan = anti-Hts, yellow = anti-Vas, red = PI, green = anti-pMad in (A–D) and anti-pHH3 in (F–I). Scale bars A–D = 10 μm and F–I = 25 μm. The data underlying this figure can be found on Dryad at doi.org/10.7291/D1DT2C. GSC, germline stem cell; pHH3, phospho-Histone H3.

https://doi.org/10.1371/journal.pbio.3002335.g002

Staining revealed an increase in the average number of GSCs per mei-P26[1] germarium with wMel infection, relative to uninfected germaria in young, 4- to 7-day-old flies (infected/hypomorph: 1.0 versus uninfected/hypomorph: 0.58, p< = 2.9e-4 Wilcoxon rank sum test, S8 Table), but GSC maintenance did not reach OreR wild-type rates (infected/WT: 2.3 and uninfected/WT: 1.9, p< = 1.2e-4 to 2.8e-4 Wilcoxon rank sum test, Fig 2A–2E and S8 Table). wMel-infected mei-P26[1] germaria also had more cells manifesting other GSC properties, such as a large cytoplasm and physical attachment to the cap cells of the somatic niche [46], than uninfected germaria. In contrast to the mei-P26[1] allele, RNAi knockdown of mei-P26 did not significantly affect the numbers of GSCs per germarium (infected/RNAi: 2.1 and uninfected/RNAi: 1.9, S7A–S7C Fig and S8 Table), suggesting that the modest reductions in mei-P26 dosage have a stronger impact on differentiation than GSC maintenance. OreR wild-type germaria did not exhibit different numbers of GSCs due to wMel infection (Fig 2C–2E and S8 Table).

The number of GSCs per germarium in wMel-infected mei-P26[1] females converges on OreR wild-type values in aged 10- to 13-day-old germaria (S7C Fig). As wild-type flies age, the number of GSCs per germarium declines naturally [47], even in wMel-infected flies (aged/infected/WT: 1.6 versus young/infected/WT: 2.3, p< = 0.019 Wilcoxon rank sum test, Figs 2C–2E and S7C and S8 Table). Both infected and uninfected mei-P26[1] females appear to retain some of their GSCs as they age (aged/infected/hypomorph: 1.5 and aged/uninfected/hypomorph: 0.94, p< = 0.013–0.014 Wilcoxon rank sum test, S7C Fig and S8 Table). While OreR wild-type infected and uninfected flies lose from 3% to 31% of their GSCs in the first 2 weeks, infected and uninfected mei-P26[1] flies increase their GSC abundance by 43% and 63% in this time (S7C Fig), potentially through recruiting differentiated GSCs [48]. Taking both age-dependent trajectories in GSC retention into account, by the time wMel-infected mei-P26[1] germaria are 10 to 13 days old, they contain the same average number of GSCs per germarium as wild-type flies and significantly more GSCs than uninfected mei-P26[1] flies (p< = 0.015 Wilcoxon rank sum test, S7C Fig and S8 Table). Therefore, Wolbachia enhances stem cell maintenance in both young and old mei-P26[1] females.

Infected mei-P26[1] GSCs are functional, as indicated by their ability to divide by mitosis (Fig 2F–2J and S9 Table). We identified nuclei in mitosis by positive anti-phospho-histone H3 serine 10 (pHH3) staining, a specific marker of Cdk1 activation and mitotic entry [49]. Significantly fewer GSCs were in mitosis in uninfected mei-P26[1] germaria than infected mei-P26[1] and uninfected OreR wild-type germaria (0.09% versus 6%, p = 4.7e-2 and 4.6e-2 Fisher’s exact test, Fig 2F–2J and S9 Table). There was no difference in the frequency of GSCs in mitosis between uninfected and infected wild-type germaria (5% to 6%, Fig 2H–2J and S9 Table). Mitotic cystoblasts and cystocytes were only significantly enriched in infected mei-P26[1] germaria compared to infected wild-type germaria (p< = 1.3e-2 Wilcoxon rank sum test, S7D Fig and S10 Table), despite over-replication during transit-amplifying (TA) mitosis being a common phenotype in uninfected mei-P26[1] germaria (discussed below and [50]).

Regulation of host Sxl expression is rescued in wMel-infected germaria

Proper Sxl expression is required for GSC maintenance and differentiation [51]. High levels of Sxl in the GSC maintains stem cell quiescence [52]. When the GSC divides and the CB moves away from the niche, Sxl cooperates with Bam and Mei-P26 to bind to the nos 3′ untranslated region (UTR) and down-regulate Nos protein levels to promote differentiation [50,53].

We quantified Sxl localization patterns in whole ovaries stained with anti-Sxl antibodies by confocal microscopy of whole oocytes, revealing that Sxl dysregulation due to mei-P26 loss is mitigated by wMel infection (Fig 3). Less Sxl is expressed in uninfected mei-P26[1] germarium region 1 and more Sxl is expressed across regions 2a and 2b, relative to OreR wild-type (p< = 5.7e-10 to 5.3e-5 Wilcoxon rank sum test, Fig 3A–3B’ versus 3F–3H’ and 3I, S7E Fig, and S11 Table). Infection with wMel increased Sxl expression in germarium region 1, had no impact on region 2a, and suppressed expression in region 2b relative to uninfected mei-P26[1], replicating an expression pattern similar to that seen in wild-type germaria (p< = 5.7e-10 to 8.4e-3 Wilcoxon rank sum test, Figs 3A–3B’ versus 3C–3E’ and 3I and S7E and S11 Table). Interestingly, wMel infection may impact Sxl expression in OreR germaria, as the protein is significantly elevated in region 2a of wMel-infected relative to uninfected germaria (Fig 3I). Thus, Sxl dysregulation due to mei-P26 loss can be partially rescued by the presence of wMel, suggesting that wMel’s mechanism for mei-P26 rescue may underlie how Nos regulation in wMel-infected Sxl mutants [31,34].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Infection with wMel mitigates the consequences of dysfunctional mei-P26 on Sxl expression in the germarium.

(A–H’) Confocal mean projections of D. melanogaster germaria stained with antibodies against Sxl and pMad. Three phenotypes for each mei-P26[1] infection state are shown, (left) one as similar to wild-type as possible, (middle) one representative of GSC loss, pMad-negative, and (right) one representative of a developmentally altered phenotype retaining pMad-positive GSCs. We sampled the fluorescence channels serially and overlaid them as indicated on each set of images: cyan = anti-Sxl, red = PI, green = anti-pMad. Scale bars = 25 μm. (I) Bar-scatter plots of relative Sxl fluorescence expression levels across the germarium, by region (yellow outlines over Sxl-channel images in A’–H’). Colors as labeled in Figs 1A–1D and 2E, from left to right: dark gray = wMel-infected OreR, light gray = uninfected OreR, dark pink = wMel-infected mei-P26[1], light pink = uninfected mei-P26[1]. Wilcoxon rank sum test * = p<0.05, ** = 0.01, *** = 0.001, **** = 1e-4. (J) Model of wMel Sxl expression rescue in mei-P26[1] germaria. Green dashed line represents wMel’s up-regulation of pMad expression in the GSC (Fig 2A–2E). Purple dashed line represents wMel’s up-regulation of Sxl in the GSC/CB and down-regulation of Sxl by germarium region 2b. Arrows with + symbols indicate wMel’s action on gene expression. The data underlying this figure can be found on Dryad at doi.org/10.7291/D1DT2C. CB, cystoblast; GSC, germline stem cell; OreR, Oregon R; PI, propidium iodide.

https://doi.org/10.1371/journal.pbio.3002335.g003

Bam expression is partially restored in wMel-infected mei-P26 mutant germaria

In wild-type female flies, Bam expression begins immediately after the cystoblast daughter cell moves away from its undifferentiated sister, which remains bound to the GSC niche ([50,54] and illustrated in Fig 1P). Bam binds to the fusome and stabilizes CyclinA to promote TA mitosis, producing 16 cyst cells from a single CB [55]. Immediately following these 4 mitoses, Bam expression is down-regulated in wild-type ovaries to limit germline cysts to 16 germline-derived cells.

Through anti-Bam immunofluorescence confocal imaging and quantification, we determined that wMel-infection mitigates aspects of Bam dysregulation in mei-P26[1] germaria (Fig 4). Loss of Mei-P26 deregulates and extends Bam expression past germarium region 1 (Fig 4A–4B” versus 4C–4F” and 4G; S7F Fig and S12 Table), producing excess nurse-like cells [50]. Infection with wMel enables Bam up-regulation at the CB-to-16-cell stage in mei-P26[1] germaria relative to uninfected germaria (p< = 0.016, Fig 4G). Bam may also be down-regulated in stage 2a and 2b wMel-infected mei-P26[1] germaria because uninfected germaria express significantly more Bam than wild-type (p< = 0.019 Wilcoxon rank sum test, Fig 4G and S12 Table), whereas wMel-infected germaria do not (Fig 4G and S12 Table). Unfortunately, variance in mei-P26[1] Bam fluorescence intensity among samples is too large to resolve whether uninfected and infected germaria differ in Bam expression after the 16-cell stage with this dataset.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Infection with wMel mitigates the consequences of dysfunctional mei-P26 on Bam expression in the germarium.

(A–F”) Confocal mean projections of D. melanogaster germaria stained with antibodies against Bam and pMad. Two phenotypes for each mei-P26[1] infection state are shown, (left) one as similar to wild-type as possible and (right) one representative of developmentally altered phenotypes. (G) Bar-scatter plots of relative Bam fluorescence expression levels across the germarium, by region (yellow outlines over Bam-channel images in A’–F’). We subdivided region 1 into GSC and CB/cystocyte subsections because Bam expression ramps up in the CB. (H) Bar-scatter plot of the ratio of relative Bam to pMad fluorescence. As pMad and Bam expression are mutually exclusive in the GSC and CB/cystocytes, respectively, OreR wild-type values are near zero. Values significantly higher than this reflect dysregulation of the stem cell state. (I) Model of wMel rescue of Bam expression in mei-P26[1] germaria. Colors as labeled in Figs 1A–1C, 2E and 3I, from left to right: dark gray = wMel-infected OreR, light gray = uninfected OreR, dark pink = wMel-infected mei-P26[1], light pink = uninfected mei-P26[1]. Wilcoxon rank sum test * = p<0.05, ** = 0.01, **** = 1e-4. Scale bars = 25 μm. The data underlying this figure can be found on Dryad at doi.org/10.7291/D1DT2C. CB, cystoblast; GSC, germline stem cell; OreR, Oregon R.

https://doi.org/10.1371/journal.pbio.3002335.g004

Infection with wMel alters Bam expression in the GSC and CB to achieve an expression profile similar to and more extreme than wild-type germaria, respectively (Fig 4H and 4I). Both infected and uninfected mei-P26[1] germaria exhibit lower Bam expression than wild-type in the GSC niche (Fig 4G). This contrasts with expectations from the literature: bone morphogenic protein (BMP) signaling from the somatic niche induces pMad expression in the GSC, which directly represses Bam transcription, preventing differentiation [56]. Loss of mei-P26 function in the GSC should derepress Brat, resulting in pMad repression and inappropriate Bam expression [38]. However, we see lower Bam expression in both wMel-infected and uninfected mei-P26[1] GSCs relative to OreR wild-type (p< = 1.3e-6 to 1.4e-3 Wilcoxon rank sum test, Fig 4G and S13 Table). The elevated relative Bam/pMad expression ratio in uninfected mei-P26[1] GSCs (p< = 2.5e-4 to 4.0e-4 Wilcoxon rank sum test, Fig 4H and 4I and S13 Table) may explain this departure from expectations. Although GSC Bam expression is lower than in wild-type, Bam expression is clearly dysregulated and up-regulated relative to pMad expression, as expected for germline cells that have lost their stem cell identity [57,58]. Normalizing against pMad expression also reinforces the finding that wMel elevates Bam expression even higher than wild-type levels in mei-P26[1] hypomorphs at the CB-to-16-cell stage (p< = 1.2e-6 to 2.0e-2 Wilcoxon rank sum test, Fig 4H and 4I and S13 Table).

wMel rescues normal germline cyst and oocyte development impaired by mei-P26 knockdown

Mei-P26-deficient flies over-proliferate nurse cells and partially differentiate GSCs to produce tumorous germline cysts ([38,39] and Fig 5A–5D versus 5E–5H, Fig 5I and S14 Table). In RNAi knockdown ovaries, the formation of tumorous germline cysts is significantly mitigated by wMel infection (35.3% versus 16.3% of cysts with more than 15 nurse cells, respectively; p = 4.5e-3 Fisher’s exact test, Fig 5E and 5F versus 5G and 5H and S14 Table). In contrast, germline cyst tumors found in mei-P26[1] allele germaria were not rescued by wMel infection, despite the bacterium’s ability to rescue the rate at which these eggs hatch into progeny (Fig 5I versus Fig 1A and 1C), but consistent with their dysregulation of Bam expression in germarium regions 2a and 2b (Fig 4G). The stronger nature of the mei-P26[1] allele relative to the nos-driven RNAi knockdown likely underlies the difference in the number of tumorous cysts. Lack of tumor rescue indicates that either tumorous cysts produce normal eggs at some rate or wMel infection rescues tumors later in oogenesis, perhaps through some somatic mechanism [59–62].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. wMel infection rescues cyst tumors and restores oocyte specification in mei-P26 mutants.

(A–I) Loss and overexpression of Mei-P26 induces the formation of tumorous cysts, which contain more than 15 nurse cells. wMel infection mitigates nos>mei-P26 RNAi-knockdown and exacerbates nos>mei-P26 OE phenotypes. (A–H) Confocal max projections of approximately 25 μm-thick sections of D. melanogaster oocyte cysts stained with PI. (I) Stacked bar chart of the proportion of germline cysts containing an abnormal number of nurse cells (less than or greater than 15 nurse cells). (J–R) Hypomorphic mei-P26[1] cysts infected with wMel significantly restored Orb translational regulation relative to uninfected cysts. Oocyte specification occurs in regions 2b-3 of the germarium (see diagram in Fig 1P), when Orb expression becomes restricted to the nascent oocyte. (J–Q’) Confocal mean projections of D. melanogaster germaria stained with antibodies against Orb and Vas. (R) Stacked bar chart of the proportion of germline cysts containing oocyte-specific Orb staining. Sample sizes are written on the bar charts. N = no, M = maybe, and Y = yes. Fisher’s exact test ** = p<1e-2, **** = p<1e-4. Scale bars = 50 μm. The data underlying this figure can be found on Dryad at doi.org/10.7291/D1DT2C. OE, overexpression; PI, propidium iodide.

https://doi.org/10.1371/journal.pbio.3002335.g005

Overexpression of mei-P26 produces tumors in both uninfected and infected ovaries; however, infected ovaries exhibit significantly more tumors than uninfected ovaries (p = 1.6e-3 Fisher’s exact test, Fig 5I and S14 Table). Finding that an excess of mei-P26 recapitulates the tumorous phenotype induced by a lack of mei-P26 indicates that this gene exerts a concentration-dependent dominant negative effect on its own function. Given that infection with wMel makes this dominant negative phenotype more severe, we hypothesize that wMel may make a factor that acts to enhance Mei-P26 function, potentially through a direct interaction or redundant function.

Infection with wMel increases the rate of oocyte-specific Orb localization in mei-P26[1] germline cysts (Figs 5J–5R and S7G and S15 Table), suggesting that wMel rescues functional cyst formation. Oocyte specification in region 2a of the germarium (see diagram in Fig 1P) is inhibited by the loss of mei-P26 through the derepression of orb translation [38]. Mei-P26 binds to the orb 3′ UTR and prevents its translation. Following cyst patterning and development, one of the 16 germline-derived cells is designated as the oocyte through specific oo18 RNA-binding protein (Orb) expression [63]. In region 2a, the germline cyst cell fated to become the oocyte activates orb translation [38]. Loss of mei-P26 results in orb derepression and nonspecific expression [38]. Infection with wMel restores oocyte-specific Orb expression in mei-P26[1] cysts relative to uninfected cysts (p = 5.9e-9 Fisher’s exact test, Fig 5L–5N’ versus 5O–5Q’ and S15 Table). We observed oocyte-specific Orb staining 93% to 94% of the time in wild-type cysts, whereas mei-P26[1] reduced this frequency to 25%, and wMel infection recovered mei-P26[1] cysts to 61% (counts in S15 Table). Oocyte-specific expression of Orb in wMel-infected cysts is robust, even in those that are developmentally aberrant (e.g., the double oocyte in Fig 5M–5M’). These results suggest that wMel rescues oocyte-specific Orb translation enhancing or duplicating Mei-P26’s function in orb translational repression.

OreR wild-type host fertility is beneficially impacted by wMel infection

Consistent with wMel’s ability to rescue the partial to complete loss of essential germline maintenance genes encompassing a range of functions, we discovered that this intracellular symbiont can reinforce fertility in D. melanogaster stocks displaying full wild-type fertility (Figs 1C and S8A–S8C and S4–S6 Tables). We analyzed egg lay, egg hatch, and overall offspring production rates independently to detect wMel effects on multiple components of fertility. Overall offspring production per female per day was elevated in wMel-infected nos:Gal4/CyO females relative to uninfected females by 49% (27 (uninfected) versus 40 (infected) offspring/female/day, p< = 2.2e-3 Wilcoxon rank sum test, S8A Fig and S4 Table). Infection with wMel increased the number of eggs laid per female per day for the nos:Gal4/CyO balancer stock by 49% (42 versus 28 eggs/female/day, p< = 2.2e-3 Wilcoxon rank sum test, S8B Fig and S5 Table), but did not have any detectable effect on egg lay rates in other genotypes with wild-type fertility. Following embryogenesis, infection with wMel elevated the rate that OreR and nos:Gal4/CyO wild-type eggs hatch into L1 stage larvae by 5.2% and 2.5%, respectively (88% and 91% (infected) versus 83% and 89% (uninfected) of eggs hatched, p< = 1e-4 to 2.1e-3 Wilcoxon rank sum test, Figs 1C and S8C and S6 Table). These results suggest that beneficial impacts of wMel infection are evident, yet variable among D. melanogaster stocks with wild-type fertility.

We measured OreR fertility across its lifetime (approximately 50 days) and found that wMel may produce higher lifetime fecundity relative to uninfected flies (Fig 6A–6C and S7 Table). OreR females were aged as indicated along the x-axis of Fig 6A–6C and mated to 3 to 7 days old males of matching infection status. wMel-infection maintains elevated OreR egg hatch rates as flies age (p< = 7.1e-9 Kolmogorov–Smirnov test, Fig 6C and S7 Table), while maintaining similar levels of egg production as uninfected flies (Fig 6B and S7 Table). This culminates in age ranges in which infected flies may produce an excess of offspring relative to uninfected flies (see gap between the regression confidence intervals when flies were approximately 20 to 30 days old, Fig 6A). The efficiency gains imparted on hosts by infection-induced elevated hatch rates could potentially accumulate for higher lifetime fitnesses, as resources are more efficiently converted to offspring.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The wMel strain of Wolbachia reinforces OreR wild-type fertility.

(A–C) D. melanogaster OreR fecundity versus female age, fit with a local polynomial regression (dark gray bounds = 95% confidence intervals). Infection with wMel may elevate (A) offspring and (B) egg offspring production at time points during the host lifespan through continuously elevating (C) hatch rate. (D–F’) Confocal mean projections of D. melanogaster (D–D’) infected and (E–E’) uninfected germaria and (F–F’) an infected ovariole. Intracellular wMel FtsZ localizes to germline (Vas+, outlined in dashed yellow) and somatic (Vas-) cells at high titers in infected OreR wild-type flies, with low background in uninfected flies. Bacteria in the GSC (* near the Hts-bound spectrosome) are indicated with arrows. Fluorescence channels were sampled serially and overlaid as indicated on each set of images as follows: cyan = anti-Hts, yellow = anti-Vas, blue = anti-FtsZ, and red = PI. Scale bars F–G = 10 μm, H = 50 μm. The data underlying this figure can be found in S3 Table and on Dryad at doi.org/10.7291/D1DT2C. GSC, germline stem cell; OreR, Oregon R; PI, propidium iodide.

https://doi.org/10.1371/journal.pbio.3002335.g006

Immunostaining of anti-FtsZ confirms that wMel exhibits high titers in the OreR wild-type germarium and concentrates in the oocyte in early cysts (Fig 6D–6F’). The intracellular bacteria are continuously located in the germline, starting in the GSC and proceeding through the end of embryogenesis and fertilization. Thus, wMel are located in the right place at the right time to reinforce female host fertility through interactions with the GSC and oocyte cyst, as well as to correct for perturbations caused by male CI-related alterations [64].

The wMel strain of Wolbachia in its native D. melanogaster host exhibits weak CI that decreases with age (S9A–S9D Fig), suggesting low maintenance of CI mechanisms [8]. Mating wMel-infected OreR males to infected (a “rescue cross”) and uninfected OreR virgin females (a “CI cross”) revealed that wMel reduces uninfected egg hatch by 26% when males are 0 to 1 day old (p< = 4.6e-4 Wilcoxon rank sum test, S9A Fig and S6 Table), but this moderate effect is eliminated by the time males are 5 days old, on average (S9C Fig and S6 Table). Only young males significantly impact offspring production (p< = 1.7E-2 Wilcoxon rank sum test, S9B and S9D Fig and S4 Table). In contrast, the Wolbachia wRi strain that naturally infects Drosophila simulans (the Riv84 line) and induces strong CI [10,65], reduces uninfected egg hatch by 95% when males are 0 to 1 day old (p< = 5.6e-8 Wilcoxon rank sum test, S9E Fig and S6 Table) and only loses some of this efficacy by 5 days (75% hatch reduction, p< = 3.6e-4, Wilcoxon rank sum test; S9G Fig, S6 Table). This significantly impacts overall offspring production (p< = 6.5E-7 and 5.0E-2 Wilcoxon rank sum test, S9F and S9H Fig and S4 Table). Paternal grandmothers of CI offspring were 3 to 7 days old (i.e., relatively young), contributing to weak CI in the D. melanogaster crosses [17]. Overall, these results suggest that wMel’s beneficial reproductive manipulations may affect its fitness more than its negative reproductive manipulations in nature.

wMel-mediated mei-P26[1] rescue may be enabled through rescuing and perturbing host transcription

Exploring wMel’s effect on host gene expression in OreR and mei-P26[1] ovaries with dual-eukaryotic and bacterial RNA sequencing (dual-RNAseq) revealed that infection significantly alters host gene expression in a variety of ways (Fig 7A–7C and S16–S19 Tables), which indicate a few potential mechanisms for wMel-mediated fertility rescue. Of the 35,344 transcripts in the D. melanogaster genome, 10,720 were expressed at high enough levels in at least 5 of 6 samples of each genotype+infection group to be included in the analysis. Of the 1,286 genes in the wMel genome, 663 were expressed in D. melanogaster ovaries (Fig 7D and S20 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Differential expression analysis of wMel and D. melanogaster genes from host ovaries reveal that mei-P26 knockdown and infection influence global expression patterns.

(A–D) Volcano plots of log2-fold change in gene expression versus log10-FDR adjusted p-value (padj) for the test performed in each panel. (E–K) Normalized transcript count plots for (E) the 3 top genotype-associated (~G) Wald Test hits by padj, and (F–K) D. melanogaster genes significantly associated (padj< = 1e-3) with infection (~I) or jointly, genotype and infection (~G*I), organized by the differential expression pattern. See S10–S15 Figs for other significant and insignificant count plots. In all barplots, dark gray = wMel-infected OreR, light gray = uninfected OreR, dark pink = wMel-infected mei-P26[1], light pink = uninfected mei-P26[1]. The data underlying this figure can be found at NCBI, under BioProject number PRJNA1007602. OreR, Oregon R.

https://doi.org/10.1371/journal.pbio.3002335.g007

Both uninfected and wMel-infected hypomorphic mei-P26[1] ovaries exhibited a 5-fold depletion in mei-P26 expression relative to OreR ovaries (S2B and S2D Fig) and significant dysregulation of 1,044 other genes due to the mei-P26[1] allele after FDR correction (Wald test FDR-adjusted p-value (padj)< = 0.05, Figs 7A, 7E and S10A and S17 Table). The changes in gene expression due to the loss and dysfunction of mei-P26 far outnumber and outpace the changes in gene expression due to infection, in either D. melanogaster genotype, and highlight how important mei-P26 is as a global regulator of gene expression. This number is well in excess of the 366 genes thought to be involved in GSC self-renewal [66], underscoring how many processes mei-P26 is involved in, from germline differentiation [50] to somatic development [61]. Across biological process, cell component, and molecular function GO categories, loss of mei-P26 impacted processes involving chromatin, recombination, protein–protein interactions, and muscle cell differentiation (S10B–S10G Fig), supporting mei-P26’s known role in genetic repression, differentiation, and meiosis [38,40,50].

Infection with wMel significantly alters the expression of hundreds of D. melanogaster genes to either restore or compensate for mei-P26-regulated gene expression. We tested for genes significantly associated with infection state in DESeq2 under the full model: ~genotype + infection + infection*genotype, revealing 422 significantly differentially expressed genes (Wald test padj< = 0.05, Fig 7B and S18 Table). Testing for an effect of the interaction between infection and genotype yielded another 20 significantly differentially expressed genes (Wald test padj< = 0.05; Fig 7C and S19 Table). The most significantly differentially expressed genes are featured in Fig 7F–7K and other significant genes are plotted in Figs 8, S11 and S12. Binning gene expression patterns revealed that there are at least 6 main rescue gene expression phenotypes induced by wMel-infection in mutant ovaries: full rescue, overshoot, undershoot, wild-type effect, inverse, and novel (Figs 7F–7K, 8A, 8C, S11A, S11B, S12A and S12B). In “full rescue” phenotypes (Figs 7F and S11A), mei-P26[1] wMel gene expression is indistinguishable from OreR levels and significantly different from uninfected mei-P26[1] levels. “Overshoot” phenotypes (Fig 7G) exhibit expression that is significantly in the opposite direction of the uninfected mei-P26[1] knockdown effect relative to OreR expression. Oppositely, “undershoot” phenotypes (Fig 7H) are partial corrections towards OreR expression in mei-P26[1] wMel ovaries. The one “wild-type effect” gene, an ER and Golgi cytoskeletal membrane anchor protein, demonstrated differential expression only between wMel-infected and uninfected OreR ovaries (Fig 7I). Five genes exhibit “inverse” differential expression phenotypes, such as a sodium symporter (Fig 7J) and the transcription factor Chronophage (S12A Fig), in which mei-P26[1] and OreR ovaries show opposite changes in differential expression relative to the uninfected genotype. In total, only these 6 “wild-type effect” and “inverse” genes were differentially expressed due to infection in OreR ovaries. Lastly, “novel” genes (Fig 7K) are those that only exhibit differential expression in wMel-infected mei-P26[1] ovaries, suggesting a novel rescue or compensation pathway. For example, suppression of retinal degeneration disease 1 upon overexpression 2 (sordd2) ubiquitin ligase mRNA expression is up-regulated, without rescue, in both infected and uninfected mei-P26[1] ovaries (Fig 7E). One of the novel rescue candidates, SKP1-related C (skpC), is a down-regulated component of SCF ubiquitin ligase (Fig 7K), which might compensate for the up-regulation of sordd2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Mei-P26 interactions predicted from the literature are supported by our ovary dual-RNAseq results.

(A, B) Predicted mei-P26 (A) protein and mRNA physical interactions and (B) genetic interactions. Symbols near gene names indicate significant DE interactions: * = significance by genotype; r = significance by infection, indicating a rescue phenotype; _ = significance by uncorrected p-values only. Red “X”s in A indicate genes that were not expressed in our dataset (CG6304 and mir-137). (C) Normalized transcript count plots for the genes designated with symbols in A and B. In all barplots, dark gray = wMel-infected OreR, light gray = uninfected OreR, dark pink = wMel-infected mei-P26[1], light pink = uninfected mei-P26[1]. (D) Cellular component GO terms for the 87 (~G*I) DE genes reveal an enrichment for early germline cytoskeletal structures and membrane proteins. Wald test association for listed padj and p-values are given in parentheses, as in Fig 7: genotype (~G), infection (~I), genotype*infection (~G*I). The data underlying this figure can be found at NCBI, under BioProject number PRJNA1007602. OreR, Oregon R.

https://doi.org/10.1371/journal.pbio.3002335.g008

GO terms associated with infection and the interaction between infection and genotype (Figs S11C–S11G, S12C–S12G and 8D) were consistent with wMel’s known ability to associate with host actin [67], microtubules, motor proteins [68–70], membrane-associated proteins [71], and chromatin [72]. Infection-associated genes were enriched in GO terms suggesting a cytoskeletal developmental function with chromatin interactions (S11C–S11G Fig). Across biological processes, cellular compartment, and molecular function categories, genes were enriched in terms suggesting interactions with actin, myosin, contractile fibers, and muscle cell differentiation. Chromatin interactions were indicated by associations with condensin complexes, centromeres, and mitotic chromosomes. Genotype-by-infection interaction-associated genes were also enriched in cytoskeletal and chromatin-interaction GO terms, in addition to plasma membrane/cell junction-associated terms such as cell junction and synapse structure maintenance, cell cortex, and lipid and calmodulin binding (S12C–S12G Fig).

Interestingly, 2 chorion proteins (Fig 7F), an ecdysteroid 22-kinase associated protein (S11A Fig), and an estrogen-related receptor (ERR) (S12A Fig), which are likely involved in chorion formation [73], were rescued by wMel. Considering that we have found mei-P26[1] embryos to have weaker corions then OreR wild-type embryos when treated with bleach, this finding suggests a novel chorion-associated function for mei-P26 and a novel rescue phenotype for wMel.

None of the 663 transcribed wMel genes were significantly differentially expressed between the OreR and mei-P26[1] host genotypes, after p-values were FDR-corrected (Figs 7D, S13 and S14 and S20 Table), suggesting host regulation by wMel’s constitutively expressed genes. On average, wMel transcriptomes were sequenced at 7.8 to 42.4× depth of coverage and D. melanogaster transcriptomes were sequenced at 25.9 to 35.9× (S16 Table). While these coverage ranges overlap, the 663 wMel genes detected only spanned −1.5-fold depletion to 1.3-fold up-regulation (S20 Table). In contrast, the 10,720 expressed D. melanogaster genes spanned an order of magnitude more differential expression, with −10.041-fold depletion to 9.866-fold up-regulation (S17–S19 Tables). Thus, wMel genes are weakly differentially transcribed compared to D. melanogaster genes, suggesting that wMel may alter host phenotypes through downstream impacts of constitutively expressed proteins. To explore the constitutive effects of wMel expression, we tested for GO pathway enrichment in the expressed transcriptome (S13 Fig). Pathways for DNA replication, protein expression, membrane transport, and heterotrophy (e.g., oxidative phosphorylation) were detected in abundance. While many of these genes are likely involved in normal bacterial cell maintenance and replication, a significant enrichment in pathways for oxidative phosphorylation, translation, transcription, and membrane functions (S13A Fig) suggest that some of these genes could be co-opted for host manipulation or processing host resources.

Among the 1,044 significantly differentially regulated D. melanogaster genes impacted by mei-P26 knockdown and infection were 6 genes (of 24) predicted in the literature (padj< = 0.05 Wald test and 9 of 24 by p-value< = 0.05; Figs 8 and S1 and S1 and S2 Tables). Loss of mei-P26 function caused dysregulation in oo18 RNA-binding protein (orb), gawky (gw), sex-lethal (sxl), benign gonial cell neoplasm (bgcn), tumorous testis (tut), and bantam (mir-ban) (padj< = 0.05 Wald test, Fig 8C and * in Fig 8A and 8B). Bam, Pumilio (pum), and Justice Seizure (jus) expression may be altered by mei-P26 knockdown, but additional samples are needed to resolve high infection variance (e.g., mir-ban) and low differential expression change (e.g., gw) (p-value< = 0.05 Wald test, Fig 8C and “*” in Fig 8A and 8B). Two of 6 genes exhibited evidence of wMel rescue with an “undershoot” rescue phenotype (“r” in Fig 8A and 8B): bgcn and tut by padjust (Fig 8C). If we consider genes with significant unadjusted p-values, gw may also exhibit evidence of “undershoot” rescue (Fig 8C, “r” in Fig 8A). Bantam or mir-ban expression suppression may be rescued as well, but high variance among wMel mei-P26[1] ovaries precludes DE rescue detection with this dataset (Fig 8C). In contrast, orb, sxl, bam, and nos transcript levels are not altered by mei-P26 knockdown or wMel infection, and therefore must be regulated posttranscriptionally. In total, these differential expression results reinforce and expand our understanding of how wMel rescues and reinforces host reproduction at transcriptional and posttranscriptional levels.

Candidate gene selection

Leveraging what is known about wMel’s abilities to rescue essential maintenance and differentiation genes [31,32] with empirical data on protein–protein and protein–mRNA interactions among these genes (esyN [82] networks in Figs S1A–S1D and 8; references in S1 Table) and data on interactions between host genes and wMel titers [37], we identified the essential germline translational regulator meiotic P26 (mei-P26) as a potential target of bacterial influence over GSC maintenance and differentiation pathways (S1A and S1B Fig). Loss of mei-P26 in uninfected flies produces mild to severe fertility defects through inhibiting GSC maintenance, meiosis, and the switch from germline cyst proliferation and differentiation [38–40].

Drosophila stocks and genetic crosses

Flies were maintained on white food prepared according to the Bloomington Drosophila Stock Center (BDSC) Cornmeal Food recipe (aka “white food,” see https://bdsc.indiana.edu/information/recipes/bloomfood.html). The wMel strain of Wolbachia was previously crossed into 2 D. melanogaster fly stocks, one carrying the markers and chromosomal balancers w[1]; Sp/Cyo; Sb/TM6B, Hu and the other carrying the germline double driver: P{GAL4-Nos.NGT}40; P{GAL4::VP16-Nos.UTR}MVD1. These infected double balanced and ovary driver stocks were used to cross wMel into the marked and balanced FM7cB;;;sv[spa-pol], null/hypomorphic mutants, and UAS RNAi TRiP lines to ensure that all wMel tested were of an identical genetic background. The D. melanogaster strains obtained from the BDSC at the University of Indiana were: Oregon-R-C (#5), w1118; P{UASp-mei-P26.N}2.1 (#25771), y[1] w[*] P{w[+mC] = lacW}mei-P26-1 mei-P26[1]/C(1)DX, y[1] f[1]/Dp(1;Y)y[+]; sv[spa-pol] (#25716), y[1] w[1] mei-P26[mfs1]; Dp(1;4)A17/sv[spa-pol] (#25919), y[1] sc[*] v[1] sev[21]; P{y[+t7.7] v[+t1.8] = TRiP.GL01124}attP40 (#36855). We obtained the UASp-mRFP/CyO (1-7M) line from Manabu Ote at the Jikei University School of Medicine. Drosophila simulans w[–] stocks infected with the Riv84 strain of wRi and cured with tetracycline were sourced from Sullivan Lab stocks [83,84]. All fly stocks and crosses were maintained at room temperature or 25°C on white food (BDSC Cornmeal Food) because the sugar/protein composition of host food affects Wolbachia titer [84].

Wild-type fertility stocks: Paired wMel-infected (OreR_wMelDB) and uninfected Oregon R (OreR_uninf) stocks were made by crossing males of Oregon-R-C to virgin females from paired infected and uninfected balancer stocks of the genotype w[1];CyO/Sp;Hu/Sb. To ensure that no differences arose between these 2 D. melanogaster genotypes during the balancer cross or subsequently, we backcrossed males of the OreR_wMelDB stock to females of the OreR_uninf stock for 10 generations and repeated egg lay and hatch assays (F10_OreR_uninf). Paired infected and uninfected nosGal4/CyO and nosGal4/Sb flies were made by crossing to paired infected and uninfected balancer stocks, as described above.

The genomic and phenotypic consequences of the mei-P26 alleles studied here and previously (e.g., [31]) have been characterized [40]. The hypomorphic mei-P26[1] allele was created by the insertion of a P{lacW} transposon in the first intron of mei-P26, which codes for the RING domain. The mostly null mei-P26[fs1] allele tested by Starr and Cline in 2002 was generated by a second P{lacW} insertion in mei-P26[1]’s first insertion. The fully (male and female) null mei-P26[mfs1] allele arose by deleting the P{lacW} insertion from mei-P26[1], along with approximately 2.5 kb of flanking sequence. According to Page and colleagues (2000), these 3 alleles form a series, with increasing severity: mei-P26[1] < mei-P26[fs1] (and other fs alleles) < mei-P26[mfs1].

Fecundity crosses

Overview.

SLR performed the 3,002 fecundity crosses continuously between October 2020 and January 2022, in batches based on their eclosion date (Table 1). Infected and uninfected OreR lines were maintained continuously to (1) control the age of grandmothers of CI males; (2) regularly have OreR virgins for mei-P26 and CI fecundity assays; (3) collect and age WT females for the aged fertility assay; and (4) obtain large sample sizes from mei-P26 fertility mutants. The full fecundity dataset is contained in S2 Table and plotted in S3 Fig.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Fecundity cross genotypes, sexes, ages, and control crosses.

https://doi.org/10.1371/journal.pbio.3002335.t001

Non-disjunction assays

We placed females homozygous for the mei-P26[1] allele bearing the yellow mutation (y[1] w[*] mei-P26[1];;; sv[spa-pol]) in vials with males bearing a wild-type yellow allele fused to the Y chromosome (y[1] w[1] / Dp(1;Y)y+). We made 7 or 8 vials of 1 to 10 females mated to a count-matched 1 to 10 males for infected and uninfected mei-P26 hypomorphs, respectively. After eclosion, we screened for non-disjunction in the progeny by the presence of yellow males (XO) and normally colored females (XXY). Rates of non-disjunction (NDJ) were calculated to account for the inviable progeny (XXX and YO) with the equation NDJ rate = 2 *NDJ offspring/all offspring = (2XXY + 2X0)/(XX+XY+2XXY+2X0), as in [41].

Ovary fixation and immunocytochemistry

Within a day or 2 or eclosion, flies were transferred to fresh food and aged 3 to 7 days, or longer as indicated in the text. For long aging experiments, flies were transferred to fresh food every week and investigated for mold every few days. We dissected the ovaries from approximately 10 flies from each cross in 1× PBS and separated the ovarioles with pins. Ovaries were fixed in 600 μl heptane mixed with 200 μl devitellinizing solution (50% v/v paraformaldehyde and 0.5% v/v NP40 in 1× PBS), mixed with strong agitation, and rotated at room temperature for 20 min. Oocytes were then washed 5× in PBS-T (1% Triton X-100 in 1× PBS) and treated with RNAse A (10 mg/ml) overnight at room temperature.

After washing 6 times in PBS-T, we blocked the oocytes in 1% bovine serum albumin in PBS-T for 1 h at room temperature, and then incubated the oocytes in the primary antibodies diluted in PBS-T overnight at 4°C (antibodies and dilutions listed below). The following day, we washed the oocytes 6 times in PBS-T and incubated them in secondary antibodies diluted 1:500 in PBS-T overnight at 4°C. On the final day, we washed the oocytes a final 6 times in PBS-T and incubated them over 2 nights in PI mounting media (20 μg/ml PI (Invitrogen #P1304MP) in 70% glycerol and 1× PBS) at 4°C. Overlying PI medium was replaced with clear medium, and oocytes were mounted on glass slides. Slides were stored immediately at −20°C and imaged within a month. Infected and uninfected, as well as experimental and control samples were processed in parallel to minimize batch effects. Paired wMel-infected and uninfected OreR stocks were used as wild-type controls.

Primary monoclonal antibodies from the Developmental Studies Hybridoma Bank (University of Iowa) were used at the following dilutions in PBS-T: anti-Hts 1:20 (1B1) [45], anti-Vas 1:50 [44], anti-Orb 1:20 (4H8) [63], anti-Bam 1:5 [36], and anti-Sxl 1:10 (M18) [85]. Primary monoclonal antibodies from Cell Signaling Technology were used at the following dilutions: anti-Phospho-Smad1/5 (Ser463/465) (41D10) 1:300 (#9516S) and anti-Phospho-Histone H3 (Ser10) Antibody 1:200 (#9701S). Anti-Wolbachia FtsZ primary polyclonal antibodies were used at a 1:500 dilution (provided by Irene Newton). Secondary antibodies were obtained from Invitrogen: Alexa Fluor 405 Goat anti-Mouse (#A31553), Alexa Fluor 488 Goat anti-Rabbit (#A12379), and Alexa Fluor 647 Goat anti-Rat (#A21247).

Confocal imaging

Oocytes were imaged on a Leica SP5 confocal microscope with a 63× objective. Optical sections were taken at the Nyquist value for the objective, every 0.38 μm, at variable magnifications, depending on the sample. Most germaria were imaged at 4× magnification and ovarioles and cysts were imaged at a 1.5× magnification. Approximately 10 μm (27 slices) were sampled from each germarium and 25 μm (65 slices) were sampled from each cyst for presentation and analysis.

PI was excited with the 514 and 543 nm lasers, and emission from 550 to 680 nm was collected. Alexa 405 was imaged with the 405 nm laser, and emission from 415 to 450 nm was collected. Alexa 488 was imaged with the 488 laser, and emission from 500 to 526 nm was collected. Alexa 647 was imaged with the 633 laser, and emission from 675 to 750 nm was collected.

Image analysis

Germaria and ovarioles were 3D reconstructed from mean projections of approximately 10 μm-thick nyquist-sampled confocal images (0.38 μm apart) in Fiji/ImageJ. Germline cyst developmental staging followed the standard conventions established by Spradling [86] and the criteria described below. We also scored and processed the confocal z-stacks for analysis as follows in the sections below.

Transcriptomics

We collected OreR and hypomorphic mei-P26[1] flies uninfected and infected with wMel and wMel-infected nos:Gal4>UAS:mei-P26 RNAi and nos:Gal4/CyO flies for RNAseq. Flies were collected after eclosion and moved to fresh food until they were 3 to 5 days old. Ovary dissections were performed in 1× PBS, in groups of 20 to 30 flies to obtain at least 10 mg of ovary tissue for each sample. After careful removal of all non-ovary somatic tissue, ovaries were promptly moved to RNAlater at room temperature, and then transferred to −80°C within 30 min for storage. Frozen tissue was shipped on dry ice to Genewiz Azenta Life Sciences for RNA extraction, cDNA synthesis, Illumina library preparation, and Illumina sequencing. The RNAi and control samples—both nos:Gal4>UAS:mei-P26 RNAi and nos:Gal4/CyO genotypes—were processed to cDNA with T-tailed primers, recovering only host transcripts. The mei-P26[1] and OreR samples were processed to cDNA using random hexamers and ribosomal sequences were depleted with sequential eukaryotic and bacterial rRNA depletion kits (Qiagen FastSelect). Illumina dual-indexed libraries were made from these cDNAs and sequenced as 2 × 150 bp reads.

We processed and analyzed RNAseq datasets for differential expression using standard computational approaches and custom perl parsing scripts. Briefly, following demultiplexing, we trimmed adapter fragments from the RNAseq reads with Trimmomatic [87]. To generate sitewise coverage data to examine read alignments directly, we used the STAR aligner [88] and samtools [89], and plotted read depths across samples in R. We quantified transcripts by pseudoalignment with Kallisto [90]. The choice of reference transcriptome was key to recovering the maximum number of alignments: we merged the NCBI RefSeq assemblies for the wMel reference genome CDSs and RNAs from genomic (accession GCF_000008025.1) and the D. melanogaster reference genome RNAs from genomic (accession GCF_000001215.4; Release_6_plus_ISO1_MT) to obtain alignments against the full, non-redundant host-symbiont transcriptome. Simultaneous mapping to both genomes was performed to avoid cross-species mismapping [91]. This reference transcriptome was indexed at a kmer length of 31 in Kallisto (version 0.45.1) [90] and reads were pseudoaligned against this reference with the “kallisto quant” command and default parameters. Host and symbionts have distinct transcriptome distributions [92], necessitating the separation of the 2 transcriptomes prior to transcript normalization and quantification in DESeq2 [93], which we performed with a custom perl script.

We imported the subset D. melanogaster and wMel Kallisto transcriptome quantifications separately into R with Tximport [94] for DESeq2 analysis [93]. Transcript-level abundances were mapped to gene IDs to estimate gene-level normalized counts. For each transcriptome, we filtered out low-count/coverage genes across samples by requiring at least 5 samples to have a read count of 10 or more. We modeled interactions among our experimental groups as a function of genotype, infection, and the interaction between genotype and infection (~genotype + infection + genotype*infection) and performed Wald Tests to detect differential expression. Briefly, the maximum likelihood gene model coefficient for each gene’s expression count was calculated and divided by its standard error to generate a z-statistic for each gene under the full and reduced models. These z-statistics were compared to the values obtained under standard normal distribution for p-value calculations. FDR/Benjamini–Hochberg corrections were performed on these p-values to reduce the number of false positives. Normalized counts for each gene were output with the plotCounts() function in DESeq2 for plotting in R.

We manually investigated the top hits for each test for evidence of germline expression in the Fly Cell Atlas project [95] through Flybase. GO categories for differentially expressed genes returned from each DESeq2 Wald Test and tested for significance against the reference transcriptome set of IDs with ShinyGO0.77 [96].

Transcriptomic data generated in this study are available through NCBI BioProject number PRJNA1007602.

Plotting and statistical analysis

Fecundity data were plotted, analyzed, and statistics were calculated in R. Differences in lay, hatch, and offspring production rates were evaluated with the nonparametric Wilcoxon rank sum test. When this was infeasible because non-zero samples were so few (e.g., null alleles), we compared the binomially distributed categorical groups of females who laid or did-not-lay eggs with the Fisher’s Exact test. All sample sizes, means, and p-values are presented in S2 Table. Fecundity was plotted across fly ages and fitted with local polynomial regression. Single age pots were made with base R and the beeswarm [97] package and fecundity-vs-time plots were made with the ggplot2 [98] package.

Confocal micrograph fluorescence intensities were analyzed and plotted, and statistics were calculated in R. Relative fluorescence intensities between oocyte genotypes were compared with the nonparametric Wilcoxon rank sum test. Counts of GSCs, mitotic GSCs, tumorous germline cysts, and orb-specific cysts were compared with Fisher Exact tests. Plots were made with the ggplot2 [98] package.

We analyzed and plotted the dual-RNAseq results from Kallisto and DESeq2 in R. Bar and scatter plots were made with the ggplot2 package [98] and volcano plots were made with the EnhancedVolcano package (release 3.17) [99].