Abstract
The alphaproteobacterium Wolbachia pipientis infects arthropod and nematode species worldwide, making it a key target for host biological control. Wolbachia-driven host reproductive manipulations, such as cytoplasmic incompatibility (CI), are credited for catapulting these intracellular bacteria to high frequencies in host populations. Positive, perhaps mutualistic, reproductive manipulations also increase infection frequencies, but are not well understood. Here, we identify molecular and cellular mechanisms by which Wolbachia influences the molecularly distinct processes of GSC self renewal and differentiation. We demonstrate that wMel infection rescues the fertility of flies lacking the translational regulator mei-P26, and is sufficient to sustain infertile homozygous mei-P26-knockdown stocks indefinitely. Cytology revealed that wMel mitigates the impact of mei-P26 loss through restoring proper pMad, Bam, Sxl, and Orb expression. In OreR files with wild-type fertility, wMel infection elevates lifetime egg hatch rates. Exploring these phenotypes through dual-RNAseq experiments revealed that wMel infection rescues and offsets many gene expression changes induced by mei-P26 loss at the mRNA level. Overall, we show that wMel infection beneficially reinforces host fertility at mRNA, protein, and phenotypic levels, and these mechanisms may promote the emergence of mutualism and the breakdown of host reproductive manipulations.
Graphical Abstract
Highlights
● The wMel strain of Wolbachia restores fertility in females and males deficient for the essential translational regulator meiotic-P26
● Mei-P26’s germline maintenance and oocyte cyst differentiation functions are genetically rescued by wMel infection
● Perturbed pMad, Sxl, Bam, and Orb protein expression are mitigated by wMel infection
● wMel infection elevates lifetime egg lay and hatch rates in OreR flies
● Ovary transcriptomes reveal wMel infection restores or offsets changes in gene expression induced by the loss of functional mei-P26
Significance Statement Wolbachia bacterial symbionts are being harnessed as biological control agents through the use of their costly manipulations to host fertility. Here, we reveal that the wMel strain can have a beneficial impact on fruit fly development in both mutant and wild-type flies. This phenotype is essential to the long-term use of these bacteria for host population control.
Introduction
Endosymbiotic bacteria have evolved diverse strategies for infecting and manipulating host populations [1,2], which are now being leveraged for biological control applications [3]. Many of these bacteria reside within host cells and navigate female host development to colonize offspring, thus linking their fitness to that of their hosts through vertical transmission [4]. Inherited endosymbionts with reproductive manipulator capabilities go a step further by altering host development in ways that rapidly increase the frequency of infected, reproductive females in the host population. Reproductive manipulator strategies include cytoplasmic incompatibility (CI), where infected sperm require rescue by infected eggs, male-killing, feminization, and parthenogenesis [5,6]. These manipulations can be highly effective at driving bacterial symbionts into host populations, regardless of costs to the host individual or population [7].
Despite the success of parasitic reproductive manipulation, natural selection favors symbionts that increase the fertility of infected mothers, even if these variants reduce the efficacy of the parasitic mechanisms that initially drove the infection to high frequency [8]. In associations between arthropods and strains of the alphaproteobacterium Wolbachia pipientis, this scenario may be common. For example, measured fecundity in populations of Drosophila simulans infected with the wRi strain of Wolbachia swung from -20% to +10% across a 20-year span following wRi’s CI-mediated sweep across California in the 1980s [9]. This transition from fitness cost to benefit coincided with weakening of CI strength over the same time-frame [10]. Fertility-enhancing mechanisms may be at work in other strains of Wolbachia that have reached high infection frequencies in their native hosts, yet do not currently exhibit evidence of parasitic reproductive manipulation [11]. Importantly, these fitness benefits may have also played a role in the early stages of population infection, when infected hosts are at too low of frequencies for CI to be effective [12].
Currently, the wMel strain and its encoded CI mechanism are successfully being used to biologically control non-native hosts [3]. In Aedes aegypti mosquitoes, CI causes nearly 100% mortality of offspring born to uninfected mothers [13]. However, in its native host, the fruit fly Drosophila melanogaster, wMel exhibits CI that rarely exceeds 50% mortality and is extremely sensitive to paternal age [14–16], as well as grandmother age because titer increases with age [17,18]. Despite weak CI in its native host, wMel is found at moderate to high infection frequencies in populations worldwide [19,20]. Other data suggest that these frequencies may be explained by some emergent beneficial function that increases host fitness [21–25]. The molecular basis for these beneficial functions could be related to the loss of CI efficacy in D. melanogaster. Given that wMel’s use in non-native hosts relies on strong and efficient CI, it is essential that we learn the basis for its beneficial functions that could ultimately undermine CI function.
Host germline stem cell (GSC) maintenance and differentiation pathways are powerful targets for Wolbachia-mediated reproductive manipulation. Wolbachia strains have strong affinities for host germline tissues [26,27], positioning them at the right place to manipulate and enhance host fertility. In the strains that form obligate associations with Brugia filarial nematodes [28] and Asobara wasps [29], Wolbachia is required in the germline to prevent premature differentiation and achieve successful oogenesis (reviewed in [30]). In the facultative wMel-D. melanogaster association, the wMel strain can partially rescue select loss of function alleles of the essential germline maintenance and differentiation genes sex lethal (sxl) and bag-of-marbles (bam) in female flies [31–33]. It is known that wMel encodes its own factor, Toxic manipulator of oogenesis (TomO), that partially recapitulates Sxl function in the GSC through derepression and overexpression of the translational repressor Nanos (Nos) [34,35]. However, Nos expression is negatively correlated with Bam expression in the early germarium [36]. Therefore, Bam’s function in cyst patterning and differentiation in wMel-infected mutant flies cannot be explained by shared mechanisms with sxl rescue or TomO’s known functions.
In a previous screen, our lab identified the essential fertility gene meiotic-P26 (mei-P26) as a host factor that influences wMel infection intensity in D. melanogaster cell culture, potentially through modulating protein ubiquitination [37]. Mei-P26 is a Trim-NHL protein that confers a wide range of functions through its multiple domains: its protein-binding NHL and b-box domains allow it to act as an adapter for multiple translational repressor complexes (e.g., Sxl, Nanos (Nos), Argonaute-1 (AGO1), see figs S1-2 and tables S1-2). Its E3-ubiquitin ligase domain likely enables it to modulate proteolysis [38]. Given mei-P26’s in vitro role in infection [37] and its in vivo role in GSC maintenance [38], differentiation [39,40], and meiosis [40,41], we selected it as a candidate gene for modulating wMel-host interactions.
We report that D. melanogaster flies infected with the wMel strain of Wolbachia partially rescue mei-P26 mutations and exhibit reinforced fertility. In flies homozygous for hypomorphic mei-P26, infection with wMel elevates fertility in both males and females to a level sufficient to maintain a stable stock, whereas the uninfected stock is unsustainable. Infection rescues GSC maintenance and cyst differentiation by mitigating the downstream effects of perturbed mei-P26 function on other protein and mRNA expression. Specifically, wMel infection restores a wild-type-like expression profile for phosphorylated Mothers against decapentaplegic (pMad), Sxl, Bam, and Oo18 RNA-binding (Orb) proteins, as well as tut and bgcn mRNAs. We find evidence of wMel’s beneficial reproductive manipulator abilities in OreR wild-type flies, illustrating how bacterially-mediated developmental resilience may be selected for in nature. These results are essential to understanding how wMel reaches high frequencies in natural D. melanogaster populations and these beneficial functions may be able to be harnessed for biological control applications.
Results
Here, we explore the effects of wMel Wolbachia infection on D. melanogaster’s essential fertility gene mei-P26 through dosage knockdown with an RNAi construct, disrupted function with a hypomorphic allele, mei-P26[1], and full knockdown with a null allele, mei-P26[mfs1]. Using fly fecundity assays, immunocytochemistry, and dualRNAseq, we show that wMel can compensate for the loss of mei-P26 to significantly rescue host fertility.
Infection with wMel rescues female D. melanogaster fertility in mei-P26 deficient flies
Infection with wMel Wolbachia significantly rescued fertility defects induced by the loss of mei-P26 function relative to the Oregon R (OreR) wild-type strain, as measured by offspring produced per female D. melanogaster per day (Fig 1A-E, fig S3; p<=2.2e-16 to 2.9e-2, Wilcoxon rank sum test; see tables S2-6). 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,C) 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).
wMel infection rescues the loss of essential host fertility gene mei-P26 in females (and males, see Fig S4E-G). 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 transheterozygotes 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) Bar plots 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 tables S4-6). 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. 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 germline stem cell (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 fig S5), and N,O) mei-P26[mfs1]-induced oogenesis defects. Fluorescence channels were sampled serially and overlaid as indicated on each set of Figs as follows: cyan=anti-Hu Li Tai Shao (Hts), yellow=anti-Vasa (Vas), blue=wMel anti-Filamenting temperature-sensitive mutant Z (FtsZ), and red=propidium iodide (PI) DNA staining. Scale bars: H-I’=10 µm, L-O=50 µm, J-K=100 µm. J) 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 cystoblast (CB) is formed, moves away from the niche, and divides into cystocytes (CCs).
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, table S4 and Fig 1A,E) due to a higher rate of egg production (p<=6.99E-05 Wilcoxon rank sum test, table S5 and Fig 1B,E), opposed to a higher hatch rate (Fig 1C,E and table S6). 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, table S4 and Fig 1A,E). 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, table S4 and Fig 1A) through increasing both the rate of egg lay (p<=2.82E-02 Wilcoxon rank sum test, table S5 and Fig 1B) and egg hatch (p<=2.55E-03 Wilcoxon rank sum test, table S6 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, table S4 and Fig 1A,D). 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, table S4 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; table S6 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, table S4 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, table S4, 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 Supplemental Methods, tables S2-6, and fig S3 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, table S7 and Fig 1F,G). However, the underlying number of eggs laid per day and the hatch rate do vary considerably with age for both infection states and genotypes (fig S4A-D). 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 (fig S4E-G; p<=6.1e-6 to 2.9e-2 Wilcoxon rank sum test, tables S3-S5). 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 (Fig S4H,I). Given the severe, yet significantly rescuable nature of the mei-P26[1] allele (Fig 1A-C and fig S2B,D), 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-H’, compared to control in Fig 1I-I’), 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 Fig1H-H’). 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 (Fig 1J-O and fig 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,M), the germline-derived interior (Fig 1M,N), 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%, respectively, Fig S6A,B). In the following sections, we analyze wMel’s ability to rescue mei-P26 function at each of the critical timepoints 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 phosphorylated Mothers Against Dpp (pMad) and Hu Li Tai Shao (Hts) in “young” 4-7 day old (Fig 2) and “aged” 10-13 day old females (fig S7A-C). 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].
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 and allows for partial scores. Wilcoxon rank sum test *=p<0.05, **=0.01, ***=0.001, ****=1e-4. F-J) Mitotic GSCs detected by phospho-histone H3 (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.
Staining revealed an increase in the average number of GSCs per mei-P26[1] germarium with wMel infection, relative to uninfected germaria in young, 4-7 day old flies (infected/hypomorph: 1.0 vs uninfected/hypomorph: 0.58, p<=2.9e-4 Wilcoxon rank sum test, table S8), 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-E and table S8). 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-C and table S8), 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-E and table S8).
The number of GSCs per germarium in wMel-infected mei-P26[1] females converges on OreR wild-type values in aged 10-13 day old germaria (fig S7C). 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, Fig 2C-E, fig S7C, and table S8). 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, fig S7C and table S8). While OreR wild-type infected and uninfected flies lose from 3% to 31% of their GSCs in the first two weeks, infected and uninfected mei-P26[1] flies increase their GSC abundance by 43% and 63% in this time (fig S7C), 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 10-13 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, fig S7C and table S8). 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-J and table S9). 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-J and table S9). There was no difference in the frequency of GSCs in mitosis between uninfected and infected wild-type germaria (5-6%, Fig 2H-J and table S9). 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, fig S7D and table S10), 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 cystoblast (CB) moves away from the niche, Sxl cooperates with Bam and Mei-P26 to bind to the nos 3’ untranslated region (UTR) and downregulate 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-B’ vs F-H’ and I, fig S7E, and table S11). 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, Fig 3A-B’ vs C-E’ and I, fig S7E, and table S11). 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].
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 Figs: 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. Colors as labeled in Figs 1A-D 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 upregulation of pMad expression in the GSC (Fig 2A-E). Purple dashed line represents wMel’s upregulation of Sxl in the GSC/CB and downregulation of Sxl by germarium region 2b. Arrows with + symbols indicate wMel’s action on gene expression.
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 1J). Bam binds to the fusome and stabilizes CyclinA to promote TA mitosis, producing 16 cyst cells from a single CB [55]. Immediately following these four mitoses, Bam expression is down-regulated in wild-type ovaries to limit germline cysts to sixteen 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-B” versus C-F” and G, fig S7F, and table S12), producing excess nurse-like cells [50]. Infection with wMel enables Bam upregulation 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 downregulated 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 table S12), whereas wMel-infected germaria do not (Fig 4G and table S12). 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.
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. 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 is 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-C, 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.
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-I). 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 table S13). 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-I and table S13) may explain this departure from expectations. Although GSC Bam expression is lower than in wild-type, Bam expression is clearly dysregulated and upregulated 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-I and table S13).
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-D vs E-H, Fig 5I, and table S14). In RNAi knockdown ovaries, the formation of tumorous germline cysts is significantly mitigated by wMel infection (35.3% vs 16.3% of cysts with more than 15 nurse cells, respectively; p=4.5e-3 Fisher’s exact test, Fig 5E-F vs G-H and table S14). 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 vs Fig 1A,C), 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].
wMel infection rescues cyst tumors and restores oocyte specification in mei-P26 mutants. A-I) Loss and over-expression of Mei-P26 induces the formation of tumorous cysts, which contain more than 15 nurse cells. wMel infection mitigates nos>RNAi-knockdown and exacerbates nos>overexpression (OE) phenotypes. A-H) Confocal max projections of ∼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 1J), 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.
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 table S14). 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 (Fig 5J-R, fig S7G, and table S15), suggesting that wMel rescues functional cyst formation. Oocyte specification in region 2a of the germarium (see diagram in Fig 1J) 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 non-specific 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-N’ vs O-Q’ and table S15). We observed oocyte-specific Orb staining 93-94% 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 table S15). 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-M’). 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 (Fig 1C, fig S8A-C, and table S4-6). 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) vs 40 (infected) offspring/female/day, p<=2.2e-3 Wilcoxon rank sum test, fig S8A and table S4). Infection with wMel increased the number of eggs laid per female per day for the nos:Gal4/CyO balancer stock by 49% (42 vs 28 eggs/female/day, p<=2.2e-3 Wilcoxon rank sum test, fig S8B and table S4), 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, Fig 1C, fig S8C, and table S6). 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 (∼50 days), and found that wMel may produce higher lifetime fecundity relative to uninfected flies (Fig 6A-C and table S7). OreR females were aged as indicated along the x-axis of Fig 6A-C and mated to 3-7 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 table S6), while maintaining similar levels of egg production as uninfected flies (Fig 6B and table S6). 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 ∼20-30 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.
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 Figs as follows: cyan=anti-Hts, yellow=anti-Vas, blue=anti-FtsZ, and red=PI. Scale bars F-G=10 µm, H=50 µm.
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-F’). 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 (fig S9A-D), suggesting low maintenance of CI mechanisms [8]. Mating wMel-infected OreR males to infected and uninfected OreR virgin females revealed that wMel reduces uninfected egg hatch by 26% when males are zero to one day old (p<=4.6e-4 Wilcoxon rank sum test, fig S9A and table S6), but this moderate effect is eliminated by the time males are five days old, on average (fig S9C and table S6). Only young males significantly impact offspring production (p<=1.7E-2 Wilcoxon rank sum test, fig S9B,D and table S4). In contrast, the Wolbachia wRi strain that naturally infects Drosophila simulans and induces strong CI [10,65], reduces uninfected egg hatch by 95% when males are zero to one day old (p<=5.6e-8 Wilcoxon rank sum test, fig S9E and table S6) and only loses some of this efficacy by five days (75% hatch reduction, p<=3.6e-4, Wilcoxon rank sum test; fig S9G, table S6). This significantly impacts overall offspring production (p<=6.5E-7 and 5.0E-2 Wilcoxon rank sum test, fig S9F,H and table S4). Paternal grandmothers of CI offspring were 3-7 days old, 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-RNAseq (see Supplemental Methods and table S15) revealed that infection significantly alters host gene expression in a variety of ways (Fig 7A-C and table S16-19), which indicate a few potential mechanisms for wMel-mediated fertility rescue. Of the 35344 transcripts in the D. melanogaster genome, 10720 were expressed at high enough levels in at least five of six samples one each genotype/infection group to be included in the analysis. Of the 1286 genes in the wMel genome, 663 were expressed in D. melanogaster ovaries (Fig 7D and table S20).
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 three 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 Supplemental Figures S10-S13 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].
Both uninfected and wMel-infected hypomorphic mei-P26[1] ovaries exhibited a 5-fold depletion in mei-P26 expression relative to OreR ovaries (fig S2B,D) and significant dysregulation in 1044 other genes due to the mei-P26[1] allele after FDR correction (Wald test FDR-adjusted p-value (padj)<= 0.05, Fig 7A,E, fig S10A, and table S17). 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 germline stem cell 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 (fig S10B-G), 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 table S18). 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 tables S19). The most significantly differentially expressed genes are featured in Figure 7F-K and other significant genes are plotted in Fig 8 and figs S11-12. Binning gene expression patterns revealed that there are at least six main rescue gene expression phenotypes induced by wMel-infection in mutant ovaries: full rescue, overshoot, undershoot, wild-type effect, inverse, and novel (Fig 7F-K, Fig8, fig S11A,B, and fig S12A,B). 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 are partial corrections towards OreR expression in mei-P26[1] wMel ovaries (Fig 7H). 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 (fig S12A), in which mei-P26[1] and OreR ovaries show opposite changes in differential expression relative to the uninfected genotype. In total, only these six “wild-type effect” and “inverse” genes were differentially expressed due to infection in OreR ovaries. Lastly, “novel” genes (Figure 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 upregulated, 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 upregulation of sordd2.
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).
GO terms associated with infection and the interaction between infection and genotype (fig S11C-G, fig S12C-G, and Fig 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 (fig S11C-G). 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 (fig S12C-G). Interestingly, two chorion proteins (Fig 7F), an ecdysteroid 22-kinase associated protein (fig S11A), and an estrogen-related receptor (ERR) (fig S12A), 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 (Figure 7D, fig S13-14, and table S20), suggesting host regulation by wMel’s constitutively expressed genes. On average, wMel transcriptomes were sequenced at 7.8-42.4x depth of coverage and D. melanogaster transcriptomes were sequenced at 25.9-35.9x (table S16). While these coverage ranges overlap, the 663 wMel genes detected only spanned -1.5-fold depletion to 1.3-fold upregulation (table S20). In contrast, the 10720 expressed D. melanogaster genes spanned an order of magnitude more differential expression, with -10.041-fold depletion to 9.866-fold upregulation (tables S17-S19). 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 (fig S13). 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 (fig S13A) suggest that some of these genes could be co-opted for host manipulation or processing host resources.
Among the 1044 significantly differentially regulated D. melanogaster genes impacted by mei-P26 knockdown and infection were six genes (of 24) predicted in the literature (padj<=0.05 Wald test, and nine of 24 by p-value<=0.05; Fig 8, fig S1, and table S1-2). 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,B). 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,B). Two of six genes exhibited evidence of wMel rescue with an “undershoot” rescue phenotype (“r” in Fig 8 A,B)): 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 post-transcriptionally. In total, these differential expression results reinforce and expand our understanding of how wMel rescues and reinforces host reproduction at transcriptional and post-transcriptional levels.
Discussion
The wMel strain of Wolbachia is a successful and widespread symbiont: naturally in D. melanogaster populations and novelly through lab-generated trans-infections of wMel into non-native hosts used for biological control. However, we know remarkably little about the processes that shape how symbiont and host co-evolve and find mutually beneficial mechanisms for their reproduction. In this work, we confirm that wMel’s CI strength is typically weak in D. melanogaster, (fig 9A-D, [14–18]) and show that wMel confers benefits on host fertility through reinforcing host germline stem cell maintenance and gamete differentiation (Figs 1-8).
Here, we demonstrate that wMel infection can reinforce host fertility in both OreR wild-type and mei-P26 mutants by correcting perturbed gene expression patterns, and we shed light on the long-standing mystery of how wMel beneficially impacts host reproduction. Previously, wMel was shown to rescue the partial loss of Sxl [31] and Bam [32], two essential genes for two very different stages of early oogenesis: GSC maintenance and CB differentiation and TA-mitosis, respectively. These findings left unresolved how wMel suppresses these mechanistically and temporally distinct cellular processes, and suggested that mei-P26 was not involved [31]. We show with an array of alleles and developmental assays that infection with wMel rescues defects associated with mutant mei-P26, a gene required for both GSC maintenance and differentiation. Infection with wMel rescues all stages of oogenesis in mei-P26 RNAi knockdowns, as well as hypomorphic and null alleles (Fig 1-5 and 7-8, figs S4-8 and S11-12, and tables S4-S15 and S18-S19). Mei-P26 is a TRIM-NHL protein that regulates gene expression via mRNA translational inhibition through the Nos mRNA-binding complex, interactions with the RNA-induced silencing complex (RISC), and protein ubiquitination [38,74]. Importantly, Mei-P26 interacts with Sxl in the GSC and CB and Bam in the CB and cystocytes [36,50]. These interactions suggest that the mechanism wMel employs to rescue mei-P26 function may also be responsible for rescuing aspects of Sxl-dependent GSC maintenance and CB differentiation and Bam-dependent cyst differentiation.
In this in-depth molecular and cellular analysis of wMel-mediated manipulation of GSC maintenance and differentiation, we discovered that wMel rescues mei-P26 germline defects through correcting perturbed pMad, Sxl, Bam, and Orb protein expression and bgcn and tut mRNA expression towards OreR wild-type levels (diagrammed in Fig 9A). Infection with wMel partially rescued GSC-specific pMad expression relative to wild-type flies (Fig 2A-E), suggesting that the bacteria restored BMP signaling from the somatic germline stem cell niche [38]. Finding that these cells are also able to divide (Fig 2F-J), further supports their functionality and stemness. Given that mei-P26 is likely involved in Drosophila Myc (dMyc) regulation, and overexpression of dMyc induces competitive GSCs [75], wMel’s GSC rescue mechanism may involve dMyc upregulation.
Model of wMel’s interactions with essential host genes in early GSC maintenance and germline cyst formation. A) In mei-P26 mutants, wMel restores normal Bam, Sxl, and Orb expression through various mechanisms, most of which are unknown (???=unknown wMel factors). Protein and mRNA expression rescue correlates with cytological rescue from GSC maintenance through germline cyst differentiation. B) Sxl rescue in the GSC is mediated through TomO’s interaction with RNPs containing nos mRNA, upregulating Nos translation. Later, in stage 8 of cyst oogenesis, TomO was found to bind orb mRNA, displacing the translational repressor Cup, and upregulating Orb translation [34,35]. Dashed arrow in A indicates TomO’s predicted function in enabling Orb derepression in early cyst differentiation. The factor that restores mei-P26 repression of Orb translation in mei-P26 mutants has not been identified.
Although, our transcriptomic data indicate that dMyc is not rescued at the transcriptional level (fig S15 and tables S17-19). In the germline cells that left the niche, wMel recapitulated the properly timed changes in pMad, Sxl, and Bam expression that are normally influenced by mei-P26 and are required during cystoblast differentiation to form the 16-cell germline cyst [39,40,50,74] (Fig 3-4, S5E,F). While Bam downregulation is not fully rescued by wMel (Fig 4), finding that tut mRNA expression is partially reduced in wMel-infected mei-P26[1] germaria (Fig 8) suggests that there may be an unreported regulatory interaction between Tut, Bam, and Mei-P26 in females [76], which could rescue the overexpression of Bam in late germline cyst development.
This is the first time wMel has been found to rescue oocyte differentiation post-cyst formation, suggesting that wMel has pervasive impacts throughout oogenesis that may reinforce embryogenesis. We confirmed the downstream consequences of Bam expression rescue in wMel-infected mei-P26[1] ovaries by finding fewer germline cyst tumors in infected flies (Fig 5A-I). These germline cysts appeared to be functional, based on restored oocyte-specific expression of the essential oocyte differentiation protein Orb in infected mei-P26[1] cysts, relative to uninfected cysts (Fig 5J-R). Finding that wMel rescues Cp16 and Cp19 chorion protein, tut, bgcn, and other transcript expression in ovary tissues (Figs 7F-8 and figs S11-12) further supports the conclusion that wMel can rescue the downstream impacts of mei-P26 loss. Future work is needed to explore how the overshoot and novel DE categories mitigate the loss of mei-P26, which may occur through some alternate or compensatory pathway. In addition to the mechanism discussed previously for ubiquitin ligase rescue, the loss of Ionotropic receptor 76a (Ir76a) or nicotinic Acetylcholine Receptor α5 (nAChRα5) expression (fig S10A) could be rescued by the novel upregulation of the Wunen-2 lipid phosphate phosphatase as a receptor or membrane transporter in wMel-infected mei-P26[1] ovaries (fig S12B). These functions could underlie wMel-induced developmental resilience in OreR embryogenesis, as demonstrated by elevated hatch rates in wMel-infected OreR flies (Fig 1C and 6C).
The wMel strain’s rescue of D. melanogaster mei-P26 is likely independent of the Wolbachia TomO protein. Prior work on TomO indicates that the bacterial protein’s primary mode of action is through destabilizing ribonucleoprotein (RNP) complexes [35] (diagrammed in Fig 9B). This is how TomO elevates Nos expression in the GSC [34] and how TomO inhibits the translational repressor Cup from repressing Orb translation in mid-oogenesis (stage 8) [35,77]. In contrast, Mei-P26 interacts with Ago1 RISC to inhibit Orb translation through binding the orb ‘3-UTR miRNA binding sites [38]. Although Mei-P26 and TomO produce opposite outcomes for Orb expression and TomO is not significantly upregulated in mei-P26[1] ovaries (fig S14C), TomO-orb binding may underlie wMel’s ability to rescue Orb repression, if wMel produces other factors that modulate the interaction. Alternatively, wMel may make another RNA-binding protein. Given that Sxl is also repressed through the sxl 3’-UTR by Bruno [78], such a protein could be involved in restoring Sxl regulation in mei-P26 mutants. Neither orb or sxl expression is rescued at the mRNA level by wMel infection (Fig 8C), lending further support for rescue at the translational or protein-interaction/modification level.
Additional bacterial factors besides TomO must be involved in reinforcing host fecundity in wild-type flies and fertility mutants. First, Mei-P26 encodes three functional domains that confer at least three different functional mechanisms for modulating gene expression (mRNA-binding, miRNA-mediated, and ubiquitination). In a bacterial genome, these domains/functions are likely to be encoded by more than one protein [79]. Second, TomO rescues Sxl’s GSC maintenance functions by binding nos mRNA, increasing levels of Nos translation, but it cannot fully rescue Sxl loss [34]. Immediately after the GSC divides to produce the CB, Sxl interacts with Bam, Mei-P26, Bgcn, and Wh to inhibit Nos translation [36,74]. Thus, a wMel factor that inhibits Nos translation, counteracting TomO’s function to destabilize Nos translational repressors through RNA-binding [34,35], remains to be identified. Furthermore, Nos upregulation cannot explain Bam rescue because Nos and Bam exhibit reciprocal expression patterns during TA mitosis [36]. Third, exacerbation of the tumor phenotype induced by nos:Gal4-driven Mei-P26 overexpression by wMel infection (Fig 5I) suggests that wMel synthesizes a protein that directly mimics Mei-P26’s functions in differentiation, and can reach antimorphic levels. Misregulated expression of a mei-P26-like factor could explain why wMel-infected mei-P26 RNAi females produce more eggs and offspring than OreR wild-type flies (Fig 1A,B). While TomO may recapitulate some of Mei-P26’s functions, such as orb-binding, at least one other factor is needed to recapitulate proper Orb, Nos, and Bam regulation.
Understanding how Wolbachia interacts with host development at the molecular level is essential to deploying these bacteria in novel hosts and relying on vertical transmission to maintain them in host populations. CI has been a powerful tool for controlling host populations [80] and driving Wolbachia to high infection frequencies [3]. The continuous presence of bacterial reproductive manipulators in host germline stem cells opens the opportunity for compensatory developmental functions to evolve, as we have shown for mei-P26. While it is not known if any of the naturally occurring mei-P26 alleles [32,81] confer loss of function that is remedied by the presence of Wolbachia, natural variation in developmental genes theoretically could have provided the selection pressure for wMel to evolve its beneficial reproductive functions. Given that selection will favor the loss of CI if conflicting beneficial impacts on host fertility are realized [8], it is imperative that we understand the mechanisms underlying beneficial reproductive manipulations such as fertility reinforcement.
Materials and methods
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 fig S1A-D and Fig 8; references in table S1) 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 of bacterial influence over germline stem cell (GSC) maintenance and differentiation pathways (fig S1A,B). 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 two 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 Bloomington Drosophila Stock Center 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 25C 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 two 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 ten 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 ∼2.5 kb of flanking sequence. According to Page et al. 2000, these three 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 3002 fecundity crosses continuously between October 2020 and January 2022, in batches based on their eclosion date (raw data in table S2). 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 table S2 and plotted in fig S3.
Cross conditions
Food (vials of Bloomington’s white food recipe), laying media (grape food), and incubation conditions were made in-house in large batches monthly. Grape spoons: Approximately 1.5 mL of grape agar media (1.2x Welch’s Grape Juice Concentrate with 3% w/v agar and 0.05% w/v tegosept/methylparaben, first dissolved at 5% w/v in ethanol) was dispensed into small spoons, allowed to harden, and stored at 4°C until use. Immediately prior to use in a cross, we added ground yeast to the surface of each spoon.
Preparation of flies for fecundity crosses
Paired wMel-infected and uninfected genetic crosses were performed in parallel for the four mei-P26 mutants (RNAi, [mfs1], [1], [mfs1/1]) to produce homozygous virgin females of both infection states for parallel fecundity crosses. Mutant genetic crosses were performed multiple (three or more) times across the 14 months to produce homozygotes from many different parents and matings. CI crosses were performed with males from infected grandmothers that were 3-7 days old. Males were aged either 0 (collected that day) or 5 days, depending on the cross. We collected males from vials over multiple days, so the majority of males were not the first-emerged.
Virgin female flies collected for fecundity crosses were transferred to fresh food and aged an average of 5 days at room temperature (from 3 to 7), or longer for the aged OreR, mei-P26 RNAi, and mei-P26[1] fecundity crosses. Males and virgin female flies were stored separately and males were aged zero or 3-7 days. Long-term aged virgin females were kept as small groups in vials of fresh white food. Every few days the vials were inspected for mold and flies were moved on to fresh food.
Fecundity cross protocol
In the afternoon of the first day of each cross, one male and one female fly were knocked out on a CO2 pad, added to a vial containing a grape spoon, allowed to recover, and then transferred to a 25°C constant humidity incubator on a 12 light/dark cycle to allow for courting and mating. The following day, each spoon was replaced with a fresh spoon, and the vials were returned to 25°C for more mating. On the third day, we removed the flies from the vials, counted the number of eggs laid on the spoon’s grape media, and replaced the spoon in the vial at 25°C for two days. After approximately 40 hours, we counted the number of hatched and unhatched eggs. The exact times and dates for all steps in all crosses were recorded (table S2) and plotted to show consistent results across the timeframe (fig S3).
Fecundity cross analysis
Fecundity data were parsed using perl scripts and plotted in R. Individual crosses were treated as discrete samples. Hatch rates were calculated from samples that laid 20 or more eggs (for 3-7 day old female plots) and 10 or more eggs (for age vs. hatch rate plots). Egg lay and offspring production rates were calculated from raw counts, divided by the fraction of days spent laying (metadata in table S2). Crosses with zero laid eggs (and offspring) were not rejected, as the fertility mutants often laid zero to few eggs. Our cross conditions were highly consistent and run daily, lending confidence to these zero-lay/offspring samples (figure S3). Drosophila fecundity vs female age data were fit in R using the stat_smooth loess method, formula y∼x, with 0.95 level confidence interval.
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 seven or eight vials of 1-10 females mated to a count-matched 1-10 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 two or eclosion, flies were transferred to fresh food and aged 3-7 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 ∼10 flies from each cross in 1xPBS 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 1x PBS), mixed with strong agitation, and rotated at room temperature for 20 min. Oocytes were then washed 5x in PBS-T (1% Triton X-100 in 1x PBS), and treated with RNAse A (10 mg/ml) overnight at room temperature.
After washing six times in PBS-T, we blocked the oocytes in 1% bovine serum albumin in PBS-T for one hour at room temperature, and then incubated the oocytes in the primary antibodies diluted in PBS-T overnight at 4C (antibodies and dilutions listed below). The following day, we washed the oocytes six times in PBS-T and incubated them in secondary antibodies diluted 1:500 in PBS-T overnight at 4C. On the final day, we washed the oocytes a final six times in PBS-T and incubated them over two nights in PI mounting media (20 μg/ml propidium iodide (PI, Invitrogen #P1304MP) in 70% glycerol and 1x PBS) at 4C. Overlying PI medium was replaced with clear medium, and oocytes were mounted on glass slides. Slides were stored immediately at -20C 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 63x 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 4x magnification and ovarioles and cysts were imaged at a 1.5x 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.
Propidium iodide 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.
GSC quantification
GSCs were scored by the presence of pMad and an Hts-labeled spectrosome in cells with high cytoplasmic volumes adjacent to the somatic germ cell niche and terminal filament [56]. When antibody compatibility prevented both pMad and Hts staining (e.g., with anti-pHH3 staining), only one was used for GSC identification. The spectrosome was distinguished from the fusome by its position anterior of the putative GSC nucleus and the presence of a posterior fusome that continues into a posteriorly dividing cystoblast. We calculated the number of GSCs per germarium as the average number of pMad and Hts-spectrosome-expressing cells. Non-whole increments of GSCs indicate situations where a putative GSC expresses one attribute, but not the other.
Mitotic GSC quantification
Putative GSCs were identified as described above and scored for the presence of anti-phospho-Histone H3 (pHH3) staining. The number of pHH3-positive cystocytes in region 1 was also quantified across germaria.
Oocyte cyst tumor quantification
Using a 10x objective, we manually counted the number of nurse cells in each stage 6-10b cyst. Each count was repeated three times consistently and the cyst tallied as having less than 15, 15, or more than 15 nurse cells per cyst.
Bam and Sxl expression measured by fluorescence
We summed 27-slice nyquist-sampled z-stacks of each germarium in Fiji/ImageJ. Fluorescence intensity was measured by setting ImageJ to measure: AREA, INTEGRATED DENSITY and MEAN GRAY VALUE. Germarium regions were delimited as in [42] using Vas expression to indicate the germline. Three representative background selections were measured for subtraction. The corrected total cell fluorescence (CTCF) was calculated for each region of the oocyte as follows: CTCF = Integrated Density – Area of selected cell * Mean fluorescence of background readings. We controlled for staining intensity within a germarium and compared relative values among germaria.
Orb expression
Anti-Orb-stained oocyte cysts were manually scored for stage and Orb oocyte-staining in confocal z-stacks ImageJ.
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 3-5 days old. Ovary dissections were performed in 1xPBS, in groups of 20-30 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 2x150bp 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 two 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 five 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 BioProjectnumber PRJNA992140.
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 Table S2. 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, 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].
Author Contributions
SLR, study conception and design, data production and analysis, and writing. JRC, data production and analysis, and writing. WTS, study design and writing.
Supplementary Materials
Supplementary Figures 1-15
Supplementary Tables 1-20
Acknowledgements
We thank the UCSC Life Sciences Microscopy Center (RRID:SCR_021135) and Ben Abrams for training and the use of their microscopes. We thank Irene Newton for aliquots of anti-Wolbachia FtsZ antibodies and Manabu Ote for a stock of UASp-mRFP/CyO (1-7M) flies. We thank Russ Corbett-Detig for his comments and feedback throughout this project. This work was supported by UC Santa Cruz, the UC Chancellor’s Postdoctoral Fellowship, and the NIH (R00GM135583 to SLR; R35GM139595 to WTS).
Footnotes
These edits represent the outcome of the first round of revision, following three reviewers' and one editor's comments. The dualRNAseq assay was added to address the reviewer question about the extent of mei-P26 knockdown with our mutant flies. Since much more data were generated than that, we performed a full differential expression analysis.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.
- 23.
- 24.
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
