The Cif proteins from Wolbachia prophage WO modify sperm genome integrity to establish cytoplasmic incompatibility

Inherited microorganisms can selfishly manipulate host reproduction to drive through populations. In Drosophila melanogaster, germline expression of the native Wolbachia prophage WO proteins CifA and CifB cause cytoplasmic incompatibility (CI) in which embryos from infected males and uninfected females suffer catastrophic mitotic defects and lethality; however, in infected females, CifA expression rescues the embryonic lethality and thus imparts a fitness advantage to the maternally transmitted Wolbachia. Despite widespread relevance to sex determination, evolution, and vector control, the mechanisms underlying when and how CI impairs male reproduction remain unknown and a topic of debate. Here, we use cytochemical, microscopic, and transgenic assays in D. melanogaster to demonstrate that CifA and CifB proteins of wMel localize to nuclear DNA throughout the process of spermatogenesis. Cif proteins cause abnormal histone retention in elongating spermatids and protamine deficiency in mature sperms that travel to the female reproductive tract with Cif proteins. Notably, protamine gene knockouts enhance wild-type CI. In ovaries, CifA localizes to germ cell nuclei and cytoplasm of early-stage egg chambers; however, Cifs are absent in late-stage oocytes and subsequently in fertilized embryos. Finally, CI and rescue are contingent upon a newly annotated CifA bipartite nuclear localization sequence. Together, our results strongly support the Host modification model of CI in which Cifs initially modify the paternal and maternal gametes to bestow CI-defining embryonic lethality and rescue.


Introduction
Numerous animal species harbor heritable microorganisms that alter host fitness in beneficial and harmful ways. The most common, maternally inherited bacteria are Wolbachia that typically reside intracellularly in reproductive tissues of both male and female arthropods. Here, they induce reproductive modifications with sex specific effects such as cytoplasmic incompatibility (CI) that can selfishly drive the bacteria to high frequencies in host populations. CI also notably yields important consequences on arthropod speciation [1][2][3] and vector control strategies [4][5][6][7][8][9] by causing lethality of embryos from Wolbachia-infected males and uninfected females. As CI is rescued by Wolbachia-infected females with the same strain [10,11], the phenotype accordingly imparts a relative fitness advantage to infected females that transmit the bacteria [12]. Two genes, CI factors cifA and cifB, occur in Wolbachia prophage WO within the eukaryotic association module enriched for arthropod functions and homology [13][14][15]. We previously demonstrated that dual, transgenic expression of cifA and cifB from wMel Wolbachia in Drosophila melanogaster males induces CI, while single expression of cifA in females rescues CI [16,17]. These results form the basis of the Two-by-One genetic model of CI for several, but not all, strains of Wolbachia [13,15,18,19]. At the cellular level, CI-defining lethality associates with chromatin defects and mitotic arrest within the first few hours of embryonic development. Normally, after fertilization, the sperm-bound "protamines" are removed in the embryo and replaced by maternally supplied "histones," resulting in the rapid remodeling of the paternal chromatin [20]. However, during CI, there is a delay in the deposition of maternal histones onto the paternal chromatin, resulting in altered DNA replication, failed chromosome condensation, and various mitotic defects that generate embryonic death [13,[21][22][23][24][25][26][27].
The incipient, prefertilization events in the reproductive tissues that establish CI and rescue remain enigmatic and under recent debate, namely whether (i) Cifs modify the paternal genome during spermatogenesis (Host modification model) or embryogenesis (Toxin-antidote model) and (ii) rescue occurs or does not occur by CifA binding CifB in the embryo [28,29]. These 2 key observations can conclusively differentiate the mechanistic models of CI, although they have not been explicitly addressed to date. Notably, paternal transmission of the proteins from sperm to embryo may occur under either mechanistic model [11,29]. Recent work proposed the Toxin-antidote model is operational using transgenic, heterologous expression of Cif proteins from wPip Wolbachia [30]. In this study, the CifB wPip protein paternally transfers to the fly embryo and associates with DNA replication stress of the paternal genome in the embryo. However, without the ability to rescue this non-native, transgenic CI and thus visualize CifA-CifB binding in the rescue embryo, these interesting results do not yet resolve the predictions of the 2 models. Additionally, the paternal DNA replication defects observed in the embryos may be established before fertilization by the Cif proteins, in concordance with Host modification model.
Here, we develop antibodies to localize Cif proteins from wMel Wolbachia during D. melanogaster gametogenesis and embryogenesis and then perform genome integrity measurements of developing sperm across transgenic, mutant, and wild-type treatment groups. We describe the following cell biological and gametic chromatin events underpinning the Host modification model of CI and rescue: (i) CifA and CifB proteins localize to the developing sperm nuclei from early spermatogonium stage to late elongating spermatids; (ii) in mature sperm, CifA associates with sperm tails and occasionally occurs in the acrosome, whereas CifB localizes to the acrosome in all mature sperms; (iii) Cifs increase histone retention in developing spermatids and decrease protamine levels in mature sperms; (iv) both CI and rescue are dependent upon a newly annotated bipartite nuclear localization signal (bNLS) in CifA that impacts nuclear localization and sperm protamine levels; (v) during copulation, both Cif proteins transfer with the mature sperm exhibiting reduced protamine levels; importantly, protamine mutant flies enhance wild-type CI; (vi) in the ovaries, CifA is cytonuclear in germline stem cells and colocalizes with Wolbachia in the nurse cell cytoplasm; and (vi) CifA is absent in the embryos, and, thus, rescue must be established in oogenesis independently of CifA's presence in the embryo. Taken together, results demonstrate that prophage WO-encoded Cif proteins from Wolbachia invade gametic nuclei to modify chromatin integrity at the histone to protamine transition stage in males. At the mechanistic level, results support the Host modification model of CI and rescue whereby native Cifs wield their impacts prior to fertilization.

CifA and CifB invade sperm nuclei during spermatogenesis and spermiogenesis
To evaluate the cellular localization of the Cif proteins, we generated monospecific polyclonal antibodies for visualizing the proteins in reproductive tissues (S1 Fig). In D. melanogaster males, the sperm morphogenesis process is subdivided into 2 events: (i) spermatogenesis including mitotic amplification and meiotic phases; and (ii) spermiogenesis, a postmeiotic phase. During spermatogenesis, the germline stem cell undergoes 4 rounds of synchronous mitotic divisions to produce 16 precursor cells called spermatogonia. The spermatogonia then grow and become spermatocytes [31]. After the growth phase, the spermatocytes divide by meiosis and differentiate into 64 haploid round onion spermatids. Postmeiosis, the round sperm nuclei elongate to gradually change their shape accompanied by reorganization of the chromatin during the canoe stage [32]. This results in an individualization complex forming slim, needle-shaped sperm nuclei with reduced volume [31,32]. Elongation and individualization of the spermatids is the final stage of spermiogenesis, after which the mature sperms are transported to the seminal vesicle [31,33].
CifA, but not CifB, localizes in the germline stem cells at the apical end of testes in <8-hour-old cifA and cifB transgene-expressing (Fig 1) and wild-type wMel+ males (S2 Fig). CifA and CifB were both detected in the nuclei of mitotic spermatogonium and spermatocytes. CifA is more abundant than CifB in the spermatogonium stage (S3A Fig and S1 Table). In the postmeiotic, round onion spermatids, clusters of both CifA and CifB are adjacent to the nuclei. In the elongating canoe-shaped spermatids, CifA and CifB localize apical to the sperm head nucleus, in what is likely the acrosome (Figs 1 and S2). CifB is present in all of the spermatid nuclei, whereas CifA is present on average in 39% of the elongating spermatids per sperm bundle (S3B Fig). During the elongating canoe stage, chromatin-bound histones are typically removed and replaced with protamines to yield compact nuclear packaging and chromatin organization of sperm DNA [32]. After nuclear compaction is complete, neither of the Cif proteins are detectable in late spermatid needle-shaped nuclei (Fig 1) and in the mature sperms from the seminal vesicle (S4 Fig), indicating either the Cif proteins are fully stripped or they might not be accessible by the antibodies when the chromatin is tightly compacted [34,35]. To evaluate Cif presence/absence in the mature sperm, we decondensed sperms after isolations from seminal vesicles of <8-hour-old males (see Methods) and stained them with the respective Cif antibodies. CifA is common along sperm tails in a speckled pattern (Figs 1 and S2) and infrequently present in the acrosome region, on average in 45% or 0% of the mature sperm heads depending upon the sampled seminal vesicles (S3B Fig). CifB is present in all of the acrosomal tips of the sperms and not localized to the sperm tails (Figs 1 and S2).
During spermatogenesis, Wolbachia are stripped into the cytoplasmic waste bags, which eliminate excess cytoplasmic material during the process of spermatid elongation [23]. Here, we show that some CifA and CifB proteins are also stripped by the individualization complex into the cytoplasmic waste bag (S5 Fig). Since Wolbachia are not present in the mature sperms [36,37], these data suggest that the Cif proteins exit Wolbachia cells during spermatogenesis to possibly interact with and modify sperm DNA (see below). Taken together, these findings demonstrate CifA and CifB proteins access Drosophila sperm nuclei throughout development.

Cifs cause abnormal histone retention and protamine deficiency
Since the Cifs localize to developing sperm nuclei during spermatogenesis, we hypothesized that they may incipiently interact with nuclear DNA to impact sperm genome integrity-a central prediction of the Host modification model of CI [11,29]. At the histone-to-protamine transition stage during spermiogenesis [38], histones normally undergo various posttranslational modifications (PTMs) for removal and replacement by smaller protamines for tight chromatin reorganization [38][39][40][41]. Lack of PTMs can lead to histone-bound chromatin with improper protamine deposition that causes paternal chromatin defects, male infertility, and embryonic lethality [27,42,43]. Thus, the incipient defects initiated in the testes can lead to postfertilization catastrophes.
Utilizing a core histone antibody, we investigated histone abundance within spermatid bundles at the late canoe stage in CI-and non-CI causing males. We detected significantly increased histone retention in both wMel+ and cifAB-expressing testes from <8-hour-old males compared to the negative controls (Figs 2A and S6 and S1 Table). Single transgenic expression showed significantly less histone-retaining bundles at this stage similar to wMel− negative controls (S7A Fig and S1 Table). To detect if abnormal histone retention is linked with protamine deficiency in mature sperms, we next used the fluorochrome chromomycin A3 (CMA3) stain that fluoresces upon binding to protamine-deficient regions of DNA [44,45]. Mature sperms isolated from wild-type seminal vesicles of young (high CI inducing) wMel+ males exhibit increased protamine deficiency relative to wMel− males (Fig 2B and S1 Table).
To investigate if lack of protamines associates with CI, we isolated sperms from <8-hour-old males with a protamine A and B knockout mutant line. We show that protamine mutants, both in the presence (ΔProt+) and absence (ΔProt−) of Wolbachia, also exhibit a significant increase in fluorescence relative to wMel− (Fig 3A and S1 Table). Moreover, a key outcome of higher protamine deficiency in ΔProt+ males with Wolbachia is an increase in CI compared to wildtype wMel+ CI, under the same experimental setup (Fig 3B and S2 Table). These findings from <8-hour-old males of wMel+, wMel−, and transgenic cifAB lines were dissected and immunostained to visualize and quantify spermatid bundles with histone retention (purple) during late canoe stage of spermiogenesis. DAPI stain (blue) was used to label spermatid nuclei. Total spermatid bundles with DAPI signals and those with retained histones were manually counted and graphed. Compared to the negative control wMel−, wMel+ Wolbachia and dually expressed cifAB transgenic lines show abnormal histone retention in the late canoe stage. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by pairwise comparisons based on Kolmogorov-Smirnov test. (B) Mature sperms isolated from seminal vesicles (n = 15) of <8-hour-old males reared at 21˚C were stained with fluorescent CMA3 (green) for detection of protamine deficiency in each individual sperm nucleus. Individual sperm head intensity was quantified in ImageJ (see Methods) and graphed. wMel+ and transgenic cifAB lines show enhanced protamine deficiency levels compared to wMel− control. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S1 Table. The experiments were performed in 2 independent biological replicates and samples were blind-coded for the first run. Raw data underlying this figure can be found in S1 Data file. CMA3, chromomycin A3. indicate an additive effect by Wolbachia and ΔProt knockouts on the protamine deficiency and CI penetrance. Notably, ΔProt− males do not recapitulate CI on their own; thus, the protamine deficiency is not the sole cause of CI and must operate in conjunction with other CI modifications. Consistent with these results, 7-day-old wMel+ males that express almost no CI exhibit a similarly weak protamine deficiency level to wMel− males, as expected (S8A and S8B Fig and S2 Table). Moreover, transgene analyses specify both single and dual expression of CifA and CifB cause protamine deficiencies at significantly higher levels than negative controls of wMel− and a non-CI transgene (S7B Fig and S1 Table). Since singly expressed Cifs do not cause CI in D. melanogaster [16] (S7C Fig and S2 Table), additive effects on the protamine deficiency and/or histone retention due to abnormal PTMs may be required to fully establish CI.

Both CI and rescue are dependent upon a CifA bNLS
Based on the cNLS mapping tool for nuclear localization signals [46], CifA amino acids harbor a predicted bipartite nuclear localization sequence (S3 Table) in the most conserved region of the protein [13,47,48] that is under strong purifying selection [17]. As nuclear localization signals bind to the extended surface groove of nuclear transport protein importin-α, also known as karyopherin-α [49], we hypothesized that sperm nuclear localization of CifA, and CI and rescue are dependent on the bNLS. To test this hypothesis, we mutagenized 2 bNLS sequences with alanine substitutions (aa189-190 for NLS1 (denoted cifA 189 ) and aa224-225 for NLS2 (denoted cifA 224 )), and we additionally deleted the entire bNLS region (cifA ΔbNLS ) ( Fig 4A). The bNLS deletion also corresponds to the weakly predicted catalase-rel domain in CifA [47,48].
Each bNLS mutant, individually and together (cifA 189;224 ), was dually expressed in testes with transgenic, intact cifB to assess CI and singly expressed in females to assess rescue. Transgenic cifA 189 expression significantly reduced CI and rescue as previously reported (Fig 4B and  4C and S2 Table) [48]. Conversely, transgenic cifA 224 expression showed no significant difference from the controls in either CI or rescue, suggesting this region has little to no impact. , Sco/CyO) males exhibit significantly increased CMA3 fluorescence indicative of protamine deficiency compared to both wild-type wMel+ and wMel−. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S1 Table. (B) CI hatch rate analyses of male siblings used in CMA3 assays (panel A) validate that ΔProt+ males with increased sperm protamine deficiency cause stronger (rescuable) CI levels than wMel+. ΔProt− males do not cause CI. Asterisks indicate statistically significant (p < 0.05) differences as determined by pairwise Mann-Whitney tests. All of the p-values related to CI assay are reported in S2 Table. Raw data underlying this figure can be found in S1 Data file. CI, cytoplasmic incompatibility; CMA3, chromomycin A3.
https://doi.org/10.1371/journal.pbio.3001584.g003  Table). These results highlight the importance of the nuclear localization sequence in inducing CI as well as rescue. To determine if the lack of CI induction is due to nonnuclear localization of CifA protein, we used the deletion mutant cifA ΔbNLS with wild-type cifB to demonstrate that in contrast to its normal, nuclear localization (Fig 1), it mislocalizes to the cytoplasm of onion stage spermatids rather than the nuclei ( Fig 4D). Additionally, to test if deletion of the bNLS impacts sperm genomic integrity, we performed a CMA3-based protamine deficiency assay (as described above) and found reduced protamine deficiency levels in matures sperms due to non CI-causing cifA ΔbNLS ;B line compared to CI-causing cifAB (Fig 4E and S1 Table), providing further evidence that protamine deficiency is linked to CI. Overall, these data provide previously unknown findings that a functional nuclear localization sequence and CifA nuclear localization impacts CI, rescue, and sperm protamine levels.

Cifs cause a paternally transferred protamine deficiency
To investigate if Cif proteins and/or the genome integrity modifications transfer paternally to the female reproductive tract, single male:female pairwise matings were set up for CI and rescue crosses. After 4 hours of mating, we isolated the whole uterus ( Fig 5A) including the sperm storage organs (spermathecae (SP) and seminal receptacle (SR)). Following sperm decondensation, antibody staining, and microscopy, we observed 2 key results. First, both CifA and CifB proteins transfer with sperm tails and heads, respectively, to the sperm storage organs of Wolbachia-free females ( Fig 5B). Second, the CI-associated sperm protamine deficiency induced by wMel+ and cifAB-expressing males transfers and persists in the sperms isolated from SP and SR of Wolbachia-free, mated females ( Fig 5C and 5D and S1 Table). These findings connect a paternally transferred sperm modification with the activity of Wolbachia and the Cifs themselves. Results strongly support the Host modification model of CI since a Cif-induced sperm modification established in the testes transfers to the female reproductive tract.

CifA is present in early oogenesis and absent from mature eggs and rescue embryos
Expression of cifA alone in the ovaries rescues CI [16,17], yet how CifA protein mediates rescue is unknown and central to further differentiating the mechanistic models of CI. For instance, CifA in females may modify reproductive cell biology to nullify CI-inducing sperm modifications in the embryo (Host modification), or, alternatively, CifA may directly bind CifB in the embryos to prevent its proposed CI toxicity (Toxin-antitoxin). Using CifA antibodies, we show in wMel+ and cifA transgene-expressing ovaries that CifA protein is cytonuclear and specifically localizes to cyst DNA in region 1 of the germarium (Fig 6A), indicative of reported in S2 Table. (D) Antibody labeling (green) and DAPI staining of onion stage spermatids in the testes of the bNLS mutant line (cifA ΔbNLS ) reveals that the deletion ablates CifA's localization to the nucleus, and CifA thus remains in the surrounding cytoplasm. The imaging experiment was conducted in parallel to nos;cifAB line shown in Fig 1. (E) Mature sperms isolated from seminal vesicles (n = 15) of <8-hour-old males of transgenic cifA Δbnls B line shows reduced fluorescence indicative of less Protamine deficiency compared to cifAB. To control for any background confounding effects of nos-Gal4VP16 driver line, wMel− fathers were prior crossed to nos-mothers to generate males with nos;wMel− genotype. CMA3 fluorescence levels of sperms isolated from nos;wMel− males were similar to wMel− wild-type lines used in previous assays in this study. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S1 Table, and raw data underlying this figure can be found in S1 Data file. bNLS, bipartite nuclear localization signal; CI, cytoplasmic incompatibility. https://doi.org/10.1371/journal.pbio.3001584.g004 nuclear access in ovaries similar to that in testes (Fig 1). Cystoblast in the germarium undergoes rounds of mitotic divisions to produce oocyte and nurse cells [50,51]. Along egg chamber Transgenic cifABexpressing and wMel− males were crossed to wMel− females. Four hours postfertilization, sperms isolated from females were decondensed and immunostained for localizing CifA (green) and CifB (red). DAPI stain (blue) was used to label nuclei. CifA is absent in sperm heads (empty arrowheads) and puctae are seen along the sperm tails (arrows). CifB is present in apical acrosomal tip of all of the sperm heads (solid arrowheads), with more distant signal in the more decondensed sperm nuclei. No Cifs are present in the sperms transferred from wMel− negative control males. (C) Individual sperm intensity quantification shows that protamine deficiency of sperms from wMel+ and transgenic cifAB males persists after transfer in the females compared to wMel− males. Sperm protamine deficiency from transgenic cifAB males also persists in the reproductive tract of wMel+ females. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S1 Table, and raw data underlying this panel can be found in S1 Data file.   (Fig 6A). While Wolbachia are abundant in the stage 10 egg chamber, CifA is notably absent. In transgenic females, CifA is also primarily detected in the germarium and cytoplasm of the early egg chambers and absent in late egg chambers (Figs 6A and S9A). Presence of high levels of CifA in Wolbachia-infected eggs is proposed to rescue CI; importantly, we did not detect CifA in approximately 30-to 60-minute-old rescue embryos during early mitotic divisions (Fig 6B). Moreover, CifA was not detected in 1-to 2-hour-old embryos (S9B Fig), whereas the positive control histone signals colocalize with embryonic DNA at both embryonic developmental stages.
CifB from wPip was recently shown to be paternally transferred to the CI embryos [30]. We evaluated if wMel CifB colocalizes with mitotic, embryonic DNA after fertilization. CifB is absent from embryonic nuclei and the cytoplasm of CI embryos (Fig 6B), suggesting that CifB is not inherited with the paternal DNA to the embryo. The absence of CifA and CifB in late wild-type and transgenic eggs and embryos indicates that host gametic changes prime the embryo for CI and rescue before fertilization, as predicted by the Host modification model of CI. This inference is also consistent with previous studies where ovarian, rather than embryonic, transgenic cifA expression rescues CI [13,[15][16][17]52].

Discussion
At the genetic level, dual expression of cifA and cifB or single expression of cifB can recapitulate CI, and cifA rescues CI [13,[15][16][17][18][19]52]. However, the cellular and mechanistic bases of CI and rescue remain unresolved and the subject of several questions: When and where do the Cif proteins localize in testes to potentiate CI? Are the Cifs transferred to the embryo already modified for CI-defining defects? Do CifA and CifB bind in the embryo to rescue lethality? Here, we establish that both CifA and CifB proteins invade nuclei of developing spermatids and modify paternal genome integrity by altering the histone-protamine transition process. Specifically, dual CifA and CifB expression induces abnormal histone retention and protamine deficiency in CI-causing male gametes to induce CI. Moreover, knocking out protamines enhances wild-type CI, and a nuclear localization sequence in CifA is essential for CI, rescue, and protamine deficiency. Finally, binding of CifA and CifB in the embryo is not evident.
Sperm genome compaction is normally achieved during the postmeiotic canoe phase of spermiogenesis, when histones are replaced by protamines [53]. This compaction process is highly conserved from flies to humans [54][55][56] and plays a crucial role in successful fertilization and embryonic development [57,58]. Histone marks are carriers of transgenerational epigenetic information [59], and changes in the sperm epigenome can lead to detrimental consequences including early embryonic lethality and birth defects [59,60]. Histones undergo various PTMs such as ubiquitination, methylation, phosphorylation, and acetylation before degradation and removal from the sperm chromatin [38]. Therefore, abnormally retained histones could result from aberrant PTMs in Cif-expressing flies. In a protein interactome screen, both ubiquitin and histone H2B were determined as binding host candidates to CifB [52]. Thus, it is possible that inhibition of the histone ubiquitination process mediates abnormal note the observed autofluorescence in green channel outlining the tissue morphology does not signify CifA signals. Ovariole images were manually adjusted in Affinity designer software to align egg chamber stages in the same plane. (B) Immunofluorescence of CifA (green) and CifB (red) in approximately 30-to 60-minute-old embryos obtained from rescue (cifAB male × wMel+ female) and CI crosses (nos;cifAB male × wMel− female). Histone antibody labeling core-histones (magenta) was used as a positive control. Histone signals were detected colocalizing with host DNA, labeled with DAPI (blue), whereas no CifA and CifB signals were detected. Dotted white embryonic periphery is drawn around the embryo shape. White arrows indicate dividing nuclei post fertilization and arrowheads indicate polar bodies. CI, cytoplasmic incompatibility.
https://doi.org/10.1371/journal.pbio.3001584.g006 PLOS BIOLOGY histone retention. Additionally, histone acetylation during the canoe stage of spermiogenesis is required for histone eviction in Drosophila [61]. Indeed, reduced acetylation levels lead to abnormal histone retention and protamine deficiency, which causes embryonic inviability in flies [61] and in mammals [40,62,63]. Future research investigating what specific PTMs are altered at the histone-protamine transition stage will help elucidate the molecular pathway(s) leading to CI.
The paternally derived Cif proteins travel with the sperms to the female reproductive tract, where CifB is present in the acrosomal region and CifA occurs along the tail. Notably, both Cifs are not evident in the embryos. While the presence of paternally transferred Cifs is ambiguous to the mechanistic models of CI [11,29] and recently confirmed in another transgenic study [30], it is the presence of CifA-CifB binding in the rescue embryo that would support the Toxin-antitoxin model. However, there is no evidence of this binding phenomenon in the embryos to date. Thus, we conclude that Cifs act before fertilization to prime the sperm chromatin and incipiently launch CI. Paternal effect proteins can modify sperm in various systems to bestow embryonic defects, even though the proteins themselves do not transfer to the embryos [60,64,65].
In Drosophila, the sperm enters the egg with an intact membrane [66]. Therefore, absence of Cifs in the embryos raises the question at what point CifA along the tail and CifB in the acrosome are lost before the sperm enters the egg. One possible explanation is that the Cif proteins are released from the sperm upon exocytosis of the acrosome. The acrosome, best known as a secretory vesicle, undergoes exocytosis and releases its contents to facilitate sperm-egg binding in mammals [67,68]. Although not well characterized in insects, studies in the house fly Musca domestica suggest loss of the sperm plasma membrane before entry into the egg, followed by exocytosis of acrosomal contents during passage of the sperm through the egg micropyle [69].
CifA is absent in Wolbachia-infected and transgenic embryos, which indicates that rescue is established during oogenesis under the Host modification model, and CifA thus does not bind and nullify CifB from wMel Wolbachia in the embryos. Indeed, a CifA mutant in the newly annotated nuclear localization sequence ablates rescue, suggesting that access to ovarian nuclei is important for rescue. Interestingly, structures of CifA and CifB support binding of the 2 proteins [70], yet in light of these results here, the CifA-CifB binding may be central to CI induction instead of rescue. Moreover, mutating amino acid sites across the length of the CifA protein, including binding and nonbinding residues, generally ablates CI and rescue [48,70]. Thus, it is possible that CifA mutants that ablate rescue do so by altering the protein structure, function, and/or location of ovarian targets to modify, rather than the embryonic binding of CifA to CifB in vivo. Future work will be important to resolve how CifA primes oogenesis to alter specific cell biological and biochemical events that underpin rescue.
Once fertilization occurs, protamines from the paternal chromatin must be removed and replaced by maternal histones to decondense and activate the chromatin of the developing embryo [71]. Interestingly, postfertilization delays in maternal H3.3 histone deposition occur in CI embryos [21]. The delay may in part be due to preloaded paternal histones or altered paternal epigenome information leading to mistiming of maternal histone deposition, hence causing CI. Thus, we propose that a genome integrity network involving histones, protamines, and possibly other factors in the gametes may be a common and defining feature underpinning the onset of CI and rescue.
Altogether, discovery of nuclear-targeting Cif proteins in male and female gametes establishes new insights on the early cell biological and biochemical steps that underpin the CI drive system with relevance to arthropod speciation and pest control [11]. In addition to disentangling the reproductive events of the Cif proteins that control gametogenesis and embryogenesis, the evidence specifies that the Cif proteins modify sperm genomic integrity and transfer paternally, but they themselves do not enter and bind each other in the embryo to enable rescue. These findings are consistent with the Host modification model of CI by wMel Wolbachia. More generally, as there are no previous reports of prophage proteins invading animal gametic nuclei to impair the histone-protamine transition during spermiogenesis, the findings have implications for expanding an appreciation of prophage-bacteria-eukaryote interactions into the realm of animal reproduction.

Cif proteins antibody development
Conserved amino acid regions of CifA and CifB proteins from wMel Wolbachia were previously identified [47]. Using these regions, monospecific polyclonal antibodies were commercially generated by Pacific Immunology (Ramona, CA) through injection of 3 synthesized and conserved short (20 aa) peptides for each protein into rabbits. Sequences of peptides were Cys-EYFYNQLEEKDKEKKLTE for CifA and Cys-DENPPENLLSDQTRENFRR for CifB. The resulting α-CifA and α-CifB antibodies were evaluated using an enzyme-linked immunosorbent assay, and titers were determined to be higher than 1:500,000 for each antibody. Using standard protocols of the Invitrogen WesternDot kit (#W10142, Carlsbad, CA), antibody specificity to wMel+ samples was verified using western blots (1:1,000-fold antibody dilution) on protein isolated from homogenates of 50 testes pairs (0-to 8-hour-old males) and 10 ovary pairs (6-day-old females) from wMel+ (positive), wMel− (negative control), and cifAB transgenic (positive) flies. The correct size band was only detected from wMel+ and cifAB reproductive tissues (S1 Fig). Because the antibodies were generated in the same animal, all subsequent labeling was done with individual antibodies.

NLS identification
CifA amino acid sequences from known Wolbachia and close relatives were input into the cNLS Mapper software [72] to identify putative NLS sequences within each protein (S3 Table). cNLS Mapper identifies sequences specific to the importin α/β pathway. A cutoff score of 4 was applied to all sequences. Higher scores indicate stronger NLS activities. Scores >8 indicate exclusive localization to the nucleus, 7 to 8 indicate partial localization to the nucleus, 3 to 5 indicate localization to both the nucleus and the cytoplasm, and score 1 to 2 indicate localization exclusively to the cytoplasm. Predicted NLS sequences are divided into monopartite and bipartite classes. Monopartite NLSs contain a single region of basic residues, and bipartite NLSs contain 2 regions of basic residues separated by a linker region.

Development of transgenic lines
A cifA variant was synthesized de novo at GenScript and cloned into a pUC57 plasmid as described previously [48]. Site-directed mutagenesis was performed by GenScript to produce the mutants outlined in Fig 5. The cifA 189 variant was first described in Shropshire and colleagues [48] as cifA 2 . UAS transgenic cifA mutant flies were then generated using previously established protocols [13]. Briefly, GenScript subcloned each gene into the pTIGER plasmid, a pUASp-based vector designed for germline-specific expression. Transgenes were then integrated into y 1 M{vas-int.Dm}ZH-2A w � ; P{CaryP}attP40 attachment sites into the D. melanogaster genome using PhiC31 integrase via embryonic injections by BestGene. At least 200 embryos were injected per transgenic construct, and successful transformants were identified based on red eye color gene included on the pTIGER plasmid containing the transgene. All sequences are reported in S4 Table. Fly rearing and strains D. melanogaster stocks y 1 w � (BDSC 1495), nos-GAL4:VP16 (BDSC 4937), UAS transgenic lines homozygous for cifA, cifB, cifAB, WD0508 [13], and Protamine mutant (w[1118]; ΔMst35B[floxed], Sco/CyO) [73] were maintained on a 12-hour light/dark cycle at 25˚C and 70% relative humidity on 50 mL of standard media. Uninfected protamine mutant line was generated by 3 generations of tetracycline treatment (20 μg/ml in 50 ml of fly media) as described in previous studies [13], followed by 2 rounds of rearing on standard food media before using in the experiments. Infection status for all lines was regularly confirmed by PCR using Wolb_F and Wolb_R3 primers [74].

Hatch rates
Parental flies were either wild-type uninfected (wMel−) or infected (wMel+) with Wolbachia or transgene-expressing with no Wolbachia infection. Uninfected transgenic flies were generated previously [13,17]. Paternal grandmother age was controlled to 9 to 11 days for expression of naturally high penetrance of wMel CI [75]. Parental transgenic males were generated through crossing nos-Gal4:VP16 virgin females (aged 9 to 11 days) to UAS-cif transgenic, uninfected males [75]. Mothers were aged 6 to 9 days before crossing, while father males first emerged between 0 and 8 hours were used in hatch rates and tissue collections to control for the younger brother effect associated with lower CI penetrance [13,76].
Hatch rates were set up as described previously [13,17]. Briefly, a male and female pair was placed in an 8 oz, round bottom, polypropylene Drosophila stock bottle with a grape juice agar plate containing a small amount of yeast placed at the base and secured with tape. These bottles were then placed in a 25˚C incubator overnight to allow for courting and mating. The following day, these plates were discarded and replaced with new grape juice agar plates with fresh yeast. After an additional 24 hours, the plates were removed, and the embryos were counted. The embryo plates were then incubated for 36 hours at 25˚C before the total number of unhatched embryos were counted. Any crosses with fewer than 25 embryos laid were discarded from the analyses. Significant differences (p < 0.05) were determined by pairwise Mann Whitney U tests or by a Kruskal-Wallis test and Dunn multiple test correction in GraphPad Prism 7. All p-values are listed in S2 Table.

Immunofluorescence: Testes and seminal vesicles
Siblings from the hatch rate (males 0 to 8 hours) were collected for testes dissection in ice-cold 1× PBS solution. Tissues were fixed in 4% formaldehyde diluted in 1× PBS for 30 minutes at room temperature and washed in 1× PBS-T (1× PBS + 0.3% Triton X-100) 3 times for 5 minutes each. Tissues were then blocked in 1% BSA in PBS-T for 1 hour at room temperature. They were then incubated with 1˚antibody (α-CifA 1:500 OR α-CifB 1:500) overnight at 4˚C rotating. After washing in 1× PBS-T 3 times for 5 minutes each at room temperature, they were incubated with 1:1,000 dilution Goat anti-rabbit Alexa Fluor 594 secondary antibody (Fisher Scientific, Cat#A11037, CA, USA) for 4 hours at room temperature in the dark. Tissues were then washed 3 times for 5 minutes each in 1× PBS-T and mounted on slides. To stain the nuclear DNA, 0.2mg/mL of DAPI was added to the mounting media before the coverslip was gently placed over the tissue and excess liquid wiped away. Slides were allowed to dry overnight in the dark before viewing on the Zeiss LSM 880 confocal microscope. All images were acquired with the same parameters for each line and processed in ImageJ as described in [77].

Decondensation of mature sperm nuclei
Squashed seminal vesicles collected from male flies (aged 0 to 8 hours) were treated with 10 mM DTT, 0.2% Triton X-100, and 400 U heparin in 1× PBS for 30 minutes [34]. The slides were then washed quickly in 1× PBS before immunofluorescence staining (see above).

Immunofluorescence and quantification: Histones
Testes from male flies (aged 0 to 8 hours) were fixed and stained as described above for testes. The tissues were stained with a core histone antibody (MilliporeSigma, Cat#MABE71, USA) (1:1,000) and imaged on a Keyence BZ-800 microscope. Total late canoe-stage sperm bundles were quantified in each testis, and those that retained histones were determined. Ratios of late canoe-stage bundles containing histones relative to total bundles from each individual testis were graphed in GraphPad Prism 7. Statistical significance (p < 0.05) were determined by pairwise comparisons based on Kolmogorov-Smirnov test and multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction in GraphPad Prism 7.

Sperm isolation and CMA3 staining/quantification
Seminal vesicles were collected from male flies (aged 0 to 8 hours for 1-day-old flies and 7 days for older flies) and placed on a microscope slide in ice-cold 1× PBS. Sperm was extracted on the slide using forceps and fixed in 3:1 vol/vol methanol:acetic acid at 4˚C for 20 minutes. Excess solution was then removed, and the slide was air dried. Each slide was treated in the dark for 20 minutes with 0.25 mg/mL of CMA3 in McIlvain's buffer, pH 7.0, with 10 mM MgCl2. Sperm was then washed in 1× PBS, mounted, and imaged using a Keyence BZ-X700 Fluorescence microscope. All images were acquired with the same parameters for each line and did not undergo significant alteration. Fluorescence quantification was performed by scoring fluorescent pixels in arbitrary units (A.U.) within individual sperm heads using ImageJ as per the details described in [77], and calculated fluorescence intensity per sperm head was graphed. Statistical significance (p < 0.05) was determined by a Kruskal-Wallis test and Dunn multiple test correction in GraphPad Prism 7. All of the experiments involving CMA3 staining were performed at 21˚C instead of 25˚C. CI hatch rate assays were run in parallel to ensure that CI and rescue phenotypes are not impacted due to changed temperature conditions.

Immunofluorescence: Ovaries
Ovaries from females (6 days old) were dissected in 1× PBS on ice and processed as described previously [76,78]. Tissues were blocked in 1% BSA in PBS-T for 1 hour at room temperature and were first incubated with α-CifA (1:500) primary antibody at 4˚C overnight. After washing in 1× PBS-T 3 times for 5 minutes each at room temperature, they were incubated with 1:1,000 dilution Alexa Fluor 488 secondary antibody (Thermo Fisher Scientific, Cat#A11034, USA) for 4 hours at room temperature in the dark. Samples were then rinsed properly and blocked again before incubating with α-ftsZ (1:150) primary antibody (a kind gift from Dr. Irene Newton) to stain Wolbachia at 4˚C overnight. After washing in 1× PBS-T 3 times, samples were incubated with second secondary antibody (Alexa Fluor 594) for 4 hours in the dark. Since both CifA and ftsZ antibodies were generated in the same animal, we used secondary antibodies conjugated to 2 distant fluorophores to distinguish specific signals. Tissues were then washed 3 times for 5 minutes each in 1× PBS, stained with DAPI to label nuclear DNA and mounted on slides. Slides were allowed to dry overnight in the dark before viewing on the Zeiss LSM 880 (USA) confocal microscope.

Immunofluorescence: Embryos
After 24 hours of mating, plates were switched, and embryos were collected every 30 minutes. Embryos were collected in a 100-μm mesh basket in embryo wash solution. To remove the chorion, the basket was placed in 50% bleach for 3 minutes and then rinsed with 1× PBS. The embryos were then transferred to 50:50 4% paraformaldehyde (PFA) and heptane in a microcentrifuge tube and rotated for 20 minutes at room temperature. Tubes were then removed from the rotator, and the heptane and PFA were allowed to separate before the bottom PFA phase was carefully removed. Methanol was added to the remaining heptane, and the tube was shaken vigorously for 20 seconds before the embryos settled to the bottom and solution was removed. A new volume of methanol was added to the embryos, and they were allowed to settle to the bottom of the tube. Methanol was removed, and all blocking, staining, and imaging steps were carried out for testes and ovary tissues above. from <8-hour-old males of single transgeneexpressing lines cifA, cifB, and a non CI-causing control gene WD0508 were dissected to quantify spermatid bundles with histone retention (purple) during late canoe stage of spermiogenesis. DAPI stain (blue) was used to label spermatid nuclei. Total spermatid bundles with DAPI signals and those with retained Histones were manually counted and graphed. Single transgenic expressing lines showed significantly less histones similar to the negative control wMel − at the late canoe stage. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. (B) Mature sperms isolated from seminal vesicles (n = 15) of <8-hour-old males reared at 21˚C were stained with fluorescent CMA3 (green) for detection of protamine deficiency in each individual sperm nucleus. Individual sperm head intensity was quantified in ImageJ (see Methods) and graphed. cifA-and cifB-expressing lines showed significantly higher fluorescence indicative of reduced levels of protamines compared to wMel− and WD0508 control lines. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S1 Table. The experiments were performed in parallel to the ones shown in Fig 2. (C) CI hatch rate analyses of transgenic male siblings used in CMA3 assays (Figs 2B and S6) validate that CI crosses (black circles) yielded significantly less embryonic hatching compared to non CI-inducing ones, when reared at 21˚C. Letters to the right indicate statistically significant (p < 0.05) differences as determined by multiple comparisons based on a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S2 Table. Raw data underlying this figure can be found in S1 Data file. CI, cytoplasmic incompatibility; CMA3, chromomycin A3. (TIF)

S8 Fig. Seven-day-old males do not cause CI and are not protamine deficient. (A)
Sperms from the 7-day-old wild-type Wolbachia-infected (wMel+) males show similar level of protamine levels as of wMel−. Vertical bars represent mean, and error bars represent standard deviation. Letters indicate statistically significant (p < 0.05) differences as determined by pairwise Mann-Whitney test. All of the p-values are reported in S1 Table. (B) CI hatch rate analyses of male siblings used in CMA3 assays (panel A) validate that 7d old wMel+ do not induce CI that correlates with their normal levels of sperm protamine levels. Letters to the right indicate statistically significant (p < 0.05) differences as determined by multiple comparisons calculated using a Kruskal-Wallis test and Dunn multiple test correction. All of the p-values are reported in S2 Table. Raw data underlying this figure can be found in S1 Data file. CI, cytoplasmic incompatibility; CMA3, chromomycin A3.