Impact of a selfish B chromosome on chromatin dynamics and nuclear organization in Nasonia

Since their discovery, B chromosomes have persisted as intriguing components of many eukaryotic genomes. Found in over two thousand plant and animal species so far (Palestis et al., 2004), B chromosomes are diminutive, extra chromosomes that are distinct from the normal chromosomes. Because they are not a part of the wild-type genome, B chromosomes are, in most cases, not essential for the fitness of the organism (Camacho et al., 2000; Jones et al., 2008). Many B chromosomes contain large amounts of highly repetitive, non-coding DNAs, including satellite repeats, transposable elements, and degenerate rDNA genes (Jones, 1995; Jones et al., 2008). This quality owes to the likelihood that B chromosomes derive from the centric and pericentric regions of the normal chromosomes, where these sequences are normally located (Jones, 1995; Perrot-Minnot and Werren, 2001; Jones et al., 2008). Largely free from functional constraints, B chromosomes can change rapidly in their sequence composition over short evolutionary periods, and they have been implicated in major genome events such as the origin of sex chromosomes (Gamero et al., 2002). Thus, B chromosomes are regarded as both genome parasites and potentially important facilitators of genome evolution.

A fundamental question is how B chromosomes persist despite the fact that they are largely non-essential. This question is especially compelling given that many B chromosomes are known to be unstable during meiosis or mitosis (Jones, 1995). Some B chromosomes counter their instability through meiotic and mitotic drive mechanisms (Kayano, 1957; Fröst, 1969; González-Sánchez et al., 2003). For example, one B chromosome in the plant Lilium callosum positions itself on one side of the metaphase plate during female meiosis so that it can preferentially segregate into the meiotic product that will become the gamete (Kayano, 1957; Jones et al., 2008). B chromosomes in plants also can utilize non-disjunction during mitosis in order to accumulate in the generative (pollen) products and not the vegetative ones (Jones et al., 2008). Such drive mechanisms normally are effective only when aspects of the chromosome segregating system (e.g. the microtubule-based spindle apparatus) are intrinsically asymmetrical and can be taken advantage of, as is the case in many plants.

An alternative, and more extreme, mode of B chromosome propagation is exemplified by paternal sex ratio (PSR), a B chromosome found in the jewel wasp Nasonia vitripennis. PSR is transmitted to new wasp progeny through the sperm (Nur et al., 1988). In the presence of PSR, the paternal chromatin becomes hyper-condensed and fails to resolve into distinct chromosomes during the first mitotic division of the embryo (Werren et al., 1987; Reed and Werren, 1995). In insects, the paternal and maternal chromosome sets remain physically separated during the first mitosis by an incompletely broken down layer of nuclear envelope surrounding each set (Callaini and Riparbelli, 1996; Kawamura, 2001). This unique property gives rise to a bipartite spindle apparatus known as the ‘gonomeric’ spindle (Kawamura, 2001). Because the two chromosome sets do not mix during this first mitotic division, the maternally derived chromosomes are unaffected by the PSR-altered paternal chromatin and segregate normally to form two haploid nuclei (Reed and Werren, 1995). The paternal chromatin never resolves into distinct chromosomes or segregates, and is eventually lost. However, PSR is somehow not affected and escapes this destruction (Reed and Werren, 1995). N. vitripennis and all other insects belonging to the insect order Hymenoptera (i.e. ants, bees and wasps) exhibit a specialized mode of reproduction known as haplo-diploidy, in which females develop as diploids from fertilized eggs while males arise as haploids from unfertilized eggs. By completely destroying the paternal chromatin, PSR effectively converts fertilized N. vitripennis eggs that should become females into PSR-transmitting males (Werren et al., 1987). Owing to its extreme effect on the paternal half of the wasp genome, PSR is considered to be the most extreme selfish genetic element known in nature (Werren and Stouthamer, 2003). PSR was discovered in N. vitripennis, but subsequently, similar B chromosomes have been found in distantly related wasps belonging to the genus Trichogramma (Stouthamer et al., 2001).

A critical aspect of PSR propagation is the ability of PSR to alter the paternal chromatin so that it fails to resolve into distinct chromosomes and segregate. However, little is known about the underlying mechanism. One possibility is that PSR specifically disrupts some aspect of chromosome condensation. Alternatively, PSR may block paternal chromosome condensation secondarily by affecting one of several developmental steps preceding the first embryonic mitosis. During spermiogenesis, the spherical nucleus of the post-meiotic spermatid becomes highly elongated into a ‘needle’ shape (Tokuyasu, 1974). Subsequently, the sperm chromatin is packaged into an extraordinarily condensed state with paternal histone-like proteins known as protamines (Balhorn, 2007; Rathke et al., 2007). Once the sperm nucleus enters the egg cytoplasm, its nuclear envelope dissolves, the protamines are removed, and the paternal DNA is repackaged with conventional histones (Poccia and Collas, 1996). These events are believed to be essential for the paternal chromatin to undergo replication synchronously with the maternal chromatin so that the two nuclei can enter together into the first mitosis (Poccia and Collas, 1996). PSR could inhibit one or more of these important processes, thus altering normal cell cycle progression and, ultimately, blocking paternal chromosome resolution. Interestingly, PSR somehow evades the same fate that it imposes on the paternal set in order to be transmitted, despite the fact that both entities consist of chromatin and reside closely together in the sperm nucleus.

Another important aspect of PSR propagation is the extremely high transmission rate of this chromosome from father to offspring. Previous studies showed that PSR is transmitted to progeny at frequencies ranging from 94% to 100% (Beukeboom and Werren, 1993b). These numbers suggest either that PSR segregates stably during cell division in the male germ line or, alternatively, that PSR is propagated through previously undetected drive mechanisms. Perhaps the most critical period for PSR transmission is during the first mitotic division of the embryo. In order to be transmitted, PSR must first dissociate from the paternal chromatin. This goal is particularly noteworthy because the paternal set remains as a three-dimensional mass of chromatin (Werren et al., 1987; Reed and Werren, 1995). How PSR is capable of avoiding entanglement with the paternal chromatin mass (PCM) is currently unknown. Once free, PSR must then associate with the maternal chromosomes. A possible scenario is that PSR, unlike the paternal chromatin, resolves into distinct sister chromatids that segregate with the maternal chromatids during the first mitotic division.

In this study we utilized cytological methods to address if PSR affects the ability of the paternal chromatin to progress through several key developmental stages leading up to the first mitotic division. Additionally, we performed the first-ever visual analysis of PSR during early development to address how it becomes transmitted at the cellular level. Our experiments suggest that PSR does not prevent the paternal chromatin from undergoing replication during S phase or from entering into mitosis, but, instead, alters phosphorylation of Histone H3 and proper loading of the chromosome condensation machinery. Additionally, we found that PSR maintains a position at the periphery of the paternal nucleus, which we propose is important for its escape from the paternal chromatin. PSR also localizes at the anterior-most region of the elongated nucleus of all sperm, in striking contrast to the normal chromosomes, which are positioned broadly across the sperm nucleus. Thus, PSR may manipulate or bypass normal nuclear organization in the wasp sperm, thus adding a new dimension to its selfish nature.

We first analyzed young N. vitripennis embryos produced from crosses between wild-type females and males carrying PSR. As expected, these crosses yielded broods of all male progeny that were nearly identical in size to broods from control crosses that produced both sexes (t = 0.08, P = 0.94, d.f. = 21; Table 1). Examination of 0–1.5-hour embryos revealed previously described nuclear defects that occurred during the first mitotic division of the embryo (Werren et al., 1987; Reed and Werren, 1995). In PSR-positive embryos, the maternal and paternal nuclei became juxtaposed next to one another in a manner that was indistinguishable from control (wild-type) embryos (Fig. 1A). Additionally, both nuclei exhibited similar chromatin densities at this stage (Fig. 1A). During prometaphase, both chromatin sets underwent condensation but the paternal chromatin became abnormally denser than the maternal chromatin (Fig. 1A). Moreover, in contrast to the maternal chromatin, which resolved into individual chromosomes at metaphase, the paternal chromatin remained unresolved throughout the first mitotic division (Fig. 1A). During anaphase and telophase the maternal chromatids segregated normally into daughter nuclei, while the paternal chromatin, now considered the PCM, remained at the metaphase plate (Fig. 1A). The PCM persisted near the plasma membrane in older embryos that had undergone several rounds of mitosis (Fig. 1B), but eventually disappeared in later stages (not shown).

We addressed the possibility that PSR might induce formation of the PCM secondarily by preventing the paternal chromatin from progressing normally through the first cell cycle. In order to investigate this possibility, we examined the localization of proliferating cell nuclear antigen (PCNA) in young PSR-positive embryos. PCNA is a highly conserved component of the replication complex and has served as a marker for incompletely replicated DNA (Yamaguchi et al., 1991; Easwaran et al., 2007; Landmann et al., 2009). We observed an equal intensity of PCNA on both the maternal and paternal chromatin in both wild-type embryos (not shown) and PSR-positive embryos undergoing their first round of replication (Fig. 2A). Additionally, PCNA disappeared synchronously from both sets at the onset of mitosis (Fig. 2A). These observations suggest that the PSR-altered paternal chromatin undergoes DNA replication properly during the first mitotic division.

Although we detected no defects in replication of the paternal set preceding the first mitotic division, we did find that the PCM fails to become PCNA-positive during entry into the second round of DNA synthesis (Fig. 2A). Additionally, we detected high levels of broken DNA within the PCM during this stage that resemble levels present in yolk nuclei that are undergoing apoptosis (Fig. 2B). These findings suggest that the PCM may accrue chromatin defects during the first division. Nevertheless, proper replication of the paternal set before the first mitosis implies that the processes leading up to S phase, such as the removal of protamines from the paternal DNA and its repackaging with maternal histones, also occur normally.

We next addressed if the paternal chromatin is capable of entering properly into mitosis. To do this, we stained embryos with antibodies that recognize phosphorylated serine 10 of Histone H3 (PH3S10), a well-established chromatin marker of mitosis (Wei et al., 1999; Hsu et al., 2000; Tram and Sullivan, 2002). In these embryos, both the paternal and maternal chromatin became PH3S10 positive in a synchronous manner (Fig. 3A), demonstrating that the paternal nucleus is capable of entering normally into mitosis. However, the paternal chromatin became more enriched for PH3S10 relative to the maternal chromatin during metaphase (Fig. 3A). Additionally, the PCM abnormally retained this strong PH3S10 signal as the daughter nuclei derived from the maternal chromatin set appropriately lost this mark upon exit from mitosis (Fig. 3A). In later stage embryos the PCM retained PH3S10 inappropriately during interphase when the other nuclei were devoid of this mark (Fig. 3B). These observations suggest that the presence of PSR leads to the abnormal retention of PH3S10 on the paternal chromatin and its failure to exit properly from mitosis.

The PH3S10 mark is functionally associated with the proper condensation of chromatin into chromosomes at the onset of mitosis (Koshland and Strunnikov, 1996; Strahl and Allis, 2000). Specifically, this mark is though to be necessary for the loading of Condensins to chromatin (Ono et al., 2003; Hirano, 2005; Cobbe et al., 2006). At least two forms of the Condensin complex, each defined by different subsets of associating proteins, appear to play distinct but interrelated roles in the conversion of chromatin into distinct chromosomes (Hirota et al., 2004; Cobbe et al., 2006). At the core of these complexes are two proteins, SMC2 and SMC4 (Hirano, 2005; Cobbe et al., 2006). We reasoned that PSR-induced hyper-condensation of the paternal chromatin may involve abnormal Condensin localization as a result of inappropriate PH3S10 persistence. To test this idea, we stained wild-type and PSR-positive embryos with an antibody raised against Drosophila melanogaster SMC2 (Cobbe et al., 2006). This antibody strongly labeled metaphase chromosomes but not interphase chromatin in older wild-type embryos (Fig. 4), patterns that reflect normal Condensin localization in D. melanogaster (Cobbe et al., 2006). However, in PSR-positive embryos, SMC2 localized at a much higher level to the PCM than to the condensed maternal chromosomes (Fig. 4). This pattern mirrors the bright staining of DNA and high, persisting levels of PH3S10 that we observed on the paternal chromatin. Thus, PCM formation likely involves overloading of Condensins, in addition to abnormal phosphorylation of Histone H3.

We sought to determine when and how PSR dissociates from the PCM to join the maternal chromosomes for transmission. To do this, we designed fluorescently labeled DNA probes against two previously identified satellite DNA repeats present on the PSR chromosome (see Materials and Methods) (Eickbush et al., 1992). When hybridized to meiotic metaphase chromosomes, these probes highlighted two small blocks of satellite DNA, one on each end of PSR, thereby delineating its entire chromosomal length (Fig. 5A). The fact that these two satellite blocks are located on the chromosome ends supports the idea that PSR consists largely of non-coding repetitive sequences because these sequences are normally located in the pericentric regions and not within the chromosome arms in many higher eukaryotes. Furthermore, this finding is consistent with the idea that PSR derives evolutionarily from the pericentric region of a normal chromosome (Perrot-Minnot and Werren, 2001).

We also generated probes that specifically recognize two different normal chromosomes. One of these probes is directed against a satellite DNA repeat located in the rDNA spacer region (Stage and Eickbush, 2010). This probe highlights a single locus that overlaps with a chromosomal constriction that we interpret to be the Nuclear Organizer Region (NOR; Fig. 5A). This probe did not, however, localize to PSR, suggesting that PSR, unlike many other B chromosomes (Camacho et al., 2000; van Vugt et al., 2005), does not contain rDNA. To confirm this conclusion, we performed hybridizations with a combination of three probes to the highly conserved 18S region of the rDNA locus (see Materials and Methods). Like the rDNA spacer probe, these 18S probes localized to a single locus on a N. vitripennis chromosome but not to PSR (not shown). In addition to these rDNA probes, we generated a probe that recognizes a single satellite block on a normal wasp chromosome that is distinct from the one containing the NOR (Fig. 5A).

Using the PSR-specific probes, we analyzed PSR dynamics during the first embryonic mitosis (Fig. 5B). As expected, we observed a single PSR-specific signal in only one of the two juxtaposed nuclei of post-fertilization embryos (Fig. 5B). It is highly likely that the PSR-positive nucleus is the paternally derived one because in many embryos this nucleus often appeared elongated, as does the sperm nucleus just after fertilization. A single PSR-specific signal was observed within the paternal chromatin before entry into the first mitosis and during the first metaphase (Fig. 5B). However, during anaphase and telophase, a single PSR-positive signal was associated with each of the two maternally derived daughter chromatid sets undergoing segregation but not with the lagging PCM (Fig. 5B). These observations demonstrate that PSR escapes the defective paternal chromatin and associates with the viable maternal chromosomes during anaphase of the first mitotic division. The fact that PSR can effectively segregate with the maternal chromosomes at this time demonstrates that the centromeres of the PSR sister chromatids are functional. It is also likely that the centromeres of the paternal chromatin are functional because the lagging paternal chromatin often appeared oblong as if under tension, with pointed ends stretching toward the two maternally derived nuclei (Fig. 1A, Fig. 5C). We also noticed that the PCM occasionally segregated with one of the two daughter chromatid sets derived from the maternal nucleus (Fig. 5C). This co-segregation has the potential of destroying the associated chromatid set. However, embryos in which this occurs are likely to survive due to the presence of the second, free daughter set.

PSR was found to localize to the periphery of the paternal nucleus in all young embryos analyzed between fertilization and the first prophase (n = 19/19; Fig. 5B). Additionally, PSR often was present in a position immediately adjacent to the maternal chromosomes (Fig. 5B). In several cases we observed PSR residing in a condensed form just outside and separate from the uncondensed paternal chromatin during metaphase (Fig. 5C). The rDNA locus also was found at the nuclear periphery, suggesting that localization of PSR at the nuclear periphery is a general characteristic of pericentric chromosomal regions in N. vitripennis (Fig. 5C). Nevertheless, we speculate that this peripheral position allows PSR to effectively escape the PCM for association with the maternally derived chromosomes.

We analyzed the nuclear position of PSR in developing and mature sperm (Fig. 6). PSR was present near the anterior tip of the elongated nucleus in all of the mature sperm examined (Fig. 6A,D). This position is at the end of the sperm that contains the acrosome vesicle and first enters the egg during fertilization. In contrast to PSR, two different wasp chromosomes localized broadly across the elongated sperm nucleus (Fig. 6B–D). Earlier, during the process of spermatid nuclear elongation, PSR localized to the edge of the nucleus that forms one of the two elongating nucleus ends (Fig. 6E), showing that PSR is at the position that will become the anterior nucleus tip from the beginning of elongation. In contrast, the rDNA locus was present in different regions within the elongating spermatid nucleus (Fig. 6E). This disparity in localization patterns between PSR and the normal chromosomes suggests the intriguing possibility that PSR may alter or bypass normal nuclear organization during sperm nucleus elongation to obtain its peripheral position at the tip of the sperm nucleus.

Previous genetic studies suggested that PSR transmission does not involve meiotic or mitotic drive mechanisms (Beukeboom and Werren, 1993a). To cytologically confirm this conclusion, we visualized PSR in whole testes. The majority of germ cells in the testis are cysts of mitotically dividing (pre-meiotic) spermatocytes and post-meiotic spermatids. We observed a single PSR signal within each nucleus of these cysts (Fig. 6F). We also found exceptionally large clusters of PSR signal within the giant nuclei of somatic cells (Fig. 6F). We interpret these PSR clusters to be multiple chromosome copies due to the polyploid nature of these and other somatic cells in N. vitripennis (Rasch et al., 1977). A single region of PSR signal was detected in nearly all post-meiotic spermatids (n = 206/210; Fig. 6F) and mature sperm (n = 146/149). Similar patterns were found for the chromosome carrying the rDNA locus (Fig. 6F). These observations demonstrate that PSR is stably transmitted during the mitotic and meiotic divisions in the male germ line, a finding that is consistent with the idea that PSR transmission does not involve pre- or post-meiotic drive.

A fascinating aspect of B chromosomes is how they can persist within a given eukaryotic genome without contributing to the fitness of the organism. The B chromosome, PSR, exhibits a unique means of achieving this goal – it takes advantage of the haplo-diploid reproduction of N. vitripennis, completely destroying the paternal nuclear material of this organism so that all embryos fertilized by PSR-bearing sperm, which should become diploid females, are transformed into haploid, PSR-transmitting males. Previous studies demonstrated that, in the presence of PSR, the paternal chromatin does not properly condense into chromosomes during the first mitotic division and is, therefore, rendered incapable of segregating with the maternal chromosomes (Nur et al., 1988; Reed and Werren, 1995). A fundamental question is specifically how PSR induces this effect on the paternal half of the genome while, at the same time, escaping this fate and becoming transmitted to new progeny at near perfect frequencies.

Our results suggest that PSR does not destroy the paternal chromatin indirectly by altering its progression into the first mitotic division. Specifically, we found that both maternal and paternal sets synchronously undergo DNA replication and entry into mitosis while in the presence of PSR. Instead, our experiments suggest that the process of chromosome condensation is specifically affected. We found that SMC2, a primary Condensin component, appears highly enriched on the paternal chromatin relative to the maternal chromosomes during the first mitotic division. Additionally, Histone H3 of the paternal chromatin remains phosphorylated at Serine residue 10 long after the maternal chromosomes have exited from the first mitotic division. In fact, the PH3S10 mark persists on the paternal chromatin after multiple mitotic divisions, suggesting that the PCM fails to exit properly from mitosis. Our experiments do not allow us to conclude if persistent PH3S10 leads directly to Condensin overloading. However, this hypothesis is consistent with the likelihood that the PH3S10 mark is necessary for proper Condensin loading to chromatin (Ajiro et al., 1996a; Ajiro et al., 1996b; Van Hooser et al., 1998). We speculate that PSR does not directly cause abnormal Histone H3 phosphorylation because the paternal set is not PH3 positive before entry into the first mitosis. Instead, PSR may induce other, more upstream ‘epigenetic’ marks that ultimately affect proper Histone H3 phosphorylation.

Previous groups have speculated that PSR ‘imprints’ the paternal chromatin some time during spermatogenesis (Werren and Stouthamer, 2003). During later stages of this developmental period, histones are largely stripped away from the paternal DNA before it is packaged with protamines. However, a fraction of histones are known to remain within the sperm chromatin in a number of organisms, including Drosophila (Gatewood et al., 1987; Dorus et al., 2006; Govin et al., 2007). In wasps, PSR could modify any remaining histones during sperm development in a way that specifically affects their phosphorylation and ability to facilitate chromatin condensation following fertilization. Alternatively, PSR could directly alter methylation of the paternal DNA in the testis. Recent experiments have uncovered substantial Cytosine methylation at a number of genes in N. vitripennis (Park et al., 2011). In other organisms, DNA methylation is closely linked with proper Histone H3 methylation (Jackson et al., 2002). Such marks, if altered, could conceivably affect downstream chromatin processes such as chromosome condensation. It is intriguing to speculate that PSR, while thought to be largely heterochromatic and gene-poor, encodes a DNA- or Histone-modifying enzyme that is expressed in the testis and is capable of modifying the paternal chromatin but not its own chromatin. This scenario is plausible, first, because largely heterochromatic chromosomes, such as the Y and 4th chromosomes in D. melanogaster, are known to contain protein-coding genes (Adams et al., 2000). Second, a number of chromatin remodeling enzymes are known to act on distinct types of chromatin. For example, the Histone methyltransferase 1 (HMT1) in human, the G9a Histone H3 methyltransferase in mouse, and the Histone H4 acetyltransferase MOF in Drosophila are associated primarily with euchromatin (Akhtar and Becker, 2000; Wang et al., 2001; Tachibana et al., 2005; Kleefstra et al., 2006). A PSR-linked enzyme that acts solely on euchromatin would be expected to affect the normal wasp chromosomes but not PSR. Alternatively, the presence of PSR could cause the mis-regulation of a euchromatin-modifying enzyme encoded in the wasp genome. Testing these hypotheses will be greatly facilitated by genome-wide studies of epigenetic marks and gene expression during spermatogenesis and early embryogenesis in PSR-carrying wasps.

Our cytological analyses of PSR during early development are summarized in Fig. 7. These experiments revealed that PSR escapes the PCM and associates with the viable maternal chromosomes during anaphase of the first mitotic division. This finding is not unexpected because association of PSR with the maternal chromosomes during the later mitotic divisions would likely result in mosaic animals. However, what remains to be determined is precisely how PSR utilizes the gonomeric spindle for segregation. It is well established that, in addition to the centrosomes – the primary organizers of the spindle apparatus – chromatin also can play an active role in microtubule organization. A primary example is the centrosome-less meiotic spindle apparatus in the D. melanogaster oocyte, in which the chromosomes alone nucleate the microtubule-based spindle apparatus (Matthies et al., 1996). Additionally, in N. vitripennis embryos in which the centrosomes have been inactivated by the male-killing bacterium Arsenophonus nasoniae, the chromosomes alone are capable of nucleating rudimentary spindles (Ferree et al., 2008). An interesting question is whether the PCM, along with the centrosomes, can support the formation of its half of the gonomeric spindle. Future attempts to answer this question will require careful analyses of the spindle apparatus in young, PSR-positive wasp embryos.

We found that PSR localizes to the outer surface of the paternal nucleus in all young embryos examined. This pattern is likely to be highly important for the successful transmission of PSR. For example, an alternative finding, such as the presence of PSR within the center of the paternal nucleus, would likely result in the entanglement of PSR with the unresolved paternal chromatin. We never observed PSR to remain with the PCM following anaphase of the first division, demonstrating the proficiency of this chromosome in escaping the PCM. The localization of PSR to the nuclear periphery likely stems from the general tendency of certain centric and pericentric regions to associate with the inner nuclear envelope (Rae and Franke, 1972; Marshall et al., 1996; King et al., 2008). In support of this idea, the pericentric region of the autosome containing the rDNA locus also was found to associate with the nuclear periphery in N. vitripennis. A large portion of the euchromatic arms of the autosomes likely extends into the interior of the nucleus where the entangled chromatin mass forms. PSR, in contrast, contains little or no euchromatin. Thus, its entire contents reside at the nuclear periphery where it can resolve freely from the paternal chromatin.

The fact that a single PSR chromosome resides within the nucleus of each germ cell and mature sperm demonstrates that PSR stably segregates during cell division in the testis. These observations are consistent with the idea that PSR transmission does not involve meiotic or post-meiotic drive mechanisms. The fact that PSR stably segregates in the male germ line stands in contrast to many other B chromosomes, which are highly unstable and prone to loss during cell division (Jones, 1991). Some of these B chromosomes have evolved ways of amplifying or segregating asymmetrically, including directed non-disjunction, in order to counter their instability (Jones, 1991). A more universal problem of B chromosomes is that, unlike the normal chromosomes, they lack pairing partners during meiosis. PSR may be transmitted stably in the male germ line because meiosis in this sex occurs as a modified mitotic division due to the haploid state of males (Whiting, 1968). This idea predicts that PSR should not be stably transmitted during female meiosis, which involves homolog pairing (Beukeboom and Werren, 1993b). Support for this prediction stems from the unstable nature of several PSR derivatives produced from genome fragmentation or mutagenesis (Beukeboom and Werren, 1993a; Perrot-Minnot and Werren, 2001). These chromosomes were successfully obtained in females due to their reduced ability to destroy the paternal chromatin. However, the instability of these chromosomes to persist in the female germ line may have been caused either by damage incurred to the centromeres or other chromosomal regions that are important for their stability or by a lack of a pairing partner in this tissue.

One of our most striking findings is that PSR localizes at the anterior tip of the elongated sperm nucleus in all analyzed sperm. This pattern is in contrast to the normal chromosomes, which localize broadly across the nucleus. An important question stemming from this observation is how PSR is able to achieve such a high level of localization to the anterior tip of the sperm nucleus. One possibility is that this region is where any extra chromatin that arises in the genome is placed. An alternative and more intriguing possibility is that PSR localizes to the sperm nucleus tip in an active manner. How might this occur?

Sperm nuclear elongation likely involves the coordination of extra-nuclear events, such as formation of a microtubule bundle on one side of the nucleus that generates the elongating force (Fawcett et al., 1971), with intra-nuclear events that are less clearly understood. Early studies in birds documented the presence of dense foci of DNA, presumed to be regions of heterochromatin, at the position of elongation initiation in the nuclei of post-meiotic spermatids (Dressler and Schmid, 1976). In mice, foci of heterochromatin were observed in a position beneath the nuclear envelope that eventually becomes the anterior tip of the sperm nucleus (Czaker, 1987). Although the mechanisms are unknown, these observations suggest that heterochromatin plays a role in determining the initial site of sperm nucleus elongation. In light of this possibility, we propose that PSR, being largely heterochromatic, may seed nucleus elongation preferentially over the pericentric regions of the autosomes.

A second important question is whether there is any functional relevance of PSR localization to the anterior tip of the sperm nucleus. A similar pattern of B chromosome localization also was discovered in the sperm of maize (Rusche et al., 1997). It was proposed that the localization of a B chromosome to the sperm nucleus tip confers a fertilization advantage to the subset of sperm that carry this chromosome (Rusche et al., 1997), although no compelling mechanistic explanations were offered (Jones et al., 2008). In N. vitripennis, all sperm from PSR-positive males contain a single copy of this B chromosome, arguing against the need for a fertilization advantage in this organism. Alternatively, it is possible that placement at the tip of the sperm nucleus may confer some epigenetic advantage to the PSR chromosome. Such a hypothesis has been proposed for the sex chromosomes, which localize invariably to discrete regions within the sperm of several mammals (Greaves et al., 2003; Foster et al., 2005).

The paternal transmission of PSR is possible due to the haplo-diploid reproduction of N. vitripennis. Because males normally arise as haploids from unfertilized eggs, the destruction of the paternal half of the genome by PSR successfully converts what should become diploid females into transmitting haploid males. It has been suggested that haplo-diploid organisms in general may be prone to sex ratio distortion by similar, paternally transmitted B chromosomes (Werren and Stouthamer, 2003). This hypothesis is supported by the previous discovery of another paternally transmitted B chromosome in a different hymenopteran insect, Trichogramma kaykai (Stouthamer et al., 2001). It is highly likely that this B chromosome in T. kaykai and PSR in N.vitripennis arose through independent evolutionary events due to their large sequence differences (Eickbush et al., 1992; McAllister, 1995; van Vugt et al., 2005) and the distant relatedness of these two insect species. An intriguing question is whether these B chromosomes are transmitted through similar mechanisms. This idea appears to be supported at the morphological level – the B chromosome in T. kaykai causes the paternal genome to undergo hyper-condensation and chromosome resolution failure in a manner that is very similar to PCM formation in N. vitripennis (van Vugt et al., 2003). Further studies in N. vitripennis and T. kaykai will be important for exploring if PCM formation and B chromosome transmission involve similar chromatin-based mechanisms, and more broadly, how such systems of genome conflict can evolve at the molecular level.

Experimental crosses involved AsymC (wild-type, Wolbachia-uninfected) virgin females with either AsymC males or AsymC males carrying PSR (AsymC-PSR). Wasps were allowed to mate en mass overnight with 50% honey in water. Subsequently, mated females were separated and individually allowed to deposit embryos into fresh blowfly hosts.

Embryos were removed from host pupae and fixed in a solution of 5 ml heptane, 1.5 ml 1× PTX (1×phosphate-buffered saline with 0.5% Triton X-100) and 600 ml 37% formaldehyde (final concentration of 4% in the aqueous layer) for 26 minutes. Fixed embryos were removed, placed on a small piece of Whatman paper, allowed to air-dry for ∼1 minute, and adhered to double-sided adhesive tape on the surface of a 22 mm Petri plate. Approximately 2 ml of 1× PTX was immediately placed into the dish to rehydrate the embryos. A 28-gauge hypodermic needle was used to remove embryos from their vitelline membranes. Embryos were transferred to a 0.6 ml microfuge tube, washed three times with 1× PTX and treated with RnaseA for 1–2 hours at 37°C before fluorescence in situ hybridization or antibody staining.

The following sequences were used for fluorescence in situ hybridization probes: 5′-TCAAAAGTCTTGACTTTGGCTTACACGCTT-3′ and 5′-TATAATCAAAAGTCTCGACTTATGATTGGA-3′ are regions of two unique satellite DNA repeats on PSR (Eickbush et al., 1992); 5′-TTTTTAAGACGTTTTATTGTTACTGGTAGT-3′ is a region of a satellite DNA repeat that is found on a single autosome (Eickbush et al., 1992); and 5′-GTATGTGTTCATATGATTTTGGCAATTATA-3′ is a 240-bp spacer repeat of the rDNA locus (Stage and Eickbush, 2010). Additionally, three sequences corresponding to regions of the 28S rDNA gene also were used for probes: 5′-TGGTTGCAAAGCTGAAACTTAAAGGAATT-3′, 5′-ATTTAGAGGAAGTAAAAGTCGTAACAAGG-3′, 5′-ATGTTTTCATTAATCAAGAACGAAAGTTAG-3′. All of these sequences were chemically synthesized (IDT, Inc., USA) and 5′-end-labeled with either Cy3 or Cy5 for detection. Mapping the hybridization sites of the synthesized probes was performed on meiotic chromosomes from testis tissues. Whole testes were dissected in 1× PBS and fixed in 4% paraformaldehyde and 50% acetic acid for 4 minutes. Immediately following fixation, testes were squashed, immersed into liquid nitrogen for 5 minutes, dehydrated in 100% ethanol for 10 minutes, and air-dried for at least 1 hour. 125 ng of each probe was diluted in 20 ml of 1.1× hybridization buffer, which was placed onto the tissue and slide and covered with a coverslip. Samples were incubated overnight at 30°C, washed three times for 20 minutes each wash in 0.1× SSC buffer, and mounted in Vectashield mounting medium with DAPI (Vector Labs, Inc., USA). Probe hybridizations to embryos were carried out as described in (Ferree and Barbash, 2009).

Primary antibodies were incubated with fixed embryos overnight at 4°C for the following dilutions: rabbit anti-PH3 (Santa Cruz Biotech, USA) at 1∶200; rabbit anti-PCNA (a gift from Paul Fisher, Stony Brook University) at 1∶500; rabbit anti-SMC2 (a gift from Margarete Heck, University of Edinburgh) at 1∶200. TUNEL reactions were performed on whole embryos using the In Situ Cell Death Detection kit, TMR red (Roche, Inc., USA) according to the manufacturer's instructions for tissues in liquid medium.

Images of chromosomes were captured on an Olympus IX81 epifluorescence microscope and ImagePro 6.3 imaging software. Images of whole mount tissues (testes and embryos) were taken with a Zeiss DMIRB confocal microscope. Constant gain and pinhole settings were used to collect all images within a given experiment. All images were subsequently processed with Adobe Photoshop CS5 version 12.

We thank John Werren for the wasp lines, Margarete Heck for the anti-SMC2 antibody, and Paul Fischer for the anti-PCNA antibody. We also thank John Werren, Richard Stouthamer, and Emily Wiley for helpful discussions. Finally, we thank the Harvey Mudd Biology Department for the use of their confocal microscope.