Giant meiotic spindles in males from Drosophila species with giant sperm tails

The spindle is a microtubule (MT)-based machine that mediates accurate chromosome segregation during both mitosis and meiosis (reviewed in Dumont and Mitchison, 2009; O'Connell and Khodjakov, 2007). Although the spindle shows little size variation between cells of the same type, spindle length can vary dramatically both within and between species (reviewed in Goshima and Scholey, 2010). It has been proposed that spindle length is affected by both extrinsic and intrinsic factors. Well-known extrinsic factors are cell size and tubulin availability (Dumont and Mitchison, 2009; Goshima and Scholey, 2010). For example, during Xenopus laevis embryonic development, cell size decreases from 1200 to 12 μm, and in cells smaller than 300 μm, spindle length scales linearly with cell size (Wuhr et al., 2008). This suggests that constraints imposed by cell size and, presumably, tubulin availability set the spindle length. However, in the largest X. laevis blastomeres spindle length has an upper limit of 60 μm, implying the existence of intrinsic factors that govern spindle size (Wuhr et al., 2008). Further evidence for spindle size regulation by intrinsic factors is provided by the observation that spindles assembled in vitro from embryonic extracts of related Xenopus species display different species-specific sizes (Brown et al., 2007). Another example of spindle size regulation by intrinsic mechanisms is found in the 500 μm-long Drosophila melanogaster syncytial embryo, which displays spindle lengths of 10 μm throughout the first 14 rounds of embryonic divisions (Kwon and Scholey, 2004; Sullivan and Theurkauf, 1995).

Studies carried out in the past few years have identified several cell-size-independent intrinsic factors that regulate spindle length. These factors include proteins that mediate MT polymerization or depolymerization, MT sliding and MT clustering (Dumont and Mitchison, 2009; Goshima and Scholey, 2010). However, perturbations of these factors result in relatively small changes in spindle length (Goshima and Scholey, 2010), suggesting that intrinsic factors are not sufficient to account for the large variations in spindle size observed in animal cells.

Although it is clear that spindle size depends on both intrinsic and extrinsic factors, the reasons for spindle size variability are largely unknown. Here we show that variations in spindle size can be linked to a precise developmental requirement. We found that meiotic spindles from males of different Drosophila species have extremely different sizes, and that the size of these spindles positively correlates with the length of the sperm tail. This suggests that Drosophila primary spermatocytes manufacture and store most of the tubulin needed for sperm tail formation and use it for meiotic spindle assembly.

Although male meiotic spindles and somatic cell spindles of mammals are comparable in size (Manandhar et al., 2000), spermatocyte spindles of D. melanogaster are much larger than those of mitotic cells (Cenci et al., 1994). In D. melanogaster primary spermatocytes, metaphase chromosomes appear as a small mass at the center of a large bipolar MT assembly (Fig. 1). By contrast, in D. melanogaster embryonic cells, larval neuroblasts, ganglion mother cells and female meiotic cells, metaphase chromosomes occupy the entire equatorial region of the spindle or even protrude from the spindle (Bonaccorsi et al., 2000; Kwon and Scholey, 2004; Sullivan and Theurkauf, 1995; Theurkauf and Hawley, 1992) (Fig. 2). These differences in the basic shape of male meiotic spindles with respect to those in other fly cells are intriguing because there is no evidence that male meiotic spindles mediate chromosome segregation by any but normal mechanisms.

What is then the reason for the large size of the meiotic spindle in D. melanogaster males? We hypothesized that Drosophila spermatocytes accumulate a substantial fraction of the tubulin needed for sperm tail assembly and use it for meiotic spindle formation. This hypothesis is testable, because various Drosophila species have dramatically different sperm flagella. D. persimilis, D. pseudoobscura, D. melanogaster, D. nannoptera and D. hydei sperm tails are 0.32, 0.36, 1.91, 15.74 and 23.32 mm long, respectively; the D. bifurca flagellum is more than 30-fold longer than its D. melanogaster counterpart, with the extraordinary length of 58.29 mm (Pitnick et al., 1995a; Pitnick et al., 1995b). We examined male meiosis in these six Drosophila species to determine whether a correlation exists between sperm length and the size of the meiotic spindle. Dividing spermatocytes obtained from pupae were squashed under mild pressure and then stained for tubulin and DNA. To assess the spindle size, we measured both the length and the area of the spindle (see supplementary material Fig. S1). We did not measure the fluorescence of the spindle MTs to estimate the amount of tubulin, because the commercial anti-tubulin antibody we used (raised against chicken α-tubulin) might react to the tubulins of distant Drosophila species with varying efficiencies. The results of our analysis unequivocally showed that the six Drosophila species differ dramatically in the length and the area of the spindles of the first meiotic division (Fig. 1 and supplementary material Table S1). Similar differences in size were observed in spindles of the second meiotic division (supplementary material Fig. S2 and Table S1). In all cases, we found a highly significant positive correlation between the spindle size of the meiotic figures and the length of the sperm flagellum (Fig. 1 and supplementary material Fig. S2). These observations support the hypothesis that primary spermatocytes of Drosophila species store amounts of tubulin that are proportional to the sperm tail length, and use a substantial part of this tubulin to assemble the meiotic spindle.

Spermatocytes result from four rounds of mitotic divisions of a single founding gonial cell. The 16 postmitotic spermatocytes enter a 90-hour growth phase during which undergo extensive transcription and translation accompanied by a 25-fold increase in volume (reviewed in Fuller, 1993; Schafer et al., 1995; White-Cooper, 2010). Before completion of this growth phase, spermatocytes degrade the β1-tubulin isotype and switch to the testis-specific β2 isoform. Following this switch, all MT functions are carried out by a single tubulin heterodimer composed of the α-84B and β2-tubulin isoforms (Hutchens et al., 1997; Kemphues et al., 1982). Collectively, these results indicate that spermatocytes are enriched in tubulin and other proteins needed for spermiogenesis, and that the same tubulin heterodimer is used for both meiotic spindle and sperm tail formation.

D. melanogaster male meiosis produces four spermatids, each comprising a nucleus and a mitochondrial derivative called the nebenkern. Elongating spermatids contain two classes of MTs: the MTs that comprise the axoneme, which is dispensable for sperm elongation, and the cytoplasmic MTs that drive sperm tail growth (Basto et al., 2006; Noguchi et al., 2011). The cytoplasmic MTs interact with the mitochondrial derivatives and mediate the extension of these structures, which are thought to serve as cell shape templates during spermatid elongation (Noguchi et al., 2011). Transcription is strongly reduced in D. melanogaster spermatids and only a few genes appear to be transcriptionally active (Barreau et al., 2008; Vibranovski et al., 2010; White-Cooper, 2010). Spermatids inherit many stabilized mRNAs from spermatocytes, including mRNAs that encode β2-tubulin, and undergo protein synthesis until late in spermiogenesis (reviewed in Fuller, 1993; White-Cooper, 2010). However, electron microscopy studies have shown that spermatids and mature sperm have a similar volume (Lindsley and Tokuyasu, 1980; Tokuyasu, 1975). This suggests that spermatocytes deliver to spermatids a substantial fraction of the material needed for sperm differentiation. It should be noted that tubulin and mitochondria delivery are closely related events. During meiotic telophase I and II, mitochondria associate with the central spindle and are equally partitioned between the two daughter cells upon execution of cytokinesis (Cenci et al., 1994; Fuller, 1993). Thus, a spindle proportionate to sperm length ensures proper transmission to the spermatids of the two major components of the sperm tail: the tubulin and the mitochondria.

To control for the possibility that the differences in meiotic spindle size could be governed by species-specific factors other than the amount of tubulin needed to form the sperm tail, we examined mitotic division in larval brain squashes from all six Drosophila species. We observed that the spindle length in larval neuroblasts does not vary significantly between species, and that the mitotic spindle size does not correlate with the sperm length (Fig. 2). We also noted that the large differences in meiotic spindles cannot be explained by obvious differences in chromosomal complements. Although these species exhibit different karyotypes, the number of chromosomes varies only between 8 and 12 (Fig. 3), and the diploid cells of all six species contain comparable amounts of DNA (Bosco et al., 2007; see also supplementary material Table S2). We thus conclude that there are no major species-specific factors that influence mitotic spindle size, making it unlikely that meiotic spindle size is regulated by factors other than the amount of tubulin accumulated in spermatocytes.

The metaphase spindle of D. bifurca primary spermatocytes with its 63 μm of length is probably the largest spindle described to date in any organism (Goshima and Scholey, 2010). The meiotic spindles of D. hydei and D. nannoptera are also very large. Our results strongly suggest that the main extrinsic factor responsible for the large size of these spindles is tubulin availability. However, we believe that the assembly of such large spindles requires a number of intrinsic factors. One of these factors could be the centriole and/or centrosome size. In C. elegans embryos, the size of the centrioles and centrosomes correlates with spindle length (Greenan et al., 2010). Centriole and spindle lengths also appear to be correlated in different D. melanogaster cell types; the centrioles of embryonic cells, tissue culture cells and spermatocytes are 0.2, 0.2 and 2.5 μm long, respectively (Gonzalez et al., 1998; Baker et al., 2004), and the metaphase spindle lengths in the same cells are 11.8, 10.0 and 23.0 μm, respectively (Goshima and Scholey, 2010). We were unfortunately unable to test whether this correlation also held for male meiotic cells of the Drosophila species examined, because we found that antibodies raised against various D. melanogaster centrosome components failed to react either totally or partially with the centrioles of some species (we tested antibodies directed against Asterless, γ-tubulin, DSpd2 and centrosomin). However, our cytological observations indicated that species with long sperm tails and large meiotic spindles also have large asters (Fig. 1 and supplementary material Fig. S2). To assess the relationships between aster and spindle size, we measured the radius of the asters of late prophase–early prometaphase primary spermatocytes; this radius positively correlated with the area of the metaphase spindle (Fig. 4), suggesting that the centrioles and centrosomes of large spindles have a higher nucleating ability than those of small spindles. Consistent with this finding, EM studies have shown that the centrioles of D. hydei spermatocytes are approximately threefold longer than their D. melanogaster counterparts (Hennig and Kremer, 1990).

A positive correlation between the centrosome nucleating ability and spindle size implies a major role of the centrosomes in male meiotic spindle formation in Drosophilidae. This view is supported by previous studies in D. melanogaster that indicated that centrosome-nucleated MTs are crucial for spindle assembly in spermatocytes but not in somatic cells. Mutants in which centrosome function is inhibited form anastral but otherwise functional mitotic spindles by exploiting MTs nucleated near the chromosomes (Basto et al., 2006; Bonaccorsi et al., 2000; Giansanti et al., 2008; Megraw et al., 2001). However, although the same mutants nucleate MTs around the meiotic chromosomes, their male meiotic spindles are highly defective, leading to male sterility (Bonaccorsi et al., 1998; Giansanti et al., 2008). Consistent with these findings, D. melanogaster secondary spermatocytes have the peculiar ability to assemble a spindle in the complete absence of chromosomes, exploiting only MTs nucleated by the centrosomes (Bucciarelli et al., 2003). Collectively, these results support the hypothesis that the centrosome is an important intrinsic factor in spindle size determination in Drosophila spermatocytes.

The reason why some Drosophila species evolved giant sperm tails is still matter of debate. The most popular current hypothesis asserts that the gametes and reproductive structures of males and females coevolved in response to sexual selection (Bjork and Pitnick, 2006; Joly and Schiffer, 2010; Miller and Pitnick, 2002). Our results underscore the strength of the selective forces that drove sperm length increase during Drosophila evolution. Formation of the giant sperm tails was not simply achieved through an increase in postmeiotic translation of sperm proteins, but involved the evolution of giant spindles that utilize a substantial fraction of the tubulin that will be used after meiosis for sperm tail differentiation. The assembly of such spindles is likely to result from a series of concerted evolutionary changes in the regulation of spindle-associated proteins and in centrosome structure and function.

The D. melanogaster wild-type stock used in this study was an Oregon R strain. The other species were all obtained from the Drosophila Species Stock Center of the University of California, San Diego (https://stockcenter.ucsd.edu). D. melanogaster, D. persimilis, D. pseudoobscura and D. hydei were reared at 25°C on standard yeast–sucrose–agar medium. D. nannoptera and D. bifurca were grown on banana medium supplemented with opuntia cactus powder; the recipe for the medium can be found at https://stockcenter.ucsd.edu.

DAPI-stained metaphases and larval neuroblast spindles were prepared according Gatti et al. and Bonaccorsi et al., respectively (Gatti et al., 1994; Bonaccorsi et al., 2000). Meiotic spindle preparations were obtained from pupal testes fixed and processed according to published methods (Cenci et al., 1994). Brain and testis preparations were incubated overnight at 4°C with a monoclonal anti-α-tubulin antibody (diluted 1:1000 in PBS; Sigma-Aldrich), which was detected by 1 hour of incubation at room temperature with fluorescein isothiocyanate (FITC)-conjugated anti-mouse IgG+IgM (diluted 1:20 in PBS; Jackson Laboratories). Immunostained preparations were mounted in Vectashield medium H-1200 (Vector Laboratories) containing the DNA dye DAPI, and examined with a Zeiss Axioplan fluorescence microscope equipped with a CCD camera (Photometrics CoolSnap HQ).

Spindle and aster measurements were carried out on digital images at defined enlargement. The spindle length is the distance between the two spindle poles, as indicated in supplementary material Fig. S1. To calculate the spindle area we used the MetaMorph program (Meta Imaging Series Software from Molecular Devices). We outlined the spindle perimeter as shown in supplementary material Fig. S1 and the program calculated the spindle area.

Regression analysis was performed using the GraphPad software. The r² value measures the goodness-of-fit of linear regression; the P value evaluates the probability of the null hypothesis.

We thank Mike Goldberg for comments on the manuscript and the Drosophila Species Stock Center (UCSD) for fly strains.