Klp10A modulates the localization of centriole-associated proteins during Drosophila male gametogenesis

ABSTRACT

Mutations in Klp10A, a microtubule-depolymerising Kinesin-13, lead to overly long centrioles in Drosophila male germ cells. We demonstrated that the loss of Klp10A modifies the distribution of typical proteins involved in centriole assembly and function. In the absence of Klp10A the distribution of Drosophila pericentrin-like protein (Dplp), Sas-4 and Sak/Plk4 that are restricted in control testes to the proximal end of the centriole increase along the centriole length. Remarkably, the cartwheel is lacking or it appears abnormal in mutant centrioles, suggesting that this structure may spatially delimit protein localization. Moreover, the parent centrioles that in control cells have the same dimensions grow at different rates in mutant testes with the mother centrioles longer than the daughters. Daughter centrioles have often an ectopic position with respect to the proximal end of the mothers and failed to recruit Dplp.

KEYWORDS:

• centriole-associated proteins
• centrioles
• male gametogenesis
• Drosophila

Introduction

The centrioles are cylindrical structures formed by singlet, doublets or triplets microtubules usually arranged in a beautiful ninefold symmetry.^1,2 These organelles are instrumental to the assembly of a functional centrosome since they play crucial roles to spatially organize the proteins involved in the recruitment of γ-TuRC complexes need to nucleate the microtubule network during interphase and mitosis.^3-5 Therefore, decipher the centriole structure and behavior during the cell cycle is crucial to understand the centrosome function and dynamics.

The knowledge of the molecular mechanisms of centriole assembly and centriole-to-centrosome conversion have greatly increase in the last years,^5-9 and super-resolution immunoflorescence microscopy techniques greatly help us to decipher the localization of the main centriole-associated proteins involved in centriole organization and PCM recruitment in both Drosophila and human somatic cells.^10-13

The somatic centrioles are useful models to look at the dynamics of the centriole-associated proteins during progression through the cell cycle and centrosome maturation.⁵ However, they are too short to appreciate in detail eventual differences of protein localizations along their length. To circumvent these issues we focused our attention on centriole-associated proteins during Drosophila male gametogenesis where the centriole length increases significantly.

The centrioles of the germ line stem cells and spermatogonia in the Drosophila testis look like centrioles of somatic cells in that they are short and duplicate once during each cell cycle in concert with DNA replication. However, germ line centrioles in Drosophila are composed by triplet microtubules whereas the somatic centrioles have doublets.^14,15 Interestingly, all centrioles in the Drosophila tissues retain their cartwheel. By contrast, the cartwheel is only transiently present within the daughter centrioles of vertebrate cells and it is lost during daughter-mother transition at the onset of mitosis.¹⁶ Moreover, there is a clear structural asymmetry between the parent centrioles of germline stem cell. The mother one that is localized close to the hub region is composed of triplets, whereas the daughter displays mixed doublets and triplets and will acquire the full triplet number later.¹⁵

Early spermatocytes inherit at the end of the fourth spermatogonial mitosis 2 orthogonally arranged centrioles that duplicate and move to the cell periphery to organize distinct cilium-like projections that persist through the meiotic divisions.^17-19 Thus, each primary spermatocyte displays 4 ciliary processes suggesting that all the centrioles have the same competence.²⁰ This is in contrast with the prevailing view that only the mother centriole is able to nucleate a ciliary axoneme that is disassembled in vertebrate cells at the end of the interphase.^21,22

During the first prophase the cilium-like regions grow by an unidentified mechanism independent by the intraflagellar transport and the centrioles elongate concurrently to reach a 10-fold length at the onset of prometaphase.²³ This aspect contrasts with the findings that the basal bodies of ciliated cells do not increase dimensions during ciliogenesis. Since the distal region of the spermatocyte centrioles is engaged in the assembly of the ciliary axoneme it is still unclear how the centriole may elongate. Therefore, the process driving centriole growth in Drosophila spermatocytes does not simply relies on the molecular pathways involved in centriole assembly and elongation in other organisms.²⁴

Klp10A, a microtubule-depolymerising Kinesin-13, involved in microtubule dynamics during both interphase and mitosis,^25-27 seems to play a major role in the dynamics of the Drosophila centriole. Loss of Klp10A leads, indeed, to uncontrolled centriole length and significant instability in germ line and somatic Drosophila cells.^28,29 However, it is still unclear how this protein works to control centriole length. To better understand the role of Klp10A on centriole structure and add insights into the mechanism of PCM recruitment we asked whether the localization of Dplp and Spd2, 2 centriole associated proteins critical for PCM organization,^3,6,30,31 and Sak-Plk4 and Sas-4, essential to centriole duplication and assembly,^24,32-34 may be affected in overly long centrioles of mutant cells.

We show here that Klp10A, beyond its established role in centriole growth during the spermatogonial divisions and in young primary spermatocytes, is differently required by parent centrioles to achieve their full length. We found, indeed, that mother centrioles are longer than daughters after Klp10a loss. Moreover, we also observe that Dplp, Sas-4 and Sak/Plk4 have a broad localization along the length of the mutant centrioles, whereas in control germline cells these proteins are restricted to the proximal regions of the centrioles.

Results

Parent centrioles have different length in Klp10A mutant testes

The Drosohila centrioles are short in stem cells and spermatogonia (Fig. 1A) and elongate in young primary spermatocytes (Fig. 1B) to reach their full length at the end of the first prophase (Fig. 1C). Loss of Klp10A leads to uncontrolled centriole elongation and significant centriole instability. Centrioles were unusually long during the last spermatogonial division (Fig. 1D) and in young primary spermatocytes (Fig. 1E) and often displayed growing microtubule bundles at their distal ends (Fig. 1D) as previously observed in Klp10A wing disc cells.²⁹ Unlike control tests in which the parent centrioles have the same size, the mother centrioles in mutants were more elongated than daughters (Fig. 1E). Moreover, the position of the daughter centrioles respect to the mother ones was unusual: in control testes the procentrioles always formed at the proximal end of the mothers, whereas in mutant testes the daughter centrioles were displaced away from the proximal end of the mothers (Fig. 1E). The centrioles that reached the cell surface assembled abnormal and short cilium-like projections (Fig. 1F).

Figure 1. Klp10A loss leads to overly long centrioles. Centrioles in control (A–C) and mutant (D–F) testes. (A–C) Centrioles elongate gradually in control testes and reach the full length at the end of the first prophase when their distal end organizes a cilium-like projection. Mother (B, arrow) and daughter (B, arrowheads) centrioles have the same length in control germ cells. Mutant centrioles are unusually long with blurred distal ends that often prolonged in single or doublet tubules (D,E, arrowheads); the daughter centrioles (E, arrowhead) are shorter than the mothers (E, arrow) and are displaced from the proximal end of the mothers. Scale bars: 250 nm.

Klp10A mutation affects the localization of Dplp

To analyze the centriole-associated protein dynamics during male gametogenesis we firstly look at the localization of Dplp that in cultured S2 cells is tightly associated with the centriole wall where it organizes a fibrillar network to scaffold centrosomal proteins around the centriole.¹² To better understand the distribution of Dplp we counterstained control and mutant testes with an antibody against the PCM protein Spd-2 that recognized the whole Drosophila centrioles. In control testes, Dplp localized on the whole centrioles of stem cells and spermatogonia where it formed a distinct ring around Spd-2 (Fig. 2A, B). In late prophase spermatocytes the Dplp staining was restricted at the basis of the giant centrioles (Fig. 2B).^10,19

Figure 2. The recruitment of Dplp increases in Klp10A mutant testes. Low magnification of the hub region and early primary spermatocytes in control (A) and mutant (C) testes; centrioles are dot-like in control germline stem cells and early primary spermatocytes, whereas they are very elongated in mutant germ cells (Spd-2, green; Dplp, red; DNA, blue; GSCs, germline stem cells). Representative centrioles in control (B) and mutant (D) male germ cells at indicated stages of development were stained to reveal Spd-2 (green) and Dplp (red). Dplp forms in control testes a short cylinder (B, arrowheads) that elongates in mutant centrioles (D, arrows) and surrounds Spd-2. The paired signals found in mature primary spermatocytes correspond to V-shaped pairs of centrioles. Mother centrioles (D, m) are longer than daughters (D, d) and stained for Spd-2 and Dplp, whereas during the earlier stages of spermatogenesis the daughters often stained only for Spd-2. Acetylated-tubulin labeling (green) reveals distinct cilium-like projections in control mature primary spermatocytes (E, arrowheads) but not during the earlier phases of spermatogenesis in mutant testes; centrioles are stained with Spd-2 (red); GSCs, germline stem cells.Scale bars: A,C = 5 µm; B,D,E,F = 1 µm.

The more dramatic effects of Klp10A loss were observed during the early stages of male gametogenesis. Unusually long centrioles were, indeed, found in germline stem cells, in 16-cell cysts of spermatogonia and in young primary spermatocytes (Fig. 2C,D). The localization of Dplp was highly perturbed and acquired an expanded distribution along the centriole length (Fig. 2D).

To exclude the possibility that the overly long centrioles found in germline stem cells and spermatogonia may be the result of the premature assembly of cilium-like projections, we counterstained mutant testes with antibodies against acetylated α-tubulin that recognize the stable microtubules within the ciliary projections of Drosophila spermatocytes (Fig. 2E).^19,35 However, this antibody failed to recognize such structures at the distal ends of the elongated centrioles in mutant germ cells (Fig. 2F).

Both electron microscopy (EM) and immunofluorescence analysis showed that the parent centrioles had the same length in control testes (Figs. 1B and 2B). By contrast, in mutant germ cells most of the daughter centrioles examined (201 out of 236) were shorter than mothers (Figs. 1E and 2D). This was dramatically evident in germ line stem cells where we find overly long mother centrioles (48 out 97) close to the hub region, whereas their daughters always appeared as small spots (Fig. 2C, D). We also find that the daughter centrioles of mutant 16-cell cyst spermatogonia and young primary spermatocytes in some cases (39 out 85) lack Dplp signal whereas the mother centrioles stained for both Dplp and Spd-2 (Fig. 2D). This asymmetric localization was not longer seen in mature primary spermatocytes where both parents display a distinct Dplp signal (Fig. 2D).

Sas-4 is not required to maintain Dplp and Spd-2 at the centriole during meiosis

In fly somatic cells the PACT domain of Dplp is close to the centriole wall, where Sas-4 links the cartwheel to the peripheral microtubules promoting their elongation during centriole assembly.^12,36-38 We then asked whether there was a correlation between Dplp and Sas-4. Sas-4 co-localized with Dplp and Spd-2 on the wall of the centrioles in somatic and in male germline stem cells (Fig. 3A, B).¹⁰ The distribution of Sas-4 did not change significantly in mature primary prophase spermatocytes despite the centrioles had elongated 10 times (Fig. 3A, B). Sas-4 was restricted to the proximal region of the giant centrioles (Fig. 3A) in correspondence of the Dplp staining (Fig. 3A). However, although, Dplp and Sas-4 signals overlap at the basis of the centrioles, their specific localization differs with Dplp forming a thick cylinder around Sas-4 (Fig. 3A). All the centrioles of the 16-cell cyst spermatogonia displayed a distinct Sas-4 signal, whereas one centriole of the pairs often lacks Spd2 and Dplp (Fig. 3A, B).

Figure 3. Sas-4 distribution in Klp10A depleted testes. Centrioles in control (A,B) and mutant (C,D) male germ cells at different stages of development were stained to reveal Sas-4 (red), Dplp (green, left column) and Spd-2 (green, right column). (A,B) Dplp and Spd-2 co-localize with Sas-4 during the spermatogonial mitoses although one of the just duplicated centrioles within each pair often lacks Dplp and Spd-2 signals. In mature primary spermatocytes the Sas-4 signal co-localizes with Dplp (A) and is found at the proximal end of the centrioles fully stained by the anti-Spd-2 antibody (B). In Klp10A mutant testes the Sas-4 localization extends over Dplp (C) and Spd-2 (D) signals in 16-cell cyst spermatogonia. (C,D) In mature primary spermatocytes the Sas-4 signal shrinks. Scale bars: 1 µm.

We reasoned that if there was interdependence between the localization of Sas-4 and Dplp, then an increase of Sas-4 signal in Klp10A mutants must correspond to similar changes in Dplp distribution. We find, indeed, that the centrioles and centriole fragments were strongly stained by both anti-Sas-4 and anti-Dplp antibodies in 16-cell cyst spermatogonia and in young primary spermatocytes (Fig. 3C). However, Sas-4 was present along the entire length of the abnormal centrioles in these earlier stages, whereas the Dplp signal did not exceed their proximal half (Fig. 3C). The localization of Sas-4 during the last spermatogonial mitoses and in early primary spermatocytes also extended beyond the region occupied by Spd-2 along the abnormal centrioles (Fig. 3D). However, the intensity of the Sas-4 signal in mutant testes decreased as prophase progressed and became restricted to the proximal region of the centrioles in mature spermatocytes (Fig. 3C, D). By contrast, Dplp and Spd-2 maintained a broad localization.

The proximal region of the centriole is a distinct domain

The unusual position of the daughter centrioles placed away from the proximal ends of the mothers in mutant testes (Fig. 1E) prompted us to verify if Klp10A loss may also affect the localization of Plk4 (Polo-like kinase 4) that plays a main role in the assembly of the procentrioles and in the sequential recruitment of some centriole-associated proteins.^32,33,39,40 Plk4 colocalized in control testes with Dplp at the proximal end of the centrioles (Fig. 4A) where also Sas-4 was found. Plk4 was not restricted to the proximal region of the centrioles in 16-cell cysts of the spermatogonia, and in early and mature primary prophase spermatocytes but extended beyond this region inside an elongated Dplp cylinder (Fig. 4B). Remarkably, Plk4, Dplp and Sas-4 signals roughly matched in control testes the cartwheel position at the basis of the centriole. By contrast, the centrioles examined in Klp10A mutant spermatogonia and young primary spermatocytes rarely had a distinct cartwheel (Fig. 4D; 9 out of 67) as that found in control cells (Fig. 4C). Rather, mutant centrioles lack the cartwheel (Fig. 4D; 25 out of 67) or it was abnormal, with indistinct hub and incomplete spokes (Fig. 4D; 33 out of 67). Plk4, Dplp and Sas-4 were also localized on the whole centriole fragments that lack cartwheel components frequently found in young primary spermatocytes.

Figure 4. The proximal region of the centriole. Centrioles in control (A) and mutant (B) male germ cells were stained to reveal Sak/Plk4 (red) and Dplp (green). Sak/Plk4 and Dplp co-localize in control testes to the short spermatogonial centrioles or at the proximal end of the giant primary spermatocyte centrioles (A), but expand along the abnormal centrioles in mutant testes (B). Cross sections through the proximal end of control (C) and mutant (D) centrioles: a distinct cartwheel is visible in control centrioles, whereas it is rarely found in mutant centrioles that often lack a cartwheel or display blurred cartwheels. Cartoons depicting the architecture of the basal region of control (C) and mutant (D) centrioles. Scale bars: A,B = 1 µm; C,D = 100 nm.

Discussion

The centrosome in vertebrate cells contains one pair of centrioles with distinctive morphologies and functions.⁴¹ The mother centriole displays characteristic distal appendages involved in microtubule binding and assembles the primary cilium.⁴² By contrast, the parent centrioles in Drosophila lack a distinct structural dimorphism and both nucleate a cilium-like projection although they maintain the cartwheel that represents a hallmark of immaturity.^23,43 Thus, they may be only recognized by their reciprocal position with the daughter centriole orthogonal to the proximal end of the mother.

The parent centrioles have a different length in the absence of Klp10A in 16-cell cysts of the spermatogonia and in young primary spermatocytes. The mother centrioles, indeed, are much longer than daughters. This could be explained with the role of Klp10A on centriole stability.²⁸ Klp10A loss leads, indeed, to centriole fragmentation. Therefore, the daughter might be more instable than the mother during the elongation process and, therefore, it could undergo to a delayed growth. Accordingly, Dplp is always associated in mutant testes to the mother centriole, but it is recruited to the daughters later when they have already duplicated and are growing. By contrast, in control testes Dplp is always present on parent centrioles. Thus, Klp10A seems to play opposite roles in the dynamic of the parent centrioles, allowing the elongation of the mother, but delaying the growth of the daughter. This observations point to a diverse requirement for Klp10A in the behavior of the parent centrioles thus uncovering new intrinsic structural dimorphisms that was not morphologically apparent during normal development.

Since the reduced length of the daughter centrioles is associated with the failure to recruit Dplp, an intriguing hypothesis is that Dplp could play some role in centriole elongation likely due to its involvement in the recruitment of centriole-associated proteins. Depletion of Dplp affects the recruitment of interphase PCM components in fly cultured cells, and its association with Cnn is required to maintain the integrity of the PCM and to drive centrosome separation during the early embryonic divisions.^12,31 Moreover, in the absence of Dplp the centrioles lose their orthogonal arrangement and start to partially fragment suggesting that Dplp may help the maintenance of centriole integrity. Remarkably, Dplp is localized to the basal body and not to the proximal centriole in Type I sensory neurons, consistent with a role for Dplp in basal body function.^44,45

The effect of Klp10A loss is not limited to centriole length as previously reported, but it also results in an abnormal distribution of Plk4, Sas-4 and Dplp.^28,29 We find, indeed, that these proteins spread along the length of the centrioles whereas they are restricted in control testes to the proximal region. An extended localization of the proximal centriole-associated proteins including Sas-4 and Dplp has been recently described in male germ stem cells of Asterless (Asl) mutants that also display overly long centrioles.⁴⁶ The finding of a similar phenotype in such different mutants point to the presence of a common target checking the localization of the basal proteins and monitoring centriole architecture and polarity.

How can the loss of Klp10A influence the localization of the basal proteins? A correlation between pericentrin and the minus end directed motor protein cytoplasmic dynein has been reported in human cultured cells and in Xenopus embryos where microtubule depolymerisation or dynein inactivation lead to failed pericentrin accumulation at the centrosome.^47,48 However, the observation that microtubules are dispensable for the localization of Dplp to the centrosome in early Drosophila embryos, points to alternative mechanisms in flies.⁴⁹ It has been suggested that the organization of internal centriole proteins may be involved in the proper positioning of the PCM components in the long Asl mutant male GSC centrioles.⁴⁶ Thus, mutations affecting the centriole architecture may result in the abnormal distribution of basal centriole-associated proteins.⁴⁶

The meiotic centrioles in Drosophila have a distinct proximal domain that keeps the overall organization seen in the short mitotic centrioles found in stem cells and spermatogonia. The main proteins involved in centriole assembly and centrosome organization are restricted to this region that closely matches the position of the cartwheel. The mid and distal regions of the centrioles are devoid of basal proteins and look like simple extensions of the centriole wall rather than true centrioles. Thus, the centriole functions appear to be essentially achieved by the basal domain.

A comparative analysis of germline stem cells, spermatogonia and primary spermatocytes, revealed that the cartwheel is disorganized in most of the mutant centrioles examined. Thus, the cartwheel or cartwheel associated components might play some roles to restrict the distribution of the basal proteins to the proximal region of the centriole. Sak/Plk4, Dplp and Sas-4 were, indeed, localized on the whole centriole fragments that lack cartwheel components. A possible hypothesis could be that the excessive elongation of the centrioles may depend by the fragmentation of the cartwheel leading to the loss of the basal properties.

However, both Asl and Klp10A elongated centrioles showed a distinct Sas-6 signal pointing to a remnant cartwheel.^28,46 Nevertheless, from conventional immunofluorecence analysis it is not possible to appreciate the true cartwheel structure.

The position of the daughter centrioles respect to the mother ones was unusual: in control testes the procentrioles always formed at the proximal end of the mothers, whereas in mutant testes the daughter centrioles had an ectopic position. This variation might be explained with the abnormal distribution of Plk4, the master protein involved in the assembly of the procentrioles and in the sequential recruitment of some centriole-associated proteins.^24,50-55

The broad distribution of Plk4 in mutant testes may enable more sites to recruit centriole-assembly proteins. Accordingly, the procentrioles in cep97 mutants were not always placed at the proximal end of the overly long mothers that also showed an extended localization of the basal proteins.⁴⁶ Therefore, Klp10A might be also involved in earlier stages of centriole duplication. Accordingly, the main defects in centriole organization were seen during the spermatogonial mitosis and in young primary spermatocytes when centrioles duplicate and start to elongate.

Sas-4 localization was more extended than Spd2 and Dplp in centrioles from 16-cell cysts of the spermatogonia and young primary spermatocytes in Klp10A testes, but as prophase progressed its distribution significantly reduces. By contrast, the localization of Dplp and Spd2 does not decrease. Therefore, Sas-4 is not required to maintain Dplp and Spd2 at the centriole. Fly centrioles, indeed, can still organize the PCM in the absence of Sas4.^56,57 This is surprising since Sas-4 would provide a scaffold for Dplp tethering at the mitotic centrosome.⁵⁸ The Cdk1-dependent phosphorylation of Sas-4 allows, indeed, the mitotic centrioles to duplicate and assemble functional centrosomes by promoting the incorporation of Asterless (Asl) that in turn plays a main role in recruiting Spd-2 during centrosome maturation.^3,4,51,34,59 Our observations suggest, therefore, that Sas-4 might play different roles in PCM organization during mitosis and meiosis and that its function is differently required during earlier and later stages of the male gametogenesis.

Materials and methods

Drosophila strains

OregonR flies were used as wild type. The Klp10A mutant line was previously described.⁶⁰ Observations were also made on a strain in which the male fertility was rescued by mobilization of the P-element with a P-element transposase.²⁸ In these flies we never observed abnormal localization of centriole associated proteins. Flies were raised on standard Drosophila medium at 24°C.

Antibodies

The following antibodies were used: chicken anti-Dplp (1:800,⁶¹); rabbit anti-Spd2 (1:500,⁴⁰); mouse anti-Sas-4 (1:50,⁵⁸); rabbit anti Sak/Plk4 (1:50,⁶¹); mouse anti-acetylated-tubulin (1:400; Sigma-Aldrich). Secondary antibodies used were Alexa Fluor 488 and 594 (1:800; Invitrogen).

Immunofluorescence preparations

Testes from mid aged pupae were dissected in phosphate buffered saline (PBS) and placed in a small drop of 5% glycerol in PBS on a glass slide. Testes were squashed under a cover glass and frozen in liquid nitrogen. The samples were then immersed in methanol for 10 min at −20°C. For antigen localization the samples were washed 15 minutes in PBS and incubated for 1 hour in PBS containing 0.1% bovine serum albumin (PBS-BSA) to block non specific staining. The samples were then incubated overnight at 4°C with the specific antisera in a humid chamber. After washing in PBS-BSA the samples were incubated for one hour at room temperature with the appropriate secondary antibodies. DNA was visualized following incubation of 3–4 minutes in 1 μg/ml Hoechst 33258 (Sigma). Samples were mounted in small drops of 90% glycerol in PBS.

Image acquisition

Images were taken by using an Axio Imager Z1 (Carl Zeiss) microscope equipped with an HBO 50W mercury lamp for epifluorescence and with an AxioCam HR cooled charge-coupled camera (Carl Zeiss). Gray-scale digital images were collected separately and pseudocolored and merged using Adobe Photoshop 7.0 software (Adobe Systems).

Transmission electron microscopy

Testes were isolated from mid-aged pupae and transferred in 2.5% glutaraldehyde buffered in PBS overnight at 4°C. Samples were subsequently rinsed in PBS and post-fixed in 1% osmium tetroxide in PBS for 2 hour at 4°C. The material was carefully washed in PBS, dehydrated in a graded series of Ethanol, embedded in a mixture of Epon-Araldite resin, and then polymerized at 60°C for 48 hr. Thin sections (60–70 nm thick) were obtained with a Reichert Ultracut E ultramicrotome equipped with a diamond knife, mounted upon copper grids, and stained with samarium triacetate and lead citrate. Samples were observed with a Tecnai Spirit Transmission Electron Microscope (FEI) operating at 100 kV and equipped with a Morada CCD camera (Olympus).

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors would like to thank A. Rodrigues-Martins and J. Gopalakrishnan for generously providing the antibodies.

Funding

This work was supported by a grant from MIUR (Grant 2012N7TX98 to GC).