Mst27D binds to Nup358

To identify interaction partners of Mst27D, we analyzed proteins that were co-purified with Mst27D-EGFP from testis extracts by mass spectrometry (MS). Mst27D-EGFP was expressed from a transgene (g-Mst27D-EGFP) under control of 5’ cis-regulatory sequences derived from the Mst27D locus [32]. Around 9000 testes were isolated for each of the four replicate experiments from late larvae and early pupae using mass isolation [35]. Affinity-purification (AP) from testes extracts was achieved with anti-EGFP beads. To dismiss proteins that bind to GFP or to anti-GFP beads rather than to the Mst27D part of Mst27D-EGFP, control AP-MS experiments were done in parallel with testes expressing only EGFP. Moreover, data obtained in analogous AP-MS experiments with testes expressing MNM-EGFP, SNM-EGFP and UNO-EGFP [35] was used for comparison. Our results indicated that Nup358 was most efficiently co-purified specifically with Mst27D (Fig 1D and S1 Table). In addition, RanGAP, which is known to bind to Nup358 [36], was also among the proteins co-purified specifically with Mst27D (Fig 1D). Nup358 is the largest constituent of the metazoan NPC. Human Nup358 has 3224 amino acid (aa) residues. The D. melanogaster ortholog, which contains fewer zinc fingers and lacks E3 ligase and cyclophilin domains, still has 2718 aa residues (Fig 1E). The α-helical N-terminal domain (NTD) and the adjacent oligomerization element (OE) of human Nup358 were shown to mediate binding as homo-pentamers on the cytoplasmic periphery of the NPCs [37,38] (Fig 1E). The reminder of Nup358 extends as a major component of the cytoplasmic filaments of the NPC as far as 60 nm into the cytoplasm and includes four Ran-binding domains (RanBDs) [37] (Fig 1E). D. melanogaster Nup358 binds RanGAP with a 23-aa stretch that is present within Nup358-PB, the major isoform [36] (Fig 1E).

Apart from Nup358 and RanGAP, the product of CG2233 appeared to be co-purified specifically with Mst27D-EGFP (Fig 1D). CG2233 is an uncharacterized gene with homologs detectable only in dipterans. According to transcriptomic analyses, it is strongly expressed in several tissues and also in testes although at slightly lower levels [39]. Further characterization will be required to confirm a potential CG2233 protein association with Mst27D. Beyond the proteins listed in Fig 1D, our AP-MS analyses identified some additional proteins that were co-purified with Mst27D-EGFP but not with control proteins (S1 Table). However, low numbers of detected peptides make future validation of these additional candidate interactors even more critical.

For confirmation of the interaction between Mst27D and Nup358, we performed co-immunoprecipitation experiments after transient transfection of cultured S2R+ cells with expression constructs. Mst27D was expressed with EGFP fused to the C terminus (Fig 2A) and Nup358 with mCherry at the N terminus (Fig 2B). Antibodies against mCherry allowed successful immunoprecipitation of mCherry-Nup358 (Fig 2C). Immunoblotting with anti-EGFP demonstrated that Mst27D-EGFP was co-immunoprecipitated efficiently with mCherry-Nup358 (Fig 2C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Interaction between Mst27D and Nup358 is mediated by their C-terminal regions.

(A-E) Identification of interacting regions in Mst27D and Nup358 by co-immunoprecipitation experiments after transient co-expression in S2R+ cells. (A,B) Schemes illustrating the structure of the analyzed fragments of Mst27D (A) and Nup358 (B). (C-E) Co-immunoprecipitation after transient co-expression of the indicated fusion proteins. Presence or absence of proteins in the extracts used for immunoprecipitation (input) or in the samples immunoprecipitated with anti-mCherry (IP) were analyzed by immunoblotting with anti-mCherry, anti-EGFP and anti-lamin. Positions of molecular weight markers are indicated.

https://doi.org/10.1371/journal.pgen.1010837.g002

Co-immunoprecipitation experiments with Mst27D protein fragments fused to EGFP (Fig 2A) demonstrated that the C-terminal (CT) part of Mst27D (aa 151–424) binds to Nup358 (Fig 2C). In contrast to Mst27D_CT-EGFP, Mst27D_CH-EGFP, i.e., the N-terminal Calponin homology (CH) domain (aa 1–150) fused to EGFP, was not co-immunoprecipitated with mCherry-Nup358 (Fig 2C).

To identify the region within Nup358 that binds to Mst27D, an initial series of N- and C-terminal truncations was generated (Fig 2B; N1-N5 and C1-C5). The truncated Nup358 fragments with N-terminal mCherry were co-expressed with Mst27D_CT-EGFP in S2R+ cells. After immunoprecipitation of the Nup358 fragments with anti-mCherry, immunoblotting with anti-EGFP indicated that it is the C-terminal region of Nup358 that binds to Mst27D_CT-EGFP (Fig 2D). Deletion of the C-terminal most region (aa 2342–2718) abolished co-immunoprecipitation of Mst27D_CT-EGFP (Fig 2D). Conversely, this most C-terminal Nup358 region was co-immunoprecipitated with Mst27D_CT-EGFP, albeit with lower efficiency compared to larger C-terminal fragments (Fig 2D). Microscopic analysis of the subcellular localization of the mCherry-Nup358 fragments indicated that NE localization was dependent on the presence of NTD and predicted OE (S1 Fig), as shown previously for Hs Nup358 [37,38]. Mst27D-EGFP also co-immunoprecipitated mCherry-Nup358 fragments that were not localized at the NE, suggesting that the interaction between Mst27D and Nup358 does not depend on other Nups and is thus likely direct. Further dissection with additional C-terminal truncations (Fig 2B; N6-N8) confirmed the importance of the C-terminal region of Nup358 for Mst27D binding. The additional truncations identified sequences C-terminal to aa 2538 as indispensable for Mst27D_CT-EGFP binding (Fig 2E, left). With full length Mst27D-EGFP, we obtained analogous results as with Mst27D_CT-EGFP (Fig 2E, right).

Overall, the results of our AP-MS analysis with extracts from g-Mst27D-EGFP testes and the co-immunoprecipitation experiments with S2R+ cells indicated that Mst27D binds to Nup358. This interaction is mediated by the C-terminal regions of these proteins.

Mst27D dimers can localize to the NE and on MTs

Beyond binding to NPCs via Nup358, Mst27D is predicted to associate with MTs based on the aa sequence similarity of its CH domain with that of Eb1. To assess these predictions, we analyzed the localization of Mst27D-EGFP after co-expression with mCherry-Nup358 in S2R+ cells. Indeed, as expected, Mst27D-EGFP was observed to be enriched at the NE and on MTs (Fig 3A). Mst27D_CT-EGFP, encompassing the C terminal region but not the N-terminal CH domain of Mst27D, exhibited a stronger enrichment at the NE and a dissimilar association with MTs, compared to full length Mst27D-EGFP (Fig 3A). Mst27D_CT-EGFP appeared to highlight comets, comparable to growing MT plus ends visualized by Eb1 [40], while full length Mst27D-EGFP associated with MTs all along their length (Fig 3A and 3C). Somewhat unexpectedly, Mst27D_CH-EGFP, i.e., the N-terminal CH domain tagged with EGFP, did not associate with MTs (Fig 3A and 3C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Subcellular localization of Mst27D in S2R+ cells at the NE and on MTs.

(A) S2R+ cells expressing mCherry-Nup358 and either Mst27D-EGFP, Mst27D_CT-EGFP or Mst27D_CH-EGFP were fixed and labeled with anti-α-tubulin. Some distinctive EGFP signals on MTs (arrows) and at the NE (arrowheads) are pointed out. (B) Evidence for Mst27D dimerization. Mst27D-mCherry was co-expressed in S2R+ cells with EGFP fusion proteins as indicated. The presence or absence of proteins in the extracts used for immunoprecipitation (input) or in the material immunoprecipitated with anti-mCherry (IP) was analyzed by immunoblotting with anti-mCherry and anti-EGFP. Proteins co-immunoprecipitated with Mst27D-mCherry are indicated (arrows), as well as the positions of molecular weight markers. (C) Subcellular localization of the indicated EGFP fusion proteins was analyzed by live imaging with S2R+ cell lines. Still frames display representative cells during interphase and mitosis, respectively. Scale bars = 5 μm.

https://doi.org/10.1371/journal.pgen.1010837.g003

Coiled coil-mediated homodimerization stimulates MT binding and tracking of growing MT plus ends strongly in case of EB1 family proteins [41–45]. Unlike full length D. melanogaster Eb1, its N-terminal CH domain failed to bind to MTs except after fusion to a leucine zipper dimerization motif [43]. Therefore, MT binding by the CH domain of Mst27D might require dimerization. While aa sequence similarity between D. melanogaster Eb1 and Mst27D is very low outside the N-terminal CH domain, secondary structure predictions indicated a region with coiled-coil forming propensity within the C-terminal part of Mst27D (Fig 1C). To assess whether Mst27D might dimerize, we performed co-immunoprecipitation experiments after co-expression of Mst27D-mCherry and Mst27D-EGFP in S2R+ cells. In parallel, Mst27D-mCherry was also co-expressed with either Mst27D_CH-EGFP, Mst27D_CT-EGFP or D. melanogaster Eb1-EGFP. Our results indicated that Mst27D-mCherry co-immunoprecipitated Mst27D-EGFP and Mst27D_CT-EGFP, but neither Mst27D_CH-EGFP nor Eb1-EGFP (Fig 3B). Thus, Mst27D appears to form homodimers mediated by the CT region.

To address whether a homodimerizing version of the CH domain of Mst27D might bind to MTs, we expressed an EGFP-tagged chimeric Mst27D-Eb1 protein in S2R+ cells. The N-terminal domain in this chimera was the CH domain of Mst27D and the C-terminal region was that of D. melanogaster Eb1 (Fig 3C). The subcellular localization of Mst27D_CH-Eb1_CT-EGFP was analyzed by time-lapse imaging of live cells stably expressing the chimeric protein. For comparison, additional S2R+ cell lines expressing either full length Mst27D-EGFP, Mst27D_CH-EGFP, Mst27D_CT-EGFP or Eb1-EGFP were analyzed analogously (Fig 3C). The chimeric Mst27D_CH-Eb1_CT-EGFP protein was strongly associated with MTs during both interphase and mitosis (Fig 3C). In contrast, Mst27D_CH-EGFP without the Eb1_CT region was not MT associated (Fig 3C), as previously observed with fixed preparations (Fig 3A). The localization of the chimeric Mst27D_CH-Eb1_CT-EGFP protein on MTs was similar to that of full length Mst27D-EGFP (Fig 3C) and clearly distinct from that of Eb1-EGFP (Fig 3C). While Eb1-EGFP highlighted moving comets, as expected [43,46], Mst27D-EGFP and the chimeric Mst27D_CH-Eb1_CT-EGFP were all along the MTs (Fig 3C).

Overall, these results demonstrate that the CH domain of Mst27D mediates binding to MTs and suggest that efficient MT binding depends on Mst27D homodimerization. Moreover, Mst27D binds all along MTs, contrasting with Eb1’s preference for growing MT plus ends.

Time-lapse imaging of S2R+ cells expressing Mst27D_CT-EGFP (Fig 3C) clearly confirmed an enrichment on the NE. However, in addition, EGFP signals were also present on moving comets, weaker but highly similar to Eb1-EGFP (Fig 3C). Although heterodimerization of Mst27D-mCherry and Eb1-EGFP was not observed in our co-immunoprecipitation experiments (Fig 3B), we speculate that heterodimerization of Mst27D_CT-EGFP with endogenously expressed Eb1 or perhaps with the uncharacterized Eb1 family protein encoded by CG18190, which appears to be expressed in S2R+ cells [47], might explain the comet localization.

The experiments involving expression of Mst27D-EGFP in S2R+ cells also provided evidence suggesting that Mst27D might have MT bundling activity. In cells with high expression levels, Mst27D-EGFP was localized in extended strong cables that were also intensely labeled with the tubulin stain SPY555-tubulin (S2 Fig). Intracellular cables were also present in S2R+ cells expressing Mst27D-mCherry at high levels (S2 Fig), indicating that MT bundling by Mst27D does not depend on the presence of EGFP, which has a weak propensity to form dimers, while mCherry is essentially monomeric [48]. Intracellular MT cables with Mst27D_CH-Eb1_CT-EGFP were also present in cells expressing high levels of this chimeric MT binding protein (S2 Fig). In contrast, cells expressing high levels of Mst27D_CH-EGFP, Mst27_CT-EGFP or D. melanogaster Eb1-EGFP did not display such cables.

Interestingly, the MT cables in cells with high levels of Mst27D-EGFP also contained mCherry-Nup358, if this protein was co-expressed (S3 Fig). In contrast, mCherry-Nup358 was not recruited into the cables induced by Mst27D_CH-Eb1_CT-EGFP (S3 Fig). These findings provide further support for the conclusion that Nup358 binds to the CT region of Mst27D, which is present in Mst27D-EGFP but not in Mst27D_CH-Eb1_CT-EGFP. Further support for the interaction between Nup358 and the CT region of Mst27D was obtained from an analysis of the effect of Nup358 depletion by RNAi on the localization of Mst27D_CT-EGFP in S2R+ cells (S3 Fig). Nup358 depletion abolished the localization of Mst27D_CT-EGFP at the NE (S3 Fig).

Mst27D association with the NE during spermatogenesis depends on Nup358

To analyze the subcellular localization of Mst27D in the organism, we generated a fly line expressing Mst27D-mCherry and EGFP-Nup358 from transgenes. Expression of these transgenes, g-Mst27D-mCherry and g-EGFP-Nup358, was controlled by cis-regulatory sequences derived from the endogenous gene loci. As described in further detail below, the transgenes rescued Mst27D and Nup358 null mutants, respectively, indicating that the tagged protein products were functional. While Nup358 appears to be expressed ubiquitously, Mst27D expression is testis-specific [32,39]. Whole mount preparations of testes from males expressing g-Mst27D-mCherry and g-EGFP-Nup358 revealed an onset of Mst27D-mCherry accumulation in spermatocytes at the S5 stage (Fig 4A). Mst27D-mCherry was clearly enriched at the NE. High magnification revealed precise co-localization of Mst27D-mCherry and EGFP-Nup358 at the NE (Fig 4A). Mst27D-mCherry and EGFP-Nup358 were present in elongated patches on the NE. Thus, NPCs appear to be clustered within the NE of late spermatocytes, in contrast to the even distribution of NPCs in the NE of cultured mammalian cells [49]. In D. melanogaster, however, super-resolution microscopy has recently revealed pronounced NPC clustering also in various additional tissues [50]. In contrast to EGFP-Nup358, which localized primarily at the NE, Mst27D-mCherry was also present in the cytoplasm (Fig 4A). The cytoplasmic Mst27D-mCherry signals were diffuse. Whether MT collapse during fixation contributed to the diffuse appearance of the cytoplasmic signals remains unresolved.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Nup358 is required for NE localization of Mst27D-mCherry in late spermatocytes.

(A) Co-localization of Mst27D-mCherry after onset of expression in S5 spermatocytes with EGFP-Nup358. Whole mount preparations of testes from males with g-Mst27D-mCherry and g-EGFP-Nup358 were fixed and labeled with a DNA stain. A maximum intensity projections of optical sections through the most peripheral germline cells in the testis tube are displayed in the left panel, with early stages in the apical testis region on the left and late spermatocytes on the right side. Single optical sections grazing the NE are displayed at high magnification in the right panel. (B-D) The deGradFP system [52] was used for analysis of Mst27D-mCherry localization after EGFP-Nup358 degradation in Nup358 mutant spermatocytes. (B) The indicated genotypes, -deGrad and +deGrad, were used for comparative analyses. (C) After live imaging of cysts released from early pupal testes, S6 spermatocytes were selected for quantitative analysis. Single optical sections through representative S6 spermatocytes are displayed. (D) Signal intensities of EGFP-Nup358 and Mst27D-mCherry at the NE and in the cytoplasm were quantified in regions of interests (ROIs) as indicated. For each cell, the intensities in the two cytoplasmic ROIs were averaged, as well as those in the two ROIs on the NE. The intensities observed within the ROI in center of the nucleus were considered to reflect background, which was subtracted from the intensities detected in the cytoplasm and at the NE. Swarm plots display signal intensities detected in individual cells as well as their mean (+/- s.d.); n = 49 from 9 distinct cysts (-deGrad) and 57 from 5 distinct cysts (+deGrad). Scale bars = 10 μm (A, left), 5 μm (A, right) and 10 μm (C).

https://doi.org/10.1371/journal.pgen.1010837.g004

To assess the dynamics of the association of Mst27D with the NE in vivo, we generated a g-Mst27D-Dendra2 transgenic fly line. After locally restricted photoconversion in a small region of the NE in S6 spermatocytes, the converted red fluorescent Mst27D-Dendra2 was detected within minutes throughout the NE (S4 Fig). In contrast, at least in cultured mammalian cells, where lateral NPC mobility within the NE has been analyzed, NPCs are essentially immobile [51]. Quantification of the dispersal of converted Mst27D-Dendra2 suggested that Mst27D in the cytoplasmic pool diffuses rapidly with intermittent episodes of transient association with and dissociation from the NE (S4 Fig).

To determine whether the association of Mst27D with NPCs is mediated by Nup358, we applied deGradFP [52]. This method permits elimination of GFP fusion proteins by expression of Nslmb-vhh4-GFP, an anti-GFP nanobody fused to the F-box of Slmb that is assembled into a GFP-specific ubiquitin ligase [52]. Hence, an analysis of Mst27D-mCherry localization after EGFP-Nup358 degradation in spermatocytes lacking endogenous Nup358 was envisaged. However, the original deGradFP system involves expression of Nslmb-vhh4-GFP with the UAS/GAL4 system which does not promote efficient expression in spermatocytes during mid- to late stages. Therefore, alternative expression of Nslmb-vhh4-GFP using the transgenes betaTub85DP-Nslmb-vhh4-GFP and exumP-Nslmb-vhh4-GFP transgenes was explored (S6 Fig). These transgenes exploit spermatocyte-specific cis-regulatory sequences derived from the gene loci betaTub85D and exuperantia (exu), respectively [53,54]. The transgene under control of exumP (exumP-deGrad in the following) resulted in earlier degradation during spermatocyte maturation and worked better for EGFP-Nup358 elimination just before the onset of g-Mst27D-mCherry expression (S5 Fig). The two genotypes that were compared for our analysis of the effects of EGFP-Nup358 elimination on Mst27D-mCherry (Fig 4B) will be designated as “-deGrad” and “+deGrad”. The exumP-deGrad transgene was present only in the latter, but otherwise these two genotypes were identical. For analysis, we used testes dissected at the early pupal stages. As expected, EGFP-Nup358 signals were strongly reduced in late +deGrad spermatocytes, compared to the -deGrad control spermatocytes (S5 Fig). Presumably reflecting some unanticipated low-level expression of exumP-deGrad even at earlier stages of spermatogenesis, EGFP-Nup358 was also slightly weaker in +deGrad compared to -deGrad in the anterior of pupal testes where cells at early stages reside (S5 Fig).

For careful quantification of the effects of EGFP-Nup358 degradation on Mst27D-mCherry localization, spermatocyte cysts were released from pupal testes for live imaging with a spinning disc confocal microscope. We focused on cysts at the S6 stage, which has a short duration and can be identified readily due to the spherical shape of the nuclei. Signal intensities in the EGFP and mCherry channels at the NE and in the cytoplasm were quantified (Fig 4C and 4D). EGFP-Nup358 signals at the NE were almost completely abolished in +deGrad spermatocytes (Fig 4C and 4D). Moreover, the Mst27D-mCherry signals at the NE, which were clearly above the cytoplasmic level in -deGrad spermatocytes, were no longer above the cytoplasmic level in +deGrad spermatocytes (Fig 4C and 4D), indicating that Mst27D-mCherry was either no longer localized at the NE or at least strongly reduced. These results indicate that Nup358 is required for normal localization of Mst27D to the NE. Beyond the effect on localization of Mst27D-mCherry, Nup358-EGFP degradation was also accompanied by an apparent reduction of the total level of Mst27D-mCherry (Fig 4C and 4D).

Co-localization of Mst27D and Nup358 in spermatids during nuclear elongation

To compare the localization of Mst27D and Nup358 throughout spermatogenesis, we studied testes of males with the g-Mst27D-mCherry and g-EGFP-Nup358 transgenes with fixed samples and by live imaging. In addition, we also analyzed an emerald-GFP (emGFP) knock-in allele of Nup358 [55] combined with g-Mst27D-mCherry, which gave indistinguishable results.

During the meiotic divisions, nuclear envelope breakdown (NEBD) is incomplete in D. melanogaster spermatocytes [11]. Similar as also in other tissues and developmental stages [56–59], NPCs are disassembled during M I and M II, but a fenestrated NE encloses spindles and chromosomes almost completely during the meiotic divisions except for more prominent openings beneath the centrosomes at opposite spindle poles. Interestingly, EGFP-Nup358 was present on the spindle envelope during the meiotic divisions, while Nup58-EGFP was diffusely distributed throughout the cell (S6 Fig). In contrast, Mst27D-mCherry was primarily enriched on the spindle (S6 Fig). After completion of M II, EGFP-Nup358 first adopted an essentially symmetric distribution throughout the NE around the spherical nuclei of early round spermatids. However, within about two hours, the symmetric EGFP-Nup358 localization was transformed into a polarized asymmetric distribution where EGFP-Nup358 was restricted to a hemisphere of the NE (S6 Fig and S1 Movie). During this NE polarization process, Mst27D-mCherry was strongly enriched at the EGFP-Nup358-containing region of the NE (S6 Fig and S1 Movie).

At the start of nuclear elongation in spermatids, Mst27D-mCherry and EGFP-Nup358 were co-localized on the NE in a hemispherical cap. The two proteins remained co-localized during nuclear elongation that was accompanied by the spatial transformation of the hemispherical cap into a stripe running in a grove extended all along the elongated spermatid nucleus (Fig 5A). After completion of nuclear elongation, Mst27D-mCherry and EGFP-Nup358 lost their association with spermatid nuclei. The elimination of these two proteins from the elongated spermatid nucleus occurred in a process designated here as NPC-NE shedding. This process, which could also be monitored in g-Nup58-EGFP testes, occurred around the time when the F actin-containing individualization cones started to form (S7 Fig). Time-lapse imaging of spermatid cysts expressing g-Nup58-EGFP and a g-ProtB-DsRed transgene indicated that NPC-NE shedding was completed within approximately one hour soon after the onset of nuclear accumulation of ProtB-DsRed (S7 Fig and S2 Movie). The NPC-NE was observed to move in basal direction and was detached eventually from the spermatid nucleus as a single large vesicle (S7 Fig and S2 Movie). Thereafter, this vesicle was displaced along the axoneme into the waste bag at the end of the sperm tail. After NPC-NE shedding, Nup58-EGFP was absent from spermatid nuclei except for a very weak dot signal at the basal end of the elongated nucleus that remained detectable also in mature sperm (S7 Fig). During NPC-NE shedding, EGFP-Nup358 and Mst27D-mCherry were also co-localized (Fig 5A). However, after completion of NPC-NE shedding, Mst27D-mCherry signals disappeared rapidly rather than moving along with the NPC-NE vesicles into the waste bag (Fig 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Localization of Mst27D and Nup358 in spermatids during nuclear elongation and NPC-NE shedding.

(A) Testes from males with g-Mst27D-mCherry and g-emGFP-Nup358 were fixed and labeled with a DNA stain. Regions from cysts during nuclear elongation and NPC-NE shedding are displayed. An isolated fully elongated nucleus is presented in the inset on top of the third column. (B-D) Antibodies against Mst27D. (B) The indicated peptides were used for production of affinity-purified polyclonal rabbit antibodies. (C) Total extracts from parental S2R+ cells (-) or from S2R+ cells expressing Mst27D-EGFP (+) were probed by immunoblotting and the indicated antibodies. Re-probing with anti-lamin was done for comparison of loading. (D) Immunofluorescent labeling of whole mount preparations of testes with the indicated antibodies against Mst27D and a DNA stain. Testes were isolated from either control (Mst27D⁺) or Mst27D null mutant males (Mst27D^-). Single optical sections displaying spermatids at the onset of nuclear elongation (early) and at the canoe stage (late) are displayed. Scale bars = 5 μm; while the bar on the left in (A) applies to the first two columns and the inset, the bar on the right applies to all other images.

https://doi.org/10.1371/journal.pgen.1010837.g005

For verification of expression pattern and subcellular localization revealed by the g-Mst27D-mCherry transgene (or identically with g-Mst27D-EGFP), we generated antibodies against three distinct Mst27D peptides (Fig 5B). The antibodies readily detected Mst27D-EGFP expressed in S2R+ cells, except for those directed against a C terminal peptide (Fig 5C). Immunostaining of testes with all our anti-Mst27D antibodies resulted in comparable specific signals, i.e., those not observed in Mst27D null mutant testes (Fig 5D; see also below), indicating that the antibodies directed against the C terminal peptide recognize wild-type Mst27D but not Mst27D-EGFP.

The specific anti-Mst27D signals observed in testes confirmed the findings made with the transgenes although not completely. The fluorescently tagged Mst27D proteins expressed from the transgenes appeared to accumulate much earlier than endogenous Mst27D protein detected by anti-Mst27D. While the transgene products were already detectable in S5 spermatocytes (Fig 4A), specific anti-Mst27D signals emerged at the onset of nuclear elongation (Fig 5D).

In principle, the later onset of Mst27D accumulation revealed by anti-Mst27D staining might reflect reduced detection sensitivity compared with fluorescence of Mst27D fusion proteins expressed by the transgenes. Alternatively, the cis-regulatory sequences in the transgenes might direct an expression pattern distinct from that of the endogenous Mst27D gene. In g-Mst27D-EGFP [32] and analogously in our g-Mst27D-mCherry transgene, SV40 terminator sequences were used instead of the downstream sequences present at the endogenous Mst27D locus. For clarification of the significance of the downstream sequences, we generated a g-Mst27D-EGFP-endo3’ transgene, in which the endogenous Mst27D downstream sequences were present rather than SV40 terminator sequences. Like anti-Mst27D signals, Mst27D-EGFP derived from g-Mst27D-EGFP-endo3’ was not detectable before the onset of nuclear elongation, but it was present during the canoe stage at levels even higher than those resulting with the SV40 terminator containing transgenes (S8 Fig).

In conclusion, based on the concurrent results obtained with anti-Mst27D and g-Mst27D-EGFP-endo3’, endogenous Mst27D accumulation starts after meiosis in early spermatids and increases during nuclear elongation before disappearing at the onset of spermatid individualization.

Mst27D is required for normal male fertility and normal nuclear elongation in spermatids

To determine the function of Mst27D, we characterized the phenotype of Mst27D mutants. A first mutant allele, Mst27D^LL01973 (hereafter abbreviated as Mst27D^LL) was generated in a large-scale effort for isolation of piggyBac transposon insertion lines [60]. We confirmed the presence of a piggyBac insertion within the Mst27D coding sequence in the Mst27D^LL allele by polymerase chain reaction (PCR) and DNA sequencing (Fig 6A). The predicted protein product that the Mst27D^LL allele could express in principle comprises the first 172 aa residues of Mst27D followed by seven extra aa residues (PRKIIIL) and a premature stop codon. However, the antibodies N1 and N2 directed against an N-terminal peptide did not produce specific signals in Mst27D^LL mutant testes, possibly because of non-sense mediated decay of Mst27D^LL transcripts. A second Mst27D allele (hereafter designated as Mst27D^cc) was generated using CRISPR/Cas9. Mst27D^cc, which carries a deletion that starts 42 bp upstream of the initiation codon in the 5’UTR and ends 191 bp before stop codon (Fig 6A), is predicted to be a protein null allele. Indeed, anti-Mst27D immunostaining of Mst27D^cc testes did not reveal any specific signals (Fig 5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Loss of Mst27D function is detrimental for male fertility and nuclear elongation in spermatids.

(A) Wild-type and mutant Mst27D alleles. Mst27D (top) is located within an intron of Mnn1 (bottom). The Mst27D^LL allele carries a piggyBac insertion (red triangle) within the coding region, and the Mst27D^cc allele an intragenic deletion (red box), eliminating start codon and the majority of the coding region. Genomic regions present in the indicated transgenes are shown as well (grey boxes). (B-D) Fertility of Mst27D mutants. Males or females of the indicated genotypes were mated with w¹¹¹⁸ flies and the number of resulting adult F1 progeny flies was counted. Df(2L)ade3 (Df) deletes Mst27D. The results of replicate crosses are indicated by filled circles; averages (+/- s.d.) are indicated as well. (E) Time-lapse imaging of nuclear elongation in Mst27D^cc/ Mst27D^LL mutants (Mst27D^-) and in Mst27D^cc/ + controls (Mst27⁺) expressing EGFP-Nup358. Still frames are displayed with nuclear clusters at the indicated time points (h:min). (F,G) Abnormal nuclear shape in Mst27D mutant spermatids during the stages characterized by nuclear Tpl94D (F) or ProtB (G). Whole mount preparations of testes expressing either g-Tpl94D-mRFP or g-ProtB-DsRed in a background that was either Mst27D^cc/ Df(2L)ade3 (Mst27D^-) or Mst27D^cc/ + (Mst27⁺) were labeled with a DNA stain. Single optical sections display spermatid nuclei at high magnification. Scale bars = 5 μm.

https://doi.org/10.1371/journal.pgen.1010837.g006

To assess whether Mst27D function is required for development to the adult stage, the progeny resulting from an inter se cross of flies with Mst27D^cc over a balancer chromosome (CyO) was analyzed. Mst27D^cc homozygotes comprised 34% of all eclosing flies, indicating that Mst27D function is not required for development to the adult stage. Mst27D^LL homozygosity was also found to be compatible with development to the adult stage. Flies homozygous for Mst27D^LL or Mst27D^cc did not display obvious morphological abnormalities.

To analyze effects on fertility without potential compounding effects of second-site mutations potentially present on the Mst27D^LL and the Mst27D^cc mutant chromosomes, we crossed the alleles over the deficiency Df(2L)ade3, which deletes the Mst27D gene. F1 hemizygous Mst27D mutant males and females were crossed to w¹¹¹⁸ flies and the number of resulting adult F1 progeny was determined. Thereby, hemizygous Mst27D^LL males were found to have a severely reduced fertility (Fig 6B and 6C). Average male fertility was about 20% of control (w¹¹¹⁸) fertility. In contrast, female fertility was not reduced in Mst27D^LL hemizygotes (Fig 6B). Comparable results were obtained with Mst27D^cc (Fig 6D). As residual Mst27D mutant male fertility appeared to be age-dependent, the apparent difference between Mst27D^LL and Mst27D^cc males might reflect age variation in test males rather than differences in allele strength.

Mst27D is within an intron of Menin 1 (Mnn1) (Fig 6A), which encodes a subunit of the Drosophila COMPASS-like complex [61,62]. Flies homozygous for an Mnn1 protein null mutation were reported to be viable and fertile [63,64]. Thus, the severe reduction of male fertility that was observed in Mst27^cc and Mst27^LL mutants is unlikely due to effects of the alleles on Mnn1 function. To exclude effects on Mnn1, we crossed g-Mst27D-EGFP into hemizygous Mst27D^LL males. The transgene restored normal fertility to hemizygous Mst27D^LL males (Fig 6B). Male fertility was also restored when g-Mst27D-mCherry was crossed into hemizygous Mst27D^LL males, but not with the transgenes expressing only parts of Mst27D (g-Mst27D_CH-mCherry and g-Mst27D_CT-mCherry) (Fig 6C). Overall, our analyses demonstrate that Mst27D is crucial for male fertility.

To resolve whether loss of Mst27D compromises spermatogenesis, mutant testes were characterized microscopically using both fixed and live preparations. Until after meiosis, abnormalities were not detectable. Moreover, time-lapse imaging of transheterozygous Mst27D^cc/ Mst27D^LL mutants expressing g-EGFP-Nup358 failed to reveal abnormalities during NE polarization and the subsequent polarization of the spermatid cysts, which clusters all the haploid nuclei at one end, with basal bodies docked in the center of the NPC-containing NE hemisphere and consistently oriented towards the other end of the cyst. However, during the following stages of nuclear elongation that are accompanied with Mst27D accumulation in wild-type spermiogenesis, Mst27D mutants displayed obvious defects (Fig 6E). In the mutants, nuclear elongation was limited and EGFP-Nup358 adopted a patchy distribution on the NE, while controls presented the normal spatial NPC-NE transformation from hemisphere to stripe along the elongation axis (Fig 6E).

The abnormality of nuclear elongation in Mst27D mutants appeared to vary from cyst to cyst. Nuclear elongation appeared to proceed to a variable extent, but the axis of elongation was rarely straight and the nuclear diameter usually varied along the axis. For a statistical comparison of control and mutant cysts at comparable stages, we analyzed fixed testes expressing marker transgenes for spermatid staging. A first marker transgene was g-Tpl94D-mRPF [65], expressing red fluorescent transition protein Tpl94D under control of Tpl94D cis-regulatory sequences. A second marker transgene, g-ProtB-DsRed, expressed red fluorescent Protamine B (ProtB) under control of ProtB cis-regulatory sequences [66]. For histone replacement during normal spermiogenesis, Tpl94D and ProtB accumulate sequentially in the spermatid nuclei during the early and late canoe stages, respectively [2]. While Tpl94D accumulation is transient, ProtB remains until after fertilization. Average number of spermatid cysts positive for either Tpl94D-mRFP or ProtB-DsRed were similar in Mst27D mutants and control testes (mean number per testis tube with Tpl94D-mRFP signals = 19.1 +/- 3.6 s.d. in control and 19.7 +/- 3.2 in Mst27D mutants, n = 10 for each genotype; mean number with ProtB-DsRed signals = 23.5 +/- 4.6 s.d. in control and 26.5 +/- 4.8 in Mst27D mutants, n = 6 and 7 for control and Mst27D mutant, respectively). The spatial arrangement of cysts differing in nuclear marker protein accumulation and DNA signal compactness also supported the conclusion that the histone replacement process proceeds without gross temporal perturbations in Mst27D mutant testes. As in sibling control testis tubes, clusters of spermatid nuclei positive for either His2Av-mRFP, Tpl94D-mRFP or ProtB-DsRed were spatially ordered within the distal region with increasing compactness of the DNA signals. While histone replacement proceeded in spermatid nuclei of Mst27D mutants, nuclear shapes were highly abnormal. Instead of a normal canoe form, Tpl94D-mRFP positive nuclei in Mst27D mutants presented a far more variable, tadpole-like spatial appearance (Fig 6F). In controls, only 1.4% of the Tpl94D-mRFP positive cysts contained nuclei with irregular non-canoe shapes (n = 191 cysts from 10 testes). In contrast, 98% of the Tpl94D-mRFP positive cysts in Mst27D mutants displayed abnormal nuclear shapes (n = 197 cysts from 10 testes). Similar findings were made with the ProtB-DsRed marker (Fig 6G). In controls, only 1.4% of the ProtB-DsRed positive cysts contained nuclei with irregular non-canoe shape (n = 141 cysts from 6 testes), while 95% of the ProtB-DsRed positive cysts displayed abnormal nuclei in Mst27D mutants (n = 159 cysts from 7 testes). The low frequency of cysts with Tpl94D-mRFP or ProtB-DsRed and apparently normal nuclear shapes in Mst27D mutants (2 and 5%, respectively) appeared in line with their residual male fertility.

Analogous to male fertility, nuclear elongation was restored back to normal in Mst27D mutants expressing g-Mst27D-mCherry, while g-Mst27D_CH-mCherry and g-Mst27D_CT-mCherry did not result in normal nuclear elongation (S9B-S9D Fig). The Mst27D fragments expressed from the latter two transgenes in males with Mst27D function also displayed a subcellular localization (S9A Fig) that was either severely (in case of CH) or subtly (in case of CT) distinct from that of the full length protein (S6 Fig), similar as also evident after expression in S2R+ cells (Fig 3A and 3C).

In summary, our phenotypic characterizations demonstrated that Mst27D function is required for normal nuclear elongation in spermatids.

Mst27D is required for formation of the MT bundle of the dense complex

The conspicuous strong bundle of MTs in the DC, initially revealed by ultrastructural analyses [20], was proposed to be crucial for nuclear elongation. Since our findings argued for Mst27D functioning as a linker between DC-MTs and the NPC-NE, a careful comparison of these DC-MTs in wild type and Mst27D mutants appeared of interest. While our immunofluorescent labeling with anti-tubulin antibodies after fixation with either formaldehyde or methanol resulted in weak signals in the DC, possibly because of accessibility problems, live imaging with testes expressing either α- or β-tubulin tagged with GFP generated considerably stronger signals in control spermatids with Mst27D⁺ function. MT organization revealed by live imaging of spermatids appeared to be somewhat distinct from that reported based on immunofluorescent analyses with fixed testes [67]. Moreover, live imaging revealed the temporal dynamics of MT organization in the DC. Intense cyst motility during this phase resulted in frequent displacement of the clustered nuclei out of focus and field of view, precluding successful tracking of individual spermatid nuclei throughout the complete nuclear elongation process. However, MT behavior in the DC of control spermatid cysts was revealed with multiple movies, covering 4–8 hours each. Nuclear elongation appeared to start at around 6 hours after the end of telophase II and lasted for approximately 10–14 hours (S10A Fig and S3 Movie).

EGFP-β-tubulin signals immediately after completion of M II (S10A Fig, t = 0:00) were most prominent on the centrosome and weak on an aster emanating from the centrosome and a coexisting cytoplasmic MT network, similar as observed in fixed cells [67]. Around 2–3 hours after M II completion, a single prominent straight MT bundle per spermatid appeared in the cytoplasm (S10A Fig, t = 2:00). This bundle, which has not been mentioned in the study with fixed cells [67], did not seem to be nucleated by the centrosome/basal body, which was docking on the NE during this period. Thereafter, extension of the tail MTs away from the basal body started (S10A Fig, t = 3:40). Concomitantly, periodic waves of contraction started to move across the cysts at intervals of about 20 minutes, while cyst polarization was observed. Spermatid nuclei were gradually redistributed to one end of the cyst and tail MTs extended towards the other end. Accompanied by the contraction waves, EGFP-β-tubulin signals became increasingly stronger in the region of the DC and nuclear elongation proceeded (S10A Fig, t = 6:39–19:18).

To characterize MT formation near nuclei during the elongation process in further detail, we performed time-lapse imaging with spermatid cysts expressing His2Av-mRFP in addition to α- or β-tubulin tagged with EGFP (Fig 7A–7C). Both EGFP-tubulin fusions yielded comparable results. At the start of nuclear elongation (around 6–10 hours after the end of M II), several MT bundles appeared extending away from the basal body and surrounding the spermatid nucleus similar to the ribs of an umbrella (Fig 7B and S4 Movie). Over time, the angle between the MT ribs decreased (as in a closing umbrella) and the MT bundles fused progressively into fewer and more prominent MT bundles, and finally into a single massive bundle (Fig 7B and 7C and S5 Movie). In parallel, an increasing deformation of the spherical nuclei occurred inside the MT ribs, resulting in nuclear extension eventually along the single prominent DC-MT bundle (Fig 7A–7C). A subtle but reproducible enrichment of the His2Av-mRFP signals in the subnuclear region next to the MT bundles accompanied nuclear elongation (Fig 7A). Time-lapse imaging of spermatid cysts expressing EGFP-α-tubulin and Mst27D-mCherry revealed extensive co-localization in the DC region. During the progressive fusion of MT bundles in the region of the DC that accompanied nuclear elongation, Mst27D-mCherry was clearly enriched on the MT bundles (Fig 7D and S6 Movie).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Mst27D⁺ function is required for formation of the dense complex MT bundle.

(A-E) MT organization in the DC during nuclear elongation was analyzed by time-lapse imaging of spermatids with Mst27D function (Mst27⁺) (A-D) or without (Mst27D^-) (E), expressing the indicated fluorescent proteins (His2Av-mRFP, EGFP-α-tubulin, EGFP-β-tubulin and Mst27D-mCherry). Still frames display regions with clustered spermatid nuclei (A and E) or individual nuclei at high magnification (B-D) with time indicated (h:min). Scale bars = 5 μm. (F) Cartoon summarizing MT and chromatin organization accompanying nuclear elongation in control and Mst27D mutants. (G) Mst27D function during nuclear elongation. The N-terminal CH domain of Mst27D binds to MTs and its C-terminal domain binds to the C-terminal region of Nup358 that is estimated to be located within the cytoplasm about 60 nm away from the NE, as 8 x 5 copies of Nup358 form the NPC’s cytoplasmic filaments. Mst27D forms dimers (or even higher oligomers) and thus might also function as an MT bundling protein (not illustrated). (H) Working model for formation of the DC. After docking of the basal body into the center of the NPC-NE hemisphere in early round spermatids, perinuclear MTs are nucleated, forming a radially symmetric array. Mst27D accumulation results in the binding of the MTs to NPCs. The MT bundling activity of Mst27D (or other MT associated proteins) breaks up the radial symmetry by progressive reorganization of the perinuclear MTs into a single bundle. Bundling is also proposed to promote stabilization and growth of the MTs within the bundles, presumably by increasing the local concentration of regulators of MT dynamics. MT bundling and extension in combination with attachment to the NPC-NE transform the shape of the nucleus from spherical to elongated needle shape.

https://doi.org/10.1371/journal.pgen.1010837.g007

Our observations in control spermatids suggested that nuclear elongation is driven by the progressive bundling of MTs into increasingly robust and longer bundles that culminated in the formation of a single strong extended bundle in the DC. As the MT bundles of the DC extended away from the basal body, the bundles are extended likely by growth at distal MT plus ends. However, recent studies have reported the presence of γ-tubulin not only at the basal body but also at the opposite acrosomal end of the elongating spermatid nucleus, although at far lower levels [67–69]. Thus, sliding apart of hypothetical anti-parallel MTs in the DC by tetrameric Kinesin-5 motor proteins might be an alternative potential mechanism for nuclear elongation. By live imaging of spermatids expressing GFP-tagged tubulin, we did not detect outgrowth of MTs away from the acrosomal end. In an attempt to visualize MT growth in the DC, we analyzed spermatids expressing Eb1-tdGFP by time-lapse imaging. Moving comet-like structures highlighted by fluorescent versions of EB1 family proteins have successfully visualized growing MT plus ends in many studies. However, within the region of the DC, Eb1-tdGFP comets were not evident (S10C Fig). Instead, high, uniform and temporally stable Eb1-tdGFP signals were present throughout the entire DC, while far weaker moving comets were detectable in other regions as previously reported [70]. After photo-bleaching of Eb1-tdGFP in the central region of the DC, signals recovered within less than one minute uniformly throughout the bleached region (S10C Fig). Importantly, recovery was not accompanied by movement of Eb1-tdGFP comets across the bleached region. Thus, the high Eb1-tdGFP signals in the DC most likely reflect highly dynamic MT lattice binding. Analogous experiments for analysis of fluorescence recovery after photobleaching (FRAP) with spermatids expressing EGFP-α-tubulin, revealed far slower recovery in comparison to Eb1-tdGFP, indicating that DC-MTs are stable (S10C Fig).

To address the role of MT dynamics for DC elongation, demecolcine, an MT inhibitor, was added to spermatid cysts expressing EGFP-α-tubulin just before the start of time-lapse imaging. Demecolcine was added at a concentration that readily eliminated dynamic MTs like those assembling the meiotic division spindles, as also demonstrated previously [71]. However, the EGFP-α-tubulin signals in the DC were not weakened substantially by demecolcine, confirming that DC-MTs are stable, similar to those in the axoneme [67]. However, demecolcine inhibited the formation of a single extended MT bundle in the DC, as well as nuclear elongation (S10B Fig), suggesting that these processes depend on dynamic MT polymerization.

To assess the role of Mst27D for the organization of MT bundles in the DC, we analyzed Mst27D mutants (Mst27D^cc/ Df(2L)ade3) expressing EGFP-α-tubulin and His2Av-mRFP. Interestingly, time-lapse imaging indicated that prominent MT bundles failed to form adjacent to the nucleus in Mst27D mutant spermatids during the stages of the abnormal nuclear elongation (Fig 7E and 7F and S7 Movie). Position and shape of the nuclei became increasingly abnormal in the Mst27D mutant spermatids (Fig 7E and 7F) compared to control (Fig 7A and 7F). In Mst27D mutant spermatids, the axoneme was observed to bypass the nucleus laterally, and the normal polarized enrichment of the His2Av-mRFP signal was not evident. In conclusion, the Mst27D mutant phenotype clearly revealed that Mst27D⁺ function is required for the MT bundling process that normally proceeds in the DC. While still attempted, residual nuclear extension did not occur in a strictly linear direction in the mutants, indicating that the DC serves as a stiff rod that guides nuclear elongation during normal spermiogenesis.

The spermiogenic processes following nuclear elongation, i.e., NPC-NE shedding and spermatid individualization, were also abnormal in Mst27D mutants (S11 Fig), presumably at least in part as a secondary consequence of the defective nuclear elongation. Formation of investment cones (ICs), which mediate sperm individualization, was clearly observed in Mst27D mutants by staining of F-actin with fluorescent phalloidin (S11 Fig). Moreover, before completion of IC assembly, NPC-NE shedding was observed in hemizygous Mst27D^LL mutants expressing Nup58-EGFP (S11 Fig). However, while in controls Nup58-EGFP was stripped off from the spermatid nuclei completely (except for a remaining very weak basal dot), substantial amounts of Nup58-EGFP remained on the NE around the spermatid nuclei of Mst27D mutants (S11 Fig), indicating partial individualization defects.

Mst27D and the sperm-specific SUN domain protein Spag4 function independently

Our results indicated that Mst27D, as a linker between NPCs and MTs, is required for proper nuclear elongation along the straight line defined by the DC MT bundle. In mammalian spermiogenesis, nuclear shaping is thought to depend on connections between NE and MTs that are provided by LINC complexes containing testis-specific SUN domain proteins [26,72]. Drosophila Spag4, a SUN domain protein expressed exclusively during spermatogenesis, was reported to display a localization similar to that of Mst27D. In spermatocytes, Spag4 localization is still symmetric throughout the NE but after NE polarization it is confined to the hemispherical NPC-NE, where it is required for dynein-dynactin recruitment to this NE region in early round spermatids. During nuclear elongation, Spag4 remains enriched on the NPC-NE although with decreasing intensity [73]. To evaluate potential hierarchies in the localization of Spag4 and Mst27D, we analyzed Mst27D-EGFP localization in spag4 mutants, as well as Spag4-EGFP localization in Mst27D mutants (S12 Fig).

In spag4 mutants, Mst27D-EGFP displayed a normal localization during NE polarization in early round spermatids and also during the subsequent nuclear elongation, which appeared to be normal (S12A Fig). Nuclear morphology was still normal even in late spag4 mutant spermatids soon after NPC-NE and Mst27D-EGFP shedding at the start of individualization. However, in very late spermatids, the needle shaped nuclei displayed a highly abnormal curled morphology in spag4 mutants (S12A Fig), as reported [73].

The localization of Spag4-EGFP was not affected in Mst27D mutants (S12B Fig). In both control and Mst27D mutants, Spag4-EGFP localization was slightly distinct from that of Mst27D-EGFP in early round spermatids. While both Mst27D-EGFP and Spag4-EGFP were restricted to the NPC-NE (S12B Fig), Spag4-EGFP was most strongly enriched in the basal body region, which was largely devoid of Mst27D-EGFP (S12B Fig). This preferential localization of Spag4-EGFP on the basal body increased further during nuclear elongation in controls and also during the corresponding stages in Mst27D mutants (S12B Fig).

In conclusion, the proteins Mst27D and Spag4, which both connect NE with MT structures in spermatids, localize independently and provide distinct functions. While Spag4 maintains the connection between needle shaped nucleus and basal body/axoneme during the final stages of spermiogenesis [73], Mst27D mediates the attachment of the NPC-NE to the DC-MTs to permit normal nuclear elongation.

Drosophila lines

The source of the mutant alleles and transgenes that were used for this study is given in S2 Table. Transgenes under control of cis-regulatory regions derived from the endogenous genomic loci are named here with the prefix “g-”(as in g-EGFP-Nup358 for example). Standard crossing and meiotic recombination were used to generate strains with combinations of these mutant alleles and/or transgenes. The genotypes of the flies analyzed are described in detail in S3 Table, and additional explanations are provided in the following. Flies analyzed were usually raised at 25°C. w¹ or w¹¹¹⁸ were used as wild-type control unless stated otherwise.

Lines with g-Nup58-EGFP generated with a pCaSpeR4 construct have been described previously [82]. For generation of lines with the transgenes g-Mst27D-mCherry, g-Mst27D-Dendra2, g-Mst27D_CH-mCherry, g-Mst27D_CT-mCherry or g-Mst27D-EGFP_endo3’, the pCaspeR4 constructs described further below were injected into w¹¹¹⁸ embryos (BestGene Inc., Chino Hills, CA, USA). The transgenic g-EGFP-Nup358 line was obtained after microinjection of pattB_gEGFP-Nup358 (see below for plasmid description) into y¹ w⁶⁷c²³; P{CaryP}attP40 embryos. A homozygous viable and fertile g-EGFP-Nup358 stock was established.

The Nup358^12A002 mutant allele [83] has an A to T mutation at nucleotide position 1669 resulting in a premature stop codon (K557stop). The mutant Nup358 proteins that might be expressed from this allele do not include the C-terminal region required for Mst27D binding. Nup358^0175-G4 [84] contains a piggyBac insertion in the 5’ UTR 12 bp upstream of the start codon. Trans-heterozygous Nup358^12A002/ Nup358^0175-G4 flies did not develop to the adult stage. However, with the g-EGFP-Nup358 transgene present in this trans-heterozygous background, the mutants developed into viable and fertile adults.

For comparison of the localization of Nup358 and Mst27D, the line w*; P{w⁺, g-EGFP-Nup358} attP40/ CyO; PBac{w⁺, IT.GAL4}Nup358^0175-G4, P{w⁺, g-Mst27D-mCherry} III.7/ TM6B, Tb, Hu was generated. Males without balancer chromosomes were used for whole mount preparations and live imaging (Fig 4A). Alternatively (Fig 5A), we used males without balancers arising in the stock w*; P{w⁺, g-Mst27D-mCherry} II.3/ CyO; emGFP-Nup358, PBac{y[+mDint2] = vas-Cas9}VK00027/ TM6B, Tb, Hu.

To assess effects on the subcellular localization of Mst27D-mCherry after EGFP-Nup358 elimination by deGradFP (Figs 4B–4D and S5B), g-Mst27D-mCherry was first recombined with the Nup358^0175-G4 allele and then combined with g-EGFP-Nup358. The genotype of the resulting fly stock was w*; P{w⁺, g-EGFP-Nup358} attP40/ CyO, Dfd-YFP; PBac{w⁺, IT.GAL4}Nup358^0175-G4, P{w⁺, g-Mst27D-mCherry} III.7/ TM6B, Tb, Hu. In addition, the Nup358^12A002 allele was combined with exumP-Nslmb-vhh4-GFP. The genotype of the resulting fly stock was w*; P{w⁺, exumP-NSlmb-vhhGFP4}attP40/ CyO; P{w⁺, FMRFa-EGFP.Tv}3, P{w⁺, UAS-myr-mRFP}2, Nup358^12A002/ TM6B, Tb, Hu. The two stocks were then crossed to generate the analyzed trans-heterozygous Nup358 mutant genotype w*; P{w⁺, g-EGFP-Nup358} attP40/ P{w⁺, exumP-NSlmb-vhhGFP4}attP40; PBac{w⁺, IT.GAL4} Nup358^0175-G4, P{w⁺, g-Mst27D-mCherry} III.7/ P{w⁺, FMRFa-EGFP.Tv}3, P{w⁺, UAS-myr-mRFP}2, Nup358^12A002 (“+deGrad” genotype). In addition by crossing the first stock with a third stock w¹¹¹⁸; P{w+, FMRFa-EGFP.Tv}3, P{w+, UAS-myr-mRFP}2, Nup358^12A002/ TM6B, P{w+, Dfd-EYFP}3, Sb Tb ca, the additional analyzed control genotype w*; P{w⁺, g-EGFP-Nup358} attP40/ +; PBac{w⁺, IT.GAL4} Nup358^0175-G4, P{w⁺, g-Mst27D-mCherry} III.7/ P{w⁺, FMRFa-EGFP.Tv}3, P{w⁺, UAS-myr-mRFP}2, Nup358^12A002 (“-deGrad” genotype) was generated.

Mst27D mutant alleles were generated with CRISPR/Cas9. The plasmids pCFD5-sgRNA-Mst27D-1_4 and pCFD5-sgRNA-Mst27D-2_3 (see below) were injected individually into embryos of the stock yw;;nos-cas9(III-attP2) (BestGene Inc., Chino Hills, CA, USA). Resulting adult G0 males were crossed out singly with three w*; Gla/CyO virgins. The primer pair LP043/LP044 (see S4 Table for oligonucleotide sequences), which annealed 86 bp upstream and 147 bp downstream of the two gRNA-targeting sites, respectively, were used for identification of deletions by PCR in genomic DNA isolated from a pool of some of the F1 progeny. Eventually, the Mst27D^cc1-4 allele was established as a balanced stock. Molecular characterization by PCR and sequencing indicated the presence of an intragenic deletion. Sequences from 42 bp upstream of initiation codon until 191 bp before the stop codon were replaced by a novel short 7 bp sequence (5’-TTAGAAT-3’), resulting in a sequence across the breakpoints of 5’- … TACTCCTTACCTTAGAATAAACGGCAAG- … 3’.

The dynamics of NPC-NE shedding in late spermatids (S7 Fig) was analyzed by time-lapse imaging with the genotype w*; P{w⁺, g-Nup58-EGFP}35B (12.4); P{w⁸, ProtB-DsRed-M1}50A (III). The transgene g-ProtB-DsRed [66] carries a genomic 4.2 kb fragment containing the Protamine B gene region and 1.3 kb of sequence upstream of the transcription start site and 1.2 kb of sequence downstream of the polyadenylation site.

For analyses with g-Mst27D-EGFP_endo3’ in a spag4 null mutant background, the alleles spag4¹ and spag4⁶ were used in trans. These spag4 alleles were generated by replacing the wild-type spag4 gene with w⁺ using ends-out homologous recombination [73]. The spag4-GFP transgene [73] was analyzed in the hemizygous Mst27D mutant background Mst27D^cc1-4/ Df(2L)ade3. The genotype w*; Mst27D^cc1-4/ +; P{w⁺, spag4-GFP} 3/ + was used as wild type control (S12 Fig).

Plasmids

The gRNA expression plasmids pCFD5-sgRNA-Mst27D-1_4 and pCFD5-sgRNA-Mst27D-2_3 were made using the vector plasmid pCFD5 [85] and the synthetic DNA fragments LP049 and LP050 (gBlocks Gene Fragments obtained from IDT, Leuven, Belgium) (for sequences see S4 Table). The synthetic DNA fragments and pCFD5 were digested with BbsI before ligation. An online tool (http://targetfinder.flycrispr.neuro.brown.edu/) was used for identification of gRNA targets sites. Each pCFD5 derivative contained two distinct gRNA sequences targeting different regions of Mst27D.

The plasmids pCaSpeR-gMst27D-mCherry, pCaSpeR-gMst27D_CH-mCherry, pCaSpeR-gMst27D_CT-mCherry, pCaSpeR-gMst27D-Dendra2 and pCaSpeR-gMst27D-EGFP-endo3’ were made as follows. For pCaSpeR-gMst27D-mCherry, a fragment containing the Mst27D coding region and 1045 bp upstream sequences was amplified from w¹ genomic DNA with the primer pair LP017/LP018. The fragment was digested with BssHII and BglII and ligated into pUASt-mcs-mCherry [86] that had been digested with MluI and BglII. Sequencing of the resulting construct (pCaSpeR-gMst27D-mCherry) revealed the presence of a mutation that induces an aa sequence change (N92T) compared to the sequence predicted by the FlyBase reference genome. Sequencing of an independent PCR product amplified from w¹ genomic DNA with the primer pair LP011/LP021 revealed an identical single base change, indicating that it reflects a single nucleotide polymorphism present in the w¹ genome. Our cloning strategy eliminated the sequences containing the multimerized UAS_GAL4 binding sites and the minimal heat shock promoter that were present in the pUASt-mcs-mCherry vector. However, the 3’ sequences downstream of the mCherry coding sequence were the SV40 terminator sequences derived from pUASt-mcs-mCherry. The final construct pCaSpeR-gMst27D-mCherry was therefore analogous to the g-Mst27D-EGFP transgene construct of Gärtner et al. (2019) [32] except that the mCherry rather than the EGFP coding sequence were fused C-terminally.

The construct pCaSpeR-gMst27D-mCherry was modified the generate pCaSpeR-gMst27D_CH-mCherry and pCaSpeR-gMst27D_CT-mCherry. For cloning of these derivatives, the synthetic insert DNA fragments CL342 and CL343, respectively, and the vector pCaSpeR-gMst27D-mCherry were digested with SpeI and BglII before ligation. A similar strategy was used for the generation of pCaSpeR-gMst27D-Dendra2. The synthetic DNA fragment CL344 and the vector pCaSpeR-gMst27D-mCherry were ligated after digestion with KpnI and XbaI.

To generate pCaSpeR-gMst27D-EGFP-endo3’, pUASt-mcs-EGFP and pCaspeR-gMst27D-mCherry were both digested with XhoI and XbaI to produce an insert fragment from pCaspeR-gMst27D-mCherry and vector fragment from pUASt-mcs-EGFP that were ligated, yielding the cloning intermediate pCaSpeR-gMst27D-EGFP. The SV40 terminator sequences in this intermediate were replaced with genomic 3’ sequences derived from the Mst27D locus that were obtained by amplification from w¹ genomic DNA with the primers CL396/CL397. The resulting DNA fragment comprising 288 bp of Mst27D 3’ sequences downstream of the stop codon and pCaSpeR-gMst27D-EGFP) were digested with XbaI and BamHI before ligation which generated pCaSpeR-gMst27D-EGFP-endo3’.

For the construction of pattB_gEGFP-Nup358, the primer pair CL405/CL406 was used to amplify a fragment containing 827 bp of upstream sequences with the Nup358 5’UTR region, intron 1 and the beginning of exon 2 from w¹ genomic DNA. The PCR fragment was ligated as insert into pattB after digestion of insert and vector with BamHI and EcoRI, yielding the cloning intermediate 1 (pattB-Nup358_upstream). The EGFP coding sequence (736 bp) was amplified from pUASt-mcs-EGFP using the primer pair RAS42/CL385. The PCR product was then ligated into pattB-Nup358_upstream after digestion of insert and vector with EcoRI and NotI, yielding intermediate 2 (pattB-Nup358_upstream-EGFP). The plasmid pUbiqP_mCherry-Nup358-ry_2 (see below) was digested with NotI and SalI to generate an insert fragment that was ligated into pattB-Nup358_upstream-EGFP, which had been digested with NotI and XhoI, yielding the final construct pattB_gEGFP-Nup358.

The plasmids pMT-Mst27D-EGFP_bla, pMT-Mst27D_CH-EGFP_bla, pMT-Mst27D_CT-EGFP_bla, pMT-Mst27D-mCherry-bla and pMT-Mst27D_CH-EB1_CT-hl-EGFP_bla used for transfection of S2R+ cells were made as derivatives of pMT-bla [35]. In case of pMT-Mst27D-EGFP_bla, the primer pair CL349/CL231 was used for amplification of an Mst27D-EGFP fragment (1995 bp) from genomic DNA of g-Mst27D-EGFP flies [32]. The PCR fragment and pMT-bla were digested with BglII and NotI before ligation, yielding pMT-Mst27D-EGFP_bla. For construction of pMT-Mst27D-CH-EGFP_bla, an insert was produced by overlap PCR. The region including the sequences coding for the CH domain was amplified from pMT-Mst27D-EGFP_bla with the primer pair CL349/LP037 and the EGFP coding sequence was amplified from pMT-Mst27D-EGFP_bla with the primer pair LP038/CL231. A CH-EGFP fusion fragment was then produced using a mixture of these two fragments as template and the primer pair CL349/CL231 in the overlap PCR. The overlap PCR product was ligated into pMT-bla after digestion of vector and insert with BglII and NotI. For the generation of pMT-Mst27D-CT-EGFP_bla, the insert fragment was amplified from pMT-Mst27D-EGFP_bla with the primer pair LP039/CL231. The PCR product was then ligated into pMT-Mst27D-EGFP_bla after digestion of insert and vector with SpeI and NotI. For pMT-Mst27D-mCherry-bla, the plasmid pMT27D-CT-mCherry (vector) and pCaSpeR-Mst27D-EGFP-endo3’ (insert) were digested with SpeI and NotI before ligation. For pMT-Mst27D_CH-EB1_CT-hl-EGFP_bla, the synthetic DNA fragment CL381 was ligated into pMT-hl-EGFP-bla after digestion with BglII and NotI.

For the construction of pUbiqP_mCherry-Nup358-ry_2, the primers CL350 and CL351 were annealed to produce a linker that was inserted into pUbiqPry [87] that had been digested previously with Asp718 and SpeI. Thereby, the first cloning intermediate (pUbiqPry-linker_2) was obtained. The mCherry coding sequence was amplified from pUASt-mcs-mCherry with the primer pair CL352/CL353. After digestion of the resulting fragment and pUbiqPry-linker_2 with SpeI and NotI, a ligation generated the second cloning intermediate (pUbiqP-mCherry-ry_2). Thereafter, pUbiqP-mCherry-ry_2 and the Nup358 expressed sequence tag (EST) plasmid pOT2_LD43045 (Drosophila Genome Resource Center, Indiana University, Bloomington, Indiana, USA) were digested with HpaI and SacI for subsequent ligation, yielding cloning intermediate 3 (pUbiq-mCherry-Nup358-4-ry_2). In a next step, pUbiq-mCherry-Nup358-4-ry_2 and the additional Nup358 EST plasmid pOT2_LD24888 (Drosophila Genome Resource Center, Indiana University, Bloomington, Indiana, USA) were digested with HpaI and EcoRI, and ligated to generate cloning intermediate 4 (pUbiq-mCherry-Nup358-2/3/4-ry_2). Finally, an additional Nup358 fragment was amplified with CL358 and CL359 from pOT2_LD24888 and ligated into pUbiq-mCherry-Nup358-2/3/4-ry_2 after digestion of fragment and vector with NotI and EcoRI, resulting in the final construct pUbiq_mCherry-Nup358-ry_2.

To generate plasmids for expression of a series of C- and N-terminal Nup358 truncations, a mutagenic amplification method [88] was used to engineer NheI target sites into pUbiqP_mCherry-Nup358-ry_2 at suitable positions within the Nup358 coding region. The plasmid pUbiqP_mCherry-Nup358-Nhe1 was generated with this method using the primer LP023 and had the introduced NheI site around the codon of aa residue 2342 of Nup358, pUbiqP_mCherry-Nup358-Nhe2 generated with the primer LP024 around aa residue 1970, pUbiqP_mCherry-Nup358-Nhe3 generated with the primer LP025 around aa residue 1498, pUbiqP_mCherry-Nup358-Nhe4 generated with the primer LP026 around aa residue 1269, and pUbiqP_mCherry-Nup358-Nhe5 generated with the primer LP027 around aa residue 830. To create plasmid derivatives for expression of N-terminal fragments, the five pUbiqP_mCherry-Nup358 versions with introduced NheI sites were digested with NheI and AvrII. Thereafter, a linker obtained by annealing the primers LP030 and LP031 was ligated into the opened plasmids to generate pUbiqP_mCherry-Nup358-N1 (aa 1–2342), -N2 (aa 1–1970), -N3 (aa 1–1498), -N4 (aa 1–1269),- and -N5 (aa 1–830). To create plasmid derivatives for expression of C-terminal fragments, the five pUbiqP_mCherry-Nup358 versions with introduced NheI sites were digested with NheI and NotI, and closed by ligation with a linker obtained by annealing the primers LP028 and LP029, yielding pUbiqP_mCherry-Nup358-C1 (aa 2342–2718), -C2 (aa 1970–2718), -C3 (aa 1498–2718), -C4 (aa 1269–2718), and -C5 (aa 830–2718). The additional constructs pUbiqP_mCherry-Nup358-N6 (aa 1–2484), -N7 (aa 1–2538), and -N8 (aa 1–2698) were generated by double digestion of pUbiqP-mCherry-Nup358-Nhe1 (vector) and the synthetic DNA fragments LP054, LP055 and LP056 (inserts) with NheI and AvrII before ligation.

Fertility tests

The fertility of Mst27D^LL hemizygous flies was analyzed after crossing Mst27D^LL/CyO and Df(2L)ade3/CyO, P{w+, Dfd-YFP} to generate the genotypes Mst27D^LL/Df(2L)ade3 and Mst27D^LL/CyO, P{w+, Dfd-YFP} as well as Df(2L)ade3/CyO. The fertility of these genotypes was compared to that of w¹¹¹⁸ flies. Moreover, for additional comparison, the fertility of Mst27D^LL/Df(2L)ade3; g-Mst27D-EGFP/+ flies, generated with additional crosses, was analyzed in parallel. To assess male fertility, 10 single males of each genotype were crossed individually with three virgins of w¹¹¹⁸ in a vial at 25°C for 4 days and turned over into second vial for 4 days. To determine female fertility, five replicate crosses were set up. In each cross, three virgin females were mated with three w¹¹¹⁸ males. The number of adult F1 progeny eclosing in the second vials was counted.

To determine to which extent the g-Mst27D transgene variants (g-Mst27D-mCherry, g-Mst27D_CH-mCherry or g-Mst27D_CT-mCherry) improve the low fertility of Mst27D^LL hemizygous males, fertility tests were performed with the genotypes w*; Mst27D^LL/Df(2L)ade3; g-Mst27D transgene variants/ +, as well as with the genotype Mst27D^LL/Df(2L)ade3. In addition, for comparison w¹¹¹⁸ control flies were analyzed in parallel. Four replicate crosses were set up for each genotype, except for the w¹¹¹⁸ control, where only three replicate crosses were started. For each cross, three males were mated with three virgins of w¹¹¹⁸ in a vial at 25°C for 3 days, and then turned over into another fresh vial for 3 days. All adult F1 flies that eclosed in the second vials were counted.

To assess the effect of the Mst27D^cc null mutation on fertility, homozygous null mutant males and females, respectively, were mated with w¹¹¹⁸ flies and the number of the resulting adult F1 progeny was determined. For comparison, analogous inter se crosses of w¹¹¹⁸ flies were analyzed. To assess male fertility, 10 single males were crossed individually with three w¹¹¹⁸ virgins at 25°C for 2 days, and then turned over into a second vial for 3 days. To determine female fertility, five replicate crosses were set up with three virgins and three w¹¹¹⁸ males at 25°C for 2 days, and then turned over into a second vial for 3 days. Adult F1 progeny eclosing from the second vials were counted.

Testis mass isolation

Testis mass isolation was performed as described [35]. Flies with genotype w*;att22A{exum-mCherry}/ CyO; g-Mst27D-EGFP/ TM6B, Tb, Hu were expanded to start cultures in 25 large bottles (4.7 cm diameter × 9.5 cm high, with fly food filling the bottom 2.5 cm). The resulting flies were distributed into two collection cages (80 × 80 × 80 cm). In each cage, four round plates containing standard fly food (14 cm diameter × 2 cm height) were placed for egg collection. Every 8 hours at 25°C or 16 hours at 18°C, the plates were replaced with four fresh plates. The plates with eggs were cultured at 25°C for 2 days. Afterwards, the food and larvae from the plates were transferred to fresh food plates and plexiglass puparation cylinders (14 cm diameter × 50 cm height) were inserted on top. After 18 hours incubation at 25°C, early pupae and wandering stage 3^rd instar larvae were dislodged from the wall of the cylinders with a silicon spatula and a flow of water. Larvae and early pupae were collected in a first sieve with 700 μM pore size. Afterward, pupae and larvae were re-suspended in Ringer Buffer (182 mM KCl; 46 mM NaCl; 3 mM CaCl2, 10 mM Tris-HCl, pH 7.2) and minced in a metal mill attached to a food processor (KSM125 equipped with KGM, Kitchen Aid). The disrupted tissue pieces were washed out of the mill by a continuous flow of Ringer buffer that was passed through a second sieve with 355 μM pore size and a third sieve with 100 μM pore size. The material in the second sieve was re-loaded into the mill for 2–3 additional rounds of mincing. All material from the third sieve was washed away with Ringer buffer and collected in Falcon tubes for further purification.

Ficoll gradient density centrifugation was used for a first gonad purification step. The gradient was prepared with one bottom layer of 20 ml of 25% Ficoll PM400 and one upper layer of 20ml 12% Ficoll PM400 in Ringer buffer. After settling of the tissue material in the Falcon tubes for 15 minutes, supernatant was discarded and sediments were transferred on the top of the Ficoll gradient. The gradient tubes were centrifuged at 4500 × g for 1 hour at 4°C. Afterwards, the boundary regions between the 25%/12% and 12%/0% gradient phases, as well as the material in the region in between was collected and diluted 1:5 with Ringer buffer. Fluorescence-based particle sorting was performed as an additional purification step as described [89] with the following modifications. After a settling period of 30 minutes, supernatants were discarded, and the remaining sediments were applied to a Biosorter (Union Biometrica; FOCA 1000; standard settings). Depending on particle density, the suspension was further diluted in Ringer solution for optimal sorting. Gating for selection of testes was based on measurements of the following parameters: time of passage through optical detection cell, extinction of excitation light, green and red fluorescence intensity. The quality of testis selection was assessed by inspection of gated samples with a stereomicroscope. Sorted testes were recovered in Ringer Buffer in a 50 ml Falcon tube with Ringer solution. The testes were collected by centrifugation at 4500 × g for 30 seconds. The supernatant was discarded, and the sediment was transferred to a 1.5 ml low binding Eppendorf tube (Eppendorf, #0030108116) and snap frozen in liquid nitrogen after settling of testes and removal of supernatant.

Affinity purification of EFGP fusion proteins

Affinity purification of Mst27D-EGFP and control EGFP fusion proteins was performed as described [35]. Four replicates, each with approximately 9000 testes, were processed for each protein. Testes were homogenized in 1.5 ml lysis buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 2 mM MgCl₂, 0.1% Nonidet NP-40 [Sigma Aldrich IGEPAL CA-630], 5% glycerol, 1 tablet complete protease inhibitor cocktail [Roche #04693159001]/10 ml, 1 mM DTT and 50 U/ml Benzonase) using a Potter-Elvehjem homogenizer with a glass pestle. The lysate was centrifuged at 16000 × g for 15 minutes at 4°C. The supernatant was transferred to a new low binding Eppendorf tube and the centrifugation step was repeated. Thereafter, the supernatant was added to 50 μl μMACS anti-GFP beads (Miltenyi Biotec, Solothurn, Switzerland, #130-091-125), and incubated for one hour on a rotating wheel at 4°C. The suspension was transferred into a pre-equilibrated μColumn (Miltenyi Biotec, #130-042-701). Exploiting the magnetic field provided by the magnetic stand (Miltenyi Biotec, #130-042-602 and #130-042-303), the column was washed four times with 200 μl lysis buffer and once with Tris-HCl (pH 7.5). All steps were performed at 4°C.

The PreOmics iST kit (PreOmics GmbH, Planegg/Martinsried, Germany, #P.O.00001) was used for preparation of samples for MS analyses. At first, 25 μl of LYSE buffer (PreOmics iST kit) were added to the column. The column was then removed from the magnetic stand and placed on top of a 1.5 ml Eppendorf tube. Another 50 μl of LYSE buffer were added to the column to elute the beads and proteins onto the cartridge (PreOmics iST kit). Reduction and alkylation were performed at 60°C for 10 minutes in a shaker at 1000 rounds per minute (rpm). Digestion with Trypsin plus Lys-C and the following purification steps were performed according to the manufacturer’s instructions.

Liquid chromatography, mass spectrometry and sequence comparison

Mass spectrometry was performed as described [35]. Samples were injected by an Easy-nLC 1000 system (Thermo Scientific) and separated on an EasySpray-column (75 μm x 500 mm) packed with C18 material (PepMap, C18, 100 Å, 2 μm, Thermo Scientific). The column was equilibrated with 100% solvent A (0.1% formic acid (FA) in water). Peptides were eluted using the following gradient of solvent B (0.1% FA in acetonitrile): 5% B for 2 minutes; 5–25% B in 60 minutes; 25–35% B in 10 minutes; 35–95% B in 5 minutes at a flow rate of 0.3 μl/minutes. High accuracy mass spectra were acquired with an Orbitrap Fusion (Thermo Scientific) that was operated in data dependent acquisition mode. All precursor signals were recorded in the Orbitrap using quadrupole transmission in the mass range of 300–1500 m/z. Spectra were recorded with a resolution of 120’000 at 200 m/z, a target value of 5E5 and the maximum cycle time was set to 3 seconds. Data dependent MS/MS were recorded in the linear ion trap using quadrupole isolation with a window of 1.6 Da and HCD fragmentation with 30% fragmentation energy. The ion trap was operated in rapid scan mode with a target value of 8E3 and a maximum injection time of 80 ms. Precursor signals were selected for fragmentation with a charge state from +2 to +7 and a signal intensity of at least 5E3. A dynamic exclusion list was used for 25 seconds and maximum parallelizing ion injections was activated. After data collection the peak lists were generated using Proteome Discoverer 2.1 (Thermo Scientific).

All MS/MS data were analyzed using Mascot 2.4 (Matrix Science, London, UK). MS data were searched against a decoyed database from UniProt (organism:7227, date:20131216) concatenated with an in-house build contaminant database. Precursor ion mass tolerance was set to 10 ppm and the fragment ion mass tolerance was set to 0.6 Da. The following search parameters were used: trypsin digestion (two missed cleavages allowed), fixed modifications of carbamidomethyl labelled cysteine, and as variable modification oxidation of methionine, transformation of N-terminal glutamine to pyroglutamic acid (Gln→pyro-Glu), and deamidation of asparagine and glutamine. Data was filtered using Scaffold (v4.10.0) with the settings Protein threshold 95%, Minimal number of peptides 2 and Peptide Threshold DTs. The number of the peptides displayed in Fig 1D corresponds to the rounded Quantitative Value (Normalized Weighted Spectra) provided by the scaffold software.

For the comparison of amino acid sequence similarity between human EB1 and the D. melanogaster EB1 protein family members (Fig 1C), sequences were retrieved from UniProt (Q15691 for human EB1/MAPRE1) and FlyBase (Eb1-PA, CG32371-PA, CG18190-PB, CG15306-PA, CG2955-PA and Mst27D-PA). After aligning the sequences using Clustal Omega with default settings provided by the EMBL-EBI online services, the cladogram of the guide tree was downloaded for incorporation into Fig 1C.

Cell culture, transfection and microscopic analyses

S2R+ cells were cultured as described [90]. S2R+ cells were grown in Schneider’s medium (Gibco, #21720) supplemented with 10% fetal bovine serum (Gibco, #10500–064) and 1% Penicillin-Streptomycin (Gibco, #15140) at 25°C. For transfection, S2R+ cells were seeded in a 25-cm² flask (Greiner Bio-ONE, #690160) with 2.6 × 10⁶ cells in 4 ml complete medium, or in a 35 mm glass bottom dish (MatTek Corporation, #P35G-1.5-14-C) with 6 × 10⁵ cells in 2 ml, or in a single well of 24-well plate (with a sterilized coverslip at bottom) with 1.5 × 10⁵ cells in 0.5 ml medium. After allowing cells to adhere for 1 hour at 25°C, they were transfected. The transfection mixture for a 25-cm² flask contained 2 μg of plasmids, 8 μl FuGENE HD (Promega, #E2311) and Schneider’s medium (not supplemented) added to a total volume of 200 μl; in case of a 35 mm glass bottom dish 1 μg of plasmid, 4 μl FuGENE HD and Schneider’s medium to 100 μl total volume, and in case of a single well of 24-well plate 0.25 μg plasmids, 1 μl FuGENE HD and Schneider’s medium to 25 μl total volume. The transfection mixture was incubated at room temperature for 15 minutes and added dropwise to the cells, followed by an incubation at 25°C for 48 hours. For the induction of expression from pMT plasmid constructs under control of the Metallothionein A (MtnA) promoter, CuSO₄ was added (500 μM final concentration) 24 hours after transfection followed by additional 24 hours of incubation. Cells were then either fixed or analyzed by live imaging or harvested for immunoprecipitation experiments. For the establishment of stable cell lines, medium was replaced 48 hours after transfection with fresh medium containing blasticidin S (25 μg/ml final concentration). Selection was continued for 10–16 days.

To detect expression of fluorescent fusion proteins in S2R+ cells grown on coverslip or in a glass bottom dish, cells were fixed with either 4% formaldehyde in phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 1.47 mM KH₂PO₄, 6.46 mM Na₂HPO₄, pH 7.4) at room temperature or cold methanol at -20°C for 10 minutes. After a brief wash with PBS, cells were permeabilized with 0.5% Triton-X-100 in PBS for 3 minutes. After two washes with 0.1% Tween-20 in PBS, DNA was stained with Hoechst 33258 (1 μg/ml) for 5 minutes. After two washes with PBS, around 10–20 μl mounting medium (70% glycerin, 50 mM Tris-HCl pH 8.5, 10 mg/ml propyl gallate, 0.5 mg/ml phenylendiamine) as described above was added onto a slide, and the coverslip with the cells was inverted and lowered onto the mounting medium. In case of glass bottom dishes, mounting medium was added and covered with a coverslip.

For immunofluorescent staining of MTs, cells were fixed with cold methanol. Mouse monoclonal antibody DM1A anti-α-tubulin (Sigma, #T9026; 1:8000) was used as primary antibody and Cy5-conjugated goat anti-mouse antibody (Invitrogen, #A10524; 1:500) as secondary antibody.

The slides were imaged using an inverted wide-field fluorescence microscope (Zeiss Cell Observer HS) with 40×/1.3, 63×/1.4 or 100×/1.4 oil immersion objectives. To analyze the expression of fluorescent fusion proteins by live imaging, preparations in glass bottom dishes were mounted on a spinning disc confocal microscope (VisiScope with a Yokogawa CSU-X1 unit combined with an Olympus IX83 inverted stand and a Photometrics evolve EM 512 EMCCD camera, equipped for red/green dual channel fluorescence observation; Visitron systems, Puchheim, Germany) with a 40×/1.3 or 60×/1.42 oil immersion objective at 25°C.

RNAi with cultured cells

The template for Nup358 dsRNA production was amplified from w¹ genomic DNA with the primer pair LP045/LP046. The plasmid pUR291 was used for the amplification of the template for lacZ control dsRNA production with the primer pair CL298/CL299. PCR products were purified with a gel extraction kit (QIAGEN, #28706). The purified products were used to template in vitro transcription using the MEGAscript T7 Transcription Kit (Thermo, #AM1334). The dsRNA samples were purified with the MEGAclear Transcription Clean-Up Kit (Thermo, #AM1908).

Depletion with Nup358 or lacZ dsRNA was performed with S2R+ cells stably expressing mCherry-Nup358 or Mst27D_CT-EGFP as described [91]. Cells were harvested and re-suspended in Schneider’s medium (without supplements) at a density of 1 × 10⁶ cells/ml. 1.5 ml cell suspension was mixed with 15 μg dsRNA and seeded into a 35 mm glass bottom dish. After the incubation of cells with dsRNA for 45 minutes, 3 ml complete medium was added. Cells were incubated at 25°C for 3 days before microscopic analysis.

Immunoprecipitation and immunoblotting

For co-immunoprecipitation experiments, S2R+ cells were collected by a cell scraper from two 25-cm² flasks and centrifuged at 580 × g for five minutes. All further steps were performed on ice or at 4°C. Cells were re-suspended with 1 ml PBS and transferred to a 1.5 ml low binding Eppendorf tube. This tube type was also used for all additional steps. Cells were sedimented by centrifugation at 600 × g for 5 minutes at 4°C. Cells were re-suspended in 600 μl lysis buffer (LB) and incubated on ice for one hour. Insoluble material was removed by centrifugation at 16100 × g for 15 minutes. A small fraction (10%) of the supernatant was reserved for later analysis by immunoblotting (input samples). The rest of the supernatant (540 μl) was incubated with 25 μl anti-RFP beads (ChromoTek, RFP-Trap agarose, rta-20) for 1 hour on a rotating wheel at 15 rpm. Beads were collected by centrifugation at 2500 × g for two minutes and washed 3 times with 500 μl LB by repeating the centrifugation steps. For the final wash, beads were transferred into a fresh tube. After centrifugation, beads were re-suspended in 80 μl 3x Lämmli Buffer (62.5 mM Tris-HCl, 10% glycerol, 5% β-mercaptoethanol, 3% SDS, 0.01% Bromophenol Blue) and boiled for 10 minutes at 96°C. After two minutes of incubation on ice, samples were centrifuged to sediment the beads and the supernatant was distributed into aliquots for latter analysis by immunoblotting (IP samples). All samples were snap frozen in liquid nitrogen and stored at -80°C until immunoblotting analysis.

For immunblotting, samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Input and IP samples were heated to 96°C for 5 minutes and centrifuged at 13100 × g for 5 minutes before loading the supernatant onto the SDS-PAGE gels with 6.5% or 10% polyacrylamide. Electrophoresis was performed at 80 V during about 2 hours. Prestained Protein Ladder (Thermo Fisher, #26619) was used as molecular weight marker. Thereafter, proteins were transferred from the gel to Hybond ECL nitrocellulose membranes (Amersham Protran 0.45 μm, #10600002) by tank blotting overnight using 30 V. Membranes were stained with Ponceau S solution (Sigma, P7170) to assess loading quantities and electro-transfer efficiencies. After brief washing with PBS, membranes were blocked by incubation in 5% dry milk (w/v) in PBS with 0.02% NaN₃ for 30 minutes at room temperature. The primary antibody was added into this solution for an incubation for 2 hours at room temperature or overnight at 4°C. Three washes with 5% dry milk (w/v) in PBS were performed before incubation with the secondary antibody, which was applied in 5% dry milk (w/v) in PBS at room temperature for 1 hour in a dark environment. After 3 washes with 5% dry milk (w/v) in PBS and 2 washes with PBS, 0.1% Tween-20, signals were detected with ECL reagents (WesternBright ECL, Advansta, #K-12045-D50) in an Amersham Imager 600 system.

Mouse monoclonal antibody anti-RFP (ChromoTek, #6g6) was used at 1:1000 as primary antibody and HRP-conjugated AffiniPure goat anti-mouse IgG polyclonal antibody (Jackson ImmunoResearch, 115-035-003) was used at 1:1000 as secondary antibody. This secondary antibody at the same dilution was also used to detect mouse anti-Lamin ADL67.10 (Developmental Studies Hybridoma Bank) used as primary antibody at 1:200. In addition, rabbit polyclonal antibodies anti-GFP diluted 1:1000 (ChromoTek, #pabg1) or 1:2000 (Torrey Pines Biolabs, #TP401) were used as primary antibodies and HRP-conjugated AffiniPure goat anti-rabbit IgG polyclonal antibody (Jackson ImmunoResearch, #111-035-003) at 1:1000 as secondary antibody. This secondary antibody at the same dilution was also used to detect affinity-purified rabbit anti-Mst27D N1, I1 or C1 (Moravian Biotechnology, Brno, Czech Republic) used as primary antibodies at 1:10000, 1:3000 and 1:2000, respectively.

Fixed preparations of testes

For whole mount preparations, testis dissection was performed in testis buffer (183 mM KCl, 47 mM NaCl, 10 mM Tris-HCl, pH 6.8). For each slide, 10–20 testes were dissected. For permeabilization, testes were incubated twice in PBST-DOC (PBS containing 0.3% Triton-X100 and 0.3% sodium deoxycholate) for 15 minutes each. Testes were fixed in PBS containing 4% formaldehyde and 0.1% Triton X-100 for 20 minutes on a rotating wheel. After a rinse in PBS containing 0.1% Triton X-100 (PBST), blocking solution (PBST containing 5% fetal bovine serum) was added for 30 minutes. Thereafter, primary antibody was added into fresh blocking solution for incubation overnight at 4°C, or for two hours at room temperature on a rotating wheel. Testes were then washed four times in PBST for 15 minutes each. Secondary antibody was added in blocking solution for one hour at room temperature. After four washes with PBST, testes were incubated for 10 minutes in PBST containing 1 μg/ml Hoechst 33258. After three additional washes with PBS, testes were transferred into 10 μl of mounting medium on a slide before adding a cover slip. For staining of F-actin (S7A and S11 Figs), Alexa647-conjugated Phalloidin (Invitrogen, #A22287) was applied at 1:500 for 30 minutes at room temperature after the incubation with secondary antibody and before DNA staining. To determine the expression pattern and subcellular localization of Mst27D, affinity-purified rabbit antibodies N1, I1 or C1 against Mst27D were used as primary antibodies at a dilution of 1:1000. Alexa488-conjugated goat anti-rabbit IgG (Invitrogen, #A11008) was used as secondary antibody at 1:500 (Fig 5). Beyond the specific signals illustrated in Fig 5, the different anti-Mst27D antibodies resulted in signals that did not represent binding to Mst27D, as revealed by staining of Mst27D null mutants, to variable degrees at various locations within the testes. Specific anti-Mst27D signals were stronger in whole mount preparations compared to squash preparations of testes.

Testis squash preparations were made and stained essentially as described previously [92], according to protocol 3.3.2, using the mounting medium described above. Before fixation and immunofluorescent labeling, late larval and pupal testes were dissected in testis buffer. Testes were transferred to a drop of testis buffer (20 μl) on a poly-L-lysine-coated slide and opened with a tungsten needle. Cysts were dispersed and flattened by placing a 22 mm x 22 mm cover slip onto the drop for two minutes before freezing the preparation in liquid nitrogen. Thereafter, the cover slip was removed with a surgical scalpel and the slide was immersed into ethanol for 10 minutes at -20°C. The region with the squashed testes was fixed by adding 0.4 ml of 4% formaldehyde in PBS for 10 minutes within an area marked by a hydrophobic marker pen. Permeabilization was done twice in PBST-DOC for 15 minutes each. After a rinse in PBST, blocking solution (PBST containing 5% fetal bovine serum) was added for 30 minutes. Primary antibody was added in the marked area and the slide was then incubated in a humid chamber overnight at 4°C, or for two hours at room temperature. After a quick rinse with PBST, slides were incubated in PBST for 15 minutes four times. The secondary antibody was added for one-hour at room temperature. After washing with PBST, Hoechst 33258 (1 μg/ml in PBS) was applied for DNA staining for five minutes followed by three washes with PBS. Finally, the preparation was mounted with a drop of mounting medium and a cover slip.

Images were acquired using a Zeiss Cell Observer HS wide-field microscope. Z-stacks were acquired with 40×/1.3, 63×/1.35 and 100×/1.42 oil immersion objectives for analysis at high resolution. Spacing between focal planes was 280 nm. To visualize the structure of the investment cones (S7A and S11 Figs), images were acquired using an Olympus FluoView 1000 laser-scanning confocal microscope using a 40x/1.3 oil or 60×/1.42 objective and a Kalman filter of 3.

Live imaging of testes

Time-lapse imaging of cysts released from testes was performed as described [93]. Depending on the stages to be analyzed, testes were dissected from pre-pupae or 1-day old adult males in complete Schneider’s Medium in a depression well. Pre-pupal testes were separated from fat body remnants and transferred into 45 μl of medium in a 35 mm glass bottom dish. In the case of adult testes, 150 μl of medium was used. Each testis was ruptured with tungsten needles to release and separate cysts. Thereafter, 15 μl of 1% w/v methylcellulose (Sigma, #M0387) in complete Schneider’s medium was applied to pre-pupal testes preparations and 50 μl to adult testes preparations to settle down the floating samples. A filter paper soaked with water was then placed along the inner wall of the dishes. The lid was closed and sealed with parafilm before mounting the dish on the stage of a spinning disc confocal microscope. Imaging was performed at 25°C in a room with precise temperature control. Images were acquired with a 100×/1.4 or 60×/1.42 oil immersion objective, and the acquired z-stacks comprised 30 to 40 focal planes spaced by 500 nm usually recorded at 2 min time intervals except for S6B Fig (45 sec intervals), Figs 7A–7C, 7E and S10A (90 sec intervals).

For live tubulin staining in S2R+ derived cell lines, SPY555-tubulin [94] (Spirochrome, obtained from Lubio Science, #SC203) was added to the preparations at a final dilution of 1:1000. Demecolcine (Sigma, #D6165) was dissolved in dimethyl sulfoxide (10 mM) before further dilution of this stock solution to a final concentration of 10 μM in complete Schneider’s Medium. After demecolcine addition, samples were immediately mounted on the microscope and imaging was started about 20 minutes after demecolcine addition.

For the FRAP and Dendra2 photoconversion experiments, an Olympus FluoView 1000 laser-scanning confocal microscope was used. For FRAP with Eb1-tdGFP or EGFP-alphaTub84B, testes were isolated from 1-day-old males with a genotype that was either w*; EB1-tdGFP(II) or w*; P{w+, Ubq11-EGFP-alphaTub84B}; P{w+, gMst27D-mCherry} III.7. Image stacks consisting of three z-sections spaced by 500 nm were acquired. A circular ROI was selected to cover around a quarter of the medial region of the dense complex in spermatids with nuclei that were already considerably elongated. After acquisition of five initial image stacks, photobleaching was completed during one second with 405 nm light using 100% intensity. Thereafter, 20 additional image stacks at three seconds intervals were acquired in case of Eb1-tdGFP and 50 at 20 seconds intervals in case of EGFP-alphaTub84B. In the Mst27D-Dendra2 photo-conversion experiments, 20 image stacks with 10 z-sections spaced by 500 nm were acquired at three minutes interval. A circular ROI was selected to cover around a quarter of the NE rim and photo-conversion was started manually during acquisition at the second time point using 405 nm light with 1.0% laser intensity during one second.

Image processing and analysis

Maximum intensity projections were generated using the software Fiji or ZEN (version 2.3) (Carl Zeiss AG, Oberkochen, Germany) for wide-field images and IMARIS (Bitplane; versions 8.4.0, 9.2.0, 9.7.2) for most of the spinning disk and laser scanning confocal images. For export of projections as movies or as still frame images after live imaging the option of interpolated image display was activated when using IMARIS. Moreover, sample movements were corrected using drift correction with IMARIS before exporting the S3–S3 Movies.

For quantification of the effects of Nup358 degradation by deGradFP in S6 spermatocytes (Fig 4), single optical sections through the equatorial region of the nucleus were used for quantification of signal intensities with Fiji. Five ROIs (cytoplasm left, NE left, nuclear center, NE right, cytoplasm right) were chosen as illustrated (Fig 4D). For ROI selection, a line with a width set to 16 was drawn through the center of the nucleus. Signal intensities along this line were inspected to identify the position where the line crosses the NE on the left and on the right of the nuclear center based on local EGFP-Nup358 intensity maxima. The values of five positions along the line centered around the position of maximal intensity value were averaged for determination of signal intensities at the NE (NE_left and NE_right, respectively). For background correction of these signal intensities, the intensities at the five central positions along the line (nuclear center) were averaged and subtracted. For estimation of the signal intensities in the cytoplasm, the average value of a window of five positions along the line shifted away from the NE by around 20 pixels was determined, and for background correction of these cytoplasmic signal intensities we also subtracted the average of the intensity values in the five central most positions (nuclear center). For quantification of the effects of Nup358 depletion by RNAi (S3C Fig), an analogous approach was used. Single optical sections through the equatorial region of S2R+ cells expressing Mst27D_CT-EGFP were quantified. Integrated intensity of the Mst27D_CT-EGFP signals was measured in the ROIs selected in the cytoplasm, at the NE and in the central region of the nucleus. For each cell, the intensities in the two cytoplasmic ROIs were averaged, as well as those in the two ROIs on the NE. The intensities observed within the ROI in center of the nucleus were considered to reflect background, which was subtracted from the intensities detected in the cytoplasm and at the NE.

For the quantification of Mst27D-EGFP signal intensities (S8 Fig) resulting with g-Mst27D-EGFP and g-Mst27D-EGFP-endo3’, respectively, maximum intensity projections of image stacks covering the entire cluster of nuclei of a given cyst were analyzed with Fiji. In the case of cysts with early spermatids, an oval ROI was selected manually around a given nucleus. In the case of late spermatids, ROIs around the entire cluster of nuclei were selected manually using the freehand selection tool. The integrated intensity within this ROI was measured. Thereafter, the size of the ROI was dilated by the width of 3 pixels. The integrated intensity measured within this larger ROI was used for background correction using the following formulas: background intensity per pixel = (RawIntDen in large ROI–RawIntDen in small ROI/(area of large ROI–area of small ROI); background corrected intensity = RawIntDen of small ROI—(background intensity per pixel × area of small ROI).

For quantification of signal intensities obtained in the FRAP or photo-conversion experiments, the ROI used for bleaching and photo-conversion, respectively, was also used for signal quantification at time points before and after bleaching/photo-conversion. However, the position of the ROI was manually adjusted at each time point to compensate sample movements. Sum slice projections were made with Fiji for the quantification of the Mst27D-Dendra2 signal intensities (S4 Fig).

Data obtained after quantification with Fiji or IMARIS was exported and further processed using Microsoft Excel or GraphPad Prism (8.0.1) for plotting. P values were calculated using the Wilcoxon rank sum test (* = p < .05; ** = p < .01; *** = p < .001). Adobe Photoshop, Illustrator or Inkscape were used for assembly of figures.