Separation and Loss of Centrioles From Primordidal Germ Cells To Mature Oocytes In The Mouse

Oocytes, including from mammals, lack centrioles, but neither the mechanism by which mature eggs lose their centrioles nor the exact stage at which centrioles are destroyed during oogenesis is known. To answer questions raised by centriole disappearance during oogenesis, using a transgenic mouse expressing GFP-centrin-2 (GFP CETN2), we traced their presence from e11.5 primordial germ cells (PGCs) through oogenesis and their ultimate dissolution in mature oocytes. We show tightly coupled CETN2 doublets in PGCs, oogonia, and pre-pubertal oocytes. Beginning with follicular recruitment of incompetent germinal vesicle (GV) oocytes, through full oocyte maturation, the CETN2 doublets separate within the pericentriolar material (PCM) and a rise in single CETN2 pairs is identified, mostly at meiotic metaphase-I and -II spindle poles. Partial CETN2 foci dissolution occurs even as other centriole markers, like Cep135, a protein necessary for centriole duplication, are maintained at the PCM. Furthermore, live imaging demonstrates that the link between the two centrioles breaks as meiosis resumes and that centriole association with the PCM is progressively lost. Microtubule inhibition shows that centriole dissolution is uncoupled from microtubule dynamics. Thus, centriole doublets, present in early G2-arrested meiotic prophase oocytes, begin partial reduction during follicular recruitment and meiotic resumption, later than previously thought.


Primordial germ cells and oogonia in fetal ovaries contain GFP CETN2-labeled centrioles.
Since centriole elimination is reported to occur in fetal stages 4 , we first investigated primordial germ cells (PGCs) in the nascent gonads in GFP CETN2-expressing CB6F1 females, with sex identified by y chromosome PCR (see Methods and Materials), and traced their history through fetal oogonia formation in developing ovaries prior to birth. Isolated PGCs from e11.5 post-coitus gonadal tissues were identified from undifferentiated or somatic cells by the expression of the germ line cell surface marker Stage-Specific Embryonic Antigen-1 (SSEA-1; Fig. 1) 46 . Initial findings on this transgenic mouse reported GFP CETN2 expression from e14.5 forward (Jackson Laboratory Mouse Database Description at https://www.jax.org/jax-mice-and-services#) 45 , so we were initially surprised to find that all e11.5 PGCs also contained GFP CETN2-labeled centrioles within microtubule arrays, whether mitotic or interphase ( Fig. 1A-D). When PGCs stopped mitotic divisions, and entered pre-meiotic G2 arrest around e12.5 post-coitus, twin pairs of GFP CETN2 foci, in association with microtubules, were identified in SSEA-1-positive e14.5 and e18.5 oogonia, as well as in surrounding somatic cells (Fig. 1E-H). We conclude that GFP CETN2 centrin detects centriole pairs throughout fetal development, including their presence in female primordial germ cells.
Diminished GFP CETN2 punctate foci, perhaps centriole remnants, are typically observed in metaphase-I and -II spindle poles. We traced the fate of the GFP CETN2-labeled structures during meiotic maturation. At meiotic resumption, separated GFP CETN2-labeled pairs embedded in γ-tubulin ribbons were observed at the assembling spindle poles ( Fig. 2A-G). Cumulus cells, serving as an internal somatic cell control, also showed GFP CETN2-labeled centriole pairs within γ-tubulin PCM, including at the spindle poles of a rare bipolar mitotic cell ( Fig. 2A,C: arrowheads). Using correlative light and electron microscopy (CLEM), we investigated whether classical 9-triplet microtubule organization could be identified in oocytes at this stage (Suppl Fig. S1). Analysis revealed linear MTOC-like structures with associated GFP CETN2 foci (Suppl Fig. S1B, inset) in live oocytes. EM sectioning through the imaged site, however, showed only linear MTOC structures and multi-vesicular PCM aggregates (13; Suppl Fig. S1B-F), but no classical centrioles with microtubule walls, despite careful examination of hundreds of 80-nm-thick sequential serial sections. Typical canonical centrioles were identified in adhering cumulus cells attached to the same oocyte (Fig. 2E, inset; Suppl Fig. S1G-I). Collectively, the data suggest that significant oocyte centriole dissolution begins with meiosis onset. Altered oocyte centrioles still display GFP CETN2 labeling but are either too fragile to survive EM processing protocols or have undergone partial disassembly. This observation is supported by finding classical centrioles with 9 + 0 triplet microtubules in the surrounding somatic cumulus cells. Support for oocyte centriole alterations occurring during meiotic resumption was also derived from measuring changes in PCM and GFP CETN2 foci area sizes (in μm 2 ) during maturation (Suppl Fig. S2D,E). After germinal vesicle breakdown (GVBD) and through complete maturation to metaphase-II arrest, we measured  shows the presence of the germ cell marker SSEA-1 (red) at the cell periphery, along with two pairs of GFP CETN2 foci (inset) on opposite sides of the condensed prometaphase chromosomes (green, arrowheads). The lower somatic interphase gonadal cell lacks SSEA-1, with a single pair of GFP CETN2 focus (inset) adjacent to the nucleus (green, arrowheads). All GFP CETN2 centrioles are associated with microtubules, either at the center of microtubule asters at each nascent spindle pole (B: upper cell, red) or assembled on the nuclear surface connected to the more robust cortical microtubule interphase network (B: lower cell, red). (C,D) A divided e11.5 dpc primordial germ cell with SSEA-1 at the cell periphery of both daughter cells (C: red) and expressing GFP CETN2 centrioles (C: green, arrowheads) adjacent to the nuclei (C: blue, DNA). Microtubule assembly is observed only from the top daughter cell (D: red). (E,F) Early mouse oogonium from an e14.5 dpc gonad with cortical SSEA-1 (E: red) and GFP CETN2 centriole pairs (E: green, arrowheads; blue, DNA). Cortical microtubule bundles are pronounced at the site of the GFP CETN2 centriole pair (F: red, microtubules, arrowheads). (G,H) Cells isolated from the sex cord of an e18.5 dpc fetal gonad, showing a single oogonial cell labeled with SSEA-1 (G: red, arrow) with a pair of GFP CETN2 centrioles (G: green, double arrowheads) at the nucleus (blue, DNA) but without apparent microtubule assembly (H: red, double arrowheads). The remaining SCIeNtIfIC RePORTS | (2018) 8:12791 | DOI:10.1038/s41598-018-31222-x increased γ-tubulin area as the GV-associated MTOC foci were stretched into expanded ribbons and fragmented, as previously reported 16,18 (Suppl Fig. S2D). However, significantly decreased GFP CETN2 areas were measured after GVBD (Suppl Fig. S2E), suggesting structural changes in oocyte centrioles with diminishing GFP CETN2 detection. Interestingly, areas for GFP CETN2 centrioles in somatic cumulus cells were consistent in primordial follicles, growing antral follicles, and surrounding mature GVs after follicular release (0.47 ± 0.17, 0.42 ± 0.23, and 0.43 ± 0.19 µm 2 , respectively). Despite reduced areas in GFP CETN2 foci after meiotic resumption, most spindle poles in metaphase-I and -II oocytes still contained GFP CETN2 foci at the ends of γ-tubulin short segments or foci on the outer edges of the barrel-shaped spindles ( Fig. 2H-W, Suppl Fig. S2F).
Tracking GFP CETN2 foci in first and second meiotic spindles showed many localization patterns. A minority of oocytes at metaphase-I or -II were found with GFP CETN2 foci at both poles ( Fig. 2H-O), even after first polar body extrusion (Fig. 2P-W; panels enhanced in Suppl Fig. S8). Most metaphase-I spindles, however, contained single GFP CETN2 foci at the extreme edges of the barrel-shaped spindle (Suppl Fig. S3A-G), unlike centrosomes in somatic cell spindles that reside more centrally in the fusiform spindle pole. Diminished GFP CETN2 foci association with the γ-tubulin PCM was also observed at this stage, sometimes with foci residing within the spindle lattice near the aligned bivalents (Suppl Fig. S3H-V). In metaphase-II-arrested oocytes, a minority of GFP CETN2 foci were observed in the extruded first polar body (Pb1) or in the microtubule-based asters (cytasters) [20][21][22] in the cytoplasm (Suppl Fig. S4). However, the majority of GFP CETN2 foci were found in association with meiotic spindles throughout maturation to metaphase-II arrest (Suppl Fig. S2F). Taken together, our analyses suggest that centrioles show significant reduction during meiotic progression, looser association with the PCM, and could not be identified as traditional centrioles with 9-fold symmetry in EM serial sections, consistent with centriole dissolution prior to the onset of meiosis. However, GFP CETN2 foci identified in fetal stages survive in mature oocytes and are found in γ-tubulin-containing meiotic spindle poles through metaphase-II arrest following meiotic resumption. GFP CETN2 foci are found in atypical spindle pole positions relative to traditional localization of centrioles in somatic cells.
Our analysis of the fate of GFP CETN2 centriole doublets, traced from e11.5 PGCs isolated from the genital ridge through mature metaphase-II-arrested oocytes is summarized in Fig. 3 and Supplemental Fig. S5. GFP CETN2 foci associated with γ-tubulin or pericentrin PCM in PGCs, fetal oogonia, and early oocytes from P4 neonate ovaries showed double pairs of GFP CETN2 centrioles (Fig. 3a- Interestingly, around 3-10% of incompetent or competent GV oocytes showed more than two doublet pairs ( Fig. 3d-g, graph, green bars). No evidence of centriole duplication in these quiescent oocytes was found, suggesting these aberrant oocytes were produced by fetal-stage mitotic errors or, perhaps, by early oocyte cyst breakdown inconsistencies. Often, oocytes with abnormal doublets showed widely dispersed GFP CETN2 foci in the cytoplasm or cortex or along the GV surface, with distances >12 µm generally observed. Regardless, centriole dissolution appears to accelerate following meiotic resumption, with an increase in the number of single GFP CETN2 pairs at the spindle poles of after GVBD ( Fig. 3f-h, graph; Fig. 3F-H, lower panel). Collectively, the data show that GFP CETN2 centrioles embedded within PCM are maintained as two doublets until follicular recruitment in post-adolescent females, when GV-arrested oocytes within growing follicles first separate into GFP CETN2 pairs and, later, single-pair dissolution occurs. Following meiotic resumption, single-centriole-pair dissolution accelerates. No evidence of centriole duplication was found during meiotic maturation. Of the metaphase-II-arrested oocytes with detectable GFP CETN2 foci at the spindle poles (~38%; Suppl Fig. S2A), nearly all showed diminished single pairs at their spindle poles. Immature, growing, and mature oocytes contain GFP CETN2 foci in γ-tubulin PCM. Fixed neonatal Day-4 post-birth female ovaries (P4), a stage at which centrioles have been identified by EM 10 , showed two tightly apposed doublets of GFP CETN2 foci within γ-tubulin PCM, adjacent to the GV (Fig. 4A). Pre-antral and early antral follicles in adult ovaries maintained GFP CETN2 foci, with some displaying doublet pair separation, beginning with antral formation (Fig. 4B,C). Some cytoplasmic GFP CETN2 aggregates appeared in early follicular stages and could be distinguished from GFP CETN2 foci in centrosomes by their exclusion from co-localization with γ-tubulin.
Tracking GFP CETN2 doublet fate in ovarian sections is challenging and provides the rationale for isolating immature oocytes of assorted sizes during follicular growth (Suppl Fig. S6). Incompetent GV oocytes, which could not initiate meiotic maturation, showed GFP CETN2 foci embedded within γ-tubulin PCM in all oocytes, mostly without microtubule aster assembly (Fig. 5A). Only 70% of fully competent mature GV oocytes maintained GFP CETN2 foci with γ-tubulin (Suppl Fig. S2A). Although numerous GFP protein aggregates in growing GV oocytes were present, these supernumerary foci did not interact with γ-tubulin or pericentrin, nor assemble microtubules. The GFP CETN2 foci in association with γ-tubulin or pericentrin maintained consistent area sizes during oocyte growth, although their distribution within the oocyte changed from the GV nuclear surface in small follicle GVs to the cytoplasm during mid-follicular growth, before returning to the GV surface just before meiotic resumption (Fig. 5A7-A8; enhanced image in Suppl Fig. S8; graphic analysis, Suppl Fig. S2B,C). somatic interphase gonadal cells are SSEA-1-negative but express GFP CETN2 centriole pairs (G: green foci) at their nuclear surfaces near the site of interphase microtubule assembly (H: red, microtubules). Confocal images of GFP CETN2 (green) are counterstained with antibodies to SSEA-1 (A,C,E,G: red), microtubules (B,D,F,H: color assigned red) and DNA (blue). Insets in (A,C,E,G) details of GFP CETN2 centrioles. dpc: days post-coitus. Bars as marked.
GFP CETN2 foci (Fig. 5C). Collectively, GFP CETN2 foci, present from early pre-antral follicle oocytes, are maintained through follicular growth to full maturity. These structures remain within the γ-tubulin or pericentrin PCM as it undergoes decondensation, stretching, and fragmentation along the GV surface in preparation for meiotic spindle assembly.
Since GFP CETN2 doublets appear to separate during GV growth, we investigated whether the expression of Cep135, a centrosomal scaffolding protein that anchors cNAP1 in somatic cell centrioles and whose disruption or overexpression causes premature centrosome separation 47 , might be the underlying mechanism (Fig. 6). Adhering cumulus cells, as well as incompetent GVs isolated from small, medium, and large growing follicles, showed bright GFP CETN2 foci co-localized with pericentrin and Cep135 ( Fig. 6A-F). As oocytes transitioned toward meiotic resumption, GV oocytes expanded and the MTOC fragmented along the GV surface as Cep135 detection was significantly reduced in the PCM and the GFP CETN2 doublet pairs separated ( Fig. 6G-J). cNAP1, anchored by Cep135 at somatic cell centrioles, has been identified in mouse oocyte MTOCs 48 , and PLK1 has been implicated in release of cNAP1 to permit MTOC expansion in the mouse GV oocyte 18 . Therefore, we utilized the PLK1 inhibitor BI 2536 to investigate whether the inhibitor impacts Cep135 localization at MTOCs and subsequent GFP CETN2 doublet separation. A 2-hr treatment of mature GV oocytes with BI 2536 inhibitor blocked Cep135 reduction from MTOCs at the GV surface, which still slightly expanded (Fig. 6K). Remarkably, despite Cep135 retention in the MTOC, the GFP CETN2 doublet pairs separated to opposite ends of the PCM (Fig. 6K-M). Even the combined treatment of BI 2536 with the microtubule inhibitor nocodazole, which blocks MTOC Figure 3. Tracking GFP CETN2 centriole loss from PGCs through maturation to metaphase-II arrest. Nested bar depiction of oocyte populations in PGCs, oogonia, adolescent (P4) primary oocytes, various adult oocytes during follicular growth, and meiotic oocytes showing the percentage of GFP CETN2 foci as doublets (1 pair), two doublets (2 pair), or > two doublets (i.e. more than 3 doublets) at each stage. Tightly apposed GFP CETN2 doublets in association with PCM are observed in all e11.5 PGCs, fetal stage oogonia (e14.5 and e18.5), and neonate early oocytes prior to sexual maturity (graph: a-c, red bars; image panel: 3A-C, green). In adult ovaries, however, follicular recruitment under the influence of sex hormones alters the pattern of GFP CETN2 doublets during oocyte growth to full maturity, reducing both the number of doublets (graph: d-e, red bars) and their tight association (image panel: D-E, green; arrows point to GFP CETN2 doublets; arrowheads: GFP aggregates). GFP CETN2 doublets continue to decline with the onset of meiotic maturation (graph: f-h, red bars), with a significant increase in GFP CETN2 doublets or single pairs within the PCM at a single spindle pole in metaphase-I and -II oocytes (graph: f-h, blue bars; image panel: F-H, green). Other GFP CETN2 configurations include multiple doublets observed in immature, fully grown, and early prometaphase-I oocytes (graph: d-f, green bars), but these are significantly decreased by the end of first meiosis and not visible in metaphase-II-arrested oocytes. Minimum of three trials, except for e14.5 and primary follicles (two trials). Image panel: direct detection of GFP CETN2 doublets and singles after fixation and co-labeling with PCM antibody (γ-tubulin or pericentrin; not shown) and DNA (not shown). Images are at similar exposure and size for comparative analysis. The drop in GFP signal background (F-H) reflects dilution of GFP signal as the oocyte volume increases during follicular growth. Bar = 0.4 µm.
SCIeNtIfIC RePORTS | (2018) 8:12791 | DOI:10.1038/s41598-018-31222-x expansion and fragmentation, showed slightly elongated MTOCs on the GV surface, with retained Cep135 in the PCM and GFP CETN2 foci separated into four pairs ( Fig. 6N-P). Collectively, these results suggest that GFP CETN2 doublet separation does not require the loss of Cep135 centrosomal core protein at the MTOC. Centriole doublet separation and dissolution may involve other unknown mechanisms occurring upstream of meiotic resumption. GFP CETN2 foci present a lose association with the PCM. To verify that the GFP CETN2 foci observed in fully grown mouse oocytes were not an artifact due to fixation, we also followed them in living oocytes. These investigations, which were performed in France on similar and identical mouse strains, obtained independently, provide confirmation of the reliability of both the research resources and the experimental strategies. For these experiments, oocytes expressing mCherry-Plk4 (either from transgenic mice or from cRNA injection) 49 , a PCM marker 19 , together with GFP or Venus-tagged CETN2 (also from either transgenic mice or cRNA injection) were utilized. No differences in behavior or localization were observed between cRNA and transgenic expression of these constructs (Fig. 7, compare A and D). Live imaging also revealed the presence of GFP CETN2 doublets in follicular cells surrounding the oocyte coming from transgenic GFP CETN2 mice (Fig. 7A upper panels) as well as GFP CETN2 foci co-localizing with the cortical PCM in incompetent oocytes (Fig. 7A lower panel). The distance between the two doublets increased while the major PCM foci moved toward the nuclear envelope (Fig. 7B), arguing that anchoring of the major MTOC to the nuclear envelope favored the building of forces able to perturb its internal architecture, as observed by Luksza et al. 16 . This distance increased even further in competent oocytes able to resume meiosis (Fig. 7C). This is fully consistent with observations made in fixed oocytes (Suppl Fig. S5). Importantly, as meiosis I progressed, live imaging showed that the association between GFP CETN2 foci and the PCM, labeled with mCherry-Plk4, was weak ( Fig. 7D and Suppl Movie S1) at all stages of meiosis I spindle assembly (Suppl Movies S2 and S3). Foci of GFP CETN2 could be extruded into the first polar  (Fig. 7D and Suppl Movie S1) as also observed in fixed oocytes (Fig. S2F). Collectively, live imaging demonstrated GFP CETN2 foci in association with organize PCM, although co-localization appears to be progressively lost as the oocytes resumed meiosis.  GFP CETN2 foci that function in meiotic spindle reassembly after microtubule inhibition are present in PCM. Since meiotic spindle assembly is initiated by multiple MTOCs, we explored whether GFP CETN2 foci within fragmented MTOCs would reassemble microtubules following disruption with the microtubule inhibitors nocodazole and paclitaxel (Fig. 8). In prometaphase-I oocytes, 3.5-hr nocodazole exposure increased MTOC volumes near the condensed bivalents with embedded GFP CETN2 foci but no microtubule assembly (Fig. 8A-F). Upon a 5-min rescue from nocodazole, the MTOC with GFP CETN2 foci expanded and assembled a large, well-arrayed microtubule aster, suggesting that GFP CETN2-containing centrosomes can function as MTOCs that could potentially participate in spindle assembly (Fig. 8G-L).
Paclitaxel enhances microtubule assembly. After GVBD, paclitaxel initiated microtubule aster assembly around the GV nucleus from fragmented, dispersed MTOCs. The largest MTOC retained two embedded GFP CETN2 foci from which the best organized microtubule aster assembled next to the condensing bivalents ( Fig. 8M-R). All other identified cytoplasmic and cortical microtubule asters assembled from MTOCs but without detectable GFP CETN2 foci (Fig. 8R). In later prometaphase-I, GFP CETN2 foci within many small fragments of γ-tubulin PCM were observed around the circular bivalents and extensive bundles of microtubules assembled outward toward the cortex in a large ring pattern (Fig. 8S-X). Thus, paclitaxel-induced microtubule enhancement prevents PCM assembly at opposing spindle poles. GFP-expressing CETN2 foci remain with the scattered MTOCs around the condensing bivalents, but these centrosomes neither nucleate microtubules nor assemble bipolar spindles.
Analysis of changes to GFP CETN2 foci areas during exposure to nocodazole and paclitaxel is presented in Supplemental Fig. S7. GV oocytes exposed continuously to nocodazole or paclitaxel for 3.5 hrs in the presences of dbcAMP to retard meiotic resumption showed significant decreases in overall GFP CETN2 foci areas, perhaps indicating GFP CETN2 foci instability in the face of altered microtubule dynamics (Suppl Fig. S7A, green bars). Similar observations were seen when dbcAMP was removed to permit GV oocytes to progress to prometaphase-I in the presence of nocodazole but not after paclitaxel exposure (Suppl Fig. S7A, blue bars). Regardless, GFP CETN2 foci areas decreased following resumption of meiotic maturation in all microtubule inhibitor treatments when analyzed with their corresponding GV controls (Suppl Fig. S7B). Collectively, disruption of microtubule assembly or disassembly reduces GFP CETN2 areas, even in arrested GV oocytes, suggesting that structural alterations have already occurred prior to oocyte full maturation that render the oocyte centrioles susceptible to changes in microtubule dynamics before meiotic resumption.

Discussion
The transgenerational contribution of centrosomes and centrioles was thought to have been solved by Boveri 50,51 who wrote that: "The ripe egg possesses all of the elements necessary for development save an active division-center (centriole). The sperm, on the other hand, possesses such a center but lacks the protoplasmic substratum in which to operate". Later, the pioneering ultrastructural imaging of centrioles by Fawcett and Phillips 52 , previously depicted as mere dots, recognized that centrioles and basal bodies were composed of 9-triplet microtubules. However, predictions that molecular genetic approaches would answer definitively inheritance questions, following on the discovery of DNA in mitochondria 53,54 , failed. Without a permanent tracer molecule, as DNA in mitochondria, it has not been possible to answer this question conclusively, until recently. Now, breakthroughs in understanding the nature and dynamism of the core protein components of both centrosomes and centrioles 55 , and the realization that their assembly is tightly regulated and orchestrated provide the tools to address inheritance. With considerable simplification, centrioles assemble upon a SAS-6 cartwheel 56 , upon which the other components of the 9-triplet structures assemble and onto which centrosomal PCM later concentrates 36 . In consequence, the question as to whether centrosomes and centrioles are inherited, strictly speaking, may need to be examined more stringently. Findings demonstrating that they are elaborate, dynamic protein assemblies support the concept of "transmission", as thoughtfully proposed by Ross and Normark 57 . Without the DNA of chloroplasts, mitochondria and other plastids, perhaps the term "inheritance" should not be applied casually to organelles, assembled at times de novo, which lack their own genomic material.
Centriole disassembly, studied here in oocytes during the last two meiotic divisions, remains challenging, though elimination mechanisms have been discovered in flies 58 and starfish 59 . In mice, two centriole pairs persist from fetal stages 11 and in oocytes within adolescent ovaries 10 . During maturation, the centriole pairs first separate within the surrounding PCM, with the disassembly of one of the two doublets prior to full oocyte growth. Centriole adhesion does not appear to depend on the Cep135 centriolar core protein that anchors cNAP1 at the centrioles. Centriole dissolution accelerates upon meiotic resumption, as the number and area significantly decrease following GVBD ( Fig. 3; Suppl Fig. S2E). Perhaps these centriole-like remnants anchor the stretching PCM to the GV surface 16,18 . Later, the metaphase-I and -II spindles typically display at least a single pair near or at inhibitor, showing reduced MTOC stretching and fragmentation (red, arrowheads) with retained Cep135 (K: red) and pericentrin (L: red). Despite Cep135 retention, the GFP CETN2 doublets separate to opposing MTOC ends (green, arrows). (M) Magnified view of GFP CETN2 (green, arrows), Cep135 (blue, arrowheads), and pericentrin (red) organization. (N-P) A mature GV co-incubated 2 hrs simultaneously in BI 2536 and the microtubule inhibitor nocodazole (10 µM). The elongated MTOC retains highly co-localized Cep135 and pericentrin (red, arrowheads), and the GFP CETN2 doublets separate with the PCM (green, arrows). (P) Detailed view of Cep135 (blue), pericentrin (red), and split GFP CETN2 doublets (green, arrows). All images are triple-labeled for pericentrin (red), Cep135 (color assigned red in A,D,G,K,N; blue in C,F,I,J,M,P), and DNA (blue). GFP CETN2 (green) is by direct fluorescent detection. Insets: details of GFP CETN2 foci with Cep  135 (A,D,G,K,N) or pericentrin (B,E,H,L,O). Bars, as marked.
SCIeNtIfIC RePORTS | (2018) 8:12791 | DOI:10.1038/s41598-018-31222-x a spindle pole. Overall, this suggests that the maternal centriole is not lost until the completion of oocyte maturation, as in starfish oocytes 59,60 . Centrioles were found in first or second mouse meiotic spindles neither by Szollosi et al. 7 , nor in our CLEM investigations here. Locating centrioles within an oocyte is akin to looking for a needle in a haystack, what with the enormous cytoplasmic volume and the lack of any fiduciary marks to help locate the centriole precisely. Here however, the CLEM approach defined the site of the GFP CETN2 pairs, and still, only osmiophilic PCM was observed (Suppl Fig. S1). This suggests that the canonical 9-triplet centriole structure disassembles prior to meiotic maturation, while the adjacent cumulus cells with intact centrioles provide ideal controls (Fig. 2E, inset and Suppl Fig. S1).
The GFP CETN2 doublets identified here likely fall into the category of centrioles described as "advertisements, " "remnants", "passenger" or "zombie" structures [61][62][63][64] ; i.e., vestigial, non-functional, and non-replicating centrioles. Perhaps centriole alterations mirror events seen in sperm distal centriole dissolution, where proximal centriole-like structures identified in the sperm of flies and humans lose their 3-dimensional 9-triplet microtubule  : two GV oocytes cultured for 3.5 hrs to induce GVBD before exposure to 1-10 µM PT for 15-30 min. An early prometaphase-I oocyte (M-R) shows two GFP CETN2 doublets (green, arrows) in fragmenting γ-tubulin (red) around condensing bivalents (blue). Enhanced microtubule aster assembly (P: red) is observed from the largest remaining γ-tubulin PCM (M: red) having embedded GFP CETN2 centrioles (green, arrows). Abundant cortical cytasters (P: red, arrowheads) are also present. In later-stage prometaphase-I oocytes (S-X: blue), GFP CETN2 doublets separate (S: arrows) within fragmented γ-tubulin PCM (S: red). Microtubule assembly radiates from the cytoplasm outward toward the cortex, not from the GFP CETN2 foci within γ-tubulin PCM (V: red). (N,Q), T1-T3, W1-W3: GFP CETN2 (green) in γ-tubulin PCM (N, T1-T3: red) or microtubules (Q, W1-W3: red) details. All structure, but retain PCM antigens and the ability to organize into functional MTOCs under the right conditions, such as after fertilization in the activated zygotic cytoplasm 64,65 . Indeed, the destruction of first one, and then the second, centriole in mouse sperm as these transit through the epididymis 25 underscores the complex behavior of centrosomes and centrioles in gametes even after they depart from the gonad. While these GFP CETN2 doublets in maturing oocytes appear non-functional, their microtubule nucleating activity can be revived after recovery from nocodazole microtubule disassembly (Fig. 8). Upon this revival, robust microtubule arrays reassemble around the now reactivated GFP CETN2 doublets, which themselves, are surrounded by PCM foci.
The precise mechanism of CETN2 doublet separation is not currently known. Cep135, important in centriole biogenesis, is present in the MTOCs of incompetent oocytes and is lost when the PCM begins to expand and fragment prior to resumption of meiosis (Fig. 6). In somatic cells, Cep135 disruption leads to centrosome splitting and abnormal mitotic spindle phenotypes through its interaction with cNAP1, a centrosomal linker protein implicated in centriole-centriole cohesion 17,66,67 . cNAP1 is also in mouse GV MTOCs, although disruption of cNAP1 by microinjection of specific double-stranded RNAs did not affect first or second meiotic spindle assembly 48 . Recently, the initiation of MTOC stretching on the GV surface before nuclear envelope breakdown 16 , critical for PCM fragmentation and first meiotic spindle assembly, was shown to be dependent on PLK1 phosphorylation of cNAP1, although the presence and functional role of cNAP1 in these MTOCs was unclear since they were not thought to have centrioles 18 . Our results suggest that maturing oocytes split the two CETN2 doublets early in follicular growth before attaining full maturation (Fig. 3), with Cep135 and cNAP1 continuing to be expressed until just before meiotic resumption. Thus, perhaps other proteins play a key role in mouse centriole-centriole cohesion upstream of Cep135 and cNAP1. Investigations on the precise molecules essential for centriole persistence and PCM functionality in the doublets and their fates during oocyte growth and maturation to metaphase-II arrest await future studies.
Centriole destruction may be prerequisite for terminal differentiation, and the retention of centrioles is associated with proliferation, regeneration, pluripotency, perhaps even totipotency. Skeletal muscle differentiation results in centriole loss 68,69 as during neuron differentiation 70 . Cardiomyocyte differentiation is particularly instructive, since centrioles are lost during terminal differentiation in murine hearts, which do not regenerate 71,72 . Remarkably, centrioles are retained in the regenerating hearts of fish and newts 73 . Here, centrioles are lost upon the onset of meiotic maturation, with immature oocytes retaining their double paired centrioles (Fig. 3, graph). Perhaps the last phase of oogenesis (i.e., oocyte maturation) is another example of terminal differentiation, including with the elimination of centrioles and centrosomes. Interestingly, while it appears necessary to undergo two mitotic cycles to generate centrioles, perhaps during both spermatogenesis and oogenesis, the last two meiotic cycles also are prerequisite for centriole destruction.

Methods and Materials
Mouse husbandry, handling, and Institutional oversight. All 45 were obtained from the Jackson Laboratory (Bar Harbor, ME) as juveniles, bred and analyzed as described previously 25 . All mice were housed in an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)-accredited mouse facility or in the CIRB (French agreement number C75-05-12). We investigated isolated germ cells from 14 fetal gonads, follicle and mature GVs from >36 females, and produced more than 500 oocytes for analysis. Fetal stage sex determination by PCR. Genomic DNA was isolated for Y-chromosome detection in GFP CETN2-expressing e11.5, e14.5 and e18.5 mouse tail tip tissues and PCR performed using MyTaq Extract-PCR Kit (Bioline, Taunton, MA). A 331-bp gene segment was amplified using the Jarid primers (forward: 5′ CTG AAG CTT TTG GCT TTG AG; reverse: 3′ CCG CTG CCA AAT TCT TTG G; Life Technologies, Carlsbad, CA). Β-actin loading control used the primers: 5′ GAT GAC GAT ATC GCT GCG CTG GTC G 3′ (forward), and 5′ GCC TGT GGT ACG ACC AGA GGC ATA CAG 3′ (reverse). PCR conditions were: 95 °C for 3 min, followed by 35 cycles of 95 °C for 15 seconds, 60 °C for 15 seconds, and 72 °C for 20 seconds. PCR product was analyzed in 2% agarose gel and visualized using ethidium bromide. Lanes expressing two bands were classified as males, the remainder female. Direct immunofluorescence of GFP CETN2 expression from non-gonadal cells at e11.5, e14.5, and e18.5 stages was accomplished as previously described 25 .
Cytoskeletal and PLK1 inhibitors. 10 mM stocks of nocodazole, paclitaxel and BI 2536 PLK1 inhibitors (Sigma Aldrich) were prepared in DMSO and stored at −80 °C. Inhibitors were diluted to final concentrations in KSOM. Rescue experiments from inhibitors were performed by rinsing treated oocytes 3x in M2 before returning to inhibitor-free KSOM for recovery at the times specified.
Immunocytochemistry. Isolated primordial germ cells, oogonia from fetal gonads, and zona-denuded Immunohistochemistry of intact ovaries for primary oocyte detection. P4 neonatal and adult excised ovaries were fixed in toto in 4% pFA (Electron Microscopy Services) in mHTF-Hepes overnight at 37 °C, rinsed 3x in sterile PBS, air dried at −20 °C for 2-4 hr, and then embedded in optimal cutting temperature compound (OTC; Tissue-Tek; VWR, Bridgeport, NJ) prior to storage at −80 °C. 7-µm sections were cut on a cryostat at −20 °C (CM1850 UV cryostat; Leica, Buffalo Grove, IL), floated onto clean glass slides (MidSci, Valley Park, MO), and stored at −80 °C. For immunostaining, slides were warmed at room temperature for 5 min, treated for 10 secs with Surgipath O-Fix (Thermo Fisher), and washed briefly in distilled H 2 O. Immunostaining was performed as described above.
Equipment, Imaging, analysis, and settings. Fixed slide imaging was accomplished with a Nikon A1 four-laser line confocal microscope 25 . Images were collected at 1024 × 1024 size at ¼ frames per second, a 79.2 μm pinhole, and a z-depth of 0.25 μm through the entire oocyte, with a differential interference contrast (DIC) Plan Fluor ×100 (1.3 NA) objective. The same laser photomultiplier tube settings for each channel across specimens helped facilitate comparison between oocytes. Fluorescent intensity ratios, mean intensities, area or volume measurements were performed on binarized images, using Elements software threshold tool and region-of-interest statistical menu, and downloaded to Microsoft Excel for statistical analysis. For image panels, a subtracted background image from outside of the oocyte was performed on generated confocal nd2 files. Deconvolution software (Landweber; supplied by Nikon USA), using the point scan confocal command and same filter (clear) at eight iterations was applied and selected deconvolved.tiff images prepared in Adobe Photoshop (Adobe Systems, San Jose, CA) for image panels.
Correlative light and electron microscopy. CLEM on GFP CETN2 mouse oocytes was adapted from a published protocol 75 . GFP CETN2 foci were imaged on an inverted microscope (Eclipse Ti; Nikon, Tokyo, Japan) with a spinning-disk confocal head (CSUX Spinning Disk; Yokogawa Electric Corporation, Tokyo, Japan) and 200-nm-thick Z-sections spanning the entire oocyte recorded to register GFP CETN2 doublets positions. Adhering oocytes were then perfused with 2.5% glutaraldehyde, marked with a diamond scribe, and rinsed extensively in PBS. Next, samples were stained with 2% osmium tetroxide and 1% uranyl acetate, dehydrated, and then embedded in EMbed 812resin. Oocytes identified by light microscopy were then serially sectioned (200-nm-thick), transferred onto copper slot grids, stained with uranyl acetate and lead citrate, and imaged by transmission electron microscope (H-7650; Hitachi, Tokyo, Japan).
Mouse strains and genotyping. Both CB6-Tg (CAG-EGFP/CETN2)3-4Jgg/J and ZP3 Cre+; mCherry-Plk4+ female mice were used. The [C57BL/6-Tg(Zp3-cre)93Knw/J] breeding pairs were obtained from Jackson Laboratories. mCherry-Plk4 flox/wt mice were generated by random insertion of a pCAG-loxCATloxmCherryPlk4SV40pA construct in the genome of C57BL/6 N mice 49 . Microinjection. Injection of in vitro-transcribed cRNAs into the cytoplasm of Prophase I-arrested oocytes was performed using an Eppendorf FemtoJet microinjector as described 76 , and oocytes kept for 1-3 hrs in milrinone to allow fusion protein expression. Oocytes were released from prophase I arrest by washing and transferring into milrinone-free M2.