The Male Mouse Meiotic Cilium Emanates from the Mother Centriole at Zygotene Prior to Centrosome Duplication

Cilia are hair-like projections of the plasma membrane with an inner microtubule skeleton known as axoneme. Motile cilia and flagella beat to displace extracellular fluids, playing important roles in the airways and reproductive system. On the contrary, primary cilia function as cell-type-dependent sensory organelles, detecting chemical, mechanical, or optical signals from the extracellular environment. Cilia dysfunction is associated with genetic diseases called ciliopathies and with some types of cancer. Cilia have been recently identified in zebrafish gametogenesis as an important regulator of bouquet conformation and recombination. However, there is little information about the structure and functions of cilia in mammalian meiosis. Here we describe the presence of cilia in male mouse meiotic cells. These solitary cilia formed transiently in 20% of zygotene spermatocytes and reached considerable lengths (up to 15–23 µm). CEP164 and CETN3 localization studies indicated that these cilia emanate from the mother centriole prior to centrosome duplication. In addition, the study of telomeric TFR2 suggested that cilia are not directly related to the bouquet conformation during early male mouse meiosis. Instead, based on TEX14 labeling of intercellular bridges in spermatocyte cysts, we suggest that mouse meiotic cilia may have sensory roles affecting cyst function during prophase I.


Introduction
Centrioles are conserved microtubule-based intracellular structures [1,2]. Centrosomes are made up of two orthogonally barrel-shaped centrioles embedded in an electron-dense material called the pericentriolar matrix [3]. In mammals, each of the centrioles is composed of nine microtubule (MT) triplets around a central proteinaceous structure whose shape resembles a cartwheel [4]. In contrast, centrioles are composed of either MT triplets or MT doublets in flies [5] or a MT singlet in nematodes [6]. The function of centrosomes in most animal cells is crucial, since they behave as the MT organizing centers (MTOCs), controlling the configuration of the MT cytoskeleton of interphase cells, as well as regulating the formation of the bipolar spindle during cellular division [7][8][9][10]. For this reason, centrosome regulation is tightly coordinated with cell cycle progression [2]. In G1, the centrosome is composed of two parental centrioles, but it undergoes a duplication during the S phase, with each of the parental centrioles giving rise to a newly formed procentriole. Thus, by G2, cells contain two centrosomes, each consisting of an older and a younger centriole (known as the mother and daughter centrioles, respectively). Later, during mitosis, the two centrosomes occupy opposite poles of the dividing cell, where they shape the bipolar spindle [11].
Ciliopathies are often associated with infertility [2,37]. Moreover, some studies have revealed that primary cilia dysfunction is associated with impaired male reproductive tract development. However, these effects were attributed to the malfunction of the primary cilium in somatic cells of the testicle, like Sertoli and peritubular myoid cells or cells of the epididymis [38]. So far, information about the presence of cilia in meiocytes is scarce. Two recent studies reported the presence of cilia in meiotic cells in zebrafish oogenesis [39] and spermatogenesis [40]. These reports indicated that the meiotic cilium might be an important regulator of chromosome polarization and recombination. In addition, these recent findings indicated that cilia might be present in male and female meiosis in mice [39], but a morphological or functional description is lacking.
This study presents the first thorough description of cilia in male mouse meiosis. We uncovered the precise stage at which the cilium is assembled, revealing that meiotic cilia are very transient structures. In addition, contrary to zebrafish, our results suggest that the role of the cilium in mammalian meiosis is not related to chromosome polarization.

Materials
Testes from adult C57BL/6 (wild-type, WT) and genetically modified Ankrd31 −/− [41] male mice were used for this study. All animals' procedures were approved by local and regional ethics committees (UAM Ethics Committee for Research and Animal Welfare) and performed according to European Union guidelines.

Squashing Procedure and Immunofluorescence Microscopy
Seminiferous tubules were fixed and processed following previously described protocols for the squashing technique [42].
Immunofluorescence images and stacks were collected on an Olympus BX61 microscope equipped with epifluorescence optics, a motorized Z-drive, and Olympus DP74 digital cameras controlled by Cellsens software (Olympus Life Science, Hachioji, Tokyo). Finally, images were processed with ImageJ (National Institute of Health, Bethesda, MD, USA; http: //rsb.info.nih.gov/ij accessed on 11 September 2022) or/and Adobe Photoshop software.

Histology
For histological cryosections, dissected testes were included in OCT (Sakura Finetek Europe, AV Alphen aan den Rijn, The Netherlands) and immediately frozen. The OCT blocks were cut into 10 µm thick sections using a cryostat. The sample slides were allowed to dry at room temperature and hydrated in PBS three times, followed by fixation with 4% PFA for 10 minutes. Slides were washed in PBS twice and permeabilized in PBS/0.1% Tri-ton X-100 for 20 minutes. Slides were washed in PBS three times and blocked with PBS/2% BSA for 30 minutes before being used for immunofluorescence protocol.

Quantification Analysis
The presence of cilia was quantified in 100 zygotene spermatocytes of three biological replicates. The length of cilia was quantified in a minimum of 20 spermatocytes per stage. Statistical analysis was performed using a One-way ANOVA followed by Holm-Sidak's post hoc test. Significant differences were considered when p < 0.05. The relation between cilia and bouquet configuration was quantified in a total of 50 spermatocytes. The number of cilia per cyst was quantified in a total of 25 cysts.

Mouse Meiotic Cilia Are Present in Zygotene spermatocytes prior to Centrosome Duplication
Ciliary structures in mouse seminiferous tubules have been recently suggested to be very short projections, barely extending beyond the basal body [39]. However, the actual size, cycle of extension, and retraction, or even the precise stages at which cilia develop, remain unclear. We here present a detailed study of the localization and dynamics of the mouse's meiotic cilium. For this purpose, we first studied the localization of the main component of its axoneme, acetylated tubulin (AcTub), during both meiotic divisions in squashed spermatocytes. We triple immunolabelled AcTub with a marker of centrioles, Centrin 3 (CETN3), and a marker of the progression of meiotic synapsis, the Synaptonemal Complex Protein 3 (SYCP3). CETN3 allowed us to determine whether centrioles have been duplicated, while SYCP3 allowed us to identify the meiotic stage, especially during prophase I.
At early prophase I, in leptotene, AcTub labeled the centrioles colocalizing with CETN3 ( Figure ??A and A ). In the transition from leptotene to early zygotene, two alternative patterns could be seen. Some spermatocytes were not ciliated (Figure ??B and B ), while others showed an elongated hair-like structure labeled with AcTub, emanating from one of the CETN3 centriole signals (Figure ??C and C ). Quantitative analysis showed that around 20% of the spermatocytes at zygotene showed this hair-like structure (n = 100 spermatocytes at zygotene per individual, three biological replicates) (Figure ??D). This hair-like staining pattern of AcTub in spermatocytes at early zygotene is consistent with the solitary projection formed when primary cilia assemble in both somatic [14] and meiotic cells [39]. These zygotene cilia were present when only two centrioles were detected, which indicated that mouse meiotic cilia are assembled prior to centrosome duplication.
Centrioles duplicate during zygotene [33], as revealed by the presence of four different signals labeled with CETN3. Cilia were no longer observed in spermatocytes at late zygotene after centrioles had duplicated (Figure ??E and E ). At pachytene, AcTub was detected as a single signal per centrosome associated with the duplicated centrioles ( Figure ??F and F ). At diplotene (Figure ??G and G ) and diakinesis (Figure ??H and H -H ), when centrosomes started migrating towards opposite poles, the signal of AcTub expanded to surround the two pairs of separated centrioles. During prometaphase I, AcTub still localized to the centrosomes (Figure ??I and I -I ). Once centrosomes had completely migrated to opposite poles and the meiotic bipolar spindle was established at metaphase I, AcTub labeled the two pairs of centrioles and the MTs of the bipolar spindle, while CETN3 only detected the centrioles (Figure ??J and J ). The same pattern persisted at anaphase I when homologous chromosomes segregated (Figure ??K and K ). At telophase I, CETN3 still labeled the centrioles, and AcTub labeled the centrioles and the midzone MTs (Figure ??L and L ). During interkinesis, the AcTub signal again accumulated at the centrosomes, as duplicated centrioles were clearly seen by the four CETN3 signals upon the second centrosome duplication (Figure 2A and A ). Cilia were not detected again prior to the second centrosome duplication ( Figure 2E). At metaphase II ( Figure 2B and B ) and anaphase II ( Figure 2C and C ), AcTub was observed at the centrioles and the meiotic bipolar spindle II. At telophase II, AcTub labeled the centrioles and the midzone MTs ( Figure 2D and D ).  round the two pairs of separated centrioles. During prometaphase I, AcTub still localized to the centrosomes ( Figure 1I, I′ and I″). Once centrosomes had completely migrated to op-posite poles and the meiotic bipolar spindle was established at metaphase I, AcTub labeled the two pairs of centrioles and the MTs of the bipolar spindle, while CETN3 only de-tected the centrioles ( Figure 1J,J′). The same pattern persisted at anaphase I when homologous chromosomes segregated ( Figure 1K,K′). At telophase I, CETN3 still labeled the centrioles, and AcTub labeled the centrioles and the midzone MTs ( Figure 1L,L′). During interkinesis, the AcTub signal again accumulated at the centrosomes, as duplicat-ed centrioles were clearly seen by the four CETN3 signals upon the second centrosome duplication (Figure 2A,A′). Cilia were not detected again prior to the second centro-some duplication ( Figure 2E). At metaphase II ( Figure 2B,B′) and anaphase II ( Figure 2C,C′), AcTub was observed at the centrioles and the meiotic bipolar spindle II. At tel-ophase II, AcTub labeled the centrioles and the midzone MTs ( Figure 2D,D′).  Tubulin is a protein whose distribution has been extensively studied in mammalian meiosis. Therefore, it was surprising that the meiotic cilium was not observed before. In order to compare the distribution of AcTub with that of overall tubulin, we performed a colabeling with an antibody against αTubulin (αTub). A careful comparison of nonacetylated αTub and AcTub leads us to conclude that they show a different localization pattern in meiosis (Figure 3). At leptotene, αTub was mainly localized to cytoplasmatic MTs, while AcTub accumulated only at the centrosome ( Figure 3A). In those zygotene spermatocytes in which the cilium was present, AcTub labeled this structure, while αTub was only located along cytoplasmatic MTs ( Figure 3B). This indicates that αTub does not label meiotic cilia and may explain why these structures were not detected in previous studies. At metaphase I and metaphase II, both non-acetylated and acetylated tubulin labeled the spindle MTs, but only AcTub was detected at the centrosomes ( Figure 3C,D).
order to compare the distribution of AcTub with that of overall tubulin, we performed a co-labeling with an antibody against αTubulin (αTub). A careful comparison of non-acetylated αTub and AcTub leads us to conclude that they show a different localization pattern in meiosis (Figure 3). At leptotene, αTub was mainly localized to cytoplasmatic MTs, while AcTub accumulated only at the centrosome ( Figure 3A). In those zygotene spermatocytes in which the cilium was present, AcTub labeled this structure, while αTub was only located along cytoplasmatic MTs ( Figure 3B). This indicates that αTub does not label meiotic cilia and may explain why these structures were not detected in previous studies. At metaphase I and metaphase II, both non-acetylated and acetylated tubulin la-beled the spindle MTs, but only AcTub was detected at the centrosomes ( Figure 3C,D).  To further characterize the organization and composition of the mouse meiotic cilium, we studied the localization of AcTub and ADP-ribosylation factor-like 13B (ARL13B) in histological cryosections of mouse testis. ARL proteins belong to the ARF family of GTPases, which are involved in ciliogenesis and ciliary intraflagellar transport (IFT) in somatic cells [43,44]. More specifically, ARL13B mediates primary cilia function and/or Hedgehog signaling [45]. In testis sections, spermatocytes located at the base of the semi-niferous tubule were the only ones that presented with cilia and an SYCP3 signal compatible with early prophase I ( Figure 4I.(Aa-d)). However, AcTub also labeled flagella of spermatids facing the lumen of the seminiferous tubule ( Figure 4I.(A)). ARL13B was present along the meiotic cilia visualized at prophase I spermatocytes located at the base of the tubule ( Figure 4I.(Ba-d)). Triple immunolabeling of ARL13B, AcTub, and SYCP3 in squashed spermatocytes corroborated that ARL13B and AcTub colocalized in zygotene spermatocytes ( Figure 4II.(A)). These results confirm that the fiber-like structures observed at zygotene spermatocytes are bona fide cilia. Nevertheless, given that mouse flagella also incorporate AcTub [13], an additional corroboration was conducted in order to disprove that the cilia observed in zygotene could be mistaken with the flagella of nearby spermatozoa after the squashing procedure. For this, we studied the localization of AcTub in spermatocytes of Ankrd31 −/− [41], a sterile mouse model that lacks spermatozoa. Our results showed that Ankrd31 −/− presents zygotene cells with AcTub-labelled cilia ( Figure S1A,B), with a morphology similar to that of the wild type (WT) individuals ( Figure S1C,D). In contrast, only WT mice present early spermatids ( Figure S1E) and elongated mature spermatids ( Figure S1F) with AcTub-labelled flagella.
In conclusion, these results lead us to conclude that cilia are present in mouse spermatocytes at zygotene prior to centrosome duplication.

Meiotic Cilia Appear at the Onset of Synapsis
We next wanted to assess the temporal correlation between the formation of this structure and other typical meiotic processes, such as chromosome synapsis. For this, we triple immunolocalized AcTub with SYCP3 and SYCP1, a component of the transverse filaments of the synaptonemal complex [46]. At leptotene, when synapsis had not yet started and SYCP1 labeling was absent, AcTub was only detected at the centrioles ( Figure 5I.(A)). The cilium was first observed in late leptotene spermatocytes when SYCP3-labeled axial elements were already formed, but SYCP1 protein was not yet loaded to chromosomes. The average length of the cilium at this stage was approximately 4,7 µm, presumably polymerizing ( Figure 5I.(B),II.). Synapsis initiation marked the beginning of zygotene. At this stage, short filaments of SYCP1 were seen partially colocalizing with SYCP3. The cilium elongated conspicuously at this stage, reaching up to 22 µm, with an average length of 15µm at early-mid zygotene ( Figure 5I.(C,D), and Figure 5II). When synapsis has partially progressed and longer filaments of SYCP1 were clearly seen (late zygotene), the meiotic cilium appeared to shrink, measuring an average of 9,3 µm long ( Figure 5I.(E),II.). In the quantitative analysis of the cilia length ( Figure 5II), One-way ANOVA analysis showed significant differences between the length of the cilia among the different stages (p < 0.0001). As previously indicated, the cilia were present in only a fraction (≈20%) of zygotene cells. Therefore, most spermatocytes at this stage did not present cilia. In those cases, AcTub only labeled centrioles ( Figure 5F).
These results led us to conclude that mouse meiotic cilia start to polymerize at the transition from leptotene to zygotene, and they are fully formed by mid zygotene.

Mouse Meiotic Cilia Emanate from the Mother Centriole Prior to Centrosome Duplication
We then tested if mouse meiotic cilia emanate from the mother centriole. For this purpose, we first studied the dynamics of CEP164, a protein of the distal appendages required for cilia formation in somatic cells [24,47] that has also been identified at the mother meiotic centriole in mouse spermatocytes [48]. Our results corroborated that CEP164 labeled one of the two centrioles before centrosome duplication at early prophase I. At leptotene, CEP164 labeled a circular signal around one of the two CETN3-labeled centrioles of the not yet duplicated centrosome ( Figure 6I.(A,A )). Centrosome duplication occurs during zygotene [33,48], and four signals of CETN3 were observed, but CEP164 still labeled only one of the four centrioles at this stage ( Figure 6I.(B,B )). We then triple immunolabelled CETN3, CEP164, and AcTub. We found that the cilia emanated from the centriole that possessed CEP164 at early zygotene, i.e., from the mother centriole of the not yet duplicated centrosome ( Figure 6II.(A)). The CETN3 signal did not overlap with AcTub cilium labeling, suggesting that when the cilium is fully formed, the centrioles are not acetylated ( Figure 6II.(A )). These results indicate that mouse meiotic cilia appear at late leptotene to early zygotene transition when centriole duplication has not yet occurred and that the cilium emanates from the mother centriole.   labelled CETN3, CEP164, and AcTub. We found that the cilia emanated from the centriole that possessed CEP164 at early zygotene, i.e., from the mother centriole of the not-yet duplicated centrosome ( Figure 6II.A). The CETN3 signal did not overlap with AcTub cilium labeling, suggesting that when the cilium is fully formed, the centrioles are not acetylated ( Figure 6II.A′). These results indicate that mouse meiotic cilia appear at late leptotene to early zygotene transition when centriole duplication has not yet occurred and that the cilium emanates from the mother centriole. To corroborate the pattern of distribution of CEP164 and its implication during male mouse meiosis, we continued studying its localization during both meiotic divisions. We observed that CEP164 was sequentially loaded to the centrioles until the four centrioles assembled distal appendages at the end of the first meiotic division ( Figure S2). Therefore, the four centrioles present CEP164 in spermatocytes at metaphase I and telophase I ( Figure S2E,F), according to previous results in mouse spermatocytes [48]. There was no incorporation of additional distal appendages in the daughter centrioles during the second meiotic division ( Figure S3). CEP164 labeled only one of the two centrioles per centrosome in opposite poles of the bipolar meiotic spindle II ( Figure S2B). When spermiogenesis progressed, centrioles were disorganized in late spermatids, and no CETN3 signals were seen; however, the CEP164 signal was still present as previously reported [48,49], indicating the formation of a basal body for the growing flagellum ( Figure S3A-E).
Overall, our results prove that the meiotic cilium appears prior to the first centrosome duplication at zygotene. The cilium is not reassembled prior to nor during the second meiotic centrosome duplication at interkinesis.

Meiotic Bouquet Conformation and Cilia Formation Are Not Concurrent Events
Meiotic cilium in zebrafish oocytes favors the polarization of the chromosome telomeres, which are anchored to the nuclear envelope (NE), forming the so-called bouquet configuration [39]. To test whether the mouse zygotene cilium in spermatocytes is related to the bouquet conformation, we performed a double immunodetection of the telomeric double-stranded TTAGGG repeat binding proteins (TRF2) and AcTub. In spermatocytes at zygotene, we observed TRF2 as small and rounded signals randomly distributed, presumably over the NE, regardless of whether cilia labeled by AcTub were absent ( Figure 6III.(A)) or present ( Figure 6III.(B)). After triple immunolocalization of AcTub, SYCP3, and TRF2, only a semi-bouquet could be seen in a small fraction of spermatocytes at zygotene, where almost all telomeric ends were polarized to one region of the nucleus. This agrees with previous reports on mouse meiosis [50][51][52]. However, no cilia were observed in this situation ( Figure 6III.(C)). Accordingly, no bouquet nor semi-bouquet configuration was presented in any ciliated spermatocyte at zygotene ( Figure 6III.(D)). A graphical representation of the number of ciliated spermatocytes at zygotene (n = 50) that present the representative mouse semi-bouquet configuration is included ( Figure 6IV). Therefore, there seems not to be a direct relationship between the presence of a cilium and the configuration of the telomeric bouquet in mouse spermatocytes.

Zygotene TEX14 Connected Cysts
Classical studies on the organization of the seminiferous tubules showed that spermatocytes form cysts of cells synchronously advancing through meiosis. These spermatocytes present cytoplasmic bridges that presumably provide biochemical and physiological coordination of the cells in a cyst [53,54]. As our results showed that only ≈20% of the spermatocytes at zygotene had a formed cilium, we then wanted to test if spermatocytes displaying the cilium could form specific cysts. For this, we detected TEX14, the product of testis-expressed gene 14 (Tex14), a protein that localizes to germ cell intercellular bridges in male spermatocytes [55]. Our results showed that TEX14 labeled ring-shaped intercellular bridges between spermatocytes at different stages. Cysts of early prophase I spermatocytes were seen at the base of the seminiferous tubules in testis cryosections ( Figure 7I.(Aa)). In most of these spermatocyte cysts, the cilium could be detected in only one of the spermatocytes ( Figure 7I.(Ab)).
displaying the cilium could form specific cysts. For this, we detected TEX14, the product of testis-expressed gene 14 (Tex14), a protein that localizes to germ cell intercellular bridges in male spermatocytes [55]. Our results showed that TEX14 labeled ring-shaped intercel-lular bridges between spermatocytes at different stages. Cysts of early prophase I spermat-ocytes were seen at the base of the seminiferous tubules in testis cryosections ( Figure 7I.Aa). In most of these spermatocyte cysts, the cilium could be detected in only one of the spermatocytes (Figure 7I.Ab).  The association of spermatocytes in cysts was also preserved in preparations of squashed seminiferous tubules. We double immunolabeled TEX14 and VASA, a germspecific cytoplasmic marker [56], and corroborated a cytoplasmic continuity between the spermatocytes at the places where TEX14 bridges were present ( Figure 7II.(A)). Our results suggested that cysts of connected spermatocytes presented mono-ciliation ( Figure 7II.(B)). Furthermore, quantitative analysis showed that only one ciliated spermatocyte was observed per cyst (n = 25 cysts of spermatocytes at early zygotene) ( Figure 7II.(C)). This demonstrates that cells bearing a cilium are not specifically connected in a cyst, but on the contrary, they preferentially form connections with other spermatocytes lacking the cilium.

The Meiotic Cilium in Mouse Spermatogenesis
The primary cilium is an outgrowth of the plasma membrane located in many eukaryotic cell types. These immobile and solitary projections have diverse functions related to cell communication [57] and have traditionally been depicted as structures whose continued presence is incompatible with cell cycle progression [28]. However, although studies on the structure of primary cilia in various somatic tissues are extensive and diverse, their presence in vertebrate meiosis has only been reported in two recent works. The first one described the zygotene cilium in zebrafish oocytes, pointing to an evolutionary conservation of this structure in mammals, but suggested that ciliary structures in mouse spermatocytes were very short, barely extending beyond the basal body [39]. Our results demonstrate that the ciliary structures in mouse spermatocytes are not very short, on the contrary, cilia are strikingly large (up to 23 µm). This finding has been addressed by providing a detailed morphological analysis of adult mice in a reproductive stage, data that was lacking in that previous study [39]. The second study proposed that germ-cell-specific depletion of ciliary genes resulted in compromised double-strand break repair, reduced crossover formation, and increased germ-cell apoptosis in zebrafish spermatogenesis [40]. We here showed a detailed description of the mouse's meiotic cilium, and we hypothesized that they are primary cilia. We found that around 20% of the spermatocytes at zygotene presented cilia as a long, solitary protrusion that disassembled once spermatocytes progressed into pachytene. This provides a scenario that challenges the assumption that primary cilia are antagonistic with cell division, as it has to be taken into account that cells at zygotene have already started meiotic cell division [58].
Apart from our work and the recent advances in zebrafish gametogenesis [39,40], another previous report also pointed to the presence of primary cilia during D. melanogaster spermatogenesis. Moreover, it suggested that these cilia were stable through both meiotic divisions in dividing spermatocytes with yet unknown functions [59]. Therefore, it appears that, although somatic primary cilia in vertebrates are sensory machinery during interphase, which was considered a "sleeping beauty" to date during cell division [60,61], cells that undergo meiosis are capable of polymerizing cilia once cell division has started. This may be related to the dynamics of the mitotic versus the meiotic centrosome. In somatic cells, centrosome duplication occurs, like DNA duplication, during the S phase, i.e., before mitosis [62]. In contrast, during gametogenesis, DNA duplication occurs in the premeiotic interphase, and in mice, centrosome duplication happens twice once meiosis begins, at zygotene, and again at interkinesis [33,48]. This implies that somatic centrosome duplica-tion occurs after ciliogenesis and before entry into mitosis. In contrast, meiotic centrosome duplication occurs after ciliogenesis, with both processes taking place during meiosis.

Meiotic Centrosomes and Their Relation to Ciliogenesis and Flagelogenesis
Our results demonstrate that zygotene cilia in mouse spermatocytes contain acetylated tubulin -AcTub-, similar to the zygotene cilia in zebrafish spermatocytes and oocytes [39,40]. This was expected since this posttranslational modification of tubulin is a well-known marker in both motile and primary cilia in somatic cells [63][64][65]. As expected, we detected that tubulin was also acetylated in the flagellar axoneme of spermatids [13]. Given that the cilium emanates from the larger rounded signal of the two centrioles marked by CETN3, where CEP164 distal appendages are assembled, we concluded that zygotene cilia emanate from the mother centriole, as in somatic cells [66,67], pointing to conservation of the process of ciliogenesis between somatic and germ cells.
However, it is important to note that ciliogenesis from the mother centriole only occurs prior to the first centrosome duplication at zygotene, whereas another meiotic ciliogenesis from the new distal appendages does not occur prior to the second centrosome duplication at interkinesis. Our results reveal that CEP164 distribution was consistent with the function of distal appendages in cilia and flagella [68,69]. On the one hand, we show that CEP164 served to dock the basal body during zygotene ciliogenesis, sharing a similar role to its function in somatic cells [24,26,27]. On the other hand, spermatids presented CEP164 distal appendages to promote the polymerization of the flagellar axoneme, as previously reported [48]. Hence, the dynamics of CEP164 appear to be essential for the progression of spermatogenesis. This is consistent with the sterile phenotype of CEP164 deficient mice -FoxJ1-Cre;CEP164fl/fl-that showed sperm agglutination and aberrant ciliogenesis of motile multicilia in the somatic cells of the efferent ducts [49]. With the present work, we now suggest that CEP164 participates not only in flagelogenesis but also in mouse meiotic ciliogenesis. For further clarification, we present a scheme representative of the progression of male mouse meiosis in relation to centrosome dynamics, ciliogenesis, and flagelogenesis ( Figure 8).

Tubulin Acetylation during Male Mouse Meiosis
We here showed that centrosomes presented AcTub partially colocalizing with CETN3 during the entire meiotic division, but nonacetylated α-Tub was never detected at centrosomes. CETN3 is a component of meiotic centrioles [33,48]; therefore, we can conclude that meiotic centrioles possess acetylated tubulin. This might promote their stability since experiments using biotinylated tubulin demonstrated that centriolar MTs do not exhibit significant exchange with the cytoplasmic protein pool [70]. This agrees with centriolar MTs being heavily acetylated and polyglutamylated, which are hallmarks of stable MTs [71]. In this sense, acetylation of MTs promotes the recruitment of PLK1 to the centrosomes in mitosis [72]. This kinase is located at meiotic centrosomes regulating centrosome maturation, separation, and migration [33,48]. Interestingly, our results also demonstrate that when meiotic cilia are formed, centrioles do not show AcTub. Future research should clarify the relationship between PLK1 and the acetylation of centriolar tubulin and the role this kinase might exert on meiotic ciliogenesis.
On the other hand, during mitotic metaphase, AcTub is enriched at interpolar and kinetochore MTs, but not at astral MTs, and AcTub becomes concentrated on the midbody during telophase and cytokinesis [73]. This acetylation is an evolutionarily conserved post-translational modification of α-Tub associated with the increased stability of various MT populations, including the mitotic spindle [65]. Moreover, tubulin acetylation does modulate the ability of mitotic MTs to bind to MAPs (Microtubule Associated Proteins) and motor proteins and may regulate MT stability and function [74,75]. Our results show that once the bipolar spindles are shaped at metaphase I and II, spindle MTs are acetylated. We also observed the labeling of AcTub in the midbody of telophase I and II. It has been described that the post-translational modifications of tubulin are related to the integrity of the spindle in oocytes [76] and the stability of the spermatozoa [13]. Specifically, an altered acetylation level of tubulin leads to defective spindle assembly and chromosome misalignment in mouse oocyte meiosis [77,78]. However, the function of AcTub in male mouse meiosis is not yet described. We suggest that the acetylation of tubulin in meiotic MTs might play similar roles, contributing towards the stability of the centrioles throughout the entire meiosis and towards the stability of the meiotic spindles I and II at metaphase I/II and the midbodies during telophase I/II.

Tubulin Acetylation during Male Mouse Meiosis
We here showed that centrosomes presented AcTub partially colocalizing with CETN3 during the entire meiotic division, but non-acetylated α-Tub was never detected at centrosomes. CETN3 is a component of meiotic centrioles [33,48]; therefore, we can conclude that meiotic centrioles possess acetylated tubulin. This might promote their stability since experiments using biotinylated tubulin demonstrated that centriolar MTs do not exhibit significant exchange with the cytoplasmic protein pool [70]. This agrees with centriolar MTs being heavily acetylated and polyglutamylated, which are hallmarks of stable MTs [71]. In this sense, acetylation of MTs promotes the recruitment of PLK1 to the centrosomes in mitosis [72]. This kinase is located at meiotic centrosomes regulating centrosome maturation, separation, and migration [33,48]. Interestingly, our results also demonstrate that when meiotic cilia are formed, centrioles do not show AcTub. Future research should clarify the relationship between PLK1 and the acetylation of centriolar tubulin and the role this kinase might exert on meiotic ciliogenesis.
On the other hand, during mitotic metaphase, AcTub is enriched at interpolar and kinetochore MTs, but not at astral MTs, and AcTub becomes concentrated on the midbody during telophase and cytokinesis [73]. This acetylation is an evolutionarily conserved post-translational modification of α-Tub associated with the increased stability of various

Zygotene Cilia and Bouquet Formation Are Not Directly Related in Male Mouse Meiosis
From our results, it emerged that the cilium appeared in primary spermatocytes at the transition from leptotene to zygotene prior to centriole duplication. This indicates that the cell signaling processes in which the cilium may be involved might be related to the initiation of synapsis at zygotene [46,79], especially because cilia are no longer detected when synapsis is complete at pachytene. During zygotene, the bouquet polarization is a transient conformation of the chromosomal ends that favors the pairing process between homologous chromosomes [52]. However, this structure significantly differs between species. In zebrafish, the bouquet configuration is obvious and easily detected [80]; In C. elegans, the bouquet does not exist per se as the chromatin concentrates at one side of the nucleus adopting the characteristic half-moon shape [81]. In marsupial mammals, the bouquet stage is conspicuous [82], but in contrast, the bouquet chromosome polarization in mice is incomplete. Indeed, it behaves as a semi-bouquet, as it does not include all the chromosome ends, and it is extremely short-lived as it is only present in a time window limited to ∼0.5% of spermatocytes [50,51]. Recent reports in zebrafish oogenesis demonstrated a correlation between the zygotene cilia and the bouquet conformation, suggesting that the absence of cilia or their incorrect formation affects synapsis progression [39]. However, we could not find any direct correlation between the presence of cilia and bouquet conformation in mice.
Our results indicate that the extremely short-lived bouquet in mice most likely precedes the formation of the cilia. Therefore, we conclude that there is no direct relation nor concurrency between the representative mouse semi-bouquet and the presence of meiotic cilia. Additionally, the zebrafish cilium has also been related to recombination regulation [40]. We cannot make any considerations in this regard, as the relationship between cilia and recombination in mice should require additional approaches.

Signaling Functions of the Cilia Could Be Spread to Cyst-Connected Spermatocytes
Primary cilia are found in many mouse cell types, and their morphology and size vary depending on their cellular functions. For example, primary cilia of cardiac fibroblasts are essential for heart development [83], and primary cilia of epithelial cells from the proximal renal tubule participate in the regulation of the glomerular filtration barrier [84].
In spermatocytes, we have observed the presence of a remarkably long cilium, approximately 15 µm, compared to the 6 µm described in zebrafish spermatocytes [39]. All aspects of cilium biology are regulated: their presence or absence on the cell surface, their morphology, their length, and ultimately their function [85]. Considering the structure we describe is a primary cilium, its length is similar to those of mouse renal epithelial cells, neurons, and olfactory cells [86,87]. In these somatic cells, primary cilia have a sensory function, capturing information from the extracellular environment and triggering an intracellular response. A greater surface area of the ciliary membrane could harbor more sensory transmembrane proteins and, therefore, more signaling receptors. That is why longer ciliary length has been related to greater sensitivity and response [88], and modifications of the ciliary length affect cilium integrity and, ultimately, its functions [86]. Given that the mouse meiotic cilium has a mean length equivalent to other long cilia in mouse somatic cells, we suggest that it may also perform important cell signaling functions during spermatogenesis. In parallel, we have also shown that mouse zygotene cilia present ARL13B, a GTPase of the ARL family known to mediate ciliary protein trafficking and regulate ciliary Hedgehog signaling [45]. Therefore, it is plausible that mouse zygotene cilia have an active Hedgehog pathway, which should be clarified in detail in future studies.
The histology of the testis might explain the absence of cilia in all spermatocytes. The organization of the epithelia of the seminiferous tubules is evolutionarily conserved in mammals and present cysts of cells undergoing spermatogenesis [89]. In mice, spermatogonia are linked by cytoplasmic bridges whose absence leads to infertility [54,55]. These bridges contain, among other proteins, the testis-expressed gene 14 product (TEX14) [55,90]. These bridges are partially released once spermatogonia evolve to primary spermatocytes upon entry to meiosis [91]. The proposed roles for the intercellular bridges include germ cell communication and synchronization [53] and sharing of gene products. An example of this function is the sharing of proteins encoded by Akap genes of the X chromosome between the four spermatids resulting from meiosis so that they can all form the fibrous sheath of the sperm [92]. The fact that only 20% of the spermatocytes at zygotene present cilia leads us to suggest that in a cyst of spermatocytes at zygotene with cytoplasmic continuity, not all the spermatocytes need to be ciliated to progress through spermatogenesis. Therefore, this leads us to hypothesize that spermatocytes within the same cyst could share the signaling events provided by a single ciliated spermatocyte. For further clarification, a scheme representing the seminiferous epithelia is presented (Figure 9). chromosome between the four spermatids resulting from meiosis so that they can all form the fibrous sheath of the sperm [92]. The fact that only 20% of the spermatocytes at zygotene present cilia leads us to suggest that in a cyst of spermatocytes at zygotene with cytoplasmic continuity, not all the spermatocytes need to be ciliated to progress through spermatogenesis. Therefore, this leads us to hypothesize that spermatocytes within the same cyst could share the signaling events provided by a single ciliated spermatocyte. For further clarification, a scheme representing the seminiferous epithelia is presented (Figure 9).  In conclusion, we argue that the need to maintain intercellular connections in spermatocyte cysts for achieving fertility [53], together with the unique anchoring junctions in the testis -Sertoli-Sertoli, Sertoli-spermatogonia, Sertoli-spermatocyte and Sertoli-spermatid cell unions [93,94]-might require complex signaling that helps the progres-sion of cells from base to lumen of the seminiferous tubule during spermatogenesis and subsequent spermiogenesis (Figure 9). One might think that the long zygotene cilia could be implicated in the signaling pathway between spermatocytes in coordination with Sertoli cells, at least during the first stages of prophase I. Given that ciliopathies present various anomalies in the assembly or regulation of the ciliary and flagellar axoneme [95], more studies will be necessary to analyze the relationship between meiotic cilia and sterility.
The present work has described the zygotene cilium in mouse spermatocytes suggesting cyst monociliation in the seminiferous epithelium, opening the line of research for future studies to unravel the importance of meiotic ciliogenesis.
Institutional Review Board Statement: All animal procedures of this study were conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board: UAM Ethics Committee for Research and Animal Welfare (PROEX n 301/19).

Informed Consent Statement: Not applicable.
Data Availability Statement: The raw data for immunofluorescence images and statistical data supporting the conclusion of this article will be made available by the authors upon request.