The WD40-protein CFAP52/WDR16 is a centrosome/basal body protein and localizes to the manchette and the flagellum in male germ cells

Development of spermatozoa requires remodelling and formation of particular structures. In elongating spermatids, the transient microtubular manchette contributes to the formation of the head–tail coupling apparatus (HTCA) and the sperm tail. The HTCA derives from the centrosome in that the proximal centriole inserts into the nuclear indentation and the distal centriole gives rise to the sperm flagellum. Although impairments in the formation of HTCA and sperm tail cause male infertility their molecular constituents are only partially known. The WD40-protein CFAP52 is implicated in motile cilia, but its relevance for male germ cell differentiation is not known. Here we show that CFAP52 is widespread expressed and localizes to a subset of microtubular structures. In male germ cells, CFAP52 is a component of the transient manchette and the sperm tail. However, expression of Cfap52 is not restricted to motile cilia-bearing cells. In NIH3T3 cells, CFAP52 localizes to the centrosome, the basal body, and the mitotic spindle poles, but not to the primary cilium. Our results demonstrate that CFAP52 is not restricted to motile cilia but instead most likely functions in constituting the centrosome/basal body matrix and the sperm tail.


Results
Widespread expression of Cfap52. In order to verify the reported expression of Cfap52, we performed RT-PCR using total RNA isolated from mouse tissues, and in addition from the mouse fibroblast cell line NIH3T3. Amplification of exons 1-14 of Cfap52 revealed the expected fragment of 1871 bp in testis and ovary but not in any of the other probes (Fig. 1a), whereas amplification of Gapdh in all probes verified the quality of cDNAs (Fig. 1c). These results confirmed the previous observation of a testicular expression of Cfap52 but additionally demonstrated that Cfap52 is also highly expressed in ovary, which to the best of our knowledge do not harbor cells with motile cilia. However, to detect even low expression levels we performed a nested PCR on the first PCR product amplifying a region comprising parts of exons 4 to 6. A DNA fragment of the expected size of 255 bp was found in all samples (Fig. 1b). Sequencing of the PCR products obtained from brain, ovary and epididymis confirmed amplification of Cfap52. The different intensities obtained for the secondary PCR product are caused by differing amounts of first PCR used as template. These results demonstrate that Cfap52 is expressed in all tissues investigated including the fibroblast cell line. Cfap52, therefore, seems to be ubiquitously expressed, including somatic tissues and cells that to not generate motile cilia, although at a low level. Expression of Cfap52 is therefore not restricted to motile cilia bearing cells.

CFPA52 localizes to the spermatid manchette and to the sperm tail. Testicular expression of
Cfap52, previously reported in the rat, was confirmed by our investigation in the mouse. However, besides an overall and unspecific cytoplasmic distribution of CFAP52 in spermatocytes no further specification is currently available 7 . We therefore intended to investigate the subcellular localization of CFAP52 during mouse spermatogenesis using mouse testicular cell suspensions decorated by a commercially available antibody. We first validated the anti-CFAP52 antibody for immunocytology by generating a CFAP52::EGFP fusion construct, transfected the plasmid into NIH3T3 cells, and decorated the expressed fusion protein with the anti-CFAP52 antibody ( Fig. 2a-d). Antibody staining was exclusively found in transfected cells, identified by their green autofluorescence of the CFAP52::EGFP fusion. Furthermore, the antibody co-localized with the green fluorescent CFAP52::EGFP fusion protein.
Scientific RepoRtS | (2020) 10:14240 | https://doi.org/10.1038/s41598-020-71120-9 www.nature.com/scientificreports/ Having the anti-CFAP52 antibody validated, we performed immunocytology on mouse testicular cell suspensions (Fig. 3). The acrosome was decorated by peanut lectin-FITC staining to enable identification of spermatid differentiation steps (Fig. 3). A weak expression of CFAP52 with a few strong dots and a shallow indication of fibrous structures at the nuclear surroundings was first found in early round spermatids ( Fig. 3a-d). These weak structures are most likely the first indications of the forming manchette, as fibre like structures with a strong CFAP52 reactivity are present in round spermatids of a more advanced step (Fig. 3e-h). Thereafter, a clear manchette with its characteristic parallel bundles was visible in round spermatids ( Fig. 3i-l) and elongating spermatids ( Fig. 3m-x) that is strongly stained for CFAP52 indicating an association of CFAP52 with microtubules. In mature spermatozoa, CFAP52 located to the sperm tail with a somehow stronger intensity in the principal piece than in the mid-piece (Fig. 3y-y″′). No antibody staining was detected in elongating spermatids when anti-CFAP52 antibody was omitted ( Fig. 3z-z″′).
Colocalization of CFAP52 with the manchette microtubules was demonstrated by alpha-tubulin decoration in round and elongating spermatids (Fig. 4a-h). Furthermore, CFAP52 showed a similar distribution as the LINC components SUN3 and SUN4 ( Fig. 4i-p). Enrichment of CFAP52 at the head-to-tail coupling apparatus (HTCA), which was decorated by γ-tubulin staining, was not found (Fig. 4q-t). Additionally, CFAP52 is not enriched at the perinuclear ring, which was decorated for the microtubule plus-end binding protein EB3 (Fig. 4u-x), but showed a faint ring-like concentration anteriorly to the perinuclear ring (Fig. 4u, x, arrows) 23 .
A similar distribution in the manchette and the sperm tail was observed for the outer dense fibre protein 1, ODF1 31 , posing the question whether the observed colocalization indicates interaction or interdependency. We, therefore, investigated CFAP52 localization in testicular sections using both wild-type as well as Odf1-deficient mice 28 . In testis-sections of Odf1 +/+ -mice we observed CFAP52-decoration in the manchette (Fig. 5a-d, i-l, arrows). Odf1-deficient testis-sections exhibited the same strong CFAP52-staining in the manchette (Fig. 5e-  www.nature.com/scientificreports/ Figure 3. Subcellular localization of CFAP52 in male germ cells. CFAP52 is located to manchette and sperm tail in mouse male germ cells. Suspension preparations of adult mouse testis were incubated with a commercial antibody against CFAP52 (red) and the acrosome decorated with PL-FITC. The developing manchette is decorated with CFAP52 in round and elongating spermatids (a-x). In spermatozoa, CFAP52 locates to the tail with a more intense staining of the principal piece than the mid-piece (z-z″′). No staining of elongating spermatozoa was evident when the anti-CFAP52 antibody was omitted (control, y-y″′).
Scientific RepoRtS | (2020) 10:14240 | https://doi.org/10.1038/s41598-020-71120-9 www.nature.com/scientificreports/  Endogenous CFAP52 localizes to centrosome/basal body, intercellular bridge and spindle poles in somatic cells. We have proven prevalent low-level expression of Cfap52 by nested RT-PCR. In order to substantiate these findings, we performed immunoblotting using total proteins obtained from mouse tissues as well as from NIH3T3 cells. CFAP52, in the expected molecular mass of ~ 68 kDa, was clearly detectable in all probes demonstrating once more that CFAP52 is not restricted to specific tissues (Fig. 6). Our results are corroborated by Western blot validation of anti-CFAP52 (www.cusab io.com). Next, we used NIH3T3 cells www.nature.com/scientificreports/ to isolate the coding sequence of Cfap52 and generated a fusion protein with EGFP by cloning Cfap52 in frame to the N-terminal end of Egfp in plasmid pEgfp-N1. The plasmid pCfap52::egfp was transfected into NIH3T3 cells, and its distribution analyzed. Besides a more or less uniform background fluorescence we found a specific concentration of the CFAP52::EGFP fusion protein in two closely associated spots in the vicinity of the nucleus. Immuno-decoration with anti-γ-tubulin antibodies as a centrosomal marker revealed that CFAP52::EGFP colocalized with the centrosome in NIH3T3 cells ( Fig. 7A a-d). We then asked whether the centrosomal location is caused by overexpression of the fusion protein, or otherwise is a feature of the endogenous protein characterizing CFAP52 as a novel centrosomal protein. To this end, NIH3T3 cells were treated for immuno-decoration and double stained with anti-CFAP52, and either anti-γ-tubulin or anti-acetylated tubulin antibodies (Fig. 7B). We detected the endogenous CFAP52 protein in the centrosome by colocalization with the centrosomal marker protein γ-tubulin (Fig. 7B a-d). Furthermore, CFAP52 is found in the cytoplasmic bridge linking the two daughter cells (Fig. 7B e-h). In this case, the cytoplasmic bridge is highlighted by acetylated tubulin. CFAP52 did not colocalize with acetylated tubulin but instead seemed to be concentrated at the tip of the cytoplasmic bridge near the plasma membrane. NIH3T3 cells generate non-motile, solitary cilia known as primary cilia that are also easily identifiable by immuno-decoration for acetylated tubulin. The primary cilium as well as its associated daughter centriole at its base were identified by acetylated tubulin (Fig. 7B i-l). CFAP52 was not found in the ciliary axoneme but is highly concentrated at the basal body and at the daughter centriole. Additionally, CFAP52 is concentrated at the spindle poles that are highlighted by acetylated tubulin staining when captured at low intensity ( Fig. 7B m-p). Omitting the anti-CFAP52 antibody and using both secondary antibodies for incubation in the controls demonstrated decoration of the centrosome with anti-acetylated tubulin antibodies (in red) but did not show any green fluorescence (Fig. 7B q-t).
Centrosomal recruitment of CFAP52 isoforms. We have identified CFAP52 as a novel centrosomal protein in NIH3T3 cells. CFAP52 is encoded by the Wdr16/Cfap52 gene on chromosome 11 (11B3) in mice. The gene consists of 14 exons, and encodes three isoforms by alternative splicing. The longest isoform (NP_082239.2) consists of 620 amino acids (aa) encoded by exons 1-14. The isoform denominated X1 (XP_006534328.1) consists of 577 aa by skipping of exon 4. Isoform 203 (Q5F201) has a postulated length of 342 aa and is encoded by exons 1-8. We cloned the isoforms from NIH3T3 cells and generated in fusion plasmids to Egfp. Additionally, we generated another fusion protein named isoform X1-Nterm that consists of 309 aa of CFAP52 by skipping exons 4 and all C-terminal exons 9-14 (Fig. 8). According to ProSite (https ://prosi te.expas y.org) the full-length sequences CFAP52 and CFAP52-X1 contain two WD-repeats regions comprising aa 60-195, and 328-620 (related to the full-length isoform), and five WD-repeats-2. Referring to the full-length isoform, the WD-repeats-2 sequences comprise aa 107-142, 413-446, 457-490, 541-582, and 583-620. The isoform 203 contains one WD-repeat region and one WD-repeat-2, comprising aa 107-142. The WD-region of the isoform X1-Nterm differs from that of isoform 203 by a deletion of 43 aa (Fig. 8). To investigate the impact of the WDdomains for centrosomal recruitment of CFAP52 the egfp-fusion plasmids were transfected into NIH3T3 cells and the fusion proteins detected by their green auto-fluorescence (Fig. 9). The centrosome was decorated by immunostaining for their marker proteins either γ-tubulin or pericentrin. As observed previously for the fulllength CFAP52::EGFP fusion protein (Fig. 7A) both, the EGFP-fusion proteins of isoform 203 as well as of the N-terminal end of isoform X1 (X1-Nterm), are concentrated in the centrosome (Fig. 9).Taken together, our data indicate that (1) the C-terminal region comprising the second WD-repeats domain is not essential for centrosomal recruitment and that (2) parts of the first WD-repeats domain encoded by exon 4 are also not essential for centrosomal recruitment. Thus, a centrosomal targeting sequence, if at all present, must reside in the N-terminal end of CFAP52, represented by isoform X1-Nterm, or most likely the WD-repeats domains taken as a whole are responsible to direct CFAP52 to the centrosome. www.nature.com/scientificreports/ www.nature.com/scientificreports/

Discussion
WDR16/CFAP52 is a member of the WD40-protein family characterized by the WD-repeat domain. The WD40repeats, also known as GH-WD repeats, are repetitive motifs consisting of a conserved core bracketed by GH (Gly-His) and WD, and additional sequences of variable lengths 1,38 . The WD40-domains were first described in β subunits of transducin and other G proteins isolated from bovine tissues, and in the SCF (SKP1-CUL1-F-box protein) ubiquitin ligase complex component CDC4. WDrepeat proteins fulfill diverse biological functions including RNA synthesis/processing, signal transduction, cytoskeleton assembly, mitotic spindle formation, vesicular trafficking and cell growth 6,39 . WD-repeat domains are hence widespread and highly conserved in eukaryotes. Particularly, the WD40-repeat protein CFAP52/ WDR16/WDRPUH was initially identified and found to be upregulated in human hepatocellular carcinomas.  -fluorescence (green, a, e, i, m). Decoration of the centrosomes by immunostaining for the centrosomal marker proteins either γ-tubulin (anti-gammatubulin, b, j) or Pericentrin (anti-Pericentrin, f, n) (both in red). Nuclear counterstain with DAPI (blue). Arrows pointing to the centrosomal location of the fusion proteins. Bars are of 2 µm (a-p).
Scientific RepoRtS | (2020) 10:14240 | https://doi.org/10.1038/s41598-020-71120-9 www.nature.com/scientificreports/ Its overexpression in NIH3T3 cells accelerated cell growth, whereas inhibition of WDRPUH/CFAP52 reduced the growth of human liver carcinoma cells and induced apoptosis 3,7 . Previous investigation of Cfap52 indicated that its expression correlated with the presence of motile cilia 7 . Furthermore, WDR16/CFAP52 was annotated in the Ciliome Database and was identified in mouse ciliated tissues 40 . Moreover, the Wdr16 gene was identified as FOXJ effector gene altogether indicating a correlation of WDR16/CFAP52 with ciliation especially with the generation of motile cilia 41 . Proteins containing specific domains, especially coiled-coil domains, WD40-repeat domains, or tetratricopeptide repeat domains have all been linked to cilia 40,42,43 . In this regard, the intraflagellar transport component IFTA-1 is a WD repeat-containing protein 44 . However, WDR16/CFAP52 is not annotated in Syscilia Gold Standard (https ://www.sysci lia.org/golds tanda rd.shtml ) and therefore seems not to be an essential ciliary component.
The reported association of the cilia and flagella associated protein 52 (CFAP52/WDR16) with ciliation and its high expression in testis offered the unique opportunity to investigate its sub-cellular distribution in more detail. The formation of the flagellum starts in haploid spermatids and is well known step-by-step from the migration of the paired centrioles towards the caudal end of the nucleus, their insertion into the implantation fossa, and their progressive transformation into the basal body, the connecting piece, and eventually the outgrowth of the flagellum. The proximal centriole inserts into the implantation fossa and mediates the proper linkage between head and tail by the head-to-tail coupling apparatus (HTCA). The distal centriole, on the other hand, gives rise to the basal body that in turn will seed flagellum formation 10,45 . Furthermore, flagellum and connecting piece are both discernable at light-microscopic resolution and are easily identifiable using immuno-decoration.
Formation of HTCA and sperm tail requires many proteins that are presumably transported by a transient microtubular structure, the manchette 24,27 . The HTCA is essential for functional sperm, and its malformation causes acephalic spermatozoa and male infertility due to fragile attachment of sperm head and tail or even detachment 12,46 . Albeit essential for fertilisation, the knowledge of the molecular composition of the HTCA is just in the beginning. We therefore asked, whether CFAP52 as an annotated ciliary component is associated with the flagellum or its development. We found that, opposed to previous reports, CFAP52 locates to the transient manchette in round and elongating spermatids. CFAP52 might thus be involved in the stabilization of the structure or in supporting delivery of cargos, or is a cargo protein itself. Later on, CFAP52 was found in the sperm tail with the strongest staining in the principal piece. A weaker decoration was observed in the middle piece that might be explained by a reduced antibody accessibility due to the fact that mitochondria and outer dense fibres encapsulate the central region of the axoneme. Although, location of CFAP52 to manchette microtubules and sperm tail resembles that of the major outer dense fibre protein ODF1, recruitment of CFAP52 to these structures is independent of ODF1.
Expression analyses of Cfap52 by RT-PCR, notably nested RT-PCR, revealed transcription in all tissues investigated including those that do not harbour motile cilia, e.g. ovary. Additionally, Cfap52 was also expressed in the fibroblast cell line NIH3T3. Cfap52 cDNAs were isolated from NIH3T3 cells and cloned in frame to egfp to generate EGFP-fusion proteins when ectopically expressed. Investigations of the subcellular distributions of CFAP52::EGFP-fusion proteins as well as of the endogenous CFAP52 protein demonstrated location of CFAP52 in the centrosome, the mitotic spindle poles, the tip of the mid-bodies, and at the base of primary cilia. However, the primary cilium axoneme was not decorated by CFAP52. We observed centrosomal location of full-length CFAP52 and its isoforms including its N-terminal part comprising a single truncated WD40-domain. We propose that centrosomal recruitment most likely is mediated by the WD40-domain that generally functions in mediating protein-protein interactions 6 .
Our data indicate, that CFAP52 associates with a subset of microtubular structures, especially centrosome/ centriole derived structures. As WD-repeat proteins mediate protein interactions, CFAP52 might likewise act in assembly or stabilization of protein complexes constituting the centrosomal/basal body matrix and the sperm tail 6 . Furthermore, the location of CFAP52 to the sperm tail but not to the axoneme of primary cilia suggests that CFAP52 fulfils specific functions related to the stability, maintenance, or motility of the sperm tail. A loss-offunction of CFAP52 might therefore primarily affect the sperm tail and in turn male fertility. It is currently not known whether mutations in CFAP52 affect male fertility although mutations in WD-repeat proteins, as e.g. in WDR66/CFAP251, WDR96/CFAP43, and WDR52/CFAP54, causing human male infertility due to the sperm flagella defects have been reported [47][48][49] . Knock down or mutation studies of CFAP52 in zebrafish and humans, respectively, indicated impaired motile cilia in the ependym causing hydrocephalus, and impaired motile, solitary cilia in the node causing left-right symmetry disorders 7,8 . However, it has not been investigated whether primary cilia and the sperm tail are also affected, causing ciliopathies and male infertility due to sperm motility disorders, respectively. We anticipate our results to be a starting point for a more sophisticated analysis of the role CFAP52 plays in the formation of centrosome-derived structures and in male germ cells using knock out approaches.
Western blotting. Proteins were prepared from mice tissues (testes, brain, liver, kidney) and NIH3T3 cells.

Ethics statement.
All mouse experiments were reviewed and approved by the animal welfare commission of the University Medical Faculty and Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit. Licence for animal experiments has been obtained by the Institute of Human Genetics. The guidelines