Analysis of the male reproductive system

MaleCfap43mice are infertile, which hints to defects in sperm production or function. There-fore sperm was isolated from epididymis. Fewer sperm could be prepared from knock-out epididymis as compared to wild type. Although the effect was not quantified, the result is consistent with reduced amount of sperm fromCfap43knock-out mice (6.8×₁₀⁶ sperm cells from one wt epididymis vs. 0.2×₁₀⁶ sperm cells from one mutant epididymis) observed by Tang et al. (2017). Isolated sperm was diluted in methylcellulose to slow sperm movements and allow video-based analysis without use of a high-speed camera. Movement of single cells was documented by video microscopy and tracked using ImageJ manual tracking soft-ware. Tracked sperm cells from one video are shown as an overlay with the start-picture of the videos for wild type and mutant sperm (figure 2.30). Wild type sperm moved rapidly, mostly in wavy lines, in contrast most knock-out sperm cells did not move at all. In total only two out of 71 cells were found to be slightly motile (e.g. green line in figure 2.30,Cfap43-/-).

In addition, isolated sperm from knock-out animals appeared to be malformed.

To find causes for disturbed motility ofCfap43knock-out sperm and define potential mal-formations more precisely, sperm cells were isolated, spread on glass slides and stained for markers of different regions of the sperm tail (figure 2.31). In general, the flagella of mu-tant sperm seemed to be severely shortened and partly looped (asterisks, b, c, f, h, j, o) or displayed split axonemes (arrowheads, e, k, n). Additionally the connection between head and flagellum (neck, connecting piece) appeared to be angled or in the wrong position (ar-rows, d, i) in some cases. The acrosome of mutant sperm, which was marked by PNA-lectin,

Figure 2.30: Sperm motility analysis. Sperm movement was tracked in short videos of isolated epididymal sperm. Each line represents the movement of a sperm cell within 10 s. While wild type sperm covered a long distance, in this video only oneCfap43knock-out sperm cell (green line) moved at all.

2.5 Analysis of murine Cfap43 in vivo

was comparable to the acrosome in wild type sperm regarding the hook shaped form and positioning along the head. In both, wild type and mutant sperm malformed heads were occasionally observed (see table 2.5, no exemplary picture shown). COX IV labels mitochon-dria and thus marks the midpiece in wild type sperm. In mutant sperm the mitochonmitochon-dria containing section was strongly shortened, but the mitochondria were aligned along the ax-oneme and looping or other malformations were not observed within the midpiece. Septin7 localizes to the annulus, which marks the border between midpiece and principal piece. The annulus was present in wild type as well as mutant sperm, but positioned very close to the head. Thus, like staining of shortened stretches of mitochondria by COX IV also the posi-tioning of the annulus suggests a shortened midpiece in mutant sperm. Malformations of the axonemes, as looping or splitting, appeared always behind the midpiece, mostly begin-ning right after the annulus. AKAP3 is a structural component of the fibrous sheath, aligned along the principal piece of the sperm flagellum. In Cfap43 knock-out sperm AKAP3 was either missing completely or localized to the mid of looped or split axonemes. The structural

Figure 2.31: Malformation of Cfap43 knock-out sperm. Mutant sperm displayed shortened and looped (asterisks) sperm tails, sometimes split axonemes (arrowheads) or miss-positioned necks (ar-rows) could be observed. Since the phenotype is varying, more than one mutant sperm is depicted for each marker. AThe midpiece, which contains mitochondria (marked by COX IV), appeared short-ened in mutant sperm, but otherwise normal. BSeptin7 localizes to the annulus, which is the border between midpiece and principal piece. In agreement with the shortened midpiece labeled by COX IV, the annulus was located closer to the head. CThe fibrous sheath, marked by AKAP3, was either completely missing or was restricted to loops formed by the principal piece or appeared between split axonemes.

Table 2.5:Quantification of sperm malformations

Split Coiled Angled Malformed Total No tail Short tail axoneme axoneme neck head

Cfap43+/+ 185 7 2 0 1 2 4

% 3.78 1.08 0 0.54 1.08 2.16

Cfap43-/- 489 36 442 80 188 63 21

% 7.36 90.39 16.36 38.44 12.88 4.29

defects of spermatozoa were quantified and are summarized in table 2.5. Similar axone-mal axone-malformations and misslocalization of proteins of the fibrous sheath were described as dysplasia of the fibrous sheath (DFS) in sterile patients (Chemes et al., 1987; Moretti et al., 2016).

For better description and understanding of the defects occurring inCfap43-/- sperm epi-didymal sperm was analyzed by TEM (performed by Dr. Jan Hegermann, Institute for func-tional and applied anatomy, Hannover medical school). Cross sections of wild type sperm (figure 2.32 a-c) show the regularly arranged axoneme, outer dense fibers (ODFs, arrows), mitochondria (M) and the fibrous sheath (FS). In contrast, microtubuli are disorganized in knock-out sperm (figure 2.32 e-g). Since the axoneme is heavily disturbed, it is not possible to judge, whether the central pair is absent, as shown for humanCFAP43-/-sperm (Tang et al., 2017). The ODFs appeared rather normal in the midpiece (marked by mitochondria, e) and missarranged as well as split into more than nine fibers in the proceeding parts of the flagel-lum (f, g). The fibrous sheath appeared partly normal (f) but mostly absent (g). In addition longitudinal sections show the wild type axoneme to be lined by ODFs and mitochondria in the mid-piece or ODFs and the fibrous sheath in the principal piece (d). A longitudinal section through mutant sperm shows the disorganized arrangement of outer dense fibers, mispositioned mitochondria and an electron dense material, which might be dislocalized fi-brous sheath components. Hence, as suggested by immunofluorescence analysis of mutant sperm also TEM images show defects which are associated with DFS (Chemes et al., 1987).

Although the severe malformations observed in TEM images evolved most likely during spermiogenesis, the male reproductive system was histologically analyzed to determine the onset of the phenotype (figure 2.33). As expected first differences between wild type and Cfap43 knock-out sperm were already observed in the seminiferous tubules of testes (A (a, b vs c, d)). Flagella from mutant sperm were shorter and not protruding into tubule lu-men, but interspersed with spermatids. In addition the lumina from mutant seminiferous tubules were filled with cellular debris (A (c)), which was also observed in the tubules of the epididymal caput (B (d)). Consistent with fewer isolated sperm cells from knock-out epi-didymis described above less sperm was observed in both analyzed areas of the epiepi-didymis

2.5 Analysis of murine Cfap43 in vivo

as compared to wild type (B and C (a, e vs. c, g)).

To define the onset of sperm malformations more precisely, single stages of spermiogene-sis were identified as described by Oakberg (1956), and formation of wild type and mutant sperm was evaluated (figure 2.34). In early stages of spermiogenesis before flagellum for-mation (S1-S9) no differences between wild type and knock-out spermatids are evident by nuclear (DAPI, blue) and acrosome (PNA, green) staining (a-f and a’-f’). In sections stained for the axoneme (acetylated tubulin, red, figure 2.34 A) the most obvious defect in spermio-genesis ofCfap43-/- mice were the malformed flagella, which were observed before (figures 2.31 and 2.33). In addition mutant S16 spermatogonia were not able to align in phase VIII as seen in wild type seminiferous tubules (f vs. f’) and were not completely released until phase IX (g vs. g’). The first expression of AKAP3, which is a structural component of the fibrous sheath in mature sperm (Vijayaraghavan et al., 1999), was observed in S9 spermatids in wild type and mutant seminiferous tubules (figure 2.34 B (g and g’)). This protein is transported to the flagella during S10 (h and h’). Clear differences of the distribution of AKAP3 was visible in S11-12 spermatogonia (figure 2.34 B (i vs. i’)). Hence, the first defects in sperm morphol-ogy observed so far occur between spermiogenesis stages S10 and S12. Earlier defects might

Figure 2.32: TEM analysis of sperm fromCfap43knock-out mice. TEM images of wild type sperm (a-c) show the regularly arranged axoneme containing nine outer microtubule doublets (arrowheads) and the central pair (asterisks). ODFs (arrows) are visible throughout the mid-piece and parts of the principal piece (a, b, d). Mitochondria (M) localize to the mid-piece (a, d), whereas the fibrous sheath (FS) surrounds the axoneme in the principal piece (b-d). In mutant sperm (e-f) the axonemal structure is dissolved, but microtubuli are present. ODFs appear to be organized in the proximal parts of the flagellum (e, h), but circular arrangements is loosened more distally (f-h). Mitochondria were found to be arranged normally (e) or dislocalized (h), and the fibrous sheath might be present (f), absent (g), or disorganized (h).

Figure 2.33: Analysis of Cfap43knock-out testis and epididymis sections. A HE and fluorescent staining of each one seminiferous tubule showed long flagella within the lumen in wild type, whereas flagella from mutant sperm were short and interspersed with the developing spermatids. The lumen of mutant seminiferous tubules was filled with cellular debris. B Tubuli in the caput of wild type epididymis were filled with sperm, but only few spermatozoa (arrowheads) were visible between the debris in knock-out tubules. CThe cauda of wild type epididymis is mostly filled with sperm (arrowheads). In contrast, most tubules in Cfap43-/-cauda are empty, and sperm in filled tubules appeared to be less densely packed.

be detected using markers for different sperm components, which might be affected earlier than AKAP3.

2.5 Analysis of murine Cfap43 in vivo

Figure 2.34: Stages of sperm development. Spermiogenesis stages were judged based on nucleus (DAPI, blue) and acrosome (PNA, green). AThe axoneme (acetylated tubulin, red) of mutant sper-matogonia appears malformed. 16Cfap43-/-spermatogonia are not aligned at the seminiferous tubule lumen, as seen for the wild type (f’, f). Not all mature spermatogonia are released from mutant sem-iniferous tubules in phase IX (g’). BFirst expression of AKAP3 (red) occurs in S9 spermatogonia (g, g’). It is relocated to the flagellum during S10 (h, h’). Clear differences between wild type and mutant distribution of AKAP3 are visible from S11-12 (i, i’)

In section 2.2 the antibody P4 was described to detect a protein which partly colocal-izes with EZRIN and is downregulated inCfap43 knock-out mice. Among others, this anti-body was also tested in epididymis sections, where it stained actin-based cellular protrusions called stereocilia (figure 2.35 A (a)). As shown for trachea (figure 2.10) the signal was fainter and dislocalized (b) or absent (c) in sections of knock-out tissue, although stereocilia are still present in the knock-out (figure 2.33 C (b, d)). Arrowheads point to stereocilia, where the protein stained by P4 is localized in wild type tissues. Since the protein detected by P4 resembles the localization of EZRIN and EBP50, epididymis sections were stained for both proteins. EBP50 and EZRIN (described before by Höfer and Drenckhahn, 1996) localized to stereocilia in the epididymis tubuli (figure 2.35 B and C (a)). Hence, as in trachea the protein detected by P4 colocalizes with EBP50 and EZRIN. Interestingly, as the protein stained by P4, also EBP50 and EZRIN were absent or reduced in epididymal stereocilia in the mutants (B, C (b and c)). P4 was shown to detect other protein(s) than CFAP43 in immunofluores-cence (see section 2.2). Hence, localization of CFAP43 to the stereocilia is not proven by this immunofluorescence. However localization of EBP50 and EZRIN in epididymal stereocilia appears to be dependent on the presence of CFAP43, suggesting further functions of CFAP43 in addition to its contribution for motile cilia function.

2.5 Analysis of murine Cfap43 in vivo

Figure 2.35: Immunofluorescence analysis of stereocilia in the epididymis. AImmunofluorescence staining of epididymis sections using the antibody P4 resulted in a signal in stereocilia of epididymal tubuli (a). In knock-out tissue the signal can no longer be localized to stereocilia (b, c; two individ-uals). Arrowheads indicate expected sites of P4 signal. BEZRIN localizes as well to the epididymis stereocilia (a). In sections from knock-out animals (b, c; two individuals) the signal is weaker as compared to the wild type. Arrowheads point to stereocilia.

Cfap43 was identified in a set of microarray analyses as a potential FOXJ1 target gene. As such CFAP43 is a likely candidate, which might contribute to biogenesis or function of motile cilia. Although previously a possible connection to the actin cytoskeleton was suggested, the molecular and physiological function of CFAP43 remained elusive (Mai, 2012). Therefore different approaches were chosen to address CFAP43 function in motile cilia. Evolutionary conservation was evaluated by investigation of CFAP43 function in zebrafish. Expression domains of CFAP43 were investigated in more detail, as well as its subcellular localization.

Moreover, the function regarding motile cilia was investigated using murine knock-down and knock-out model systems.

3.1 CFAP43 function in zebrafish and evolutionary conservation

Before investigating the effect ofCfap43knock-down in zebrafish, its expression pattern was analyzed (figure 2.1). Therefore five antisense RNA probes derived from different regions of the Cfap43 cDNA were designed for detection of Cfap43 mRNA. Surprisingly, in situ hybridization using these probes showed no distinct localization of Cfap43. The staining achieved with these probes can be explained either by ubiquitous expression of Cfap43 in the zebrafish or by non-specific binding of each of the probes. Also in medaka, another well established fish model organism in biological research, Cfap43 expression was analyzed by in situhybridization (cooperation with Beate Wittbrodt, Institute for developmental biology and physiology, University of Heidelberg). Figure 3.1 shows widespread staining all over the embryo similar toin situhybridizations in zebrafish. Therefore, it appears unlikely, that

Figure 3.1: In situhybridization of Cfap43 in medaka. As for zebrafish,in situ hybridization us-ing a probe directed againstCfap43 in medaka does not show distinct localization of the transcript.

Experiments were performed by Beate Wittbrodt.

3.1 CFAP43 function in zebrafish and evolutionary conservation

of five independent RNA probes in zebrafish and one in medaka none was able to detect Cfap43. Thus, it appears possible, thatCfap43is ubiquitously expressed in embryos of lower vertebrates and strong expression becomes restricted to tissues containing motile cilia (e.g.

inXenopusand mice) later in evolution. This hypothesis is supported by low level expression of Cfap43 in murine tissues containing no motile cilia, as heart, salivary glands or muscles.

Furthermore this expression pattern suggests additional functions of CFAP43, which are not connected to motile cilia. Despite difficulties to localize Cfap43 to distinct structures or or-gans in the fish embryo, its expression was evident in isolated pronephric ducts, which carry among others motile cilia.

Subsequent to localization studies CFAP43 was knocked down using two different mor-pholinos. Knock-down by both morpholinos led to the same phenotypes, which appeared to be more severe after knock-down using the ATG-morpholino blocking translation from an existing mRNA (figure 2.3). The apparent higher efficacy of MO ATG could be explained by the different mode of action of the used morpholinos. In contrast to the splice site mor-pholino, which mediated knock-down by interfering with the splicing process and thus of freshly transcribed RNAs, translation blocking morpholinos can additionally act on maternal mRNAs. Although presence Cfap43 mRNA in the 1 hpf embryos was not detected by RT-PCR, low amounts of maternal mRNA could be present in early embryos. Using quantitative RT-PCRs a reduction of the Cfap43 transcript to 60% was evident in i4 MO injected fish by 48 hours after injection. Thus, the ATG morpholino might be able to mediate more effective knock-down than the splice site morpholino.

Specificity of phenotypes seen after morpholino injection is controversially discussed (re-viewed in Blum et al., 2015). After knock-down of CFAP43 using both morpholinos ventral body axis curvature, hydrocephali, pericardial edema, missaranged otholiths and dilated pronephric ducts were observed. The first argument for specificity of the morpholinos is the fact that both morpholinos resulted in the same phenotypes. Second, these phenotypes have been specifically linked to impaired cilia motility before (e.g. Kramer-Zucker et al., 2005; Krock and Perkins, 2008; Lunt et al., 2009). The most striking argument for specificity of the knock-down effect is the outcome of the rescue experiments. Except for body axis curvature caused by injection of the morpholino i4, all observed phenotypes were partially rescued by injection of murine Cfap43 mRNA. Thus, body axis curvature might partly be a result of artifacts due to morpholino injection. But all other phenotypes, and also strong ven-tral curvature caused by MO ATG, appeared to be specifically caused byCfap43knock-down, because they were less severe and occurred less frequently, when murine CFAP43 was ectopi-cally expressed in the morphants. Remarkably, the rescue could be performed by injection of the murine transcript, indicating that mouse CFAP43 is partly able to replace the function

of the homologous fish protein. Hence, the biochemical function of CFAP43 appears to be conserved between fish and mouse despite the apparent divergent expression pattern. The amount of injected mRNA was not titrated to simultaneously achieve the optimal amount for highest rescue effect and production of the lowest overexpression artifacts. Thus, it is not clear, whether the incomplete rescue is due to low amounts of the murine protein or its inability to completely fulfill the functions of zebrafish CFAP43.

In addition to ciliary function of CFAP43 in zebrafish,Xenopus Cfap43was determined to be expressed in tissues and cells containing motile cilia (cooperation with Tim Ott). High throughput studies in various organisms, which are summarized in CilDB (Arnaiz et al., 2014), detected CFAP43 in cilia of rat, Ciona intestinalis, Chlamydomonas andParamecium as well as its expression during differentiation of human cells containing motile cilia (Arnaiz et al., 2009; Baker et al., 2008; Nakachi et al., 2011; Pazour et al., 2005), suggesting an essential contribution of CFAP43 to motile cilia function from the earliest development of motile cilia during evolution.

In summary, although specific expression Cfap43 in tissues containing motile cilia could not be definitely shown in zebrafish, morphants displayed phenotypes that are associated with motile cilia dysfunction. Additionally these phenotypes were partially rescued by ec-topic expression of murine CFAP43 in the morphants. Thus, cilia-related phenotypes were specifically caused by CFAP43 knock-down and the function of the protein was proven to be evolutionary conserved between fish and mice. Taken in consideration that loss of CFAP43 results in comparable sperm defects in humans and mice (compare Tang et al. (2017) and section 2.5), functional conservation in all vertebrates appears plausible.