Identification of zebrafish oocyte-to-embryo transition mutants

To isolate mutations in genes regulating the oocyte-to-embryo transition in zebrafish, we induced mutations with the chemical mutagen N-ethyl-N-nitrosourea (ENU) and conducted a four-generation adult maternal-effect screen for egg and embryo phenotypes [13,36]. We examined 716 mutagenized genomes, and identified 9 recessive oocyte-to-embryo transition maternal-effect mutants (Table 1), which failed to produce eggs that were competent to undergo early embryonic development. In addition, we identified 9 maternal-effect mutants disrupting the cleavage stage of embryogenesis in zebrafish [37]. Five of the oocyte-to-embryo transition mutants displayed phenotypes indicating a primary defect during oogenesis, including one new mutant allele of the over easy (ovy^p37ad) gene [13]. Four additional mutant genes were identified that disrupt egg activation.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Zebrafish Maternal-effect Oocyte-to-Egg Transition and Egg Activation Mutant Genes.

https://doi.org/10.1371/journal.pgen.1011343.t001

Oocyte maturation genes

The largest group of mutants identified in our screen exhibits defects in oocyte maturation (Fig 1A). In zebrafish, the clarity of the yolk is a cytoplasmic indicator of this process. Prior to oocyte maturation in wild type, the yolk is opaque and the bulk of the cytoplasm is intermingled between the YGs in the oocyte. During oocyte maturation, the yolk proteins are cleaved, generating the translucency of the egg (Fig 1A) [38]. Three mutants reported here, p33bjta, p26ahubb, and p35aluc, produced eggs that were opaque and lacked a blastodisc, more closely resembling wild-type oocytes than eggs (Fig 1A). The other two mutants of this class, p26ahubg and p25bdth, only produced degenerated eggs (Figs 1B and S1A), indicating a more severe or distinct defect in the oocyte-to-egg transition.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Zebrafish oocyte maturation mutants.

A. Wild-type stage IV oocyte and egg in incident light. Eggs of p33bjta (1570 from 21 females), poac^p26ahubb (1620 from 35 females), and p35aluc (2088 from 25 females) resemble wild-type oocytes in their opacity, while eggs of p26ahubg (546 from 8 females) and blac^p25bdth (1020 from 17 females) were degenerated. B. Images of 5 minutes post-egg activation (mpa) wild-type (n = 68), p26ahubg (n = 56) and blac^p25bdth (n = 61) eggs, showing lysed mutant eggs. bd, blastodisc. C. Non-complementation test showed that females carrying the ovy^p37ad mutation in trans to the p35aluc mutation produced opaque mutant eggs (n = 1382 from 9 females), showing that p35aluc is a new ovy allele. D. Coomassie stained gel of major yolk proteins in wild-type and mutant eggs showing that the cleavage of yolk protein is impaired. kD, kilodaltons. E. Low (top row) and high (bottom row) magnification histological sections of wild-type and mutant stage III oocytes stained with H&E revealing YG sizes. Scale bar = 96 μm (top row) and 16 μm (bottom row). F. Scatter plots of YG diameters measured in a defined region of the oocyte. The mean diameter of the YGs in the mutants was significantly different to that of wild-type oocytes (WT = 14.6 μm, p33bjta = 3.87 μm, p35aluc = 5.95 μm, p25bdth = 9.25 μm). The number of YGs analyzed in this study was: wild type (n = 188 from 4 oocytes), p33bjta (n = 320 from 2 oocytes), ovy^p35aluc (n = 503 from 2 oocytes), and blac^p25bdth (n = 177 from 2 oocytes). Data are means ± SEM. ****p<0.0001 and **p = 0.0066 in parametric statistical Tukey’s test. ns, not significant. G. Confocal z-projections of isolated acid fixed oocytes analyzed by fluorescent in situ hybridization and also DiOC₆ staining (green) in lower panels. In stage II oocytes, cycB1 transcript localized to the animal pole in wild-type (n = 32), p33bjta (n = 22), poac^p26ahub (n = 47), ovy^p35aluc (n = 38), and blac^p25bdth (n = 28) oocytes. In stage I oocytes, dazl mRNA co-localized with DiOC₆ to the vegetal Balbiani body (Bb) in wild-type (n = 32), p33bjta (n = 23), poac^p26ahub (n = 63), ovy^p35aluc (n = 31), and blac^p25bdth (n = 29) oocytes. Scale bar = 35 μm.

https://doi.org/10.1371/journal.pgen.1011343.g001

To determine if the mutations represent alleles of the same or distinct genes, we determined the chromosomal positions of each mutation (Table 1). The p35aluc and p26ahubb mutations mapped to a similar chromosomal location as the ovy^p37ad mutation, which also causes an opaque egg phenotype [13]. Hence, we conducted complementation tests between these alleles. We found that p35aluc failed to complement ovy^p37ad (Fig 1C), indicating that p35aluc is an allele of ovy^p37ad. In contrast, we found that p26ahubb complemented ovy^p37ad, indicating that this mutation disrupts a different gene, which we named poached (after a type of cooked egg_; hereafter called poac^p26ahubb). Among the mutants identified in this screen, only p35aluc (hereafter called ovy^p35aluc) represented a new allele of a previously reported zebrafish maternal-effect gene. We did not find linkage between the p33bjta mutation and the SSLP markers tested. However, our analysis indicates that the p33bjta mutation is not linked to the chromosomal intervals of the other opaque egg mutations reported here or previously [13], suggesting that it disrupts a distinct gene.

Homozygous females of the p25bdth and p26ahubg mutations produced an identical phenotype, lysed eggs at egg laying (Figs 1B and S1A). We found that both mutations mapped to the same chromosomal position (Table 1) and failed to complement each other (see Materials and Methods), thus indicating that these two alleles disrupt the same gene (hereinafter called black caviar (blac)).

Oocyte cytoplasmic maturation of YGs

The zebrafish stage III oocyte is opaque due to the accumulation of membrane-enclosed crystalline yolk; however, at the end of oocyte maturation the yolk becomes non-crystalline and translucent [16,38]. This process is driven by the proteolytic processing of the major yolk proteins during oocyte maturation in stage IV oocytes, changing their refractive index from opaque to translucent. The persistent opaque yolk in these mutant eggs suggests that cleavage of the major yolk proteins during oocyte maturation is impaired. To investigate this directly, we prepared cytoplasmic extracts from wild-type immature stage IV vitellogenic oocytes and eggs, and compared the yolk protein profiles to those of the oocyte maturation class of mutant eggs. We found that the yolk protein composition of mutant eggs of blac^p25bdth, poac^p26ahubb, p33bjta and ovy^p35aluc resembled wild-type oocytes rather than wild-type eggs (Fig 1D), indicating that cleavage of the vitellogenic proteins was indeed defective during oocyte maturation in these mutants.

To further understand the YG maturation defect, we asked whether YG morphology and integrity was also affected. We therefore examined the size of YGs in wild-type and mutant stage III oocytes in hematoxylin and eosin (H&E) stained histological sections (Fig 1E). We found that the YGs of wild-type stage III oocytes displayed a mean diameter of 14.6 μm (Fig 1F). In p33bjta, ovy^p35aluc and blac^p25bdth mutants, the YGs were on average 3.7-, 2.5-, and 1.6-fold smaller than in wild type, respectively (Fig 1E and 1F). The p33bjta phenotype was more severe than ovy^p35aluc, displaying a much greater amount of small YGs and no large YGs (Fig 1E and 1F). In blac^p25bdth mutant oocytes numerous, smaller YGs were present, but some large YGs also formed, with a few larger than in wild type (over 40 μm) (Fig 1E and 1F). Thus, quantification of YG size confirmed a significant difference between wild type and mutants (Fig 1F), likely associated with altered YG formation during oogenesis in ovy^p35aluc and blac^p25dbth mutants.

Histological sections of earlier stage I and II oocytes revealed no overt differences between mutants and wild type (S1B Fig). H&E staining indicated that previtellogenic oocytes were normal; cortical granules (CGs) accumulated and later localized to the oocyte cortex in ovy^p35aluc, p33bjta, and blac^p25bdth mutant stage III oocytes (S1B Fig).

These data indicate that YG size was severely compromised in ovy^p35aluc and p33bjta mutants; thus, suggesting that the products of the ovy and p33bjta genes promote the formation of large YGs. Taken together, the anomalies observed in these mutants strongly suggest that p33bjta, ovy and blac regulate the initial steps of YG formation and sizing, and yolk protein processing, independent of other aspects of oocyte cytoplasmic maturation.

Oocyte maturation mutants exhibit defects in egg activation and early embryogenesis

All mutants of this class produced eggs that failed to segregate cytoplasm to the animal pole to form the blastodisc and showed no overt signs of egg polarity (Fig 1A–1C). Therefore, we examined markers of the animal-vegetal axis to investigate whether failed blastodisc formation was due to a defect in oocyte polarity. We found that, like in wild type, cyclinB1 (cycB1) mRNA localized to the animal pole of p33bjta, ovy^p35aluc, poac^p26ahubb and blac^p25bdth stage II oocytes and deleted in azoospermia-like (dazl) mRNA localized to the vegetal Balbiani body of stage I oocytes (Fig 1G). Therefore, impaired oocyte polarity does not cause the lack of animal-ward segregation of cytoplasm that forms the blastodisc.

We also investigated whether p33bjta, poac^p26ahubb and ovy^p35aluc mutant eggs completed meiosis following egg activation by examining second polar body formation [39]. We found that meiosis resumed normally at 5 minutes post-egg activation (mpa) in the mutants, with sister chromatids evident in anaphase II (Fig 2A). By 10 mpa, an actin associated second polar body was detected in wild-type and mutant eggs (Fig 2A), indicating normal completion of meiosis II.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Meiosis II completion and egg activation defect in dark eggs mutants.

A. Confocal micrographs of the animal pole show anaphase of meiosis II at 5 mpa and the extruded polar body at 10 mpa. Top row: DAPI-stained wild-type (n = 24), p33bjta (n = 14), poac^p26ahubb (n = 22) and ovy^p35aluc (n = 21) sister chromatids during meiotic anaphase II. Bottom row: Phalloidin-stained actin (red) and DAPI staining (green) of second polar bodies in wild-type (n = 14), p33bjta (n = 9), poac^p26ahubb (n = 12) and ovy^p35aluc (n = 10) activated eggs. Scale bar = 4 μm. B. Egg-to-chorion index (ECI) is the ratio of egg area (A₁) to chorion area (A₂) as a measure of chorion elevation. Scatter plots of the ECI measured in activated eggs at 30 mpa. The wild-type egg ECI value (ECI = 0.61; SD = 0.02; n = 65) defines the numerical value for normal chorion elevation. The ECI for the eggs from p33bjta (ECI = 0.64; SD = 0.05; n = 44, 2 females) and ovy^p35aluc (ECI = 0.63; SD = 0.03; n = 20, 2 females) mutant females are significantly different to wild type, while the eggs of poac^p26ahubb (ECI = 0.59; SD = 0.02; n = 25, 2 females) mutant females show a mean value similar to wild type. SD = standard deviation. Data are means ± SD. *p = 0.0131 and ***p = 0.0004 in parametric statistical Tukey’s test. C. Confocal z-projections (5 μm depth) of acid fixed and MPA stained wild-type and mutant activated eggs. In wild-type (n = 60), p33bjta (n = 55), ovy^p35aluc (n = 25), and poac^p26ahubb (n = 27) eggs, numerous large and small CGs were distributed throughout the cortex at 0 mpa. At 30 mpa, wild-type (n = 21), p33bjta (n = 28), poac^p26ahubb (n = 20) and ovy^p35aluc (n = 18), eggs displayed fewer CGs, indicating that they were released following activation. However, some CGs persisted in p33bjta (5/28), poac (11/20) and ovy^p35aluc (6/18) mutant eggs. D. Scatter plots of CG retention measured in a determined lateral region (DLR) of activated wild-type and opaque mutant eggs. The total area of retained CGs per total DLR area in all dark egg mutants was significantly increased compared to wild type at 30 mpa. n, number of eggs used in this analysis from 3 females. Data are means ± SEM. *p = 0.0249, **p = 0.0485 between wild type and poac, and ***p = 0.0042 between wild type and p33bjta in a nonparametric statistical Kruskal-Wallis test. ns, not significant; mpa, minutes post-activation. E. Scatter plots of CG number (top graph) and median size (bottom graph) measured in the DLR of the wild-type and dark activated eggs. The average number of CGs in the mutants was comparable to wild type at 0 mpa but significantly different in the p33bjta and poac^p26ahubb mutants at 30 mpa. Unactivated (wild type (n = 7), p33bjta (n = 7), poac^p26ahubb (n = 7) and ovy^p35aluc (n = 5)) and activated (wild type (n = 8), p33bjta (n = 5), poac^p26ahubb (n = 4) and ovy^p35aluc (n = 6)) eggs from 3 females per genotype were used in this analysis. Data are means ± SEM. Top graph: *p = 0.0353 and **p = 0.0012 in a nonparametric statistical Kruskal-Wallis test. Bottom graph: *p = 0.0250 in a nonparametric statistical Kruskal-Wallis test, and **p = 0.0011 in a parametric statistical Tukey’s test. F. Top row: Animal views of DAPI stained p33bjta/+ (n = 34, 2 females), and p33bjta (n = 36, 2 mutant females) 4 hpf embryos, showing abnormal chromatin organization. Middle row: Lateral views of 4-cell stage embryos from p33bjta/+ (n = 77/111, N = 2 females), and p33bjta (n = 29/59, N = 3) females. Bottom row: Lateral views of 4 hpf embryos from p33bjta/+ (n = 77/111, N = 2) and p33bjta (n = 29/59, N = 3) females. G. Top row: Animal views of DAPI stained 4 hpf embryos from ovy^p35aluc/+ (n = 19, N = 2), and ovy^p35aluc (n = 9, N = 2 females), showing abnormal chromatin organization. Middle row: Lateral views of 4- and 8-cell stage embryos from two heterozygous (n = 110), and one homozygous (n = 69) ovy^p35aluc females. Bottom row: Lateral views of 4 hpf embryos from 2 heterozygous (n = 110), and one homozygous (n = 10) ovy^p35aluc females. hpf, hours post fertilization. Scale bar = 4 μm (A), 40 μm (C), 210 μm (F, G).

https://doi.org/10.1371/journal.pgen.1011343.g002

To address whether other aspects of egg activation were altered in the mutants, we investigated CG exocytosis and chorion elevation. We first defined the egg-to-chorion index (ECI) as a readout of chorion elevation. To calculate ECI, activated egg and chorion areas were measured at 30 mpa (see Materials and Methods). Quantitative analysis of the ECI of poac^p26ahubb eggs showed normal chorion elevation (p>0.05) (Fig 2B). However, p33bjta and ovy eggs displayed a significant, variably expressive reduction in chorion elevation (Figs 1C, 2B, and 2F).

Since chorion elevation is mediated by CG exocytosis, we next examined CG size, area, and number in unactivated eggs, as well as their reduction in activated eggs as an indicator of their exocytosis. We found in wild-type eggs that both small and large CGs were distributed at the cortex before activation with few remaining after egg activation, indicating their exocytosis (Fig 2C–2E). Unactivated eggs of p33bjta, poac^p26ahubb, and ovy tended to have a reduced CG area and ovy a reduced CG number compared to wild type, although the difference was not statistically significant (Fig 2D and 2E). At 30 mpa, however, a substantial quantity of CGs and/or CG area remained in p33bjta, poac^p26ahubb, and ovy mutant eggs (Fig 2C–2E), which may cause the variably expressive chorion elevation phenotype. In wild type, we measured a median CG diameter of 22.3 μm and 6.2 μm before and after activation, respectively (Fig 2E, right). In contrast, in p33bjta unactivated mutant eggs, the median size of CGs was significantly smaller, 11.4 versus 22.3 μm in wild type. On the other hand, at 30 mpa, the CG average median size in the ovy^p35aluc mutant was 32.7 μm, a value significantly larger than in wild type (Fig 2E). Notably, poac^p26ahubb activated eggs exhibited normal chorion elevation (Fig 2B), while a significant number of small (mean 5.3 μm diameter) CGs persisted (Fig 2E), indicating that the release of the larger ones is sufficient to induce normal chorion elevation.

To examine whether the cytoplasmic defects in oocyte maturation and egg activation impacted embryonic development, we examined the embryos of two of the mutants, p33bjta and ovy mutant females (Fig 2F and 2G). We found that modest cleavage furrows were evident in some mutant embryos at the 4- and 8-cell stages, however, a blastoderm was not evident at 4 hours post fertilization (hpf). DAPI staining at 4 hpf revealed three classes of blastulae for each mutant compared to wild type, characterized by a single nucleus or pronounced clumping or fragmentation of nuclei at the surface of the opaque yolk, indicating that nuclear divisions proceeded in some embryos but that cytokinesis failed (Fig 2F and 2G). As blac eggs degenerate after laying, we did not perform meiosis resumption, egg activation and DAPI staining experiments of this mutant.

These data show that features of egg activation are perturbed in the p33bjta, poac^p26ahubb, and ovy mutants. While the capacity to undergo meiotic divisions and polar body extrusion is normal, defective cytoplasmic segregation and embryonic progression suggest that specific egg activation traits are regulated by bifurcating or distinct genetic networks.

ovy encodes adaptor-related protein complex 5, mu 1 subunit (Ap5m1)

To determine the molecular nature of the over easy gene, we performed RNA sequencing (RNA-seq) of ovy^p37ad mutant eggs, an approach used previously to identify mutant genes in zebrafish [40–43]. A search of the top ten downregulated genes in the mutant egg transcriptome identified only one gene in the chromosomally defined ovy interval on Chr 17, the adaptor-related protein complex 5, mu 1 subunit (ap5m1) gene (log₂ fold change = -7.0632) (Fig 3A). It encodes one of four subunits of the fifth Adaptor Protein complex (AP5) [44], the mu or μ subunit. The AP5 heterotetramer complex localizes to a late endosomal/lysosomal compartment and functions in sorting late endosomes back to the Golgi in cultured mammalian cells [44–47].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Molecular nature of the ovy gene.

A. Genetic and physical map of the ovy locus and schematic representation of the zebrafish ap5m1 gene, which consists of 8 coding exons and 7 introns. z7170 and z8980 are SSLP markers flanking the ovy mutation. In parenthesis, the number of recombinants/total analyzed meioses defining the interval of ovy. Exons are shown as red boxes and introns as black lines. Sizes are not to scale. B. DNA sequencing analysis of the ovy^p37ad and ovy^p35aluc mutant alleles. Left: genomic DNA sequence of the wild-type and ovy^p37ad allele indicates a single point mutation in the splice donor site of intron 1 of the ap5m1 gene (red rectangle). Right: cDNA sequence of the wild-type and p35aluc allele indicates a single point mutation in exon 2 of the ap5m1 gene (red rectangle). C. Left: predicted tertiary structure of the zebrafish Ap5m1 protein. The α- and β-helixes are colored in purple and pink, respectively, and the connecting loops in cyan. N- and C-terminal domains are indicated. The boundary of the two domains is at residue 205 (arrow head), which is found in the flexible loop comprised by residues 204–208. The approximate volumetric density map of the protein is shown in transparent gray. Right: predicted domain architecture of the Ap5m1 N-terminal portion containing residues Arginine 24 (R24) and Threonine 27 (T27). The amino acid change and premature stop codon caused by the ovy^p35aluc and ovy^p37ad mutations, respectively, are shown. D. Multiple amino acid alignments of the Ap5m1 protein and representative members of the vertebrate lineage. Top: Schematic of the protein alignment. The overall percentage (%) identity decreases from top to bottom. The red box indicates the first 49 amino acid residues of Ap5m1. Bottom: Detailed amino acid alignment showing high conservation in fish, amphibians and mammals. Note the high conservation of the Threonine (T) amino acid (red asterisk), which is mutated to Lysine (K) in ovy^p35aluc (indicated in red in the lower panel, along with the splice site mutation, causing a premature stop codon). E. Structural superposition between zebrafish Ap5m1 (pink, amino acids 205–475) and the crystal structure of human Ap4m1 C-terminal domains (purple, amino acids 185–453). The RMSD value corresponds to 2.45 Å, thus demonstrating structural identity in overall 3D structure.

https://doi.org/10.1371/journal.pgen.1011343.g003

Next, we amplified the wild-type and mutant ap5m1 cDNA in three overlapping amplicons to sequence the entire coding region (S2A Fig and S1 Table). The middle and 3’ end overlapping PCR products showed no differences in sequence between wild type and mutant. However, the most 5’ ap5m1 amplicon amplified successfully from wild type, but was not obtained from mutant cDNA, despite repeated attempts, suggesting a possible splicing defect. Sequencing the 5’ and 3’ ends of intron 1 of ap5m1 in ovy^p37ad genomic DNA (gDNA) confirmed a splice donor site mutation (G to A) in the exon1-intron 1 junction (Figs 3B and S2B). This results in predicted retention of the first intron in the transcript. Using primers within the first intron and either exon 1 or exon 2, we found that indeed the ovy^p37ad cDNA retained the intron, resulting in a premature stop codon in the intronic sequence (S2B and S2C Fig), expected to cause a loss of all functional domains of the Ap5m1 protein. We then sequenced the ap5m1 gDNA and cDNA from the second ovy^p35aluc allele and identified a missense mutation (C to A transversion), changing amino acid threonine 27 to lysine (Fig 3B and 3C). This residue is highly conserved in all 9 vertebrate species examined (Fig 3D), suggesting its importance to Ap5m1 function.

To molecularly characterize the Ap5m1 protein, we generated a homology-based structural model of the zebrafish Ap5m1 protein (Fig 3C). Our model indicates that Ap5m1 is comprised of an N-terminal α-helical domain (residues 1–205) and β-sheet structured C-terminal domain (residues 206–492) (Fig 3C). Immunoprecipitation experiments suggest that the N-terminal domain interacts with another subunit of the AP5 complex [48]. The ovy^p35aluc mutation impacts a peptide loop in the N-terminal domain (Fig 3C). The threonine to lysine substitution likely provokes a conformational change that would affect its binding activity [46]. In contrast, the C-terminal or Mu homology domain may be the site of specific cargo recognition and interaction [44]. Importantly, the AP5m1 C-terminal domain showed high conservation with acidic residues of the adaptor protein complex 4 mu4 (AP4m4) subunit C-terminal domain and its overall predicted structure (Fig 3E). AP4m4 is a subunit of the human AP4 complex (another heterotetramer involved in vesicle trafficking), and plays a role in mediating the transport of cargo from the Golgi to endosomes [46,49]; thus suggesting that AP5m1 could function in endosomal trafficking during oogenesis.

The morphological and molecular phenotypic features of the ovy/ap5m1 mutant suggest a functional relationship between the endolysosomal route, YG size acquisition, and yolk protein cleavage during zebrafish oogenesis, which was not known previously in egg-laying organisms. Thus, we provide the first phenotypic study in an animal model system of ap5m1 gene function.

Egg activation genes

We identified a second group of four maternal-effect mutants with defects in the egg-to-embryo transition (Table 1): p30ahub (henceforth called krang^p30ahub, named for a fictional supervillain standing in a confined space), p08bdth (hereinafter called spotty^p08bdth, named for its spotty cortical pockets of cytoplasm in the yolk cell cortex), p28tabj, and p26thbd (henceafter called kazukuram (kazu^p26thbd), which in the Mapuche language, Mapudungun, means gray (kazu) egg (kuram)). We examined two processes of egg activation in these mutants, chorion expansion and cytoplasmic segregation. Immediately after wild-type eggs are laid and prior to their activation, the chorion is closely associated with the egg surface (Fig 4A). Following egg activation, the chorion elevates due to CG exocytosis and the blastodisc forms by the initiation of cytoplasmic transport to the animal pole. We found that the krang^p30ahub mutant showed a severe, highly penetrant small chorion phenotype (Fig 4A). The three other mutants, spotty^p08bdth, p28tabj, and kazu^p26thbd, displayed abnormalities in cytoplasmic subcellular reorganization, subsequent cell cleavage and blastoderm formation, but did not compromise chorion elevation (Fig 4A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Egg activation mutants.

A. Bright-field images showing early developmental stages of live wild-type and mutant eggs and embryos. All unactivated eggs collected after gently squeezing gravid females showed no detectable defects. After egg activation, however, the krang mutant egg displays a pronounced chorion elevation defect (black arrows). As development proceeds, the cytoplasm is abnormally distributed in the yolk cell (yc) of spotty mutant eggs and early embryos (black arrowheads) and a defective blastoderm (bt) is formed. krang (1887 eggs from 50 females), spotty (1273 eggs from 40 females), and p28tabj (2506 eggs from 34 females) and kazu (1608 eggs from 32 females). B. Lateral view of acid-fixed wild-type and mutant eggs and early embryos showing the organization of the central cytoplasm (bright). In wild type, the blastodisc (bd) grows by animal-ward transport of cytoplasm from the yolk cell (yc) and by the 4-cell stage, the cytoplasm in the yolk cell is no longer evident. In contrast, spotty, p28tabj and kazu mutant eggs exhibited an abnormal distribution and severe retention of cytoplasm in the yolk cell (black asterisks). Wild type (n = 22), spotty (n = 17), and p28tabj (n = 19) and kazu (n = 12). C. Lateral view of acid-fixed and DiOC₆-stained wild-type and mutant eggs and early embryos showing the organization of the cortical cytoplasm (green) and yolk (dark). Wild type (n = 20), spotty (n = 20), and p28tabj (n = 14) and kazu (n = 18). Scale bar = 830 μm (A, columns 1 and 2 from left to right), 150 μm (A, columns 3 and 4 from left to right), 190 μm (B, C).

https://doi.org/10.1371/journal.pgen.1011343.g004

We mapped the krang, spotty, and kazu mutations to chromosomal loci and determined that p28tabj was not linked to the intervals of the other mutations reported here or previously (Table 1. See Material and Methods). Based on our phenotypic and mapping analysis, we conclude that these mutations disrupt distinct genes required for chorion elevation, cytoplasmic segregation, and early embryo formation. Importantly, we identified genes that appear to control divergent processes acting in the zebrafish egg-to-embryo transition.

Cytoplasmic reorganization and blastoderm formation are controlled by maternal egg-to-embryo transition genes

In most animals with telolecithal eggs and discoidal cleavage, including zebrafish, egg activation is characterized by blastodisc elevation, which then forms the blastoderm and ultimately the embryo (reviewed in [50]). During blastodisc growth, the cytoplasm is transported to the animal pole through the process of cytoplasmic segregation in two main phases: the slow (20–30 minutes post fertilization, mpf) and fast (30–40 mpf) flow of cytoplasm [29,32]. The spotty and p28tabj mutants displayed apparent normal blastodisc formation at 20 and 40 mpf, while kazu mutant eggs appeared normal at 20 mpf but displayed a smaller blastodisc at 40 mpf (Fig 4A and 4B). For all three mutants the subsequent cell cleavage stage was morphologically abnormal, with altered cell size and organization, and the spotty mutant also displayed cortical pockets of cytoplasm within the yolk cell (Fig 4A and 4C).

To further explore the cytoplasmic segregation process, we visualized the organization of cytoplasmic domains and organelles. In acid-fixed wild-type eggs, the cytoplasm is distributed in the blastodisc and across most of the yolk cell, intermingled with YGs (Fig 4B). As development proceeds, the blastodisc grows by transport of cytoplasm from the yolk cell via specialized transportation pathways or streamers to the animal pole (Fuentes and Fernandez, 2010). The spotty, p28tabj and kazu mutant zygotes showed marked changes in the organization of cytoplasm distributed in the yolk cell during the first phase of cytoplasmic segregation (Fig 4B, 20 mpf). During the fast phase of cytoplasmic segregation when axial streamers form, the distribution of membranous organelles was also affected in the mutant zygotes (Fig 4C, 40 min, 1-cell stage). Cytoplasmic patches in various regions of the yolk cell were present at the 4- and 8-cell stages in these mutants but not wild type (Fig 4B and 4C). Thus, defective cytoplasmic organization and animal-ward streaming, likely explain its retention in the yolk cell in all mutant zygotes studied. This suggests that the mechanisms regulating cytoplasmic segregation are compromised in these mutants.

Maternal Krang plays a role in cortical granule biology during egg activation

The eggs of krang females exhibited a fully penetrant small chorion phenotype, as determined by quantitating the extent of chorion elevation (Fig 5A). To investigate whether the pronounced chorion elevation defect in the krang mutant results from a defect in CG synthesis, translocation or exocytosis, we performed MPA-staining to specifically label and track CGs. We found that both small and large CGs formed and were cortically distributed in krang eggs before activation (Fig 5B). However, at 20 mpa, CGs persisted in the mutant egg, along with numerous small CG-like vesicles showing reduced MPA staining within the vesicle (Figs 5B, 5C, and S3A and S1 Video). By counting the number of CGs per DLR (determined lateral region, see Methods) at different time points after egg activation, we found, as expected, rapid CG exocytosis in wild type, while no such phenomenon occurred in the mutant eggs (Fig 5C). Although unactivated krang mutant eggs did not exhibit a significantly reduced count of CGs, the number remaining at multiple time points following egg activation showed a significant difference (Fig 5C). Furthermore, in krang^p30ahub mutants, the rate of CG exocytosis was slower from 0 to 20 mpa (Fig 5C). Quantitative analysis of CG size showed that their mean diameter at 30 mpa is significantly larger in the mutant at 34.3 μm, than in wild type, at 11.9 μm (S3B Fig). In wild type, the median CG size was 10 μm with a similar 11.3 μm median size in krang mutants (S3B Fig). These measurements suggest a retention of larger CGs, accompanied by many small CGs, unlike in wild type (Figs 5B, S3A, and S3B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Characterization of the maternal-effect krang mutant phenotype.

A. Quantification of the chorion elevation phenotype. Scatter plots of the measured ECI value show a penetrant chorion elevation defect in the krang mutant egg at 30 mpa (ECI = 0.64 and 0.83; SD = 0.006 and 0.005; for the wild-type (n = 48, 2 females) and mutant (n = 48, 2 females) egg, respectively). SD = standard deviation. Data are means ± SD. ****p<0.0001 in statistical clustering analysis in a nonparametric unpaired t test of chorion elevation measurements. B. Confocal z-projections (5 μm depth) of MPA stained CGs showing their distribution in wild-type (top row, n = 68 eggs from 3 females) and krang mutant (bottom row, n = 55 eggs from 3 females) eggs at two different time points after activation. Images were taken in a determined lateral region (DLR) of each activated egg. Notice that smaller, possibly CGs lacking MPA-staining lectin accumulate in the mutant egg at 20 mpf. C. Scatter plots of CG exocytosis timing measured in the DLR of the activated egg. The average number of CGs in the krang mutant was not reduced compared to wild type at 0 min but is significantly higher at later time points. Normalized total area curves (upper right graph) showing CG exocytosis rate in wild-type and krang mutant activated eggs. Five eggs from 3 females per genotype and per time point were used in this analysis. Data are means ± SEM. *p = 0.0131 and **p = 0.0252 in a parametric statistical two-way ANOVA test. ns, not significant. D. Hematoxylin and Eosin stained sections of intact stage II (left column) and stage III (right column) oocytes. Wild-type (n = 2 ovaries), and krang (n = 2 ovaries) oocytes, were phenotypically comparable. CGs are formed (stage II) and accumulated at the cortex (stage III). E. Confocal micrographs of actin- and DAPI-stained fertilized eggs showing normal meiosis completion and second polar body formation (wild type (n = 6, 2 females) and mutant (n = 15, 2 females). F. Most krang mutant embryos failed to undergo cell divisions. Top row: wild-type embryo with a typical symmetric blastoderm at 2 hpf (left panel, n = 87/93 embryos from 2 females). In contrast, krang mutant embryos display relatively normal (middle panel, n = 78/227) or abnormal (right panel, n = 147/227) blastoderm formation (n = 3 mutant females). Bottom row: high magnification confocal micrographs showing the cell architecture in wild-type and mutant blastoderms. cg, cortical granule; yg, yolk globules. Scale bar = 40 μm (B), 25 μm (D, left column), 95 μm (right column), 50 μm (E, left column), 12 μm (E, right column), 160 μm (F, top row), 85 μm (F, bottom row).

https://doi.org/10.1371/journal.pgen.1011343.g005

To explore CG formation and localization during oogenesis, we examined ovaries of homozygous mutant females. We found that oogenesis in krang mutants appeared morphologically and histologically normal, where CGs formed in stage II oocytes and then moved and docked to the cell cortex in later oogenesis stages (Fig 5D). Collectively, these results indicate that CGs form and translocate in the krang^p30ahub mutant as in wild type. Possible changes in CG translocation dynamics, content, or exocytosis after egg activation might impact the magnitude of chorion elevation.

We next investigated development of krang mutant embryos. We found that krang^p30ahub eggs were fertilized, meiosis II resumed and completed, as revealed by segregating anaphase II chromosomes and second polar body formation (Fig 5E). Also, the initiation of the first mitosis ensued in krang mutant zygotes (Figs 5F and S3E). However, during early embryonic development, many krang^p30ahub embryos failed to undergo cleavage or did so abnormally, often dying during blastula stages (Figs 5F and S3C). The krang^p30ahub embryos that survived to 1 dpf exhibited a wild-type phenotype, or a reduced or ventralized body axis (S3C Fig). To test whether defective chorion elevation impairs the cleaving ability of krang early embryos, we removed the chorion by pronase treatment after fertilization. Removal of the chorion resulted in an increased number of cleaving mutant embryos that reached the gastrula and 1 dpf stages, mostly displaying a wild type-like phenotype (S3C and S3D Fig).

To investigate whether defects in chorion biogenesis were associated with the chorion elevation defect, we examined its ultrastructure in cross-sectioned wild-type and krang whole ovaries by transmission electron microscopy. The chorion is derived from the vitelline envelope, which is made during oogenesis. In the zebrafish, the structure of the vitelline envelope consists of three zones organized horizontally around the oocyte [38,51]. In analyzing the vitelline envelope ultrastructure, we did not find any morphological variation between wild-type and krang oocytes (S3F Fig). These results indicate that the chorion elevation defect, but not its biosynthesis, influences the developmental potential of mutant embryos to progress. When removed, it improved cell cleavage and axis formation (S3C and S3D Fig).

Krang encodes a novel protein

To identify the molecular nature of the krang mutant gene, a combination of positional cloning and next generation sequencing strategies was carried out. We mapped the krang^p30ahub mutation to chromosome 18 between markers z65576 and z10008. In testing 1380 females for recombination in the interval, we finely mapped the critical interval of the mutation to 0.87 cM between markers CR925798-1 (38.93 cM) and z58289 (39.8 cM) (Fig 6A). We performed a genomic DNA sequence capture of the entire 1.2 Mb interval from a krang^p30ahub homozygote and wild-type control [52–55], followed by next generation sequencing of the captured DNA. Sequence analysis revealed only 3 single base pair (bp) changes within the physical interval of the mutant gene. Two of the changes were in a large first intron of a gene and the third was in the coding sequence of the same gene. This coding sequence change did not alter the Gly amino acid encoded, but was a wobble codon change from GGC to GGT. None of the changes appeared deleterious, but all were present in the same gene, kiaa0513. Thus, we cloned and sequenced the cDNA of kiaa0513 from krang^p30ahub and wild-type ovaries to determine if it was affected. Interestingly, sequencing of the cDNA revealed a 35 nucleotide (nt) deletion at the 3’-end of exon 3, precisely at the position of the wobble alteration (Fig 6B). Thus, the wobble mutation (C to T) created an earlier splice donor site 35 nts upstream of the wild-type donor splice site (Fig 6A and 6B), generating a transcript lacking 35 nts of the coding sequence.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Molecular characterization of the Krang maternal factor.

A. Schematic representation of the molecular lesion and generation of a premature donor splice site 35 nt upstream of the wild type (wt) donor splice site in the krang/kiaa0513 (mutant) transcript. B. Predicted tertiary structure of the full-length zebrafish Krang protein (left) and its truncated version (right). The α-helixes and the connecting loops are colored in pink and green, respectively. N- and C-terminal domains are indicated. The SBF1/2 functional domain of the protein is shown in yellow. The approximate volumetric density map of the protein is shown in translucent gray. C. Krang protein sequence alignment of multiple species represented in E. Top: Schematic representation of the protein sequence alignment. The overall percentage (%) similarity among sequences decreases from top to bottom. The yellow box indicates the SBF1/2 functional domain in Krang. Bottom: Detailed sequence alignment, showing high conservation of this domain in fungi, insects, fish, birds and mammals. Percent similarity is color coded by the scale bar at the right. D. Reduced neighbor-joining phylogenetic tree of Krang homologs found in public databases indicating its evolutionary history from invertebrate to vertebrate species. The scale bar in the middle represents evolutionary distances based on residue substitutions per site. E. In situ hybridization showing the cortically-restricted distribution of krang mRNA in a cryosectioned stage III oocyte (n = 18). Notice that the expression of the transcript (arrowheads) is associated with MPA-stained CGs. FCs, follicle cells. Right panels show selected areas (arrowheads) in separate channels. F. Whole-mount in situ hybridization showing krang transcript localization. Top row: wild-type stage III oocyte (n = 43), where krang mRNA is peripherally distributed in the cell, presumably overlapping with cortical granules (white asterisks in high magnification images of the same oocyte). Bottom row: stage II oocyte (n = 19), where the krang transcript is associated with nascent CGs (bright field and fluorescent high magnification images). Control sense probe did not show transcript signal in stage III oocytes (n = 23, top row, left). Boxes show selected magnified areas. Scale bar = 20 μm (E), 140 μm (F, stage III oocyte), 9 μm (F, top row high magnification images), 55 μm (F, stage II oocyte), 20 μm (F, bottom row high magnification images).

https://doi.org/10.1371/journal.pgen.1011343.g006

The kiaa0513 transcript encodes a 424 amino acid protein with no previously known function, although highly conserved from arthropods to humans (Fig 6C–6E). The 35 nt deletion in the krang^p30ahub cDNA results in a frameshift, which generates a premature stop codon following 24 aberrant amino acids (S3G Fig). As a result, the predicted mutant protein lacks 255 of 424 amino acids, including the highly conserved SET binding factor 1/2 (SBF1/2) or myotubularin-related domain (Fig 6C). SBF1 and SBF2 proteins (Myotubularin-related protein 5 and 13, respectively) have been previously described as homologous to phosphatases but lacking phosphatase activity and instead acting as pseudo-phosphatases regulating the enzymatic activity of other phosphatase myotubularin-related proteins [56–58]. However, the non-catalytic phosphatase domain of these factors was not found in Krang or Kiaa0513 orthologs. Amino acid alignment of zebrafish Krang to other animal orthologs revealed high similarity within the SBF1/2 domain (Fig 6D) and phylogenetic analysis shows the conserved evolutionary relationship of krang homologs in vertebrate and invertebrate organisms (Figs 6E and S4). Little is known about Krang/Kiaa0513 to date; therefore, how its SBF1/2 domain physiologically functions in egg activation will need further investigation.

To validate that krang is encoded by kiaa0513, we generated a second mutant allele, krang^p14del, using a CRISPR/Cas9 approach. The new krang mutant allele is a 14 bp deletion affecting exon 5, which generates a premature stop codon (S5A Fig). Mutant females also produced eggs that fail to elevate the chorion (S5B Fig). To evaluate if krang^p30ahub and krang^p14del alleles do not complement, we generated trans-heterozygous loss-of-function females (krang^p30ahub/krang^p14del), whose eggs and early embryos displayed the krang chorion elevation and cell cleavage defect (S5B Fig). In addition to the tight chorion phenotype, transheterozygous females yielded ventralized embryos, altogether showing that these two alleles do not complement and are mutations in the same gene, krang.

To gain insights into Krang function, we examined its mRNA expression pattern during oogenesis. In previtellogenic stage II oocytes, krang transcript puncta were primarily detected around or in nascent CGs, whereas in vitellogenic stage III oocytes, it is peripherally localized in the vicinity of CGs (Fig 6F and 6G). A similar localization pattern has been described for other mRNAs encoding exocytosis and vesicular release proteins in neurons [59]. Taken together, the abnormalities observed in the mutant egg and embryo, and the spatial localization of its mRNA, suggest that Krang functions in regulating crucial components involved in CG dynamics or integrity, which subsequently act in chorion elevation.

The maternal-effect kazu gene controls cell volume acquisition and blastomere adherence during embryogenesis

In cleavage stage embryos, the initial cytoplasmic volume at the one-cell stage is cleaved at each cell division in half, with no increase in total cytoplasmic volume until much later in development [60,61]. In zebrafish the extent of cytoplasmic segregation from the yolk into the blastodisc also determines the cleavage stage cell sizes. We found that the eggs of the maternal-effect kazu^p26thbd mutant display a small blastodisc, and the early cell divisions are characterized by asymmetric and irregular cleavage, generating smaller and less compacted blastomeres (Fig 7A and 7B). Additionally, the cytoplasm persisted within the yolk and cell detachment was also observed in kazu^p26thbd cleavage stage mutants (Figs 3B and 7A, 5 hpf). Most kazu^p26thbd mutant embryos failed to develop beyond the sphere stage and died during this period. However, those that survive generally show a reduced body axis (S6A Fig). These results are consistent with a defect in fully segregating the cytoplasm from the yolk cell, with cytoplasm persisting within the yolk at the 4-cell stage and beyond.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Phenotypic characterization of the kazu^p26thbd mutant early embryo.

A. Bright field images showing early developmental stages of whole-mount live wild-type (n = 99/99 from 2 females) and kazu^p26thbd (n = 185/265 from 2 females) embryos. The mutant phenotype reveals a smaller blastodisc (bd) with smaller cells during cleavage. These abnormalities in cell size acquisition, together with asynchronous and unequal cell divisions, leads to the formation of a defective blastula with many detached blastomeres. B. Higher magnification images showing a cellularized wild-type blastoderm (n = 125/125). In contrast, kazu^p26thbd embryos exhibit a less compacted blastoderm containing smaller cells (n = 62/91). C. Lateral and animal views of whole-mount wild-type (n = 31/35) and mutant (n = 39/47) early embryo that show the distribution of α-Tubulin at 2 hpf. In wild-type embryos, α-Tubulin is mainly distributed in the blastoderm (bt) and dividing blastomeres. In kazu^p26thbd mutants, α-Tubulin is distributed in different sized cells and throughout the yolk cell (yc, asterisks). D. Animal views of whole-mount wild-type (n = 26) and mutant (n = 8) early embryos that show the localization of β-Catenin at 2 hpf. In wild-type embryos, β-Catenin localized to the blastomere boundaries. In kazu^p26thbd mutants, β-Catenin is aberrantly localized in different cytoplasmic domains. E. Animal views of DAPI, α-tubulin, and centrosome staining showing abnormal blastoderm formation at the sphere stage. Cell boundaries and mitotic figures are not observed in most of the mutant (n = 25/33) compared to wild-type (n = 30/35) embryos examined. Notice the formation of a syncytial nuclei layer in the 5 hpf kazu^p26thbd embryo. Scale bar = 180 μm (A), 110 μm (B), 230 μm (C), 165 μm (D), 100 μm (E).

https://doi.org/10.1371/journal.pgen.1011343.g007

We further examined the impact of defective cytoplasmic segregation and a small blastodisc to early embryo development. Immunostained embryos revealed α-Tubulin in mutant embryos extending from the blastomeres into the yolk cell at the 16- to 32-cell stage, whereas it was restricted to the blastomeres in wild-type (Fig 7C). In large parts of the yolk cell, a honeycomb pattern of α-Tubulin was evident, not observed in wild-type, likely reflecting the cytoplasm remaining in the yolk cell that surrounds the yolk globules (Fig 7C). In addition, cell-cell contact was absent in kazu early embryos, as revealed by delocalized aggregates of membrane β-Catenin (Fig 7D). In the wild-type sphere stage embryo, microtubules (MTs) and centrosomes were mainly organized around the nucleus or in the mitotic spindle apparatus of well-defined cells (Figs 7E and S6B). In contrast, kazu^p26thbd embryos displayed a severely altered blastoderm morphology at the same developmental stage, revealing a syncytial layer-like blastoderm expanded over the yolk cell, including diffuse MT distribution and multiple centrosomes (Figs 7E and S6B).

These results indicate that Kazu is a critical factor in regulating the incorporation of cytoplasm into the developing blastodisc after egg activation and in the formation of adherent blastomeres cells to form a multi-layer blastoderm during early embryogenesis. Thus, the kazu^p26thbd mutant and other cytoplasmic segregation mutants [13] provide excellent genetic entry points to understand the molecular control of cytoplasmic volume acquisition in blastomere size regulation and in their adherence during early vertebrate embryogenesis.

spotty regulates chromosome integrity, MTOC number, and microtubule nucleating activity during the egg-to-embryo transition

Images of live one-cell stage spotty showed that at 40 mpf the blastodisc grew after the phase of massive transport of cytoplasm (Fig 8A). However, staining of the yolk cell with DiOC₆, which labels the YGs due to their high lipid content, revealed multiple, small cytoplasmic domains in cortical regions of the yolk cell between peripheral YGs of the spotty mutant, unlike in the wild-type egg (Fig 8B, asterisks). These results point to a cytoplasmic subcellular organization defect in the spotty mutant zygote.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Chromosome, γ-Tubulin and microtubule organization in the spotty^p08bdth mutant egg and zygote.

A. Ectopic cytoplasmic domain formation in the spotty mutant. Instead of a regular organization of cytoplasm intermingled with YGs in the wild-type yolk cell (n = 74 embryos from 2 females), the spotty mutant (n = 65 embryos from 2 females) shows numerous, peripheral cytoplasmic domains (black arrows). B. Animal viewed confocal images of DiOC₆- stained zygotes reveal that YG arrangement is affected, coinciding with peripheral patches of cytoplasm (white asterisks). Higher magnification DAPI- and γ-Tubulin-stained images show perturbed nucleus formation and γ-Tubulin assembly in spotty (n = 37 from 3 females) compared to wild-type (n = 33 from 3 females) zygotes at 25 mpf. Boxes show magnified areas below. C. DAPI-stained wild-type (n = 20 from 2 females) and mutant (n = 48 from 2 females) unactivated eggs showing chromosome organization during metaphase II of meiosis. D. α-Tubulin- and DiOC₆-stained confocal z-projections showing MT and YG distribution at the cortex of wild-type (n = 34) and spotty (n = 41) eggs. Higher magnification images reveal the magnitude of such MT structures and YG organization in the mutant egg. E. α- and γ-Tubulin co-staining indicates that MTs emanate from numerous MTOCs across the mutant egg (n = 14). Image 4 was selected from a z-stack containing 31 x 1 μm-thick optical sections across the mutant egg. Boxes show magnified (B, D and E) areas. bd, blastodisc; yc, yolk cell; yg, yolk globules; cp, cytoplasmic pocket; A, animal pole; V, vegetal pole; mpf, minutes post fertilization. Scale bar: 190 μm (A), 200 μm (B, top row), 50 μm (B, bottom row), 5 μm (C), 180 μm (D, low magnification), 38 μm (D, high magnification), 180 μm (E, top row), 60 μm (E, bottom row).

https://doi.org/10.1371/journal.pgen.1011343.g008

During the transition from meiosis to mitosis, the centrosome and astral MT activity is coupled to zygotic nucleus formation and cell-cycle progression [35,39]. The spotty mutant shows abnormally cleaved cells (Fig 3A), suggesting altered mitosis. Given that this process is accompanied by γ-tubulin assembly, and the nucleation and organization of MTs, we performed immunofluorescent staining of γ-tubulin and MTs immediately after fertilization. In the wild-type zygote, γ-tubulin is perinuclearly dispersed (Fig 8B). However, the pattern of γ-tubulin distribution was dramatically altered in the mutant zygote with numerous stained aggregates localized around an unusual nuclear architecture in the blastodisc (Fig 8B). Such nuclear organization suggested defective pronuclei behavior and nucleus formation. Fixed DAPI stained wild-type unactivated eggs revealed the establishment of a metaphase plate during meiosis II (Fig 8C). However, several small DAPI spots, possibly dispersed individual or fragmented chromosomes were detected in spotty mutant eggs. This suggests that the maternal DNA integrity is compromised by a likely defective meiotic cell cycle in the oocyte. These results reveal that Spotty is essential to: a) regulate or maintain meiosis II metaphase plate integrity and thus chromosome segregation and mononuclear assembly after fertilization, and b) regulate spatially restricted γ-tubulin localization during the first mitotic cell cycle.

In examining MTs in the egg, we found unusual MT arrays in the spotty mutant. Multiple unipolar aster-like MT structures were visualized at the cortex at 5 mpf (Fig 8D). Higher magnification optical sections revealed apparent astral-like MT organization (Fig 8D). The localization pattern of γ-tubulin confirmed that these MT-generating structures in the spotty mutant were indeed microtubule organizer centers (MTOCs) (Fig 8E). The striking γ- and α-tubulin organization phenotypes in this mutant suggest that the factor encoded by the spotty gene regulates MTOC organization, MT architecture and/or dynamics in the egg and zygote.

Altogether, these findings reveal severe defects in MTOC biology in the spotty mutant; thus, spotty may represent the first described mutation in a vertebrate gene that regulates MTOC number. The identification of the spotty gene offers the opportunity to simultaneously understand the mechanism of how MTOCs are regulated in the egg and spatially organized through maternal control during the egg-to-embryo transition.

Molecular genetics of the oocyte-to-embryo transition

The mutants reported here disrupt genes that are required during oogenesis or activation of the egg. They genetically uncouple nuclear and cytoplasmic aspects of oocyte maturation, and spatially distributed events of egg activation. In addition, they demonstrate that the process of preparing an egg for embryonic development is not a simple linear series of events. Rather, multiple parallel, coordinated events that do not all depend on successful completion of earlier steps in the process. For example, most of the opaque egg mutants can complete nuclear maturation (meiosis) even though cytoplasmic maturation is compromised during oogenesis. The maternal-effect krang, kazu, p28tabj and spotty mutants show overtly normal development of the oocyte, including maturation of the cytoplasm and nucleus. However, these mutants do not successfully undergo egg activation. Like opaque egg mutants, the distinct defects observed in these mutants are consistent with oocyte maturation regulated independently of other aspects of egg activation.

The zebrafish has become a useful model system to identify genetic factors controlling human female reproduction and associated diseases (reviewed in [62]). Taking advantage of future genotype-phenotype association studies of oocyte and egg biology in this vertebrate model will enhance our knowledge of the maternal gene bio-network regulating the oocyte-to-embryo transition. Importantly, the accessibility to genetic models will illuminate such maternal factors’ biological function in vertebrate reproduction and development, as well as our understanding of the molecular causes of infertility, and eventually improve clinical applications. However, further studies are needed to fully elucidate the specific functions of these maternal genes and their participation in oocyte maturation, egg activation, and subsequent embryonic development.

Genetic modulation of YG formation and maturation by egg-laying organisms

The link between YG size acquisition and cleavage of yolk proteins during embryogenesis has been elegantly investigated in invertebrate animals, including insects, starfish and sea urchin [63–67]. However, how YGs form and how yolk protein processing is regulated during oogenesis is still unknown in oviparous vertebrates. Here, we identified four maternal factors controlling YG size and yolk protein cleavage during zebrafish oocyte maturation.

The small YG phenotype in the ovy/ap5m1, p33bjta, and blac mutants suggests that their gene products regulate fundamental events in vesicle trafficking and formation, respectively, during and/or after the yolk protein precursor internalization process of vitellogenesis. These studies may provide entry points for future investigations of the molecular control of YG biology [68,69]. Based on the lysosomal nature of YGs and their proteolytic hydrolase content, it is possible that discovering the molecular regulators of YG formation and function may contribute to our understanding of lysosomal related pathologies in humans (reviewed in [10–12]).

Yolk protein cleavage constitutes a critical event to egg-laying species, generating maternal nutrients essential for embryonic growth: the proteolytically cleaved products of Vitellogenin (Vtg), Lipovitellin (Lv) and Phosvitin (Pv) [70]. In addition, antimicrobial activity of Pv and its derived cleavage product has been demonstrated in vitro and in vivo, including evidence in the immunity of zebrafish embryos [71,72], which opens new avenues for the study of the immunological role of yolk protein cleavage in oviparous animals during oogenesis.

The opaque egg mutant class now comprising 7 mutant genes (here and [13]), all fail in blastodisc formation. Hence, it is possible that the blastodisc defect is a consequence of the failed yolk maturation. However, we cannot exclude that it reflects a secondary function of this class of genes in blastodisc formation. In the former case, two mechanisms could explain the defect in blastodisc formation: i) yolk maturation is a prerequisite for the reorganization of cytoplasm to the animal pole or ii) the persistence of unmodified or immature yolk interferes with blastodisc formation. Despite an initial successful transition from meiosis to mitosis, the progeny of these mutants do not complete development underscoring the requirement for cytoplasmic maturation to produce viable progeny.

Although the molecular identification of regulators of yolk biogenesis and proteolytic processing in multicellular organisms has begun to emerge, much of their functional relevance to embryogenesis remains a puzzle ([13,73–75]; this work). Dissecting the events mediating YG formation and function may help understand: i) conserved or lost functions during oocyte maturation and how different reproductive strategies and nutritional reserves have evolved among vertebrates [7,76], and ii) the temporal and spatial coordination of oocyte maturation and its link to external sources of oocyte-specific signaling molecules including sexual hormones and growth factors in amniotic and non-amniotic vertebrates.

Advances in female gametogenesis, together with the need to decipher molecular phenotypes during this process, may provide new insights into the mechanisms underlying yolk biology prior to oocyte maturation and the contribution to the study of oviparity-to-viviparity transition in metazoans. In this scenario, knowledge of the underlying molecular mechanisms modulating yolk composition and function in egg-laying organisms might also extend previous findings that link the loss of egg yolk genes and the emergence of new genes in the origin of specific traits such as lactation and placentation [7]. Thus, deciphering the molecular circuit controlling yolk biology and function might contribute to understanding the complex cellular energy, nourishment and immunological resource strategies developed in higher organisms during evolution.

The role of Ap5m1 in reproductive function

The reduced YG size and failure of yolk protein cleavage in the ovy/ap5m1 mutant suggest that Ap5m1 is involved in YG formation and maturation. Ap5m1 might act in the formation of large YGs and in regulating in some manner the activity of YG proteases that cleave Vtg during zebrafish oocyte maturation [68,69]. Our findings and the Ap5m1 molecular pathway provide evidence of a potential role for Ap5m1 in intracellular trafficking and the sorting processes required for YG biogenesis in oviparous organisms.

At the cellular level, the Ap5 complex acts in intracellular trafficking [44–47]. It consists of a heterotetrameric complex including the ζ and β5 (large), μ5 (medium or m5), σ5 (small) subunits [44]. Our molecular modeling of the zebrafish Ap5m1 factor and its homology with the human AP4 complex m4 subunit indicate that it may function in endosomal trafficking during zebrafish oogenesis. The ovy^p35aluc T27K missense mutation is expected to disrupt the domain of Ap5m1 that interacts with the Ap5 β subunit, as observed in in vitro studies where that same domain has been mutated [46].

The AP5 complex specifically interacts with proteins SPG11 and SPG15/Spastizin in mammalian cell culture. Mutations in these proteins are implicated in Hereditary Spastic Paraplegia (HSP) disease [46,48]. Intriguingly, siRNA-mediated knockdown of Spg11 and Spg15/Spastizin results in the aggregation of early endosomes, consistent with AP5 functioning with these factors in trafficking in mammalian cells [45]. Each of the five AP complexes localizes to and functions in distinct subcellular endosomal trafficking compartments, with the Ap5 complex shown in cell culture to act at later stages of the endocytic pathway by sorting endosomes back to the Golgi [47]. Our ovy mutant represents the first multicellular model for ap5m1 gene function and concomitantly, the Ap5 complex, providing new avenues to study the oogenesis endolysosomal-YG route in a possible vertebrate model of HSP.

Interestingly, a role for Spg15/Spastizin in reproduction has been reported in zebrafish [73]. The maternally deposited Spastizin factor regulates cortical granule secretory granule maturation and like the ovy/ap5m1 mutant the spastizin mutant also displays a similar opaque egg phenotype and a defect in yolk protein processing [13,73]. Therefore, it will be interesting in future studies to investigate how Ap5m1 and Spastizin association functions in YG biogenesis and major yolk protein processing.

In this work, we identified a new regulator, the ap5m1 gene, as a valuable genetic tool for the study of lysosome-related organelles and its function in the endolysosomal pathway in development. Indeed, our results indicate a novel role for Ap5 complex function in female reproduction. Sorting of Cathepsins and ATPases into lysosomes is essential for yolk protein cleavage [68,69]. In fact, the role of maternal Arl8b protein in Cathepsin-L sorting and lysosomal-mediated degradation of maternal factors in visceral yolk sac has been described in mammalian embryos [77].

Our results highlight the importance of endolysosomal trafficking and protein cleavage during vertebrate embryogenesis. We hypothesize that YG biogenesis relies on Ap5 complex function and its interaction with Spastizin. Epistasis analysis and biochemical approaches involving the sorting of proteases and electrogenic pumps into lysosomes might help elucidate Ap5m1-mediated function during oogenesis.

Spatial and temporal control of maternal gene function in early embryogenesis

The kazu, p28tabj, and spotty mutant phenotypes show that while chorion elevation and blastodisc formation are both triggered by egg activation, they can be genetically separated and are linked to the formation of compartmentalized cytoplasmic domains. This supports a model in which regionalized and distinct or divergent pathways in the egg regulate these aspects of egg activation. While separable genetically, other maternal-effect mutants show defects encompassing all or many aspects of egg activation, reflecting components acting in early steps of egg activation [20,78,79]. For example, brom bones/ptbp1a is defective in inositol 1,4,5-trisphosphate (IP₃) production and the Ca²⁺ wave that triggers egg activation, thus affecting all aspects of egg activation [20]. The maternal-zygotic dachsous 1b mutant exhibits slower activation of the egg, chorion elevation, cell cleavage, and early development [79]. The nature by which this atypical cadherin regulates egg activation processes and early embryonic development broadly is unclear. The maternal-effect aura/mid1ip1l mutant alters only a subset of egg activation processes with its primary function in regulating the cytoskeleton [78]. Hence, downstream of Ptbp1a and IP₃, Dachsous and Aura, likely act in divergent egg activation pathways. Our findings of mutants affecting either CG biology or cytoplasmic segregation further show that egg activation-associated events are regulated independently.

Krang: An uncharacterized and conserved regulator of cortical granule biology

Following exocytosis, the contents of CGs modify the vitelline envelope causing it to expand and enlarging the perivitelline space, which is a well-characterized phenotypic trait to analyze high-quality eggs in assisted reproductive technology (ART) [27,28,80–82]. In addition, mutants with compromised CG exocytosis or vitelline envelope formation generate eggs with a smaller perivitelline space compared to wild type [20,51,73,78] or delayed enlargement of the space [79]. Therefore, insight into the genetic program governing CG biology and vitelline envelope formation may provide biological markers for evaluating oocyte and egg quality in human ART. Recent findings have revealed new insight into the regulation of CG maturation and exocytosis during the vertebrate egg-to-embryo transition [20,28,73,81,83,84], this work). However, a more comprehensive understanding of the molecular basis of how key maternal factors regulate CG biogenesis, dynamics and content release remains to be revealed.

The loss of Krang function impairs egg activation, specifically in chorion elevation. We do not know the nature of Krang’s function in modulating secretory granule biology during female reproduction. Clues for its function are revealed by krang transcript localization, which is closely associated with CGs, strongly supporting a spatially restricted function for this factor to CGs. These findings might shed light on a maternal post-transcriptional regulation machinery that localizes translation, possibly to facilitate protein localization and function to, or within, CGs to regulate their translocation, content or function during egg activation. To understand Krang’s function in CG biology will require further investigation.

The human ortholog of Krang, KIAA0513 (71% identity), has been predicted to function during neural development and its transcript to be dysregulated in individuals with schizophrenia and Alzheimer’s Disease [85,86]. The krang/kiaa0513 transcript is also zygotically expressed in the zebrafish hatching gland and nervous system (Thisse and Thisse, 2004). Indeed, the identification of KIAA0513 splicing variants in the brain of related vertebrates, including the zebrafish, has also brought to light a possible exclusive and tissue-specific function of Krang in neural development [85,86]. If Krang acts in an oocyte-specific manner in vertebrates remains to be explored.

Our data strongly support a novel role of Krang during the zebrafish egg-to-embryo transition, presumably modulating its function by specific subcellular localization. Importantly, as this factor constitutes a biological marker of schizophrenia and AD in humans, insights into Krang function might contribute to advancing our current understanding for the still obscure cellular etiology of these neuropathologies and their treatments.

Maternal determinants of cytoplasmic organization during early embryogenesis

Following fertilization, broadly in animals, early embryonic cells divide and maintain the total cytoplasmic volume inherited prior to the first cell cycle. These early cells are also unusually large, and subcellular structures, such as the nucleus and mitotic spindle, scale to a large extent to the larger cell size [87–91]. Despite these findings, how the total volume of cytoplasm is regulated and distributed, remains poorly understood. We identified the kazu^p26thbd mutant, which displays defects in blastodisc volume acquisition and small cells, thus representing a regulator of cytoplasmic domain formation during early embryogenesis.

After egg activation, a secondary and slower calcium transient observed in the central egg region coincides with and signals the spatial reorganization and animal-directed flow of cytoplasm [21,29]. It is possible that Kazu acts in blastodisc size acquisition by regulating the rate of the animal-ward cytoplasmic flow from the yolk cell, possibly in response to a calcium signal to appropriately allocate the subcellular yolk cell cytoplasmic domains to the zygote (reviewed in [50]). Actin, MTs, molecular motors and their regulators have been implicated in cytoplasmic segregation and RNA translocation through inhibitor and genetic studies [19,29,32,33,79,92,93]. Therefore, it is possible that kazu^p26thbd and other cytoplasmic segregation mutants disrupt upstream regulators or downstream targets of Actin/Myosin contractility and/or MT dynamics during egg activation [19,29,32,33,93–96]. Ultimately, these cytoplasmic transportation defects may affect the arrival of maternal determinants to the developing blastodisc [29,94,95]. Future transcript localization analysis in kazu^p26thbd, and other cytoplasmic segregation mutants, would test this hypothesis.

Cell survival critically depends on cell volume regulatory mechanisms. Impaired steady-state cell volume regulation underlies physiological and pathological abnormalities such as cancer and hypertrophy (reviewed in [97–99]). In zebrafish, mutant genes that result in small blastodisc and blastomere phenotypes are candidates for acting in cytoplasmic domain formation and cell volume (reviewed in [50]). Thus, the kazu^p26thbd mutant provides the opportunity to study the maternal control of spatial distribution of embryonic cytoplasm and cell volume acquisition, which may also have relevance in disease.

Maternally-controlled MTOC assembly and microtubule nucleating activity during early embryogenesis

In most animal cells, the centrosome orchestrates the nucleation of both cytoplasmic MTs and the mitotic spindle during interphase and mitosis, respectively [100–103]. During meiosis in most animals, however, an acentrosomal MTOC (aMTOC) regulates the meiotic divisions, due to elimination of the maternal centrosome during oogenesis. Despite their importance as MTOCs, the genetic regulation of many aspects of MTOC and aMTOC assembly, structure, composition, and function remain elusive. Intriguingly, multiple MTOCs/centrosomes are a hallmark in a growing list of human cancers, emphasizing the importance of regulating MTOC properties and number in health and disease [104–106]. Our understanding of how MTOCs acquire the ability to control their copy number and orchestrate the nucleation of cytoplasmic MTs, impacting on genome stability and embryonic progression, is poorly understood.

During meiosis and early embryogenesis, little is known about the genetic program controlling how the oocyte, egg, and early embryo: a) transitions from centriolar to an acentriolar MTOC, and then back to a centriolar MTOC in the embryo, and b) how they keep MTOC de novo biogenesis repressed. This may be a particularly challenging task in the egg, which is chock full of the factors needed to generate the multitude of centrosomes and MTs that function during early cleavage divisions prior to zygotic genome activation. There are few known regulators of centrosome elimination and copy number control. In Drosophila, centrosome elimination is determined by Polo kinase activity and the reduction of pericentriolar material (PCM) components, which causes centriole loss before meiosis initiates [107]. We do not know if the ectopic MTOCs in spotty represent centrosomal or acentrosomal MTOCs.

The excessive number of aster-like MT structures with γ-Tubulin aggregates in the spotty mutant egg suggests that Spotty functions in repressing MTOC number and MT organization during the egg-to-embryo transition. Additionally, the temporal, geometric and spatial organization of MTs and γ-Tubulin in the spotty mutant reveal two important features of its genetic product and the mechanisms behind its activity. First, in addition to the recruitment of γ-Tubulin and other MTOC components, the gene product may be necessary for assembling and positioning the functional MTOCs after fertilization. Second, it may be critical for MT dynamics by negatively controlling the nucleating activity of the MTOC. Determining the molecular identity of the spotty gene is expected to identify a new repressor of supernumerary MTOCs and MT assembly during the oocyte-to-embryo transition. However, the origin of the defect in spotty mutant will require further investigation during oogenesis. Here, we present a genetic and functional model for the in vivo study of MTOC copy number regulation and MT architecture control during early vertebrate embryogenesis.

Ethics statement

All animal protocols were approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC) and University of Concepcion Institutional Animal Care (CEBB 818–2020).

ENU mutagenesis and fish husbandry

TÜ and AB strain males were mutagenized with 3.3 mM ENU as described in [108]. The mutagenesis efficiency was determined by crossing mutagenized males to females that were triple mutant for three mutations that disrupt pigment formation, albino (0.04%), golden (0.12%) and sparse (0.12%) as in [108]. Through the natural mating strategy, we generated 503 F3-families, each bearing two independently mutagenized genomes. The 716 genomes screened for maternal-effect mutations was calculated according to the formula for pooled F3 embryos as previously described [13].

Female maternal-effect mutants were identified by crossing F3 females to sibling/cousin or wild-type males and examining their F4 progeny. Phenotyping analysis was performed immediately after egg collection for fertilization and egg activation morphological defects and later at 1-day post fertilization for morphological defects and viability. Females that produced greater than ~40% phenotypic progeny were retested in crosses with wild-type males to confirm that the phenotype was a strict maternal-effect. No mutants exhibited maternal-zygotic genetics. For further phenotypic characterization and mapping, female maternal-effect mutants that reproducibly produced phenotypic offspring when crossed with a wild-type male were recovered typically from 10–15 intercrosses of the F2 generation of each family. Adult wild-type and mutant fish were maintained for a 14 hr light/10 hr dark photoperiod at 28°C.

Genome mapping, and sequencing for ovy/ap5m1 and krang/kiaa0513 genes identification

The oocyte-to-egg transition mutations were mapped to a chromosomal locus as described [13,109,110]. Positional mapping and cloning were carried out through bulk segregant analysis using SSLP markers [111]. The ovy^p35aluc and poac^p26ahubb mutations were both mapped to Chr 17 but to distinct genetic intervals between 51 and 61 cM, and 58 and 79 cM, respectively (Table 1). By testing 26 females for recombination, the ovy^p35aluc interval was defined to an interval of 0.7 cM between markers z7170 (2/26 recombinants) and z8980 (7/26 recombinants). Linkage analysis of p35aluc and ovy^p37ad mutations located them to the same interval on chromosome (Chr) 17. The ap5m1 gene exon 1-intron 1 and intron 1-exon 2 junctions were amplified by PCR from ovy^p37ad gDNA using primers targeting the 5’ and 3’ ends of intron 1 (S1 Table). The ap5m1 missense mutation was identified by sequencing the first overlapping gene amplicon from the p35aluc mutant background using specific primers (S1 Table). The blac^p25bdth mutation was mapped first to Chr 5 and then the blac^p26ahubg allele was mapped to the same locus on Chr 5 between 47 and 64 cM (Table 1).

The krang^p30ahub, spotty^p08bdth, and kazu^p26thbd mutations were mapped to chromosome (Chr) 18 (between 33 and 47 cM), Chr 2 (between 44 and 62 cM), and Chr 8 (between 95 and 104 cM), respectively (Table 1). For the krang^p30ahub mutation, the CR925798-1 marker was generated for fine mapping (see S1 Table). Chromosomal locations of the SSLP markers were determined using the zebrafish reference genome Zv9.

For next generation sequencing, enriched and captured sequences (~800 kb) were extracted from gDNA of wild-type and krang^p30ahub females as previously described [54]. The libraries were then sequenced using an Illumina Genome Analyzer II.

Generation of krang^p14del mutant

Genome editing was carried out by targeting krang/kiaa0513 in exon 5. A CRISPR target site 5′-gattgacagATGTCACCAGGTGG-3′ in intron 4 and the fifth coding exon of krang/kiaa0513 was selected. The sgRNA construct was purchased from the University of Utah Mutation Generation and Detection Core. For injection, the synthetized sgRNA was incubated with Cas9 protein (PNA Bio, Cat# CP02-250) to form an RNP complex. 1–3 nl of 2.5 μM RNP complex was injected into 1-cell stage embryos. To evaluate cutting efficiency, we performed HRM (High Resolution Melt assay) analysis on F0 single injected embryos using MeltDoctor HRM Master Mix (Applied Bio-systems). Optimized flanking primers for HRM are indicated in S1 Table. Remaining injected embryos were raised to adulthood to produce F1 families. To identify the krang/kiaa0513 molecular lesion, we sequenced the amplified PCR product from transheterozygous fish genomic DNA using primers flanking the CRISPR/Cas9 target site (Exon 4-Intron 5, S1 Table).

Gene complementation and PCR-based genotyping

Complementation tests were performed by crossing two poac^p26ahubb heterozygous females to two homozygous p35aluc males and two heterozygous p35aluc females to two ovy^p37ad homozygous males and raising the progeny to adulthood. Transheterozygous p35aluc/ovy^p37ad females produced 100% mutant opaque eggs, whereas transheterozygous p35aluc/poac^p26ahubb females produced all wild-type embryos (36/36 and 38/38 females, respectively). Genetic complementation analysis between the two blac mutations was conducted crossing three p26ahubg heterozygous females to either p25bdth homozygous males (n = 2) or a heterozygous male (n = 1). Fish were raised to adults and females tested for the lysed egg phenotype. These mutations failed to complement each other, generating transheterozygous mutant females from all 3 crosses in the expected ratios (10/20, 15/31, and 2/10 females for the three respective crosses). Also, a krang^p14del heterozygous female was crossed to a krang^p30ahub homozygous male to generate transheterozygous adult fish. Five transheterozygous and sibling krang^p30ahub heterozygous females each were tested for phenotype. All transheterozygote females produced mutant embryos with small chorions, whereas sibling heterozygotes produced wild-type embryos. All transheterozygous females were tested at least twice and up to 5 times, reproducibly showing the mutant phenotype.

poac^p26ahubb, blac^p25bdth, blac^p26ahubg, and kazu^p26thbd alleles were genotyped using flanking SSLPs designed for positional cloning. ovy^p35aluc, ovy^p37ad and krang^p30ahub alleles were genotyped using KBiosciences Competitive Allele-Specific PCR genotyping system (KASP, KBiosciences) (see S1 Table).

Whole-mount fluorescent in situ hybridization and CG staining

Fluorescent in situ hybridization of cycB1 [112], dazl [113] and krang (this work) was performed in whole-mounted oocytes and cryosectioned ovaries (30 μm thick). Oocytes were isolated and fixed according to [114]. For staining of frozen tissue, ovaries were dissected, fixed, and stored in 30% sucrose as described in [115]. Isolated and cryosectioned oocytes were pre-hybridized, hybridized, washed and blocked as described in [114]. Samples were then incubated in the Anti-POD antibody (1:500; Roche) and washed 4 times for 25 min in 1X PBS. For fluorescent mRNA detection, samples were developed in TMR-Tyramide Signal Amplification (TSA) system in staining buffer (1:50, PerkinElmer), 10 min at room temperature (RT) in the dark. After 3 times washing for 5 min in 1X PBS, samples were incubated in Maclura pomifera Agglutinin (MPA) diluted to 50 μg/ml as described in [116,117] for 20 min in the dark at RT. Then washed in 1X PBS 4 times for 5 min each and mounted in Vectashield mounting medium with DAPI. The krang digoxigenin-labeled probe was synthesized in vitro using DIG RNA labeling kit (Roche) with T3 (sense) or T7 (antisense) polymerase from kiaa0513 full length DNA sequence cloned into the pBluescript II SK+ vector.

For CG staining, unactivated and activated eggs were collected from gravid females by squeezing and by natural crosses, respectively. Unactivated eggs were kept as described [20], and activated in a hypotonic solution of 1X E3 buffer. Eggs were then acid-fixed at 0, 5, 20 and 30 mpa intervals in 5% formaldehyde and 5–8% glacial acetic acid at room temperature, manually dechorionated and placed in gently agitation for 2 hrs, washed in 1X Phosphate-Buffered Saline (PBS) 4 times for 10 min each [118]. Samples were labeled with fluorescent conjugated MPA for 30 min as described above. Then washed in 1X PBS 4 times for 10 min each and either placed in small petri dishes for in toto observation or mounted on glass slides.

F-actin and meiotic chromosomes labeling

For actin labeling, zygotes were fixed, washed and blocked as in [117]. Then, TRITC-Phalloidin (Sigma) was diluted to 0.5 μg/ml in PBT. Samples were mounted in Vectashield mounting medium with DAPI on glass slides.

Histology

Ovaries dissected from euthanized females were fixed overnight at 4°C in 4% paraformaldehyde. The next day the fixed tissue was washed in PBS and dehydrated in MeOH. Ovaries were embedded in JB-4 Plus plastic resin (Polysciences) and 5–10 μm sections were cut using a Leica RM 2155 microtome. Sectioned ovaries were stained with Hematoxylin (Sigma-Aldrich), washed in distilled water, stained in Eosin Y (Polysciences), washed in distilled water, and cleared with 50% EtOH. Stained sections were coated with Permount (Fisher), and coverslipped.

SDS-PAGE analysis of major yolk proteins

Ovaries were dissected and sorted oocytes were obtained as described in [13]. Sorted oocytes and eggs were homogenized in sodium dodecyl sulfate (SDS) buffer with bromophenol blue and glycerol. Samples were heated at 100°C for 5 minutes, vortexed, centrifuged, and proteins were resolved on 8% or 12% acrylamide gels and visualized with Coomassie blue.

YG diameter, ECI, and CG number and dimensions measurements

For YG size quantification, the mean diameters of YGs from wild-type and mutant stage III oocytes were measured on micrographs taken in whole sectioned and stained ovaries (n = 2 wild-type and 2 mutant ovaries). 3 different cytoplasmic regions (DCRs) of 96 x 96 μm (height x width) within the same wild-type (n = 12 total DCRs from 4 oocytes) and mutant (n = 6 total DCRs per allele from 2 mutant oocytes) oocyte were determined to calculate the individual YG diameters.

To quantitatively examine chorion elevation, the surface area of the egg or chorion was measured. Then, the ratio of the chorion relative to the wild-type 30 mpa egg, which was defined as the egg-to-chorion index (ECI), was determined. The ECI for wild-type eggs was defined as the numerical value for normal chorion elevation (Figs 2B and 5A).

Wild-type and p33bjta, poac^p26ahubb, ovy^p35aluc and krang^p30ahub CGs were counted in a determined lateral region (DLR) of 60 x 212 x 212 μm (depth x height x width) in egg Z-stacks at 0, 20 and 30 mpa from 2 wild-type and 2 mutant females. To automate the assessment of CG number and dimensions, Z-stacks underwent denoising using the PURE-LET algorithm (https://doi.org/10.1016/j.sigpro.2009.07.009) with the maximum cycle count. Subsequently, background and shading corrections were executed across the stacks using the BaSiC tool [119]. Each Z-stack was transformed into a 2D image via a maximum projection in Fiji [120]. Individual CGs were then pinpointed using the deep learning-driven StarDist method (https://doi.org/10.1007/978-3-030-00934-2_30, https://doi.org/10.1109/WACV45572.2020.9093435), specifically employing the "Versatile (fluorescent nuclei)" built-in model within Fiji. Any segmentation errors in the images were rectified manually.

Prism 7 (GraphPad) software was used to make graphs and for statistical analysis of the data. Samples sizes and statistical values are shown in corresponding figure images or legends. Data are represented as mean ± SD or SEM.

Cytoplasmic domains visualization

Organization and distribution of cytoplasmic domains were examined under a Leica MZ 12.5 stereomicroscope after acid-fixation of wild-type, krang, kazu, p28tbja, and spotty zygotes and early embryos as described [29,118]. Central and cortical cytoplasmic domains were transformed from dark into whitish cytoplasm, and from transparent into dark yolk by inverting the contrast using the Adobe Photoshop CC 2017 software. Zygotes and early embryos were incubated in DiOC₆ (5 μg/ml), which following acid-fixation marks central and cortical cytoplasmic domain organization. DiOC₆-stained images were obtained using a Zeiss LSM 880 confocal microscope and the Plan-Apochromat 10×/0.45.

Whole-mount immuno- and actin staining

For microtubule and actin labeling, eggs, zygotes, and early embryos were fixed, washed and blocked as in [121] and [117], respectively. For staining of cryosectioned tissue, ovaries were dissected, fixed, and stored in 30% sucrose as described [115]. Primary antibodies: mouse anti-α-tubulin (1:500, Sigma) and rabbit anti-γ-tubulin (1:2000, Sigma). Secondary antibodies: anti-mouse IgG1, or anti-rabbit IgG, Alexa 488, Alexa 594, Alexa 633 (All 1:1000, Molecular Probes). When needed, samples were counterstained with DiOC₆ (5 μg/ml), then washed in 1X PBS 4 times for 10 min each and mounted in Vectashield mounting medium with or without DAPI on glass slides.

Pronase treatment

Fertilized wild-type and krang^30ahub eggs from 2 females each were treated with 0.25 mg/ml pronase solution in 1X E3 buffer for 10 minutes at 28°C. Control wild-type and mutant eggs were incubated in 1X E3 pronase-free buffer. To finish dechorionation, embryos were washed with 1X E3 buffer three times by gentle trituration using a Pasteur pipette, and examined at gastrula stage under a stereomicroscope. Cleaving rate was calculated using the Prism 7 (GraphPad) software.

Microscopy, imaging and image processing

Live specimen images were captured using a Leica MZ 12–5 and Zeiss Stemi 305 stereomicroscope equipped with a Leica IC80 HD and Zeiss Axiocam 208 HD camera, respectively, and control software on an Apple Macintosh Computer. Histological section images were obtained using a Zeiss AXIOSKOP microscope and ProgRes® Mac CapturePro 2.6 (Jenoptik) software. Fluorescent labeled z-stacks were obtained using a Zeiss LSM 780 or 880 confocal microscope and the Plan-Apochromat 10×/0.45, the Plan-Apochromat 20×/0.8, or the Plan-Apochromat 40×/1.2 immersion water objectives.

For transmission electron microscopy, whole wild-type and krang^p30ahub ovaries were dissected and fixed in 2.5% glutaraldehyde, 2.0% paraformaldehyde, and 0.1 M sodium cacodylate overnight at 4°C. Washes, post-fixation, dehydration, embedding, thin sections, and staining were performed by using standard methods at the UPenn EM core facility. Stained sections were examined with a JEOL JEM 1010 electron microscope.

Images were processed using Fiji and Adobe Photoshop CC 2017 software.

Data values

All individual data points for graphs in all figures or of data presented are included in spreadsheets in S1 Data Values.

Protein sequence analysis and molecular modeling of the zebrafish Ap5m1 and Krang proteins

The similarity of the amino acids in the sequence alignments were scored using the BLOSUM substitution matrix with the software Multiseq [122].

The protein sequences of zebrafish Ap5m1 (Accession XP_005158951.1) and Krang/Kiaa0513 (Accession AAH74034) were searched in public non-redundant databases using iterated PSI-BLAST v.2.2.32+ (3^rd iteration), entries with ‘hypothetical’, ‘putative’, ‘modeled’ and ‘environmental samples’ were filtered out using entrez tools [123]. The resulting 270 sequences were filtered with a cut-off minimum of 20% identity and E-score of 1E-5, including full and partial alignments. Most of these sequences consisted of uncharacterized proteins for both Ap5m1 and Krang, respectively. The distance matrix was calculated using MUSCLE v.3.8 [123]. Neighbor joining protein sequence-based phylogenetic tree of Krang/Kiaa0513 homologs was handled using FigTree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/). The tree was reduced from its original size by allowing a maximal sequence difference (Max Seq Difference) in fraction of mismatched residues of 0.85, resulting in 179 sequences. Bootstrap values from 500 iterations are shown for each node in the resulting cladogram (S4 Fig).

The Ap5m1 and Krang proteins secondary structure, solvent accessibility, flexibility, among other properties, were determined online using PredictProtein Server [124]. The tertiary structure of Ap5m1 and Krang were modeled online using I-TASSER Protein Structure and Function Prediction Server v.4.3 [125]. The Ap5m1 N- and C-terminal domains were defined based on the tridimensional structure of the best model. For Ap5m1 C-terminal domain modeling and structure alignment, the top ranked model based on the crystal structure of AP4m1 [49]; Protein Data Bank accession no. 3l81.PDB) was selected. The overall Root Mean Square Deviation (RMSD) of the protein structure alignment between zAp5m1 and hAp4m1 C-terminal domains was 2.45 Å, with an alignment length of 227 amino acids used in the superposition. The structural alignment was made using the TM-align server [126]. To generate 3D representations of the Ap5m1 (zebrafish), AP4m1 (human) and Krang (zebrafish) proteins, we utilized Visual Molecular Dynamics (VMD) 1.9.3 software.