Slow calcium waves mediate furrow microtubule reorganization and germ plasm compaction in the early zebrafish embryo

In the zebrafish Danio rerio, as in many animal species, primordial germ cell specification is mediated by the maternal inheritance of the germ plasm, a specialized cytoplasm containing specific RNAs and proteins (Wylie, 1999; Strome and Lehmann, 2007; Eno and Pelegri, 2013). Zebrafish germ plasm ribonucleoparticles (RNPs) are inherited as spherical units, typically of less than 1 μm in diameter, that are initially dispersed through the blastodisc. During the first several embryonic divisions, these units multimerize, gathering into clusters while appearing to retain their initial spherical structure (Theusch et al., 2006; Eno and Pelegri, 2013; Nair et al., 2013; Eno and Pelegri, 2018).

During the early embryonic cell divisions, RNP aggregates move to the forming furrows to generate typically four germ plasm masses (Yoon et al., 1997; Pelegri et al., 1999; Knaut et al., 2000). These aggregates subsequently become incorporated into primordial germ cells (Olsen et al., 1997; Yoon et al., 1997; Braat et al., 1999; Knaut et al., 2000), where they are necessary and sufficient for germ cell specification (Hashimoto et al., 2004; Bontems et al., 2009; Miranda-Rodríguez et al., 2017).

The assembly of germ plasm masses at the furrows involves sequential steps. First, RNP multimers assemble even prior to furrow initiation (pre-aggregation). Second, RNP pre-aggregates gather locally towards the forming furrows (furrow recruitment). Both of these processes appear to be mediated by the outward growth of astral microtubules, the tips of which are associated with germ plasm RNPs (Theusch et al., 2006; Eno and Pelegri, 2013; Nair et al., 2013). Germ plasm recruitment during furrow formation results in the accumulation of RNP aggregates collectively forming an elongated, rod-like structure along most of the length of the furrow. At the same time, microtubules form the furrow microtubule array (FMA) (Danilchik et al., 1998; Jesuthasan, 1998), which is thought to be homologous to the mammalian midbody involved in cell abscission (Otegui et al., 2005). FMA tubules are initially arranged in parallel along the length of the furrow, oriented perpendicular to the cleavage plane, with germ plasm RNP aggregates recruited along the furrow maintaining their association with microtubule ends. A third step in the germ plasm segregation process, ‘distal compaction’, occurs during furrow maturation and involves coordinated changes in both RNP aggregates at the furrow and the FMA (Fig. 1). During this step, RNPs recruited at the furrow translocate towards the nearest furrow distal end, where they form compact RNP masses (Pelegri et al., 1999). At the same time, FMA tubules become enriched and undergo bundling in the distal furrow region, while their ends, which maintain their association with germ plasm RNPs, acquire a characteristic outward tilt. As the furrow is completed, FMA microtubules become disassembled leaving the compacted germ plasm mass at the furrow distal end (Pelegri et al., 1999).

Germ plasm pre-aggregation, furrow recruitment and distal compaction therefore achieve the purpose of gathering maternally inherited single RNPs into discrete structures that will be subsequently inherited by primordial germ cells. In the accompanying article (Eno and Pelegri, 2018), we show that mutations in the maternal effect gene mid1ip1l (also known as aura), which encodes the cytoskeletal regulator Mid1ip1l, result in defects in the recruitment of germ plasm RNPs to the furrows. This function is mediated through mid1ip1l-dependent cortical F-actin dynamics and contractions associated with germ plasm RNPs, which allow RNP retention at the cortex prior to furrow formation and likely also facilitate RNP multimerization during furrow maturation. Here, we focus on the process of distal compaction and document a calcium-dependent mechanism required for the enrichment of germ plasm RNPs, together with associated FMA, to the distal ends of the furrow.

In both zebrafish and medaka, waves of intracellular calcium increase are associated with zebrafish cytokinesis during the first several cell cleavage cycles (Fluck et al., 1991; Chang and Meng, 1995; Reinhard et al., 1995; Webb et al., 1997; Créton et al., 1998). These waves have been termed slow calcium waves (SCWs) (Fig. 1) owing to their relatively low speed of propagation (0.1-1 µm/s) (Jaffe, 1993). Two primary furrow-associated SCWs have been described. A superficial ‘initiation wave’, generated just prior to the appearance of the furrow cleft, is required for furrow formation. Subsequently, a larger ‘maturation wave’ penetrates deeper into the cytoplasm and has a role in events associated with furrow maturation, such as vesicle exocytosis and the formation of F-actin-dependent lamellipodia (Lee et al., 2003; Li et al., 2008).

The intracellular calcium signaling network is complex and the release of calcium from internal stores is controlled by three primary sets of ligands and calcium channels: inositol trisphosphate (IP₃) and cyclic adenosine diphosphate ribose (cADPR) through IP₃ and ryanodine receptors, respectively, both at the endoplasmic reticulum (ER), and nicotinic acid adenine dinucleotide phosphate (NAADP) through NAADP-sensitive channels on lysosome-like organelles (Berridge et al., 2003). Previous studies have shown the presence of calcium stores activated by IP₃ and NAADP, but not cADPR, in early zebrafish embryos, and that inhibition of IP₃ receptors affects furrow formation in early blastomeres (Lee et al., 2003).

Embryos from homozygous nebel mutant females exhibit normal nuclear division and furrow initiation, but the forming furrows either fail to develop a fully adhesive membrane septum between daughter cells or undergo regression, revealing a defect in the completion of cytokinesis (Pelegri et al., 1999). In nebel mutant embryos, germ plasm RNP aggregates form compact masses and FMA tubule tips undergo bundling, but these events occur in medial, rather than distal, regions of the furrow.

Several factors are known to mediate RNP aggregate and FMA dynamics in early embryonic blastomeres. In myosin-inhibited embryos, the RNP aggregate and FMA are arrested in a conformation corresponding to furrow initiation, namely as an elongated RNP aggregate along the furrow with the FMA retaining its original parallel conformation (Urven et al., 2006). A similar reorganization defect occurs in mid1ip1l mutant embryos, likely because of the function of this gene in F-actin reorganization (Eno et al., 2016; Eno and Pelegri, 2018). Recent studies have shown that inhibition of the small GTPase RhoA and its effector ROCK result in aberrant enrichment of germ plasm in medial regions of the mature furrow, a defect similar to that observed in nebel mutants (Miranda-Rodríguez et al., 2017). However, mechanisms underlying germ plasm and FMA reorganization, including cues conferring medial-to-distal directionality along the furrow plane, remain largely unknown.

Here, we show that during the early cleavage stages, nebel mutant embryos show reduced intensity and temporal dynamics of furrow-associated SCWs. Treatments that interfere with SCW formation also result in defects in germ plasm RNP and FMA distal enrichment. Under these conditions, FMA bundles inward, as in later blastomeres that lack SCWs. Germ plasm RNP aggregate compaction and FMA bundling direction correlate with the medial-to-distal directionality of SCWs and are dependent on Calmodulin activity. Our studies indicate that SCWs in the early embryo modify RNP distribution and microtubule reorganization during furrow maturation, in order to facilitate the formation of multiple germ plasm masses capable of inducing the germ cell fate.

Because of the proposed role for SCWs in early teleost embryo furrow formation (Chang and Meng, 1995; Webb et al., 1997; Créton et al., 1998; Li et al., 2008), we visualized SCWs in wild-type and nebel mutant embryos with intracellular calcium-sensitive dyes. We first used the ratiometric indicator Fura-2-Dextran, which allows for quantitation of intracellular calcium levels. The patterns of intracellular calcium concentration in wild type and nebel mutants were markedly different. As previously reported (Chang and Meng, 1995; Webb et al., 1997; Créton et al., 1998; Li et al., 2008), intracellular calcium increases appear in wild-type embryos immediately prior to furrow initiation and during furrow maturation, corresponding to initiation and maturation SCWs, with the maturation SCW showing calcium release of greater intensity and depth (Fig. 2A, Fig. S1A, Movie 1). By contrast, nebel mutants exhibit intracellular SCWs of reduced amplitude and frequency (see below) compared with wild type (Fig. 2B, Fig. S1B, Movie 2). Embryos mutants for mid1ip1l, another maternal effect mutation affecting late cytokinesis (Pelegri et al., 2004; Eno et al., 2016), exhibit normal furrow-associated SCWs (data not shown), indicating that the SCW defect in nebel is not a general consequence of failed completion of cytokinesis.

To better study the relationship between SCWs and furrow development, we scored wild-type and mutant embryos for defects during furrow initiation and completion (Table S2). Nearly all cleaving nebel mutant embryos (98%, n=255) underwent normal furrow initiation (Fig. 2D) at a time and extent similar to wild type (Fig. 2C). However, these mutant embryos invariably showed defects in furrow completion, such as a lack of formation of a cell adhesive wall or furrow regression (Fig. 2G,H, compared with 2F). Only in rare cases (2%) did blastomeres exhibit defects in furrow initiation, as reflected by lack of furrowing in one of two blastomeres in the 2-cell stage embryo (Fig. 2E). These observations confirm our previous conclusions that nebel largely affects late stages of cytokinesis (Pelegri et al., 1999), and further suggest that nebel function is required to achieve high calcium release levels characteristic of the maturation SCW. As opposed to the two primary SCWs found in wild type, with the maturation SCW lasting most of furrow formation, the SCWs of reduced intensity observed in nebel mutants appeared to fluctuate, with up to four waves of calcium release observed in a single cycle and a relatively constant periodicity of ∼7 min (Fig. 2B).

The defects in intracellular calcium release were also observed using the calcium-sensitive dye Oregon Green 488 BAPTA-Dextran, which does not allow measurement of absolute intracellular calcium concentration but does allow improved visualization and quantification of SCWs. Imaging of wild-type embryos with this dye readily reveals SCWs at the furrow (Fig. 2I), with kymograph analysis showing an average time between peaks of 76 s (s.d.=16.9 s, n=38 peaks from 8 embryos; Fig. 2I′,L, Fig. S2). By contrast, nebel mutants exhibit a higher frequency of SCWs, with an average period between peaks of 27 s (s.d.=8.1 s, n=56 peaks from 8 embryos; Fig. 2J,J′,L, Fig. S2). As expected, exposure of embryos to the IP₃ receptor inhibitor 2-APB results in the rapid cessation of calcium release at the furrow (Lee et al., 2003) (data not shown). Thus, our data show that nebel mutant embryos exhibit defects in furrow-associated SCWs, such as a reduction in magnitude and increase in frequency of intracellular calcium release.

If the nebel mutant phenotype is caused by reduced intracellular calcium release, the furrow defect may be rescued by an increase in intracellular calcium. Initial experiments involving CaCl₂ injections into cleaving embryos did not result in significant rescue of the nebel mutant cellularization defect (data not shown), possibly because of the need for precise spatiotemporal regulation of calcium release at the forming furrow. However, we took advantage of differences between unfertilized eggs from wild-type and nebel mutant females (Pelegri et al., 1999) to test whether calcium can rescue the nebel mutant phenotype (Table S3). After egg activation, unfertilized eggs from wild-type females typically exhibit asymmetric membrane ingression, also known as pseudocleavages (Fig. 2M) (Kane and Kimmel, 1993), possibly through the unregulated exocytosis of intracellular membrane (data not shown; see also Danilchik et al., 1998). In contrast to wild type, activated yet unfertilized eggs from nebel mutant females typically fail to form pseudocleavages (6%, n=125; Fig. 2N). Injection of 0.1 mM or 0.2 mM CaCl₂ into eggs from nebel mutant females significantly increased pseudocleavage formation [to 42% (n=111) or 54% (n=115), respectively; Fig. 2O], whereas control injection did not have this effect (6%, n=123; Fig. 2P). Thus, the absence of pseudocleavage formation in unfertilized nebel mutant eggs can be reversed by introducing calcium, consistent with a reduction in intracellular calcium as an underlying basis for the nebel mutant phenotype.

Previous studies have demonstrated the ability of calcium buffering by mitochondria to influence the magnitude and patterns of intracellular calcium waves (Jouaville et al., 1995; Babcock et al., 1997). Specifically, mitochondrial activity is directly correlated to the amplitude and inversely correlated to the frequency of IP₃-induced propagating calcium waves in the Xenopus oocyte cortex (Jouaville et al., 1995). Because these effects are reminiscent of those found in furrow-associated SCWs in nebel mutants, we tested whether inhibition of mitochondrial activity leads to defects in SCWs in wild-type embryos. Indeed, wild-type embryos exposed during furrow formation to the mitochondrial respiratory Complex II inhibitor 3-nitropropionic acid (Alston et al., 1977) exhibited reduced intensity and increased frequency of SCWs (Fig. 2K,K′,L, Fig. S2). Thus, similar to propagating calcium waves in Xenopus oocytes, furrow-associated SCWs in early zebrafish embryos require functional mitochondrial activity.

Previous studies have shown that IP₃- and NAADP-dependent, but not cADPR-dependent, calcium stores are present and active in the early zebrafish embryo (Lee et al., 2003). To test whether nebel mutants contain functional calcium stores, the calcium release response to injection of IP₃ or NAADP was compared with that of wild-type embryos (Fig. 3A). Agonist was injected into unfertilized embryos in order to reduce experimental variation caused by the uneven distribution of calcium stores at the vicinity of forming furrows (Berridge, 1993; Berridge et al., 2003; Lawrence and Mandato, 2013) (see also Fig. 4). As expected from previous studies (Lee et al., 2003), both IP₃ and NAADP injections (Fig. 3D,F), but not mock injections (Fig. 3B), elicit a calcium release response in wild type. These two agonists (Fig. 3E,G), but not control injections (Fig. 3C), elicited a calcium release response similar in magnitude to that in nebel mutants (Fig. 3H). Consistent with the proposed role of IP₃ in long-range calcium wave propagation through calcium-mediated calcium release (Finch et al., 1991; Galione et al., 1993), the spatial extent of calcium release after IP₃ injection expands into most of the blastodisc (Fig. 3D,E), whereas it is comparatively confined after NAADP injection (Fig. 3F,G), with wild type and nebel mutants exhibiting a similar behavior (Fig. 3I). Together, our data indicate that nebel mutant eggs are able to elicit an apparently normal calcium release response to IP₃ and NAADP, suggesting that the IP₃- and NAADP-sensitive calcium stores and the corresponding receptors are present and functional in mutant eggs and embryos.

Since nebel mutant embryos exhibit reduced SCWs, yet the eggs exhibit a robust response to calcium agonists, we tested whether the localization patterns of calcium stores are affected in mutant embryos. Labeling with DiOC₆ shows an enrichment of ER at the furrows of early wild-type embryos (Fig. 4A). By contrast, the ER fails to become enriched at the furrows in nebel mutants, being found instead in a diffuse pattern throughout the blastodisc (Fig. 4B). Using Lysotracker as a labeling dye, we find that lysosomes also become aggregated to the forming furrows in wild-type embryos (Fig. 4C). nebel mutant embryos, on the other hand, lack an enrichment of lysosomes at the furrow, which instead accumulate in peripheral regions of the embryo (Fig. 4D). Labeling with Mitotracker shows that mitochondria exhibit a similar pattern to that of ER, with enrichment at the furrows in wild type (Fig. 4E) but distributed in a diffuse pattern in nebel mutants (Fig. 4F). Thus, nebel mutants show a general defect in the enrichment of calcium stores at the forming furrow, including the ER, lysosomes and mitochondria. Given the essential role for these stores and their localization in regulated calcium release (Jouaville et al., 1995; Babcock et al., 1997; Lee et al., 2003), this defect is likely to be a primary cause for the observed effects on furrow-associated SCWs in these mutants.

We tested whether inhibition of calcium release phenocopies the germ plasm segregation and FMA reorganization defects observed in nebel mutants, visualizing germ plasm RNPs with antibodies against serine 19 phospho-myosin light chain (P-myosin) (Nair et al., 2013), colabeled for microtubules, or by in situ hybridization to detect the RNA for dead end (dnd) (Weidinger et al., 2003) (Fig. 5). In wild-type embryos, germ plasm RNPs exhibit distal compaction, involving the transition from a rod-like structure along the forming furrow to a compacted mass at its distal end (Pelegri et al., 1999; Eno and Pelegri, 2013; reviewed by Eno and Pelegri, 2016) (Fig. 5A,K), which is coupled to the outward tilting and distal enrichment of associated FMA (Jesuthasan, 1998; Pelegri et al., 1999; Urven et al., 2006). In nebel mutants, germ plasm RNPs become recruited normally along the furrow and the resulting aggregate typically transitions into a more compact mass, but this mass and associated FMA fail to become enriched in distal furrow regions (Pelegri et al., 1999) (Fig. 5C,M). As expected considering the role of calcium in furrow initiation (Chang and Meng, 1995), inhibition of calcium release with an injection of the calcium chelator BAPTA results in a reduction in both furrow formation and germ plasm RNP furrow recruitment (Fig. 5E,O).

We next tested the effects on germ plasm and associated FMA in wild-type embryos of inhibitors predicted to interfere with SCWs (Lee et al., 2003). Exposure to the IP₃-mediated calcium release inhibitors 2-APB [an IP₃ receptor inhibitor (Bilmen and Michelangeli, 2002)] (Fig. 5G,Q) and U73122 [a Phospholipase C inhibitor that reduces IP₃ production (Macmillan and Mccarron, 2010)] (Fig. 5I,S) resulted in germ plasm RNPs and associated FMA retaining an arrangement characteristic of initiating furrows, namely a distribution of RNPs along the furrow without subsequent compaction, and an FMA perpendicular to the furrow without tilting or distal enrichment. Similar effects on RNPs and FMA were observed after exposure to the mitochondrial respiratory Complex II inhibitor 3-NP (Fig. 5B,L). Thus, treatments that interfere with SCWs in wild-type embryos lead to defects in enrichment at the distal furrow of germ plasm RNPs and associated FMA – effects similar to those observed in nebel mutants.

We further investigated FMA dynamics in live embryos under conditions that inhibit SCW formation using EMTB::EGFP transgenic fish (Fig. 6). In wild-type embryos at the 2- to 4-cell stage, the FMA translocates distally during furrow formation, where it subsequently undergoes disassembly (9 of 10 embryos, Fig. 6A) (Jesuthasan, 1998; Pelegri et al., 1999; Urven et al., 2006; Eno et al., 2016). By contrast, in nebel mutant embryos the FMA becomes enriched in the medial regions of the furrow, either at a single site at the center of the blastodisc (3 of 10 embryos, Fig. 6B) or at two sites abutting the center of the blastodisc (6 of 10 embryos, not shown) (see also Pelegri et al., 1999). Similar to our observations in fixed embryos, in live embryos in which IP₃ receptors are inhibited with 2-APB the FMA fails to become enriched in distal furrow regions, remaining instead in its original conformation along the length of the furrow (Fig. 6C). Similar effects are observed in live embryos after inhibition of mitochondrial function with the Complex II inhibitor 3-NP (data not shown).

Furrow-associated SCWs are detected in early zebrafish embryos during the earliest cell cycles, leading to the 8-cell stage, but become undetectable during subsequent cell cycles (Chang and Meng, 1995). Such later cell cycles (leading to the 16-cell through 500-cell stages) continue to exhibit an FMA-like structure during early stages of cytokinesis. However, bundling of FMA tubules in wild type at these later stages occurs in medial regions of the furrow (Fig. 6D,E), similar to the behavior of coalescing midbodies in most cell types (Otegui et al., 2005; Steigemann and Gerlich, 2009). Thus, the presence of furrow-associated SCWs in the early wild-type embryo correlates with distally oriented reorganization of the FMA and associated germ plasm RNPs.

The requirement for intracellular calcium to implement the medial-to-distal reorganization of the FMA and germ plasm RNP aggregation is consistent with previous suggestions that SCWs originate in the center of the furrow and propagate towards its distal ends during furrow maturation (Chang and Meng, 1995; Webb et al., 1997; Créton et al., 1998), which we confirm using a transgenic line carrying the intracellular calcium reporter construct Tg[βactin2:GCaMP6s] (Chen et al., 2017) (Fig. 7A, Fig. S3A,G). Analysis of microtubules in live wild-type embryos using the EMTB::EGFP transgene shows that the distally oriented calcium increase is mirrored by an increase in FMA intensity in distal regions at later stages of furrow formation (Fig. 7C, Fig. S3C,K). By contrast, similar analyses in nebel mutant embryos show a lack of enrichment in the distal furrow for both intracellular calcium (Fig. 7B, Fig. S3B,H) and microtubules (Fig. 7D, Fig. S3D,L). Inhibition of SCWs in wild-type embryos using either the IP₃ receptor inhibitor 2-APB (Fig. S3E,I,M) or the mitochondrial activity inhibitor 3-NP (Fig. S3F,J,N) also interferes with the distal enrichment of both intracellular calcium and FMA. These analyses provide further support for a medial-to-distal directionality of SCWs, and show that it is associated with the distally oriented reorganization of the FMA and germ plasm RNP aggregates during furrow formation.

Our studies suggest that SCWs convey information to polarize the cytoskeleton during furrow formation. We therefore considered as potential mediators of furrow polarization two factors whose activity is known to be calcium dependent: non-muscle myosin (Scholey et al., 1980) and Calmodulin (Albert et al., 1984; Clapham, 2007). Treatment with either of two Calmodulin inhibitors, W7 and W5, results in the FMA failing to undergo distal reorganization and in RNP aggregates remaining distributed along medial regions of the furrow (Fig. 5D,N; data not shown). Labeling of wild-type embryos with anti-Calmodulin and anti-Calmodulin kinase (CamK) antibodies shows enrichment of these proteins at the furrow in wild-type embryos (Fig. 8A,C). Furrow localization for both factors is reduced in nebel mutant embryos or 2-APB-treated embryos (Fig. 8B,D; data not shown), in which SCWs are inhibited. Inhibition of Rho activity by exposure to C3 exoenzyme also results in defects in RNP distal compaction (Fig. S4), as recently reported (Miranda-Rodríguez et al., 2017).

As previously reported (Urven et al., 2006), treatment of wild-type embryos with inhibitors of myosin activity, such as blebbistatin and ML7, results in a lack of FMA reorganization and germ plasm RNP distal compaction during furrow formation (Fig. 5F,H,P,R). Moreover, germ plasm RNPs are labeled with the P-myosin antibody (Nair et al., 2013), consistent with active myosin as one of their components. In contrast to Calmodulin and CamK, however, anti-P-myosin labeling is not reduced under conditions in which calcium is strongly reduced, as is also the case in nebel mutants (Fig. 5C′) or in wild-type embryos treated with inhibitors of IP₃ release (2-APB, U73122; Fig. 5G′,I′) or of mitochondrial activity (3-NP; Fig. 5B′) (see Discussion). Treatment of wild-type embryos with phalloidin indicates that stabilization of furrow F-actin also interferes with the distal movement of germ plasm RNPs (Fig. 5J,T).

Three-dimensional reconstructions show that, in control wild-type embryos, F-actin in the contractile ring forms indentations (Fig. 9A, inset), which during furrow maturation are progressively found in a more distal location, mirroring the enrichment of RNPs during distal compaction (Fig. 9B, bracket) (see Eno and Pelegri, 2018). In nebel mutants, in which germ plasm RNPs become recruited normally to the furrow (Fig. 9C) but remain in medial regions of the furrow after maturation (Fig. 9D, bracket), furrow F-actin indentations appear to be reduced (Fig. 9C, inset). Inhibition of calcium release by treatment of wild-type embryos with 2-APB also results in reduced furrow F-actin indentations (Fig. 9E, inset) and aberrant distal RNP enrichment (Fig. 9F, bracket). Similar effects are observed in embryos treated with Calmodulin or myosin inhibitors (Fig. 9G,I, insets; Fig. 9H,J, brackets). F-actin indentations continue to be apparent in phalloidin-treated embryos in spite of defects in germ plasm RNP distal movement (Fig. 9K-L′).

Altogether, these results and the accompanying manuscript (Eno and Pelegri, 2018) indicate that calcium- and myosin-dependent contractility, as well as F-actin dynamics, are essential for the distal reorganization of germ plasm RNPs and associated FMA.

In this study, we show that nebel mutants exhibit changes in furrow-associated SCWs, specifically a reduction in calcium release levels and an increase in release frequency. Our studies show that directional calcium release during the early mitotic cell cycles has a key role in the reorganization of furrow-associated microtubules and associated RNPs in the early embryo, leading to their enrichment at the furrow distal ends.

Wild-type embryos exhibit two primary SCW waves: an initiation wave prior to furrow indentation and a deeper and more intense wave during furrow maturation (Chang and Meng, 1995; Reinhard et al., 1995; Webb et al., 1997; Créton et al., 1998). By contrast, nebel mutants exhibit SCWs of comparatively low intensity. Intracellular calcium release is essential for both furrow initiation and maturation (Lee et al., 2003), but nebel mutant embryos are, in most cases, able to initiate furrow ingression (Pelegri et al., 1999; this report). These observations suggest that reduced SCWs in nebel mutants are sufficient for furrow initiation but not furrow maturation.

The calcium release response after injection of IP₃ and NAADP receptor antagonists is robust in nebel mutants, indicating that calcium stores are present in the embryos and that IP₃ and NAADP receptors are functional. However, nebel embryos show defects in the localization of calcium stores such as the ER and lysosomes (which contain IP₃ and NAADP receptors, respectively). In wild type, these calcium-rich organelles are enriched in the forming furrows, whereas in nebel embryos they are distributed throughout the blastomeres. We also find that nebel embryos lack furrow enrichment of mitochondria, another organelle involved in calcium release and buffering (see below). Segregation analysis shows that nebel is located between markers z21679 and z14228 on chromosome 3 (our unpublished data), and we are currently attempting to unambiguously identify the affected gene and determine its molecular function.

Precise patterns of calcium release are known to depend on multiple cellular calcium stores often interacting in close proximity (Penny et al., 2015). Thus, it is likely that reduced proximity and density of calcium stores at the furrow results in the observed changes in SCWs of nebel mutants (Fig. 10A). Our results highlight the importance of the spatial enrichment of calcium stores for the generation of SCWs.

Previous studies have shown that the oocyte cortex exhibits autocatalytic properties characteristic of an excitable medium, where IP₃-dependent calcium release above an activation threshold results in further calcium release through calcium-induced calcium release (CICR) (Lechleiter and Clapham, 1992). In addition, calcium release channels are repressed by high concentrations of calcium, creating a negative-feedback loop. Together, CICR and calcium-mediated channel inhibition are thought to mediate the spatial and temporal progression of calcium waves (Chang and Meng, 1995; Reinhard et al., 1995; Webb et al., 1997; Créton et al., 1998).

Mitochondria also accumulate and release calcium (DeLuca and Engstrom, 1961; Ichas et al., 1997). Calcium uptake into mitochondria is coupled to the proton electrochemical gradient across its inner membrane (Gunter and Pfeiffer, 1990) and raises the calcium concentration threshold for CICR excitation, resulting in calcium waves of higher amplitude and decreased frequency (Jouaville et al., 1995). Conversely, a decrease in mitochondrial activity results in calcium waves of reduced amplitude and increased frequency, as observed in nebel mutants.

Because of the low sensitivity of mitochondria to calcium concentrations, the action of mitochondria has been proposed to rely strongly on the close proximity of mitochondria to IP₃-dependent receptors in the ER (Jouaville et al., 1995). Thus, the defective furrow enrichment of mitochondria is also likely to contribute to the SCW defects in nebel mutants.

The FMA in early vertebrate embryos such as Xenopus and zebrafish has been proposed to be homologous to the midbody (Danilchik et al., 1998; Jesuthasan, 1998), a microtubule-based structure found in other cell types essential for membrane abscission during the final stages of cytokinesis (Skop et al., 2004). Reorganization of the canonical midbody involves the entire set of midzone microtubules undergoing inward bundling, towards the center of the dividing cell (reviewed by Mierzwa and Gerlich, 2014). By contrast, the FMA in early zebrafish embryos undergoes bundling of two populations, each representing half of its tubules, towards the distal furrow ends. We find that, in embryos with reduced SCWs, the FMA fails to undergo distally oriented bundling. An inwardly bundling FMA, resulting in a single medial structure, is also observed in wild-type blastomeres at later stages of development, when furrows are no longer associated with SCWs. These observations, together with the medial-to-distal directionality of SCWs along the furrow (Chang and Meng, 1995; Webb et al., 1997; Créton et al., 1998; this study), suggest a role for SCWs in modifying the FMA, a structure observed in large cells that is analogous to the midbody, into two distally bundling structures (Fig. 10B). The distal FMA enrichment is coupled to the formation of large germ plasm masses at the furrow distal ends.

Germ plasm has been found to aggregate at cleavage furrows in other vertebrates, such as in the chick embryo (Tsunekawas et al., 2000). In Xenopus, germ plasm also shows an association to furrow ends (Whittington and Dixon, 1975). Pluripotency shares common features with the germ cell state, such as a Blimp-1-positive stage (reviewed by Welling and Geijsen, 2013), and is associated with midbody inheritance (Ettinger et al., 2011; Kuo et al., 2011). Thus, studies on germ plasm segregation in zebrafish might provide valuable clues to aid in understanding the segregation during cell division of cell fate determinants associated with midbody-like structures.

Germ plasm RNP segregation during furrow formation involves not only the reorganization of the germ plasm aggregate itself but also the reorganization of associated microtubule and F-actin networks (Fig. 10C) (see also Eno and Pelegri, 2018). Medial-to-distal SCW directionality is associated with the observed distally directed reorganization, suggesting that the direction of calcium release itself constitutes a polarizing cue along the furrow. SCWs are associated with the localization to the furrow of Calmodulin and its potential downstream regulators such as CamK (this report) and Rho (Miranda-Rodríguez et al., 2017; this report), enrichments that are likely to rely on the presence of binding partners and which are thought to allow subcellular functional specificity (Lloyd-Burton et al., 2007; Bielak-Zmijewska et al., 2008; Fok et al., 2008; Mikl et al., 2011; Abe et al., 2012; Liu and Murray, 2012; Braun and Olayioye, 2015).

SCWs could provide a directional cue along the furrow by acting directly on F-actin. A potential candidate effector to mediate the effects of SCWs is non-muscle myosin II (reviewed by Conti and Adelstein, 2008). Our previous studies demonstrated a requirement for myosin function in distal compaction (Urven et al., 2006), and we determine that furrow F-actin indentations, associated with aggregating RNPs, depend on myosin activity. Antibodies that recognize the serine 19 phosphorylated form of myosin light chain, diagnostic of myosin-II activation (Ikebe and Hartshorne, 1985), label aggregating germ plasm RNPs (Nair et al., 2013). However, P-myosin labeling in RNPs remains at apparently normal levels in nebel mutants or calcium-inhibited wild-type embryos, indicating that this particular phosphorylation is unlikely on its own to mediate a directional response to SCWs. It is possible that SCWs, via Calmodulin, directly influence the F-actin cytoskeleton network, for example through the action of the Rho and ROCK pathway (Wolff et al., 1999; Harada et al., 2000; Fransson et al., 2003; Okabe et al., 2003; Miranda-Rodríguez et al., 2017; this report) or other Calmodulin mediators (Zhang et al., 2012; Zhao et al., 2012).

Directional reorganization may also be mediated by directional bundling of FMA tubules through processes analogous to those occurring during midbody formation in smaller cell types (Mierzwa and Gerlich, 2014), and indeed the integrity of FMA microtubules is important for the distal aggregation of RNPs (Pelegri et al., 1999). In neuronal synapses, activation of CamK by intracellular calcium promotes microtubule stabilization and bundling through the regulation of microtubule-associated proteins (MAPs) (reviewed by McVicker et al., 2015). It is possible that calcium may be similarly acting to promote distal stabilization of FMA tubules in the zebrafish embryo. Recent studies have shown a role for microtubule ends in promoting F-actin polymerization, by recruiting the mDia formins (McVicker et al., 2015; Henty-Ridilla et al., 2016), and it will be interesting to test a potential role for the FMA in modulating F-actin dynamics.

Zebrafish germ plasm RNPs are thought to interact via fluid phase properties mediated by intrinsically disordered proteins (Brangwynne et al., 2009, 2015), such as Bucky ball (Marlow and Mullins, 2008; Bontems et al., 2009; Heim et al., 2014; Riemer et al., 2015; Boke et al., 2016). It is possible that such forces also contribute to RNP reorganization and multimerization during furrow formation. Indeed, the transition of RNP aggregate arrangement from an elongated rod to a compact mass during furrow maturation is expected to involve an overall increase in RNP neighbor-neighbor interactions, presumably towards a reduced energy state. The observation that germ plasm RNPs undergo compaction, albeit in medial regions, in embryos in which the microtubule network has been reduced through nocodazole treatment (Pelegri et al., 1999) further supports an intrinsic ability of RNPs to undergo reorganization through inter-particle interactions.

Understanding the precise molecular mechanisms underlying this complex process, involving cellular mediator signals, F-actin, microtubules and RNPs, will provide insights into how organelles exhibiting fluid phase behavior reorganize in the context of complex networks.

In summary, we show that SCWs contribute to overriding the normal tendency of microtubules during late cytokinesis to bundle as a single unit in the center region of the furrow, facilitating instead the formation of two half-midbody equivalents, one at each of the distal furrow ends. This cytoskeletal reorganization allows the generation of multiple large germ plasm masses, each capable of inducing the primordial germ cell fate. Our studies provide insights into how processes related to the cell division machinery can be modified by overlying cellular signals to facilitate the directed segregation of cellular determinants.

Stocks of wild-type AB and nebel (neb^t023) mutant (Pelegri et al., 1999, 2004; Pelegri and Schulte-Merker, 1999) lines were raised and maintained under standard conditions at 28.5°C (Brand et al., 2002). Homozygous mutant nebel fish were identified by genotyping flanking SSLP markers z21679 and z14228. End-point genotyping (Integrated DNA Technologies) was used to maintain the line with custom-designed primers and probes (Table S1). Mutant embryos were obtained by crossing homozygous nebel females to AB males. Embryos were collected and allowed to develop in E3 embryonic medium (Pelegri and Schulte-Merker, 1999) and were staged according to time to egg activation/fertilization and standard morphological landmarks (Kimmel et al., 1995). Embryos in all experiments were at 26°C, with the exception of embryos for the morphological analysis of furrow initiation and completion, which were transferred within several minutes after fertilization to 22°C to increase phenotypic penetrance (Pelegri et al., 1999). Activated eggs were obtained through exposure of mature, extruded eggs to E3 medium in the absence of sperm.

Analysis with calcium-sensitive dyes was performed by injecting either the ratiometric calcium-sensitive dye Fura-2-Dextran (Slusarski et al., 1997; Westfall et al., 2003) or the calcium indicator Oregon Green 488 BAPTA-Dextran (OGB) (Thermo Fisher O6812) at the 1-cell stage, followed by imaging and analysis. For Fura-2-Dextran, embryos were microinjected with ∼3 nl Fura-2-Dextran solution (Molecular Probes B6810), oriented in a glass-bottomed dish on a Zeiss Axiovert, and image pairs at 340 nm and 380 nm excitation wavelengths (510 nm emission) were collected at 15 s intervals, for a total of 300 image pairs. The ratio image – a pixel-by-pixel match of both excitation wavelengths – was calculated using Ratio Tool (Innovision) software. For OGB, embryos were injected with a mixture of 100 µM OGB and 40 µM Rhodamine (Ashworth et al., 2001) and imaged using a Zeiss Axioplan2 epifluorescence microscope. Embryos were imaged for OGB and Rhodamine at one focal plane every 3 s. For each time point, the Rhodamine channel was subtracted from the OGB channel using FIJI. Kymograph and Plot Profile (FIJI) were used to obtain values for analysis of calcium pulses using AxoGraph software.

Drug exposure was initiated by exposure of dechorionated embryos in E3 medium within 15 min of fertilization, with the exception of BAPTA, which was microinjected (∼1 nl at 5 mM; Sigma A1076) after furrow initiation at 35 minutes postfertilization (mpf) to avoid karyokinesis and/or furrow initiation defects (Chang and Meng, 1995). For drug exposure, inhibitor concentrations in E3 medium were: IP₃ receptor, 500 μM 2-APB (Sigma D9754); mitochondrial Complex I inhibitor, 100 μM rotenone (Sigma R8875); mitochondrial Complex II inhibitor, 10 mM 3-NP (Sigma N5636); Calmodulin, 400 μM W7 (Tocris 0369); Phospholipase (C), 100 μM U73122 (Tocris 1268); Rho, 20 µg/ml C3 exoenzyme (Calbiochem 341208). Control embryos were exposed to the same concentration of carrier solvent (DMSO).

For agonist injection into unfertilized eggs and embryos, solutions of IP₃ (Sigma I9766) and NAADP (Sigma N5655) were injected at 10 μM and 0.5 μM by 15 mpf in ∼0.25 nl. Injection of CaCl₂ (Sigma C1016) at 0.1 and 0.2 mM in ∼1 nl was carried out together with 0.1 M KCl to prevent egg damage (Gilmour et al., 2002), with controls injected with KCl alone at 0.2 mM.

In situ hybridizations were carried out as described previously (Eno and Pelegri, 2013) with digoxigenin-labeled probes for germ plasm RNAs dnd (dnd1) (Weidinger et al., 2003) and vasa (ddx4) (Yoon et al., 1997). Immunofluorescence was carried out as previously described (Lindeman and Pelegri, 2012; Eno et al., 2016) with primary antibodies against phospho-myosin light chain (S19) (Cell Signaling Technology 3671L, 1:50) (Nair et al., 2013), Calmodulin (Millipore 05-173, 1:50) and alpha-tubulin (Sigma T5168, 1:2500). Percentage RNP distal enrichment (compaction) was determined using the number of furrows at the appropriate stage of formation (1-4 furrows per embryo where RNPs are detected). Imaging was acquired using either an Axioplan2 epifluorescence microscope or a Zeiss LSM510 confocal microscope. All images were analyzed using FIJI.

Fluorescent dyes, diluted in E3 medium, were Lysotracker Red (Thermo Fisher L7528, 1:100), MitotrackerCMXRos (Thermo Fisher M7512, 1:100) and DiOC₆ (Thermo Fisher D273, 1:2500). Exposure to dyes was initiated by 15 mpf and dyes were washed out with several changes of E3 medium before imaging at 30 mpf. Transgenic lines to visualize microtubules and intracellular calcium were EMTB-EGFP (Wühr et al., 2010) and Tg[βactin2:GCaMP6s] (Chen et al., 2017), respectively. Live imaging employed an Axioplan2 epifluorescence microscope. Image processing and analysis were carried with FIJI modules Z-project (projections and 3D reconstructions), Reslice (kymographs) and Plot Profile (quantitation).

We thank members of our laboratories for advice during various stages of this work, as well as animal husbandry staff for the care of the aquatic facility. We thank Erica Weber for assistance with image acquisition. We are also grateful to Drs Liliana Solnica-Krezel (Washington University School of Medicine), Martin Wühr (Harvard University, Princeton University) and Tim Mitchison (Harvard University) for generously providing transgenic lines.