An interphase contractile ring reshapes primordial germ cells to allow bulk cytoplasmic remodeling

Developing and differentiating cells often inherit unneeded gene products or organelles. Such components can be eliminated individually, such as in the targeted degradation of transcription factors (Page et al., 2007), or via the autophagic removal of entire organelles (Sato and Sato, 2017). Alternatively, some cells discard contents in bulk by creating a compartment that is filled with undesired cellular material and eliminated. For example, during erythropoiesis in mammals, erythroblasts form a subcellular compartment containing the nucleus, which is subsequently removed and digested by macrophages to leave behind an anucleate immature red blood cell (Moras et al., 2017). Similar remodeling events occur in the germ line, such as during spermatogenesis, when many cellular components are discarded in the residual body (Nishimura and L’Hernault, 2017). We recently described a bulk remodeling event in Caenorhabditis elegans embryonic primordial germ cells (PGCs), which produce large lobes that are filled with mitochondria and other cellular components; subsequently, adjacent intestinal cells cannibalize the lobes, remodeling the PGCs and their contents (Abdu et al., 2016). In each of these examples, the discarded compartment forms through changes in cell shape involving localized constrictions at the cell surface (Abdu et al., 2016; Breucker et al., 1985; Koury et al., 1989). The cellular mechanisms used to create and eliminate such compartments are not well understood.

Forces driving constriction of the plasma membrane typically arise from the localized contraction of myosin on cortical microfilaments (Martin and Goldstein, 2014; Munjal and Lecuit, 2014). Cytokinesis is a well-studied example. During cytokinesis, a contractile ring enriched in myosin, F-actin, and cross-linking and anchoring proteins such as anillin and septin, enriches in a zone at the cell equator (Glotzer, 2017; Srivastava and Robinson, 2015). The contractile ring narrows in circumference, generating forces that cause the plasma membrane to ingress. Centralspindlin, a complex of the kinesin-6 MKLP1 and MgcRacGAP (Basant and Glotzer, 2018), accumulates at the spindle midzone and at the future site of furrow formation, where it activates a cortical pool of the RhoGEF Ect2. In turn, Ect2 locally activates RhoA, which activates myosin and the actin nucleator formin, leading to assembly and contraction of the contractile ring (Green et al., 2012). Contractile rings have also been described in a few types of nonmitotic cells, such as enucleating erythroblasts, ascidian notochord cells, and budding Drosophilamelanogaster pole cells (Cinalli and Lehmann, 2013; Koury et al., 1989; Sehring et al., 2015). The cues responsible for forming and positioning contractile rings in nondividing cells are poorly understood. Here, we show that a nonmitotic contractile ring induces lobe formation in C. elegans PGCs, and we identify cellular and molecular pathways that induce its formation. Our findings provide a cell biological mechanism for forming transient compartments that are used to discard cellular content in bulk.

The two C. elegans PGCs form lobes several hours after their birth (Abdu et al., 2016; Sulston et al., 1983). We visualized PGC lobe formation by acquiring movies of embryos expressing PGC-specific membrane-localized mCherry (Mem^PGC; Video 1). Movies began before lobes formed (bean stage), when both PGCs were visible within a plane (Fig. 1 a). During the period of lobe formation (by comma stage), embryos rotated 90° such that only one PGC was visible (Fig. 1 a’’). Just before lobes formed, PGCs transitioned from a nearly spherical to an elliptical shape with an increased aspect ratio (Fig. 1, a’, b, and b’). Subsequently, PGCs constricted centrally to adopt an hourglass shape with a large lobe (L) on one side of the cell and the cell body containing the nucleus (*) on the other (Fig. 1 a’’; see also Fig. 5 e). This transition occurred within an hour. Lobe cannibalism (not depicted) occurs at a later stage of embryogenesis (Abdu et al., 2016).

PGCs form lobes without changes to cell volume and when cultured in isolation, indicating that lobe formation is an autonomously driven cell shape change (Abdu et al., 2016). Although PGCs are arrested in G2 (Fukuyama et al., 2006), lobe formation resembles an incomplete cytokinesis. To determine if a contractile ring could provide the forces for lobe formation, we examined contractile ring components in PGCs as they formed lobes (Fig. 1 c). We first examined the myosin heavy chain NMY-2 using a functional nmy-2-gfp knockin (Dickinson et al., 2013) and a PGC-specific nmy-2-yfp transgene (NMY-2-YFP^PGC). In addition to accumulating at the intercellular bridge that connects the two PGCs (the result of an incomplete cytokinesis; Goupil et al., 2017; Sulston et al., 1983), NMY-2-GFP and NMY-2-YFP^PGC concentrated at the base of PGC lobe necks within a ring (Video 2), which when viewed in cross section appear as two foci (Fig. 1, e–e’’’, arrowheads). Both fusion proteins colocalized at the PGC lobe neck ring (Fig. 1, e’’’ and f), and puncta of NMY-2-YFP^PGC appeared at incipient lobe necks as they first began to form (Fig. S1, a–a’’). These observations suggest that myosin might provide forces that induce the PGC membrane to ingress.

Anillin is a contractile ring cross-linking protein that binds to F-actin, myosin, and septins (Piekny and Maddox, 2010). The anillin homologue ANI-1, like NMY-2, concentrated in a ring at the base of PGC lobes (Fig. 1, g–g’’’ and h). Septins are filamentous proteins that enrich in contractile rings (Mostowy and Cossart, 2012). C. elegans contains two septins, UNC-59 and UNC-61, whose products associate interdependently (Nguyen et al., 2000). UNC-59 and NMY-2 also coenriched at the base of lobe necks (Fig. 1, i–i’’’ and j). We examined PGC F-actin localization indirectly by expressing the actin-binding protein Moesin-GFP specifically in PGCs. Before PGC lobes formed, GFP-Moesin^PGC was enriched uniformly at the cell cortex (Fig. 1 k’). After lobes formed, GFP-Moesin^PGC was distributed throughout the cell cortex (Fig. 1, l–l’ and m) and also accumulated within the lobe, suggesting that F-actin may have an additional function within lobes. We conclude that a contractile ring enriched in myosin, anillin, and septin assembles at the PGC lobe neck.

To determine whether contractile ring components are needed for lobe formation, we used temperature-sensitive alleles of genes required for cytokinesis. cyk-1 encodes the C. elegans Diaphanous family formin (Mi-Mi et al., 2012), which promotes unbranched actin polymerization and is critical for cytokinesis (Bohnert et al., 2013; Davies et al., 2014). We upshifted synchronized cyk-1 mutant embryos to the restrictive temperature just before lobes normally form, capturing an image stack at this time (t = 0 min) and nearly 2 h later (t = 108 min), when PGCs in WT embryos have formed lobes (Fig. 2 a; temperature-shift regime I). PGCs failed to form lobes in most cyk-1 mutants (no lobes, Class I; Fig. 2, c’ and f), in contrast to WT control embryos imaged on the same slide (Fig. 2, b’ and f), suggesting that formin-mediated linear F-actin polymerization is required for lobe formation.

We next examined the requirement for myosin using the nmy-2(ne3409ts) temperature-sensitive allele, which contains a missense mutation within the S2 region (Liu et al., 2010) important for dimerization and motor activity (Tama et al., 2005). PGCs formed lobes in only 5% of nmy-2 embryos (Fig. 2 g), and mutants fell into two phenotypic classes. In a small fraction of embryos, PGCs remained spherical and did not form lobes, similar to cyk-1 mutant embryos (no lobes, Class I; Fig. 2, d’ and g). The majority of nmy-2 mutants showed an additional defect, forming a bubble-shaped expansion at the connection between the PGCs (no lobes, Class II; Fig. 2, e’ and g), which we suspected was an expansion of the intercellular bridge connecting the PGCs. To test this hypothesis, we expressed the centralspindlin component ZEN-4, a marker of division remnants (Raich et al., 1998), specifically in PGCs. ZEN-4-YFP^PGC accumulated at the division remnant between the PGCs in WT embryos (Fig. 2 h) but was not present at PGC lobe necks (Fig. 2 i). In most nmy-2 embryos that were upshifted at a stage before lobes normally form (Fig. 2 j), a bubble developed between the two PGCs (17/22 embryos) that invariably contained a focus of ZEN-4-YFP^PGC (17/17 embryos; Fig. 2, k–k’’). These findings indicate that nmy-2 is required both to maintain the PGC intercellular connection, as previously shown (Goupil et al., 2017), and to form lobes, suggesting that myosin-mediated contraction of F-actin in the contractile ring provides the forces that promote lobe neck ingression.

In dividing cells, RhoA triggers contractile ring formation by activating formin (Otomo et al., 2005; Watanabe et al., 2008) and myosin (Kosako et al., 2000). Given that formin and myosin are required for lobe formation, we considered the RhoA homologue RHO-1 a likely activator of the lobe neck contractile ring. We examined the requirement for RHO-1 indirectly using a temperature-sensitive allele of its activator ECT-2 (Zonies et al., 2010), which localizes through the cortex of PGCs (Fig. 3, a and a’). 71% of the ect-2 mutant embryos that were upshifted using temperature-shift regime I failed to form PGC lobes (Fig. 3, c‒e). These included Class I embryos specifically defective in lobe formation (Fig. 3, c and c’); Class II embryos with an expanded intercellular bridge; and Class III embryos, which contained a binucleate PGC lacking lobes (Fig. 3, d and d’). The finding that ECT-2 is required for PGC lobe formation strongly suggests that active RHO-1 induces PGC lobe formation.

Two mechanisms have been shown to promote ECT-2 activation during cytokinesis. The predominant ECT-2 activator is centralspindlin (Basant and Glotzer, 2018), whereas the novel protein NOP-1 contributes to ECT-2 activation redundantly but, on its own, is dispensable for cytokinesis (Tse et al., 2012). C. elegans centralspindlin is a complex of CYK-4/MgcRacGAP and ZEN-4/MKLP1 (Mishima et al., 2002). Temperature-sensitive alleles of zen-4 and cyk-4 that disrupt the interaction between the two proteins [zen-4(or153ts)] or that prevent CYK-4 membrane localization [cyk-4(or749ts)] effectively block cytokinesis (Davies et al., 2014; Pavicic-Kaltenbrunner et al., 2007; Zhang and Glotzer, 2015). Surprisingly, cyk-4 or zen-4 embryos that were upshifted using temperature-shift regime I showed normal PGC lobe formation (Fig. 4, b’, c’, f, and g). To confirm that similar temperature upshifts of cyk-4 and zen-4 were able to block cytokinesis, we examined division of P₄, the parent cell of the PGCs. The majority of cyk-4 or zen-4 mutant embryos upshifted before P₄ mitosis failed in P₄ cytokinesis (Fig. S2, a–f). These results, together with the lack of ZEN-4 localization at PGC lobe necks (Fig. 2 i), strongly suggest that PGC lobe formation does not require centralspindlin.

To determine whether ECT-2 might be activated instead by NOP-1, we examined nop-1 mutant embryos, which are mostly viable (Rose et al., 1995). In nop-1 mutant embryos, most PGCs lacked lobes; however, a few developed defects in the maintenance of the PGC intercellular bridge that we noted in nmy-2 and ect-2 mutant embryos (Fig. 4, e’ and h). Thus, ECT-2 regulation differs during PGC lobe formation and cytokinesis: PGC lobe formation requires NOP-1, whereas cytokinesis requires centralspindlin.

Since the PGC nucleus is always opposite the lobe, we wondered whether it moves to a stereotypical location within the cell before lobe formation. We examined nuclear dynamics by filming embryos expressing Mem^PGC and a PGC-specific nuclear marker (GFP-H2B). To quantify nuclear position relative to the PGC cell body, we defined the PGC-PGC contact as the medial side of each PGC and the opposite end of each cell as the lateral side. Prior to lobe formation, the nucleus was positioned near the center of the PGC (Fig. 5, a and b). As PGCs lengthened before lobe formation (Fig. 1 a’), the nucleus migrated to the lateral edge (Fig. 5, c and d). Centrosomes, which we detected using SPD-2-GFP, invariably localized on the lateral-facing side of the nucleus (27/27 PGCs; Fig. 5, g‒i). However, time-lapse movies of the plus-end binding protein EBP-2-GFP (Sallee et al., 2018) revealed that centrosomes in PGCs do not function as major microtubule organizing centers; rather, microtubules appeared to grow predominantly from the plasma membrane (38/38 embryos; Video 3). These observations raise the possibility that nuclear asymmetry and lobe formation are coupled and that nuclear movement may drive PGC elongation.

After nuclear migration, PGC lobes invariably formed toward the medial side of the cell rather than extending laterally (Fig. 5, e and f), suggesting that the nucleus might locally inhibit lobe formation. To determine if and where lobes form when the nucleus is abnormally positioned, we used zen-4 mutants to create embryos containing a binucleate PGC. For these experiments, we upshifted zen-4 mutants before the P₄ cell division (t = 0 min) and captured images at two subsequent time points (Fig. 5 j). At time point 1 (t = 110 min), the P₄ cell had completed division in all WT embryos (17/17) but had not yet formed lobes (Fig. 5, l–l’). At time point 2 (t = 300 min), PGCs in WT embryos had formed lobes (17/17; Fig. 5, m–m’). Nearly all (39/40) zen-4 embryos exposed to this temperature-shift regime contained a single binucleate PGC at time point 1 (Fig. 5, n–n’), and by time point 2, the binucleate PGC developed deep lobe-like invaginations on the side of the cell opposite the clustered nuclei (33/39 embryos; Fig. 5, k and o–o’). Movies captured with higher time resolution revealed that the paired nuclei moved to one side of the PGC before lobe formation was initiated on the opposite side (7/7 embryos; Fig. S3 a). As in WT PGCs, NMY-2-YFP^PGC accumulated at the base of lobe necks in binucleate zen-4 PGCs (Fig. S3 b’’, arrowheads). These findings are consistent with a role for nuclei in positioning the PGC lobe.

We suggest the following model for PGC lobe formation. First, the PGC nucleus migrates to the lateral edge of the PGC, which we propose induces the formation of a contractile ring between the nucleus and the medial side of the cell. Ring constriction causes the plasma membrane to ingress, creating the lobe. Our findings reveal clear similarities and differences in the regulation of PGC lobe and cytokinetic contractile rings. One similarity is the requirement for RhoA activity (Jantsch-Plunger et al., 2000). However, a key difference is that PGC lobe formation occurs independently of centralspindlin, which is essential for cytokinesis. Whereas centralspindlin functions in cytokinesis by activating ECT-2, this role appears to be largely fulfilled in PGCs by NOP-1. The dispensability of centralspindlin may not seem surprising given that PGCs are nonmitotic and therefore lack a spindle. However, a spindle-independent role for centralspindlin component CYK-4 was recently described in oocyte cellularization (Lee et al., 2018).

What provides the spatial information that positions the lobe contractile ring? PGC-PGC and PGC‒somatic cell interactions are not required for lobe formation, as lobe formation occurs in a single binucleate PGC (Fig. 5, n–o’) or in PGCs cultured in isolation (Abdu et al., 2016). Because the lobe contractile ring always forms on the medial side of the asymmetrically positioned nucleus, we consider it likely that the nucleus or attached centrosomes contribute to contractile ring positioning. If so, PGC lobe formation may be mechanistically similar to pseudocleavage formation, which occurs in the C. elegans zygote in the absence of a mitotic spindle and also depends on NOP-1 (Albertson, 1984; Rose et al., 1995; Tse et al., 2012). Pseudocleavage is prompted by a cue from the centrosome, which inhibits actomyosin contractility at the posterior cortex, causing actomyosin to contract anteriorly into a cap that constricts into a furrow at its posterior border (Cowan and Hyman, 2004; Hird, 1996; Munro et al., 2004; Strome, 1986). It is plausible that centrosomes also induce formation of the PGC contractile ring since they invariably face away from the site of lobe formation, although our EBP-2-GFP imaging experiments suggest that they are unlikely to do so by functioning as major sites of microtubule polymerization.

PGC lobe formation also shares similarities and differences with erythroblast enucleation. In erythroblasts, the nucleus migrates to one side of the cell, which subsequently forms and constricts an equatorial contractile ring to separate the cytoplasmic compartment from the nuclear compartment (Moras et al., 2017; Ovchynnikova et al., 2018). However, a key difference is in how the contractile ring is regulated. Whereas erythroblasts require Rac to form the enucleation contractile ring (Ji et al., 2008; Konstantinidis et al., 2012), our findings implicate RhoA. Thus, while contractile rings similar to those that separate cells during cytokinesis can be reemployed during interphase to create cellular compartments used for remodeling, divergent regulatory inputs can direct their formation.

Cannibalism of the PGC lobes dramatically reduces PGC volume and mitochondrial content (Abdu et al., 2016). While the importance of PGC remodeling is unknown, one role may be to eliminate most PGC mitochondria, which are high in potentially damaging superoxides relative to mitochondria in other embryonic cells (Abdu et al., 2016). It is interesting to note that nop-1 mutants have reduced fecundity and ∼20% embryonic lethality (Rose et al., 1995; Tse et al., 2012), raising the possibility that these phenotypes could be consequences of defects in PGC remodeling.

C. elegans strains were maintained at room temperature on nematode growth medium plates seeded with Escherichia coli strain OP50. Temperature-sensitive strains and FT1909 were maintained at 15°C. The following strains were used: DP38: unc-119(ed3) (Maduro and Pilgrim, 1995), FT404: ect-2(gk44); unc-119(ed3); xnIs162 [Pect-2::ect-2-gfp, unc-119(+)], FT1258: him-8(e1489); xnIs360 [Pmex-5::mCherry-PH::nos-2^3′UTR, unc-119(+)] zuIs70 [Pend-1::gfp-caax, unc-119(+)], FT1614: xnIs360; Ppie-1::gfp-H2B::nos-2^3′UTR, unc-119(+) (D’Agostino et al., 2006), FT1729: avr-14(ad1302); xnSi49 [Pmex-5::nmy-2-yfp::nos-2^3′UTR, unc-119(+)]; unc-119(ed3); avr-15(ad1051) glc-1(pk54), FT1749: xnSi49; xnIs360, FT1851: zen-4(or153ts); xnIs360, FT1852: nmy-2(ne3409ts); xnIs360; him-8(e1489), FT1853: cyk-1(or596ts); xnIs360, FT1909: xnIs360; zuIs178 [Phis-72::his-72-gfp::his-72^3′UTR, unc-119(+)], FT1927: cyk-4(or749ts); xnIs360, FT1938: itSi922 [Pspd-2::gfp-spd-2, unc-119(+)]; xnIs360, FT1974: nop-1(it142); xnIs360; zuIs70, FT2000: xnSi50 [Pmex-5::zen-4-yfp::nos-2^3′UTR, unc-119(+)]; xnIs360, FT2051: nmy-2(ne3409ts); xnSi50; xnIs360, FT2088: ebp-2(wow47: ebp-2-gfp); xnIs360, FT2096: nmy-2(cp13[nmy-2-gfp + LoxP]); xnSi49; xnIs360, FT2107: zen-4(or153ts); xnSi49; xnIs360, FT2108: ect-2(ax751ts); xnIs360, FT2109: unc-119(ed3); xnIs101 [Ppie-1::gfp-Moesin::nos-2^3′UTR, unc-119(+)]; xnIs360, and WM186: avr-14(ad1302); ttTi5605; unc-119(ed3); avr-15(ad1051) glc-1(pk54); neEx15 [myo-2::RFP, myo-2::avr-15(+), unc-119(+)] (Shirayama et al., 2012).

Gene, transgene, and fusion protein nomenclature in the text and figures relates to the following alleles and transgenes: cyk-1 = cyk-1(or596ts), cyk-4 = cyk-4(or749ts),ect-2 = ect-2(ax751ts), nmy-2 = nmy-2(ne3409ts),nop-1= nop-1(it142),zen-4 = zen-4(or153ts), F-actin^PGC = xnIs101 [Ppie-1::gfp-Moesin::nos-2^3′UTR, unc-119(+)], Mem^PGC = xnIs360 [Pmex-5::mCherry-PH::nos-2^3′UTR, unc-119(+)], NMY-2-GFP = nmy-2(cp13[nmy-2::GFP + LoxP]), NMY-2-YFP^PGC = xnSi49 [Pmex-5::nmy-2-yfp::nos-2^3′UTR, unc-119(+)], and ZEN-4-YFP^PGC = xnSi50 [Pmex-5::zen-4-yfp::nos-2^3′UTR, unc-119(+)].

Pmex-5::nmy-2-yfp::nos-2^3′UTR was assembled from plasmid Pmex-5::hmr-1-gfp::nos-2^3′UTR, which was constructed using Multisite Gateway (Invitrogen) from vector pCFJ150 (Addgene #19329) (Frøkjær-Jensen et al., 2008), 5′ entry clone pJA252 (Pmex-5; Addgene #21512; Zeiser et al., 2011), middle entry clone pJN527 (hmr-1-gfp; Chihara and Nance, 2012), and 3′ entry clone pDC10 (nos-2 3′ UTR; Chihara and Nance, 2012). hmr-1-gfp was replaced with nmy-2 genomic sequence and yfp in a three-fragment Gibson Assembly reaction (Gibson et al., 2009).

Pmex-5::zen-4-yfp::nos-2^3′UTR was constructed from Pmex-5::nmy-2-yfp::nos-2^3′UTR, replacing nmy-2 with zen-4 genomic sequence via Gibson Assembly.

Ppie-1::gfp-Moesin::nos-2^3′UTR was constructed from vector pKS111-His (Ppie-1::gfp-Histone::nos-2^3′UTR; a gift from Kuppuswamy Subramaniam; D’Agostino et al., 2006). pKS111-His was digested with SpeI to remove the Histone H2B coding region. cDNA encoding amino acids 438–575 of Drosophila melanogaster Moesin (isoform D) was amplified by PCR using primers containing SpeI restrictions sites and ligated with the pKS111-His vector fragment.

Ppie-1::gfp-Moesin::nos-2^3′UTR was integrated through microparticle bombardment using a Bio-Rad Biolistic PDS-1000/HE gene gun outfitted with a Hepta adaptor (Praitis et al., 2001). DNA (1–2 µg) was precipitated onto 1.0-µm-diameter gold beads (1652263; Bio-Rad) and bombarded directly onto an unseeded 10-cm agar plate containing a uniform lawn of young adult unc-119(ed3) worms. Bombarded worms were transferred to ten 10-cm peptone-enriched agar plates seeded with E. coli strain NA22 and allowed to starve. Twelve non-Unc worms were singled from each plate containing transformants and were allowed to self-fertilize to identify homozygous integrants, which were scored for GFP expression.

Pmex-5::nmy-2-yfp::nos-2^3′UTR and Pmex-5::zen-4-yfp::nos-2^3′UTR were microinjected as described (Mello et al., 1991) into WM186 worms lacking the neEx15 extrachromosomal array at a concentration of 20 ng/µl together with 50 ng/ml each of pRF4 [rol-6(su1006)] (Mello et al., 1991), pCCM416 [Pmyo-2::avr-15] (a gift from Craig Mello; Shirayama et al., 2012), and pJL44 [Phsp16.48::MosTase::glh-2^3'UTR] (19333; Addgene; Frokjær-Jensen et al., 2008). A transmitting extrachromosomal line was isolated by its Rol phenotype and integrated using the heat-shock method (Frøkjær-Jensen et al., 2008; Shirayama et al., 2012). Briefly, several thousand transgenic worms were heat shocked at 35°C for 1.5 h and grown for two generations, and F3 embryos were plated onto NGM agar plates containing 2 ng/µl of ivermectin. After 3 d, surviving non-Unc non-Rol worms were picked from each plate and tested for homozygosity of the insertion and for YFP expression.

Embryos were dissected on poly-L-lysine‒coated slides, freeze cracked, and fixed in −20°C MeOH (20 min) and 3.7% formaldehyde in salts (50 mM Pipes, 25 mM Hepes, 10 mM EGTA, and 2 mM MgCl₂; 5 min) before washing in PBS-Tween and immunostaining directly on slides in PBS-Tween containing 1% IgG-free BSA (001–000-162; Jackson ImmunoResearch). The following primary antibodies were used: chicken anti-GFP 1:10 (GFP-1010; Aves Labs), rabbit anti-UNC-59 1:100 (Maddox et al., 2005; a gift from Karen Oegema), rabbit anti-ANI-1 1:100 (Maddox et al., 2005; a gift from Karen Oegema), and mouse anti-FRM-1 1:1,000 (Choi et al., 2011; a gift from Nam Jeong Cho). The following secondary antibodies were used: AlexaFluor 488 goat anti-chicken IgY (H+L) (A-11039; Thermo Fisher Scientific), Cy3 donkey anti-rabbit IgG (H+L) (711–165-152, lot 86701; Jackson ImmunoResearch), and Cy5 donkey anti-mouse IgG (H+L) (715–175-151, lot 76920; Jackson ImmunoResearch). After washing and incubating with 4′,6-diamidino-2-phenylindole, embryos were mounted in 90% glycerol containing 1,4-diazabicyclo[2.2.2]octane (D27802; Sigma-Aldrich) anti-fade agent.

Live embryos mounted on 4% agarose pads in M9 medium and fixed embryos attached directly to slides and mounted in 90% glycerol with 1,4-diazabicyclo[2.2.2]octane anti-fade agent were imaged either on a Zeiss AxioImager.A2 using a 40× Plan-Neofluar 1.3 NA or 63× Plan-Apochromat 1.4 NA oil-immersion objective, Axiocam 503 mono camera, Uniblitz shutter (Vincent Associates), and Zen software (Zeiss) or on a Leica TCS SP8 confocal microscope equipped with HyD detectors using a 63× 1.4 NA oil-immersion objective lens and LAS AF software. Imaging was performed at room temperature except for temperature-sensitive mutants, where temperatures at each time point are indicated. Images of fixed embryos were deconvolved using ZEN software and the Fast Iterative method. Brightness, contrast, and gamma level adjustments, as well as image resizing and cropping, were performed using Fiji (National Institutes of Health) and Photoshop (Adobe).

A Nikon Ti2-E microscope with CSU-W1 Spinning Disk, 100X CFI SR HP PlanApochromat 1.35 NA silicone immersion objective, and IXON Life 888 Dual EMCCD camera and NIS Elements software were used for imaging centrosomes and microtubule growth in PGCs. For centrosomes, 12-µm image stacks (1-µm step size) of Mem^PGC and SPD-2-GFP were captured every 5 min for ∼1 h, through the period of lobe formation. For determining the site and direction of microtubule growth, Mem^PGC and EBP-2-GFP were imaged continuously for 20 s in a single plane through the center of the PGCs. Microtubule growth was determined by following EBP-2-GFP puncta using the MTrackJ Fiji macro with manual annotation of tracks (Meijering et al., 2012).

Worms were dissected at the permissive temperature (18°C), and two-cell embryos collected over a ∼10-min period were mounted on 4% agar pads in M9. Control and experimental embryos were mounted in separate, nearby clusters on the same slide for every experiment. Slides were transferred to a ∼24°C metal plate in the microscope room at the indicated time and immediately mounted for microscopy on a Zeiss AxioImager.Z2 compound microscope. Temperature of the samples was maintained at the restrictive temperature (26°C ± 0.5°C) using room temperature control and an HLS-1p objective heater and thermistor probe (Cell MicroControls), which was monitored throughout the duration of the experiment. mCherry and differential interference contrast (DIC) image stacks were acquired immediately and again at the indicated time. The presence of a nucleus was scored using DIC images.

To test temperature-sensitive mutants for a role in PGC lobe formation (temperature-shift regime I), embryos were shifted from the permissive temperature to the restrictive temperature 6 h and 30 min after the two-cell stage, and the final image was taken at 8 h 28 min. Samples were discarded if control embryos mounted on the same slide failed in lobe formation.

To examine whether nmy-2 Class II mutants (temperature-shift regime II) had a defect in maintaining the connection between PGCs, embryos were shifted from the permissive temperature to the restrictive temperature 5 h and 30 min after the two-cell stage. The final image was taken at 6 h, a stage before PGCs form lobes in WT control embryos.

To test zen-4 and cyk-4 mutants for a role in P₄ division (temperature-shift regime III), embryos were shifted from the permissive temperature to the restrictive temperature 2 h after the two-cell collections, and the final image was captured at 3 h 50 min. Samples were discarded if control embryos mounted on the same slide failed in P₄ division.

To examine lobe formation in binucleate embryos (temperature-shift regime IV), zen-4 embryos were shifted from the permissive temperature to the restrictive temperature 2 h after the two-cell collections, time point 1 was captured at 3 h 50 min, and time point 2 was captured at 7 h. For experiments with higher time resolution, DIC and mCherry image stacks were captured every 5 min, beginning at time point 1 above. 2 × 2 pixel binning was used to reduce fluorescence exposure times.

To examine lobe formation in nop-1 mutant embryos, control (FT1258) and nop-1(it142) mutant (FT1974) two-cell embryos were dissected and kept at 25°C. The first time point was captured at 4 h 45 min after the two-cell stage and every subsequent 8 min until 5 h 33 min (t = 48 min after the first time point).

Worms were grown at 25°C, and two-cell embryos that were collected over a ∼10-min period were mounted on 4% agar pads. mCherry and GFP image stacks were acquired on a Zeiss AxioImager microscope 4 h and 7 min (t = 0 min) after initial dissection and again at the indicated times. To determine nuclear localization within the PGC, the center of the cell and the center of the nucleus were measured at the first two time points using a custom macro in ImageJ. Positions were plotted on a graph with the origin representing the center of the PGC.

Sample sizes and number of experiments pooled for data shown in the graphs are indicated directly in the figures and figure legends. Fisher’s exact test was used to compare mutant embryos with control embryos. Controls mounted on the same slide were performed for each mutant genotype. P values >0.05 were considered not significant; the P value of significant differences is indicated in the figure legends.

Fig. S1 (related to Fig. 1) shows NMY-2-YFP^PGC accumulating at lobe necks as they first begin to ingress. Fig. S2 (related to Fig. 4) shows that the cyk-4 and zen-4 temperature-sensitive mutants block division of the P₄ cell at the temperature used to test whether they are required for PGC lobe formation. Fig. S3 (related to Fig. 5) shows that asymmetric nuclear positioning occurs before lobe formation in zen-4 binucleate PGCs and that NMY-2-YFP^PGC localizes to the neck of lobes in zen-4 binucleate PGCs. Video 1 (related to Fig. 1) shows Mem^PGC during PGC lobe formation in a WT embryo. Video 2 (related to Fig. 1) shows a rotated image stack of NMY-2-YFP^PGC at a PGC lobe neck to illustrate that it forms a ring. Video 3 (related to Fig. 5) shows dynamics and tracks of the microtubule plus-end binding protein EBP-2-GFP to demonstrate where microtubule growth occurs in PGCs.

We thank Julie Canman (Columbia University, New York, NY), Nam Jeong Cho (Chungbuk National University, Cheongju, South Korea), Dan Dickinson (University of Texas, Austin, TX), Jessica Feldman (Stanford University, Stanford, CA), Bob Goldstein (University of North Carolina, Chapel Hill, NC), Amy Maddox (University of North Carolina, Chapel Hill, NC), Craig Mello (University of Massachusetts Medical School, Worcester, MA), Karen Oegema (University of California San Diego School of Medicine, San Diego, CA), Geraldine Seydoux (Johns Hopkins University School of Medicine, Baltimore, MD), and Kuppuswamy Subramaniam (Indian Institute of Technology, Kanpur, India) for reagents. Thanks to Lionel Christiaen, Jane Hubbard, Ruth Lehmann, and Niels Ringstad for comments on the manuscript and Michael Cammer for ImageJ help. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440).

Funding was provided by the National Institutes of Health grant R35GM118081 to J. Nance.

The authors declare no competing financial interests.

Author contributions: C. Maniscalco and J. Nance conceptualized the project. C. Maniscalco, A.E. Hall, and J. Nance performed experiments and analyzed and interpreted the data. C. Maniscalco and J. Nance wrote the original draft, and A.E. Hall and J. Nance wrote the revised manuscript.