The seip-1(tm4221) deletion allele [seip-1(-) thereafter] was initially annotated as lethal by the National Bioresource Project (Mitani, 2017). Surprisingly, upon outcrossing with wild type worms, we discovered that seip-1(-) worms were fertile (median = 51 live progenies per animal) (column ii, Figure 1A). The significant reduction in viable progeny was due to a large number of eggs that failed to hatch after being laid. In a separate attempt to study the lipid accumulation of seip-1(-) worms, we cultured them in the presence of the vital dye BODIPY on standard nematode growth media (NGM) plates. Although the total number of eggs laid by wild type and seip-1(-) worms were comparable (Supplementary Figure S1B), BODIPY could not penetrate wild type eggs, while ∼60% of seip-1(-) eggs were stained (Figures 1B–D). The strong correlation between BODIPY staining and embryonic arrest suggested that the latter might be caused by a structural deficiency of the eggshell (Supplementary Figure S1A). Consequently, the penetration of exogenous material might interfere with embryonic development. Because some seip-1(-) embryos develop and hatch as L1 larvae, we therefore conclude that SEIP-1 acts in parallel of additional proteins to ensure embryonic viability. Interestingly, seip-1(-) eggs laid by relatively young adults were more prone to BODIPY staining (Supplementary Figure S1C). Therefore, it is plausible that the SEIP-1-independent mechanism is triggered at least 1 day after the initiation of egg laying.

The structural integrity of the C. elegans eggshell is critical for embryonic development, because defects in the eggshell can cause mitotic and polarity defects (Olson et al., 2012). Upon fertilization of the oocyte in the spermatheca, the C. elegans eggshell is formed by the hierarchical establishment of four discrete protein- or lipid-rich layers. Starting with the outermost vitellin layer (VL), the chitin layer (CL), the chondroitin proteoglycan layer (CPG), and the permeability barrier were sequentially formed toward the embryonic plasma membrane (Supplementary Figure S1A). We noted that the embryonic arrest phenotype of seip-1(-) worms was similar to a class of mutants that fail to form the permeability barrier properly. Accordingly, a large number of perm-1 and dgtr-1 embryos could be stained when they were laid on BODIPY-containing NGM plates (Supplementary Figure S1D). The permeability barrier separates the perivitelline space and the peri-embryonic space (PES) (Supplementary Figure S1A). In wild type worms, the permeability barrier retains the chondroitin proteoglycan CPG-2 in the perivitelline space and exclude it from the peri-embryonic space (Figure 1E). In contrast, we found abnormal accumulation of CPG-2 near the plasma membrane in a fraction of seip-1(-) embryos (Figure 1F), similar to perm-1 and dgtr-1 deficient embryos (Olson et al., 2012). Live imaging of wild type embryos in utero indicated the emergence of PES upon the cortical ruffling of plasma membrane (Supplementary Video S1) (Green et al., 2008). PES expands during the pseudocleavage (Goldstein et al., 1993) and is likely finalized following the first mitotic cleavage, which results in two-cell-stage embryos (Supplementary Video S1). In contrast, PES failed to emerge even after the first mitotic cleavage of some seip-1(-) embryos, whereas the exocytosis of CPG-2 was unperturbed (Supplementary Video S2). Our results are consistent with a recent report (Bai et al., 2020), and implied that SEIP-1 is necessary for the proper formation of the permeability barrier, but not other layers of the eggshell.

The Alanine 212 to Proline (A212P) mutation in human seipin causes congenital generalized lipodystrophy type 2 disease (CGL2) (Magré et al., 2001). Based on primary sequence alignment, we noted that A185 of SEIP-1 is orthologous to A212 of human seipin (Supplementary Figure S2A). To assess if the conserved alanine is important for SEIP-1 function, we generated single-copy transgenes that expressed either SEIP-1 (wild-type)::GFP (hjSi189) or SEIP-1 (A185P)::GFP (hjSi541), driven by the ubiquitous dpy-30 promoter. Wild type SEIP-1::GFP supported the formation of the embryonic permeability barrier (Figure 1H) and restored the fertility of seip-1(-) mutants (Figure 1A, column iv). Correspondingly, no BODIPY-stained embryos were observed (Figure 1B, column iv). In contrast, SEIP-1 (A185P) failed to suppress the permeability barrier defects of seip-1(-) mutant worms (Figures 1A,I, column v; Figure 1B, column v). To complement our single-copy transgene strategy, we engineered a knock-in seip-1(A185P) allele (hj158) using CRISPR. Based on our homology directed repair (HDR) methodology, an “orphan” loxP sequence was inserted between the stop codon and 3’-UTR of seip-1(A185P) (Supplementary Figure S2B). We therefore constructed a control wild-type seip-1 allele (hj156) with a loxP site inserted at the same position as in hj158 (Supplementary Figure S2B). The fertility of seip-1(hj156) animals was comparable to that of wild type. However, the fertility of seip-1(hj158) (i.e., A185P) was reduced to a level similar to seip-1(-) animals, with a corresponding increase in BODIPY-positive embryos (Supplementary Figures S2C,D). Our observations were consistent with those made using an independently generated allele that encoded SEIP-1 (A185P) (Bai et al., 2020). We conclude that the A185P mutation disrupts SEIP-1 function, similar to the effect of A212P to human seipin.

Endogenous SEIP-1 is expressed in both the intestine and the germline (Cao et al., 2019; Bai et al., 2020). In C. elegans, the proper embryonic development is dependent on the supply of intestine-derived yolk proteins (Grant and Hirsh, 1999; Ezcurra et al., 2018). To this end, we sought to determine the site of SEIP-1 action for proper embryogenesis. Germline-specific expression of SEIP-1 with the sun-1 promoter (hjSi502) fully rescued the defect of seip-1(-) embryos (Figure 1B, column iii; Figure 1G), thereby restoring the fertility of seip-1(-) worms to the wild-type level (Figure 1A, column iii). In contrast, intestine-specific expression of SEIP-1 with the vha-6 promoter (hjSi3) did not rescue the embryonic defect (Figure 1B, column vi; Figure 1J, column vi) or the fertility of seip-1(-) mutants (Figure 1A, column vi). Similarly, ubiquitous but not intestine-specific expression of human seipin rescued the defects of seip-1(-) embryos (Figure 1A, columns vii and viii; Figure 1B, columns vii and viii). Taken together, a conserved function of SEIP-1/seipin is required in the germline to support embryonic development.

We previously demonstrated that SEIP-1 regulates intestinal LD expansion (Cao et al., 2019), using the mRuby-tagged C. elegans diacylglycerol O-acyltransferase 2 ortholog, DGAT-2 (Xu et al., 2012), as the LD marker. Prompted by the germline-specific function of SEIP-1, we examined the morphology of germline LDs in seip-1(-) worms. Because DGAT-2 is primarily found in intestinal cells, we needed an alternative marker to visualize germline LDs. To this end, we used the sole C. elegans perilipin ortholog, MDT-28/PLIN-1 (for simplicity, referred to as PLIN-1 in subsequent text), which is ubiquitously expressed (Na et al., 2015). We engineered a knock-in allele (hj178) for which the two long isoforms of PLIN-1 (PLIN-1a/c) were tagged with GFP at their C-termini (Figure 2B). Consistent with our previous finding, intestinal PLIN-1-labeled LDs in seip-1(-) worms were smaller than those in wild type worms (Figures 2C,D, inset i). Surprisingly, in contrast to the more uniformly sized LDs in the wild-type germline (Figure 2C), we found abnormally large or small LDs in both oocytes (inset ii) and embryos (inset iii) of seip-1(-) mutants (Figure 2D). As a result, the size range of LDs in seip-1(-) germline was larger than that in wild type (Figure 2E). We made similar observations in seip-1(A185P) loss of function mutants (Supplementary Figures S2E–G). Such aberrant LD morphology in the SEIP-1 deficient germline is reminiscent of that in yeast and cell line models when seipin orthologs are depleted (Szymanski et al., 2007; Fei et al., 2008; Salo et al., 2016; Wang et al., 2016; Chung et al., 2019). Our results indicate that seipin deficiency causes distinct LD defects in proliferating versus differentiated cells (such as C. elegans intestinal cells).

SEIP-1 is enriched at a subdomain of the endoplasmic reticulum (ER), termed the peri-LD cage (Cao et al., 2019), in the C. elegans intestine. To investigate the localization of SEIP-1 in the germline, we first examined SEIP-1(WT)::GFP in adult worms that lacked the endogenous SEIP-1 protein. In both oocytes and embryos, SEIP-1(WT)::GFP was targeted to “ring” or “cage”-like structures (Supplementary Figure S3A). Similar observations were made in seip-1(-) worms that expressed human seipin::GFP ubiquitously (Supplementary Figure S3C). In comparison, SEIP-1 (A185P)::GFP was rarely found in equivalent structures (Supplementary Figure S3B). Such localization defect might be linked to its reduced oligomeric state, as reported for human seipin (A212P) (Binns et al., 2010; Sui et al., 2018; Yan et al., 2018). Next, we examined the localization of a SEIP-1::tagRFP fusion protein, relative to the PLIN-1::GFP LD marker. We reasoned that the SEIP-1::tagRFP fusion protein, expressed from a single-copy transgene (hjSi434), was functional since it could rescue both the fertility and the permeability barrier defect of seip-1(-) worms (Supplementary Figures S3D,E). SEIP-1::tagRFP was enriched in tubular structures around a subset of PLIN-1 positive LDs in the germline (Figure 2F). Therefore, SEIP-1 appeared to mark a subset of LDs in the germline, similar to our previous observations in the intestine. Additional subcellular compartments, such as cortical granules (Bembenek et al., 2007; Olson et al., 2012) and yolk particles (Sharrock et al., 1990) are known to contribute to early embryogenesis and their reported sizes are similar to that of LDs. Therefore, we asked if SEIP-1 could be found in the proximity of these structures. The cortical granules are exocytosed by canonical anterograde trafficking (Bembenek et al., 2007). They can be marked with COPII components. We focused on SEC-16A.1, which is one of the 2 C. elegans orthologs of mammalian and yeast SEC16 that is found at ER-exit sites (ERES) and COPII vesicles (Watson et al., 2006). Accordingly, we constructed a knock-in allele that expressed SEC-16A.1 with a C-terminal GFP tag (Supplementary Figure S3F). For the visualization of yolk particles, we used a published knock-in allele that expressed VIT-2::GFP (Perez et al., 2017). Overall, we did not observe overt colocalization of SEIP-1 and SEC-16A.1 or VIT-2 in oocytes or embryos (Supplementary Figures S3G,H). Our results imply that LDs associated with SEIP-1 (+) cages are distinct from other vesicular structures such as cortical granules and yolk particles.

We next sought to understand how SEIP-1 (+) LDs might regulate the formation of the permeability barrier. Based on live imaging, we consistently observed LDs and SEIP-1-labeled peri-LD cages near the cell cortex of +1 embryos, but not in −1 oocytes or +2 embryos (Figures 2A,C,F,G, Supplementary Video S3). To quantify this phenomenon, we compared the localization of SEIP-1 relative to the plasma membrane (Figure 2G) by measuring the shortest distance from each peri-LD cage to the plasma membrane in cells at either −1, +1, or +2 position. The median distance was reduced from ∼2 μm in -1 oocytes and ∼3.8 μm in +2 embryos to ∼1 μm in +1 embryos (Figure 2H). In sum, our data is consistent with a model that upon fertilization, SEIP-1 (+) LDs are recruited to the cortical region of the zygote to support the formation of permeability barrier. Such distribution pattern of SEIP-1 was not observed in the differentiated intestine (Cao et al., 2019), again highlighting the distinct function of SEIP-1 positive structures in proliferating versus differentiated cells.

We hypothesized that the redistribution of SEIP-1 (+) LDs near the plasma membrane supports the construction of the permeability barrier in newly fertilized embryos. To determine if such redistribution is linked to the catabolism or anabolism of SEIP-1 (+) LDs, we developed a photoconversion strategy to label preexisting SEIP-1 in the oocytes. From the endogenous seip-1 locus, we expressed the fusion of the monomeric photoconvertible fluorescent protein mKikGR (Habuchi et al., 2008) with SEIP-1 (SEIP-1::mKikGR). In the same worms, we also expressed the ubiquitous LD marker, PLIN-1::GFP. When one-day-old adult worms were exposed to 405 nm fluorescence, all green SEIP-1::mKikGR fusion proteins in the germline were converted to red (pseudocolored magenta) (Figure 3A). These worms were subsequently examined after a 1-h lag. This lag was necessitated when the following events were taken into consideration: ovulation cycle (∼15 min) (Huelgas-Morales and Greenstein, 2018), fertilization in the spermatheca (∼5 min), and development from fertilization to the first mitotic division (40 min) (Stein and Golden, 2018). Therefore, by the time of imaging, the formation of the permeability barrier and the presumptive turnover of oocyte-derived SEIP-1::mKikGR should be complete in +1 embryos. Consistent with our previous study with single-copy transgenes (Cao et al., 2019), endogenous SEIP-1::mKikGR was also targeted to peri-LD cages in the intestine (Figure 3B, inset i). In the germline, we found fewer pre-existing, photoconverted peri-LD cages in +1 embryos than in -1 oocytes (Figure 3B, inset ii and iii). Such reduction correlated with a decrease in the photoconverted SEIP-1::mKikGR to PLIN-1::GFP fluorescence intensity ratio (Figures 3A,C). In this case, PLIN-1::GFP fluorescence was used for normalization across different +1 embryos. Altogether, we propose that SEIP-1 (+) LDs mobilization to the plasma membrane and their subsequent disappearance are integral steps of permeability barrier formation.

A genetic pathway, consisting of FASN-1, POD-2, PERM-1, and DGTR-1, has previously been shown to control permeability barrier formation (Olson et al., 2012; Stein and Golden, 2018). How does SEIP-1 fit into such pathway? Interestingly, the terminal enzyme DGTR-1 is a germline-specific paralog of DGAT-2, which contributes to neutral lipid synthesis in the intestine. We propose that the lipid cargoes, synthesized by DGTR-1, for constructing the permeability barrier are contained in germline LDs. Notably, FASN-1, POD-2, PERM-1 or DGTR-1 deficiency causes embryonic lethality at high penetrance (Tagawa et al., 2001; Carvalho et al., 2011), unlike SEIP-1 deficiency (Figures 1A,B). It is plausible that SEIP-1 acts downstream of the cargo synthesis enzymes by specifying a subset of LDs that are destined to be utilized for constructing the permeability barrier.

We next sought to understand the molecular basis of incomplete penetrance and high variability of the embryonic defect of seip-1(-) worms (Figures 1A–F, Supplementary Figure S1C). FAT-3, a polyunsaturated fatty acyl-CoA (PUFA) desaturase, is required for targeting SEIP-1 to peri-LD cages in the intestine (Cao et al., 2019) and oocytes (Supplementary Figures S6A,B). Thus, FAT-3 products such as gamma linolenic acid (GLA, C18:3n6) and their derivatives appeared to be broadly required for the recruitment of SEIP-1 to peri-LD cages. Intriguingly, dietary GLA supplementation was reported to significantly rescue the permeability barrier defect of seip-1(-) worms (Bai et al., 2020), suggesting that GLA also boosts the function of a SEIP-1-independent pathway. To this end, we asked if FAT-3 deficiency could modify the phenotypes of seip-1(-) worms. Indeed, FAT-3 deficiency reduced the fertility of otherwise wild type worms or seip-1(-) worms (Figure 4C). Accordingly, seip-1(-); fat-3(-) worms displayed a highly penetrant embryonic arrest phenotype, similar to dgtr-1 or perm-1 deficient worms. We conclude that FAT-3 products and their derivatives are important for both SEIP-1-dependent and SEIP-1-independent pathways that are responsible for permeability barrier formation. Both pathways are blocked in seip-1(-); fat-3(-) worms.

Next, we expanded our analysis by comprehensively interrogating genes that regulate fatty acid (FA) synthesis and desaturation (Supplementary Figure S4A). The percentage of BODIPY-stained embryos (i.e., embryos with defective permeability barrier) was compared after individual genes were depleted by RNAi-based knockdown (KD). In seip-1(-) background, ubiquitous depletion of either fat-3 or fat-4 increased the number of embryos with permeability barrier defects (Supplementary Figure S4B). Accordingly, depleting enzymes upstream of FAT-3/FAT-4 (i.e., FAT-2, FAT-6, ELO-1, and ELO-2) enhanced the embryonic arrest of seip-1(-) mutants. Similar effect was not observed when FAT-1 or FAT-5 was depleted (Supplementary Figure S4B).

To determine the tissues from which fatty acyl-CoA desaturases modulated the phenotypes of seip-1(-) worms, we repeated our experiments with strains that restricted RNAi to the intestine or germline (Melo and Ruvkun, 2012; Cao et al., 2019; Zou et al., 2019). In seip-1(-) mutant background, intestine-specific depletion, but not germline-specific depletion, of FAT-2, FAT-3, or FAT-4 enhanced the embryonic arrest phenotype (Figures 4D, Supplementary Figure S4D). Therefore, the synthesis of PUFAs and their derivatives (such as phospholipids and lipoproteins) in the intestine appears to regulate permeability barrier formation by germline factors. We also investigated the dependence on fatty acyl-CoA desaturases for embryonic development in otherwise wild-type worms. By analyzing the distribution of a yolk protein marker, VIT-2::GFP, we found abnormal accumulation of GFP signals in the pseudocoelom of fat-3 knockdown worms (Supplementary Figures S6C–E), suggesting a partial block of yolk protein transport into the germline. Interestingly, embryonic arrest was observed only when FAT-2 was depleted ubiquitously in wild type worms (Supplementary Figure S4C). However, neither intestine- nor germline-specific knockdown of fat-2 recapitulated phenotypes observed when fat-2 was knocked down ubiquitously. Our results imply that FAT-2 products made in one tissue can diffuse to other tissues to compensate for local FAT-2 deficiency. Based on the differential requirement of fatty acyl-CoA desaturases in wild type and seip-1(-) worms, different PUFAs and their derivatives appear to contribute to germline and embryonic development in multiple ways, including but not limited to the establishment of ER subdomains and intestine-to-germline lipid transport. The distribution of these PUFA derivatives in specific membranes and vesicles will require further chemical analysis.

We inadvertently discovered that PLIN-1 inhibited an alternative mechanism that supported permeability barrier formation in seip-1(-) worms. In an attempt to tag all isoforms of PLIN-1 with the mRuby red fluorescent protein, we inserted the mRuby coding sequence to the 5’ end of the plin-1 gene by CRISPR (Supplementary Figure S5A). When we introduced the resultant allele, hj249, into seip-1(-) worms, we were surprised that the fertility of plin-1(hj249); seip-1(-) worms was similar to that of wild type worms (Supplementary Figure S5B). Accordingly, the percentage of BODIPY-positive embryos was significantly reduced (Supplementary Figure S5C). These phenotypes were not observed in plin-1(hj178) when PLIN-1A and PLIN-1C isoforms were fused at their C-terminus to GFP (Supplementary Figures S5B,C). Such discrepancy led us to speculate that the mRuby fusion to the N-terminus of PLIN-1 compromised its function, and that hj249 was a hypomorphic plin-1 allele that rescued SEIP-1 deficient embryos. To prove our hypothesis, we analyzed worms that carried the plin-1(tm1704) deletion allele (Xie et al., 2019) (referred to as plin-1(-) hereafter). The plin-1(-) worms showed normal permeability barrier formation and fertility (Figures 4A,C). Similar to plin-1(hj249); seip-1(-) worms, the fertility of plin-1(-); seip-1(-) worms was comparable to that of wild type worms, and almost all embryos had an intact permeability barrier as indicated by the lack of BODIPY staining (Figures 4B,C,G). In addition, the plin-1(-) allele rescued the embryonic defects of seip-1(A185P) mutants (Supplementary Figures S2C,D), suggesting that the suppression by plin-1(-) was not restricted to specific seip-1 loss of function alleles.

Next, we used RNAi to knockdown plin-1 ubiquitously, or tissue-specifically in the intestine or germline. We found that ubiquitous or germline-specific RNAi against plin-1 suppressed seip-1(-) (Figure 4D). However, intestine-specific RNAi against plin-1 did not show similar suppression (Figure 4D). As a complementary approach, we expressed plin-1 in the germline of plin-1(-); seip-1(-) worms, with a sun-1 promoter driven single-copy transgene. These worms showed embryonic defects similar to seip-1(-) worms (Supplementary Figures S5B,C). Thus, our results indicate that the loss of PLIN-1 function in the germline is sufficient to bypass the requirement on SEIP-1 for permeability barrier formation. Interestingly, ubiquitous knockdown of fat-2 or fat-3 and intestine-specific knockdown of fat-3 by RNAi reversed the suppression of seip-1(-) associated embryonic defects by plin-1(-) (Figure 4E, Supplementary Figure S4E). Therefore, plin-1(-); seip-1(-) worms appeared to rely on FAT-2 or FAT-3, acting in distinct tissues, to support permeability barrier formation. Our results support the notion that local and diffusible PUFAs and their derivatives are required for embryonic development of C. elegans.

Next, we sought to identify proteins that acted redundantly with SEIP-1 to ensure embryonic development. To this end, we inactivated genes encoding known or putative LD-associated proteins by RNAi. We found that ubiquitous or germline-specific rab-18 knockdown yielded BODIPY-positive, permeability barrier defective plin-1(-); seip-1(-) embryos (Figure 4E). Intestine-specific RNAi against rab-18 had no effect (Figure 4E). Similar observations were made when dgtr-1, a gene known to be essential for permeability barrier formation (Olson et al., 2012), was knocked down in plin-1(-); seip-1(-) embryos (Figure 4E). Our results suggest that RAB-18 acts in the germline, in the absence of SEIP-1, to support permeability barrier formation in a DGTR-1-dependent manner. It should be noted that loss of dgtr-1 function caused highly penetrant embryonic arrest of otherwise wild type worms (Carvalho et al., 2011). In contrast, worms harboring the rab-18 (ok2020) [referred to as rab-18(-) hereafter] loss of function allele did not arrest as embryos and produced comparable number of live progenies as wild type worms (Figure 4F). However, loss of rab-18 significantly increased the percentage of arrested embryos from plin-1(-); seip-1(-) worms and reduced the number of live progenies accordingly (Figures 4F,G). Using CRISPR, we also generated a mutant rab-18 allele that encoded the S25N mutation (Supplementary Figure S7A). The S25N mutation is analogous to S22N of human Rab18, which “locks” RAB-18 in a constitutively GDP-bound form. Based on BODIPY staining, we found that the seip-1(-); plin-1(-); rab-18 (S25N) mutants shared the same permeability barrier defect as plin-1(-); seip-1(-); rab-18(-) mutants (Supplementary Figure S7B). Our results imply that active RAB-18 is specifically required in plin-1(-); seip-1(-) worms to support embryonic development.

To elucidate the mechanism by which RAB-18 supported permeability barrier formation in the absence of SEIP-1 and PLIN-1, we investigated the localization of GFP::RAB-18 fusion protein, expressed from its endogenous locus (Supplementary Figure S7A). It did not localize to distinct structures in wild type and plin-1(-) worms (Figures 4H,I). In seip-1(-) worms, GFP::RAB-18 appeared in numerous diffraction-limited puncta throughout the cytoplasm of newly fertilized (+1, +2) embryos (Figure 4J). Finally, in plin-1(-); seip-1(-) embryos, GFP::RAB-18 was found on the LD surface as confirmed by the LDP-1::mRuby marker (Na et al., 2015) (Figures 4K,L). Interestingly, Rab18 supports lipid droplet growth only in the absence of seipin in human A431 cells (Salo et al., 2019). Furthermore, ADRP/Perilipin 2 and Rab18 appeared to compete for LD surface association in cultured mammalian cells (Ozeki et al., 2005). Therefore, it was plausible that the loss of PLIN-1 in seip-1(-) embryos similarly permitted the association of RAB-18 with LDs that contained the cargos for permeability barrier construction. Taken together, our results support the notion that RAB-18 positive LDs assume an identity that is similar to SEIP-1 (+) LDs in otherwise wild type embryos. The association of RAB-18 with LDs is modulated by other resident LD proteins, such as PLIN-1.

Because perilipin orthologs are best known for their role in regulating lipolysis (Supplementary Figure S7C), we asked if basal lipolysis contributed to the high percentage survival of plin-1(-); seip-1(-) embryos. Using multiple independent deletion alleles (Supplementary Figure S7A), we found that loss of ATGL-1/ATGL, but not LID-1/CGI-58 or HOSL-1/HSL, partially reduced the survival of plin-1(-); seip-1(-) embryos (Supplementary Figures S7D,E). However, the percentage of arrested plin-1(-); seip-1(-); atgl-1(-) embryos was lower than that of plin-1(-); seip-1(-); rab-18(-) embryos. We also noted that ATGL-1 was already present on lipid droplets in plin-1(-) single mutants (Supplementary Figures S7F,G), whereas RAB-18 was present on lipid droplets specifically in plin-1(-); seip-1(-) mutants (Figure 4K). Taken together, we conclude that ATGL-1 and RAB-18 may contribute separately to permeability barrier formation, and that RAB-18 is more critically required by plin-1(-); seip-1(-) mutants.

Bristol N2 was used as the wild-type C. elegans strain. All animals were maintained and investigated at 20°C. The following alleles and transgenes were obtained from the Caenorhabditis Genetics Center (CGC): LG III, unc-119 (ed3), rab-18 (ok2020), itIs38 [(pAA1)pie-1p::GFP::PH (PLC1delta1) + unc-119(+)]; LG IV, fat-3 (ok1126); LG V, sid-1 (qt78), ltIs44 [(pAA173)pie-1p::mCherry::PH (PLC1delta1) + unc-119 (+)]; LG X, vit-2 (crg9070). seip-1 (tm4221) V and plin-1 (tm1704) I alleles were obtained from Dr. Shohei Mitani (National Bioresource Project for the nematode). rde-1 (mkc36) V and mkcSi13 [sun-1p::rde-1::sun-1 3′UTR] II are gifts from Dr. Di Chen (Model Animal Research Center, Nanjing University). The following single-copy transgenes were used: hjSi3 [vha-6p::seip-1 cDNA::GFP_TEV_3x- FLAG::let-858 3′UTR] II, hjSi189[dpy-30p::seip-1 cDNA:: GFP_TEV_3xFLAG::tbb-2 3′UTR] II, hjSi206 [vha-6p::human seipin isoform 2 cDNA (codon optimized)::GFP_TEV_3xFLAG::let-858 3′UTR] II, hjSi223[dpy-30p::human seipin isoform 2 cDNA (codon-optimized)::GFP_TEV_3xFLAG::tbb-2 3′UTR] II, hjSi434[dpy-30p::seip-1 cDNA::tagRFP::tbb-2 3′UTR] II, hjSi494[vha-6p::sid-1 cDNA::dhs-28 3′UTR] I, hjSi502[sun-1p::seip-1 cDNA::GFP_TEV_3xFLAG::sun-1 3′UTR] II, hjSi541[dyp-30p::seip-1(A185P) cDNA::GFP_TEV_3xFLAG::tbb-2 3′UTR] II, hjSi552[sun-1p::plin-1 gDNA::GFP_TEV_3xFLAG::sun-1 3′UTR] II. The following CRISPR/Cas9-generated alleles were used: hj140[seip-1::GFP_TEV_3xFLAG] V, hj158[seip-1(A185P)::stop codon_loxP_seip-1 3'-UTR] V, hj178[plin-1a/c::GFP_TEV_3xFLAG] I, hj249[mRuby_TEV_3xFLAG::plin-1] I, hj256[sec-16A.1::GFP_TEV_3xFLAG] III, hj289[ldp-1::mRuby_TEV_3xHA] V, hj323[seip-1::mKikGR_SEC_3xFLAG] V, hj340[cpg-2 signal peptide::mCherry_TEV_3xFLAG::cpg-2 w/o signal peptide] III, hj345[GFP_TEV_3xFLAG::atgl-1] III, hj347[GFP_TEV_3xFLAG::rab-18b] III, atgl-1 (hj349, hj352) III, lid-1 (hj355, hj358) I, hosl-1 (hj360, hj361) X, hj371[GFP_TEV_3xFLAG::rab-18b(S22N)] III. The sgRNA sequences for CRISPR are listed in Supplementary Table S2.

RNA interference (RNAi) was performed by on-plate feeding according to published methods (Neve et al., 2020). The targeting sequence in each RNAi vector was either taken from the Ahringer library or constructed with primers detailed in Supplementary Table S1. The plasmids were transformed into OP50 [rnc14::DTn10 laczgA::T7pol camFRT] (Neve et al., 2020). Fresh overnight cultures of OP50 RNAi clones were seeded on NGM plates with 0.4 mM IPTG and 100 μg/ml Ampicillin. The seeded plates were stored in the dark and incubated at room temperature for 1 day prior to experiments. The intestine-specific and germline-specific knockdown was performed as previously described (Cao et al., 2019; Zou et al., 2019). In brief, the single-copy intestine-specific (hjSi494) or germline-specific (mkcSi13) transgene is used to rescue sid-1(qt78) or rde-1(mkc36) mutant animals, respectively. The seip-1(tm4221) and/or plin-1(tm1704) allele were introduced into hjSi494; sid-1 (qt78) and mkcSi13; rde-1(mkc36) by genetic crosses. Further details of the RNAi clones can be found in Supplementary Table S1.

For all fluorescence imaging, one-day-old adult animals were examined by a spinning disk confocal microscope (AxioObeserver Z1, Carl Zeiss) with a ×5 (numerical aperture (NA) 0.16) or ×63 (numerical aperture (NA) 1.4 oil) Alpha-Plan-Apochromat objective. Image stacks were acquired with a Neo sCMOS camera (Andor) and a piezo Z stage controlled by the iQ3 software (Andor). For GFP, a 488 nm laser was used for excitation and emitted signals were collected by a 500–550-nm filter. For mCherry, mRuby, tagRFP, and photo-converted mKikGR, a 561 nm laser was used for excitation and emitted signals were collected by a 580.5–653.5-nm filter. Optical sections of images at 0.5 μm intervals were exported to Imaris 8 (Bitplane) for processing and 3D reconstruction.

For imaging the germline and embryos of live C. elegans, fresh 8% agarose pads were prepared on top of microscope glass slides. 2–3 μl of 1.25% (w/v) polystyrene microspheres (Polybead Microspheres 0.05μm, Polysciences) and 0.2 mM levamisole (Sigma) in 1 × PBS were dropped at the center of the agarose pads for immobilization of adult animals. For imaging the permeability barrier, on 22 mm × 22 mm cover slips, embryos were dissected from the uterus of one-day-old adults (24 h after the mid-L4 larval stage) in 0.8 × egg buffer with 0.2 mM levamisole in spherical micro-chambers bordered by vaseline. The microscope glass slides were subsequently applied with care to avoid the compression of the embryos.

For photoconversion of mKikGR, individual plates of one-day-old seip-1(hj323) adults (24 h after the mid-L4 larval stage) were exposed to 400–440 nm fluorescence filtered by SZX2-FBV (Olympus) for 30 min on a stereomicroscope (SZX16, Olympus) equipped with a LED light source (EXFO). Worms were allowed to rest for 1 h at 20°C so that -1 oocytes were ovulated and fertilized prior to confocal microscopy.

The diameter of LDs present in −1 oocyte and +1 embryos was manually fitted using the spot function in Imaris. The shortest distance between peri-LD cages and plasma membrane was computed by the spot and surface functions of Imaris XT.

Animals carrying the seip-1(tm4221) allele were backcrossed four times with wild-type males right before measurement. L4 larval P0 animals of each genotype were singled to individual plates and transferred to new plates every 24–48 h until egg-laying ceased. Hatched F1 animals were monitored and counted 24–48 h after the removal of the P0.

Four one-day-old adults (24 h after the mid-L4 larval stage) per replicate per strain were transferred to a new plate and allowed to lay eggs at 20°C. After 3 hours, adults were removed and the plate was stained with 500 μl 1 mM BODIPY in 1 × PBS for 3 h. The percentage of BODIPY-positive embryos was calculated using a fluorescence stereomicroscope. For comparison upon knockdown of different genes, the setup was similar as described above except that the age of adults varied in order to ensure both the efficiency and specificity of the knockdown. For constitutive knockdown, L4 larval P0 animals were plated and one-day-old F1 animals were used for scoring the percentage of BODIPY-positive embryos. For intestine-specific knockdown, L4 larval animals were plated and scored 48 h later. For germline-specific knockdown, L4 larval animals were plated and scored 24 h later.

Tagging of the endogenous genes was performed as described previously (Dickinson et al., 2015). In brief, single guide RNAs (sgRNAs, detailed in Supplementary Table S2) were designed using CHOPCHOP (https://chopchop.rc.fas.harvard.edu/) and cloned into pDD162 (Addgene). The sgRNA/Cas9 vectors were injected into adult C. elegans along with the corresponding repair templates and co-injection markers. The transgenic P0s were singled and their progeny were selected with 5 mg/ml hygromycin B (InvivoGen). Viable F2s were further screened using a fluorescence stereomicroscope and PCR-based genotyping to verify knock-in alleles. For knocking out a target gene, 2-4 sgRNAs were injected along with co-injection markers. F1s from the transgenic P0s were singled and genotyped for deletion. For each knock-in/knock-out strain, at least two independent alleles were acquired and examined for consistency. All constructed alleles were backcrossed two times with wild-type males before further characterization.