Forced lipophagy reveals that lipid droplets are required for early embryonic development in mouse

Macroautophagy (hereafter referred to as autophagy) is an evolutionarily conserved system of ‘self-eating’, in which cytoplasmic content is engulfed by the autophagosome and delivered to the lysosome for degradation (Mizushima, 2007). Although autophagy was originally characterized as a non-selective bulk degradation pathway, recent studies revealed that autophagy plays crucial roles in selective degradation of cellular components, including organelles (Okamoto, 2014; Stolz et al., 2014). Autophagy adaptors, which act as cargo receptors for selective degradation, play important roles in determining selectivity for target organelles. The most well-characterized adaptor is p62 (also known as Sqstm1), which contains an LC3-recognition sequence (LRS)/LC3-interacting region (LIR), a ubiquitin-associated (UBA) domain, and a Phox and Bem1p (PB1) domain, which are responsible for dimerization and multimerization of the protein (Katsuragi et al., 2015). To activate selective autophagy, surface proteins on damaged organelles (e.g. mitochondria) or microbes are first ubiquitylated by ubiquitin ligase. Then, autophagy adaptors, such as p62, are recruited onto the damaged organelle via interactions with the ubiquitylated proteins. The adaptors on the damaged organelle are recognized by LC3 (Map1lc3b), and the target is then engulfed by the autophagosome (Anding and Baehrecke, 2017; Stolz et al., 2014). Therefore, the crucial steps in selective autophagy are recruitment of an adaptor to the substrate and the interaction of the adaptor with LC3.

Lipid droplets (LDs) are unique organelles that store lipids essential for cellular energy (metabolism) and membrane production (Thiam et al., 2013; Walther and Farese, 2012). LDs consist of a neutral lipid core, mainly made up of triglyceride (TG) and cholesterol esters, and are coated by a phospholipid monolayer and various proteins such as the perilipins (Plins) [Plin1, ADRP (Plin2; also known as adipophilin), TIP47 (Plin3), Plin4 (also known as S3-12) and Plin5 (also known as OXPAT)]. The Plins share a conserved sequence, the PAT domain, primarily at the N terminus, which is essential for the LD interaction (Bickel et al., 2009). In mammals, several Plins are widely expressed (e.g. ADRP and TIP47), whereas others are expressed in a tissue-specific manner (e.g. Plin1 in adipocytes and steroidogenic cells) (Bickel et al., 2009). LDs are present in all tissues, and serve various functions in addition to lipid storage (Fujimoto and Parton, 2011).

Lysosomes contain both proteases and acidic lipases. Indeed, LDs are also targets of a form of selective autophagy termed lipophagy. During this process in mammals, yeast and Caenorhabditis elegans, LDs are transported to the lysosomes (O'Rourke and Ruvkun, 2013; Singh et al., 2009; Wang et al., 2014), where they are broken down by acidic lipases. This suggests that regulation of lipophagy might provide a therapeutic approach to preventing metabolic syndrome. However, the molecular machinery responsible for lipophagy, including its specific adaptors, remains uncharacterized.

In the mammalian oocyte, a considerable number of LDs accumulate during oogenesis, although the amount differs across species (Sturmey et al., 2009). For example, in comparison with mouse, porcine oocytes contain massive amounts of LDs (Leese, 2012), which interferes with embryonic manipulation, especially in the context of cryopreservation (Nagashima et al., 1995). The physiological function of LDs stored in the oocyte/embryo is species specific, but, in general, LDs are required for energy (i.e. ATP) production during preimplantation embryonic development (Dunning et al., 2014). LDs stored in the oocyte dynamically change their morphology during oocyte maturation and subsequent embryonic development. In unfertilized (MII) oocytes (before fertilization), LDs are mostly detected as clustered structures, which disperse into smaller droplets soon after fertilization (Bradley et al., 2016; Watanabe et al., 2010). Interestingly, when MII oocytes are cultured without pyruvate and glutamine (i.e. energy-depleted conditions), the clustered LDs are dispersed (Bradley et al., 2016). Thus, LD morphology changes in response to cellular energy level.

In this study, we developed a system of forced lipophagy using a fusion protein of p62 and TIP47, an LD-binding domain. Using this system, we demonstrated that the amount of LDs was decreased upon induction of forced lipophagy in cultured mammalian cells expressing TIP47-p62. Our results demonstrate that maintenance of the proper amount of LDs is important for early embryonic development.

We hypothesized that expression of an artificial linker protein that connects substrates to the autophagosomal membrane would induce selective autophagy. p62 is a well-characterized adaptor that binds to LC3 via the LIR motif (Ichimura et al., 2008; Pankiv et al., 2007). Given that overexpression of p62 results in Nrf2 (Nfe2l2) activation through Keap1 binding, we used the p62^T352A mutant, which lacks the Keap1 interaction (Komatsu et al., 2010). The PAT domain of TIP47 has a high binding affinity for LD surfaces, and efficiently competes with other proteins from LD surface (Bulankina et al., 2009; Kory et al., 2015). Therefore, we fused the PAT domain with p62^T352A and the red fluorescent protein mCherry, yielding TIP47-mCherry-p62^T352A (Fig. 1A). To determine whether this fusion protein could induce lipophagy, we expressed TIP47-mCherry-p62^T352A in HeLa cells and visualized LDs using the fluorescent neutral lipid dye BODIPY 493/503. HeLa cells in regular media contain few LDs, and no significant change in the number of LDs was detected in cells expressing TIP47-mCherry-p62^T352A (data not shown). Therefore, to induce high levels of LDs, we treated the cells with oleic acid (OA). In the treated cells, the amount of LDs was markedly decreased by expression of TIP47-mCherry-p62^T352A in comparison with TIP47-mCherry (Fig. 1B). To determine whether the reduction of LDs was dependent on autophagic degradation activity, we expressed TIP47-mCherry-p62^T352A in Atg14 knockout (KO) cells, which are defective in autophagosome formation (Itakura et al., 2008). The Atg14 KO cells did not exhibit any reduction in the amount of LDs upon expression of TIP47-mCherry-p62^T352A (Fig. 1B), suggesting that TIP47-mCherry-p62^T352A induced degradation of LDs, but did not inhibit LD synthesis or recruitment of endogenous LD-associated proteins to LDs. Analysis of the fluorescence intensity of BODIPY 493/503 by flow cytometry confirmed that the amount of LDs per cell was lower in OA-treated cells expressing TIP47-mCherry-p62^T352A than in mCherry-negative cells (Fig. 1C), indicating that expression of TIP47-mCherry-p62^T352A in cells accelerates breakdown of LDs via forced lipophagy.

This finding prompted us to perform a functional analysis of cells in which LDs were depleted by forced lipophagy. For this purpose, we focused on mouse early embryonic development, for several reasons. First, because LDs accumulate during oogenesis, oocytes contain considerable numbers of LDs. Second, autophagy is highly activated after fertilization, and is essential for early embryonic development (Tsukamoto et al., 2008). Third, the physiological function of LDs during embryonic development remains largely unknown. To induce forced lipophagy after fertilization, we microinjected mRNAs encoding either TIP47-mCherry (control) or TIP47-mCherry-p62^T352A (lipophagy-induced) into the cytoplasm of one-cell embryos (Fig. 2A). The embryos were collected 4-5 h after in vitro fertilization (IVF), the time at which autophagy is highly activated (Tsukamoto et al., 2008). Induction of lipophagy resulted in formation of LD clusters, which were subsequently delivered to the cell periphery during embryonic development. No such clustering and movement of LDs was observed in control embryos (Fig. 2B, Fig. S1A, Movie 1). We confirmed that the clustered LDs observed in lipophagy-induced embryos were specifically labeled with BODIPY 493/503 (Fig. 2C). Similar changes in LD morphology after lipophagy induction occurred in embryos collected from another mouse strain, C3H (Fig. S1B). By contrast, in control embryos, LDs were distributed uniformly, as reported previously (Bradley et al., 2016; Watanabe et al., 2010), although TIP47-mCherry signals failed to localize to the LD surface and were diffused throughout the cytoplasm, probably owing to the abundance of ADRP protein in mouse oocytes and embryos, which could block access of TIP47-mCherry to LDs (see also Fig. S1C,D and Discussion).

In lipophagy-induced embryos, the content of TG, a major component of LDs, decreased approximately 50% at the eight-cell stage in comparison with control embryos (Fig. 2D). Furthermore, quantitative analysis of electron micrographs of the four-cell stage revealed that induction of lipophagy significantly decreased LD number and size (Fig. 2E). Thus, forced lipophagy caused not only morphological changes of LDs, but also a reduction in intracellular lipid storage.

Although we used the p62^T352A mutant to prevent secondary effects of p62, we were still concerned that p62 overexpression alone might induce lipophagy, largely because p62 plays multiple roles in cell signaling, including lipid metabolism (Rodriguez et al., 2006). To eliminate this concern, we synthesized mRNA encoding DsRed-p62^WT (fusion protein of DsRed and wild-type p62) and microinjected it into one-cell embryos. DsRed-p62^WT was detected as cytoplasmic puncta that stained with an anti-ubiquitin antibody that recognizes poly-ubiquitylated proteins, representing protein aggregates, indicating that DsRed-p62^WT was functional (Fig. 3A). We also confirmed that these p62-positive foci partially overlapped with LC3 (autophagosome marker) and Lamp1 (lysosome-associated membrane protein 1, a lysosome marker), indicating that, like LC3, overexpressed p62 also functions as an autophagy adaptor (Fig. 3B). Importantly, DsRed-p62^WT expression induced neither LD clustering nor a close interaction between DsRed-p62^WT and LDs throughout embryonic development (Fig. 3C). Therefore, we concluded that the clustering and movement of LDs observed in lipophagy-induced embryos was unlikely to be due to p62 overexpression, and was instead dependent on p62 localization on the LD surface.

To determine whether forced lipophagy was autophagy dependent, we first analyzed the localization of clustered LDs relative to autophagosomes and lysosomes. Immunofluorescence analysis revealed that both LC3 (autophagosome) and Lamp1 (lysosome) were in close proximity to the clustered LDs in lipophagy-induced embryos, but not in non-injected control embryos (Fig. 4A,B). In the presence of wortmannin [a phosphoinositide 3-kinase (PI3K) inhibitor, which blocks the initial step of autophagy], this close localization was lost, although the clustered LDs were still present. Furthermore, treatment with chloroquine, an inhibitor of lysosomal hydrolases, not only inhibited LD clustering and association with autophagosomes and lysosomes, but also changed the lysosome positioning: specifically, the predominantly peripheral localization of lysosomes was abolished (Fig. 4A,B).

To further assess the specificity of LD clustering upon autophagy, we microinjected siRNA targeting Atg14 along with mRNAs encoding either TIP47-mCherry or TIP47-mCherry-p62^T352A. In these embryos, LC3 dots (representing autophagosomes) were almost absent (Fig. 4C), and embryonic development was impaired, as a result of efficient inhibition of autophagy (Fig. S2A,B). Additionally, LD clustering was diminished, and lysosomal association of LDs was weaker in siRNA-treated lipophagy-induced embryos than in control lipophagy-induced embryos (Fig. 4D, Fig. S2C).

Consistent with our immunofluorescence results, electron microscopic analysis of lipophagy-induced embryos confirmed the abundance of lysosomes and autolysosomes, which were localized proximally to the clustered LDs at the cell periphery (Fig. 4E).

We next investigated whether lipophagy induction would alter endogenous autophagic activity. To this end, we cultured embryos microinjected with either TIP47-mCherry or TIP47-mCherry-p62^T352A mRNA in the presence or absence of bafilomycin A₁ (BafA₁, a lysosomal inhibitor), and then performed immunoblotting with anti-LC3 antibody. The autophagic flux did not differ significantly between control and lipophagy-induced embryos (Fig. 4F), indicating that endogenous autophagic activity was not altered after induction of lipophagy.

To further assess whether lipophagy induction was autophagy specific, we expressed TIP47-mCherry-p62^T352A in unfertilized (MII) oocytes, in which autophagic activity is low, and cultured them for 2 days in vitro. Under these conditions, LD clustering occurred, whereas LD translocation did not (Fig. 4G), indicating that p62 expression at the LD surface is responsible for LD clustering even under low (or absent) autophagic activity, whereas high autophagic activity is required for LD movement to the periphery. Taken together, these results suggested that forced lipophagy occurs in an autophagy-dependent manner.

Based on these results, we hypothesized that lipophagy-induced embryos have higher developmental potential as a result of their more efficient utilization of intracellular LDs in comparison with normal embryos. To investigate this possibility, we cultured embryos microinjected with either TIP47-mCherry or TIP47-mCherry-p62^T352A mRNA for 4-5 days in vitro. Unexpectedly, the proportion of embryos reaching the late blastocyst stage within this time frame was lower in lipophagy-induced embryos (30.7%) than in control embryos (56.7%) (Fig. 5A,B). Because the mouse embryo is particularly reliant on exogenous nutrients (Leese, 2012), we then cultured control and lipophagy-induced embryos without amino acids. The blastocyst formation rate of the lipophagy-induced embryos was not influenced by the presence or absence of amino acids, whereas amino acid deficiency significantly decreased the rate of development in control embryos (Fig. 5A,B). These results imply that utilization of intracellular LDs is important for early embryogenesis, particularly in the absence of extracellular energy sources.

The intracellular distribution of LDs may be linked to energy balance, e.g. the close association of LDs with mitochondria facilitates ATP production via β-oxidation (Herms et al., 2015). Accordingly, we analyzed the ATP content in lipophagy-induced embryos. Throughout embryonic development, ATP content did not differ significantly between lipophagy-induced and control embryos (Fig. S3A). Consistent with this, clustered LDs were not localized in close proximity to mitochondria in lipophagy-induced embryos (Fig. S3B). In addition, we observed no significant difference in cellular levels of reactive oxygen species (ROS) between control and lipophagy-induced embryos (Fig. S3C). These data indicate that lipophagy induction is not directly involved in increasing cellular ATP, even when mitochondrial function is not impaired.

We further assessed the contribution of lipolysis to lipophagy-induced embryos. To this end, we cultured either control or lipophagy-induced embryos with OA, to stimulate TG synthesis, in the presence or absence of the lipase inhibitor diethylumbelliferyl phosphate (DEUP). OA supplementation had a negative influence on embryonic development in control embryos, but not in lipophagy-induced embryos (Fig. 5A-D, Fig. S4). Moreover, in the presence of DEUP, almost all control embryos were developmentally retarded, probably owing to excess accumulation of LDs, whereas 12.7% of lipophagy-induced embryos survived even to the late blastocyst stage (Fig. 5C,D, Fig. S4).

Palmitic acid (PA, a saturated fatty acid) is more lipotoxic than OA (unsaturated fatty acid) in a variety of cell lines (Listenberger et al., 2003). To assess tolerance of lipotoxicity, we cultured control or lipophagy-induced embryos with PA to the blastocyst stage. PA treatment substantially decreased the developmental rate (i.e. the rate at which the embryos formed blastocysts) in both groups (Fig. 5A,B,E,F), probably as a result of lipotoxicity. However, the level of apoptosis, indicative of lipotoxicity, was lower in the lipophagy-induced embryos than in control embryos (Fig. 5G).

Taken together, these data indicate that forced lipophagy plays an important role in avoiding excess LD accumulation, and can confer resistance to lipotoxicity.

In this study, we developed a system for inducing forced selective autophagy of LDs, termed forced lipophagy, using a fusion protein of p62 and an LD-binding domain. p62 on the surface of LDs can be recognized as an autophagic cargo, facilitating the interaction with LC3 through its LC3-interacting region (LIR). LC3 stably binds to the autophagosomal membrane, forming a bridge between LDs and the membrane. The autophagosomal membrane then elongates to form the autophagosome, which encloses a portion of LDs, and subsequently fuses with the lysosome for degradation (see our proposed model, Fig. S5). LDs are also metabolized (i.e. degraded) by cytosolic neutral lipases such as adipose triglyceride lipase (ATGL; PNPLA2) and hormone-sensitive lipase (HSL; LIPE), both of which contain LC3-binding motifs required for their lipolytic activities (Martinez-Lopez et al., 2016). In addition, lipolysis is tightly coupled with chaperone-mediated autophagy (CMA) (Kaushik and Cuervo, 2015). Although additional unknown factor(s) and mechanism(s) such as self-oligomerization of p62 via the PB1 domain (which may promote LD clustering) and degradation of LD coat proteins (Plins) by the proteasome (which may stabilize the association of p62 on the LD surface) might contribute to this process, our results strongly suggest that expression of p62 on the LD surface is sufficient to induce selective LD degradation in mammalian cells and fertilized mouse embryos.

We observed that induction of lipophagy in mouse embryos caused LD clustering and translocation to the cell periphery. This clustering also occurred in cultured mammalian cells, whereas translocation did not (data not shown), suggesting that the latter process is mediated by an embryo-specific mechanism. Although the mechanism underlying this process is not fully understood, lipophagic activity in embryos might be required for subsequent translocation of clustered LDs to the cell periphery. Consistent with this idea, in unfertilized oocytes (in which autophagic activity is low), expression of TIP47-mCherry-p62^T352A caused LD clustering but not subsequent movement. Similar results were obtained when lipophagy-induced embryos were cultured with wortmannin, an inhibitor of autophagy, and injected with siRNA targeting Atg14. In contrast, lysosomal inhibition by chloroquine treatment shifted lysosome localization, but caused neither LD clustering nor translocation in lipophagy-induced embryos, indicating that lysosomal positioning influences the association between clustered LDs and lysosomes.

Unexpectedly, we found that forced lipophagy induction decreased embryonic viability (Fig. 5A,B), suggesting that LDs consumed by forced lipophagy at an early stage of embryonic development might be necessary at a later stage. Thus, preimplantation embryonic development requires not only endogenous lipolysis, but also the proper amount of LDs. On the other hand, the developmental potential of lipophagy-induced embryos was not significantly influenced by the presence of amino acids. At present, we know little about the molecular machinery responsible for this phenomenon, but speculate that a signaling pathway that regulates autophagy [e.g. mammalian target of rapamycin (mTOR), which negatively regulates autophagy] or cellular energy balance [e.g. AMP-activated protein kinase (AMPK), which is activated by ATP depletion] could be involved, facilitating the recycling of fatty acids (FAs) generated by forced lipophagy.

By contrast, some lipophagy-induced embryos survived under conditions in which excess LDs accumulated (i.e. in the presence of OA and a lipase inhibitor) or lipotoxic species were generated (i.e. under PA supplementation). This result indicates that forced lipophagy attenuates lipotoxicity by eliminating excess LDs. Although the precise mechanism underlying tolerance of lipotoxicity remains unknown, it is possible that unsaturated FAs generated by forced lipophagy protect the cells against lipotoxic FAs, such as PA, as previously demonstrated in cultured mammalian cells (Listenberger et al., 2003). Further analysis will be required to determine whether tolerance of lipotoxicity is related to apoptosis.

An unresolved issue, based on our observations in mouse early embryos, is that TIP47-mCherry, used as an LD marker, did not work as expected. In contrast to the lipophagy-induced embryos (microinjected with TIP47-mCherry-p62^T352A), in which fluorescence was specifically detected at the LD surface, TIP47-mCherry-injected embryos exhibited diffuse fluorescence (Fig. S1C). Because post-translational modification of Plins, including phosphorylation and ubiquitylation, are involved in stabilizing the LD localization of these proteins (Orlicky et al., 2008; Xu et al., 2005), we speculated that some modification occurs at the C terminus of TIP47, which might affect its LD localization. Therefore, we created mCherry-TIP47 (C-terminally mCherry-tagged TIP47) and repeatedly microinjected it into one-cell embryos. However, although the new fusion protein was expressed, fluorescence signals were not detected at the LD surface, but instead remained diffusely distributed in the cytoplasm (Fig. S1D). On the other hand, mRNAs encoding either TIP47-mCherry-p62^T352A or mCherry-TIP47-p62^T352A could induce lipophagy, with the mCherry signals localized specifically at clustered LDs (Fig. S1C,D), indicating that p62 expression at the LD surface might stabilize TIP47 or its interaction with LDs. Because a previous study in cultured mammalian cells showed that TIP47 fails to associate with LD when full-length ADRP is simultaneously overexpressed, but not when an N-terminally modified (mutant or GFP-tagged) ADRP is overexpressed (Orlicky et al., 2008), we speculate that access of TIP47 to LDs is dependent on the level of ADRP. Consistent with this idea, ADRP is abundant in mouse oocytes and embryos (T.T. and S.T., unpublished). Our ongoing research using ADRP KO mice should be able to determine why TIP47 protein fused to either the N or C terminus of mCherry does not function as an LD marker in fertilized mouse embryos.

Cellular lipid content varies between animal species, probably because of differences in ATP requirements between oocyte maturation and subsequent embryonic development (Dunning et al., 2014). For example, in comparison with mouse and human, porcine oocytes contain massive amounts of LDs [37.3 ng fat/nl volume of oocyte in pig, versus 6.25 ng fat/nl in mouse (Leese, 2012)], which can negatively influence embryonic manipulation, especially in the case of cryopreservation (Nagashima et al., 1995). Because autophagy is highly conserved across animal species, it would be interesting to investigate the consequences of forced lipophagy in porcine embryos.

In summary, we achieved forced selective autophagy by expressing a p62 fusion protein, thereby generating cells depleted of a specific organelle. Using this system, we revealed that the correct amount of LDs is important for early embryonic development. The same strategy could be applied to generate other organelle-depleted cells to remove unwanted organelles and thus analyze organelle functions. Therefore, this method represents a valuable tool for elucidating the roles of specific organelles.

To generate pCW TIP47-mCherry and pCW TIP47-mCherry-p62^T352A, TIP47 (1-221 amino acids) and human p62^T352A cDNA were cloned into pCW57.1 (Addgene plasmid #41393, deposited by David Root) along with mCherry. The plasmid DsRed-p62^WT was a kind gift from Dr Masaaki Komatsu (Niigata University, Japan).

HeLa cells were cultured in Dulbecco's Modified Eagle's medium (DMEM, Nacalai Tesque) supplemented with 10% fetal bovine serum (FBS, Biosera), 50 mg/ml penicillin, and streptomycin (regular medium) in a 5% CO₂ incubator. Atg14 KO HeLa cells were generated by clustered regularly interspaced short palindromic repeat (CRISPR) using pX459 GFP carrying the sgRNA expression cassette to target Atg14 (CAGAGGCATAATCGCAAACT) as previously described (Takayama et al., 2017). TIP47-mCherry and TIP47-mCherry-p62^T352A cells were generated by lentiviral transduction with pCW, psPAX2 (Addgene plasmid #12260, deposited by Didier Trono) and pCMV VSV-G (Addgene plasmid #8454, deposited by Bob Weinberg). The stable cell lines were cultured in regular media containing 1 mg/ml doxycycline. For drug treatment, cells were incubated for the indicated times in medium containing 0.5 mM OA (Sigma-Aldrich). For LD staining, cells were cultured for 30 min in the presence of 1 µg/ml BODIPY 493/503 (Thermo Fisher Scientific).

Trypsinized cells were incubated with 1 µg/ml BODIPY 493/503 (Thermo Fisher Scientific) in PBS for 30 min at 4°C, and then sedimented by centrifugation at 5000 rpm (2500 g) for 2 min. The pelleted cells were washed with PBS and then re-suspended in PBS containing 10% FBS and 1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI), passed through a 70-μm cell strainer, and subjected to flow cytometry on a CytoFLEX S flow cytometer equipped with NUV 375 nm (for DAPI), 488 nm (for GFP) and 561 nm (for mCherry) lasers (Beckman Coulter). Dead cells were detected by DAPI staining.

All mouse experiments were performed in accordance with the relevant guidelines and were approved by the Animal Care and Use Committee of the National Institute of Quantum and Radiological Science and Technology. Mice used in this study were 8- to 12-week-old C57BL/6J mice (Japan SLC), except for C3H mice (Japan SLC) used for the experiment described in Fig. S1B. Mice were housed under specific pathogen-free conditions. Metaphase II (MII) oocyte collection from superovulated females, embryo manipulation, in vitro fertilization, microinjection, and live-cell imaging were performed based on previously described methods (Kaizuka et al., 2016; Tsukamoto et al., 2014, 2013). For assessment of blastocysts, one-cell embryos were cultured for 4-5 days in vitro, and blastocyst formation was evaluated based on expansion of the blastocoel. We confirmed that 70% of control embryos (70.1±6.2%; from three independent experiments) injected with TIP47-mCherry mRNA developed to early and growing blastocysts with a cavity volume greater than half the volume of the embryo. We considered developing embryos with cavities that completely filled the embryo as late blastocysts. When necessary, embryos were cultured with KSOM lacking amino acids (Eagle's essential and non-essential amino acids), and glutamine. In some experiments, one-cell embryos were cultured with the following chemicals: 200 nM BafA₁ (WAKO Pure Chemical Industries), 100 µM DEUP (Sigma-Aldrich), 0.1 mM OA, 0.1 mM PA (Sigma-Aldrich), 10 µM wortmannin (Sigma-Aldrich) or 50 µM chloroquine (WAKO Pure Chemical Industries).

For RNAi experiments, Stealth RNAi duplex oligonucleotides (siRNAs) were designed to target Atg3 (MSS228710, Invitrogen) or Atg14 (MSS211913, Invitrogen). Stealth RNAi Negative Control Duplexes (Invitrogen) were used as negative controls. Knockdown efficiency was calculated based on conversion of LC3-I to LC3-II, as determined by immunoblot analysis (Fig. S2B), as well as by immunofluorescence analysis of LC3 (Fig. 4C). In addition, we assessed whether embryos injected with siRNAs were developmentally compromised (Fig. S2A), based on our previous observation that autophagy-deficient embryos (derived from oocyte-specific Atg5 knockout females) die before implantation (Tsukamoto et al., 2008). Based on these analyses, we chose siRNA targeting Atg14, which inhibited development more efficiently than siRNA against Atg3, for further study.

In vitro transcription and microinjection were performed as described (Tsukamoto et al., 2014). Briefly, plasmids encoding the indicated mRNAs were linearized, and capped mRNAs with poly (A) tailing were synthesized using the mMESSAGE mMACHINE Ultra Kit (Ambion). The purified mRNAs were diluted to 100 ng/µl in TE buffer, filtered, and microinjected into the cytoplasm of oocytes or embryos. For the RNAi experiment, 10 µM siRNA was microinjected simultaneously with either TIP47-mCherry or TIP47-mCherry-p62^T352A mRNA into the cytoplasm of fertilized embryos, as described above, except that these mixtures were not filtered before microinjection.

Cells expressing fluorescently tagged protein were grown on coverslips and fixed in 3.7% formaldehyde in PBS for 15 min. For LD staining, the fixed cells were incubated with 1 µg/ml BODIPY 493/503 in PBS for 30 min and observed under a confocal laser microscope (FV1000 IX81, Tokyo, Japan) using a 60× oil-immersion objective lens with a numerical aperture (NA) of 1.42.

Immunofluorescence of mouse embryos was performed as previously described (Tsukamoto et al., 2014). Briefly, fixed embryos were blocked and permeabilized overnight at 4°C with 0.4% Triton X-100 in PBS containing 10% goat serum and 3% bovine serum albumin (BSA). The permeabilized embryos were then incubated with the following primary antibodies, diluted in PBS containing 3% BSA (PBS/BSA), overnight at 4°C: anti-LC3B (Cosmo Bio; CAC-CTB-LC3-2-IC; 1:200), anti-Lamp1 (Abcam; ab25245; 1:200), anti-ubiquitin (Medial & Biological Laboratories; D058-3; 1:200) and anti-Atp5a (Abcam; ab14748; 1:200). The embryos were washed in PBS/BSA at least four times for 15 min each, and then incubated with the appropriate secondary antibodies (1:800) for 1 h at room temperature or overnight at 4°C. Embryos were then washed once more, and transferred to PBS/BSA in a glass-bottomed dish (Matsunami Glass) covered with mineral oil. Fluorescence images were acquired with a SP8 laser scanning confocal system (Leica, TCS, SP8) with a 63× Plan APO 1.4 NA oil immersion objective.

For labeling of LDs, fixed oocytes or embryos were stained with BODIPY 493/503 solution (1 µg/ml, diluted in PBS/BSA) in the dark for 10-15 min at room temperature, washed several times with PBS/BSA, and observed by laser confocal microscopy as described above. Fluorescence measurements were performed using ImageJ software (NIH). Images were adjusted in Photoshop and Illustrator (Adobe).

To measure ROS production, embryos were incubated with 10 µM aminophenyl fluorescein (APF, GORYO Chemical) for 30 min in PBS containing polyvinylpyrrolidone (PVP, Sigma-Aldrich) instead of BSA to avoid interference with fluorescence measurements. Fluorescence was imaged on a real-time cultured cell monitoring system (CCM, ASTEC Co.) with a 10× objective lens (Plan, NA0.25, Nikon), and analyzed using ImageJ software as described before.

To monitor cellular apoptosis, embryos were stained using the Annexin V-EGFP Apoptosis Kit (BioVision, #K104-25), which includes Annexin V-EGFP and propidium iodide. Fluorescence was analyzed by CCM, and fluorescence measurements were performed as described above.

Immunoblotting was performed based on previously described methods (Tsukamoto et al., 2014). The following antibodies were used: anti-LC3B (Cell Signaling Technology; #12741; 1:400) and anti-α/β-tubulin (Cell Signaling Technology; #2148; 1:500). Horseradish peroxidase (HRP)-conjugated anti-mouse and anti-rabbit IgG antibodies were purchased from Cell Signaling Technology (#7076 and #7074, respectively). For the autophagic flux assay, immunoblots were performed according to previously reported guidelines for monitoring autophagy (Klionsky et al., 2016). Briefly, 70 control or lipophagy-induced embryos were cultured in the absence or presence of BafA₁ and lysed for immunoblotting. After SDS-PAGE, proteins were blotted onto PVDF membrane using a Trans-Blot Turbo Transfer System (Bio-Rad). The blots were blocked with 3% BSA in TBS-T (Bio-Rad) for 1 h, and then incubated with primary antibodies at 4°C overnight. Subsequently, the membranes were washed three times for 10 min each with TBS-T (Bio-Rad), and then incubated with the appropriate HRP-conjugated secondary antibodies as described above. The signals were visualized using a SuperSignal West Femto Chemiluminescent Substrate (Thermo Fisher) and imaged on a ChemiDoc system (UVP).

Electron microscopy was performed as described previously (Tsukamoto et al., 2013). LD number and size were quantified using the ImageJ segmentation tool (NIH).

For TG measurements, 40 embryos per experiment were transferred into 10 µl of water (Nippon Gene) and stored at −80°C until use. TG levels were assayed using the Picoprobe Triglyceride Fluorometric Assay kit (BioVision, #K614-100). In this assay, TG is converted to glycerol and FA, and the resultant glycerol is oxidized to generate a product that reacts with the picoprobe and developer (included in the kit) to generate fluorescence. Because the resultant fluorescence is directly proportional to the amount of TG, TG content was quantitated by fluorometry. The fluorescence was measured on a QFX Fluorometer (DeNovix), and TG level was calculated using a standard curve.

Intracellular ATP levels were determined using the IntraCellular ATP assay kit (TOYO B-Net Co.). In brief, embryos cultured to the indicated stage were individually transferred into 100 µl of water, and then mixed with 100 µl of the kit reaction buffer, followed by incubation for 10 min at room temperature. ATP was measured immediately using a GENELIGHT (Microtec Co., #GL-220A). ATP content was calculated using a standard curve, prepared using the ATP provided with the kit.

Data were compared using unpaired Student's t-test in Prism 6 (GraphPad). Statistical significance is represented as *P<0.05, unless otherwise stated.

We thank Dr Noboru Mizushima (The University of Tokyo) for his mentorship and encouragement. We also thank Masaaki Komatsu (Niigata University) for providing the DsRed-p62^WT plasmid, and Ayako Wada and Tomomi Hatakeyama for technical assistance.