Concept of lipid droplet biogenesis

Abstract

Lipid droplets (LD) are functionally conserved fat storage organelles found in all cell types. LDs have a unique structure comprising of a hydrophobic core of neutral lipids (fat), triacylglycerol (TAG) and cholesterol esters (CE) surrounded by a phospholipid monolayer. LD surface is decorated by a multitude of proteins and enzymes rendering this compartment functional. Accumulating evidence suggests that LDs originate from discrete ER-subdomains, demarcated by the lipodystrophy protein seipin, however, the mechanisms of which are not well understood. LD biogenesis factors together with biophysical properties of the ER membrane orchestrate spatiotemporal regulation of LD nucleation and growth at specific ER subdomains in response to metabolic cues. Defects in LD formation manifests in several human pathologies, including obesity, lipodystrophy, ectopic fat accumulation, and insulin resistance. Here, we review recent advances in understanding the molecular events during initial stages of eukaryotic LD assembly and discuss the critical role of factors that ensure fidelity of this process.

1. Introduction

Lipid droplets (LDs) represent a functionally conserved dynamic organelle dedicated for the storage of metabolic energy and plays a crucial role in cell physiology. LD core is filled with neutral lipids (NLs)/fat, particularly, triglycerides (TAG), and sterol esters (SE), enveloped by a monolayer of phospholipids, harboring LD-specific proteins having role in lipid metabolism including lipases and acyltransferases, and structural proteins, such as perilipins (Olzmann and Carvalho, 2019; Wilfling et al., 2014). Impaired LD homeostasis manifests in various pathological conditions, such as obesity, type-2 diabetes, lipodystrophy (abnormal distribution of fat), insulin resistance, neuronal diseases, and cancer (reviewed in (Henne et al., 2019; Islimye et al., 2022; Olzmann and Carvalho, 2019; Zadoorian et al., 2023)). Growing evidence suggests that apart from lipid homeostasis, LDs play roles in protein degradation (Hartman et al., 2010; Olzmann et al., 2013), the endoplasmic reticulum (ER) stress response (Fei et al., 2009), they act as sites for assembly of infectious virions (Miyanari et al., 2007), are involved in membrane trafficking and signal transduction, and act as a temporary storehouse of proteins (Cermelli et al., 2006; Fujimoto and Ohsaki, 2006; Martin and Parton, 2006). Owing to its link with major metabolic disorders, LD biogenesis has been vividly explored in recent years and have provided novel insights. Assembly of eukaryotic LDs occurs in the ER, the mechanisms of which is not completely known (Henne et al., 2019; Olzmann and Carvalho, 2019; Walther et al., 2017; Zadoorian et al., 2023). The earliest observation indicating an ER origin of lipid bodies/oil bodies (OBs)/spherosomes dates back to ultrastructural analysis of plant seeds using conventional electron microscopy (EM) technique (Frey-Wyssling et al., 1963). Remarkably, these OBs were found to be surrounded by an unusual half-unit membrane (Yatsu and Jacks, 1972), a finding further confirmed by cryo-EM of isolated LDs from HepG2 cells (Tauchi-Sato et al., 2002).

ER localized NL synthesizing enzymes generate NLs that assimilate within the ER bilayer membrane at low concentrations, however, upon reaching a certain threshold (2.8–10 mol%), NLs coalesce into lenses/blisters via de-mixing phenomenon of phase separation as revealed by in silico and in vitro data (Hamilton et al., 1983; Khandelia et al., 2010; Thiam and Ikonen, 2021), and observed as 30–60 nm structures by electron microscopy (EM) in yeast (Choudhary et al., 2015). These lenses acquire more NLs and eventually grow into nascent LDs that emerge toward the cytoplasm delineated by a phospholipid monolayer and are decorated by specific set of LD-resident proteins (Olzmann and Carvalho, 2019; Walther et al., 2017). LDs retain close physical and functional relationship with the ER throughout their life cycle to ensure bi-directional transport of proteins and lipids between these two compartments (Choudhary and Schneiter, 2021; Jacquier et al., 2011; Salo et al., 2016). However, LDs can be detached from the ER during lipophagy, or when they are being secreted as milk globules by mammary gland epithelial cells (Schulze et al., 2020). Recent studies have revealed role of proteins, lipids, and biophysical properties that determine assembly of LDs at discrete ER subdomains, a pre-requisite for maintaining ER homeostasis. In this review, we discuss recent insights into eukaryotic LD nucleation event that is restricted to few specialized ER subdomains and discuss the critical role of LD assembly factors that ensure fidelity of this process.

2. Passive vs active model of LD biogenesis

In mammals, the terminal step of TAG synthesis is catalyzed by diacylglycerol acyltransferase (DGAT1, DGAT2) through esterification of diacylglycerol (DAG) (Lardizabal et al., 2001; Yen et al., 2008), whereas SE is catalyzed by the esterification of cholesterol by acyl-CoA cholesterol O-acyltransferase (ACAT1, ACAT2) (Cases et al., 2001; Chang et al., 1993). NL synthesis in budding yeast is under the control of four enzymes; Lro1 and Dga1 produce TAG, Are1, and Are2 produce SE. All NL synthesizing acyltransferases are ER localized, however, DGAT2/Dga1 can also localize on the surface of mature LDs. Remarkably, NLs are dispensible for yeast since a quadruple mutant missing all four NL-synthesizing enzymes (4ΔKO; dga1Δ lro1Δ are1Δ are2Δ) is viable but lacks NLs or any detectable LDs (Sandager et al., 2002). Lack of LDs sensitizes yeast toward excess fatty acids, a phenomenon known as lipotoxicity (Garbarino et al., 2009).

Whether lenses of NLs accumulate in a stochastic/passive manner in the ER, or form at predefined ER subdomains is not well understood. Our understanding of the mechanisms which regulate ordered nucleation of NL is far from complete. However, recent work suggests that certain ER subdomains are selected for efficient LD formation, and this process is influenced by a number of factors including biophysical properties of the ER, ER membrane curvature, ER membrane tension, locally enriched lipids, LD assembly factors, and protein-lipid interactions (Henne et al., 2020; Schneiter and Choudhary, 2022; Thiam and Ikonen, 2021). For NL nucleation to occur, the first criteria includes overcoming energy barrier which takes place through the phenomenon of phase separation/demixing process, where it mainly depends on the interaction between phospholipids (PLs) and TAG. By altering the ratio of PLs and TAG, the critical demixing concentration (CDC) also changes. Thus, nucleation starts at sites where the nucleation energy is less (Thiam and Ikonen, 2021).

A key LD assembly factor that regulates NL nucleation is seipin (Sei/Fld1 in yeast) that renders ER subdomains permissive for droplet biogenesis (reviewed in (Salo, 2023)). Seipin and its associated partner LDAF1/promethin assemble together in a complex at ER-LD contact sites, that copurifies with TAG (Castro et al., 2019; Chung et al., 2019; Eisenberg-Bord et al., 2018). In the absence of LDAF1, very few LDs form for a given amount of TAG, suggesting that seipin-complex induces NL nucleation at a lower TAG concentration by decreasing the energy barrier for LD nucleation (Chung et al., 2019; Prasanna et al., 2021). Consistent with this de novo LDs form at sites marked by seipin, and if seipin is absent LDs form ectopically throughout the ER (Choudhary et al., 2020; Gao et al., 2017; Salo et al., 2019). Random production of TAG results in a delayed LD formation with concomitant TAG accumulation within the ER bilayer, and/or aberrant generation of small, supersized or clustered LDs, a phenotype typically associated with seipin knockouts. Such heterogenous growth of nascent LDs in the absence of seipin can be explained by a biophysical process called Ostwald ripening which promotes transfer of NLs from smaller to larger LDs through ER membrane conduits. As a result larger LDs would grow at the expense of small shrinking LDs. Hence presence of seipin-complex counteracts ripening and facilitates uniform delivery of NLs from ER into growing LDs (Salo et al., 2019). Unlike mobile seipin foci in the ER, ER subdomains engaged in droplet formation are stable and show enrichment of seipin prior to accumulation of either NL dyes LD540, BODIPY, or LiveDrop (an early LD marker), suggesting that seipin can spatially define LD formation sites in the ER. In this regard, relocalization of seipin to a subdomain of the ER, such as nuclear ER or ER-PM junctions resulted in relocation of LD biogenesis machinery at these newly marked ER sites (Chung et al., 2019; Salo et al., 2019).

Therefore, LD assembly is indeed an active process that is spatially defined by the seipin-complex and do not occur passively through random coalescence of NL into lenses in the ER, to ensure ER and lipid homeostasis (Fig. 1). Future study will uncover the constituents of the seipin-complex that regulate its localization in the ER.

Fig. 1. Passive vs Active model of LD biogenesis.

A) Passive model: In a passive model of LD formation, neutral lipids are produced all over the ER in a random manner resulting in heterogeneous population of lipid droplets that exist either as numerous small or clustered LDs or a few large LDs. These stochastically formed LDs fail to enrich LD surface proteins. B) Active model: In the Active model, ER subdomains spatially defined by the seipin-complex recruits various LD-assembly factors and facilitates localized production of NLs and prevents its outflow into the ER membrane from these sites. At these sites lenses of NLs grow into nascent LDs. Seipin establishes functional ER-LD contacts that facilitates bi-directional transport of proteins and lipids between the two compartment.

3. Biophysical parameters of the ER membrane: indispensable for LD biogenesis

Growing evidence suggests that biophysical property of ER membrane play an important role in selection of sites where nucleation of nascent LDs might occur and/or facilitate emergence of growing LD. Data from in vitro and in vivo experiments together with in silico modelling analysis demonstrate that PL composition of ER membrane and membrane tension play a crucial role in LD biogenesis (Ben M’barek et al., 2017; Choudhary et al., 2018). An asymmetric distribution of lipids at LD biogenesis sites modulates surface tension and membrane curvature properties that affects the direction into which LD emerges from the ER (Chorlay et al., 2019; Gao et al., 2019). It has been shown that PL having negative intrinsic molecular curvature such as DAG, favour embedded state of LD within the ER bilayer, whereas lipids having positive curvature, such as lysophospholipids, reduce surface tension and promote budding of LDs from the ER (Ben M’barek et al., 2017; Choudhary et al., 2018; Gao et al., 2019). In vitro experiments using artificial LDs embedded inside giant unilamellar vesicles (GUVs), indicate that LDs emerge toward the side of the membrane having increased coverage of phospholipids and proteins, thereby resulting in LDs with lower surface tension (Chorlay et al., 2019). Consistent with this, PLs need to be replenished in a continuous manner on the cytoplasmic side of the ER membrane to ensure emergence of droplet into the cytoplasm. When cells are loaded with oleate, LDs need to expand dramatically, under such scenario the capacity to refill PLs might get overwhelmed resulting in failure of LD emergence and instead LDs would bud toward the ER lumen (Chorlay et al., 2019; Mishra et al., 2016).

Recently it has been shown that acyl chain composition of ER PLs might also affect nucleation of NL, with higher levels of saturated or short chain fatty acids impeding nucleation of NLs by interfering with phase separation, thereby NLs tend to accumulate within the ER membrane (Zoni et al., 2021a). Moreover, the architecture of ER membrane also influences LD nucleation. In agreement with this, LD biogenesis showed bias toward ER tubule rather than ER sheets. Owing to higher curvature, seipin preferentially enriches in tubular ER subdomains, rather than ER sheets, and hence controls droplet nucleation at these ER domains (Santinho et al., 2020). Moreover, TAG accumulation in ER tubules is energetically less favourable compared to ER sheets, thus promoting either outflow of TAG from ER tubules or their condensation into nascent LDs (Santinho et al., 2020). In consonance with this, LD nucleation can be achieved in vitro by enhancing membrane curvature, thus, supporting the notion that ER membrane architecture catalyzes droplet nucleation (Santinho et al., 2020).

4. Lipin complex regulates production of storage lipids

How LD biogenesis sites are regulated and which seipin-marked ER subdomains get selected for droplet assembly is not well understood. Apart from proteins, specific lipids enriched at these ER subdomains create permissive platform for the assembly of LDs. In particular, accumulation of DAG, a key substrate for ER localized TAG-synthases plays a crucial role in LD biogenesis. Accumulation of DAG at LD biogenesis sites create a platform for the recruitment of many LD biogenesis factors, including acyltransferases and hence generates an active hotspot for droplet assembly (Choudhary et al., 2020; Gao et al., 2019). Regulation of DAG production is under the control of Lipin class of lipid phosphatases (Pah1 in yeast), that determine the crucial decision for channelling phosphatidic acid (PA) toward either PL synthesis for membrane biogenesis or toward storage lipid TAG synthesis (Kwiatek et al., 2020; Zhang and Reue, 2017) (Fig. 2A). Mammals differentially express three Lipins, Lipin1–3, with Lipin-1 having additional three splice variants, Lipin-1α, Lipin-1β, and Lipin-1γ (Peterfy et al., 2005; Peterfy et al., 2001; Wang et al., 2011). Lipins show tissue specific distribution; Lipin-1 is highly distributed in adipose tissue and skeletal muscles, Lipin-2 show predominant expression in liver and brain, whereas Lipin-3 is present in gastrointestinal tract and liver (Donkor et al., 2007). All three Lipin proteins have Phosphatidic Acid Phosphatase (PAP) enzyme activity in vitro, with Lipin-1 having the highest activity (Donkor et al., 2007), and perform functions as transcriptional co-regulator inside the nucleus (reviewed in (Zhang and Reue, 2017)). Lack of Lipin-1 function in mice results in generalized lipodystrophy, with reduced body fat mass and impaired adipocyte differentaion (Peterfy et al., 2001). Cytosolic Lipins translocate to various cellular membranes including ER in response to stimuli such as fatty acids, or the presence of negative charges conferred by the accumulation of its substrate, PA (Zhang and Reue, 2017). This association of Lipin with ER and nuclear membrane is functionally conserved from mammals to invertebrates such as C. elegans and Drosophila (Gorjanacz and Mattaj, 2009; Valente et al., 2010). Multisite phosphorylation regulates the membrane translocation and activity of yeast Pah1 (Choi et al., 2011).

Fig. 2. Activation of lipin-complex triggers production of storage lipids.

A) Schematics of lipin activation. ER localized phosphatase complex Nem1-Spo7 (CTDNEP1-NEP1R1 in mammals) controls the activation of phosphatidic phosphatase, Pah1 (Lipins in mammals). Dephosphorylated Pah1 becomes active and catalyses hydrolysis of phosphatidic acid (PA) to diacylglycerol (DAG). Accumulation of DAG triggers recruitment of acyltransferases for the production of storage lipid TAG, and thus stimulates droplet formation. ER membrane protein Ice2 (SERINCs in mammals) inhibits Nem1-Spo7 complex, thereby preventing activation of Lipin/Pah1. In such a scenario, PA accumulates and acts as a precursor for phospholipid synthesis and hence shifts the balance toward membrane expansion instead of storage lipids. DAG can also serve as substrate for Kennedy pathway for the synthesis of membrane lipids. B) Colocalization between seipin- and lipin-complex creates hotspots for LD biogenesis. Seipin-complex localizes independently of the lipin-complex in the ER. Colocalization between seipin- and lipin-complex render these ER subdomains active and create hotspots for the assembly of LDs.

The Lipin/Pah1 activity hydrolyzes PA to DAG which is then used by TAG-synthases to produce TAG and hence promote LD formation and growth. Pah1 activity is tightly controlled by phosphorylation and is activated by ER membrane anchored phosphatase complex composed of catalytic subunit Nem1, and regulatory subunit Spo7 (Karanasios et al., 2013; Kwiatek et al., 2020) (Fig. 2A). The mammalian orthologue of Nem1-Spo7 complex comprises of CTDNEP1-NEP1R1 complex suggesting shared mechanism for Lipin activation (Han et al., 2012; Kim et al., 2007). Nem1-Spo7 phosphatase complex is inhibited by Ice2, a polytopic ER membrane protein thereby regulating the activity of Pah1 (Papagiannidis et al., 2021). Lack of Ice2 results in accumulation of dephosphorylated form of Pah1, and enhanced LD storage, whereas presence of Ice2 promotes TAG catabolism and regulates the flux of lipids towards membrane biogenesis (Markgraf et al., 2014; Papagiannidis et al., 2021). Ice2 belongs to the serine incorporator (SERINC) superfamily of proteins in mammals, that contains hydrophobic groove that might bind to lipids and modulate viral infectivity (Alli-Balogun and Levine, 2021).

Interestingly, lack of any member of the Lipin complex, Pah1, Nem1, Spo7 results in dramatic ER expansion with accumulation of STE class of NLs within the ER membrane (Adeyo et al., 2011). Recently it has been shown that Nem1 and Sei1 localize independently of each other. Interestingly, upon induction of TAG synthesis increased colocalization between Sei1 and Nem1 was observed, suggesting that the number of LD biogenesis sites is regulated in a dynamic manner in response to NL production (Choudhary et al., 2020) (Fig. 2B). Consistent with this increased local DAG levels were observed at Sei1/Nem1 sites upon stimulation of LD formation as revealed by ER-DAG sensor, a probe to visualize DAG distribution in the ER (Choudhary et al., 2020). Future work awaits mechanistic understanding of how these sites are turned on/off to regulate NL filling into LDs.

5. Recent insights on mechanistic role of LD assembly factors

5.1. Seipin-complex assemble into disc like structure and creates a hydrophobic environment in the ER permissive for lipid nucleation and transfer

Seipin/Sei1 is one of the crucial ER membrane protein that posits indispensable function in LD biogenesis in all eukaryotes including yeast, plants and animals (Salo, 2023). The protein is encoded by Berardinelli-Seip Congenital Lipodystrophy type 2 gene (BSCL2), mutation in which causes severe form of lipodystrophy (Magré et al., 2001), upper, lower, and peripheral motor neuronal disorders and encephalopathy, collectively referred to as seipinopathies (reviewed in (Rao and Goodman, 2021)). Its pertinent role in the disease gained attention of researchers for investigation of the detailed mechanism of its role in LD biogenesis. Seipin contains N- and C-terminal regions exposed toward the cytoplasm, two transmembrane domains (TMDs), and a highly conserved ER luminal domain that play critical role in LD nucleation (Rao and Goodman, 2021; Wang et al., 2016). Remarkably, seipin localizes to discrete ER subdomains, independently of NL synthesis, where they create a platform for the assembly of droplets by recruiting LD-biogenesis factors and/or enrichment of lipid intermediates (Choudhary et al., 2020). It remains to be determined if these seipin marked ER subdomains in yeast comprises of one or more seipin oligomer. However, it is plausible that upon rapid expansion of LDs, under conditions of oleate loading, adjacent seipin oligomers might become juxtaposed thereby resulting in fusion of two neighbouring early LDs into a larger LD, thus allowing enhanced filling of NL into a growing droplet, a process facilitated by cell death-inducing DFFA-like effector (CIDE) also known as fat-specific protein 27 (FSP27) family of proteins that localize to LD-LD contact sites and promote lipid transfer, LD fusion, and growth in hepatocytes (Xu et al., 2016), and unilocular LD formation in adipocytes (Zhou et al., 2015) (Fig. 3A-C). Future study will provide insights into the role of seipin in LD fusion.

Fig. 3. Lipid droplet assembly cascade at seipin-complex.

A) Organization of LD biogenesis machinery. Seipin-complex shows enrichment in tubular ER regions and establishes functional ER-LD contact site. Assembly of droplets begin with localized activation of the Lipin-complex that produces DAG. ER membrane proteins, FITM2 and Pex30 get recruited at these seipin marked ER subdomains. FITM2 binds to and regulates local DAG levels. Activity of Pex30 results in deformation of the ER subdomain and remodelling of lipids at LD biogenesis sites to accommodate locally produced DAG and/or TAG. These lipids together with local membrane geometry at LD biogenesis sites facilitates recruitment of TAG-synthases, DGATs to catalyse NL production at these ER subdomains. Seipin-complex traps NLs within the hydrophobic helices and the transmembrane domain region prevents these NLs to diffuse into the bulk of the ER membrane thereby triggering nucleation of a nascent LD at these sites. These LDs acquire more NLs, emerge toward the cytoplasm and mature by acquiring class I and class II LD-resident proteins. Seipin-complex also plays a crucial role in the biogenesis of SE-only LDs. B) Monomeric form of seipin. Cartoon depicting yeast and human seipin monomer with key residues that interact with neutral lipids. C) Growth and maturation of LD mediated by LD-LD fusion. Steps of fusion between two adjacent LDs while remaining associated with the seipin-complex. Under conditions of rapid expansion of LDs during oleate loading (Step I), two adjacent seipin foci might become juxtaposed (Step II) resulting in a hemi-fusion between neighbouring LDs facilitated by LD-LD contact site protein CIDE/FSP27 (Step III). A larger LD grows at the expense of smaller LD due to unidirectional transfer of neutral lipids (Step III). Once an LD fully grows, one of the seipin-complex becomes devoid of any LD and hence free seipin foci perhaps moves away to engage in building a new droplet de novo upon demand (Step IV).

Although seipin is critical for establishing proper ER-LD contact site (Chapman et al., 2019; Salo et al., 2016), the factors that determine its localization and regulation is not known. Loss of seipin results in delay in the rate of LD formation with concomitant TAG build up in the ER membrane, leading to aberrant and ectopic LD formation, that are either small, clustered or supersized (Cartwright et al., 2015; Choudhary et al., 2020; Fei et al., 2008; Grippa et al., 2015; Rao and Goodman, 2021; Szymanski et al., 2007; Wang et al., 2014; Wolinski et al., 2011). Surprisingly, aberrant LDs in seipin mutants lack the full complement of LD surface proteins and thus, are not fully functional (Choudhary et al., 2020; Salo et al., 2019; Wang et al., 2016). Recently in C. elegans it has been shown that in a subset of LDs seipin becomes enriched at ER subdomains upon exogenous supplementation of polyunsaturated fatty acids (PUFA), and these domains co-purifies with LDs, hence promoting LD diversity (Cao et al., 2019). Surprisingly, homozygous deletion of worm seipin causes penetrant embryonic lethality due to defects in embryonic eggshell formation, suggesting key role of seipin in regulating lipid homeostasis during embryogenesis (Bai et al., 2020).

More recently, seipin structure has been determined by Cryo-EM providing insights into the role of seipin function and its regulation. The core elements in human, fly and yeast proteins are conserved as it forms a large membrane-embedded oligomeric complex comprising of 11, 12 and 10 monomeric subunits respectively (Arlt et al., 2022; Klug et al., 2021; Sui et al., 2018; Yan et al., 2018). The luminal domain of seipin consists of a hydrophobic α-helix (HH) in the mammalian and fly seipin, that faces toward the inner leaflet of the ER membrane, and has affinity toward TAG. However, yeast seipin lacks HH that is important for TAG nucleation, but this function is provided by Ldb16, a seipin partner protein that forms a stable complex (Klug et al., 2021). Absence of either seipin or Ldb16 from the seipin-complex renders it non-functional, and results in LD formation defect that can be rescued by heterologous expression of human seipin (Grippa et al., 2015; Wang et al., 2014) (Fig. 3A, B).

Another interesting feature of seipin is the presence of two β-sheets, each containing four antiparallel β strands, a fold that shows similarity to lipid binding domain of Niemann-Pick type C2 protein (NPC2), as well as to the C2 domain found in PKCα and synaptotagmin 1 (Sui et al., 2018; Yan et al., 2018). Seipin has been shown to bind to anionic phospholipid PA in vitro and facilitate transfer of lipids from ER to LD in vivo (Salo et al., 2016; Yan et al., 2018). Consistent with this observation, seipin has been implicated in the regulation of localized PA production and to prevent its accumulation in the ER that could hamper with adipogenic regulation of PPAR-gamma, a key transcription factor in adipogenesis (Pagac et al., 2016; Rao and Goodman, 2021; Talukder et al., 2015).

Molecular Dynamic Simulation (MDS) studies revealed important serine residues (S165 and S166) within the HH having a key role in interaction with DAG and TAG molecules, being located at the center of the seipin ring that effectively sequesters NL molecules and thereby aid in lens formation and growth into nascent LDs (Prasanna et al., 2021; Yan et al., 2018; Zoni et al., 2021b) (Fig. 3B). In contrast, yeast seipin, Sei1 does not possess hydrophobic helices, however, TAG nucleation property is provided by the TMD region of the Ldb16 as revealed by MDS, that might complement the function of human/drosophila HH (Klug et al., 2021; Zoni et al., 2021b). It remains to be determined if yeast Sei1 TMD are the key to nucleation of TAG. It was observed that Ldb16 interacted with TAG through their hydroxyl groups of serine (S53, S55, S62) and threonine residues (T52, T61, T62). In concordance with the idea of Ldb16 having predominant role in droplet assembly, yeast with genotype sei1Δ, ldb16Δ and sei1Δldb16Δ showed phenotype which was indistinguishable with defects in LD morphology (Grippa et al., 2015; Wang et al., 2014). Moreover, the yeast seipin TMDs can exist in two separate conformations, thereby providing additional flexibility to the seipin ring to open up and accommodate additional TAG molecules when the seipin-complex is engaged in droplet biogenesis (Arlt et al., 2022). Accordingly, a locking helix or also known as switch domain is important for seipin function as it imparts more flexibility between the two conformations of seipin TMDs (Arlt et al., 2022; Klug et al., 2021). Any disruption in the switch region showed prominent effect on the Sei1 localization, which was seen as large rings within the ER that encircled LDs instead of a foci adjacent to LDs, suggesting that alteration of the switch region might perturb the interactions between the TMDs of neighbouring monomers (Arlt et al., 2022). In addition, mutated switch seipin constructs assembled into smaller oligomers, and failed to rescue growth sensitivity of sei1Δ on terbinafine (Arlt et al., 2022).

Seipin partner LDAF1 is widely conserved and are related to yeast LD Organization (Ldo45, Ldo16) proteins (Eisenberg-Bord et al., 2018; Teixeira et al., 2018). LDAF1 is induced during adipogenesis, and shows interactions with the HH of human seipin that is mediated by clustering of TAG molecules within the seipin ring (Chung et al., 2019; Prasanna et al., 2021). Importantly, seipin-LDAF1 complex co-purifies with TAG, whereas seipin alone does not, implying that a stable association of seipin and LDAF1 is necessary for TAG nucleation (Chung et al., 2019). Upon LD growth, LDAF1 dissociates from the seipin-LDAF1 complex and relocates over the LD periphery, owing to its hairpin type of membrane topology that allows its insertion in ER bilayer as well as LD monolayer (Chung et al., 2019). Taken together, seipin-complex assembles into ring-shaped oligomers that facilitates clustering/sequestration of NLs at LD formation sites in the ER, and the TMD region might be preventing the outflow of NLs from these sites into the bulk of the ER, thereby helping in NL nucleation into nascent LDs (Fig. 3A).

Though the role of seipin in the biogenesis of TAG-only LDs is well established, recent work have investigated the role of seipin in the formation of other class of NLs, such as SE or retinyl esters (RE). Surprisingly, formation of RE-only LDs does not require seipin or TAG (Molenaar et al., 2021). However, seipin and its partner Ldb16 are important for the biogenesis of SE-only LDs, and human seipin complements the defect of a yeast mutants lacking these proteins (Renne et al., 2022) (Fig. 3A). Interactions between hydroxyl-residues found in human seipin or ldb16 with the carboxyl esters present in NLs is important for the formation of SE/TAG containing LDs (Renne et al., 2022). Another study used model membranes and cultured cells to demonstrate that formation of CE containing LDs occurs at seipin enriched ER subdomains, and its nucleation if facilitated by the presence of TAG (Dumesnil et al., 2023). Remarkably, deletion of seipin in steroidogenic tissues of mice, particularly in adrenal, ovary, and testis resulted in significant decrease in LD formation and reduced accumulation of SE-rich LDs thereby impairing steroid hormone production. Further, addition of exogenous lipoprotein cholesterol did not reverse the defect of hormone production in seipin deleted mice (Shen et al., 2022). This suggests prominent role of seipin in the formation of SE-rich LDs and intracellular trafficking of cholesterol. It remains to be determined if SE synthesizing enzymes get recruited at seipin defined ER subdomains and whether SE packaging is coordinated with TAG filling into LDs. Taken together, these new findings suggest how seipin functions in the assembly of LDs containing diverse class of NLs.

5.2. FITM2 an essential gene facilitates droplet assembly

Fat storage-inducing transmembrane (FITM) proteins are evolutionarily conserved polytopic ER membrane proteins (Kadereit et al., 2008). Humans have two FITM proteins, FITM1 is primarily expressed in muscles, and FITM2 that is expressed in all tissues. FITM2 is functionally regulated by peroxisome proliferator-activated receptor gamma (PPARg), and hence is important for adipogenesis by converting pre-adipocytes into mature adipocytes (Kadereit et al., 2008). Over-expression of FITM2 in 3T3-L1 fibroblasts and in mouse liver results in accumulation of TAG rich LDs, whereas absence of FITM2 lead to smaller and fewer LDs (Kadereit et al., 2008). Interestingly, FITM2 has been shown to bind to TAG and DAG in vitro, although they do not have any TAG-synthase activity (Gross et al., 2011). In agreement with this it has been suggested that FITM2 proteins facilitate TAG partitioning from ER membrane into LDs (Gross et al., 2011; Kadereit et al., 2008). In a FITM2 knockdown mice the brown adipose tissue show large but fewer LDs causing lipodystrophy and insulin resistance which indicates their role in regulating LD storage (Miranda et al., 2014). Remarkably, FITM2 mutation in humans causes Siddiqi Syndrome characterized by deafness dystonia syndrome accompanied with motor regression, ichthyosis-like features and signs of sensory neuropathy and low body mass index (Zazo Seco et al., 2017). FITM2 is widely conserved with homologs in Drosophila, Caenorhabditis elegans, and Saccharomyces cerevisiae but not FITM1. Yeast genome encodes two FITM2 homologues, known as Yft2 and Scs3. Lack of yeast FITM2 proteins results in perturbation of ER membrane homeostasis and scs3Δ cells show auxotrophy for inositol (Moir et al., 2012; Renvoise et al., 2016). In yeast FITM2 mutants LDs remain arrested in the ER and fail to emerge toward cytoplasm, however, they still remain accessible to cytoplasmic proteins (Choudhary et al., 2015). Remarkably, whole-body knockout of FITM2 in mice causes lethal enteropathy and dysregulation of LD formation as well as bile acid transport (Goh et al., 2015). Lack of the sole FITM2 protein in C. elegans is lethal, suggesting that FITM2 fulfils a hitherto uncharacterized but essential function in lipid metabolism (Choudhary et al., 2015).

Recent work suggest FITM2 perhaps could be lipid phosphatase/phosphotransferase since mutation of conserved residues impairs FITM2 activity in vivo (Hayes et al., 2017). It has been shown that FITM2 proteins become transiently enriched at LD biogenesis sites in the ER, and might play a crucial role in regulating DAG levels at these sites, thereby controlling emergence of LDs from the ER (Choudhary et al., 2018) (Fig. 3A). Recently it has been reported that FITM2 is an acyl-CoA diphosphatase enzyme that cleaves the phosphoanhydride bond of acyl-CoA to yield acyl 4’-phosphopantetheine and adenosine-3’, 5’-bisphosphate (Becuwe et al., 2020). FITM2 protein also interacts with ER tubule forming proteins REEP5, Rtn4 and Septins in C. elegans indicating their role in early stages of LD formation (Chen et al., 2021a). Moreover, FITM2 showed higher expression in tissues with hepatocellular carcinoma (HCC) indicating its correlation with larger tumour size and high microvascular invasion. FITM2 knockdown correlated with downregulation of Caveolin-1 protein which further inhibited tumour wound healing and migration indicating that FITM2 promotes HCC by helping in formation of caveolae. (Chen et al., 2021b). Altogether, the current available data suggest important role of FITM2 in cellular lipid metabolism. It is plausible that FITM2 might act as sinks of DAG at LD biogenesis sites, due to its high affinity binding to DAG, and might facilitate transfer of DAG to TAG-synthase upon their recruitment at LD biogenesis sites. By doing so FITM2 perhaps prevents toxic build-up of unbound DAG at LD biogenesis sites and thus DAG levels are significantly elevated at LD biogenesis sites in cells lacking FITM2 proteins thereby resulting in aberrant LD emergence from the ER membrane. Future work will determine the precise mode of action of FITM2 proteins.

5.3. Activity of membrane shaping proteins play a critical role at LD biogenesis sites

ER membrane comprises of sheets and tubules. Proteins that induce ER shape, such as reticulons, atlastins, and REEPS, have been shown to play role in LD biogenesis (Klemm et al., 2013; Renvoise et al., 2016). Interestingly, nucleation of LD is favoured in ER tubules rather than sheets since the nucleation energy of these regions is lower due to high curvature stress hence proteins that modulate tubulation of ER membrane would positively impact droplet biogenesis (Kassan et al., 2013; Santinho et al., 2020). With the emergence of a recent family of ER shaping protein, Pex30/MCTP2 in mammals, it has been demonstrated that they induce tubulation of ER subdomains at which biogenesis of both peroxisomes and LDs occurs (Joshi et al., 2018; Wang et al., 2018). Deletion of sole MCTP2 homolog in C. elegans leads to reduction in the number and size of LD compared to WT animals, however, this deletion does not affect the embryo viability (Joshi et al., 2018). Another member of MCTP family, MCTP1 has recently been shown to tubulate ER subdomains and play role in LD biogenesis similar to MCTP2 (Joshi et al., 2021). MCTPs colocalize with LD biogenesis protein seipin, show affinity towards charged lipids, and are likey important for establishing proper ER-LD contact sites (Joshi et al., 2021).

Pex30 in yeast and MCTP2 harbour a reticulon homology domain (RHD) that facilitates insertion of Pex30/MCTP2 in the ER bilayer in tubules and edges of ER sheets, thereby inducing positive membrane curvature in the ER (Joshi et al., 2016; Joshi et al., 2018; Wang et al., 2018). It has been shown that Pex30 colocalize with seipin marked ER sites, and a double deletion of both proteins results in defects in LD formation with concomitant accumulation of NL in the ER membrane (Joshi et al., 2018; Wang et al., 2018). Intriguingly, lack of seipin, mislocalizes Pex30 to a single puncta in the ER, however, loss of Pex30 although does not affect localization of seipin, but these seipin foci fail to recruit TAG synthesizing enzymes and hence are non-functional (Choudhary et al., 2020; Joshi et al., 2018). Deformation of ER subdomain engaged in LD biogenesis would prime these sites to accumulate DAG and/or TAG molecules that would drive the recruitment of LD biogenesis factors thereby assisting in droplet assembly (Fig. 3A). Thus Pex30 would serve a prominent role at these discrete ER sites, lack of which impairs NL nucleation.

Apart from membrane tubulation, Pex30/MCTP2 also contain a C-terminal Dysferlin (DysF) domain (Joshi et al., 2016), that is conserved in human ferlins, including dysferlin, and myoferlin that functions in membrane repair and lipid remodelling (Bulankina and Thoms, 2020). DysF domain is crucial for Pex30 function in LD formation, implying its role in modulating localizaed ER membrane properties (Ferreira and Carvalho, 2021). Previously it has been shown that feeding yeast cells on oleate to induce large LDs also stimulates peroxisome biogenesis in a way that they invade each other, suggesting that both these organelles share an intimate relationship (Binns et al., 2006). Similarly, other peroxisome biogenesis proteins such as Pex19 and Pex3 have been also implicated in LD formation (Schrul and Kopito, 2016). Taken together biogenesis of both LDs and peroxisomes appear to be regulated by ER shaping proteins and induced by similar conditions, however, the complete mechanisms of action remain to be determined.

5.4. LD coat proteins are important to stabilize growth and maturation of LD

Once LDs are formed, they grow and mature inside the cytoplasm. Various LD-associated proteins are targeted to LD monolayer and are classified into two categories; class I and class II. Class I represents proteins that translocate along the plane of the ER bilayer onto LD monolayer via lateral diffusion due to presence of hairpin type of membrane anchor that allows bi-directional transport of these proteins. Acyltransferase Dga1 belong to class I protein, that is first inserted in the ER and later moves over the periphery of mature droplets, hence the protein can catalyse TAG production in the ER as well as on mature LDs (Fig. 3) (Jacquier et al., 2011). On the other hand Class II comprises of proteins that are targeted from cytoplasm to LD surface due to presence of amphipathic helices (AH), such as perilipins in mammals, that are cytosolic in the absence of LDs, however, are recruited to LD surface upon induction of LD formation (reviewed in (Dhiman et al., 2020; Olarte et al., 2022; Olzmann and Carvalho, 2019)) (Fig. 3A).

LD scaffolding coat proteins such as mammalian perilipins (PLIN) are the most abundant and best characterized LD-associated proteins. Mammals have five PLIN genes, and additional splice variants, with differential tissue expression (Lu et al., 2001; Miura et al., 2002). Similarly, oleosins in plants, lipid storage droplet (LDS1/2) in Drosophila, MDT-28 and DHS-3 in C. elegans, and microorganism lipid droplet small (MLDS) in bacteria function as LD coat proteins and prevent fusion of adjacent LDs (Na et al., 2015; Teixeira et al., 2003; Yang et al., 2012). Perilipin-1 (PLIN1) is a major constituent of LD coat in adipocytes and steroidogenic cells, and plays an important role in storage and turnover of neutral lipids in these cells by allowing lipases to access LD stores (Bickel et al., 2009). Overexpression of PLIN1 enhances LD formation (Brasaemle et al., 2000) whereas PLIN1 knockout mice are lean and resistant to diet-induced obesity (Martinez-Botas et al., 2000; Tansey et al., 2001). Perilipin-2 (PLIN2; previously ADRP, adipophilin) is predominantly expressed in non-adipose tissues in which it is tightly associated with the LD surface (Londos et al., 2005). PLIN3, 4, 5 show cytosolic or ER localization (Bartholomew et al., 2012; Skinner et al., 2009). Recently, Pet10 (renamed as Pln1), a yeast orthologue of mammalian perilipin was identified that coats nascent droplets and is crucial for stabilizing LD structure and functions in collaboration with seipin and FITM2 proteins (Gao et al., 2017). Cells missing Pet10 exhibit delayed LD biogenesis, decreased TAG accumulation and fusogenic LDs (Gao et al., 2017).

PLINs consists of repeat segments of AH, analogous to those found in apolipoproteins, and alpha-synuclein protein associated with Parkinson’s disease. These AH segments are important for the association with the LD surface, perhaps by recognizing lipid packing defects over LD monolayer (Copic et al., 2018; Rowe et al., 2016). A greater degree of hydrophobicity found in PLIN4 due to long AH segment enables PLIN4 association with LD core instead of phospholipid monolayer, acting as an alternative surfactant to coat LD surface (Copic et al., 2018). This property might be beneficial in times when LDs are dramatically expanding and hence supply of phospholipids to coat LD monolayer becomes limiting. Heterologous expression of oleosins and PLINS in yeast promoted accumulation of NL in the ER and induced LD formation (Jacquier et al., 2013). Interestingly, if PLINs are appended with an ER membrane protein to be anchored in the ER bilayer, they induce close apposition of ER with the LD by promoting redistribution of ER membrane around LDs (Khaddaj et al., 2022). Intriguingly, if these PLINs are expressed in cells devoid of NL synthesis, they induce membrane domain formation that harbors LD-like properties, and when LD biogenesis is stimulated in these cells, de novo LDs form at these ER subdomains (Khaddaj et al., 2023). These domains are enriched in DAG, and LD biogenesis factors essential for LD formation, such as seipin and Pex30 (Khaddaj et al., 2023). This reinforces the idea that LD forming domains are pre-defined independent of NL synthesis. In a recent study it has been shown that PLIN3 gets recruited onto the nascent LDs at the DAG-enriched sites in the presence of Perilipin-ADRP-Tip47 (PAT) domain and 11-mer AH repeats (Choi et al., 2023). Taken together, LD coat proteins play vital role in the life cycle of LDs; playing an active role in the early steps of LD biogenesis, and serves to stabilize the surface of mature LDs by shielding the NL core from unregulated access to lipases, thereby allowing proper growth and maturation of LDs.

5.5. Future perspectives

Lipid homeostasis in the ER is fundamental to cell physiology. Recent advances in understanding how cells control ordered nucleation of LDs at defined ER subdomains have provided insights into the novel mechanisms underlying LD biogenesis. A broader understanding has emerged that seipin-complex plays an instrumental role in defining the ER locations at which droplets need to be assembled. However, how these sites are selected and regulated awaits future investigation. What are the molecular cues (localized membrane lipid and protein constituents) that allows seipin to either be trafficked at these defined ER domains or does seipin itself creates such a platform in the ER? The molecular details of how seipin-complex interacts with its partner proteins to efficiently sequester NL from the ER bilayer and facilitates its transfer to growing LD remains to be determined. Future work will reveal if seipin structure changes during growth and maturation of an LD to facilitate exchange of lipids and proteins between ER and LD.

Most of the studies have been performed with TAG rich LDs, which tempts to speculate if similar mechanisms operate for SE containing LDs. How does LD biogenesis factors responsible for the assembly of TAG-only LDs differ from the ones required for packaging of SE-only LDs? Do TAG or SE filling into LDs occur simultaneously? How are these two processes regulated in real time? What are the mechanisms of altered protein trafficking in seipin deficient cells? How does lack of seipin manifests in lipodystrophy and/or seipinopathy diseases? What is the role of ER shaping proteins in regulating lipid and protein composition at LD biogenesis sites? Deciphering the role of key players in LD assembly is likely to bring novel insights into the better understanding of defects associated with lipid storage. Thus, in-depth research on LD mechanism can bring about surprising discoveries in the coming years.

Data Availability

Data will be made available on request.