Early steps in the birth of four membrane-bound organelles - peroxisomes, lipid droplets, lipoproteins, and autophagosomes

Abstract

Membrane-bound organelles allow cells to traffic cargo, and separate and regulate metabolic pathways. While many organelles are generated by the growth and division of existing organelles, some can also be produced de novo, often in response to metabolic cues. This review will discuss recent advances in our understanding of the early steps in the de novo biogenesis of peroxisomes, lipid droplets, lipoproteins, and autophagosomes. These organelles play critical roles in cellular lipid metabolism and other processes, and their dysfunction causes or is linked to several human diseases. The de novo biogenesis of these organelles occurs in or near the endoplasmic reticulum (ER) membrane. This review summarizes recent progress and highlights open questions.

Introduction

Membrane-bound organelles are a defining characteristic of eukaryotic cells. Organelles are produced in two ways. One is the growth and division of existing organelles; new mitochondria, for example, only come from other mitochondria. Most membrane-bound organelles are also produced by growth and division, including the organelles of the secretory and endocytic pathways. Some of them, however, can be formed de novo, often in response to stress, changes in nutritional status, or pathogenic infection. A few of the best-characterized instances of de novo organelle biogenesis take place in or near the ER. For example, de novo biogenesis of peroxisomes happens from the ER membrane (1) and the earliest stages of LD biogenesis occur within the bilayer of the ER membrane (2, 3). Another type of organelle biogenesis occurs close to the ER, for example, autophagosomes. Autophagosomes are thought to originate from a “seed membrane” near the surface of the ER (about 50nm) and other organelles (4, 5). Lipoproteins are born de novo in the lumen of the ER and are finally secreted in blood circulation (6, 7).

Peroxisomes, LDs, lipoproteins, and autophagosomes have important functions in cellular metabolism and organismal physiology. The de novo biogenesis of these organelles is therefore highly regulated. Aberrant biogenesis of these organelles either causes or is linked to several human diseases (8–12) and a better understanding of the mechanisms of biogenesis of these organelles could aid the development of effective therapies for these diseases.

In the last few years, there has been significant progress in our understanding of the de novo biogenesis of peroxisomes, LDs, lipoproteins, and autophagosomes. We now know in more detail how functional domains are produced in the ER and other membranes (13, 14). Interestingly, specialized domains within the ER membrane or on the surface of other organelles facilitate de novo biogenesis of the four organelles (13, 15, 16). We will discuss what we have learned about these domains and how they facilitate organelle biogenesis de novo. There has also been a significant conceptual advancement in the models of how some lipid transport proteins may facilitate the expansion of organelle membranes to support de novo biogenesis. We will discuss ‘tube-like’ lipid transport proteins and how they might contribute to organelle biogenesis.

A. Peroxisomes

Peroxisomes play critical roles in lipid metabolism, including the β-oxidation of fatty acids and the synthesis of lipids needed for myelin generation. Defects in peroxisome formation and function cause neurodegenerative diseases such as X-linked adrenoleukodystrophy and Zellweger Syndrome. More recently, it has also become clear that peroxisomes are protective organelles that play important roles in human health (17). Lipids synthesized in peroxisomes have recently been found to play a role in ferroptosis, which is an iron-dependent, non-apoptotic cell death pathway caused by oxidative damage (18). Specifically, peroxisomes help synthesize a cytoprotective class of ether glycerophospholipids known as plasmalogens, which putatively suppress ferroptosis.

It was once thought that peroxisomes, like mitochondria, were semi-independent organelles that were only formed by the growth and division of pre-existing peroxisomes. However, it was later discovered that cells devoid of peroxisomes can reform them. Cells lacking a biogenesis factor like PEX3 can reform peroxisomes when PEX3 is re-expressed, indicating that peroxisomes can be produced de novo (19). The ER plays a critical role in peroxisome biogenesis in S. cerevisiae (20). Mature peroxisomes are derived from pre-peroxisomal vesicles (PPV) that originate in specialized ER subdomains and then mature into fully functional peroxisomes (Fig. 1) (21). There may be more than one type of PPV, and they may fuse to form mature peroxisomes (20). The mechanism and regulation of PPV production are not understood. Important questions include how proteins are sorted into nascent PPVs and what machinery buds PPVs from the ER. In vitro PPV budding assays using ER-derived membranes have been developed and are providing mechanistic insight into PPV production. Recently, one such assay was used to show that ESCRT-III (endosomal sorting complex required for transport III) is required for PPV scission from the ER membrane (Fig. 1) (22). The ESCRT-III protein complex is required for other membrane scission events in cells (23). Because a good deal is known about ESCRT-III regulation, this finding opens the way to future work on how PPV budding from the ER is controlled.

Figure 1. De novo biogenesis of peroxisomes.

Peroxisomes are synthesized de novo from specialized sub-domains of the ER. Proteins that organize at these domains and regulate peroxisome biogenesis are illustrated. Peroxisomes are first synthesized as pre-peroxisomal vesicles (PPVs) that import (shown by the arrow) peroxisomal matrix proteins (black dots) to form a mature peroxisome. In human cells, PPVs are also reported to originate from mitochondria. They fuse with ER-derived PPVs to form mature peroxisomes. Question marks indicate major unanswered questions about PPV biogenesis. Whether there is more than one kind of PPV emerging from the ER is also debated. (81, 82)

Remarkably, some PPVs may arise from mitochondria in mammalian cells; these vesicles are thought to fuse with PPVs generated in the ER to form mature peroxisomes (24). Further research will determine whether PPV production on the surface of mitochondria happens ubiquitously in mammalian tissues and how it is regulated (Fig.1).

The last few years have seen progress in characterizing the sub-domains of the ER where PPVs form. Proteins with ER-shaping reticulon homology domains (RHDs) play a regulatory role in peroxisome biogenesis (25–27). Our lab found that Pex30, a protein known to control peroxisome size and ER-peroxisome contacts, has an RHD and is found at sites of peroxisome biogenesis (Fig. 1) (28). Interestingly, Pex30 is also at sites of lipid droplet biogenesis (16, 29), suggesting a link between the de novo biogenesis of both organelles (30). The mammalian RHD-containing proteins MCTP1 and MCTP2 may be functionally similar to yeast Pex30 and localize to ER subdomains where nascent LDs form and regulate LD formation (Fig.1) (31). Whether MCTP1 and MCTP2 regulate peroxisome biogenesis remains unknown. An important outstanding question is how RHD domain-containing proteins contribute to organizing ER subdomains where peroxisome and LD biogenesis occur. They may help determine the lipid and protein composition of these sites and could contribute to the coregulation of lipid metabolism by LDs and peroxisomes, which has been suggested by recent studies (32, 33).

Both the de novo formation of peroxisomes and peroxisome growth requires the expansion of PPV and peroxisome membranes. This could occur by vesicular fusion, but nonvesicular lipid transport may also play a role. Recent work has implicated VPS13D in peroxisome formation (35). The protein is a member of a newly identified family of tube-like proteins thought to facilitate lipid exchange at membrane contact sites (36). Mutations in VPS13D cause an inherited autosomal recessive disorder that is characterized by spinocerebellar ataxia (37). VPS13D probably facilitates lipid transport from the ER to nascent peroxisomes to support the growth of peroxisomal membranes, though this remains to be shown. Consistent with this, a fraction of VPS13D resides at the ER-peroxisome contact sites (Fig. 1) (38). In an independent study conducted in yeast, the VPS13D homolog has also been suggested to play a role in peroxisome membrane expansion (39).

More peroxisome biogenesis factors remain to be discovered. We do not know the machinery that generates PPVs. There are also likely to be peroxisome biogenesis regulators that are specific to mammalian cells. Most peroxisome biogenesis factors were identified by genetic screens in yeast (34). While these proteins are highly conserved in humans (1), there probably are mammalian cell-specific factors yet to be identified.

B. Lipid droplets

LDs are lipid storage organelles present in all eukaryotes. They are composed of a hydrophobic core of neutral lipids, typically triglycerides and steryl esters, encircled by a phospholipid monolayer containing proteins (40). LDs are central to lipid metabolism and disorders in LD biogenesis and function are associated with human diseases (40). LDs form de novo in the ER membrane. LD formation is thought to begin when neutral lipids in the ER membrane undergo a phase transition, forming lens-like structures between the leaflets of the ER membrane (40). As more neutral lipid is incorporated into a nascent LD, it grows and emerges from the ER membrane. How proteins modulate the nucleation and maturation of LDs is being intensely studied.

The protein seipin plays a central role in LD biogenesis (Fig. 2). Cells lacking seipin have defects in LD formation and have LDs with abnormal sizes. Mutations in seipin cause severe lipodystrophy (40). Several mechanisms have been proposed for seipin. It may directly facilitate LD nucleation, determine the direction of LD emergence from the ER membrane, and modulate the lipid and protein composition of the surface of LDs.

Figure 2. Biogenesis of LDs.

LD biogenesis occurs on the cytoplasmic side of the ER membrane from specialized ER sub-domains. The proteins that orchestrate the initiation, growth, and maturation of LD biogenesis are illustrated. (42)

Two recent structures of S. cerevisiae seipin and molecular dynamics (MD) simulation of seipin have yielded important new insights into how it could nucleate LD formation (Fig. 2A) (41–43). Seipin forms a homo-decameric supercomplex with a central cavity. The luminal domains of the seipin complex and the trans-membrane helices are essential for their function in LD biogenesis. The trans-membrane segments have neutral lipid binding sites in seipin that may facilitate LD nucleation within the center of the seipin oligomer (41, 43).

Most studies on the role of seipin in LD biogenesis have focused on how seipin facilitates TAG nucleation and incorporation into LDs. Two recent studies have investigated the biogenesis of LDs containing mostly other neutral lipids, namely cholesteryl ester (CE) or retinyl ester. It has been suggested that the formation of LDs containing only or mostly retinyl ester does not require seipin or TAG (44). Another study used bio-mimicking membranes and cultured human cell lines to show that CE nucleates LD biogenesis when a critical ratio of CE to phospholipids is achieved in membranes and can be facilitated by TAG (45). The authors speculate that a similar process occurs in the ER membranes of cells initiating lipid droplet biogenesis. Biogenesis of LDs containing primarily CE occurs at regions of the ER where seipin is localized and is facilitated by the presence of triglycerides at these sites (45). Together, these studies add to our knowledge of how seipin functions in LD biogenesis.

Fat storage-inducing transmembrane protein 2 (FIT2) also plays a role in LD biogenesis (Fig. 2). A few years ago, our lab showed that in cells lacking FIT2, LDs fail to emerge from ER membrane and instead grow within the ER membrane (3). More recent work suggests this defect may be caused by the altered lipid composition of the ER at sites of LD biogenesis (46). In mammalian cells, the interaction of FIT2 with ER-tubulating proteins Rtn4 and REEP5 could also help facilitate LD emergence from the ER (47). FIT2 has been shown to have acyl-coenzyme A diphosphatase activity, though how this relates to other phenotypes of cells lacking FIT2 remains to be determined (48).

Nascent LDs are thought to form in specialized domains of the ER. Several proteins become enriched at these sites as LDs form and mature. In budding yeast, seipin and the protein Nem1, a phosphatase that regulates lipid metabolism, may either help generate sites for LD biogenesis or arrive at them soon after LD biogenesis begins (Fig. 2) (10). Later, other proteins arrive in the following order: a FIT2 protein, Pex30, triglyceride synthesizing enzymes, and, finally, proteins found on the surface of new LDs (Fig. 2) (10). How these proteins work together, and coordinate LD formation at these ER sites remains to be determined.

How proteins are targeted to nascent LDs is another important question and there has been a good deal of progress in the last few years. Mechanisms of protein targeting to LDs must be unique because the limiting membrane of LDs is formed by a phospholipid monolayer instead of a bilayer. Some proteins reach the surface of nascent LDs from the cytoplasm, while others diffuse from the ER membrane onto the surface of nascent LDs. Interestingly, two distinct pathways allow proteins to move from the ER to LDs during biogenesis or to mature LDs. Proteins destined for LDs are either co-translationally or post-translationally inserted into the ER membrane and then targeted to LDs (49). The co-translational targeting of LD proteins is regulated by the coordinated action of the ER membrane protein complex (EMC) complex and the Sec61 translocon complex. A post-translationally targeted LD protein requires the Pex3-Pex19 system for its efficient targeting; these proteins also play a role in peroxisome biogenesis (Fig. 2) (49). This is consistent with the observation that peroxisomes and LDs share ER subdomains during their biogenesis. Another recent study has found that, in Drosophila, proteins reach nascent and mature LDs by two distinct pathways (50). Some proteins are targeted early during biogenesis while others reach the surface of LDs later. It is unclear how the arrival of proteins to LDs is determined; the current paradigm is that seipin somehow controls the access of proteins to nascent LDs. Protein targeting to mature LDs has been suggested to occur via ER-LD contacts where membrane bridges between these organelles facilitate protein trafficking. How these bridges form and how various targeting machinery work together to traffic proteins to LDs at late stages of LD biogenesis remain to be investigated (50).

C. Lipoproteins

Lipoproteins are LD-like particles that transport lipids in the circulatory system. The most abundant lipoproteins are produced de novo in the ER lumen of hepatocytes and enterocytes and are secreted out of the cells. Understanding lipoprotein biogenesis is important because elevated levels of circulating lipoproteins increase the risk of developing atherosclerosis (51). While statins remain an effective first-line treatment for reducing circulating lipoproteins, there is a need to develop additional drugs, which requires new insights into lipoprotein biogenesis.

Lipoproteins are structurally like LDs, with a core of neutral lipids surrounded by a membrane monolayer containing proteins (Fig. 3A) (52). Unlike LDs, most lipoprotein particles contain a large protein, apolipoprotein B (ApoB), that coats the surface and is required for biogenesis. The ApoB produced by hepatocytes is 540 KDa (Fig. 3A). While the basic outline of the mechanism of biogenesis of nascent VLDL (very low-density lipoprotein) and other lipoproteins that form in the ER lumen has been known for some time, much remains to be learned about the biogenesis of these particles. The lipidation of ApoB is thought to begin co-translationally, as it is being translocated into the ER lumen. ApoB lipidation requires an ER luminal protein known as microsomal transfer protein (MTP) (53) (Fig. 3A). Mutations in MTP cause an inherited disorder called abetalipoproteinemia, in which patients have highly reduced levels of lipoproteins in their blood and suffer from lipodystrophy (12, 53). MTP directly interacts with ApoB (12). In vitro, lipid transfer activity and a recent crystal structure of MTP indicate that MTP has lipid transporting activity (12, 54). Together, this means that MTP may directly lipidate ApoB during VLDL biogenesis. Another possible source of lipids for VLDL biogenesis is LDs in the ER lumen (Fig. 3A) (55). The biogenesis of these luminal LDs also requires MTP (56), though the mechanism is not known and, it should be noted, LDs formed on the cytoplasmic surface of the ER are not known to require cytoplasmic MTP-like proteins. The ER-resident protein Vmp1 has recently been shown to play a role in lipoprotein biogenesis in fish, mice, and humans (Fig. 3A) (57, 58). Cells depleted of Vmp1 accumulate neutral lipids in the ER, suggesting they have a defect in lipoprotein maturation or trafficking. Interestingly, VMP1 is a scramblase and thus could affect the availability or transfer of phospholipids onto nascent lipoproteins or perhaps luminal LDs (59, 60). Another ER membrane protein recently shown to play a role in nascent VLDL particle lipidation is TM6SF2 (Fig. 3A) (61). The mechanism remains to be determined.

Figure 3. Biogenesis of VLDL and Lamellar Bodies.

A. VLDL biogenesis begins in the lumen of the ER. Nascent VLDL particles mature in the compartments of the secretory pathway and are secreted out of the cell. ApoB is co-translationally lipidated as it being moved through the translocon (green open cylinder). MTP transfers lipids to ApoB, which is indicated by arrows. A nascent VLDL and a luminal LD are indicated. MTP transfers neutral lipids to luminal LDs (indicated by straight arrows). The lipid scramblase activity of VMP1 is indicated by bidirectional arrows. TM6SF2 (indicated in the figure) may help in the lipidation of ApoB. The question mark indicates mysteries in the mechanism of the process. (54)

B. Lamellar bodies are lipoproteins that are used to produce surfactant in the lungs. LBs are synthesized, processed, and secreted by the lung alveolar epithelial cells. The chief protein component of LBs is SP-B, which undergoes pH-dependent proteolytic cleavage (cleavage junctions are indicated by scissors) before it forms a mature lipoprotein and is secreted. The pH range of ER and Golgi is indicated (83). SP-B is initially lipidated in the lumen of the ER and is further lipidated when it reaches compartments with lower pH. Ultimately, concentric rings of membranes are generated by the SP-B protein (green).

ER-mitochondria contact sites have recently been suggested to play a role in VLDL biogenesis. In rat liver, mitochondria are often wrapped by the rough ER – termed “wrappER” (62). These may be sites of VLDL biogenesis (Fig. 3A) (62). Downregulating ER-mitochondria contact sites in rat liver cells strongly compromises VLDL secretion (62). A follow-up study suggests that VLDL biogenesis might be coordinated by a tripartite contact of peroxisomes, mitochondria, and rough ER (Fig. 3A) (63). Production of VLDL in regions of ER at contact sites may allow signals from adjacent organelles to regulate the VLDL biogenesis machinery in the ER lumen.

D. Autophagosomes

Autophagy clears aggregated proteins and dysfunctional or unnecessary organelles from cells (64). These cargoes are compartmentalized into a double membrane-bound organelle known as the autophagosome. Autophagosomes form de novo from a vesicular ‘seed’ that grows and eventually seals to form an enclosed double membrane structure. Cells can produce multiple autophagosomes at once and each autophagosome fuses with a lysosome (or the vacuole in yeast), degrading the cargo contained in the autophagosome (64).

Autophagosome biogenesis is being intensively studied by many groups. There are two main questions about the early stage of autophagosome membrane biogenesis. One is the origin of the pre-autophagosomal structure (PAS), a membranous structure that is the precursor for forming autophagosomes. The second question is about the sources of the membrane needed to grow the autophagosome, which occurs in minutes. Both questions have been hotly debated for some time. There have been several notable developments in the last few years. PAS initiates when phase-separated early autophagy machinery proteins form a droplet and bind to a membrane source (Fig. 4A) (65, 66). A recent study suggests that PAS is synthesized by the fusion of Golgi- and endosome-derived membranes (Fig. 4A) (67). The study finds that the biogenesis of PAS is regulated by the SNARE proteins STX17 and VAMP7 and the lipid-transporting protein E-SYT2, however, the mechanism and regulation of this pathway remain to be understood (67). Interestingly, the study also found that a SARS-CoV-2 infection hinders PAS biogenesis by the activity of a coronaviral protein, nsp6. Inhibition of autophagosome biogenesis is presumably required for efficient Covid infection. (67).

Figure 4. Biogenesis of autophagosomes and CASMs.

A. Early autophagy proteins form the autophagy initiation complex. They assemble into a liquid-like droplet (PAS) assembling on the surface of the vacuole. It is also reported that PAS originates from a mixed endosome- and Golgi-derived membranes. Proteins at ER-autophagosome contact sites thought to facilitate lipid transport to the nascent autophagosomes are indicated. A straight arrow through Atg2 indicates the direction of phospholipid transport. Bidirectional round arrows indicate lipid scrambling activity of a protein. Autophagosome expansion may be driven by the de novo biosynthesis of phospholipids. The pathway for this biosynthesis is indicated. The question mark indicates that the PAS biogenesis mechanism remains to be determined. (69,71)

B. CASMs are generated proximal to the Golgi complex during a bacterial or viral invasion. They are critical to a host defense mechanism termed xenophagy (76). They are characterized by a single membrane-bound organelle that is decorated by lipidated LC3. The biogenesis of non-canonical autophagosomes happens downstream to the activation of the STING pathway, in which the activated STING receptor upon binding to its ligand (cGAMP) is translocated from the ER to the Golgi. Two known essential factors for the biogenesis of CASM are the WD40 domain-containing protein ATG16L1 in complex with ATG5 and ATG12, and an active V-ATPase.

How the autophagosome membrane rapidly expands is another controversial question in autophagosome biogenesis. Autophagosomes had been thought to grow by vesicular fusion but in recent years it has been proposed that non-vesicular lipid transport could be a major contributor to the growth of the autophagosomal membrane. Atg2 is a tube-like lipid transport protein required for autophagosome biogenesis (60, 68–70). It has been proposed to act as a conduit that allows lipids to flow between closely apposed membranes (60). It forms a bridge between the ER and nascent autophagosomes and may mediate high-volume phospholipid transport to the growing autophagosome (Fig. 4A) (60, 68–70). Atg2 interacts with two scramblases in the ER, TMEM41B and VMP1, and one on the autophagosomal membrane, ATG9 (Fig. 4A) (60, 71–73). The scramblases may facilitate lipid transport by equilibrating lipids between the leaflets of the ER and autophagosomes at sites of transport (Fig. 4A) (60). What drives lipid import into growing autophagosomes remains to be determined. Lipid synthesis at the ER-autophagosome contact sites plays a role in this. In yeast, the enrichment of fatty acyl-CoA synthase at these sites has been found to promote autophagosome growth, suggesting that lipid synthesis at ER-autophagosome contacts could drive the net movement of phospholipids to autophagosomes (Fig. 4A) (74).

Future Directions

The past few years have seen tremendous progress in our understanding of how de novo biogenesis of membrane-bound organelles occurs, but many questions remain. Biogenesis of the organelles discussed in this review occurs at or near sub-domains of the ER. How these domains form and whether they have a lipid composition that differs significantly from the rest of the ER remains to be determined. Better methods for isolating and characterizing these domains are necessary. It is also becoming clear that other organelles participate in de novo organelle biogenesis by influencing the ER via membrane contact sites or, in the case of degradative organelles, by providing seed vesicles that are expanded into autophagosomes. Thus, while much remains to be learned, we look forward in the next several years to significant progress in our understanding of the mechanism and regulation of de novo biogenesis both for the organelles discussed here as well as others; Box 1 shows intriguing new recent discoveries. There are almost certainly other surprises to come.

Box 1. Recent discoveries: a novel membrane biogenesis mechanism (A) and a new puzzle in organelle biogenesis (B).

A. Lamellar bodies (LBs)

LBs are a type of lipoprotein that is secreted by alveolar type II cells to generate pulmonary surfactant. LBs contain concentric rings of membranes surrounded by a limiting membrane. The biogenesis of these organelles remained enigmatic until recently. It has long been known to require the lung surfactant protein SP-B, which is synthesized as a zymogen called pre-pro-SP-B. The protein undergoes a pH-dependent proteolytic cleavage during its maturation and secretion (Fig. 3B). A recent study shows that proSP-B protein is a lipid transporter that sequesters phospholipids to synthesize nascent lamellar bodies (Fig. 3B) (75). Cleavage of the pro-SP-B to form SP-B allows it to pack lipids into concentric membrane rings, forming the central structure of lamellar bodies (Fig. 3B) (75). Like ApoB, proSP-B probably begins acquiring lipids in the ER lumen, though it does not seem to require an accessory lipid transport protein like MTP. The nascent LBs formed in the ER are further lipidated as they traffic through the secretory pathway. This study opens the possibility that other luminal lipid-transfer proteins might synthesize membranes de novo.

B. Conjugation of Atg8 to Single Membranes (CASM)

When human cells are infected with Salmonella or the Influenza virus, they respond by producing a single-membrane organelle, currently known as CASM. It is produced by a pathway that requires only some of the proteins necessary for canonical autophagosome biogenesis (Fig. 4B) (76, 77). Production of this single-membrane autophagosome-like organelle is an innate immune response in humans that works downstream to the activation of the STING pathway (Fig. 4B) (78). Interestingly, a Salmonella effector protein, SopF, hinders the biogenesis of this organelle, allowing Salmonella to evade host defenses (76). CASM biogenesis happens near the Golgi apparatus (78). Although it has been suggested that the source of the seed membrane necessary to produce these organelles is the Golgi complex, another study suggests that the source is membranes in the endo-lysosomal system (79). There are many unanswered questions. One is whether CASM is an organelle with an independent function. A second is how CASM works as a host defense mechanism. From a biogenesis standpoint, whether lipid transport from the ER via Atg2 (80) is required to support CASM membrane expansion is an intriguing possibility.