Mechanisms of COPII coat assembly and cargo recognition in the secretory pathway

Abstract

One third of all proteins in eukaryotes transit between the endoplasmic reticulum (ER) and the Golgi to reach their functional destination inside or outside of the cell. During export, secretory proteins concentrate at transitional zones of the ER known as ER exit sites (ERES), where they are packaged into transport carriers formed by the highly conserved coat protein complex II (COPII). Despite long-standing knowledge of many of the fundamental pathways that govern traffic in the early secretory pathway, we still lack a complete mechanistic model to explain how the various steps of COPII-mediated ER exit are regulated to efficiently transport diverse cargoes. In this Review, we discuss the current understanding of the mechanisms underlying COPII-mediated vesicular transport, highlighting outstanding knowledge gaps. We focus on how coat assembly and disassembly dictate carrier morphogenesis, how COPII selectively recruits a vast number of cargo and cargo adaptors, and finally discuss how COPII mechanisms in mammals might have adapted to enable transport of large proteins.

Introduction

One out of three newly synthesised proteins in eukaryotes enters the secretory pathway to be sorted to its intra- or extracellular destination. The secretory pathway is an obligate route for most proteins that are translocated into the ER during their synthesis and then have to be relocated to their functional compartment. Proteins that depend on this pathway include most membrane proteins such as ion channels and receptors as well as soluble secreted proteins such as hormones, immunoglobulins and components of the extracellular matrix (ECM). Secretory proteins are first transported from the ER to the Golgi apparatus, where they are post-translationally modified, for example by acquiring complex glycosylation, and sorted either to other cellular membranes or to the extracellular environment (Figure 1A). ER exit, which occurs at specific domains of the ER called ER exit sites (ERES)^1,2, is mediated by the coat protein complex II (COPII), an essential and highly conserved cytosolic complex that remodels the ER membrane into coated vesicles while selectively recruiting cargo proteins for transport. Retrieval of proteins from the Golgi back to the ER is mediated by the coat protein complex I (COPI). Many of the genes involved in ER–to–Golgi transport were discovered in yeast secretion screens³ and later characterised biochemically (e.g.^2,4,5), leading to a share of the 2013 Nobel prize in Physiology or Medicine⁶. Following on from these original screens, the genes involved in the secretory pathway are denoted as ‘Sec’, followed by a numerical designation.

Figure 1. Schematic of coat protein complex II (COPII)-mediated vesicle biogenesis.

a) Overview of the secretory pathway in vertebrates. Newly synthesised proteins are transported in vesicles from the endoplasmic reticulum (ER) to the Golgi apparatus via the ER–Golgi intermediate compartment before being sorted to their functional location in intracellular organelles, the plasma membrane or the extracellular environment. The inset shows a magnified view of ER exit, where COPII mediates vesicular cargo transport to the ERGIC, and retrograde transport from ERGIC to ER is mediated by COPI. The main driver of forward transport from ERGIC to Golgi remains unclear, but COPI might be involved. B) This panels shows the steps of COPII-mediated vesicle formation according to the canonical model. 1.The process is initiated with secretory protein 12 (Sec12)-catalysed GTP exchange on secretion-associated RAS-related protein 1 (Sar1), which triggers Sar1 membrane binding through insertion of its amino-terminal amphiphatic helix. 2. Active Sar1 subsequently recruits the inner coat proteins Sec23 and Sec24, which enables cargo selection through the Sec24 subunit. 3. The inner coat then polymerises, while recruiting the outer coat heterotetramer Sec13–Sec31. 4. Assembly of both coat layers leads to membrane deformation, which also stimulates GTP hydrolysis, initiating coat detachment. 5. The membrane is progressively remodelled into a coated bud. 6. Finally, scission creates vesicles that are free to travel to the target compartment.

In brief (Figure 1B), COPII assembly begins with nucleotide exchange on the small GTPase secretion-associated RAS-related protein 1 (Sar1), a process catalysed by the ER-resident factor Sec12 (refs. ^7,8). Sar1–GTP now partially embeds in the outer leaflet of the ER membrane and recruits the Sec23–Sec24 complex to form the ‘inner layer’ of the coat^2,9,10. Ternary Sar1–Sec23–Sec24 complexes in turn recruit the outer coat, formed by heterotetramers of Sec13–Sec31 (ref. ¹¹). Concomitantly, cargo is selectively enriched in the growing bud through direct or indirect interaction with multiple cargo-binding sites on the Sec24 subunit^12–15. Coat assembly causes membrane deformation, eventually leading to the budding of 60–100 nm vesicles^2,16,17. Sec23 also acts as the GTPase-activating protein (GAP) for Sar1 (refs. ^9,10). Sec23-mediated GAP activity is accelerated by Sec31 binding¹⁸, effectively initiating the process of Sar1 detachment from the membrane and uncoating. Sec16 is an ERES-localised protein that has been proposed to act as a scaffold for coat assembly¹⁹ or as a regulator of the Sar1–GTP hydrolysis cycle^20–22. The timing and regulation of uncoating, transport and delivery to the target compartment remain poorly characterised and are thought to diverge across evolution.

Although the COPII core components are highly conserved, there are several indications of radical adaptation of the early secretory pathway throughout evolution. First, animal and plant genomes encode multiple paralogues of most COPII components, which might have diverged evolutionarily to carry out non-overlapping functions^23–25. Second, cellular morphology differs: animal cell size is significantly increased compared to fungal cells, and in vertebrates the Golgi ribbon concentrates in the perinuclear region as opposed to organising in multiple ER-juxtaposed ministacks in lower eukaryotes²⁶. ER-derived carriers must therefore traverse a range of distances in vertebrate cells. Higher eukaryotes have evolved an ER–Golgi intermediate compartment (ERGIC), which is thought to serve as a mobile cis-Golgi cisterna facilitating transport across longer distances²⁷, but the mechanisms of ERGIC formation and the spatial and morphological relationship between ERES, COPII vesicles and ERGIC remain debated. Third, many cell types in a multicellular organism secrete copious amounts of ECM components and lipoproteins, which are larger than the reported size of COPII-coated vesicles. Procollagens for instance form 300–500 nm long trimers in the ER, and are secreted by fibroblasts at a rate of 100,000 molecules per hour in a COPII-dependent manner²⁸. It remains a mystery how transport of large proteins is achieved by classical vesicular transport, and many ‘alternative’ models have been proposed^28–34 (Figure 4). Finally, there are a number of components in higher eukaryotes that are absent in fungi, which have been shown to regulate ER exit^35–37. Some of these factors, such as Transport and Golgi organization protein 1 (TANGO1) and its homologue cTAGE5 (also known as MIA2), were originally characterised in the context of large cargo secretion but are now thought to have a general role in the regulation of vesicle budding at ERES^38,39. Other such factors, for example TRK-fused Gene (TFG), are thought to have an important role in the organisation of the ERES–ERGIC interface^40–42.

Figure 4. Different models for COPII organisation at mammalian ERES.

According to the canonical model (left), COPII-coated carriers encase vesicles that bud away from ERES, undergo uncoating and then fuse with the ER–Golgi intermediate compartment (ERGIC). Incorporation of large cargo, such as procollagens, is achieved by assembly of large COPII-coated carriers which are formed with the aid of the Transport and Golgi organisation protein (TANGO) family. Alternatively, cargo of various sizes could be transported via direct connection between the ER and the ERGIC as proposed by the “tunnel” model. Finally, ERES may form extended tubular carriers which progress to form the ERGIC. Both alternative models suggest that COPII forms a gate-keeping collar rather than forming a coat. The protein TRK-fused Gene (TFG) supports the structure of the ERES–ERGIC interface in an undefined manner.

Recent technological advances in fluorescence and electron imaging have allowed us to gather new data on the mechanisms, organisation and dynamic regulation of the early steps of the secretory pathway in mammals and have led us to reconsider and refine models that had been widely accepted for decades^32,33. Here we review the various steps of COPII-mediated trafficking, highlighting new hypotheses and open questions. We focus on the regulation of COPII coat assembly and disassembly, and its role in membrane remodelling and in the selective incorporation of cargoes. Finally, we discuss adaptations of the early secretory pathway in organisms that secrete bulky cargoes.

Coat assembly and membrane remodelling

According to the canonical vesicular trafficking model (Figure 1), COPII components assemble to form two concentric layers around the concave side of the budding membrane. The inner layer consists of Sar1, which directly inserts its amino-terminal amphipathic helix into the outer ER leaflet in a GTP-dependent manner, and the membrane-apposed Sec23–Sec24 complex¹, whereas the outer layer consists of Sec13–Sec31 heterotetramers (Figure 1b). Both inner and outer coat proteins polymerise to form curved assemblies, contributing to membrane remodelling, and form flexible and morphologically variable architectures; however, how inner and outer coat assembly is regulated to achieve a certain architecture is not known. We also do not fully understand the role of architectural variability in membrane remodelling and how carrier size and shape are established, nor its physiological relevance in transporting a wide range of cargoes in multicellular organisms. In the following sections we summarise recent advances in our understanding of coat assembly and architecture and highlight where key questions remain.

Role of Sar1 and its paralogues

Sar1 is the dedicated small GTPase for COPII-mediated trafficking. When GTP-bound, it recruits inner and outer coat components to the ER membrane (Figure 1b). Sar1 is also capable of inducing membrane tubulation by itself⁴³ through insertion of its amphipathic helix in the ER outer leaflet⁴⁴. Its role in initiating COPII coat assembly and membrane budding is clearly established; however, a recent report suggested that secretion still occurs in the absence of Sar1 through a condensate-mediated transport carrier that contains the inner but not the outer coat⁴⁵. This contradicts previous studies, which showed that alternative ER exit routes in Sar1-depleted cells are COPII-independent⁴⁶. Either way, although cargo seems to reach the Golgi in cells depleted of Sar1, this is less efficient and cells do not survive or proliferate, confirming that Sar1 is essential⁴⁵.

Mammals have two paralogues of Sar1, Sar1A and Sar1B, which are 90% identical in sequence. Both are differentially expressed across tissues, with Sar1A being the dominant form⁴⁷. Sar1B is specifically increased in certain differentiation stages and cell types including in plasma cells, hepatocytes and early spermatids⁴⁸. The functional and mechanistic differences between the two paralogues are unclear. Sar1A and B exhibit different biochemical properties in vitro in terms of nucleotide exchange rates and Sec23-binding affinity²³, and it has been proposed that Sar1B is specifically involved in the secretion of large cargoes⁴⁹. Sar1B mutations affect secretion of chylomicrons [G] from the small intestine, and are associated with lipid absorption disorders⁴⁹. In addition, a collagen-secretion defective mutant of Sec23 (Phe382Leu) displays a loss-of-function phenotype in vitro in the context of Sar1B but not Sar1A⁵⁰.

Surprisingly, it was recently shown that mice carrying only two copies of the Sar1A gene (one endogenous and one inserted into the Sar1B locus) appear to develop normally, suggesting total redundancy of function between the two paralogues in vivo⁴⁷. This is consistent with studies showing that depletion of both Sar1 paralogues is required to abolish chylomicron secretion in cultured colorectal adenocarcinoma (Caco) cells⁵¹. Phenotypic differences between Sar1A and Sar1B defects could therefore be explained by different levels of transcript at the tissue level. However, B cells and spermatids switch expression from one paralogue to the other at different stages of differentiation⁴⁸, hinting at non-overlapping functional roles.

Determinants of coat assembly and architecture

The molecular structure of all COPII components has been determined (Figure 2). In brief, the outer coat proteins Sec13–Sec31 form elongated, two-fold symmetric rod-shaped heterotetramers with the Sec31 amino-terminal beta propeller domains at their tips, and alpha-solenoid domains forming the dimerization interface¹¹. Tucked between the alpha solenoid and beta-propeller domains, Sec13 forms a second beta-propeller (borrowing a blade from the Sec31 beta-propeller domains) that is thought to confer rigidity to the rods^11,52 (Figure 2a). When purified and incubated in high salt in the absence of membranes, Sec13–Sec31 rods interact through their amino-terminal beta-propeller domains to form cuboctahedral or icosidodecahedral cages^16,17. The inner coat components Sec23–Sec24 form pseudo-symmetric bowtie-shaped complexes, with basic residues on the convex face interacting with negatively-charged phospholipids on the membrane¹⁰. Sar1 interacts with Sec23 through its nucleotide-binding face, and this arrangement favours the insertion of Sec23’s arginine finger [G] to promote GTP hydrolysis¹⁰ (Figure 2a, inset). The Sec23–Sar1 interface also serves as binding platform for a region of Sec31 dubbed the ‘active peptide’, which accelerates GTP hydrolysis⁵³ (see below, Figure 2a).

Figure 2. Structure and assembly of the COPII coat.

a) Here, we provide an overview of the structure of the COPII inner and outer coat components as revealed by X-ray crystallography, cryo-electron microscopy and cryo-tomography studies (Protein Data Bank (PDB) IDs: 6zga, 2qtv, 2pm6 and EM Data Bank (EMDB) IDs: EMD-19410, EMD-19418, EMD-19421). Several interfaces have been characterised, including: i. the Sec23–Sar1 interface (see inset showing details of the Sar1-GTP-binding pocket with the Sec23 arginine finger, which has GTPase-activating functions, ii. the Sec31 ‘GAP’ peptide which accelerates GTP hydrolysis a further 10X), iii. the interface between neighbouring Sec23 molecules with the Sec31 triple-proline motif bridging this interaction, iv. additional interactions between the Sec31 disordered region and Sec23, mediated by the Sec31 GAP peptide⁵³ and charged interface⁵⁴, v. interactions between Sec31 beta-propellers to form outer coat cage vertices and vi. between the Sec31 carboxy-terminal domain and Sec31 alpha-solenoid. B) A slice through a tomogram of reconstituted budding of S. cerevisiae COPII from giant unilamellar vesicles (Image taken from the dataset published in ref. ⁵⁴). C) Representation of the assembled coat as revealed by placing inner and outer coat subunits according to subtomogram averaging of the data shown in B. D) A slice through a tomogram of reconstituted COPII from Saccharomyces cerevisiae budding from native membranes. Image reprinted with permission from ref ⁵⁷. C) As in C, but assembled from the native membranes data shown in D.

Insights into the regulation of coat assembly and membrane remodelling come from structural analysis of the yeast coat on reconstituted budded membranes. When wildtype COPII proteins are incubated with Giant Unilamellar Vesicles [G] (GUVs) consisting of a specific lipid mixture⁴, a variety of coated morphologies are formed, with a high prevalence of straight tubules of various diameters (Figure 2b). On these tubules, Sec23–Sec24–Sar1 trimers form imperfect pseudo-helical arrays^44,54,55 and the outer coat arranges into fishnet-like assemblies⁵⁴ (Figure 2c), with the Sec31 carboxy-terminal domain providing an additional interface with the outer coat cages, stabilising the outer coat interaction network (Figure 2a). There are several small interfaces between inner and outer coat visible in the reconstituted coat, all mediated by motifs in the Sec31 disordered region. These include the previously mentioned Sec31 active peptide, clusters of charged residues, and triple-proline motifs that bind to the interface between inner coat subunits, potentially promoting inner coat polymerisation and stability (Figure 2a)^44,53,54,56. Interestingly, loss of either the amino- or carboxy-terminal domain of Sec31 still supports budding of coated tubules in vitro, but does not support yeast viability, unless they are introduced against a bypass-of-sec13 (bst)-mutant background⁵⁴. This background refers to yeast mutants in which ERES are depleted of luminally-oriented cargo proteins, making the membrane more deformable⁵². This indicates that stable assembly of the outer coat contributes to overcoming membrane resistance, rather than being a main driver of curvature.

When membrane budding is reconstituted from cargo-containing microsomes [G] rather than GUVs, the morphology of coated membranes drastically shifts from tubular to mostly spherical shapes⁵⁷ (Figure 2d). The outer coat on spherical vesicles forms rod–rod interfaces similar to the polyhedral cages and fishnet assemblies described above, but lacks any regular geometry or large-range order, and in addition presents highly flexible hinges⁵⁷. The inner coat assembles with similar interfaces as on tubular lattices but only forms small patches rather than extended pseudo-helical arrays (Figure 2e). This is attributed to the presence of cargo and membrane proteins in the microsomes that create steric obtrusions to inner coat polymerisation⁵⁷. GUV budding reconstitutions performed with COPII mutants that have a weakened inner coat lattice interface result in coated structures resembling beads on a string, supporting the idea that spherical curvature arises when inner coat lattice assembly is interrupted or patchy⁵⁴.

Together, these data suggest that polymerisation and stability of the inner coat layer is the main determinant of membrane morphology, which might in turn be affected by GTP hydrolysis-dependent detachment of Sar1 from the membrane. Upon GTP hydrolysis the affinity of Sar1 for the membrane is reduced⁵⁸. Although it is unclear if this directly leads to full uncoating in cells (discussed below), it is assumed to cause at least some level of inner coat destabilisation. Given that Sec23 and Sec31 act as GAPs, this has the implication that coat destabilisation is a direct consequence of coat assembly. Therefore, fine-tuning of assembly–disassembly rates might be important in establishing coat architecture and carrier size and morphology.

Overall, coat assembly is supported by a large network of weak interactions that are expected to be highly dynamic, providing an ideal platform for regulation of membrane morphology, and timing of assembly and disassembly relative to the vesicle lifecycle. Whether and how specific cargoes or other factors affect GTP hydrolysis rates to regulate coat assembly and membrane shape remains an intriguing question that needs to be explored.

Cargo recruitment and sorting

It is estimated that, in humans, 6,000 proteins require COPII-mediated ER export^59,60, a feat achieved by bulk flow [G]^61–63 or Sec23–Sec24–mediated concentrative and selective export^13,64,65. Several cargo-binding sites have been identified within the Sec23–Sec24 dimer, which selectively recognise ER export signals^{13,15,56,65–69} (Figure 3). Cargo interaction with the inner coat can occur in a direct or an indirect manner through cargo receptors and adaptors^{13,15,61,64,65,70–72}. These factors bridge the interaction between a cargo and COPII, but whereas receptors always accompany cargo into coated vesicles, cargo adaptors are only packaged into COPII vesicles when accompanied by the cargo receptor⁷³ (Figure 3a). After delivering their cognate cargo to the Golgi, receptors are incorporated into retrograde vesicles, formed by the action of COPI, for recycling back to the ER. In this section, we discuss the mechanisms of selective export, and how they are regulated to ensure quality control.

Figure 3. Overview of cargo binding to Sec24.

a) Schematic overview of the different modes of selective incorporation of cargo in COPII-coated vesicles. Cargo can bind Sec24 directly (left), or indirectly through a cargo receptor or adaptor. Cargo receptors link cargo to Sec24 and accompany the cargo into the vesicle, whereas cargo adaptors either bridge the interaction between cargo and receptor and thus accompany the cargo into the vesicle, or bridge the interaction between cargo and Sec24 whilst remaining in the ER (right). B) Side view of the COPII inner coat complex, showing the distribution of known cargo-binding sites on mammalian Sec24A and Sec24D in dark green as well as the Sar1 and triple-proline binding site on Sec23A. C) Details of the cargo-binding sites on Sec24 (dark green), as viewed from the angles indicated in the “Views” insert. All four Sec24 paralogues contain the B-site, which recognises the DxE, LxxLE, YxxCE and ΦC motifs in cargo (see main text). The D-site, which is thought to recognise the IFRTL motif, and the C-site that recognises a structural motif in Sec22 are only found on Sec24A/B. In contrast, the DD-site known to recognise RL, KL and RI motifs, and IxM-site, which binds the IxM motif and is also thought to bind the YV motif, are only found on Sec24C/D. Insets show the region containing the B-site colour-coded by sequence similarity where blue is 0% and red is 100% similarity.

Characterisation of selective cargo

As summarised in a recent systematic Review¹⁴ several proteomic approaches have attempted to catalogue the complete repertoire of COPII cargoes^74–76. Even when extending this systematic analysis¹⁴ to the current date (Supplementary Table 1), only < 2% of the secretome was found to directly interact with Sec23–Sec24. Cargo receptors and adaptors have been shown, with varying degrees of certainty, to facilitate concentrative export of at least 100 additional cargoes^14,73 (Supplementary Table 2). Altogether, these data demonstrate concentrative selection via COPII for only < 5% of the secretome, leaving an important knowledge gap. This could be explained by the majority of the secretome being exported by bulk flow, or by our inability to detect most concentrative cargoes due to low copy numbers. For instance, it is estimated that to maintain the concentration of the most abundant plasma membrane protein in mammalian cells, the amino acid transporter SLC3A2, only < 1 copy of SLC3A2 is required per COPII vesicle⁷⁶. Furthermore, when representative candidates of a class of proteins are shown to be selectively transported, it is likely that the same applies to most members of the corresponding class. Therefore, although concentrative export has only been demonstrated for ∼ 5% of the secretome, a much greater proportion is assumed to be transported in this manner.

Mechanisms of selective export

Export motifs are short sequences or structural features that mediate interaction of cargo proteins with Sec23–Sec24 (refs. ^14,77–79). In mammals, Sec23 has two paralogues, Sec23A and Sec23B²⁴, whereas Sec24 has four paralogues, Sec24A-D, that are categorised into two subfamilies depending on sequence similarity, with Sec24A and Sec24B (referred to as Sec24A/B where interchangeable) belonging to one and Sec24C/D belonging to the other subfamily^25,65,80. This genetic expansion leads to 8 possible Sec23–Sec24 combinations in higher eukaryotes and is thought to support the increased complexity of the secretome by increasing the number of cargoes recognised by COPII and by creating a degree of redundancy (Figure 3b).

Five cargo-binding sites have been identified on the mammalian Sec24 paralogues, referred to as IxM, B, C, DD and N-terminal^{13,15,56,65–67,69} (Figure 3c). Two further sites have also been described in yeast, the A-Site^15,81 (which is functionally replaced by the IxM site in higher eukaryotes), and the D-site, whose sequence is conserved but has not been proven to act as a cargo-binding site in mammals^68,76 (discussed below).

The IxM binding site, present on Sec24C/D⁶⁸, recognises the ER export motif, IxM (Figure 3c). Several proteins have functional IxM motifs, including the soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNARE) protein syntaxin-5, membrin, Golgi pH regulator A (GPR89A/B) and Zinc finger protein-like 1 (ZFPL1)^65,76,79. Syntaxin-5’s interaction with Sec24 requires syntaxin-5 to be in its open conformation, which is stabilized when syntaxin-5 is in a complex with its partner proteins, thus promoting concentrative export of the whole SNARE complex⁷⁹. The IxM binding site may recognise additional motifs or contribute to another currently unknown binding site^76,82.

Perhaps the best documented Sec24 cargo-binding site is the B-site (Figure 3c). First described in yeast¹⁵ and functionally conserved in mammals, the B-site binds cargoes containing DxE (aspartic acid-X-glutamic acid, where X is any amino acid), LxxLE (leucine-X-X-leucine-glutamic acid), YxxCE (tyrosine-X-X-cysteine-glutamic acid) or ΦC (where the carboxy-terminal residues are either two hydrophobic residues or a single valine) ER export motifs^65,67,83. Key residues are conserved across the four mammalian paralogues of Sec24, allowing non-discriminative binding by cargoes such as p24 proteins via their ΦC motif⁸⁴. However, there are a few paralogue-specific differences affecting the B-site (Figure 3c), which give rise to cargo preference¹⁴. For example, studies across multiple cell lines have shown that the DxE and YxxCE motifs specifically or preferentially bind Sec24A/B in a binding configuration that is perturbed in Sec24C/D^14,65,85,86. The cargo receptor Surfeit locus protein 4 (SURF4) has also been shown to interact with the B-site⁸⁷ through a currently unknown region. Some studies have shown that SURF4 and its dependent cargoes⁸⁸ specifically bind to Sec24A/B⁸⁹ and to Sec24C, but not to Sec24D^76,83. This is interesting as cargo specificity within Sec24 paralogue subfamilies is rare¹⁴ and may indicate a conserved SURF4-binding site in Sec24A/B/C but not in D.

Cargo interaction with Sec24 is not only mediated by peptide sequences, but also structural signatures. The C-site, formed at the interface of the Sec23–Sec24A/B dimer, recognises a conformational epitope on Sec22 (refs. ^15,79) (Figure 3b and c). To date, Sec22 is the only protein known to bind the C-site, which might seem inefficient, and it is possible that other clients exist. However, being the v-SNARE [G], at least one Sec22 molecule is required for successful fusion of COPII vesicles with target membranes⁹⁰ and thus dedicating a specific binding site to Sec22 might ensure the functional ability of each vesicle to fuse with downstream membranes.

There is evidence for another cargo-binding site on mammalian Sec24C/D consisting of a DD motif located behind the B-site^69,91,92 (Figure 3c). The associated ER export motifs consist of a positively charged and hydrophobic amino acid pair^91–95, where the degree of hydrophobicity of the (+2) residue is correlated with Sec24 paralogue preference⁹¹. Sec24A/B contain an EE motif at this site, which is conserved in terms of charge, yet, to date, there is no evidence for its functionality.

There may be another cargo-binding site within the amino-terminal region of Sec24. Evidence shows that the alternatively spliced (1-613) variant of Sec24A, which therefore lacks the B-site, still recognises the di-positive RR motif of the Hepatitis B virus (HBV) S protein, suggesting the N-terminal hypervariable region may also contain an additional cargo binding site^66,80.

Finally, yeast Sec24 has another cargo-binding site termed the D-site, consisting of Ser491, Phe576 and Arg578, which recognises the Ile–Phe–Arg–Thr–Leu (IFRTL) motif on the cargo receptor Erv14 (ref. ⁶⁸). These residues are conserved in the Sec24A/B paralog family and mammalian Erv14 homolog, respectively. Mammalian Erv14 has been shown to be exported from the ER in a Sec24A/B-dependent manner⁷⁶. Taken together, this suggests that the D-site may also be conserved in mammals, but this has not yet been proven.

In addition to Sec24, Sec23 also binds cargoes including the Ebola virus matrix protein VP40 (ref. ⁹⁶), the cargo adaptors TMEM39A (ref. ⁹⁷), TANGO1 and cTAGE5 (discussed in the next section)^35,36, and cargo receptor ERGIC-53 (ref. ⁶⁷). ERGIC-53 contains a ΦC motif (FF) at its carboxy terminus⁹⁸, and its oligomerisation creates an FF–FF motif which is recognised by Sec23 and is required for selective ER export^98–100.

Despite the discovery of multiple export motifs and cargo-binding sites, the search for the complete repertoire of COPII-selective cargoes is difficult. Proteomic approaches have been applied to search for the presence of known export motifs, however many of these “hits” were subsequently proven to be non-functional. How COPII distinguishes between functional and non-functional export motifs is currently unknown.

Quality control and regulation of ER export

The secretory pathway is a highly regulated process, with quality control mechanisms acting at almost every level to ensure timely delivery of cargo proteins to their functional location^60,101. ER export is the second site of control within the secretory pathway, acting downstream of ER entry^102–104, and is controlled by the integrated action of retention, retrieval and release^88,105,106. The ER is packed full of proteins¹⁰⁷: the fluid equivalent of ½ ER volume is secreted every 40 mins in the mammalian cell lines CHO and MDCK⁶². Therefore, ½ of the proteins within the ER, including resident proteins, would leak out by bulk flow within the same period if it weren’t for the mechanisms of retention and retrieval.

Many ER resident proteins and cargo receptors contain both COPII–ER exit and COPI-retrieval motifs which allow them to cycle between the ER and Golgi¹⁰⁶. Cargo receptors dissociate from their cargo at the ERGIC or Golgi due to altered physiochemical conditions such as pH, calcium and zinc concentrations^108–113. These conformational changes generally restrict access to the COPII-binding site of cargo receptors and expose their COPI retrieval motif for recycling back to the ER^108–113. Some cargo receptors further aid quality control by recognising inappropriately exported proteins in the Golgi for retrieval to the ER¹¹⁴.

Quality control of ER export is also thought to be achieved by spatial regulation within the ER, although the precise mechanisms of this are largely unresolved⁸⁸. Firstly, misfolded proteins tend to be hydrophobic and associate together^73,115. Misfolded proteins, alongside their chaperones, might form a large interaction complex, which limits their diffusion into ERES for export via bulk flow¹¹⁶. Secondly, spatial regulation is thought to be achieved by the protein composition of ERES. Cargo receptors such as p24 create an exclusion barrier around ERES through their interaction with Sec24 to gatekeep ER export^84,88, and cargo crowding at ERES further restricts access and aids retention of ER resident proteins⁸⁸, intrinsically linking concentrative export and retention.

Protein structure and sequence alone are not the only features to function as a quality control check point. Post-translational modifications of the Sec23–Sec24 dimer, cargo and cargo receptors have been shown to modulate cargo affinity to COPII^117–120. Control over cargo recognition can also be tuned temporally and across different cell types by the ratio of expression of different COPII paralogues, particularly within the cargo recognition unit of COPII, Sec23–Sec24 (refs. ^13,25,121). Each paralogue displays a different tissue expression pattern (Human Protein Atlas)¹²² and has been reported to show different cargo binding preferences and specificities¹⁴ (Supplementary Table 1).

Whereas the paralogue specificity and preference of cargoes for Sec24 has been of particular interest within the field¹⁴, the difference between Sec23 paralogues has largely been overlooked. The ability of Sec23A/B to functionally complement yeast Sec23p, rescue the lethality of their deficient counterparts, the indistinguishability of their interactomes and their 84% sequence identity, suggest that Sec23A/B are functionally interchangeable²⁴. However, cargo specificity to a Sec23 paralogue was recently shown for the first time for Hepatitis B envelope particles⁶⁶. Notably, studies investigating Sec24 cargo specificity were largely based on a heterodimer with Sec23A¹²³. Whether the Sec24 cargo specificities identified in these studies are unique to the Sec23A–Sec24 dimer or shared with Sec23B–Sec24 dimers is therefore unknown.

In addition to the above described mechanisms, ERES membrane composition is also thought to have a role in regulating ER export through lipid partitioning and/or phase separation, creating a favourable environment for the thermodynamic sorting of transmembrane proteins based on their transmembrane domain length^124–126. Multiple lines of evidence support the idea that ERES contain a distinct lipid and protein membrane composition from the ER and loosely mimic downstream membranes which tend to be thicker ^33,127–131. Importantly, secretory proteins tend to have longer transmembrane domains and are thus mismatched in the thinner ER membrane, a feature which might contribute to their partitioning at ERES¹³². What triggers the formation of such domains at ERES is currently unknown. However, it has been shown that transmembrane proteins can trigger phase separation in heterogenous membranes¹³³. Therefore, recruitment of the portion of secretory membrane proteins containing export signals to ERES could trigger phase separation at ERES.

Lastly, it has been proposed that Sec24 acts as a DxE-containing cargo sensor for a regulatory mechanism known as Auto-Regulation of ER eXport (AREX)⁸⁵, thought to modify ER export and protein synthesis in response to ER load, but the mechanisms of this need to be further explored. There may be other signalling pathways, yet unidentified, which act to sense the secretory load of other cargoes.

Vesicle scission, uncoating and delivery to target compartment

Based on the classical vesicular traffic model (Figure 1), vesicles are pinched off from the ER and need to uncoat and move towards the target membrane. These events happen within a localised area of a few hundred nanometres, between the ER and ERGIC in mammals or between the ER and cis-Golgi in yeast. It is unclear whether COPII proteins mediate directional movement or tethering, and whether and how their uncoating is regulated. The inner and outer coat layers might have a differential role in the post-budding processes and be shed at different times. The uncoating process has been elusive as in vitro studies often utilise non-hydrolysable analogues of GTP, and studies in cells lack the resolution to follow the fate of individual coat units or even vesicles, even with super-resolution techniques.

Role of GTP hydrolysis in vesicle scission

Although it is clear that the GTP hydrolysis cycle is at the basis of coat assembly and disassembly, its role in vesicle scission is controversial. Use of non-hydrolysable GTP analogues or GTP-locked Sar1 mutants blocked cargo transport in vitro^134,135, and this was attributed to inefficient release of vesicles¹³⁴. On the contrary, other studies found that scission is equally or even more efficient with non-hydrolysable GTP analogues than GTP^2,8. The latter finding is consistent with a later study proposing that GTP-dependent persistence of the Sar1 amphipathic helix at the neck of the budding vesicle causes a high-energy state promoting membrane scission¹³⁶. However, in COPII budding reactions performed with GUVs in the absence of any mechanical perturbation, inhibition of GTP hydrolysis by using non-hydrolysable analogues is not sufficient to cause scission, and spherical coated profiles remain attached as ‘beads on a string’⁵⁴. When the same budding reactions are performed using microsomes, coated vesicles detach with high efficiency, suggesting that, in addition to the requirement for GTP-Sar1 in the membrane, additional factors present only in microsomal preparations but not in GUVs are needed promote scission⁵⁷. What these factors are is currently unknown.

Role of GTP hydrolysis in uncoating

At least partial uncoating is necessary for the vesicle membrane to fuse with target compartments². Unlinking of Sar1 from the membrane is thought to be necessary for uncoating, but it remains unclear whether it is sufficient, as partial coat detachment does not necessarily imply full uncoating. The hypothesis that uncoating is a direct consequence of Sar1–GTP hydrolysis has been challenged by tryptophan fluorescence measurements [G] and molecular dynamics simulations [G] suggesting that GDP–Sar1 is able to bind membranes, albeit with reduced affinity^58,137. It was shown that, in yeast, cargo can act to retain the coat on budded membranes throughout multiple cycles of GTP binding and hydrolysis by Sar1 (ref. ¹³⁸). Furthermore, coat retention on budded vesicles after Sar1–GTP hydrolysis was shown to be necessary for targeted and unidirectional delivery to the Golgi apparatus¹³⁹.

Regulation of uncoating and delivery to target compartment

In worms and mammals, TFG has been shown to localise at ERES and is proposed to have an important role in regulating vesicle uncoating and delivery. TFG knockdown significantly slows secretion by blocking cargo at ERES^40,140,141, and the distribution of COPII proteins at ERES is also affected¹⁴¹. TFG has been shown to bind to several COPII components — different studies have reported direct binding to Sec23 (ref. ⁵⁸), indirect binding to Sec31 via the small Ca²⁺-binding protein Alg-2 (ref. ¹⁴²), and direct binding to Sec16 (ref. ⁴⁰). Its sequence contains an intrinsically disordered proline-rich region with triple-proline motifs. This region has been shown to compete with Sec31 for Sec23 binding, suggesting TFG might promote detachment of the outer coat⁴². TFG forms high-order assemblies, and TFG condensates are proposed to encompass the ERES–ERGIC interface^42,140 (Figure 4). These condensates might act to form a protective ‘shell’, spatially isolating COPII vesicles and either aiding their diffusion to the target compartment as shown in a recent non-peer reviewed study¹⁴⁰ or promoting uncoating^42,143,144, or both. Evidence is accumulating that condensates might be important in short-range vesicle transport¹⁴⁵. In addition to TFG, Sec16 has also been shown to form condensates in which other ERES proteins also partition, and this process is phosphorylation-dependent¹⁴⁶. The liquid-like nature of ERES is proposed to be essential for secretion, however, the mechanisms of this are unknown.

Retention of the inner coat layer on vesicles has been shown to be necessary for tethering to the cis-Golgi in yeast, via binding of the Rab GTPase Ypt1 (the yeast homologue of Rab1) to phosphorylated Sec23 (ref. ¹³⁹). Ypt1 mediates interaction of coated vesicles with the cis-Golgi-localised tethering factor Uso1p and is necessary for delivery of cargo to the target compartment¹³⁹. At the Golgi, Sec23 is dephosphorylated by a dedicated Golgi-resident phosphatase, ensuring directionality of transport¹³⁹. Whether Rab1 performs a similar role in mammals is unclear. Rab1 is not detected at ERES, but it colocalises with ERGIC-53 on mobile ERGIC elements that take over cargo downstream of COPII and travel to the Golgi to deliver said cargo^32,147.

Alternative models of COPII-mediated secretion

COPII-dependent transport of large cargos constitutes a long-standing mystery. Canonical COPII vesicles have diameters of 60-100 nm (ref ²), too small to accommodate large molecules such as lipoproteins and extracellular matrix components. The in vitro reconstitution experiments discussed above clearly indicate that COPII can, in certain conditions, form large or tubular assemblies that would be able to accommodate bulky cargo, however, such large, coated carriers have not been unambiguously identified in cells. Moreover, bulky cargoes such as procollagens and lipoproteins are secreted in high abundance from specialised cells, and whether large individual coated carriers would be an efficient means to clear the cargo load is debated. This, together with recent data in animal cells, has led to alternative models for COPII mediated ER export (Figure 4).

Mechanisms for large cargo secretion

The presence of large structures containing both procollagen and COPII has been reported^148,149. In these studies, Sec31 ubiquitylation by the E3 ubiquitin ligase Kelch-like protein 12 (KLHL12) was identified as a factor that aids procollagen I secretion and it has been proposed that ubiquitin binding might favour the assembly of COPII outer coats into larger scaffolds^149,150. However, there is no direct evidence for ubiquitin-dependent rearrangement of outer coat architecture, and whether the large COPII- and procollagen-positive puncta are bona fide carriers remains to be established. It has alternatively been proposed that procollagen might be less rigid than previously assumed, and its transport might not require exceptionally large carriers^34,151.

Although COPII is essential, certain mutations have been identified which are tolerated in animal cells^50,152. These mutations do not interfere with the secretion of most cargoes, but large cargo ER export is impaired, suggesting that the mutated residues confer some non-essential property to the coat that specifically aids transport of bulky molecules. Phe382Leu and Met702Val mutations in Sec23 are independently associated with cranio lenticulo structural displasia¹⁵³, a genetic disease arising from defective collagen secretion. Both residues are located at the interface of Sec23 with the Sec31 GAP active peptide (see above, Figure 2) and have been shown to interfere with the stimulation of Sar1B-GTP hydrolysis rates^50,152. These findings have led to the hypothesis that regulation of GTP hydrolysis determines COPII carrier size by tuning its assembly–disassembly dynamics and therefore the coat architecture and its ability to generate large carriers^44,56.

A strong indication that COPII transports large cargoes by a non-canonical mechanism came from the discovery of TANGO1. The TANGO1 family of membrane proteins has evolved specifically in animals and all members are located in the ER¹⁵⁴ and have been implicated in the secretion of large cargoes such as procollagen and lipoprotein particles^31,35,36. The family includes TANGO1S, TANGO1L, cTAGE5 and TALI, which all contain a cytoplasmic region with coiled-coil domains that mediate binary complex formation³⁶, and a carboxy-terminal proline-rich disordered region that binds to Sec23 (ref. ⁵⁶). TANGO1 contains lumenal SH3-like MOTH domains¹⁵⁵ that recruit procollagens into COPII vesicles, acting as their cargo adaptor (while itself remaining in the ER)³⁵ (Figure 4).

TANGO1 acts in complex with other members of the family, and it has been shown to perform a number of functions at ERES beyond acting as a collagen adaptor¹⁵⁴. One of TANGO1 coiled-coil domains mediates supra-molecular assembly into ‘rings’, potentially using its transmembrane helices to act as a diffusion barrier at the base of COPII-generated tubular carriers¹⁵⁶. TANGO1 recruits Sec16 (ref. ¹⁵⁷) and cTAGE5 recruits Sec12 to ERES downstream of Sec16, and this ERES-organising function is necessary to promote procollagen VII secretion¹⁵⁸. TANGO1 also acts as a tethering factor linking ERGIC-53-containing membranes to ERES, potentially providing the membrane material needed for large carriers, or facilitating direct communication between ER and ERGIC^159,160.

The carboxy-terminal proline-rich regions of TANGO1 family proteins that contain several triple-proline motifs have been suggested to compete with Sec31 for Sec23 binding^44,56. While aiding inner coat-polymerisation by binding at the Sec23–Sec23 lattice interface in a similar fashion as Sec31 (Figure 2a), TANGO1 binding would prevent Sec23 from activating Sar1’s GTP hydrolysis activity, stabilising the formation of an extended inner coat lattice and coating of large tubular carriers.

Whereas there is evidence that the TANGO1 family acts specifically in secretion of procollagens and lipoproteins, a range of large and smaller cargoes have also been identified that depend on TANGO1 for their ER export^39,161–163. Some groups have also reported a generalised effect on secretion and ER–Golgi morphology in cells without TANGO1 (ref. ³⁸) (in fact TANGO1 was first identified in a secretion screen using overexpressed HRP as the readout reporter¹⁶⁴), proposing that TANGO1 has a general role in ER export and that subtle defects merely become apparent with collagens as they are large and more immediately affected³⁸. However, general secretion defects may themselves be a consequence of bulky cargo being specifically blocked in the ER and inhibiting other cargoes’ access to ERES¹⁶¹. A recent study reported segregation of ERES that are competent for procollagen export and ERES that only transport small cargoes, suggesting local differences in their molecular organisation. However, both types of ERES recruit TANGO1 (ref. ¹⁶⁵). More work is needed to understand the role of the TANGO1 family of proteins.

Alternative models of COPII trafficking in animals

Whereas 60-100 nm COPII-coated vesicles can clearly be identified in cryo-electron tomography (cryo-ET) data of unicellular organisms (ref ¹⁶⁶ and data deposited in the cryo-ET data portal of the Chan Zuckerberg Institute), their role in metazoan is debated. Despite suggestions that the COPII coat can assemble larger vesicles, some researchers argue that the ‘classical’ vesicular transport model does not present an optimal solution for cells that need to secrete conspicuous amounts of bulky material.

In vertebrates, COPII marks ERES and does not move away from the ER. Instead, cargo transport to the Golgi is taken over by mobile, Rab1- and COPI-positive, ERGIC elements^{32,33,147,167–169}. Even in budding yeast, where Golgi elements are dispersed, COPII seems to remain stable at ERES whereas cis-Golgi elements travel to ‘collect’ cargo¹⁷⁰.

Recent technological advances have allowed the investigation of COPII and its relationship with cargo and other compartments with high temporal and spatial resolution. Experiments using temperature blocks and retention using selective hooks [G] (RUSH) showed that upon synchronised release from the ER, Rab1-positive, cargo-containing tubular structures depart from ERES³², whereas the bulk of COPII fluorescence remains static. Further studies reported similar data, but spherical rather than tubular carriers were seen¹⁴⁷. However, in the latter study, a small amount of Sec31 is resolved departing ERES together with cargo and lost within a few seconds, possibly indicating fast uncoating¹⁴⁷.

Using cryo structured illumination microscopy [G] (cryo-SIM) and correlative focused ion beam–scanning electron microscopy [G] (FIB–SEM) it was shown that ERES consist of vesicular tubular clusters that are attached to the ER by a small neck³³. RUSH-synchronised cargo accumulated at ERES and departed in elongated pearled tubules that run along microtubules. In line with previous studies, live fluorescence imaging clearly showed that COPII does not accompany cargo beyond ERES, whereas mobile cargo-containing structures include ERGIC53 and COPI. At steady-state before cargo release, COPII and COPI label adjacent structures ∼140 nm away from each other³³.

These recent studies propose a new model for COPII function in animals, where COPII forms a collar at the base of ERES, and acts as a gatekeeper to concentrate cargo selectively into COPII-free transport carriers (Figure 4, right panel). Based on this model, COPII could arrange into tubular assemblies around carrier necks and TANGO1 could be a key player in the formation of such structures by organising COPII at these collars. This is an attractive model, which could explain efficient transport of abundant and bulky cargo, potentially also through direct connections from the ER to the ERGIC (Figure 4, middle panel), but more evidence is needed to support it. Due to the limited resolution of fluorescence microscopy, and the lack of molecular detail in plastic-embedded FIB–SEM and transmission electron microscopy (TEM) images, it is not currently possible to distinguish these models from the ‘classic’ model in which ERES and ERGIC are in close proximity, exchanging material through coated vesicles, with Rab1 and COPI-positive ERGIC membranes subsequently departing to deliver cargo to the Golgi (Figure 4, left panel). The role of COPI is also unclear: is it responsible for anterograde trafficking towards the Golgi, or just a ‘passenger’ on ERGIC membranes there to mediate retrograde recycling of cargo receptors and escaped ER-resident proteins?

Conclusions and future perspectives

Our molecular level understanding of the canonical COPII model is incredibly detailed due to work conducted mostly in yeasts and in vitro reconstitution systems. However, even in this model there are knowledge gaps that remain to be filled. For instance, the dynamics and regulation of uncoating and scission are poorly understood, especially the role of GTP hydrolysis. Does GTP hydrolysis change assembly–disassembly rates and affect coat and vesicle morphology? Is this process regulated by extrinsic factors? At what point in the vesicle lifecycle are the different coat layers shed?

Our ability to uncover the complete COPII cargo repertoire is also limited⁷⁶. Current proteomic approaches are insufficiently sensitive and individual investigation of each potential cargo is too low throughput to be economically viable. An improved understanding of cargo–COPII interactions is key to rectifying this knowledge gap and revealing how COPII distinguishes between functional and non-functional export motifs.

In addition, our understanding of the paralogue-specific functions of COPII components is currently lacking. Significant work is required to disentangle their functions, particularly in the case of the Sar1A/B and Sec23A/B paralogues and the various paralogue combinations capable of forming COPII complexes. Whether the Sar1 and Sec23 paralogues have truly different functions or if their divergence is instead focused on regulation, aiding response to a wider range of intracellular and environmental cues, is not yet understood.

Technological advances have recently led researchers to reconsider the canonical model of COPII organisation at ERES (Box 1) ^32,33. Indeed, a model that only accounts for small (< 100nm) coated vesicles does not explain the physiological variety of transported cargoes in higher eukaryotes and animals, which include bulky and highly abundant protein complexes such as procollagens and lipoproteins. Several alternative models have been proposed in the literature, but direct evidence for them is lacking. Metazoan factors located at ERES such as the TANGO1 family and TFG have been implicated in large cargo export; however, their direct involvement is debated, mainly due to the overall disruption of membrane homeostasis that their prolonged depletion with small interfering RNA (siRNA) causes. The use of CRISPR to engineer acutely degradable variants of TANGO1 and TFG¹⁴¹, and optogenetics to induce their recruitment¹⁶⁵ is likely to clarify their roles.

Box 1. Advanced imaging approaches illuminate the early secretory pathway.

These are very exciting times in cell imaging, with recent technological advances promising to soon tackle many of the open questions set out in this Review. Advances in super-resolution light microscopy are beginning to narrow the resolution gap between light and electron microscopy, and modern fluorescence microscopy approaches allow researchers to visualise multiple colours with high spatiotemporal resolution^175,176. Together with the increasing accessibility and versatility of endogenous labelling of proteins, this will enable us to follow cargo through the early secretory pathway while monitoring the location of ER, COPII and ERGIC markers.

Electron microscopy of cells is also undergoing profound transformation¹⁷¹. Volume EM, where large volumes (> 1 um) of cells and tissues are imaged at nanometer resolutions, is increasingly providing us with beautiful atlases of individual cells, outlining morphology and distribution of all organelles¹⁷⁷. Coupled with high-resolution fluorescence, this technique allows us to identify the targets of interest when their morphology is unknown or unexpected. For instance, using a correlative super-resolution fluorescence and serial focused ion beam–scanning electron microscopy (FIB–SEM) approach¹⁷⁸, it was recently proposed that ER to Golgi transport might occur via vesicular clusters that remain attached to the ER by a COPII neck, from which COPI-positive tubular extensions depart³³. Recent work has demonstrated that FIB–SEM volume imaging can provide sufficient contrast in cryo-preserved vitrified cellular samples to visualise individual organelles¹⁷⁹, opening new possibilities to couple it with cryo-transmission electron microscopy in future workflows to visualise the coat arrangement at ERES.

Cryo-FIB–SEM can be used to prepare thin biological samples such as vitrified cells and tissues down to 100–300 nm thick lamellae, allowing them to be imaged by transmission electron microscopy, either in 2D or by cryo-electron tomography (cryo-ET). Pseudo-atomic resolution information can be recovered through image processing, for example using 2D Template Matching¹⁸⁰ or subtomogram averaging^166,181, providing a detailed understanding of the structure and localisation of complexes in their native environment. The in situ structure of the COPI coat on Golgi-derived vesicles in the green alga Chlamydomonas reinhardtii has been resolved using cryo-FIB–SEM and cryo-ET¹⁶⁶, and the same approach could be applied in the future to vesicles and organelles involved in the early secretory pathway in a range of organisms. Correlation with cryo-fluorescence signal can be a powerful way to target areas of interest for data collection when these are not ubiquitous, such as ERES that cluster in discrete fluorescent puncta¹. Adapting super-resolution fluorescent techniques to cryo-conditions still remains a significant challenge¹⁸², but a number of solutions have been developed^{178,183–187}, providing significantly improved targeting resolution to correlative light and electron microscopy (CLEM) techniques. Quantitative and automated image analysis also contributes to our ability to interpret cryo-ET data. From denoising^188,189, to automated segmentation^190,191 and feature recognition^192–195, we can envision that we will be able to reconstruct the molecular landscape of individual cells.

The various models proposed do not need to be mutually exclusive. The formation of small or large carriers, fully or partially coated, and detached or connected through a coated collar can be theoretically explained by different tuning of the relative timing of coat assembly, uncoating, membrane scission, and fusion with the target compartment. Given the highly dynamic nature of all interactions, it would be unsurprising if cells evolved mechanisms to finely regulate these timings to achieve the transport of diverse cargo and to dynamically respond to changes in cargo load such as those observed during RUSH-induced secretory challenges.

Recent advances in imaging such as cryo-correlative FIB–SEM and tomography provide the possibility, for the first time, to directly visualise COPII-coated structures in animal cells, at a resolution that allows us to uncover membrane morphology, coat architecture and local cellular environment (Box 1)^166,171. Utilisation of endogenous tagging and synchronisation of secretion combined with high spatio-temporal resolution fluorescence microscopy could be used alongside cryo-correlative FIB–SEM and tomography to visualise the structures associated with bulky cargo export and to improve our understanding of the function of the TANGO1 family and TFG proteins. We anticipate that these advancements can consolidate the various models of COPII-coated structures in animal cells. These techniques could be applied to various tissue and cell types to further our understanding of the secretory pathway across different cells. Neurons provide a particularly interesting avenue for this investigation as they require incredibly long-range transport, have a notable secretory load and unique intracellular morphology^172,173. Furthermore, defects in COPII-mediated secretion in neurons are associated with several neuronal defects and diseases which are currently poorly understood¹⁷⁴.

Considering these current open questions and technological advances, we expect the field of protein traffic to continue to flourish.

Glossary

Arginine finger

This is a highly conserved and essential residue acting in trans on many GTPase and AAA+ ATPase enzymes to promote GTP hydrolysis. Often they are found in GTPase-activating proteins (GAPs) (such as Sec23).

Bulk flow

Bulk flow refers to the passive export of proteins at their prevailing concentrations within the ER.

Chylomicrons

Chylomicrons are large lipoproteins produced in enterocytes. They are composed of a central lipid core primarily consisting of triglycerides, together with esterified cholesterol and phospholipids and transport dietary fat from the intestine to the liver and peripheral tissues.

Cryo-structured illumination microscopy

Interference-based high-resolution fluorescence technique performed at cryogenic temperatures (below -160 C).

Focused ion beam–scanning electron microscopy

a dual beam technique where scanning electron microscopy to visualise samples is combined with a focussed ion beam to mill material. It can be used to prepare thin lamellae of biological samples for use in TEM, or to produce serial SEM images of the ‘face’ of the block being milled.

Giant unilamellar vesicles

spherical particles made of a lipid bilayer with sizes in the micrometer range. They can be used to mimic cellular membranes for in vitro experiments.

Microsomes

particles derived from permeabilised cells after differential centrifugation, consisting mostly of endoplasmic reticulum membranes.

Molecular dynamics simulations

a computational technique used to simulate the physical movements of atoms and molecules over time based on their energy landscape.

Retention using selective hooks

The Retention Using Selective Hooks (RUSH) system is used to synchronise protein transport within the secretory pathway. It consists of a “hook” (a streptavidin-tagged protein localised to a secretory compartment like the ER) and a “reporter” (a fluorescently tagged protein fused to Streptavidin Binding Peptide, SBP). The streptavidin-SBP interaction retains the reporter at the hook’s location, and is disrupted by biotin, causing the reporter to be released synchronously.

Tryptophan fluorescence measurements

a technique that monitors tryptophan fluorescence to measure conformational changes in proteins.

v-SNARE

SNAREs mediate the fusion of vesicles with the target membrane via the interaction of vesicle localised SNAREs (v-SNAREs) with target membrane localised SNAREs (t-SNAREs).