A systematic review of Sec24 cargo interactome

Endoplasmic reticulum (ER)‐to‐Golgi trafficking is an essential and highly conserved cellular process. The coat protein complex‐II (COPII) arm of the trafficking machinery incorporates a wide array of cargo proteins into vesicles through direct or indirect interactions with Sec24, the principal subunit of the COPII coat. Approximately one‐third of all mammalian proteins rely on the COPII‐mediated secretory pathway for membrane insertion or secretion. There are four mammalian Sec24 paralogs and three yeast Sec24 paralogs with emerging evidence of paralog‐specific cargo interaction motifs. Furthermore, individual paralogs also differ in their affinity for a subset of sorting motifs present on cargo proteins. As with many aspects of protein trafficking, we lack a systematic and thorough understanding of the interaction of Sec24 with cargoes. This systematic review focuses on the current knowledge of cargo binding to both yeast and mammalian Sec24 paralogs and their ER export motifs. The analyses show that Sec24 paralog specificity of cargo (and cargo receptors) range from exclusive paralog dependence or preference to partial redundancy. We also discuss how the Sec24 secretion system is hijacked by viral (eg, VSV‐G, Hepatitis B envelope protein) and bacterial (eg, the enteropathogenic Escherichia coli type III secretion system effector NleA/EspI) pathogens.


| INTRODUCTION
Gene duplication is one of the key drivers of functional diversification during evolution. Closely related gene pairs often exhibit functional redundancy but over time also evolve to execute distinct functions.
One such well-studied example of gene family duplication and functional diversity is seen in the coat protein complex-II (COPII) members. COPII subunits form vesicles that package and transport proteins from the endoplasmic reticulum (ER) to the Golgi, the first transport step in the classical secretory pathway. COPII cargoes include a diverse range of proteins comprising of secretory (eg, cytokines), extracellular matrix (ECM), organellar and signalling proteins as well as large complexes such as lipoproteins (500 nm). Accurate conveyance of newly synthesised proteins is a prerequisite for all aspects of cellular functions, including organellar maintenance and response to the extracellular environment.
Structurally, the core COPII complex is an icosidodecahedral coat consisting of an inner, Sar1-Sec23-Sec24 lattice and outer layer of Sec13-Sec31 heterotetramers. 1 The GTPase Sar1, activated by Sec12, is recruited at the ER membrane, which is followed by binding of Sec23-Sec24 heterodimer. Sec23-Sec24 acts as cargo selection platform and forms the pre-budding complex. Sec13-Sec31 joins the complex after cargo capture. Sec23 stimulates Sar1-GTP hydrolysis leading to vesicle fission. The fact that there is a highly diverse repertoire of cargo needed to maintain both steady-state and tissuespecific demand, as well as cater to changing cellular requirements, Sharanya Chatterjee and Ana Jeemin Choi contributed equally to this study. has likely propelled the need for different paralogs of COPII subunits.
While mammals encode two paralogs each for Sec23, Sar1, Sec31 and four for Sec24, and yeast only one for all except three for Sec24, the genome of the Arabidopsis thaliana encodes several isoforms of all COPII components (five Sar1, seven Sec23, three Sec24 and two Sec13 and Sec31). 2 Numerous examples indicate that the function of multiple COPII subunit paralogs may not be completely overlapping as loss of function of one paralog cannot be fully compensated by its counterparts. Although Sar1A and Sar1B are 90% identical at the amino acid level, mutations in Sar1B have been reported to impair the ER export of chylomicron-cargo, causing Anderson disease and chylomicron retention disease. 3,4 Sec23A is involved in secretion of collagen and its mutations have been linked to craniofacial skeletal abnormalities 5 whereas Sec23B mutations result in haematopoietic abnormalities. 6 Distinct functions of Sec23 paralogs have been attributed to differential tissue specific expression. 7 Sec24C and Sec31 are alternatively spliced, further adding to the complexity of cargo selection and packaging. 8,9 Furthermore, during murine embryonic development, deficiency of different Sec24 paralogs show unique abnormalities, such as neural tube defect for 24B and early and late embryonic lethality for 24D and 24C, respectively. 10,11 The functional bifurcation of Sec24 paralogs is also reflected in their role in autophagy. Sec24A/B has been implicated in bulk autophagy whereas Sec24C has been shown to be exclusively involved in selective ER-phagy. 12 Under starvation, when the secretory pathway is inhibited, COPII-dependent processes are involved in autophagosome biogenesis. When nutrients are replete, Sec23B is tagged by the F-box protein FBXW5 for proteasomal degradation to limit autophagy. 13 Upon nutrient depletion, UNC51-like kinase 1 phosphorylates Sec23B on serine 186, which then associates with Sec24A/B, but not with Sec24C/D to promote autophagy. 13 During ER-phagy, Sec24C is exclusively involved in the lysosomal delivery of ER-phagy receptors FAM134B and RTN3. 12 In yeast, there are three Sec24 paralogs: Sec24p, Lst1p and Iss1p.
Mammals have four Sec24 paralogs which have been broadly categorised based on sequence identity into two subgroups: Sec24A/B and Sec24C/D, 60% identity within and 25% identity across subgroups. 14 Sec24p and Sec24A/B are more closely related and Lst1p shows more homology to Sec24C/D. Amplification of Sec24 paralogs in mammals might be because of the greater diversity of secretory proteins in mammals as compared with yeast. Although few cargo/ cargo-receptors can be packaged by all paralogs, mounting evidence from recent studies also support the notion of paralog-specific packaging of cargoes/cargo-receptors. Recently, using SILAC-based mass spectrometry, it was shown that the composition of mammalian COPII vesicles differ depending on the Sec24 paralog used for in vitro vesicle reconstitution. 15 This raises the possibility of a common cargo repertoire which can be packaged and transported by all paralogs and a specific cargo subset either exclusively or preferentially packaged by a specific paralog, in order to meet specific/changing demands. This review explores relevant studies on Sec24 paralogs within the functional context of defining cargo associated to each paralog.
Considering the vast literature on ER-Golgi export studies, to comprehensively identify all known Sec24 cargo proteins and their Sec24-binding motifs, we conducted a systematic search using a PRI-SMA approach followed by a qualitative assessment of the studies.
We have extended our analysis to obtain a thorough overview of cargo recognition specificity by synthesising the existing data of ER export motifs in a paralog-specific manner. In all, this study represents a comprehensive analysis to capture both redundancy and diversity displayed by Sec24 paralogs in context of cargo recognition. paralog-specific cargo interactome, a literature search was performed on PubMed, Scopus and Web of Science to extract papers that study the interaction between Sec24 and its cargoes. Briefly, the title, abstract and keywords were searched with a curated combination of entry terms, described in Methods, to obtain research articles on Sec24-binding proteins and motifs. Papers focused on COPII coat organisation and structure or on ER exit regulation were excluded because they were not of interest to this review. Highly specific results comprised Sec24 papers that contained keywords of 'interaction' or 'binding' of 'cargoes' or 'motifs' and of 'cargo sorting' or 'ER export'. The title and abstract of the 136 records obtained from the database searches and the 33 records obtained from complementary manual searches were screened according to the exclusion criteria detailed in Methods, to obtain a total of 76 articles included for full-text review ( Figure S1).
Classification of the yeast Sec24-binding proteins based on cellular function revealed that the field is heavily skewed towards ER-Golgi transport proteins such as SNAREs and cargo receptors followed by proteins involved in ion exchange (Figure 2A and listed in Table 1).
relied on in vitro COPII vesicle formation assay and pulse-chase techniques to test paralog specificity. This was complemented with X-ray crystallography and affinity titrations by fluorescence polarisation to measure the effects of alanine substitutions (Table S1). One study of plant Sec24 (Vicia faba) used in vivo FRET to confirm Sec 24 binding. 21 Based on the methodological assessment (Table S2) To identify Sec24-binding to the POI, most studies (60%) used a single technique, while a quarter of the studies did not experimentally verify physical Sec24-POI binding ( Figure 4C). In such studies, Sec24-POI binding was either deduced from previous literature, [49][50][51][52] or by confirming Sec24-driven ER export 10,47,48,[53][54][55] as cargoes are known to be sorted for ER export by direct or indirect interactions with Sec24. 56 The most frequently used techniques to verify Sec24-POI binding were immunoprecipitation (IP) and pulldown assays, coupled with immunoblotting of the POI or liquid chromatography-mass spectrometry (LC-MS) or MS/MS for identification of the binding protein ( Figure 4D and listed in Tables S2 and S3).
Few studies used two techniques such as pulldown combined with fluorescence resonance energy transfer (FRET) 57 or with X-ray crystallography, 14 or a combination of molecular homology modelling with FRET 58 or with IP. 59 One study validated binding using pulldown, yeast two-hybrid (Y2H) assays and protein-protein interaction ELISA 60 while another confirmed binding through four assays, con- To identify Sec24-binding motifs, approximately a third of the studies relied on motifs that were previously identified in literature or identified new binding motifs from empirical testing from untargeted mutations or from targeted mutations after identification of conserved sequences ( Figure 4G). Upon identification of a motif, a variety of assays were used to confirm the motif's role in export, most commonly functional assays of cargo packaging into COPII or localization assays ( Figure 4H). Further, few studies relied on X-ray crystallographic and biochemical assays to dissect mechanistic details of cargo discrimination by specific paralogs. For mammals, as shown in Table 2, cargoes specific for   Sec24 paralogs have been reported, for example, E-cadherin reported to be a Sec24A client, Vangl2 a Sec24B client, and various compound transporters and ion channels specific for Sec24 A, C or D. The neurotransmitters, noradrenaline and glycine and γ-aminobutyric acid (GABA) transporters (DAT, NET, GLYT1 and GAT1 respectively) were identified to exclusively bind to Sec24D while serotonin transporter (SERT) is an exclusive client of Sec24C (Table 2).

| Sec24-paralog specificities of cargo proteins
Interestingly, cargoes specific for more than one paralog were also identified ( Non-cargo proteins involved in the biogenesis of COPII vesicles were also found to possess Sec24-paralog specificities, as highlighted in Table 3. The binding motifs, location, function and full methodological assessment for non-cargo proteins have been summarised in various ER export recognition sites, sequence motifs on cargoes and non-cargo proteins binding to Sec24 have been listed alongside the proteins in Tables 1 to 3. In yeast, the most commonly found motifs include the diacidic (DxE) and dihydrophobic motifs. The PM ion-exchange protein Erv14 utilises the unique IFRTL motif to bind to Sec24p. In plants, both diacidic and triacidic motifs are the most studied ones (Table S1).
In mammals, 14 sequence motifs and 2 conformational motifs have been identified so far (Table 4) respectively. 51 In case of SERT, a single substitution of lysine to tyrosine in the Sec24-binding motif switched its binding preference from Sec24C to D. 51 Further, cargo receptor-specific motifs were also identified, such as the IxM motif on syntaxin-5 and the YxxCE motif on Bet1 (Table 3). IxM motif binds specifically to a groove on Sec24C and D, demonstrating functional overlap between Sec24C and D.
Other unique motifs have been reported for specific cargoes, such as YV on claudin-1, the looptail comprising of D and S residues in Vangl2 and two synergistic motifs KW and PYRKR found on dopamine transporter ( Table 2). It is yet to be studied whether these motifs are also found on other cargoes or cargo receptors and whether the specificity of recognition is based on a particular physico-chemical characteristic rather than the specific amino acid sequences. Thus far, conformational motifs have been reported for Sec22 and syntaxin-5 (Table 3). Additionally, the ER export of cholesterol and collagen depend on the proline-rich domain of the related cargo receptors Mia2 and Mia3 (Table 2). Thus, as with the diversity of Sec24-paralog specificities of different cargoes, we further highlight the varying Sec24-paralog specificities of different motifs revealing both specificity and partial redundancy of Sec24 paralogs that is both cargodependent and motif-dependent (Table 1).
Further, from the studies included in this systematic review, binding motifs on cargoes that bind cargo receptors instead of Sec24 were also identified: the 'FF' motif on autotaxin binds p23, 49 the 'KEEL' motif on ERp29 binds the KDEL receptor 69 and the 'LS' motif on SAC1 binds the phosphoregulatory protein 14-3-3. 87 These findings highlight an increased complexity in the Sec24-cargo interactions that involve various mediators such as cargo receptors that themselves bind to a wide repertoire of cargoes through specific motifs.

| Sec24 is a target of viral and bacterial virulence factors
Among the non-cargo proteins, along with cargo receptors and ER-Golgi transport proteins, we have included non-mammalian proteins such as vesicular stomatitis virus glycoprotein (VSV-G), 14,53 Hepatitis B envelope protein (HBV-S) 78 and NleA (aka EspI) 60,82 which is injected into mammalian cells via a type III secretion system injectisome, expressed on the bacterial cell wall of enteropathogenic and enterohemorrhagic E. coli (EPEC and EHEC) 88 (Table 3)  Considering that approximately one-third of all mammalian proteins rely on the COPII-mediated secretory pathway for membrane insertion or secretion, 93 and the largest fraction of the tissue-enriched transcriptome codes for secretory proteins, 94 a lot still remains to be discovered.

| CONCLUSIONS AND FUTURE PERSPECTIVES
Delineating Sec24 paralog specific clients is experimentally challenging because of the low copy number of cargo molecules being transported at steady state levels, with <1 copy per vesicle required to maintain the most abundant transmembrane proteins. 15 Only one study which opted for proteomics of the human Sec24 paralog-specific interactome observed that the identified proteins were not cargo proteins but cargo receptors that reside in the ER-Golgi compartments and proteins involved in tethering and fusion of vesicles such as SNAREs and GOT1B, thus highlighting the technical difficulty to assay Sec24 and cargo interactions. Further, Sec24B is prone to proteolytic degradation, so Sec24B paralog specific COPII proteome could not be elucidated. 15 The methodological assessment of the included studies revealed that a high diversity of epithelial cell types from various tissues and mammalian species were used, although, importantly, not polarised cells except for MDCK cells, which were used in five papers. [69][70][71]73,74 As polarity determines protein destination, it might also affect expression and binding specificities of the cargoes. Cellular factors could also impact the binding of two proteins, making methods that use cell lysates to test binding, such as co-IP and pulldowns, more relevant than Y2H assays to show that interactions occur in cellulo. Furthermore, protein overexpression studies in transfected cells may not guarantee that the same binding occurs with endogenous expression in physiological conditions.
Because RNAi silencing methods do not lead to complete silencing, accurate assessment of paralog specific transport becomes technically challenging. Another limit is that when Sec24-dependent transport is halted, some proteins may rely on bulk flow to be transported albeit at a slower pace, as was shown for SERT. These findings underline a need for more sophisticated proteomics techniques such as cross-linking, proximity labelling and high-throughput data studies to elucidate the COPII cargo interactome.
Among the ER export motifs, diacidic and dihydrophobic motifs are the best characterised ones in both yeast and mammals. However, presence of existing ER export motifs should not be solely used to predict export of proteins. For instance, a study identified 100 cargoes that contained putative diacidic motifs and proposed that the functional ER export diacidic motifs seem to appear in a region with ordered structure and higher hydrophobicity. 95 The importance of the dihydrophobic nature, rather than the specific residues, of the ER exit signal 'FF' in ERGIC-53 was found by replacing the di-phenylalanine 'FF' motif by di-tyrosine 'YY', di-valine 'VV', di-isoleucine 'II' or dileucine 'LL' which mediated transport as efficiently if not more than the 'FF' motif. 80,96 Indeed, the possession of a motif does not guarantee ER export or Sec24 binding, as was seen in a study where a Sec24C/D-binding 'IxM' motif nested in the 'DxE' diacidic motif of VSV-G did not bind Sec24C/D. 14

| Search strategy
To conduct the systematic review, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) search strategy was adopted according to its guidelines. The PRISMA diagram ( Figure S1) depicts the different phases of the systematic review. An initial PubMed search using the term 'Sec24' yielded 364 articles. To increase the stringency of search, a finetuned combination of selected search terms were designed and optimised to extract the papers with the most potential for relevance to this review's research questions about Sec24-binding cargoes and motifs in the context of ER export and the secretory pathway: (sec24*) and (motif or cargo*) and (interact* or bind* or 'cargo-binding' or signal* or 'exit signal*' or 'export signal*') and (secretory pathway/cargo/protein or cargo sort*/adaptor/transport or traffic* or selective* export/sort* or export). This search query was then applied to Scopus and Web of Science. The literature search in PubMed was performed on April 12, 2020, and in Scopus and Web of Science on June 5, 2020.
To find all other relevant papers that would not have been found in the search strategy (for instance if the paper happened to not mention Sec24 in the title or abstract), literature cited in the papers collected from the above search strategy and reviews on COPII cargo export were manually searched, and potentially relevant studies were added to the study collection. The manual search was stopped when a loop was reached whereby no new studies were found. Of the total of 171 papers obtained from the three database searches and the manual search, 95 papers were excluded based on the exclusion criteria detailed in Figure S1, with the aim to only include studies on Sec24-dependent cargo export and Sec24-binding. The remaining 76 included articles underwent full-text review. A limitation in this review's search strategy is that, despite optimising the query terms combination, articles that studied relevant topics but did not have the searched terms in at least one of the titles, abstract or search engine keywords, would have not been identified.

| Data synthesis and analysis
To assess the findings of the included studies, the following informa-

ACKNOWLEDGMENTS
This project is supported by a grant from Wellcome Trust.

CONFLICT OF INTEREST
The authors declare no conflict of interests.