The WDR11 complex is a receptor for acidic-cluster-containing cargo proteins

Summary

Vesicle trafficking is a fundamental process that allows for the sorting and transport of specific proteins (i.e., “cargoes”) to different compartments of eukaryotic cells. Cargo recognition primarily occurs through coats and the associated proteins at the donor membrane. However, it remains unclear whether cargoes can also be selected at other stages of vesicle trafficking to further enhance the fidelity of the process. The WDR11-FAM91A1 complex functions downstream of the clathrin-associated AP-1 complex to facilitate protein transport from endosomes to the TGN. Here, we report the cryo-EM structure of human WDR11-FAM91A1 complex. WDR11 directly and specifically recognizes a subset of acidic clusters, which we term super acidic clusters (SACs). WDR11 complex assembly and its binding to SAC-containing proteins are indispensable for the trafficking of SAC-containing proteins and proper neuronal development in zebrafish. Our studies thus uncover that cargo proteins could be recognized in a sequence-specific manner downstream of a protein coat.

Graphical Abstract

In brief

In vesicle trafficking, cargo recognition primarily occurs at the donor membrane. In this study, Deng et.al demonstrate that WDR11 complex, a tethering factor located on both the TGN and vesicles, directly interacts with the super acidic clusters of cargo proteins and mediates their endosome-to-TGN trafficking, highlighting direct sequence recognition in a later step of vesicle trafficking.

Introduction

Vesicle trafficking plays a vital role in maintaining cellular homeostasis by regulating the localization of various intracellular components in eukaryotic cells.¹ Dysregulation of vesicle trafficking gives rise to a spectrum of disorders, including neurological, respiratory, immune, and metabolic disorders.² The process of vesicle trafficking involves a series of interconnected events, including cargo recognition and vesicle formation mediated by protein coats at the donor membrane, movement along the cytoskeleton by coupling to motor proteins, vesicle capture by tethering proteins at the target membrane, and subsequent fusion mediated by SNARE complexes.^3,4 The signal-dependent recognition of specific cargoes by components of protein coats (e.g., COPI, COPII, and clathrin) at the donor membrane is key to vesicular trafficking, allowing only a subset of cargoes to be selectively transported.⁵ However, it remains unclear whether other steps of vesicle trafficking involve signal-dependent recognition of cargo proteins to further enhance the fidelity of the process.

The process of vesicle trafficking from endosomes to the trans-Golgi network (TGN), also known as retrograde transport, is critical for a wide array of physiological functions, including nutrient uptake, cell signaling, and neuronal development.⁶ Multiple proteins are known to recognize specific signals in the cytoplasmic domain of cargo proteins transiting through endosomes, including the clathrin-associated adaptor protein 1 complex (AP-1), retromer, retriever, and several members of the sorting nexin (SNX) family.^7–13 AP-1 recognizes various sorting signals, including tyrosine-based, dileucine-based and acidic cluster motifs, in the cytosolic tails of transmembrane proteins to sort them into clathrin-coated vesicles (CCVs).^14–17 Several proteins are known to contain acidic cluster motifs, including the cation-independent mannose 6-phosphate receptor (CI-MPR), carboxypeptidase D (CPD), furin, and KIAA0319L. Recently, the WDR11 complex was reported to function downstream of the AP-1 complex, and to facilitate the transport of acidic-cluster-containing proteins to the TGN.^18,19 The WDR11 complex consists of WDR11 and FAM91A1 subunits in eukaryotic organisms, and a third subunit C17orf75 in vertebrates.^18,20 The WDR11 complex is localized on both the TGN and vesicles, and may contribute to vesicle tethering through its interaction with the Golgi-localized protein TBC1D23.^21–23 Although the role of WDR11 in the trafficking of acidic-cluster-containing proteins is widely accepted, it is unclear how the WDR11 complex exactly regulates this trafficking.

Emphasizing the importance of retrograde trafficking is the observation that mutations in multiple proteins involved in this process give rise to neurological disorders. For example, mutations in the σ1A and σ1B subunits of AP-1 are the causes of MEDNIK and Fried/Pettigrew syndrome, respectively.^24,25 Furthermore, TBC1D23 was identified as a cause of pontocerebellar hypoplasia (PCH).^26–29 Intriguingly, recent clinical reports described cases where loss-of-function mutations in WDR11 also give rise to a disease highly similar to PCH.³⁰ In addition, mutations in WDR11 have been linked to other medical conditions, such as congenital hypogonadotropic hypogonadism, Kallmann syndrome, and 10q26 deletion syndrome.^31–34 Both AP-1 and WDR11 are also subverted by a subset of viruses during infection.^35–39 For example, HIV-1 hijacks AP-1 via its accessory protein Nef, promoting the degradation of major histocompatibility complex I (MHC-I) and thus facilitating immune evasion.³⁵ Furthermore, WDR11 was reported to be exploited by herpes simplex virus (HSV) and human cytomegalovirus (HCMV), helping the establishment of a virion assembly compartment.^38,39 To date, the etiology of WDR11-related diseases remains poorly understood.

Here, we used cryo-electron microscopy (cryo-EM) to determine the structure of the Homo sapiens WDR11-FAM91A1 complex in the monomeric and dimeric states at 3.1 Å and 3.3 Å resolution, respectively. WDR11 consists of two classic WD40 domains followed by an α-solenoid domain. In the monomeric state, the C-terminal domain of FAM91A1 contacts both WD40 domains of WDR11, resulting in a large buried interface. In the dimeric state, WDR11 further dimerizes through intermolecular binding between two α-solenoid domains, forming a FAM91A1-WDR11-WDR11-FAM91A1 complex. Importantly, WDR11 directly and specifically interacts with a subset of acidic clusters, termed super acidic clusters (SACs). WDR11 complex assembly and SAC recognition are indispensable for the trafficking of SAC-containing proteins and proper neuronal development in zebrafish. In summary, our work uncovers the assembly mechanism of the WDR11-FAM91A1 complex and reveals that cargo selectivity can occur at multiple steps of vesicle trafficking.

Results

Cryo-EM Structure of the WDR11-FAM91A1 complex.

WDR11 and FAM91A1 are the most conserved subunits in the WDR11 complex among eukaryotes (Figure 1A and 1B). Furthermore, previous studies showed that deletion of C17orf75 does not impact either the localization or stability of WDR11 and FAM91A1.¹⁸ We thus focused on the WDR11-FAM91A1 complex in our studies. Homo sapiens WDR11 and FAM91A1 were co-expressed and purified via Ni-NTA affinity and gel filtration chromatography. The WDR11-FAM91A1 complex was eluted from a gel filtration column with an estimated molecular mass of ~400 kDa, which was significantly larger than the calculated size of one copy of WDR11 and FAM91A1 (calculated molecular mass: 231 kDa). To more precisely determine the molecular mass of the complex, we employed an 18-angle static light scatterer (SLS) to measure the molecular mass of the WDR11-FAM91A1 complex. SLS showed that the molecular mass of the complex was 462 ± 2 kDa, suggesting that the WDR11-FAM91A1 complex may contain two copies of each protein (calculated molecular mass: 462 kDa) (Figure 1C).

Figure 1. Cryo-EM structure of the human WDR11-FAM91A1 complex.

(A) Presence of WDR11, FAM91A1, and C17orf75 in a variety of common model organisms. Solid green, brown, and gray circles indicate WDR11, FAM91A1, C17orf75, or their homolog, respectively.

(B) Domain organization of WDR11 (green) and FAM91A1 (brown). The names and boundaries of domains are labeled. WD40–1 and WD40–2 indicate the first and second WD40 domain of WDR11, respectively. NTD and CTD indicate the N-terminal and C-terminal domain of FAM91A1, respectively.

(C) Molecular mass of the wild type (WT) WDR11-FAM91A1 complex measured by Static-Light-Scattering (SLS). The horizontal axis is elution volume (Superpose 6 10/300 GL), the left vertical axis is the molecular mass and the right vertical axis is ultraviolet (UV) absorption at 280nm. The Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel shows the peak of the WDR11-FAM91A1 complex from the column.

(D) Two views of the cryo-EM map (left) and model (right) of intact dimeric WDR11-FAM91A1 complex containing FAM91A1’ (red), WDR11’ (blue), WDR11 (green) and FAM91A1 (brown). The bottom view is a 180° rotation along the horizontal axis of the top view.

Subsequently, we employed single-particle cryo-EM to determine the WDR11-FAM91A1 complex in the monomeric and dimeric states, at 3.1 Å and 3.3 Å resolution, respectively (Figure S1A–D). The monomeric state contains one copy of the WDR11-FAM91A1 C-terminal domain (CTD) complex, and FAM91A1 interacts with both WD domains of WDR11. The N-terminal domain (NTD) of FAM91A1 and some α-helices of the WDR11 α-solenoid domain were not resolved, indicating the flexibility of these regions (Figure S1E). The dimeric state contains two copies of WDR11-FAM91A1 CTD (Figure S1E), with clearly resolved side chains that allowed de novo model building (Figure S2). WDR11 further dimerizes through interaction between two α-solenoid domains, resulting in an elongated dumbbell shape (about 239 × 106 × 109 ų) with two WDR11-FAM91A1 interfaces of 1051 Ų and one dimeric WDR11-WDR11 interface of 1080 Ų (Figure 1D).

WDR11 forms a dimer via its α-solenoid domain.

WDR11 consists of two WD40 domains in its N-terminus and an α-solenoid domain in its C-terminus (Figure 1B). WDR11 dimerizes predominately via the C-terminal α-solenoid domain (Figure 2A). Three α-helices (α9-α11) from the α-solenoid domain of each WDR11 monomer mediate the interaction, involving multiple hydrogen bonds, hydrophobic interactions, and Van der Waal forces. Among them, α9 is located at the center of the interaction interface (Figure S2A). A series of hydrophobic amino acids such as Y1027, Y1028, C1029, and L1032 on α9, form a hydrophobic bulge that inserts into the hydrophobic grooves formed by α9′, α10′, and α11′. The same thing happens to α9′ (Figure 2B). Moreover, the side chain of T1038 from α10′ of WDR11′ forms a hydrogen bond with the main chain of S1046 from α9 of WDR11 (Figure 2C). Besides, C1029 of α9 and C1029 of α9′ are close (2.7 Å), and might form an intermolecular disulfide bond (Figure 2D). N1056 of α10 forms a hydrogen bond with Y1027 of α9′ through their side chains. On the other side, the side chain of Y1027 of α9 forms a hydrogen bond with N1056 of α10′ (Figure 2D).

Figure 2. The WDR11-FAM91A1 complex further dimerizes via WDR11.

(A) The high-resolution map of the WDR11-FAM91A1 complex and its cartoon representation.

(B) The dimer interface of the WDR11 and WDR11′ (green and blue) with critical amino acid residues shown.

(C) (D) The view obtained by rotating the (B) view around the horizontal axis and the vertical axis. Yellow dashed lines indicate hydrogen bonds or disulfide bonds.

(E) HEK293T cells were transfected with mCherry-tagged WDR11 WT and GFP-tagged WDR11 WT or WDR11 dimer mutant (WDR11-dm), and then subjected to mCherry-nanotrap for immunoprecipitation. WDR11 and WDR11-dm were detected via antibodies against mCherry and GFP, respectively. Experiments were performed in triplicate.

To verify our structural observation, we generated a WDR11 construct in which we deleted α9 in the dimer interface (WDR11-dm). WDR11 wild-type (WDR11-WT) robustly immunoprecipitated WDR11-WT, consistent with the formation of the WDR11 dimer. In contrast, WDR11-dm failed to immunoprecipitate WDR11-WT, indicating that α9 was critical for the formation of the dimer interface (Figure 2E). This result was further confirmed by our cryo-EM analysis that showed the WDR11-dm-FAM91A1 complex was significantly smaller than the complete dimeric complex (Figure S3A–D). Notably, WDR11-dm retained normal association with FAM91A1 (Figure S3C). These data indicate that the α-solenoid domain is critical for the dimerization of WDR11.

The WDR11-FAM91A1 interface

Two WD40 domains of WDR11 are positioned close to each other, resembling two parallel donuts (Figure 3A). FAM91A1 contacts the lateral surfaces of both WD40 domains mainly through a loop (residues 552 to 568) and an α-helix (residues 584 to 597). The interaction between the WD40–1 domain and FAM91A1 involves electrostatic interactions, as well as hydrophilic and hydrophobic interactions. Residues S48 to Q52 of the WD40–1 domain form an acidic bulge, attracting basic amino acids (R558 and K559) from FAM91A1. Similarly, the side chain of K453 from WDR11 contacts D568 from the loop of FAM91A1. Multiple hydrogen bonds are observed at this interface, including two hydrogen bonds formed between the main chain of A51 and T50 of WDR11 and the side chain of R558 of FAM91A1, two hydrogen bonds between T457 of WDR11 and the main chain of L560 and the side chain of K559 from FAM91A1, and two hydrogen bonds between D348 of WDR11 to S585 and N586 of FAM91A1 (Figure 3B, 3C and S2C). In addition to these electrostatic interactions, L455 and L456 from WDR11 form a hydrophobic cluster with N586, V587, and L588 from FAM91A1 (Figure 3C).

Figure 3. Interaction between WDR11 and FAM91A1.

(A) Two views of the WDR11-FAM91A1 complex protomer are shown in cartoon. WDR11 and FAM91A1 are colored in green and brown, respectively.

(B) (C) (D) The detailed views of the interface between WDR11 and FAM91A1 showed the critical amino acid residues in the interface. Yellow dashed lines indicate hydrogen bonds.

(E) HEK293T cells were transfected with mCherry-tagged WDR11 WT and GFP-tagged FAM91A1 WT, FAM91A1 RRK^EEE, or FAM91A1 Δα mutants, and then subjected to GFP-nanotrap for immunoprecipitation. FAM91A1 and WDR11 were detected via antibodies against GFP and mCherry, respectively. Experiments were performed in triplicate.

(F) HEK293T cells were transfected with mCherry-tagged FAM91A1 WT and GFP-tagged WDR11 WT, WDR11 Δ_SITAQ, KLL^AAA, or KE^AA mutants, and then subjected to mCherry-nanotrap for immunoprecipitation. FAM91A1 and WDR11 were detected via antibodies against mCherry and GFP, respectively. Experiments were performed in triplicate.

Unlike the WD40–1 domain, the WD40–2 domain contacts FAM91A1 predominantly via hydrophilic interactions. E510 of WDR11 forms a salt bridge with R556 from the loop of FAM91A1. In addition, multiple hydrogen bonds are observed at this interface. The side chain of K509 of WDR11 forms two hydrogen bonds with the main chain of K553 and the side chain of S552 of FAM91A1, and a main chain hydrogen bond is formed between S512 of WDR11 and T596 of FAM91A1 (Figure 3D).

To validate our structure, we generated two FAM91A1 mutants by converting key residues from the loop to residues bearing the opposite charge, R556E/R558E/K559E (RRK^EEE), or deleting the α-helix that contacts WDR11, Δα. In contrast to FAM91A1-WT, both FAM91A1 RRK^EEE and Δα failed to immunoprecipitate WDR11 (Figure 3E). Similarly, we mutated several FAM91A1-contacting residues in WDR11, including replacing residues S48 to Q52 by GGGGG (Δ_SITAQ), alanine substitution of key residues from WD40–1, K453A/L455A/L456A (KLL^AAA), or WD40–2, K509A/E510A (KE^AA). Unlike WDR11-WT, all three WDR11 mutants were unable to immunoprecipitate FAM91A1 (Figure 3F).

WDR11 harbors a positively charged groove that directly interacts with acidic-cluster-containing cargoes.

The WDR11 complex is known to facilitate the transport of acidic-cluster-containing cargoes from endosomes to the TGN.¹⁸ However, its underlying mechanism remains unknown. Careful inspection of the WDR11-FAM91A1 complex revealed that WDR11 possesses a positively charged groove located on the top of its WD40–2 domain (Figure 4A and 4B). Multiple lysine residues contribute to the formation of the grove, including K9, K479, K482, K719, K761, and K808. To test whether this region could be involved in contacting the acidic clusters in the cargo proteins, we generated a series of single or double mutations on WDR11 (K9D, K479D/K482D (2KD), K719D, K761D, and K808D), and assessed the binding with CI-MPR. Our results find that WDR11 WT could precipitate CI-MPR WT from cells. In contrast, all the WDR11 mutants dramatically decreased binding to CI-MPR (Figure 4C). These results indicate that WDR11 might utilize its positively charged groove to contact the acidic clusters in the cargoes.

Figure 4. WDR11 directly interacts with the acidic-clusters-containing cargoes.

(A) The structure of WDR11 is shown in cartoon and the domains are labeled.

(B) Electrostatic surface potential map of WDR11 and zoomed-in picture of a region rich in basic amino acids. The lysine residues in this region are labeled.

(C) HEK293T cells were transfected with Venus-tagged CI-MPR cytosolic tail WT and mCherry-tagged WDR11 WT, WDR11 K9D, 2KD (K479D+K482D), K719D, K761D, or K808D mutants, and then subjected to mCherry-nanotrap for immunoprecipitation. CI-MPR and WDR11 were detected via antibodies against GFP (Venus) and mCherry, respectively. Experiments were performed in triplicate.

(D) The sequence arrangement of the wild type CI-MPR cytosolic tail (WT), acidic cluster 1 (AC1), and acidic cluster 2 (AC2).

(E) His-pulldown assays performed with the complex of His-FAM91A1 and WDR11 WT and purified GST-CI-MPR WT, GST-CI-MPR AC1, GST-CI-MPR AC2, or GST. GST-CI-MPR WT, GST-CI-MPR AC1, GST-CI-MPR AC2, or GST is marked with red stars, and the complex of His-FAM91A1 and WDR11 WT is denoted by red triangles. Experiments were performed in triplicate.

(F) Bio-layer interferometry (BLI) binding studies of the WDR11-FAM91A1 complex with CI-MPR acidic cluster 1 peptide (AC1 WT). The purified WDR11-FAM91A1 complex was fixed to the probe and different concentrations of AC-1 WT peptide were present in the pool below. Experiments were performed in triplicate.

The cytoplasmic tail of CI-MPR harbors two acidic clusters: acidic cluster 1 (AC1) and acidic cluster 2 (AC2).¹⁸ To assess whether the two acidic clusters bind to the WDR11 complex similarly, we generated two internally deleted constructs, CI-MPR-AC1 and CI-MPR-AC2 (Figure 4D). We observed that CI-MPR-WT and CI-MPR-AC1 precipitated much more WDR11 than CI-MPR-AC2 and the vector (Figure S4A). To test whether the WDR11 complex directly interacts with the acidic clusters, we purified GST fused to the cytoplasmic tail of CI-MPR, or fragments harboring either AC1 or AC2, and tested their binding to the purified WDR11 complex. We observed that GST-tagged CI-MPR, but not GST alone, robustly bound to the WDR11 complex (Figure 4E), indicating that the interaction between the WDR11 complex and CI-MPR was direct. Furthermore, like in the immunoprecipitation experiments, purified GST-CI-MPR-AC1, but not GST-CI-MPR-AC2, bound to the WDR11 complex (Figure 4E). Using bio-layer interferometry (BLI), we determined that the AC1-WT peptide bound to WDR11 with an affinity of 20.8 ± 0.2 μM (Figure 4F). This affinity was similar to that between acidic clusters and the AP-1 complex and fell within the normal range for interactions between cargo proteins and the protein transport machinery.¹⁶

In addition to CI-MPR, previous studies have identified several acidic-cluster-containing cargo proteins, including KIAA0319L, furin, and CPD.^18,40 Indeed, all of them displayed some degree of binding to WDR11 (Figure S4B). Among them, CI-MPR displayed the strongest association with WDR11, followed by KIAA0319L and CPD, with furin exhibiting the weakest interaction (Figure S4B). Taken together, our study indicates that WDR11 could function as a receptor for acidic cluster-containing cargoes by directly engaging them through its positively charged groove.

WDR11-binding proteins harbor a Super Acidic Cluster (SAC)

Protein transport machineries often interact with their cargoes transiently and weakly. In order to identify additional proteins that interact with WDR11 and undergo transport in a WDR11-dependent manner, we performed proximity biotinylation using the promiscuous biotin ligase miniTurbo to biotinylate proteins within a ~10 nm radius.^41,42 We generated two constructs expressing mCherry and miniTurbo: one with wild-type WDR11 (mCherry-WDR11-WT-miniTurbo) and the other with a 2KD mutant that displayed a decreased binding to CI-MPR (mCherry-WDR11–2KD-miniTurbo). The two constructs were transfected into HEK293T cells for 24 hours. Subsequently, cells were exposed to 100 μM biotin for 1 hour, and the biotinylated proteins were isolated and subjected to mass spectrometry analysis (Figure S5A). Expression of the tagged and immunoprecipitated proteins in each group was verified through immunoblot analysis (Figure S5B). In comparison to the WDR11-WT interactome, 183 proteins in the WDR11–2KD interactome exhibited a significant decrease (Figure 5A). Subsequent GO/KEGG analysis of these 183 proteins revealed their predominant enrichment in vesicular transport-related pathways (Figure 5B). Among these 183 proteins, 27 were classified as transmembrane proteins. Several acidic-cluster-containing proteins, including CI-MPR, KIAA0319L, and CPD, were in our list (Figure 5A). Consistent with our mass spectrometric results, immunoblot analysis indicated that CI-MPR and KIAA0319L were significantly enriched in the WDR11-WT group relative to both the vector and WDR11–2KD groups (Figure S5C).

Figure 5. Proximity proteomics reveal that WDR11-binding proteins harbor a Super Acidic Cluster.

(A) Volcano plot of quantitative analysis of comparative interactome of WDR11–2KD vs WDR11-WT across n=3 independent experiments using One-sample t-test and Benjamini-Hochberg FDR. Red dots indicate up-regulation in WDR11–2KD compared to WDR11-WT cells. Blue dots signify down-regulation in WDR11–2KD compared to WDR11-WT cells. Gray dots represent no significant difference between WDR11–2KD and WDR11-WT cells.

(B) The GO/KEGG enrichment analysis of 183 proteins (fold change < 0.667; p < 0.05, t-test) significantly reduced in the WDR11–2KD group compared to the WDR11-WT group.

(C) HEK293T cells were transfected with mCherry-tagged WDR11 and Venus, Venus-tagged CI-MPR cytosolic tail, ATG9A cytosolic tail, TMEM87B cytosolic tail, SV2A cytosolic tail, MERTK cytosolic tail, VAMP4 cytosolic tail, and VAMP7 cytosolic tail, respectively. Then subjected to mCherry-nanotrap for immunoprecipitation. WDR11 was detected via antibody against mCherry. CI-MPR, ATG9A, TMEM87B, SV2A, MERTK, VAMP4, and VAMP7 were detected via antibody against GFP (Venus). Experiments were performed in triplicate.

(D) HEK293T cells were transfected with mCherry-tagged WDR11-WT or mCherry-tagged WDR11–2KD and Venus-tagged CI-MPR cytosolic tail, ATG9A cytosolic tail, TMEM87B cytosolic tail, SV2A cytosolic tail, MERTK cytosolic tail, and VAMP4 cytosolic tail, respectively. Then subjected to mCherry-nanotrap for immunoprecipitation. WDR11-WT and WDR11–2KD were detected via antibody against mCherry. CI-MPR, ATG9A, TMEM87B, SV2A, MERTK, and VAMP4 were detected via antibody against GFP (Venus). Experiments were performed in triplicate.

(E) Alignment of the proteins with cytosolic tail containing acidic clusters bound to WDR11 (Φ = bulky hydrophobic amino acids) established by comparative analysis.

(F) Affinity results of CI-MPR-AC1 WT and its mutants with the WDR11-FAM91A1 complex.

(G) The model of the WDR11-CI-MPR-AC1 complex predicted by AlphaFold2. WDR11 was shown in an electrostatic surface potential map, and CI-MPR-AC1 was shown in cartoon. The critical amino acids in the interface were shown in sticks and labeled.

Among the 27 transmembrane proteins, we chose several for further validation: ATG9A, a phospholipid scramblase critical for autophagy;⁴³ TMEM87B, a poorly-characterized protein associated with endosome-to-TGN trafficking;⁴⁴ synaptic vesicle glycoprotein 2A (SV2A), a protein associated with synaptic vesicles;^45,46 MERTK, a receptor tyrosine kinase;⁴⁷ and VAMP4 and VAMP7, members of the vesicle-associated membrane protein (VAMP)/synaptobrevin family of SNAREs.^48–51 Immuno-precipitation experiments conducted using mCherry-tagged WDR11 and Venus-tagged cytosolic tails of these transmembrane proteins indicated that ATG9A, TMEM87B, SV2A, MERTK, and VAMP4 all exhibited substantial interactions with WDR11, whereas VAMP7 displayed a weaker binding (Figure 5C). Consistent with our mass spectrometric results, ATG9A, TMEM87B, SV2A, MERTK, and VAMP4 interacted with WDR11–2KD much more weakly than with WDR11-WT (Figure 5D). When aligning the cytoplasmic tails of these proteins, we noticed that all of them and many other WDR11-interacting transmembrane proteins in our mass spectrometric study contain at least one acidic cluster with approximately 6–10 E/D/S residues. Furthermore, the acidic amino acids are often flanked by one or two bulky hydrophobic residues (Figure 5E). Here, we designate sequences with around 6–10 E/D/S residues, flanked on each side by one or two bulky hydrophobic residues, as super acidic clusters (SACs). In contrast, the acidic clusters previously described neither include the flanking bulky hydrophobic residues, nor define the number of acidic residues.¹⁶

To validate the importance of these sequence features for WDR11 binding, we generated a series of CI-MPR-AC1 mutants and determined their affinity with the WDR11 complex using BLI. Converting acidic residues to alanine in AC1 (8A) decreased the binding by ~ 7-fold, demonstrating the importance of the acidic residues (Figure 5F). We also varied the number of acidic residues and found that the peptide with 8 acidic residues displayed the strongest binding (Figure 5F). When we decreased (4D, 6D) or increased (10D, 12D) the number of the acidic residues, the affinity decreased 2 to 6-fold relative to CI-MPR-AC1-WT (Figure 5F). In addition to these acidic residues, the flanking bulky hydrophobic residues were also critical, as replacing them by alanines (L^A and VL^AA) significantly reduced affinity (Figure 5F). Similar to WDR11, AP-1 contacted both the acidic residues and the flanking hydrophobic residues in CI-MPR (Figure S5D). The addition of AP-1 complex subunit μ1, which is known to bind specific cargoes,⁵² reduces the binding between CI-MPR and WDR11 (Figure S5E). The data suggested that both AP-1 and WDR11 recognize cargo proteins harboring the SAC motif. These data explained many of our previous observations, such as the difference between AC1 and AC2 of CI-MPR (AC2 has fewer acidic residues), and indicated that an optimal SAC sequence contains ~ 8 acidic residues flanked by bulky hydrophobic residues.

To gain more molecular insights into the interaction between CI-MPR-AC1 and WDR11, we used AlphaFold2 to generate a model for the WDR11-AC1 complex (Figure S5F).⁵³ In this model, the AC1 peptide binds to the positively charged groove in WDR11, as expected. D2411 and E2412 of CI-MPR form salt bridges with K479 and K761 of WDR11, respectively. Proximal to the positively charged groove on the surface of WDR11, two hydrophobic pockets are observed, which could accommodate L2402, V2413, and L2414 of CI-MPR, respectively (Figure 5G). The distance between the two pockets could also explain the requirement for the number of acidic residues in SAC. Furthermore, depletion of TBC1D23, golgin-97, golgin-245, or Arl1 did not alter the interaction between WDR11 and CI-MPR (Figure S5G–J).⁵⁴ Taken together, we establish SAC as a novel protein motif, which can be found in many WDR11-binding proteins.

The WDR11 complex is specifically required for the intracellular trafficking of SAC-containing cargo proteins.

Next, we sought to determine whether the assembly of the WDR11 complex and its interaction with SAC-containing proteins were required for intracellular trafficking. To this end, we generated WDR11 polyclonal knockout (KO) HeLa cells using CRISPR/Cas9 technology, with a knockout efficiency of ~ 90 % (Figure S6A). WDR11-KO cells displayed a dispersed distribution of CI-MPR relative to control cells, consistent with previous studies.¹⁸ The colocalization between CI-MPR and the TGN marker golgin-97 was significantly decreased in WDR11-KO cells, relative to control cells (Figure 6A and 6B). Next, we re-introduced WDR11 WT, WDR11 dimerization mutant (dm), FAM91A1-binding mutant (KE^AA), or the SAC-binding mutant (2KD) in WDR11 KO cells, and assessed the intracellular localization of CI-MPR. WDR11 WT, but not the three mutants or the empty vector, rescued CI-MPR mis-sorting, as indicated by the colocalization of CI-MPR and golgin-97 (Figure 6A and 6B). Similar to CI-MPR, the assembly of the WDR11 complex and their interactions were also indispensable for the proper intracellular localization of KIAA0319L (Figure S6B and S6C).

Figure 6. The WDR11-FAM91A1 complex assembly and its interaction with cargoes are required for SAC-containing protein trafficking.

(A) Subcellular location of CI-MPR in HeLa cells. The WDR11 KO cells were transfected with mCherry, mCherry-tagged WDR11 WT, dm, KE^AA, or 2KD, respectively. Cells were then incubated with antibodies against CI-MPR (purple) and golgin-97 (green). The golgin-97 is a marker for the TGN region. Scale bar: 10 μm. Experiments were performed in triplicate.

(B) Colocalization analysis between CI-MPR and golgin-97 in (A). Each dot represents Pearson’s correlation coefficients from one cell. Con: n=78 cells; mCherry: n=71 cells; WT: n=57 cells; dm: n=47 cells; KE^AA: n=55 cells; 2KD: n=60 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(C) Subcellular location of ATG9A in HeLa cells. The WDR11 KO cells were transfected with mCherry, mCherry-tagged WDR11 WT, dm, KE^AA, or 2KD, respectively. Cells were then incubated with antibodies against ATG9A (purple) and golgin-97 (green). Scale bar: 10 μm. Experiments were performed in triplicate.

(D) Colocalization analysis between CI-MPR and golgin-97 in (C). Each dot represents Pearson’s correlation coefficients from one cell. Con: n=40 cells; mCherry: n=38 cells; WT: n=41 cells; dm: n=34 cells; KE^AA: n=42 cells; 2KD: n=40 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(E) Subcellular location of CI-MPR in HeLa cells. The HeLa cells were transfected with mCherry-GRIP (a marker of trans-Golgi) and Venus-tagged CI-MPR cytosolic tails -WT, -AC2, -AC1-L^A, -AC1-VL^AA, -AC1–4D, or -AC1–12D, respectively. Cells were fixed, permeabilized, and stained for the cell nuclei before observation using fluorescence microscopy. Scale bar: 10 μm. Experiments were performed in triplicate.

(F) Colocalization analysis between CI-MPR and GRIP in (E). Each dot represents Pearson’s correlation coefficients from one cell. Vector: n=67 cells; WT: n=42 cells; AC2: n=61 cells; L^A: n=47 cells; VL^AA: n=54 cells; 4D: n=51 cells; 12D: n=42 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

To assess how WDR11 regulates the transport of SAC-containing cargoes, we focused on ATG9A, the only multi-pass transmembrane protein among the core ATG proteins, and a crucial regulator of autophagy. Although it is known that ATG9A traffics between the TGN and endosomes, it was unclear whether WDR11 plays a role in ATG9A trafficking. Deletion of WDR11 resulted in a reduced colocalization of ATG9A with golgin-97 relative to control cells, indicating that WDR11 is critical for the translocation of ATG9A to the TGN (Figure 6C and 6D). Notably, ATG9A mis-sorting could be rescued by reintroduction of WDR11 WT, but not by the assembly-deficient mutants (dm or KE^AA) or mutants lacking binding to SAC-containing cargos (2KD) (Figure 6C and 6D). Similar to WDR11, depletion of AP-1 also decreased colocalization of ATG9A with golgin-97 relative to control cells (Figure S6D, S6E, and S6F). Thus, our study indicates that WDR11 regulates the intracellular transport of ATG9A as it does for other SAC-containing cargoes.

To evaluate the functional importance of the conserved residues in SAC, we employed an assay first developed by Waguri et al. and frequently utilized by many groups.^55,56 In this assay, we combined the CI-MPR cytoplasmic tail with an endosome- and Golgi-targeting signal peptide and a fluorescent tag, resulting in a Venus-CI-MPR construct (signal peptide-Venus-Transmembrane domain-CI-MPR).^57,58 Venus-CI-MPR-WT displayed strong colocalization with GRIP (a region encompassing residues 688 to 767 of golgin-97),⁵⁹ similar to previous studies (Figure 6E).²² Introduction of multiple mutations (AC2, L^A, VL^AA, 4D, and 12D) into the Venus-CI-MPR construct, which decreased the binding to WDR11, significantly decreased the colocalization of CI-MPR with GRIP (Figure 6F). Depletion of golgin-97/245 or TBC1D23 also reduced colocalization of CI-MPR-WT, but not CI-MPR-L^A, with GRIP, indicating that WDR11 cooperates with golgin-97/245 or TBC1D23 to regulate the trafficking of CI-MPR (Figure S6G, S6H, and S6I). In addition to CI-MPR, we also found that the SAC motif of VAMP4 was necessary for its endosome-to-TGN trafficking (Figure S6J and S6K). Furthermore, a mutation within the SAC motif of VAMP4, VAMP4–9A, decreased the trafficking to the TGN and the association with its cognate t-SNARE proteins, VTI1A and STX6 (Figure S6L).⁵⁰

To determine whether WDR11 specifically regulates the trafficking of SAC-containing cargo proteins, we assess the impact of WDR11 deletion on LDLR (Low-density lipoprotein receptor) trafficking. LDLR contains a YxxΦ motif, a NPxY motif and two small acidic clusters (EDE and EDD), yet it does not possess a SAC motif.⁶⁰ LDLR undergoes endosome-to-TGN trafficking, and is found to associate with AP-1 in a proteomic study.⁶¹ Deletion of WDR11 did not affect the colocalization between LDLR and golgin-97 and the LDLR protein level, indicating that WDR11 likely only regulates endosome-to-TGN trafficking of SAC-containing proteins (Figure S7A–D). In addition to mediating endosome-to-TGN trafficking, AP-1 is also known to control TGN-to-endosome trafficking. For instance, AP-1 sorts phosphorylated STING (lacking the SAC motif but possessing the [D/E]xxxL[L/I/M] motif) from the TGN to the endolysosomal system, where STING is degraded.⁶² To test the impact of WDR11 on STING trafficking, we used human cerebral microvascular endothelial cells (which have a functional cGAS-STING pathway) using the doxycycline (dox)-inducible deletion of WDR11 and found that WDR11 did not affect the STING protein level (Figure S7E and S7F). Thus, WDR11 is specifically responsible for the endosome-to-TGN trafficking of SAC-containing proteins (Figure S7G). This process requires the proper assembly of the WDR11 complex and its interaction with SAC-containing proteins.

Assembly of the WDR11 complex and its interaction with SACs are required for neuronal development in zebrafish.

Individuals with mutations in WDR11 display multiple defects similar to PCH, including intellectual disability, microcephaly, and cerebellar hypoplasia.³⁰ As we and others have previously demonstrated that the knockdown of TBC1D23 or FAM91A1 in zebrafish can mimic many features of PCH patients, we chose zebrafish as a model to investigate the role of WDR11 in neuronal development.^22,28,29 To this end, we used one previously reported antisense morpholino (MO), which targeted the exon 3-intron 3 splicing site and resulted in an in-frame stop codon.³⁴ The injection of wdr11 MO significantly decreased the axon length of CaP motor neurons compared to control MO, in Tg[Hb9: GFP]^ml2 transgenic zebrafish. Critically, the defect could be effectively rescued by co-injection of mRNA encoding wild-type human WDR11. Conversely, co-injection of mRNAs encoding the WDR11 dimerization mutant (dm), FAM91A1-binding mutant (KE^AA), or acidic-cluster-binding mutant (2KD) failed to rescue the defect (Figure 7A and 7B).

Figure 7. The WDR11-FAM91A1 complex assembly is critical for neuronal development in zebrafish.

(A) CaP motor neuron axons of [Hb9: GFP]^ml2 transgenic zebrafish embryos at 48 hpf that were injected with wdr11 MO or coinjected with indicated mRNAs. All injections were performed at the one-cell stage of embryos. Lateral views (Top); enlarged views of one axon (Bottom). Rectangles indicate axons shown in the Bottom. Scale bar: 100 μm. Experiments were performed in triplicate.

(B) Statistical results of the length of Cap neuron axons were treated as in A. For each embryo, 3 axons were counted, and for each group, 10 Tg[Hb9: GFP]^ml2 zebrafish embryos were scored. P values were calculated using nested one-way ANOVA, Tukey’s multiple comparisons test.

(C) HuC(green) expression in Tg[HuC: GFP] transgenic zebrafish at 48 hpf. Classification of embryos was based on the expression level of HuC (elavl3) at 48 hpf. C1, normal; C2, moderate defects; C3, severe defects. Lateral views (Top); dorsal views (Bottom). Arrows indicate the morphology of neurons in the midbrain. Scale bar: 100 μm. Experiments were performed in triplicate.

(D) Percentage of embryos in each group upon injection of wdr11 MO or coinjection of indicated mRNAs. n stands for the number of embryos counted for analysis.

(E) CaP axons of Tg[Hb9: GFP]^ml2 zebrafish embryos at 48 hpf that were injected with wdr11 MO or coinjected with different concentrations of KIAA0319L mRNA. All injections were performed at the one-cell stage of embryos. Lateral views (Top); enlarged views of one axon (Bottom). Rectangles indicate axons shown in the Bottom. Scale bar: 100 μm. Experiments were performed in triplicate.

(F) Statistical results of the length of Cap axons that were treated as in E. Con：n=9 embryos; wdr11 MO: n=9 embryos; wdr11 MO + KIAA0319L 50pg: n=10 embryos; wdr11 MO + KIAA0319L 75pg: n=10 embryos; wdr11 MO + KIAA0319L 100pg: n=10 embryos. P values were calculated using nested one-way ANOVA, Tukey’s multiple comparisons test.

(G) A model showing how the WDR11-FAM91A1 complex selectively regulates the trafficking of SAC-containing cargoes. WDR11 directly recognizes the SAC motif in the cytoplasmic tail of cargo proteins, and the recognition is further enhanced by the dimerization of WDR11. FAM91A1, via its N-terminus, directly interacts with TBC1D23 which localizes on the TGN via binding to golgin-97/245. Thus, the WDR11-FAM91A1 complex and TBC1D23 cooperate to promote the endosome-to-TGN trafficking of SAC-containing cargoes. In addition to FAM91A1, TBC1D23 also binds to the WASH complex localized on the endosomal vesicles, which may explain the role of TBC1D23 in transporting proteins without the SAC motif.

Furthermore, we found that the expression of Huc (elavl3), an early marker of pan-neuronal cells, was dramatically decreased after MO knockdown of WDR11 especially in the midbrain area of the brain. Co-injection of wild-type WDR11 mRNA also partially rescued the loss of mature neurons (~50%). Consistent with what we found on the motor neuron, all mutants of the WDR11 protein failed to rescue this phenotype (Figure 7C and 7D). Taken together, our results suggest that the assembly of the WDR11 complex and interaction with cargoes are critical for its functions in regulating neuronal development.

KIAA0319L is a transmembrane protein participating in axon guidance, and undergoes endosome-to-TGN transport in a WDR11- and FAM91A1-dependent manner.^22,63 Due to the reduction in the protein level of KIAA0319L upon the deletion of FAM91A1, we postulated that the observed defects in zebrafish depleted of WDR11 could be partly attributed to the diminished levels of KIAA0319L. To validate this hypothesis, we co-injected KIAA0319L mRNA with wdr11 MO in Tg [Hb9: GFP]^ml2 transgenic zebrafish. Remarkably, co-injection of KIAA0319L at various concentrations partially rescued the axon length of CaP motor neurons in the wdr11 MO embryos (Figure 7E and 7F). In summary, our data suggest that the WDR11 complex regulates neuronal development, at least partially, through controlling intracellular trafficking of cargo proteins such as KIAA0319L.

Discussion

Selective protein trafficking is one of the most fundamental processes in biology. It is widely accepted that cargo selection is predominantly mediated by coat proteins and their associated proteins at the donor membrane. Here, we present the cryo-EM structure of the WDR11-FAM91A1 complex, which functions in vesicle tethering together with proteins such as TBC1D23. We show that WDR11 directly and specifically recognizes the SAC motif and is specifically responsible for the intracellular trafficking of SAC-containing proteins. Thus, direct sequence recognition could exist in a later step of vesicle trafficking (Figure 7G). We propose that such a sequence-specific recognition mechanism could allow the target membrane to select certain cargo proteins and further enhance the fidelity of vesicle trafficking.

Coat protein complexes and their associated proteins play a dominant role in selecting cargo for trafficking. For instance, AP-1 is known to sort cargo proteins from endosomes or the TGN into specific transport vesicles for delivery to their target destinations. AP-1 recognizes multiple sorting signals on the cytoplasmic tails of cargo proteins, including YxxΦ, [D/E]xxxL[L/I/M], and the acidic cluster. Following vesicle budding, AP-1 is then rapidly dissociated. The WDR11 complex, operating downstream of AP-1, specifically recognizes the SAC motif and is specifically responsible for the delivery of SAC-containing proteins to the Golgi. Such a selection mechanism could prioritize the delivery of SAC-containing proteins, among different cargoes, and further enhance their delivery accuracy. Although cargo selection by coats and associated proteins is a well-established dogma in the field, our studies demonstrate that sequence-dependent selection could exist in later steps of vesicular trafficking.

Our studies have identified a large set of WDR11 putative cargoes, and linked their functions with various physiological and pathological conditions. Individuals with WDR11 mutations displayed PCH-like symptoms, including cerebellar hypoplasia and intellectual disability. Our studies indicate that mis-trafficking of cargo proteins, such as KIAA0319L, could be at least partially responsible for the pathogenesis of these symptoms. Multiple SNARE proteins, such as VAMP4, VAMP7, and VTI1B, contain a SAC motif, indicating that WDR11 may be critical for SNARE complex-mediated membrane fusion processes.^64,65 Furthermore, the autophagy core protein ATG9A and E3 ubiquitin-protein ligases RNF149 and HECD4 also harbor a SAC motif, suggesting that WDR11 might also be involved in the regulation of autophagy and protein quality control.^66,67

A multitude of viruses are known to hijack either the AP-1 or the WDR11 complex to facilitate immune evasion and promote viral infection. For instance, the HIV-1 Nef protein exploits the interaction with AP-1 to help immune evasion and to enhance the release of viral particles from infected cells. Nef harbors an acidic cluster (WLEAQQEEEEV), which is important to interact with the μ subunit of the AP-1 complex.⁶⁸ This sequence conforms to our defined SAC motif, indicating that Nef may also interact with WDR11 and interfere with its functions. Furthermore, herpes simplex virus (HSV) and human cytomegalovirus (HCMV) are known to exploit WDR11 during infection. We were able to detect SAC-like motifs from multiple viral proteins known to interact with WDR11, including ICP0 from HSV (“LPDSSDSEAETEV”), pp28 (“MQETDDLDEEDTSIY”), UL44 (“FNDAKEESDSEDSV”) and UL45 from HCMV (“IGGDDSELEEGPL”).^38,39 Thus, these viral proteins could interfere with host AP-1 and WDR11 complexes through a mechanism analogous to that of HIV-1 Nef.

Many tethering factors, such as COG, GARP, and HOPS, are known to bind to SNAREs. For example, HOPS subunit Vps33 interacts with both the R SNARE Nyv1 and the Qa SNARE Vam3;⁶⁹ COG interacts with Syntaxin5a in the intra-Golgi SNARE complex,⁷⁰ and the GARP complex interacts with the STX16 TGN SNARE complex (STX16/STX6/VTI1A/VAMP4).⁷¹ Our studies differ from these previous findings from at least three perspectives. First, previous studies have focused on how tether-SNARE interactions regulate SNARE assembly and membrane fusion, whereas our studies emphasize that selective cargo recognition of incoming vesicles by tethers to facilitate vesicle trafficking. Second, other studies have reported multiple binding modes between SNARE proteins and tethers; however, the SAC motif, or other similar motifs have not been previously reported. Last but not the least, previous studies have limited the binding to SNAREs, we, however, have identified that a large set of membrane proteins contain the SAC motif and show that their intracellular transport requires this motif. Thus, we have identified a generalized principle for cargo recognition and intracellular transport through the interaction of the WDR11-FAM91A1 complex with cargoes containing the SAC motif.

While our study was under review, the Munro and Owen groups reported that the tethering factor TBC1D23 could recognize a TLY motif, in cargoes.⁷² One limitation of their study is the lack of functional data – whether the TLY motif is important for protein trafficking. In contrast, we performed comprehensive experiments and established that the SAC motif is not only directly recognized by WDR11, but is critical for intracellular trafficking of multiple proteins. Moreover, we revealed the physiological importance of the WDR11-SAC interaction by demonstrating its critical role in neuronal development. Our study, together with the one from the Munro and Owen groups, collectively reveal that sequence-dependent selection is not limited to coat proteins in vesicular trafficking. It remains to be determined whether similar mechanisms could be found in other vesicular trafficking routes.

The domain organization of WDR11 is commonly observed in certain subunits of the HOPS/CORVET tethering complex, vesicle coats, nuclear pore components, and subunits of the intraflagellar transport complex.⁷³ The protein mostly analogous to WDR11 is β′-COP in the COPI complex, which forms a triskelion via its N-terminal WD40 domain.⁷⁴ In contrast, WDR11 dimerizes through its C-terminal α-solenoid domain. It remains unknown whether the WDR11 complex could further oligomerize, similar to COPI and other coat protein complexes.⁷⁴ Future studies will be necessary to understand how the WDR11 complex is recruited to the membrane and how C17orf75 regulates the functions and organization of WDR11-FAM91A1. In summary, our study illustrates the structure and function of the WDR11-FAM91A1 complex, and discovers that cargo selection can occur at the later step of vesicular trafficking to further enhance the transport fidelity.

Limitations of the study

Previous studies have documented the formation of a ternary complex involving WDR11, FAM91A1, and C17orf75 in vertebrates. Although C17orf75 is dispensable for the location and stability of the WDR11 complex, our current studies did not provide structural and functional information of C17orf75. In vitro reconstitution systems using purified proteins and liposomes have been widely used to illustrate mechanisms of vesicle budding, vesicle trafficking, and cargo specificity. Such a system could be utilized to further understand the roles of WDR11 and its relationship with AP-1.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Da Jia (Jiada@scu.edu.cn).

Materials availability

All plasmids and cell lines used in this study are available from the lead contact with a completed material transfer agreement.

Data and code availability

• The mass spectrometry raw data were deposited to the ProteomeXchange Consortium with dataset identifier IPX0008905000 via the iProx partner repository. The data are publicly accessible in iProX at the following link: https://www.iprox.cn/page/project.html?id=IPX0008905000. All data relevant to the results of this study can be located in the article and its supplementary materials. The cryo-EM maps and associated atomic coordinate models have been deposited in the wwPDB OneDep System under EMD accession code EMD-38312 and PDB accession code 8XG3 for monomeric WDR11-FAM91A1 complex, EMD accession code EMD-38300 and PDB accession code 8XFB for partial dimeric WDR11-FAM91A1 complex, EMD accession code EMD-39863 and PDB accession code 8Z9M for dimeric WDR11-FAM91A1 complex, EMD accession code EMD-39947 and EMD accession code EMD-39943 for multi-body 1 and 2 of dimeric WDR11-FAM91A1 complex, and EMD-39949 for WDR11-dm-FAM91A1 complex.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Zebrafish

Zebrafish (Danio rerio) were kept in accordance with the University of Sichuan Institute animal welfare guidelines. All experiments were approved by the Animal Welfare Center of West China Hospital of Sichuan University. Both adult fish and embryos were raised at 28.5 °C. Tg Hb9:GFP (WT) and Tg Huc:GFP (WT) zebrafish strains were used. All zebrafish larvae at 0–48 hours post fertilization (hpf) were used. During this developmental stage, the sex cannot be determined.

Bacterial strains

E. coli DH5α cells were used to produce recombinant DNA plasmids. E. coli BL21 (DE3) cells were used for the generation of recombinant proteins. E. coli DH10 cells were used to produce bacmids.

Cell lines

Spodoptera frugiperda Sf9 cells for protein expression were grown at 27°C and 110 rpm in SIM SF medium. HEK293T(ATCC), HeLa (ATCC), and COS7(ATCC) cell lines were cultured in DMEM medium supplemented with 10% fetal bovine serum at 37°C and 5% CO₂.

METHOD DETAILS

Expression and purification of the WDR11-FAM91A1 complex

cDNAs encoding human FAM91A1 and WDR11 were sub-cloned into a pFastBacDual expression vector, downstream of the pPH and pP10 promoters, respectively. His₈ tag was fused to the N-terminus of the FAM91A1. The complex was expressed in Sf9 insect cells, similar to previous studies, and the infected cells were harvested 96 hours post-infection.^84,85 The cells were lysed in a buffer containing 20 mM Tris (pH 8.0), 500 mM NaCl, 1 mM PMSF, 20 mM Imidazole, and 5% (v/v) glycerol, and disrupted using an ultrasonic crusher. After high-speed centrifugation (20,000 g), the supernatant was retained and incubated with Ni-NTA beads at 4°C for 20 minutes. Following washing and elution, the proteins were further purified using Superdex 6 increase gel filtration chromatography in a buffer containing 20 mM Tris (pH 8.0) and 500 mM NaCl.

Expression and purification of recombinant proteins in E. coli

cDNA encoding the cytoplasmic tail of human CI-MPR (residues 2398–2491) was sub-cloned into a pGEX-4T-1 vector. Mutants were generated by Gibson cloning. cDNA encoding the AP-1 complex subunit μ1 (residues 158–423) was sub-cloned into a pET28a vector. The plasmid were transferred into DE3 cells and then cultured to OD₆₀₀ 0.6–0.8. Then, 1mM Isopropyl-β-D-thiogalactopyranoside (IPTG), was added and cultured for 16 hours at 18 °C. The cells were lysed by sonication in lysis buffer (25 mM Tris, PH8.0, 500 mM NaCl). After centrifugation at 22000 ×g, the supernatant was passed through glutathione or Ni-NTA beads. Then, the elution was purified by size exclusion chromatography.

Static light scattering (SLS) analysis

Static light scattering (SLS) experiments were conducted using an AKTA pure system, coupled with a Wyatt Dawn Heleos II Multi-Angle Light Scattering (MALS) detector (Wyatt Technology). Samples (100 μg) were loaded onto a Superdex 6 increase gel filtration column and eluted with PBS at a flow rate of 0.35 mL/min. The SLS system was calibrated using purified BSA. Data acquisition and analysis were performed using Astra 7 software (Wyatt Technology).

Cryo-grid preparation and EM data collection

Three microliters of protein solution were applied onto glow-discharged (60 s) 300-mesh Cu R2/1 Quantifoil grids. The grids were blotted for 2–3 s and rapidly frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher). Samples were loaded in a Titan Krios cryo-electron microscope (Thermo Fisher) operated at 300 kV in the following settings: condenser lens aperture 50 μm, spot size 6, magnification at 165,000× (corresponding to a calibrated sampling of 0.85 Å per physical pixel), and a K2 direct electron device equipped with a BioQuantum energy filter operated at 20 eV (Gatan). Movie stacks were collected automatically using EPU2 software (Thermo Fisher) with the K2 detector operating in counting mode at a recording rate of 5 raw frames per second and a total exposure time of 6 s, yielding 30 frames per movie stack and a total dose of 63 e–/Å2.

Cryo-EM data processing

For the monomeric WDR11-FAM91A1 complex data, 7,086 movie stacks were imported in cryoSPARC 4.31 and subjected to Patch Motion Correction and Patch CTF Estimation⁷⁷. Micrographs were subjected to Blob Picker and 2,685,091 particles were extracted. After two rounds of 2D classification and several rounds of ab initio and heterogeneous refinement (Figure S1A), 127,810 particles of monomeric WDR11-FAM91A1 complex were subjected to non-uniform (NU) refinement that yielded the final reconstruction at 3.1 Å resolution (Figure S1B). DeepEMhancer was used to process the map for presentation purpose.⁸⁶ 206,266 particles of another state containing one complete monomeric WDR11-FAM91A1 complex and the dimer interface region were subjected to another round of template-matching particle picking, heterogeneous refinement and 3D classification that resulted in 35,698 particles for local refinement to yield the final reconstruction at 3.4 Å resolution (Figure S1C). The resolution was estimated by the Fourier shell correlation (FSC) curve using the 0.143 criterion.

We collected another dataset of 10,204 movie stacks of the intact dimeric WDR11-FAM91A1 complex, followed by the same data processing workflow to generate preliminary 2D averages. Topaz particle picking resulted in 862,907particles,⁸⁷ and subsequent 2D and 3D classifications, 3D auto-refinement and Bayesian polishing resulted in 232,473 particles,⁸⁸ which were subjected to 3D multi-body refinement of two separated rigid bodies, WDR11-FAM91A1 and WDR11’-FAM91A1’.⁸⁹ Focused refinements of two separated bodies (1 and 2) resulted in final reconstructions at 3.3 Å resolution (Figure S1D). The resolution of each body was estimated by the FSC curve using the 0.143 criterion. The intact dimeric WDR11-FAM91A1 complex density map was generated by combining maps of two separated bodies using the “vop maximum” command in Chimera. All final density maps were subjected to DeepEMhancer for presentation purpose.

For the WDR11-dm-FAM91A1 complex data, 793 movie stacks were imported in cryoSPARC 4.31 and subjected to Patch Motion Correction and CTFFIND4.⁹⁰ Micrographs were subjected to Blob Picker and 191,720 particles were extracted. After 2D classification, ab initio and heterogeneous refinement (Figure S3A), 35,118 particles of WDR11-dm-FAM91A1 complex were subjected to non-uniform (NU) refinement. Subsequent Reference Based Motion Correction followed by local refinement yielded the final reconstruction at 7.4 Å resolution (Figure S3B). Statistics for data collection and model refinement are summarized in Table S1.

Model building and refinement

The initial WDR11 and FAM91A1 models were generated by the AlphaFold2.⁵³ All models were docked into the cryo-EM density map prior to DeepEMhancer processing in UCSF Chimera1.14,⁷⁸ followed by iterative manual adjustment and real-space refinement using Coot 0.8.9.2 and Phenix 1.19,^79,80 respectively. The model quality was validated by FSC curves of the final models versus the final maps, the unfiltered half map 1 (FSC_work) and the unfiltered half map 2 (FSC_test) (Figure S1B–D).⁹¹ Statistics of the final models are provided in Table S1. Figures were prepared with UCSF ChimeraX1.110 and PyMOL.^83,92 The data validation statistics are shown in Table S1.

Protein pull-down assays

Protein pull-down experiments were performed in a manner similar to previous studies.²² Briefly, 20 μg of His-FAM91A1-WDR11 complex protein, 400 μg of bait protein (GST-CI-MPR WT, GST-CI-MPR AC1, GST-CI-MPR AC2, or GST, purified from BL21 cells), and 30 μL of Ni-NTA beads were mixed in 0.5 mL of pull-down buffer (20 mM Tris (pH 8.0), 500 mM NaCl, 0.005% Triton-X100, 20 mM Imidazole) and incubated at 4°C for 1 hour. The beads were subsequently washed three times with pull-down buffer containing an additional 20 mM Imidazole and the bound proteins were separated by 12% SDS-PAGE.

Bio-layer interferometry

Bio-layer interferometry binding data were acquired using an Octet RED96 instrument (ForteBio) and analyzed using the integrated software. To determine the affinity of peptides for the WDR11-FAM91A1 complex, the His-WDR11-FAM91A1 complex was immobilized onto Ni-NTA-coated biosensors (NTA ForteBio) at a concentration of 50–100 nM in binding buffer (100 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% Tween 20) for 60 seconds. The sensors were washed with the binding buffer. Analyte proteins were diluted from concentrated stocks into the binding buffer. Baseline measurements were taken in the binding buffer alone, followed by monitoring of binding kinetics by exposing the biosensors to target protein at the indicated concentration (association step) and subsequently returning them to the binding buffer (dissociation step). The software DataAnalysisHT12 was used to analysis the BLI data.

AlphaFold2 modeling

The WDR11-CIMPR interaction model was generated by using AlphaFold2 on the Google Colab platform. The sequences of WDR11 1–830 and CIMPR 2402–2414 were input into the query sequence module. The interaction interfaces of the five models are highly consistent. Finally, we used PyMOL generate the interface figure (see figure 5G).

Immunofluorescence staining, and confocal microscopy

In the immunofluorescence experiment, cells were pre-seeded on the bottom of 24-well or 12-well plates and transfected using a transfection reagent (Yeasen, 40802ES03) according to the manufacturer’s instructions. After fixation with 4% formaldehyde in PBS, cells were permeabilized with 0.1% Triton X-100 in PBS at room temperature for 15 minutes. Then, cells were washed three times with PBS and blocked with 5% BSA at room temperature for 1 hour. Primary antibodies were diluted in 1% BSA in PBS and incubated with cells at 4°C overnight. After washing with PBS three times, appropriate fluorescently labeled secondary antibodies were incubated with cells in 1% BSA diluted PBS at room temperature for 1 hour. After washing with PBS, cell nuclei were stained with DAPI (4’,6-diamidino-2-phenylindole). Using an Olympus FV-3000 confocal microscope with a 100× oil objective (NA=1.45), confocal images were captured with X, and Y scans at 8 μs/pixel and a resolution of 1024×1024 in sequential mode (line). Consistent parameters were maintained for image acquisition and analysis across a specific experimental series. Colocalization analyses employed the JACoP plug-in in ImageJ, calculating Pearson’s coefficients. Figure legends specify the number of cells analyzed for each condition. All experiments were repeated at least three times to ensure reproducibility.

Co-immunoprecipitation

HEK293T cells, post-PBS wash, were lysed in NETN buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl, pH 8.0, 0.5% NP-40) supplemented with protease and phosphatase inhibitors. The lysis process involved ultrasonication at 15% power, with ultrasound pulses of 3 seconds followed by pauses of 6 seconds, totaling 1 minute. After centrifugation at 12,000 g for 20 minutes at 4 °C, 5% of the supernatants were set aside as lysate, and the remaining supernatants were employed for immunoprecipitation. Immunoprecipitation, conducted for 1 hour at 4 °C, utilized either anti-mCherry or Anti-GFP Nanobody Magarose Beads. The resulting immunocomplexes underwent three washes with NETN buffer and were subsequently boiled in 2× sample buffer. Separation using SDS-PAGE and analysis by immunoblotting with specified antibodies ensued.

Immunoblotting

Harvested cells were treated with 1× SDS sample buffer and boiled at 98 °C for 10 minutes. Subsequent protein separation occurred through SDS-PAGE, with transfer onto a polyvinylidene fluoride (PVDF) membrane (Millipore). After blocking with 5% (w/v) milk in TBST (0.1% (v/v) Tween-20 in TBS) for 1 hour, membranes underwent overnight probing with specific primary antibodies at 4 °C. Exposure to HRP-conjugated secondary antibodies followed. Visualization of the bands was achieved through Chemiluminescence (Millipore).

Generation of knockout cell lines.

The knockout cells were generated based on previous studies.¹⁸ The target guide RNA sequence was inserted into the hspcas9 plasmid. Subsequently, HEK293T cells were seeded in 10 cm culture dishes and cultured for 24 hours before being transfected with the aid of a lipid transfection reagent (Yeasen, 40802ES03) along with the helper plasmids (Prsv/Prre and PMD2.G). After 48 hours, the synthesized viruses were collected and used to infect HEK293T and HeLa cells. Infected cells were cultured in complete medium for 48 hours and then subjected to two generations of selection in medium containing 4 μg/ml puromycin. The knockout efficiency was determined by immunoblotting.

Generation of doxycycline-induced WDR11 knockdown cell lines.

Doxycycline-inducible depletion of WDR11 was performed similarly to previous studies.⁹³ A doxycycline-responsive element enhancer and a miR-30a-based siRNA targeting WDR11 (AGCAGTCGTATTCAGAGATAAA) were integrated into the pTSB-Tight-mcherry-EF1-tetR-F2A-Puro vector plasmid, resulting a Tet-On siWDR11 plasmid. HEK293T cells were seeded in 10 cm culture dishes and allowed to grow for 24 hours. Subsequently, HEK293T cells were transfected with the Tet-On siWDR11 plasmid together with helper plasmids (Prsv/Prre and PMD2.G) using a lipid transfection reagent (Yeasen, 40802ES03). After 48 hours, the synthesized viruses were collected and utilized to infect HCMEC-D3 cells. Infected cells were cultured in complete medium for 48 hours and underwent selection in medium containing 4 μg/ml puromycin. The addition of 1 μg/mL doxycycline was used to deplete WDR11, and cells without doxycycline treatment were used as control.

Proximity labeling and mass spectrometry

HEK293T cells were transfected with plasmids encoding mCherry-miniTurbo, mCherry-WDR11-WT-miniTurbo, or mCherry-WDR11–2KD-miniTurbo. Post trans-fection, cells were treated with 100 μM biotin for 1 hour at 37°C. After treatment, cells were washed twice with pre-chilled PBS and lysed in a lysis buffer comprising 8 M urea, 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 1% protease inhibitor. During lysis, ultrasonication at 20% power (3-second pulses with 6-second pauses, totaling 1 minute) was applied. The lysate underwent a cooling and ultrasonication cycle three times. The lysate was then centrifuged at 15,000g for 10 minutes. For biotin pull-down, supernatants were incubated with streptavidin agarose beads (Beyotime, C0222, 200 μL) at 4°C with rotation for 3 hours. Subsequently, beads were washed with lysis buffer thrice, TBS twice, and 50 mM ammonium bicarbonate (pH 8) once. Five percent of the sample was boiled for 10 minutes at 100°C in SDS loading buffer and analyzed by immunoblotting.

The tryptic peptides were dissolved in solvent A(0.1% formic acid, 2% acetonitrile/water) and loaded directly onto a homemade reversed-phase analytical column (25 cm length, 100 μm i.d.). The mobile phase consisted of solvent A and solvent B (0.1% formic acid in acetonitrile). Peptide separation occurred through the following gradient: 0–8 minutes, with solvent B concentration increasing from 9% to 24%; 8–12 minutes, from 24% to 35%; 12–16 minutes, from 35% to 80%; and 16–20 minutes, maintaining 80% solvent B, all at a constant flow rate of 450 nl/min on a NanoElute UHPLC system (Bruker Daltonics). Subsequently, peptides underwent mass spectrometry analysis on a timsTOF Pro instrument after passing through a capillary source. An electrospray voltage of 1.75 kV was applied, and precursor and fragment ions were analyzed on the TOF detector. The timsTOF Pro operated in data-independent parallel accumulation serial fragmentation (dia-PASEF) mode. The full MS scan range was set to 100–1700 m/z (MS/MS scan range), with acquisition of 10 PASEF-MS/MS scans per cycle. The MS/MS scan range was set to 400–1200 m/z, with an isolation window of 25 m/z.

DIA-NN search engine (v.1.8) were used to process the DIA data. Tandem mass spectra were searched against the Homo_sapiens_9606_SP_20230103.fasta database containing 20,389 entries, concatenated with a reverse decoy database. Trypsin/P was specified as the cleavage enzyme, allowing for a maximum of 1 missed cleavage site. Fixed modifications included N-terminal methionine excision and carbamidomethylation of cysteine residues. The false discovery rate (FDR) was adjusted to be less than 1%, ensuring stringent criteria for result reliability.

Synthesis of mRNAs

Human WDR11 wild-type and mutant encoding sequences, and zebrafish KIAA0319L encoding sequences were cloned into pcDNA3.1 vector and linearized as templates. T7 mMESSAGE mMACHINE kit (Invitrogen) was used to synthesize Capped mRNA using the templates above. mRNAs were recovered using the lithium chloride precipitation method and kept in nuclease-free buffers for further injection experiments.

Morpholino and mRNA injections

The wdr11 MO (5’-GATGGTGATCTTTTTCTTACCCTGA-3’) was designed against the exon 3–intron 3 splice site as previously described,³⁴ and purchased from Gene Tools. Control MO (5’-CCTCTTACCTCAGTTACAATTTATA-3’) is a standard, mismatched control. MO and mRNAs were injected into the yolk and cell and were performed at the one-cell stage. The injection dose for wdr11 MO and control MO was 2.5 ng per embryo, with mRNA injection doses of 50–100 pg per embryo.

Zebrafish fluorescence imaging and analysis

Zebrafish embryos were pre-anesthetized with 0.04% MS-222 and positioned in 0.8% methylcellulose solution for lateral or dorsal imaging. ZEISS AXIO Zoom was used to capture fluorescent images of CaP axons and neural Elavl3 expression. ZEN 3.1 software was used to quantify the length of CaP axons, measuring 3 site-specific axons per embryo from the root of the cell body to the bottom of the CaP axon in Tg[Hb9:GFP]^ml2 transgenic zebrafish. The average of 3 measurements per embryo was used for subsequent analysis. Qualification of different types in Tg[HuC:GFP] transgenic zebrafish was performed as previously described.²⁸ In Tg[HuC:GFP] transgenic zebrafish embryos, different phenotypes were distinguished by observing the fluorescence pattern in the midbrain region from the lateral view. C1 embryos: the midbrain region was almost completely covered by fluorescence; C3 embryos: no visible fluorescence within the midbrain region; C2 embryos: intermediate phenotype between C1 and C3, and the midbrain region was partially covered by fluorescence.

QUANTITATION AND STATISTICAL ANALYSIS

Statistical analyses were executed using GraphPad Prism 8 Software. The sample size was not determined using any specific statistical method. While assuming a normal data distribution, formal testing for normality was not conducted. Zebrafish experiments were subjected to statistical analysis employing one-way ANOVA and Dunnett’s multiple comparisons test. In cellular experiments, the significance of differences between the two groups was assessed using an unpaired two-tailed Student’s t-test. For experiments involving multiple groups, either ordinary one-way or two-way ANOVA (in the presence of two variables) was utilized, followed by Dunnett’s and Sidak’s tests, respectively, as detailed in the figure legends. A significance level of P < 0.05 was considered. All data were presented as mean ± SD.

Supplementary Material

1

Table S1. Cryo-EM map and model statistics, Related to Figure 1 and Figure 2.

2

Figure S1. Cryo-EM data processing of the WDR11-FAM91A1 complexes, Related to Figure 1 and 2. (A) The cryo-EM data processing workflow.

(B-D) The cryo-EM maps colored by local resolution, FSC curves indicating the overall resolutions and model quality, and angular distributions of (B) the monomeric WDR11-FAM91A1 complex, (C) the partial dimeric complex, and (D) the intact dimeric WDR11-FAM91A1 complex.

(E) Cryo-EM maps and models of the monomeric and dimeric complexes containing WDR11 (green), FAM91A1 (brown), FAM91A1’ (red) and WDR11′ (blue).

Figure S2. Representative regions of the cryo-EM model in density map of the intact dimeric WDR11-FAM91A1 complex, Related to Figure 1–3. (A-C) The cryo-EM model in density map that showed explicitly resolved side chain densities in (A) dimerization interface between WDR11 (green) and WDR11′ (blue) of the intact dimeric complex, (B) WD40 β-sheets and (C) WDR11-FAM91A1 interface of WDR11 (bremen blue) and FAM91A1 (orange) in the monomeric complex and WDR11 (green) and FAM91A1 (brown) in the dimeric complex

Figure S3. Cryo-EM data processing of the WDR11-dm-FAM91A1 complex, Related to Figure 2. (A) Cryo-EM data processing workflow. Representative cryo-EM micrograph of WDR11-dm-FAM91A1 complex and 2D class averages.

(B) The cryo-EM maps colored by local resolution, FSC curves indicating the overall resolutions and angular distributions of WDR11-dm-FAM91A1 complex.

(C) HEK293T cells were transfected with GFP-tagged FAM91A1 and mCherry-tagged WDR11 WT or WDR11-dm and then subjected to GFP-nanotrap for immunoprecipitation. FAM91A1 was detected via antibodies against GFP. WDR11 was detected via antibodies against mCherry. Experiments were performed in triplicate.

(D) The WDR11-dm-FAM91A1 and monomeric WDR11-FAM91A1 complex densities superimpose with the intact dimeric WDR11-FAM91A1 complex density (transparent outline). WDR11 and WDR11-dm are shown in green and FAM91A1 is shown in brown.

Figure S4. Interaction between WDR11 and different cargoes, Related to Figure 4. (A) Binding analysis between different CI-MPR constructs and WDR11. HEK293T cells were transfected with mCherry-tagged WDR11 WT and Venus-tagged CI-MPR WT, AC1, AC2, or 8A mutants, and then subjected to mCherry-nanotrap for immunoprecipitation. CI-MPR and WDR11 were detected via antibodies against GFP (Venus) and mCherry. Experiments were performed in triplicate.

(B) Binding analysis between different cargoes and WDR11. HEK293T cells were transfected with mCherry-tagged WDR11 WT and Venus-tagged KIAA0319L, Furin, CPD, CI-MPR and then subjected to mCherry-nanotrap for immunoprecipitation. cargoes and WDR11 were detected via antibodies against GFP (Venus) and mCherry, respectively. Experiments were performed in triplicate.

Figure S5. Proximity proteomics identify multiple SAC-containing proteins that interact with the positively charged groove of WDR11, Related to Figure 5. (A) Schematic illustrating the process of proximity labeling mass spectrometry.

(B) Western blot assays were conducted on samples subjected to mass spectrometry analysis from three independent experiments. WT, mCherry-WDR11-WT-Turbo; 2KD, mCherry-WDR11–2KD-Turbo. Experiments were performed in triplicate.

(C) Western blot assays were conducted to examine proximity-labeled KIAA0319L and CI-MPR in vector, WDR11-WT, and WDR11–2KD samples. Vector, mCherry-Turbo; WT, mCherry-WDR11-WT-miniTurbo; 2KD, mCherry-WDR11–2KD-miniTurbo. Experiments were performed in triplicate.

(D) The SAC motif of CI-MPR is critical for contacting AP-1. HEK293T cells were transfected with Venus-tagged CI-MPR WT, 8A, L^A, or VL^AA mutants, and then subjected to GFP-nanotrap for immunoprecipitation. CI-MPR, WDR11, and AP-1 γ were detected via antibodies against GFP (Venus), WDR11 and AP-1 γ, respectively. Experiments were performed in triplicate.

(E). AP-1 μ1 inhibits the between WDR11 and CI-MPR. HEK293T cells were co-transfected with mCherry-tagged WDR11 and Venus-tagged CI-MPR, and then lysed. The lysates were then mixed with or without 1 μg of purified His₆-tagged AP-1 μ1 (residues 158–423). After incubation, the lysates were subjected to GFP-nanotrap for immunoprecipitation, and the presence of CI-MPR, WDR11, and AP-1 μ1 was detected using antibodies against GFP (Venus), WDR11, and His, respectively. Experiments were performed in triplicate.

(F) An Alphafold2 model showing how CI-MPR-AC1 may interact with WDR11. The structural model is depicted in a color scheme corresponding to the pLDDT scores. The portion encircled in red represents the peptide segment of CI-MPR-AC1.

(G) A model showing how WDR11 complex cooperates with other tethering factors to mediate endosome-to-TGN trafficking.

(H) Depletion efficiency of TBC1D23, golgin-97, golgin-245, and Arl1 in HEK293T cells. Actin was used as a loading control. Experiments were performed in triplicate.

(I) CI-MPR and WDR11 binding was assessed in HEK293T cells with control and individual knockouts of TBC1D23, golgin-97, golgin-245, or Arl1. These cells were transfected with mCherry-WDR11 and Venus-CI-MPR, and subsequently subjected to immunoprecipitation using mCherry-nanotrap. Detection of CI-MPR and WDR11 was carried out using antibodies targeting GFP (Venus) and mCherry, respectively. Experiments were performed in triplicate.

(J) The relative abundance of CI-MPR compared to WDR11 was quantified in (I) by densitometry using Image J software. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

Figure S6. The assembly of the WDR11-FAM91A1 complex and its interaction with cargoes are essential for the trafficking of multiple SAC-containing proteins, Related to Figure 6. (A) Depletion efficiency of WDR11 in HeLa cells. GAPDH was used as a loading control. Experiments were performed in triplicate.

(B) Subcellular location of KIAA0319L in HeLa cells. The WDR11 KO cells were transfected with mCherry, mCherry-tagged WDR11 WT, dm, KE^AA, or 2KD, respectively. Cells were then incubated with antibodies against KIAA0319L (green) and golgin-97 (purple). The golgin-97 is a marker for the TGN region. Scale bar: 10 μm. Experiments were performed in triplicate.

(C) Colocalization analysis between KIAA0319L and golgin-97 in (B). Each dot represents Pearson’s correlation coefficients from one cell. Con: n=69 cells; mCherry: n=38 cells; WT: n=49 cells; dm n=49 cells; KE^AA n=67 cells; 2KD: n=58 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(D) Depletion efficiency of AP-1 μ1 and WDR11 in HeLa cells. GAPDH was used as a loading control. AP-1 γ antibody was more specific than AP-1 μ1 antibody, and was used to determine KD efficiency. Experiments were performed in triplicate.

(E) Depletion of AP-1 μ1 or WDR11 diminishes the TGN distribution of ATG9A in HeLa cells. Immunofluorescence analysis was employed to examine the co-localization between endogenous ATG9A (green) and the TGN marker golgin-97 (purple) in control, AP-1 μ1 depletion, and WDR11 depleted cells. Scale bar: 10 μm. Experiments were performed in triplicate.

(F) Colocalization analysis between ATG9A and golgin-97 in (E). Each dot represents Pearson’s correlation coefficients from one cell. Con: n=37 cells; AP-1 μ1 KD: n=47 cells; WDR11 KD: n=58 cells;. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(G) Depletion efficiency of TBC1D23, golgin-97/245 in HeLa cells. Actin was used as a loading control. Experiments were performed in triplicate.

(H) Subcellular localization of CI-MPR-WT and CI-MPR-L^A in control, TBC1D23 knockout, and golgin-97/245 knockout HeLa cells. HeLa cells were transfected with mCherry-GRIP (a marker of the trans-Golgi) and Venus-tagged CI-MPR cytosolic tails WT or L^A. Cells were fixed, permeabilized, and stained for the cell nuclei before observation using fluorescence microscopy. Scale bar: 10 μm. Experiments were performed in triplicate.

(I) Colocalization analysis between CI-MPR and GRIP in (H). Each dot represents Pearson’s correlation coefficients from one cell. n=50 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(J) Subcellular location of VAMP4 in HeLa cells. The HeLa cells were transfected with mCherry-GRIP (a marker of trans-Golgi) and Venus-tagged VAMP4 cytosolic tails WT, 9A, LL^AA, or FF^AA, respectively. Cells were fixed, permeabilized, and stained for the cell nuclei before observation using fluorescence microscopy. LL^AA: AAEDDSDEEEDFF, FF^AA: LLEDDSDEEEDAA, 9A: LLAAAAAAAAAFF. Scale bar: 10 μm. Experiments were performed in triplicate.

(K) Colocalization analysis between VAMP4 and GRIP in (J). Each dot represents Pearson’s correlation coefficients from one cell. Vector: n=60 cells; WT: n=54 cells; 9A: n=44 cells; LL^AA: n=57 cells; FF^AA: n=50 cells. Data are presented as mean ± SD, and P values were calculated using one-way ANOVA and Tukey’s multiple comparisons tests.

(L) Immunoblotting analysis of proximity-labeled VTI1A, STX6, golgin-97 in vector, VAMP4-WT, and VAMP4–9A samples. Vector, mCherry-miniTurbo; VAMP4-WT, mCherry-VAMP4-WT-miniTurbo; VAMP4–9A, mCherry-VAMP4–9A-miniTurbo. Experiments were performed in triplicate.

Figure S7. WDR11 is selectively required for the transport of SAC-containing proteins, Related to Figure 6. (A) Subcellular location of LDLR in HeLa cells. Cells were then incubated with antibodies against LDLR (green) and golgin-97 (red). The golgin-97 is a marker for the TGN region. Scale bar: 10 μm. Experiments were performed in triplicate.

(B) Colocalization analysis between LDLR and golgin-97 in (A). Each dot represents Pearson’s correlation coefficients from one cell. Control: n=59 cells; WDR11 KO: n=50 cells. Data are presented as mean ± SD, and P values were calculated unpaired Student’s t test. Scale bar, 10 μm.

(C) Depletion efficiency of WDR11 and the protein expression level of LDLR in HeLa cells. GAPDH was used as a loading control. Experiments were performed in triplicate.

(D) The relative abundance of LDLR compared to GAPDH was quantified in (C) and compared to the WT group by densitometry using Image J software. Data are presented as mean ± SD, and P values were calculated using unpaired Student’s t tests, n=5.

(E) Depletion efficiency of WDR11 via doxycycline (1 μg/mL) inducing and the protein expression level of STING in HCMEC-D3 cells (possessing the complete cGAS-STING signaling pathway). DOX(doxycycline): induction for functional activation of WDR11-shRNA, diABZI: STING activator. Histone H3 was used as a loading control. Experiments were performed in triplicate.

(F) The relative abundance of STING compared to GAPDH was quantified in (E) and compared to the cells without WDR11 knockdown group by densitometry using Image J software. Data are presented as mean ± SD, and P values were calculated using unpaired two-tailed Student’s t test, n=4.

(G) Summary of distinct types of cargoes transported by AP-1 and WDR11. “√” : AP-1 or WDR11 is involved; “×”: AP-1 or WDR11 is not involved.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
WDR11	Abcam	Cat# ab93871
CI-MPR	Abcam	Cat# ab2733
CI-MPR	Bio-Rad	Cat# MCA2048
FAM91A1	Proteintech	Cat# 27738-1-AP
golgin-97	Proteintech	Cat# 12640-1-AP
ATG9A	Abcam	Cat# ab108338
KIAA0319L	Abcam	Cat# ab105385
LDLR	Proteintech	Cat# 10785-1-AP
STING	Proteintech	Cat# 19851-1-AP
GAPDH	Proteintech	Cat# 10494-1-AP
mCherry tag	Proteintech	Cat# 26765-1-AP
GFP tag	Proteintech	Cat# 66002-1-Ig
His tag	Proteintech	Cat# 66005-1-Ig
GST tag	Proteintech	Cat# 10000-0-AP
STX6	Proteintech	Cat# 10841-1-AP
VTI1A	Proteintech	Cat# 12354-1-AP
TBC1D23	Proteintech	Cat# 17002-1-AP
golgin-245	Cell Signaling Technology	Cat# 79145
Actin	Proteintech	Cat# 66009-1-lg
Streptavidin HRP	Beyotime	Cat# A0305
Histone H3	Proteintech	Cat# 17168-1-AP
AP-1 γ	BD Biosciences	Cat# 610385
Arl1	Proteintech	Cat# 16012-1-AP
Goat anti-Rabbit IgG Secondary Antibody HRP conjugated	SAB	Cat# L3012-2
Goat anti-Mouse IgG Secondary Antibody HRP conjugated	SAB	Cat# L3032-2
Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 546	Thermo Fisher Scientific	Cat# A-11003
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 546	Thermo Fisher Scientific	Cat# A-11010
Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Thermo Fisher Scientific	Cat# A-11001
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Thermo Fisher Scientific	Cat# A-11008
Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Thermo Fisher Scientific	Cat# A-21235
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Thermo Fisher Scientific	Cat# A-21244
Bacterial and virus strains
E. coli DH5α	TSINGKE	Cat# TSC-C01
TransStbl3 Chemically Competent Cell	TRANS	Cat# CD521-01
BL21 (DE3)	TSINGKE	Cat# TSC-E01
DH10Bac	Biomed	Cat# BC112-01
Chemicals, peptides, and recombinant proteins
SIM SF expression medium	Sino Biological	Cat# MSF1
DMEM	Gibco	Cat#: C11995500BT
CI-MPR AC1-WT peptide	GL Biochem	N/A
CI-MPR AC1-LA peptide	GL Biochem	N/A
CI-MPR AC1-VLAA peptide	GL Biochem	N/A
CI-MPR AC1-8A peptide	GL Biochem	N/A
CI-MPR AC1-4D peptide	GL Biochem	N/A
CI-MPR AC1-6D peptide	GL Biochem	N/A
CI-MPR AC1-10D peptide	GL Biochem	N/A
CI-MPR AC1-12D peptide	GL Biochem	N/A
Deposited data
The raw mass spectrometry data of the identification of WDR11 and its mutant	iProX, this paper	IPX0008905000 https://www.iprox.cn/page/project.html?id=IPX0008905000
Map of monomeric WDR11-FAM91A1 complex	EMDB, this paper	EMD-38312
Map of partial dimeric WDR11-FAM91A1 complex	EMDB, this paper	EMD-38300
Map of dimeric WDR11-FAM91A1 complex	EMDB, this paper	EMD-39863
Map of dimeric WDR11-FAM91A1 complex body1	EMDB, this paper	EMD-39947
Map of dimeric WDR11-FAM91A1 complex body2	EMDB, this paper	EMD-39943
Map of WDR11-dm-FAM91A1 complex	EMDB, this paper	EMD-39949
Model of monomeric WDR11-FAM91A1 complex	PDB, this paper	8XG3
Model of partial dimeric WDR11-FAM91A1 complex	PDB, this paper	8XFB
Model of dimeric WDR11-FAM91A1 complex	PDB, this paper	8Z9M
Experimental models: Cell lines
HeLa	ATCC	N/A
HEK293T	ATCC	N/A
Experimental models: Organisms/strains
Tg[HuC:GFP]	Laboratory of Stem Cell Biology	N/A
Tg[Hb9:GFP]ml2	Laboratory of Stem Cell Biology	N/A
Oligonucleotides
gRNA targeting WDR11 sequence: TTATGGATGTCATTCACTTG	This paper	N/A
gRNA targeting TBC1D23 sequence: CTGCCAACGTCGAGCGGCGA	This paper	N/A
gRNA targeting golgin-97 sequence: AGAGGCCAGGAGGTGCTACT	This paper	N/A
gRNA targeting golgin-245 sequence: TTGTGCAAGAGAATTCACAT	This paper	N/A
gRNA targeting arl1 sequence: GAATAGTAACATCTCCAGTA	This paper	N/A
siRNA targeting WDR11 sequence: AGCAGTCGTATTCAGAGATAAA	This paper	N/A
siRNA targeting AP1G1 sequence: GCAGGAAGTTATGTTCGTGAT	This paper	N/A
Morpholino: MO-wdr11 GATGGTGATCTTTTTCTTACCCTGA	This paper	N/A
Recombinant DNA
pFastBacDual His8- FAM91A1, WDR11	This paper	N/A
pFastBacDual His8- FAM91A1, WDR11-dm	This paper	N/A
pGEX-4T-1 GST- CI-MPR C-tail (2398–2491)	This paper	N/A
pGEX-4T-1 GST- CI-MPR C-tail AC1 (2398–2491, Δ 2405–2412)	This paper	N/A
pGEX-4T-1 GST- CI-MPR C-tail AC2 (2398–2491, Δ 2482–2487)	This paper	N/A
pET28a His-AP-1 μ1 158–423	This paper	N/A
pcDNA3.1+ GLP- Venus-CI-MPR C-tail (2398–2491)	This paper	N/A
pcDNA3.1+ GLP- Venus-KIAA0319L C-tail (954–1049)	This paper	N/A
pcDNA3.1+ GLP- Venus-furin C-tail (739–794)	This paper	N/A
pcDNA3.1+ GLP- Venus-CPD C-tail (1321–1380)	This paper	N/A
pcDNA3.1+ GLP- Venus-VAMP4 N-tail (1–115)	This paper	N/A
pcDNA3.1+ GLP- Venus-VAMP4 N-tail (1–115) 9A	This paper	N/A
pcDNA3.1+ GLP- Venus- ATG9A C-tail T (493–839)	This paper	N/A
pcDNA3.1+ GLP- Venus- TMEM87B C-tail T (451–555)	This paper	N/A
pcDNA3.1+ GLP- Venus- SV2A N-tail (1–169)	This paper	N/A
pcDNA3.1+ GLP- Venus- MERTK C-tail (527–999)	This paper	N/A
pcDNA3.1+ GLP- Venus- VAMP7 C-tail (2–188)	This paper	N/A
pcDNA3.1+ GLP- Venus- CI-MPR C-tail AC2	This paper	N/A
pcDNA3.1+ GLP- Venus- CI-MPR C-tail LA	This paper	N/A
pcDNA3.1+ GLP- Venus - CI-MPR C-tail VLAA	This paper	N/A
pcDNA3.1+ GLP- Venus - CI-MPR C-tail 4D	This paper	N/A
pcDNA3.1+ GLP- Venus - CI-MPR C-tail 12D	This paper	N/A
pcDNA3.1+ GLP- Venus - CI-MPR C-tail 8A	This paper	N/A
pmCherry-C1 WDR11	This paper	N/A
pmCherry-C1 WDR11-KEAA	This paper	N/A
pmCherry-C1 WDR11-K9D	This paper	N/A
pmCherry-C1 WDR11-2KD(K479D+K482D)	This paper	N/A
pmCherry-C1 WDR11-K719D	This paper	N/A
pmCherry-C1 WDR11-K476D	This paper	N/A
pmCherry-C1 WDR11-K808D	This paper	N/A
pmCherry-C1 WDR11-miniTurbo	This paper	N/A
pmCherry-C1 WDR11-2KD-miniTurbo	This paper	N/A
pmCherry-C1 miniTurbo-VAMP4	This paper	N/A
pmCherry-C1 miniTurbo-VAMP4 9A	This paper	N/A
pmCherry-C1 FAM91A1	This paper	N/A
pEGFP-C1 WDR11	This paper	N/A
pEGFP-C1 WDR11-dm	This paper	N/A
pEGFP-C1 WDR11-ΔSITAQ	This paper	N/A
pEGFP-C1 WDR11-KLLAAA	This paper	N/A
pEGFP-C1 WDR11-KEAA	This paper	N/A
pEGFP-C1 FAM91A1	This paper	N/A
pEGFP-C1 FAM91A1-RRKEEE	This paper	N/A
pEGFP-C1 FAM91A1-Δα	This paper	N/A
pcDNA3.1+ WDR11	This paper	N/A
pcDNA3.1+ WDR11-dm	This paper	N/A
pcDNA3.1+ WDR11-KEAA	This paper	N/A
Software and algorithms
Graph pad 8.3.0	GraphPad	https://www.graphpad.com
ImageJ	National Institutes of Health	https://imagej.nih.gov/ij/download.html
Octet data analysis high throughput version 12	ForteBio	https://www.sartorius.com
AlphaFold2	Jumper et al.75	https://alphafold.ebi.ac.uk/
DIA-NN 1.8.1	DIA-NN	https://github.com/vdemichev/DiaNN
EPU 2.9.0	Thermo Fisher	https://www.thermofisher.cn/cn/zh/home/electron-microscopy/products/software-em-3d-vis/epusoftware.html
Relion 5.0	Scheres et.al.76	https://github.com/3dem/relion/tree/ver5.0
cryoSPARC 4.31	Punjani et.al.77	https://cryosparc.com/
UCSF Chimera 1.16	Pettersen et.al.78	http://www.cgl.ucsf.edu/chimera/
Phenix 1.19	Adams et.al.79	https://phenix-online.org/
Coot 0.9.8	Emsley et.al.80	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
UCSF ChimeraX 1.1	Pettersen et.al.81	https://www.cgl.ucsf.edu/chimerax/
DeepEMhancer	Sanchez-Garcia et al.82	https://github.com/rsanchezgarc/deepEMhancer
Pymol	Schrödinger, USA83	https://pymol.org/2/
Other
Superose6 Increase10/300 GL column	Cytiva	Cat# 29091596
Streptavidin Agarose	Beyotime	Cat# P2159-5ml
Anti-mCherry Nanobody Magarose Beads	Shenzhen KT Life technology	Cat# KTSM1337
Anti-GFP Nanobody Magarose Beads	Shenzhen KT Life technology	Cat# KTSM1334
Ni-NTA Beads	Smart-Lifesciences	Cat# SA004500
Glutathione Beads	Smart-Lifesciences	Cat# SA008500
200-mesh Cu R2/1	Quantifoil	Cat# Q250CR1

Highlights.

• Cryo-EM structure of human WDR11-FAM91A1 complex

• WDR11 directly recognizes the SAC motif within the cytoplasmic tails of cargoes

• The SAC motif is indispensable for protein intracellular trafficking

• WDR11 complex assembly and interaction with SAC are critical for neuronal development