Domain-specific N-glycosylation of the adhesion G-protein–coupled receptor ADGRG6 N-terminal fragment regulates trafficking, proteolytic processing, and signaling

Abstract

Adhesion G-protein–coupled receptors (aGPCRs) are characterized by long extracellular N-terminus fragments (NTF) with several adhesive domains. Many aGPCRs are cleaved at the GPCR-autoproteolysis site (GPS), enclosed within the larger GPCR-autoproteolysis-inducing (GAIN) domain. Following cleavage at the endoplasmic reticulum (ER), the NTF and C-terminal fragments (CTF) heterodimerize, and the protomer is trafficked to the plasma membrane. ADGRG6 is an aGPCR playing important roles in embryonic development. It is activated by mechanical perturbation of the NTF and several extracellular matrix (ECM) proteins. GPCRs are known to undergo several posttranslational modifications (PTM) that regulate the molecular pharmacology of these receptors. N-glycosylation is an important PTM that regulates GPCR expression, trafficking, ligand binding, and signaling bias. Although ADGRG6 is N-glycosylated, the location of the glycans remains unknown. Furthermore, are there spatial roles of N-glycosylation in ADGRG6 processing, trafficking, and signaling? To address these gaps in knowledge, we used biochemical and cell-biological approaches using cell lines overexpressing wild-type and N-glycosylation mutants of ADGRG6. We demonstrate that specific N-glycan residues in different domains of the NTF of ADGRG6 have distinct roles in ADGRG6 autoproteolysis, furin cleavage, trafficking to the plasma membrane, and cAMP production.

• Where is ADGRG6 N-glycosylated, and how does this PTM regulate the functions of this aGPCR?

• ADGRG6 is N-glycosylated in all five domains of the NTF. N-glycosylation in the CUB, SEA, and HrmR regulates basal cAMP signaling. N-glycosylation in the PTX domain is necessary for trafficking from endoplasmic reticulum and furin cleavage. N-glycosylation in the GAIN domain is required for trafficking, GPS autoproteolysis, and furin cleavage.

• These results provide novel insights into the mechanisms of ADGRG6 maturation and signaling in a cell culture model. Future studies using animal models can reveal the roles of ADGRG6 N-glycosylation in organogenesis.

INTRODUCTION

The adhesion G-protein–coupled receptor (aGPCR) subfamily, comprising 33 receptors, is the second most abundant category of GPCRs based on the GRAFS classification system (Schiöth and Fredriksson, 2005). Posttranslational cleavage at the GPCR-autoproteolysis site (GPS) generates the extracellular N-terminus fragments (NTF) and the membrane-bound C-terminal fragments (CTF), which together form the heteromeric receptor complex at the membrane (Guan et al., 1996; Krasnoperov et al., 2002; Lin et al., 2004; Volynski et al., 2004; Araç et al., 2012; Yeung et al., 2020). Isomerization of NTF on ligand binding or force-dependent detachment of NTF can expose a tethered agonist in the CTF, called the Stachel sequence, which binds to the orthosteric site in the receptor to initiate G-protein signaling (Liebscher et al., 2015; Stoveken et al., 2015; Yeung et al., 2020). The NTFs of aGPCRs project into the extracellular matrix (ECM) or blood plasma and often associate with ECM and other membrane-bound proteins in adjoining cells in trans, which modulate G-protein activation (Dunn et al., 2019).

ADGRG6, a member of the aGPCR clade, has an NTF with 5 domains: complement C1r/C1s, Uegf, Bmp1 (CUB), pentraxin (PTX), sperm protein, enterokinase, and agrin (SEA), hormone binding (HormR), and the GPCR-autoproteolysis-inducing (GAIN) domain (Leon et al., 2020). In addition to autoproteolysis at GPS, ADGRG6 is also cleaved by furin in the SEA domain (Moriguchi et al., 2004). Studies in model organisms have shown that ADGRG6 is important for myelination of peripheral neurons, angiogenesis, placentation, heart, and inner ear development (Waller-Evans et al., 2010; Geng et al., 2013; Patra et al., 2013; Cui et al., 2014; Liebscher et al., 2015; Torregrosa-Carrión et al., 2021). Genetic variations and mutations in ADGRG6 result in human disorders like scoliosis, pectus excavatum, arthrogryposis multiplex congenita, chronic obstructive pulmonary disorder, and cancers (Kou et al., 2013; Ravenscroft et al., 2015; Wu et al., 2019; An et al., 2022; Werder et al., 2023), making ADGRG6 an attractive drug target.

Intracellular signaling by ADGRG6 can be regulated by multiple mechanisms. Crystal structure and molecular dynamics simulation of zebrafish adgrg6 NTF have identified that the extracellular region of the receptor can adopt multiple conformations, and an open configuration bound to Ca²⁺ favors enhanced basal cAMP signaling (Leon et al., 2020; Dintzner et al., 2024). cAMP signaling by ADGRG6 is also regulated by mechanical stimulus and a wide variety of ligands, ranging from prion protein to 17-hydroxy progesterone (Küffer et al., 2016; An et al., 2022; Mitgau et al., 2022). Moreover, ECM proteins like collagen-IV, collagen-VI, and laminin-211 are known to modulate G-protein signaling by ADGRG6 by binding to adhesive domains in the NTF to introduce conformational changes to the extracellular region of the receptor (Paavola et al., 2014; Petersen et al., 2015; Wilde et al., 2023). This exposes the Stachel sequence at the extreme N-terminal end of the CTF, which interacts with the heptahelical bundle to regulate cAMP signaling (Liebscher et al., 2015). Thus, trafficking of ADGRG6 to the membrane is a prerequisite for mechanical stimulation or interaction with ECM components. However, the molecular determinants for ADGRG6 membrane trafficking remain unknown. Glycosylation at N residues is known to regulate ligand binding, trafficking, and signaling by several membrane proteins, including GPCRs (Patwardhan et al., 2021). This underscores the importance of N-glycosylation in regulating multiple aspects of GPCR functioning.

The roles of N-glycosylation in regulating the molecular pharmacology of ADGRD1, ADGRE2, and ADGRE5 have been studied (Hsiao et al., 2009; Huang et al., 2018; Frenster et al., 2021). Although prior work by Stehlik et al., has demonstrated that ADGRG6 expressed in lipopolysaccharide (LPS)-stimulated human umbilical vein endothelial cells (HUVEC) is N-glycosylated, it is unclear which domains of ADGRG6 are N-glycosylated and how this might affect trafficking and signaling of the receptor (Stehlik et al., 2004). We demonstrate that ADGRG6 undergoes N-glycosylation only at the NTF and not the CTF. Furthermore, we show that mutating certain N-glycosylation sites in the PTX and GAIN domains suppress furin and GPS cleavage, respectively. These mutations also lead to ADGRG6 accumulation in the endoplasmic reticulum (ER) as opposed to being enriched at the plasma membrane. We also observed that mutating N-glycosylation sites in CUB, PTX, SEA, HrmR or GAIN domains significantly reduce basal cAMP production by ADGRG6, while Stachel peptide-mediated cAMP generation is compromised by a lack of glycosylation at the PTX and GAIN domains only. Through this study, we identify that there are domain-specific roles of N-glycosylation in ADGRG6 posttranslational processing, membrane targeting, and cAMP generation.

RESULTS

NTF of ADGRG6 undergoes N-glycosylation with higher mannose modifications

To determine ADGRG6 N-glycosylation sites, we overexpressed codon-optimized human ADGRG6 with a FLAG epitope at the N-terminal in HEK293T cells (Figure 1A). Immunoblots of FLAG-ADGRG6 immunoprecipitated from HEK293T showed three bands corresponding to the full-length, NTF and a furin-cleaved fragment, allowing us to detect N-glycosylation sites in different regions of the NTF of the receptor by deglycosylating the protein (Figure 1A, lane 2). Upon treating immunoprecipitated ADGRG6 with PNGaseF, an enzyme that removes N-glycans from glycoproteins, we observed that the full-length, NTF, and furin-cleaved bands underwent a drop in molecular weight. This indicates that NTF and the stretch of amino acids upstream of the furin-cleavage site in the SEA domain are indeed glycosylated (Figure 1A, lane 4). Because we observed a drop in the molecular weight of the full-length glycoprotein, it is possible that the CTF is also N-glycosylated. We also subjected the immunoprecipitated FLAG-ADGRG6 glycoprotein to EndoH treatment, another glycosidase that removes higher mannose residues from N-glycan cores. With EndoH deglycosylation of FLAG-ADGRG6, we observed a drop in the molecular weight of the full-length and NTF bands, like in the PNGase-treated conditions. However, no shift in the furin-cleaved fragment was observed (Figure 1A, lane 5). Thus, we concluded that N-glycan cores added to the N residues upstream of the furin-cleavage site, a region spanning a fraction of the SEA, PTX, and CUB domains, are converted into EndoH-resistant varieties. We also treated immunoprecipitated FLAG-ADGRG6 with sialidase (neuraminidase) to determine whether α2,3-, α2,6-, or α2,8-linked sialic acid modifications are present in the N-glycan cores of the receptor. As we did not observe any alteration in the mobility of the full-length, NTF and furin-cleaved bands of ADGRG6, we concluded that either sialic acid modifications are absent from the N-glycans or there might be few such groups whose removal does significantly change the molecular weight of the receptor (Figure 1B, lane 4).

FIGURE 1:

ADGRG6 undergoes N-glycosylation at the NTF. (A) Schematic representation of the extracellular domains in the NTF of human ADGRG6 tagged with FLAG epitope used for immunoprecipitation assays in HEK293T. ADGRG6 immunoprecipitated with anti-FLAG M2 beads was treated with PNGaseF (lane 4) or EndoH (lane 5) and then subjected to Western blotting with anti-FLAG (polyclonal) and anti-actin antibodies. (B) FLAG-ADGRG6 was immunoprecipitated from HEK293T cells and subjected to digestion with sialidase (lane 4) or PNGase (lane 5) and then immunoblotted with rabbit anti-FLAG and anti-actin antibodies. (C) Cartoon of ADGRG6 tagged with HA and FLAG epitopes at the NTF and CTF, respectively, used for overexpression in COS7 cells. ADGRG6 immunoprecipitated from COS7 cells with anti-HA magnetic beads, deglycosylated with PNGaseF or EndoH and subjected to immunoblotting with anti-HA and FLAG antibodies. Orange and red arrows point to full-length and NTF bands, respectively. Red arrowhead points to furin-cleaved fragment. Green circles point to CTF. §, indicates residual N-glycosylated ADGRG6. All Western blots are representative of three independent experiments.

Next, we explored whether the CTF of ADGRG6 also undergoes N-glycosylation. We performed heterologous overexpression of ADGRG6 tagged with HA and FLAG at the N- and C-terminus ends (Figure 1C) in COS7 cells. Immunoblots of ADGRG6 immunoprecipitated with anti-HA antibodies showed four bands corresponding to the full-length, NTF, furin-cleaved fragment, and CTF of ADGRG6 (Figure 1B). Upon deglycosylation with PNGase and EndoH, we observed shifts in the full-length and NTF bands like the results observed in HEK293T cell experiments. Consistent with the observations in Figure 1A, we did not observe any fall in molecular weight of the furin-cleaved band with EndoH treatment. Moreover, no change in the molecular weight of the CTF was observed upon treatment with either deglycosylating enzyme (Figure 1B). Thus, we concluded that N-glycosylation of ADGRG6 takes place exclusively at the NTF. Furthermore, because we observed similar banding patterns of ADGRG6 in HEK293T and COS7 lysates, we inferred that the N-glycosylation pattern of the receptor bears a close resemblance in two different mammalian cell lines.

ADGRG6 N-glycosylation is necessary for trafficking to the plasma membrane

We then verified whether N-glycosylation of ADGRG6 is important for subcellular trafficking, as previous studies have shown that suppressing N-glycosylation with tunicamycin does not impair cell-surface trafficking of the receptor (Stehlik et al., 2004). We immunostained COS7 cells expressing ADGRG6 with HA and FLAG tags as extracellular and intracellular epitopes, respectively (Figure 2A). In control cells, we observed that nearly 80% of all transfected cells showed HA and FLAG staining at the plasma membrane, indicating that ADGRG6 is trafficked to the membrane. In addition to the plasma membrane, ADGRG6 was also observed in vesicles throughout the cytoplasm (Figure 2, A and B). However, tunicamycin treatment significantly reduced the number of cells with ADGRG6 expression in the plasma membrane and enhanced accumulation of the receptor in perinuclear aggregates (Figure 2, A and B). Upon quantifying surface expression of ADGRG6 using an anti-HA antibody under nonpermeabilizing conditions, we observed a significant drop in the surface levels of ADGRG6 following tunicamycin treatment (Figure 2C). Cell-surface biotinylation assays were performed to biochemically determine the levels of ADGRG6 on the plasma membranes with or without tunicamycin treatment. In lysates from control HEK293T-overexpressing FLAG-ADGRG6, we observed the three bands (Figure 2D, lane 2) we reported earlier in Figure 1A. In lysates from tunicamycin-treated cells, we observed a dramatic drop in the intensity of full-length, NTF, and furin-cleaved bands (Figure 2D, lane 3). We also observed a band above 100 kD, which is similar in size to the deglycosylated full-length band that we observed in the earlier immunoprecipitation experiments reported in Figure 1A. It is likely that a lack of N-glycosylation prevents the cleavage of ADGRG6 at GPS and furin sites. Moreover, tunicamycin introduced several low molecular weight bands of ADGRG6. As tunicamycin blocks N-glycosylation in cells, there is increased protein misfolding (Banerjee et al., 2011; Wang et al., 2015), triggering unfolded protein response (UPR) to remove the misfolded proteins by enzymatic cleavage. Because underglycosylation of the transcriptional activator ATF6 triggers UPR (Hong et al., 2004), we performed immunoblots to detect a lower molecular weight variant of ATF6 upon tunicamycin treatment (Supplemental Figure S1). Thus, tunicamycin initiated UPR to degrade the misfolded ADGRG6 lacking N-glycans sequestered in the ER, leading to the appearance of the lower molecular bands of ADGRG6 (Figure 2D, lane3). Upon analyzing the neutravidin-precipitated ADGRG6 in immunoblots, we observed that there is a drop in the levels of the full-length, NTF, and furin-cleaved fragments of the receptor upon tunicamycin treatment (Figure 2D, lane 7). Low levels of ADGRG6 observed on the plasma membrane under tunicamycin-treated conditions (Figure 2D, lane 7) may have accumulated before the start of the tunicamycin incubation. Moreover, the 100 kD band that appears in whole-cell lysates upon tunicamycin treatment was not pulled down (Figure 2D, lane 7), indicating that it accumulates in an internal compartment of the cell and is not biotinylated. As the results of the surface-biotinylation assays are consistent with the imaging data, we concluded that, contrary to previous reports, it is likely that a lack of N-glycosylation prevents ADGRG6 trafficking to the membrane.

FIGURE 2:

Lack of N-glycosylation of ADGRG6 prevents trafficking of the receptor to plasma membrane. (A) COS7 cells overexpressing HA-ADGRG6-FLAG incubated with tunicamycin for 12 h were immunostained with anti-FLAG and anti-HA antibodies. Images are representative of three independent experiments. Arrows indicate ADGRG6 expression on the membrane; Scale bar, 5 µm. (B) Bar-graph showing the percentage of cells transfected with HA-ADGRG6-FLAG showing distribution of the receptor either on the plasma membrane or in perinuclear aggregates following DMSO or tunicamycin treatment. Data are the mean ± STDEV from three independent experiments. For every experiment at least 30 cells transfected with ADGRG6 were analyzed for subcellular distribution of the receptor. The P values were calculated by two-way ANOVA with Sidak's multiple comparison tests. ****p < 0.0001. (C) COS7 cells expressing empty vector or HA-ADGRG6-FLAG were treated with DMSO or tunicamycin for 12 h, were stained with APC-tagged anti-HA antibodies in nonpermeabilization conditions, and flow cytometry was used to quantify surface fluorescence intensity. Fold change in fluorescence intensity was plotted as mean± SEM from three independent experiments. The P values were calculated by one-way ANOVA with Dunnett's multiple comparisons test. **p < 0.001. (D) HEK293T cells overexpressing FLAG-ADGRG6 were treated for 12 h with tunicamycin, surface biotinylated, lysed and biontinylated surface proteins were pulled-down with neutravidin-agarose. Inputs (lanes 1–4) and pulled down proteins (lanes 5–8) were analyzed by Western blotting with anti-FLAG and anti-actin antibodies. Red and green arrows are indicating full-length and NTF bands, respectively. Green and red arrowheads are pointing to the furin-cleaved fragment and the non-glycosylated full-length ADGRG6 band, respectively. The blot is representative of three independent experiments.

N-glycosylation in the PTX and GAIN domains of ADGRG6 regulates trafficking to the plasma membrane

Though blocking N-glycosylation of ADGRG6 with tunicamycin significantly suppressed trafficking of ADGRG6 from an intracellular compartment to the plasma membrane (Figure 2, A–D), the mislocalization may result from a lack of N-glycosylation in the proteins of the secretory pathway of the cell. To determine whether the intracellular retention of ADGRG6 is due to the lack of N-glycan residues in ADGRG6, we resorted to identifying putative N-glycosylation sites and generating point mutations in these N residues. The crystal structure of zebrafish NTF has revealed 11 N-glycosylation sites (Leon et al., 2020) of which two are in the CUB domain, six in the SEA and HrmR domains, and three in the GAIN domain (Supplemental Figure S2). As N-glycosylation sites in ADGRG6 have not been identified experimentally, we determined putative N-glycosylation sites by in silico methods.

Using NetNglyc-1.0, 18 N residues spanning the NTF of ADGRG6 were predicted to be glycosylated (Blom et al., 2004) (Supplemental Figure S2). Although four putative N-glycosylation sites were predicted and mutated in human ADGRG6 PTX domain, the crystal structure of zebrafish adgrg6 has not identified any N-glycans in the PTX domains. We successfully mutated 15 of the predicted N-glycosylation sites by site-directed mutagenesis (Figure 3A; Supplemental Figure S2) to study the physiological importance of these sites. As indicated in Figure 3A; Supplemental Figure S2, the CUB, PTX, and GAIN mutants have 2, 4, and 6 predicted N residues mutated to alanine, respectively. We also generated mutants with putative N-glycosylation sites encompassing more than a single domain. A SEA+HrmR mutant was generated by mutating three putative N-glycosylation sites encompassing these domains. Mut6 included all six sites in the CUB and PTX domains; Mut9 has nine mutated residues covering the SEA, HrmR, and GAIN domains; and Mut15 is a combination of mutations of all 15 residues in the CUB, PTX, SEA, HrmR, and GAIN domains (Figure 3A; Supplemental Figure S2).

FIGURE 3:

Site-specific point mutations at putative N-glycosylation sites in ADGRG6 NTF lead to mis-localized distribution of the aGPCR. (A) Cartoon representing the putative N-glycosylated asparagine residues in ADGRG6 NTF that have mutated to alanine. (B–H) COS7 cells overexpressing wild-type or N-glycosylation mutants of HA-ADGRG6-FLAG were immunostained with anti-FLAG and anti-calnexin antibodies and subjected to confocal microscopy. For every image line-scan (yellow lines in the merge panels), analysis was performed to determine colocalization of FLAG-tagged receptor and calnexin. Arrowheads in yellow lines represent origin in graphs. Every image is representative of three independent experiments. Scale bar, 10 µm. (I ) Violin plots showing the distribution of Pearson's correlation coefficient between FLAG tagged ADGRG6 and calnexin. The images and graphs are representative of three independent experiments. At least 30 ADGRG6-expressing cells for each construct were analyzed for every experiment. The p values were calculated by one-way ANOVA with Dunnett's multiple comparison test; *p < 0.05; **p < 0.01; ***p < 0.001.

Wild-type and N-glycosylation mutant constructs tagged with HA and FLAG epitopes at the N and C-terminal ends, respectively, were expressed in COS7 cells and subjected to immunofluorescence analysis to determine the subcellular localization of the mutants. Although wild-type ADGRG6 showed robust membrane distribution, some vesicular and intracellular staining was also observed (Figure 3B; Supplemental Figure S4B). Similar to cells expressing the wild-type receptor (Figure 3B), a large majority of the cells transfected with the CUB mutant showed FLAG staining at the membrane (Figure 3C; Supplemental Figure S4B). However, the CUB mutant receptor also showed some colocalization with calnexin, indicating that there may be some mislocalization of the protein (Figure 3C; Supplemental Figure S4B). However, the Pearson's correlation coefficient of the CUB mutant with calnexin (ER marker) was not statistically different from that of the wild-type ADGRG6 (Figure 3I). On the contrary, cells expressing ADGRG6 with putative N-glycosylation sites mutated in the PTX domain showed significantly reduced distribution of the receptor on the membrane (Figure 3D; Supplemental Figure S4B). Instead, the mutant protein showed strong colocalization with the ER marker calnexin (Figure 3, D and I; Supplemental Figure S4B). A similar distribution was observed in the Mut6 receptor, with mutations in all 6 putative N-glycosylation sites in the CUB and PTX domains (Figure 3, E and I; Supplemental Figure S4B). The Mut6 mutant did not show a statistically increased colocalization with calnexin over the PTX mutant. The SEA+HrmR mutant successfully trafficked to plasma membrane like the wild-type receptor (Supplemental Figure S3, A–C). Mutations in the GAIN domain alone (Figure 3F) or in combination with the SEA+HrmR domain (Mut9) (Figure 3G) or CUB, PTX, and SEA domains (Mut15) (Figure 3H) showed mislocalization of the receptor (Figure 3I; Supplemental Figure S4B). Combined mutations of N-glycosites in the GAIN, SEA and HrmR domains or the entire NTF did not statistically increase colocalization with calnexin over the single domain mutants. Although we did not observe perinuclear aggregates in Mut15 (Figure 3H)-transfected cells like the tunicamycin-treated ADGRG6-expressing cells (Figure 2A), the receptors largely showed an intracellular distribution in both conditions. Quantifying images for colocalization of wild-type or N-glycosylation mutants of ADGRG6 with the Golgi (Supplemental Figure S5, A–I) also revealed that the PTX, Mut6, GAIN, Mut9, and Mut15 constructs showed significantly reduced accumulation in the Golgi compared with the wild-type, CUB, or SEA+HrmR receptors. Based on the ER and Golgi distribution of ADGRG6 N-glycosylation mutants, we concluded that lack of N-glycosylation at the PTX and GAIN domains likely leads to misfolding of ADGRG6, leading to ER retention and preventing transport to the Golgi body. Because mutating the putative N-glycosylation sites in the CUB or the SEA+HrmR domains of ADGRG6 did not show major trafficking defects, it is possible that the predicted N121, N143, N485, N488, and N505 are not glycosylated in COS7 cells. To confirm the glycosylation status of the mutants studied in Figure 3, we performed biochemical assays described below.

Mutating N-glycosylation sites in the ADGRG6 PTX and GAIN domains prevents furin and GPS cleavage, respectively

Prior work with ADGRE5 has shown that a lack of N-glycosylation suppresses GPS cleavage (Hsiao et al., 2009). To check whether N-glycosylation of ADGRG6 regulates autoproteolysis of the receptor, we performed immunoblots with lysates from cells expressing HA-ADGRG6-FLAG with or without tunicamycin treatment. Similar to the results in Figure 2D, we observed a significant reduction in full-length, NTF, and furin-cleaved fragments. We also observed a band above 100 kD that was both HA and FLAG-tagged, indicating that it is the full-length receptor that has not undergone proteolysis (Figure 4A). The drop in the molecular weight of the nonproteolyzed band compared with the full-length band in the vehicle lane indicates that it is not N-glycosylated due to the effect of tunicamycin. We made a similar observation when performing immunoblots on lysates from cells expressing Mut15. A single band at 150 kD was seen, which was also HA and FLAG-tagged. However, unlike the uncleaved band visualized in tunicamycin-treated lysates from cells expressing wild-type ADGRG6, the Mut15 band was of a higher molecular weight (Figure 4A). This is likely due to the presence of N-glycans on the three remaining putative N residues, which we are unable to mutate in Mut15. On the contrary, tunicamycin treatment completely abolishes N-glycosylation, leading to further drop in molecular weight of ADGRG6 observed in Figure 4B.

FIGURE 4:

Site-specific ADGRG6 NTF N-glycosylation regulates furin cleavage and GPS autoproteolysis. (A) COS7 cells transfected with HA-ADGRG6-FLAG constructs were treated for 12 h with tunicamycin, lysed and subjected to Western blotting with anti-HA, anti-FLAG and anti-actin antibodies. Blot is representative of three independent experiments. (B) COS7 cells transfected with HA and FLAG tagged wild-type and N-glycosylation mutants of ADGRG6 were lysed and immunoblotted with anti-HA, anti-FLAG and anti-actin antibodies. Blot is representative of three independent experiments. Orange and green arrows indicate full-length and NTF bands, respectively. Green and red arrowheads represent furin-cleaved fragments and CTF, respectively. (C) COS7 cells transfected with HA and FLAG tagged wild-type, N-glycosylation and GPS cleavage mutants of ADGRG6 were fixed with ice-cold paraformaldehyde and stained with fluorophore tagged anti-FLAG antibody with permeabilization. The fluorescence intensity was quantified by flow cytometry analysis. Bar graph represents mean±SEM from three independent experiments. The P values were calculated by one-way ANOVA with Dunnett's test; *p < 0.05.

To assess whether there are site-specific roles of N-glycosylation on proteolytic processing of ADGRG6, we performed immunoblots with anti-HA and FLAG antibodies using lysates from cells expressing wild-type or N-glycosylation mutants. Similar to wild-type ADGRG6-expressing cells, CUB mutant showed four bands corresponding to the full-length NTF, the furin-cleaved fragment, and the CTF (Figure 4B). However, in the case of the CUB mutant, the furin-cleaved fragment did not run as a smear but collapsed into a single band, indicating that the N121 and N143 residues are indeed glycosylated. The mutations in the PTX domain of ADGRG6 abolished furin-cleavage without compromising GPS autoproteolysis, as the NTF and CTF bands were visible, while the furin-cleaved fragment was not visible (Figure 4B). A similar observation was made for Mut6. For the PTX and Mut6 constructs, we also observed that both the full-length and NTF proteins showed a lower molecular weight than the corresponding bands in the wild-type ADGRG6 lysates, indicating a lack of N-glycosylation. Immunoblots with the SEA+HrmR mutant did not show any pronounced change in molecular weight or proteolytic processing. We also did not observe any defects in furin or GPS cleavage for the SEA+HrmR mutant (Supplemental Figure S3D). Mutations in the GAIN domain abolished both GPS and furin proteolysis, as indicated by a single band that was HA and FLAG-tagged. This observation was also recapitulated for Mut9 and Mut15 (Figure 4B). Mut9, which is a combination of mutations in the GAIN, SEA, and HrmR domains, migrated at a lower molecular weight compared with the GAIN mutant (Figure 4B), indicating that N residues in the SEA+HrmR domains are likely glycosylated. We were unable to detect the change in the migration of SEA+HrmR bands (Supplemental Figure S3D), likely due to the glycan modifications in the N485, N488, and N505 sites being of low molecular weight.

Thus, putting together the data from Figures 2, 3, and 4, A and B; Supplemental Figure S4, A–D, we can infer that ADGRG6 is not trafficked to the membrane either due to a lack of N-glycosylation or due to a failure of GPS proteolysis. To tease these possibilities apart, we mutated the conserved histidine residue in the ADGRG6 GPS to generate a non–GPS-cleaved receptor (H839R) (Mitgau et al., 2022). Immunoblots with lysates from H839R showed full-length, a higher molecular weight CTF and furin-cleaved bands (Figure 4B). Thus, we concluded that furin-cleavage is not dependent on initial GPS autoproteolysis, and the loss of furin proteolysis observed in the GAIN, Mut9, and Mut15 receptors is not due to a lack of GPS autoproteolysis. Immunofluorescence results with cells overexpressing HA-H839R-FLAG showed that the mutant is trafficked to the membrane (Supplemental Figure S4, A and B), although some of the mutant also colocalized with calnexin, which was significantly higher than wild-type ADGRG6 (Figure 3I; Supplemental Figure S4B). Thus, we concluded that mislocalization of the receptors observed on tunicamycin treatment (Figure 2, A–D) or, in the case of GAIN, Mut9, and Mut15 mutants (Figure 3, F–H), is due to a lack of N-glycosylation at specific residues and not due to a loss of GPS cleavage.

As furin cleavage was not observed for the PTX, Mut6, GAIN, Mut9, and Mut15 constructs, we wanted to determine whether the proteolysis defects are due to impaired trafficking or a direct consequence of lack of N-glycosylation. To assess the intracellular location of ADGRG6 furin proteolysis, we blocked ADGRG6 anterograde trafficking from ER with BFA (Klausner et al., 1992). ADGRG6 did not accumulate in the plasma membrane unlike the control cells and was sequestered in the ER upon BFA treatment (Supplemental Figure S6, A–C). Compared with the DMSO treated cells, BFA incubation also showed significantly higher and lower levels of the full-length and NTF bands, respectively (Supplemental Figure S6, D and E). This is likely due to sustained GPS cleavage resulting from enhanced sequestration of ADGRG6 in the ER, the site of autoproteolysis for other aGPCRs (Volynski et al., 2004). Furthermore, we also saw a significant drop in ADGRG6 furin-cleavage upon BFA treatment, which indicated that ER exit is a prerequisite for furin proteolysis. Hence, we surmised that N-glycosylation of the GAIN domain is necessary for autoproteolysis of ADGRG6, and lack of N-glycosylation in the PTX or GAIN domains prevented furin-cleavage of ADGRG6 by suppressing ER exit.

In Figure 4B, we observed an increase in the intensity of the full-length and NTF bands for the PTX, Mut6, GAIN, Mut9, and Mut15 proteins. Is the lack of N-glycosylation leading to increased ADGRG6 expression or enhancing the stability of mutant ADGRG6? As it is known that the glycosylation status of aGPCRs determines the efficiency of transfer in immunoblots (Perry-Hauser et al., 2022; Bernadyn et al., 2023), it is also possible that the N-glycosylation mutants of ADGRG6 underwent better electrophoretic transfer than the wild-type receptor. To determine the expression levels of the different mutants, we performed flow cytometric analysis of cells expressing the HA and FLAG-tagged receptors by permeabilization and staining with a fluorophore-conjugated anti-FLAG antibody. Upon calculating the fluorescence intensity of the different constructs, we observed that mutating specific N-glycosylation residues in the PTX domains significantly enhances the expression of ADGRG6 over the wild-type receptor. However, mutating the CUB glycosites in combination with those in PTX reduced the expression of the Mut6 receptor to levels comparable with the wild-type. None of the other N-glycosylation mutants showed any significant difference in the expression levels compared with the wild-type receptor (Figure 4C). Taken together, these data confirmed that domain-specific N-glycosylation affects furin-cleavage, autoproteolysis, trafficking, and expression levels of ADGRG6 in a heterologous system.

N-glycosylation of CUB, SEA, and HrmR domains of ADGRG6 affects Stachel peptide-independent cAMP signaling

To determine whether N-glycosylation of ADGRG6 NTF is required for signaling, we quantified cAMP generated by the wild-type and mutant receptors by HTRF. As ADGRG6 basally produces cAMP and also upon stimulation with Stachel peptide, we quantified cAMP levels with or without Stachel peptide stimulation (Mogha et al., 2013; Liebscher et al., 2015). We observed that the basal cAMP generation was significantly reduced for all the ADGRG6 N-glycosylation mutants. However, the autoproteolysis-deficient H839R mutant did not demonstrate any deficits in Stachel peptide-independent cAMP signaling (Figure 5A; Supplemental Figure S4E). Upon Stachel peptide (P14) stimulation, a minor rise in cAMP levels was observed in the empty vector (pcDps)–transfected control cells. Likewise, we observed a minor increase in cAMP in P14-treated cells transfected with PTX, Mut6, GAIN, Mut9, or Mut15 (Figure 5B). However, the increase in cAMP for pcDps and the above mentioned mutants, was not statistically different from the DMSO controls. This can be attributed to the fact that COS7 cells endogenously express ADGRG6(Liebscher et al., 2015). Although basal activity was lost in the CUB mutant, P14-dependent cAMP generation was restored for this mutant (Figure 5B). Mutating N-glycosites across two or more domains of NTF (Mut6, Mut9, and Mut15) did not exacerbate basal or P14-stimulated cAMP generation.

FIGURE 5:

Site-specific ADGRG6 NTF N-glycosylation differentially regulates basal and Stachel peptide mediated cAMP signaling. (A) Basal cAMP signaling by wild-type and N-glycosylation mutants of ADGRG6 calculated as fold change over wild-type transfected COS7 cells. Bar graph shows mean ± SEM from four independent experiments. The p values were calculated by one-way ANOVA with Dunnett's multiple comparison test. (B) Fold change in cAMP levels of DMSO versus 1 mM Stachel peptide (p14) stimulated cAMP generated by COS7 cells overexpressing WT and N-glycosylation mutants of ADGRG6. cAMP levels for all conditions were normalized to DMSO-treated COS7 cells transfected with WT. Bar graph shows mean± SEM from three independent experiments. The p values were calculated by two-way ANOVA with Sidak's multiple comparison test. (C) HA and FLAG-tagged wild-type, N-glycosylation or GPS cleavage mutants of ADGRG6 were expressed in COS7 cells. Staining was performed with fluorophore-tagged HA without permeabilization. Graph represents mean ± SEM from three independent experiments. The p values were calculated by one-way ANOVA with Dunnett's test; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

To confirm whether the drop in basal cAMP generation by the N-glycosylation mutants (Figure 5A) was a result of poor expression levels at the plasma membrane, we compared cell-surface expression of the mutant receptors with wild-type ADGRG6 by flow cytometry. HA and FLAG-tagged receptors were stained for the external epitope without permeabilization. The CUB and the SEA+HrmR mutant receptors did not show significant differences in plasma membrane accumulation compared with the wild-type (Figure 5C; Supplemental Figure S4F). Thus, we concluded that lack of N-glycans in the CUB, SEA, and HrmR domains prevents the tethered Stachel region of the CTF from binding to the orthosteric ligand-binding site, suppressing basal cAMP signaling. However, peptidomimetic signaling by these mutants was not compromised, as they successfully generated cAMP on exogenous P14 stimulation (Figure 5B), as they accumulated in the plasma membrane and had access to the P14 peptide. On the contrary, lack of N-glycosylation in the PTX and GAIN domains likely introduces profound protein misfolding that not only prevents basal signaling (Figure 5A) but also stops ER exit. This leads to reduced plasma membrane accumulation of the PTX, Mut6, GAIN, Mut9, and Mut15 receptors (Figure 5C), which consequently prevents the exogenous P14 ligand from binding to the orthosteric site, aborting peptidomimetic signaling (Figure 5B).

DISCUSSION

ADGRG6 is a mechanosensory aGPCR with crucial roles in embryonic development and human diseases (Li et al., 2024). However, like many other aGPCRs the molecular mechanisms regulating the functions of ADGRG6 remain unclear. We demonstrate that ADGRG6 NTF is extensively N-glycosylated, and these sugar moieties drive various aspects of receptor biology. We identify novel roles for ADGRG6 N-glycosylation in regulating subcellular distribution, proteolytic processing, and cAMP generation, which have been summarized in Table 1. These discoveries build on the work of Stehlik and coworkers (Stehlik et al., 2004) and further our understanding of the molecular pharmacology of ADGRG6. It was known that N-glycosylated ADGRG6 is expressed in HUVECs on proinflammatory stimulation. Moreover, tunicamycin-mediated prevention of N-glycosylation under these conditions did not impair trafficking of the receptor to the plasma membrane. On the contrary, we observed suppression of ADGRG6 trafficking on tunicamycin treatment by multiple experimental approaches. The importance of ADGRG6 N-glycosylation in ER to plasma membrane transport was further highlighted when we observed a dramatic drop in cell surface levels in mutants lacking specific N-glycans. The difference in ADGRG6 trafficking observed by Stehlik et al., may be a cell-type–specific phenomenon. It is possible that ADGRG6 undertakes altered trafficking routes when synthesized by HUVECs experiencing acute immunological challenges. LPS stimulation in HUVECs may also lead to expression of novel splice variants of ADGRG6, whose trafficking differs from an overexpression system.

TABLE 1:

Summary of the observations made using domain-specific N-glycosylation mutants of ADGRG6 in trafficking, Western blotting, and cAMP assays.

	Membrane trafficking	Furin cleavage	GPS cleavage	Stachel peptide independent signaling
CUB	+	+	+	−
PTX	−	−	+	−
SEA+HrmR	+	+	+	−
GAIN	−	−	−	−

We also identified site-specific roles of N-glycosylation in the proteolytic processing of ADGRG6. Mutating putative N-glycosylation sites in the PTX domain of ADGRG6 abolished furin cleavage. This is possible because the PTX domain of ADGRG6 lacking N-glycans at N206, N314, N324, and N353 accumulates in the ER and does not enter the secretory pathway of the cell, while furin is active in the trans-Golgi network, plasma membrane, and endosomes (Thomas, 2002). PTX mutant ADGRG6 was also more abundantly expressed than the wild-type receptor. This could be due to slower ER degradation for this mutant version. Mutating glycosites in PTX along with CUB likely results in a severely misfolded protein triggering clearance from the ER. We observed that mutating N605, N667, N673, N750, N811, and N818 residues in the ADGRG6 GAIN domain also sequesters the protein in the ER. Unlike the PTX mutant, the GAIN domain lacking N-glycans cannot undergo autoproteolysis, an event that occurs in the ER. Because ADGRE5 (Hsiao et al., 2009) lacking N-glycosylation in the GAIN also suppresses autoproteolysis, we conclude that the lack of N-glycans at specific sites in the GAIN domain likely alters the conformation of the nascent full-length receptor, such that the nucleophilic reaction at the conserved HFT sequence of the GPS is prevented (Vizurraga et al., 2020).

We also saw an attenuation of basal or Stachel-independent signaling in the ADGRG6 N-glycosylation mutants (Table 1). This can also be attributed to an altered conformation of the NTF due to a lack of N-glycosylation. Crystal structure and molecular dynamics simulation of zebrafish adgrg6 have revealed that the NTF can exist in an open or a closed conformation, which increases or decreases cAMP generation, respectively (Leon et al., 2020; Dintzner et al., 2024). N-glycosylation of ADGRG6 NTF in the CUB, SEA, and HrmR domains likely stabilizes the open conformation, ensuring basal, nonpeptidomimetic cAMP generation. Furthermore, we observed that preventing N-glycosylation in the CUB domain does not affect P14-stimulated cAMP generation. Thus, we concluded that ligand binding to the orthosteric site of ADGRG6 is not compromised by a lack of N-glycosylation in the CUB domain. The ADGRG6 CUB domain is also known to bind to Ca²⁺ (Leon et al., 2020) and collagen-IV (Paavola et al., 2014), which is a positive allosteric modulator of ADGRG6. Because N-glycosylation of ADGRE5 NTF is critical for ligand binding (Wobus et al., 2004), further studies to determine the role of ADGRG6 N-glycosylation in interaction with Ca²⁺ or ECM ligands can be an important future pursuit.

N-glycosylation may also affect other aspects of ADGRG6 signaling. Although GPCR biased ligands favor signaling by a preferred pathway over others (Tan et al., 2018; Goth et al., 2020), signaling bias can also be introduced by certain posttranslational modifications (PTM), including N-glycosylation (Marada et al., 2015; Soto et al., 2015). Moreover, species-specific differences in GPCR N-glycosylation also alter intracellular trafficking of GPCRs (Marada et al., 2015). As much of our understanding of ADGRG6 in tissue development comes from zebrafish and mouse models, it will be interesting to compare how the variations in N-glycosylation of ADGRG6 in these two vertebrate systems can lead to organism-specific differences in signaling.

Crystal structure of zebrafish adgrg6 has identified putative O-glycosylation sites in NTF (Leon et al., 2020). Future studies determining the identity and the roles of the O-glycan residues on ADGRG6 will enhance our understanding of the molecular pharmacology of the complex receptor.

Various GPCRs are known to undergo misfolding and show altered signaling due to perturbation of N-glycosylation in diseases (Sullivan et al., 1993). A case in point is ADGRE5, expressed in skeletal and smooth muscles (Zyryanova et al., 2014). Although the EGF-like domains in ADGRE5 are not N-glycosylated in skeletal or smooth muscles from healthy donors, these NTF sites undergo N-glycosylation in patients with rhabdomyosarcoma and leiomyosarcoma (Aust et al., 2006). Because glycosylation patterns of proteins are often altered in cancers, it will be useful to learn whether ADGRG6 N-glycosylation is also different in tumors overexpressing ADGRG6 like triple-negative breast, colorectal, and bladder cancers (Cui et al., 2021; An et al., 2022; Li et al., 2024). Tumor-specific alterations in ADGRG6 N-glycosylation can be harnessed to design novel therapeutic antibodies and diagnostic methods. A complete understanding of the glycosylation landscape of ADGRG6 can further our understanding of the role of this aGPCR in human diseases. The NTF of ADGRG6 is involved in mechanosensation and binding to extracellular signaling modulators, which direct inner ear development, myelination, and sorting of peripheral neurons (Tan et al., 2018). Although the current study using overexpression of ADGRG6 N-glycosylation mutants sheds light on the function of this aGPCR in a cell culture model, understanding how N-glycosylation affects signal transduction by the NTF of this modular aGPCR in vivo can facilitate novel drug discovery efforts.

MATERIALS AND METHODS

Request a protocol through Bio-protocol

Reagents and plasmids

All chemicals were purchased from Sigma unless otherwise mentioned. HA-ADGRG6-FLAG in pcDps and FLAG-ADGRG6 (+ss) in pCMV6 were generous gifts from Dr. Ines Leibscher (University of Leipzig) and Dr. Demet Arac (University of Chicago), respectively. The pcDps empty vector was generated by deleting the HA-ADGRG6-FLAG coding sequence using the Q5 site-directed mutagenesis kit (New England Biolabs). All point mutations in HA-ADGRG6-FLAG were also generated by the same kit, and the sequences were confirmed by ABI sequencing. Please refer to the list of primers (Supplemental Information) used for generating the respective point mutations.

Cell culture and transfection

HEK293T and COS7 cells were purchased from Sigma and the National Center for Cell Sciences, Pune, India, respectively. These cell lines were routinely cultured in DMEM (HiMedia) supplemented with 10% heat-inactivated FBS (HiMedia), penicillin (100 U/ml) and streptomycin (100 µg/ml). COS7 and HEK293T cells below passage 20 were transfected with Lipofectamine 3000 (Thermo Fisher Scientific) and Polyfect (Qiagen), respectively, following the manufacturer's protocol.

Western blotting

A total of 10⁶ COS7 cells were transfected in 6-cm dishes with wild-type ADGRG6 or N-glycosylation mutants (tagged with HA and FLAG at N and C-termini, respectively). When required, transfected cells were treated with 1 µg/ml tunicamycin or 2.5 µg/ml BFA (Cell Signaling Technology) in media containing 1% FBS for 12 or 16 h, respectively. Cells were harvested with cell dissociation buffer (Life Technologies) and incubated in lysis buffer [20 mM HEPES pH 7.4, 0.1 mM EDTA, 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, 1% Triton X-100, 1x protease (Thermo Fisher Scientific), and 1% n-dodecyl-β-D-maltoside] for 30 min at 4°C. A total of 60 µg of protein was mixed with 5x sample buffer and incubated at 37°C for 10 min. SDS–PAGE was performed using 8% gels, and the resolved proteins were transferred to nitrocellulose membranes. The membrane was blocked with 1% fish gelatin for 1 h at room temperature and then incubated with primary antibodies overnight at 4°C. The primary antibodies used included: 1:200 rat anti-HA (Clone 3F10, Sigma), 1:100 rabbit anti-FLAG polyclonal (Sigma), 1:100 rabbit anti-ATF6 (Clone D4Z8V, Cell Signaling Technologies) and 1:1000 mouse anti-actin antibody (Clone C4, Santa Cruz Biotechnology). The following secondary antibodies were used: 1:2000 goat anti-rat DyLight 800 (Bio-Rad), 1:2000 donkey anti-rabbit Alex Fluor 488 (Jackson Immuno Research), and 1:5000 goat anti-mouse StarBright Blue 700 (Bio-Rad). The membranes were scanned with Bio-Rad ChemiDoc.

Deglycosylation assay

HEK293T or COS7 cells were transfected with FLAG-ADGRG6 or HA-ADGRG6-FLAG plasmids, respectively, for 48 h and then harvested with cell dissociation buffer. Immunoprecipitation was performed as described before (Pal et al., 2015). Epitope-tagged ADGRG6 protein was immunoprecipitated from HEK293T or COS7 cells with anti-FLAG M2 magnetic beads (Sigma, M8823) or anti-HA magnetic beads (Sigma, SAE0197), respectively. Following overnight immunoprecipitation at 4°C on a nutator, the beads were treated with 10% Glyco DN buffer (New England Biolabs) for 2 h at 37°C with agitation. The supernatant was collected to set up three reactions: an untreated reaction, a PNGase treatment (New England Biolabs), and an EndoH reaction (New England Biolabs) or a sialidase digestion (Sigma). The reactions were performed at 37°C for 2 h and then analyzed by Western blotting as described before. For deglycosylation assays with HA-ADGRG6-FLAG immunoprecipitated from COS7 cells, the proteins were resolved by 4 to 20% precast gradient gels (Thermo Fisher Scientific, XP04200) along with Spectra multicolor broad range protein marker (Thermo Fisher Scientific, 26623).

Cell-surface biotinylation assay

FLAG-ADGRG6 transfected HEK293T cells were biotinylated 48 h posttransfection using the Pierce Cell surface protein biotinylation and isolation kit (Thermo Fisher Scientific, A44390), following the manufacturer's protocol. Cell lysis and protein quantification were performed as described before. Biotinylated membrane proteins were isolated using the manufacturer's protocols. The eluted biotinylated proteins were analyzed by Western blotting with rabbit anti-FLAG polyclonal antibody as described before.

Immunofluorescence and imaging

Transfected COS7 cells were cultured on #1.5 coverslips (Corning) and fixed with 4% para-formaldehyde (Thermo Fisher Scientific) for 15 min at room temperature. For some experiments, the transfected cells were treated with 1 µg/ml tunicamycin or 2.5 µg/ml BFA in media containing 1% FBS for 12 or 16 h, respectively before fixing with 4% paraformaldegyde. For calnexin staining, the fixed cells were incubated with methanol at −20°C for  5 min. Cells were permeabilized with 0.3% Triton X-100 for 5 min and blocked with 5% normal donkey serum (Abcam) for 30 min. Cells were incubated with anti-HA (1:100) (Sigma, clone 3F10), anti-FLAG (1:200) (Sigma, Clone M2), anti-calnexin (1:100) (Cell Signaling Technology, clone C5C9), or anti-GM130 (1:200) (Cell Signaling Technology, clone D6B1) primary antibodies for 3 h at room temperature. Postwashing, cells were incubated with DAPI (1:5000) and fluorophore-tagged secondary antibodies raised in donkey (1:1000). All secondary antibodies were from Jackson Immuno Research. The coverslips were then mounted using Fluoromount-G (Invitrogen) and visualized using Zeiss LSM 900 (oil objective: Plan-Apochromat 63x/NA 1.40). For calculating Pearson's correlation coefficient, images were analyzed with the JACOP plugin available in FIJI.

Flow cytometry

COS7 cells transfected with pcDps, wild-type or mutant ADGRG6 plasmids tagged with HA and FLAG at the N- and C-termini were detached with cell dissociation buffer and treated with fixative available with the FIX and PERM kit (Thermo Fisher Scientific, GAS003) for 15 min at room temperature. The cells were washed and incubated with FACS buffer (dPBS+5% FBS) for 30 min. Surface labeling was performed by incubation with 1:50 dilution of mouse anti HA.11 tagged with Alexa Fluor 647 (Clone16B12, BioLegend, 682404) for 1 h at room temperature. For staining ADGRG6 on the plasma membrane and intracellular locations, transfected COS7 cells were fixed, permeabilized, and stained with a 1:50 dilution of FLAG M2-FITC antibody (Sigma, F4049). The stained cells were strained to obtain a single-cell suspension and subjected to flow cytometry using BD LSRFortessa. Data analysis was performed with FlowJo v10.

cAMP assays

The cAMP Gs dynamic kit (Revvity, 62AM4PEB) was used to quantify basal and Stachel peptide–stimulated cAMP levels. A total of 2 × 10⁵ COS7 cells transfected with 3 µg of empty vector, wild-type ADGRG6, or N-glycosylation mutants (tagged with HA and FLAG at N- and C-termini, respectively) in 12-well plates were detached with cell dissociation buffer and transferred to 96-well plates (35,000 cells/well) 24 h posttransfection. The cells were serum-starved for 3 h at the end of 48 h posttransfection, and then incubated with 1 mM IBMX in stimulation buffer for 30 min at 37°C. A total of 1 mM of p14 (THFGVLMDLPRSAS) Stachel peptide (Genscript) or 0.1% DMSO was used to stimulate the cells in triplicate for 30 min at 37°C. The cells were lysed and cAMP levels were quantified in triplicate in white 384-well plates (Greiner Bio-One) following the manufacturer's protocol. The readings were taken with an HTRF compatible SPARK multimode plate reader from Tecan.

Statistical analysis

The data were checked for Gaussian distribution and then analyzed by one- or two-way ANOVA or unpaired t tests with GraphPad Prism 10. The statistical analyses for each experiment have been reported in the respective figure legends.

Abbreviations used:

ATF6

activating transcription factor 6

BFA

brefeldin A

cAMP

cyclic adenosine monophosphate

DMSO

dimethylsulfoxide

EDTA

ethylenediaminetetraacetic acid

EndoH

endoglycosidase H

FACS

fluorescence-activated cell sorting

FITC

fluorescein isothiocyanate

HA

hemagglutinin

HEPES

4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid

HTRF

homogenous time-resolved fluorescence

IBMX

3-isobutyl-1-methylxanthine

PNGaseF

peptide-N-glycosidase F.