Introduction

GPR180 was identified in 2003 as a gene primarily expressed in vascular smooth muscle cells, playing a significant role in vascular remodeling1,2. Its expression levels correlate with intimal thickening which causes vascular restenosis after catheter injury in rabbit aorta. Hence, it was initially termed intimal thickness–related receptor (ITR)1. Other potential roles in pathophysiological processes were discovered later. GPR180 garnered attention due to its association to congenital microcoria (MCOR) a rare autosomal-dominant disorder, which leads to the inability of the iris to dilate due to absence of dilator pupillae muscle3,4. Furthermore, roles in cancer5,6,7,8,9 as well as lipid and energy metabolism10,11,12,13 have been reported. GPR180 has a positive influence on energy metabolism by regulating thermogenic actions and glucose homeostasis. In this context, Collagen triple helix repeat containing 1 (CTHRC1) was identified as a potential ligand for GPR18011.

GPR180 is a highly conserved seven-transmembrane helix (7TM) receptor (49 kDa) with a large 18 kDa N-terminal domain (NTD) (Fig. 1A). It was considered an orphan GPCR due to its topological resemblance to Rhodopsin-like G protein-coupled receptors (GPCR). The recently determined atomic structure of TMEM87A, a related 7TM receptor with a large N-terminal Golgi dynamics (GOLD) domain14 uncovered homology to other proteins with similar predicted structures consisting of a GOLD domain fused to a 7TM domain: members of the lung 7TM receptor (LUSTR) family (TMEM87B, GPR107, and GPR108) and previously unrecognized homologs TMEM145, TMEM181, Wntless (WLS) including GPR180. Members of this protein family were termed GOST proteins (GOLD domain seven-transmembrane helix) and Golgi or Golgi-network localization was shown for most members of this protein family: TMEM87A15,16, GPR10717,18, GPR10819,20,21, TMEM18122, and Wntless23.

Fig. 1: Biochemical characterization of GPR180 constructs.
figure 1

A Topology model of GPR180 shown in a detergent micelle (left) and embedded in a membrane (right). B Phylogenetic tree containing GPR180 from different species and TMEM87A and Rhodopsin as examples for two structurally very similar proteins. C GPR180 constructs used for structural, biochemical, and biophysical analysis. D Size-exclusion chromatograms of the GPR180NTD constructs used in this study (human in grey and mouse in light blue). E Size-exclusion chromatograms of the FL-GPR180 constructs (GPR180MBP in dark blue and GPR180GFP in green) used in this study. F CD-spectroscopy of GPR180NTD (grey) and GPR180MBP (dark blue). G DSF measurements with CPM dye of GPR180MBP (dark blue and light blue) and GPR180GFP (dark green and light green). The plot shows the individual data points (N = 2).

Full size image

Little is known about ligands and the exact function of GOST proteins. Despite the structural homology of their transmembrane domains (TMD) to GPCRs no evidence for GPCR-like signaling was demonstrated for any GOST family member. Rather, it was speculated that GOST proteins might function as trafficking chaperones for membrane-associated cargo. This hypothesis is based on the cryo-EM structures of WLS in complex with Wnt3a (PDB 7KC4) and Wnt8a (PDB 7DRT) which showed that the helical core of Wnt interacts with the GOLD domain and a palmitoylated hairpin is bound in a large hydrophobic cavity within the TMD of WLS. Such hydrophobic cavities within the TMD are predicted for all GOST protein family members, which could be important for ligand binding14. Additional cryo-EM density within the hydrophobic cavity of the TMEM87A transmembrane domain (PDB 8CTJ)14 was modeled as a phospholipid. In two other recently deposited structures phosphatidylethanolamine (PDB 8HSI) and gluconate (PDB 8HTT) were assigned to this density, indicating a potential lipid binding site within the cavity. Other potential ligands were proposed for GPR107 (neurostatin24) and GPR180 (L-lactate25 and CTHRC111). The putative GPR180 ligand CTHRC1 is particularly interesting since it is a collagen triple helix repeat-containing protein and the NTDs of the GOST family proteins exhibit structural similarities with collagen-binding domains of collagenases (PDB 3JQW or 1NQJ). However, so far, no experimental validation of collagen interaction of any GOLD domain-containing protein has been published26. Although GPR180’s role in physiological and pathophysiological processes has been investigated over the last 21 years, insufficient attention has been paid to GPR180 on the molecular level. Detailed structural, biochemical, and biophysical studies with recombinant GPR180 are missing. Assuming that GPR180 is a new member of the GOST family it should not be localized at the cell membrane, but rather within the Golgi network or other vesicular structures due to its potential trafficking function and it might have a similar mode of action compared to other family members. GOLD domain proteins are known for their roles in cargo recognition and transport along the secretory pathway27,28. In this context, the binding partners are called cargo, because the sole purpose for their interaction is transport. So far, it is not clear whether ligand binding to GPR180 induces a classical signaling pathway or if GPR180 functions as a cargo receptor and participates in the secretory pathway.

In this study, we provide the structural and biochemical characterization of GPR180, describe effects of pH differences on its structure, and investigate its putative binding pocket by HDX-MS. The intracellular localization of GPR180 further establishes this protein as a member of the GOST protein family. We propose a pH-dependent widening of a charged ligand-binding pocket in the transmembrane region of GPR180, possibly mediated by a highly conserved histidine patch in one of the loops connecting two transmembrane helices.

Results

Recombinant C-terminally truncated GPR180 can be purified as fusion protein

The extremely high sequence conservation between GPR180 proteins in even distant species (Fig. 1B, SI Fig. L1) indicates a very important biological role of this integral membrane protein. To elucidate its cellular function and potential role in disease we produced different recombinantly expressed variants of human and mouse GPR180 (Fig. 1C, SI Fig. S1A–D). Several constructs of the isolated NTD of GPR180 (starting at Lys23, directly after the signal peptide, and ending at residues 162–172) were expressed in insect cells to high levels and purified as monomers from the culture media (Fig. 1D). The smallest variant was used for further studies, as compact and globular domains are in general more promising to crystallize. Expression of high-quality GPR180 containing the transmembrane region was more challenging: the full-length protein without modifications could not be expressed in high enough quantities in insect or mammalian cells. Addition of fusion proteins to the N-terminus led to a high level of degradation, hampering the purification of the full-length construct. Truncations of the C-terminal residues, however, increased expression yields dramatically. Constructs with highest expression yields in insect cells contained an MBP or GFP protein instead of the last 23 residues at the C-terminus (Fig. 1C). Recombinant GPR180 was solubilized from membranes using detergent and formed a mixture of monomer and higher oligomeric species (Fig. 1E). CD spectra and melting curve of the purified GPR180MBP fusion protein prove the presence of a well folded protein (Fig. 1F, G).

GPR180 N-terminal domain is a GOLD domain

To gain insights into the biochemical and structural properties of GPR180 we produced the isolated NTD for crystallization experiments. Glycosylation at two different asparagine residues resulted in a heterogeneous protein sample which failed to crystallize — even after incubation with different glycosidases. Interestingly, the murine homolog of GPR180 (83.4% sequence identity) lacks one of those asparagine residues and is, therefore, less glycosylated. Purified murine GPR180NTD showed a much higher level of homogeneity than the human ortholog and formed very well-diffracting crystals. The structure of the NTD of mGPR180 was solved by molecular replacement using a 3D-structure predicted by AlphaFold229 as search model and refined to 1.9 Å (Fig. 2A, B, Table 1).

Fig. 2: X-ray structure of the mNTD.
figure 2

A Top view of the experimentally determined X-ray structure of the GPR180 mNTD showing the density for the glycosylation site N110 in a close-up. The remaining amino acids from the TEV cleavage site are highlighted in green, and the disulfide bridge is shown in orange. The variable helices and loops, which are unique for every GOST protein are colored in dark blue. The two beta-sheets, which resemble the core of every GOLD domain, are colored in light blue. B Front view of the experimentally solved X-ray structure of the mNTD colored in the same way as described in (A). C Overlay of the X-ray structure (light blue) and the AlphaFold model of the GPR180 mNTD (grey).

Full size image
Table 1 Data collection and refinement statistics (molecular replacement)
Full size table

GPR180NTD is composed of a β-sandwich domain decorated with four small α-helices and shows high similarity to the predicted AF2 model (Fig. 2C) (R.M.S.D = 0.45 Å between main chain atoms). The three molecules in the asymmetric unit are very similar with the only significant differences in the loop between Arg 134 and Glu 143. This loop is most likely flexible in solution and the distinct conformations are due to crystal packing effects. The β-sandwich domain consists of eight β-strands in two opposing sheets with β1, β8, β3, and β6 facing β2, β7, β4, and β5, typical for a member of the Golgi dynamics (GOLD) domain protein family27. In addition, five extra residues at the C-terminus (remaining after TEV cleavage during protein purification) are visible in the electron density map, forming an additional β-strand (β9 next to β6) and completing the β-sandwich structure (Fig. 2A, B). The intra-strand disulfide bond observed in many GOLD domains is not present in GPR180, instead, a cysteine in the loop region between β7 and β8 is covalently linked to a cysteine at the end of one of the small α-helices. As expected, clear additional electron density is visible for the glycan attached to Asn 110 (Fig. 2A).

We used our experimental structure of the GPR180NTD to identify structural homologs in the Protein Data Bank30 (PDB) and in the AlphaFold Protein Structure Database31 (AF DB, limited to human proteins only) (SI Tab. T1-2). In both databases, we identified candidate proteins that most probably feature GOLD domains (Fig. 3A). Other hits are either collagen binding domains (Fig. 3B), calpains, or the bacterial toxin PirA. Despite high structural similarities, no protein found by these searches shares any sequence similarities with GPR180.

Fig. 3: Results of the 3D-structure comparison using DALI.
figure 3

A Representative structure of a GOLD-domain containing protein (TMED1 PDB 7RRM) from the 3D-Search against the PDB. B Representative structure of a collagen-binding domain (Collagenase from Clostridium Histolyticum PDB 1NQJ). C GOST proteins GPR107, GPR108, and TMEM87B appeared as hits in the 3D-Search against the AlphaFold Database.

Full size image

In the AF DB we identified several GOST proteins containing a 7TM domain very similar to the one predicted for GPR180: GPR108, GPR107, and TMEM87A/B (Fig. 3C). Some of those were shown to be localized to the Golgi apparatus, which led to our hypothesis that GPR180 might not be localized in the plasma membrane, as previously assumed, but also plays an important role in the Golgi.

GPR180 is localized to intracellular vesicles

GOLD or GOST proteins, as their names already suggest, are intracellularly localized, more specifically confined to the Golgi. To experimentally determine the cellular localization of GPR180, we performed cell imaging studies using transiently transfected HEK293 cells with constructs of human GPR180, TMEM87A, Rhodopsin, and GPR52 with C-terminal GFP-fusions. GPR180GFP and TMEM87AGFP (Fig. 4A) were detected close to the Hoechst-stained nucleus with no visible plasma membrane staining. In contrast, RhodopsinGFP (Fig. 4A) showed clear plasma membrane staining combined with intense signals in intracellular vesicles containing the over-expressed RhodopsinGFP. GPR52GFP (unrelated GPCR control) showed homogeneous cytoplasmic localization without specific compartmentalization (Fig. 4A). To see if the intracellular localization of TMEM87AMyc and GPR180Myc overlaps with the Golgi apparatus or lysosome we used specific markers for these organelles (Golgin9732 and Lamp133). As shown in Fig. S2, TMEM87A and GPR180 showed no clear colocalization with Golgin-97 or Lamp1. Thus, we could not confirm previous findings15,16 and cannot clearly specify the localization of either protein to the trans-Golgi network or lysosomes.

Fig. 4: Localization studies in transiently transfected HEK293 cells.
figure 4

A Confocal microscopy with C-terminally GFP tagged constructs. Proteins with GFP-tag appear in green and the nucleus was stained with Hoechst (blue). Scale bar 10 µm. B Expansion microscopy with immunofluorescence staining. Myc-tagged proteins appear in green, nucleus in blue, and unspecific protein staining in red. Scale bar 10 µm. ExM scale bars are corrected for the expansion factor (3.95 x).

Full size image

To uncover more details about the intracellular distribution of our constructs, we combined expansion microscopy and immunofluorescence staining. We used the same set of constructs for transient transfection of HEK293 cells, but this time the protein constructs had a C-terminal Myc-tag (EQKLISEEDL) for tag-specific antibody recognition and signal amplification. Unspecific protein staining was used to visualize the general morphology of the cell (red). For GPR180Myc (Fig. 4B) we observed cytoplasmic localization in vesicles, partially near the nucleus. TMEM87AMyc (Fig. 4B) was also detected intracellularly, looking similar to GPR180. RhodopsinMyc was clearly located in the plasma membrane with some vesicular structures within the cell. GPR52Myc was found inside the cells without any detectable plasma membrane staining (Fig. 4B). These results indicate that GPR180 is located in vesicles of the endomembrane system, localized close to the nucleus.

The conformation of GPR180 is sensitive to pH differences

One of the central differences between the extracellular space and the lumen of many inner cellular vesicles is their pH value. Organelles of the secretory and endocytic pathways are distinguished by their luminal acidity, which can vary between pH 6.7–4.7, whereas the plasma has a neutral pH of around 7.434. Many cellular processes are regulated by these pH differences through conformational changes of relevant receptors or ligands upon pH change. We approached the question whether GPR180 activity may be related to changing environment pHs by characterizing its stability and shape as a function of different pH values. We measured the thermal stability of our GPR180 fusion protein variants in the CPM assay. CPM is a cysteine-specific binding dye and binds to buried cysteines only upon unfolding. As the only buried cysteines are present in GPR180 and not in the fusion proteins, we can specifically monitor unfolding of GPR180 using this assay. All GPR180 constructs exhibit similar melting curves at the tested pHs. We observed a decrease in stability at the low end of the tested pH range (Fig. 5A, B). In addition, size exclusion profiles and CD-spectra (Fig. S4) at different pHs are very comparable (Fig. 5C). Both experiments show that no significant pH-dependent changes occur and that the protein is properly folded under all conditions tested.

Fig. 5: pH-dependent protein stability.
figure 5

A Melting curve of GPR180MBP at pH 7.5, 6.5, 6.0, and 5.5 obtained by DSF measurements. The plot shows the individual data points (N = 2). B DSF melting curves of GPR180GFP at pH 7.5, 6.5, 6.0, and 5.5. The plot shows the individual data points (N = 2). C Size-exclusion chromatograms of GPR180MBP at pH 7.5, 6.5, and 5.5 used for HDX-MS. D X-ray scattering profile where the normalized scattering intensity is plotted against the scattering vector q. E Guinier-plot with the logarithmic scattering intensity ln(I) versus the squared scattering vector q2. F Small-angle x-ray scattering normalized pair distribution function (P(r)) of the scattering data of GPR180MBPand GPR180GFP.

Full size image

Small-Angle X-ray Scattering (SAXS) is a solution-based method that allows for the visualization of conformational changes of proteins that lead to a significant change in molecular shape. To ensure monodispersity of the samples and avoid problems derived from detergent in the buffer, we exchanged the detergent micelle with small amphipathic helices, so-called peptidiscs35, and used an online size-exclusion chromatography (SEC-SAXS) set up at the B21 beamline at the Diamond light source (DLS), Oxford, UK. Both MBP and GFP fusions of GPR180 showed an increase in the radius of gyration (Rg) by more than 5 Å when lowering the pH from 7.5 to 6.5, indicating significant conformational changes (Fig. 5E). The normalized pair distribution function (P(r)) of the scattering data (Fig. 5F) clearly shows an increase of larger distances at lower pH for both GPR180 fusion proteins. Scattering curves (Fig. 5D) and derived values are included in the supplementary data 1.

The transmembrane domain of GPR180 opens at lower pHs

The resolution of structural information from SAXS is too low to clearly identify the nature of conformational changes that occur in GPR180 upon lowering of the pH. In order to get more detailed information on the observed structural changes, we utilized technologies suitable for delivering higher-resolution structural insights. First, we used the GPR180MBP and GPR180GFP proteins to prepare grids for cryo-electron microscopy. Unfortunately, the obtained 2D classes show flexibility in the relative orientations of NTD and TMD which prohibits the determination of high-resolution structures. Generation of rigid fusions to GPR180, as successfully done to determine several high-resolution structures of GPCRs36,37,38, did not result in stably expressed high-quality protein.

Next, we performed Hydrogen/Deuterium Exchange-Mass Spectrometry (HDX-MS) experiments. HDX-MS is a powerful analytical technique that can provide useful information about solution phase protein conformation and dynamics39,40,41. Recently, it has become more and more accessible for analyzing integral membrane proteins through progress made in proteolytic peptide generation and detection42,43,44. While it is generally challenging to achieve full sequence coverage for transmembrane regions of membrane proteins, with GPR180MBP we were able to reach an extremely high sequence coverage of 95%, including large stretches of the TMD. The peptide coverage map obtained for GPR180MBP is shown in Suppl. Fig. S5. To investigate the dynamics of GPR180MBP at pH 7.5, deuterium uptake percentages (corrected for back-exchange) at specific incubation times ranging from 10 s to 53 min were mapped onto the structure of GPR180 (Fig. 6). Due to the lack of an experimental full-length structure, we used the AlphaFold2 model for this purpose. As described above, we validated the model for the NTD by an experimental X-ray structure and we assumed that also the predicted TMD structure would be of high confidence since the AF model of FL-GPR180 shares high structural similarity to the experimentally solved structure of TMEM87A. The exchange rates at pH 7.5 follow the expected pattern with a faster exchange in the solvent-exposed domains (NTD and MBP) and the loops between the TM helices and a slower exchange in the membrane/detergent-covered regions.

Fig. 6: Conformational dynamics of GPR180MBP at pH 7.5.
figure 6

Deuterium uptake percentages (corrected for back-exchange as detailed in the methods) at peptide level (considering overlapping peptides and excluding the first two residues at the N-terminus of each peptide) mapped onto the GPR180 AlphaFold model using the color code displayed.

Full size image

To identify conformational changes of GPR180MBP as a function of pH, we performed differential HDX-MS (ΔHDX) experiments at three different pHs: 7.5, 6.5, and 5.5. These pH values were selected as the relevant pH range present in different inner cellular vesicles. The thermal stability data discussed above (Fig. 5A) showed that the protein is properly folded at these pHs with slightly decreased melting temperatures at <pH 6.0. As the changes in pH affect the intrinsic chemical exchange rates, a mathematical time correction was applied to determine equivalent deuterium labeling times at the different pH values as described in the experimental section. The effects of different pH values on GPR180MBP structure, when changed from pH 7.5 to 6.5 and 5.5, were identified by using pH 7.5 as the reference state. Longer time points were excluded to avoid misinterpretation of data from excessively long incubation times of GPR180MBP under the labeling conditions at pH 6.5 and 5.5. Figure 7A shows significant HDX differences in GPR180 observed at pH 6.5 and 5.5 versus pH 7.5. Figure 7B shows representative HDX uptake plots selected from different regions of GPR180MBP (a complete set of deuterium uptake plots of GPR180 are included in the supplementary data 2).

Fig. 7: pH-induced HDX differences of GPR180MBP.
figure 7

A Regions showing significant HDX differences at pH 6.5 (left) and pH 5.5 (right) compared to the pH 7.5 reference state mapped onto the GPR180 AlphaFold model. B Deuterium uptake plots of representative peptides from the regions marked at pH 7.5 (black), 6.5 (blue), and 5.5 (green). All the time points were corrected to pH 7.5 reference state and no back-exchange corrections were applied to % deuteration. Error bars indicate the 99% confidence interval from quadruplicate measurements at each time point.

Full size image

The ΔHDX experiments confirm an influence of pH on the conformational state of GPR180 and reveal areas in the receptor which show lower (protection) or higher (deprotection) deuterium uptake compared to the reference state. Deprotection in ΔHDX indicates increased flexibility and can be explained by structural rearrangements. Changing the pH from 7.5 to 6.5 leads to a significant deprotection spread over both the NTD and TMD of GPR180 (Fig. 7A). In the NTD, mainly the side facing the TMD shows differences (peptides 1 and 2 in Fig. 7B). In the TMD the termini of the helices are affected, whereas the core shows no significant differences (peptide 3 in Fig. 7B). The extent of deprotection for TM 4–7 seems to travel deeper into the core of the TMD when lowering the pH to 5.5 (see SI Fig. S7). Interestingly, not all TM helices behave the same. Three of the TM helices (TM2, TM6, and TM7) show strongest deprotection (peptides 5 and 6 in Fig. 7B). Those three are involved in formation of the central TMD cavity, which was proposed to bind to lipids and potentially other ligands for other members of the GOST family14.

To ensure that there is no general pH influence on proteins in detergent micelle formulation we performed the same experiments with GPR52, a GPCR which also shows good peptide coverage in the TM region in HDX-MS experiments (SI Figs. S8, S9, supplementary data 3) and is thermally stable over the pH range tested (Fig. S3). In contrast to GPR180, barely any changes in HDX between pH 7.5 and pH 6.5 were observed. Only at pH 5.5, the protein seems to start unfolding as shown by an extensive increase in HDX across the TMD. The clear difference observed between GPR180 and GPR52 supports our hypothesis of a distinct pH sensitivity of GPR180 resulting in a widening of a putative ligand binding site in its TMD at acidic pH.

The protonation state of a histidine-rich TM helix connecting loop might be inducing additional conformational changes

For the structural interpretation of pH-dependent effects in the range between pH 5.5 and 7.5, histidine residues (pKa ~ 6–7) typically play a crucial role, because their protonation might lead to changes in intramolecular interactions. Comparing GPR180 sequences from different species reveals a highly conserved histidine patch located in the loop between TM4 and TM5, which is oriented toward a surface patch of the NTD also containing some histidine residues (for a sequence alignment see SI List L1). This loop is one of the few regions in the AlphaFold2 model which is predicted with a lower model confidence, and it might as well adopt a different structure or possess intrinsic flexibility (Figs. 6, 8A). The HDX-MS data showed that the loop does not exchange readily with deuterium, and this could rule out that it is entirely flexible or intrinsically disordered. At lower pH values — as present in most inner cellular vesicles – the histidine side chains become more and more protonated leading to an accumulation of positive charges at the interface between NTD and TMD. Repulsion of the two domains would facilitate access to the charged, putative ligand binding pocket in the TMD from the lumen of the vesicles as a result of the movement of the NTD away from the TMD. The calculated electrostatic surface potential of GPR180 at pH 6.5 and 7.5 using APBS (the Adaptive Poisson-Boltzmann Solver45) nicely shows the differences in charge distribution at the interface between TMD and NTD (Fig. 8B).

Fig. 8: Repulsion of both domains opens access to the TMD cavity.
figure 8

A AlphaFold model of GPR180 (colored by confidence) with low confidence in the histidine patch of ECL2. B APBS calculations for GPR180 at pH 7.5 (left) and pH 6.3 (right) show an increase in positive net charge especially at the interface between the TMD and NTD. The histidine-rich interface is shown as close-up. The model of the open conformation of GPR180 was created manually to show the otherwise covered surfaces.

Full size image

Discussion

GPR180 is highly conserved across the animal kingdom and plays a central role in vascular remodeling1,2 and energy metabolism11,12,13. Strikingly, there are no sequence-based similarities or common motifs shared with other protein classes. Available information on this protein is scarce and mainly limited to cell-based and in vivo studies. We used a combination of structural biology, biochemistry, and cell biology to reveal a potential role of GPR180 in acidic subcellular vesicles.

The X-ray crystal structure of the GPR180NTD and the search for structural homologs revealed similarities to a set of proteins of diverse nature, which would not have been identified by sequence-based searches. In addition to GOLD domains, named after their strong association with the Golgi apparatus, general structural homology to collagen binding domains46 was identified. CTHRC1, collagen triple-helix repeat containing protein 1, is the only protein-ligand proposed as possible ligand for GPR180 to date11. Plausible superimpositions of a structural homolog of a collagen-binding domain (PDB 1NQJ) with GPR180NTD are possible in two relative orientations. In one superposition the potential collagen binding groove is blocked by the small α-helices decorating the β-sandwich. A large movement of these helices would be needed to allow for an analogous collagen binding to GPR180. In the other orientation, the potential collagen binding groove is on the opposing beta-sheet (SI Fig. S10) distal to the GPR180 TMD. We conclude that there is no obvious structural evidence for collagen binding. Further, to our knowledge, collagen binding by any GOST protein has not yet been shown, although the structural similarity has been observed before14,26. Nevertheless, we cannot rule out a direct interaction in vivo as CTHRC1 might be processed in the body before binding to GPR180 or important cofactors might be missing.

The availability of the predicted full structural human proteome in the AlphaFold DB31 opens new prospects for the identification of structurally homologous proteins. Besides GOST proteins, structurally close homologs of GPR180 are GPCRs from the rhodopsin family. As already proposed by others11,14, there is no evidence for G-protein signaling for GPR180 or any other GOST protein family member. GPR180 shares no common sequence motif with any other protein and is therefore also lacking important GPCR motifs47,48,49. Furthermore, there is no predicted helix 8 which is important for G-protein or arrestin binding50,51. The functional role of GOST proteins is still enigmatic, mainly due to the lack of experimental data on this class of receptors. Roles in protein transport have been proposed for some members15,17,18,20,21,23,52,53, where both the GOLD domain and the TMD could be involved in cargo binding. Our new data fits to these theories: GPR180 does not seem to be located in the plasma membrane but shows cytoplasmic localization in vesicles near the nucleus. These vesicles seem to be distinct from Golgi vesicles or lysosomes. We are confident that these findings are not artifacts of transient over-expression of the protein, as they are in line with information derived from “The Human Protein Atlas”, which shows that endogenous GPR180 is also detected in vesicles54. The nature of these vesicles is still not clear and needs further investigation. GPR180 was shown to have diverse functions in different tissues, which makes cell type-specific differences in localization and trafficking a plausible scenario. We cannot exclude that under specific conditions GPR180 might migrate to the Golgi apparatus or the plasma membrane.

Based on these findings we investigated the influence of pH changes on the structure of GPR180. An increase of the radius of gyration can be observed in SAXS measurements for both GPR180 constructs when lowering the pH. We can exclude that the increase in particle size is due to unfolding of the protein since gel filtration chromatograms and thermal stability of the proteins prove that protein integrity and stability is given for the physiologically relevant pHs of 7.5 (extracellular space) and 6.5 (Golgi apparatus and vesicles of the trans-Golgi network). Based on our high-quality HDX MS data with almost complete peptide coverage even in the TMD (95%) we propose a pH-induced structural rearrangement of GPR180. At the physiologically relevant pH 6.5 we discovered an increased solvent exposure of the charged pocket in the center of the TMD. Since GPR52, a GPCR purified in the same buffer and used here as a control protein, does not exhibit any pH-dependent conformational change at pH 6.5 we conclude that the effects observed for GPR180 are not artifacts derived from the detergent micelle or unfolding of the protein.

Upon lowering the pH to 5.5 GPR180 deprotection proceeds to the center of a subset of its TM helices. At this pH value we cannot exclude the starting influence of acid-triggered unfolding processes (Tm-shift to 33 °C) which could further increase the level of H/D exchange in this area.

The same effect is observed for the GPR52 TMD which again could be explained by a significant decrease in thermal stability. We conclude that pH 5.5 data cannot be confidently interpreted as solely stemming from specific protein conformational changes.

Based on the combined set of biophysical data presented here we propose a model of pH-dependent conformational changes of the GPR180 structure.

First, we assume that protonation at acidic pH of the highly conserved patch of histidines in the TM4-5 connecting loop and NTD of GPR180 could lead to an opening of the cleft between the NTD and TMD of the receptor. The change in relative orientation of the two domains facilitates easier access to the region in the TMD which forms a funnel shaped cavity. Second, we propose that acidic pH has a specific widening effect on the cavity which in turn would create space to allow for receptor ligand binding.

Recent literature on two members of the GOST family is supportive for our model. The TMD cavity in WLS is predicted to be smaller in the non-liganded state compared to the experimental structure of Wnt complexes suggesting structural flexibility within its TMD and a size increase of the cavity upon ligand binding. Wnt binds with a palmitoylated peptide stretch into the hydrophobic WLS cavity (see SI Fig. S11)14,55,56. In the TMEM87A structures cryo-EM density within the TMD cavity14 was modeled as bound phospholipid. The propensity for lipid binding in the hydrophobic cavity can be further supported by the two other TMEM87A structures with phosphatidylethanolamine and gluconate assigned to this cavity.

Despite these obvious similarities, there are specific features of GPR180 that are distinct from the other family members. While the histidine patch is conserved in GPR180 proteins from all species, it is not present in other members of the GOST protein family. Further, TMEM87A14 (PDB 8CTJ) and TMEM87B (AlphaFold) possess a disulfide bond between the NTD and TMD, which probably stably tethers both domains and precludes a change their relative orientation as proposed here for GPR180. WLS (PDB 7KC4) and TMEM181 (AlphaFold) on the other hand have an additional N-terminal TM helix 8, which stabilizes the GOST domain as well. Therefore, we suggest that GPR180 and GPR107, GPR108, TMEM145 (AlphaFold) display a specific mode of action distinct from other GOST proteins, due to their flexible attached N-terminal GOLD domain.

Taken together we present new data on the still enigmatic protein family of GOST proteins, which contains highly conserved receptors with yet unknown function. GPR180 is solidified as a member of this family, having a GOLD N-terminal domain and being localized to inner cellular vesicles. Further research is needed to identify the exact nature of these vesicles and the ligand that can bind to GPR180’s central cavity.

Methods

Virus generation and insect cell expression of GPR180 constructs

Human GPR180 (GPR180-MBP = GPR180_1-417_CGSGS_MBP_TEV_HIS and GPR-GFP = GPR180_1-417_CBRIL_TEV_GFP_HIS) and GPR180-NTD (human = GPR180_melittin_h23-162_CTEVHis and mouse = GPR180_melittin_m24-162_CTEVHis) constructs in a FastBac1-vector were commercially obtained from GeneArt (ThermoFisher Scientific). MAX EfficiencyTM DH10BacTM competent cells (Invitrogen) were used to generate a Bacmid following the manufacturer’s instructions. Sf9 cells (ThermoFisher Scientific) were cultured in Sf-900 III SFM Medium (Invitrogen) and P1 virus was generated from cells transfected with Cellfectin® II Reagent (Invitrogen) according to the manufacturer’s instructions. P2 virus was then generated by infecting cells at 1 million cells per ml with P1 virus and harvested at 48 h. P3 virus was generated in a similar manner to expand the viral stock. The P3 viral stock was then used to infect H5 cells (ThermoFisher Scientific) at 1 million cells per mL for 72 h. Insect-XpressTM Protein-free Insect Cell Medium (Lonza) was used for expression. Pellets containing FL-GPR180 or supernatant containing GPR180-NTD were harvested by centrifugation (2500 rpm for 40 min). Pellets were flash-frozen in liquid nitrogen and stored at −80 °C until further use. Supernatants containing GPR180-NTD were directly used for purification.

Purification of FL-GPR180

Pellets containing FL-GPR180 were resuspended in 1/10 of the expression volume using buffer 1 A (50 mM HEPES, 100 mM NaCl, 20% Glycerol, pH 7.5) with Protease inhibitor (Complete EDTA free von Roche Diagnostics, 1 tablet per 50 mL) and DNAse I (Roche). Cells were lysed using nitrogen cavitation, utilizing an excessive pressure of 50 bar for 30 min. For protein solubilization, 1% (w/v) DDM-CHS (10% stock (w/v): 2% (w/v)) was added to the lysate and rotated for 2 h at 7 °C. The lysate was centrifuged for 60 min at 55.000 × g at 4 °C. The supernatant was loaded onto a gravity-flow column containing 2 mL Ni-NTA resin (Protino) per liter expression volume pre-equilibrated with buffer 2 A (50 mM HEPES, 300 mM NaCl, 20% Glycerol, 20 mM Imidazol, pH 7.5). The column was washed four times with 10 column volumes (CV) of buffer 2 A containing a lower amount of detergent in each washing step (1% - 0.5% - 0.25% - 0.125%). Additionally, the column was washed two more times with 10 CV of buffer 3 A (50 mM HEPES, 300 mM NaCl, 20% Glycerol, 40 mM Imidazol, pH 7.5).) containing 0.06% and 0.03% detergent. Puffer 4 A (50 mM HEPES, 300 mM NaCl, 20% Glycerol, 250 mM Imidazol, pH 7.5) was used to elute the protein in five fractions a 1 CV. Protein-containing fraction were concentrated with Amicon Ultra spin concentrator 50 kDa cutoff (Millipore). The concentrated sample was further purified by performing size exclusion chromatography (SEC) using an ÄKTA pure system with a Superose™ 6 Increase small-scale SEC column (Cytiva) in buffer 5 A (20 mM HEPES, 150 mM NaCl, pH 7.5 + 0.03% (w/v) DDM-CHS).

Incorporation of FL_GPR180 in Peptidiscs

Peptide NSPr (Nter-FAEKFKEAVKDYFAKFWDPAAEKLKEAVKDYFAKLWD-Cter)35 was purchased from WuXi AppTec. The lyophilized peptide was resuspended in ddH2O to create a 25 mg/mL Stock, which had a pH 2–3 due to residual TFA from peptide synthesis. Solubilized peptides were stored at −20 °C. For peptidisc incorporation, an on-bead exchange protocol35 was performed during GPR180 purification. The purification was similarly performed as described above until the protein capture step on the Ni-NTA column. The column was washed 2x with 10CV of buffer 2 A containing 0.06% (v/w) of detergent (2x CMC). 1 CV of NSPr (1 mg/mL) was prepared using buffer 2 A containing 0.5x CMC. The pH value was checked right before use and the NSPr solution was added to the column for 5 min at 4 °C. The column was washed 2x with 10 CV buffer 3 A without any detergent. Elution was performed using 5×1 CV of buffer 4 A without detergent. Protein-containing fractions were concentrated and subjected to SEC as described above. Purified GPR180 in peptidisc was frozen in liquid nitrogen and stored at −80 °C.

Purification of GPR180NTD

Protease inhibitor (cOmplete ULTRA Tablets, EDTA free from Roche) and DNAse I (Roche) were added to the supernatant. The supernatant was bound to Ni-NTA resin (Protino) (3 mL resin per 1 L) for 1 h at 7 °C. The resin was collected in a column and washed two times with 10x CV buffer 1B (50 mM Tris, 300 mM NaCl, 10 mM Imidazol, pH 8). GPR180NTD was eluted with 5× CV buffer 2B (50 mM Tris, 300 mM NaCl, 250 mM Imidazol, pH 8), spin concentrated with Amicon Ultra spin concentrator 30 kDa cutoff (Millipore), and then loaded onto a Superdex200 column (GE Healthcare) on an ÄKTA pure system (Cytiva) equilibrated in buffer 3B (50 mM Tris, 150 mM NaCl, pH 8). Peak fractions containing GPR180NTD were then collected, spin concentrated using Amicon Ultra spin concentrator 10 kDa cutoff (Millipore), and flash frozen in liquid nitrogen.

Differential scanning fluorimetry (DSF/NanoDSF)

NanoDSF measurements were carried out on a Prometheus Panta (NanoTemper Technologies) with high sensitivity capillaries. Protein samples were measured at different concentrations: GPR180NTD at 0.125 mg/mL, GPR180MBP at 0.250 mg/mL, and GPR52 at 0.250 mg/mL. Different buffers with the same composition (20 mM buffer substance 150 mM NaCl pH 4–10 ± 0.03% (w/v) DDM-CHS) covering a pH range between 4.0–10.0 in pH 0.5 increments were prepared: pH 4–5.0 sodium acetate, pH 5.0–5.5 sodium acetate, pH 5.5–6.0 MES, pH 6.0–6.5 PIPES, pH 6.5–7.0 BIS-TRIS, pH 7.0–7.5 TRIS, pH 7.5–8.0 HEPES, pH 8.0–8.5 BICIN, pH 8.5–9.0 TRIS, pH 9.0–10.0 SPG, pH 9.5–10.0 CHC. 25 µL of each sample were prepared and incubated for 2 h on ice. Each sample was measured in duplicates and for every protein the whole measurement was performed two times. For the measurements the following parameters were set: Excitation power = 100%, Start temperature = 20 °C, End temperature = 95 °C, Temperature slope = 1.0 °C/min.

Data analysis was performed using the PR.Panta Analysis software v1.6.3. All Tm values based on the ratio 350 nm/330 nm obtained from one pH-value (including the values from the same pH but from another buffer substance) were taken together and the mean value together with the standard deviation were plotted using GraphPad Prims v10.1.2.

DSF measurements were carried out on a MX3005P™ QPCR system (Stratagene) using a CPM (7-Diethylamino-3-(4’-Maleimidylphenyl)-4-Methylcoumarin) (Invitrogen) dye. A 100 mM Stock in 100% DMSO of the CPM dye was prepared and stored at −20 °C. Right before use the dye was prediluted 1:10 with 100% DMSO to obtain a 10 mM-Stock. The final concentration in the screen is 200 µM (0.8 µL of the 10 mM CPM-Stock in 40 µL sample). Prior to the actual measurement a protein concentration screen (5, 10, and 20 µM) was performed to determine the best concentration for the assay. In this work, all measurements were performed at a protein concentration of 20 µM. For the buffer screening a 10x protein stock was prepared and diluted 1:10 with buffers (20 mM buffer substance, 150 mM NaCl ± 0.03% (w/v) DDM-CHS) at different pHs (pH 5.5 = MES, pH 6.0 = MES, pH 6.5 = BIS-TRIS, pH 7.5 = HEPES). 0.8 µL of the CPM dye (10 mM Stock) was added and mixed by pipetting up and down several times. For dual determination, each sample was split in half and 20 µL were pipetted into each well of a 96-well plate (0.2 mL Non-skirted Low Profile 96-well PCR Plate from ThermoScientific AB-0700/W). The plate was sealed (iCycler iQ® Optical tape #2239444 BioRad) and centrifuged for 2 min at RT at 1000 × g. The measurement was started using the MxPro-Software using the following parameters (Filter = ANS, Start temperature = 25 °C, End temperature = 95 °C, Temperature slope = 1.0 °C/min).

Circular dichroism (CD) spectroscopy

CD measurements were performed on a J-715 Spectropolarimeter (JASCO) with High precision cell (0.5 mm quartz cuvette). 200 µL of each sample was prepared at a concentration of 20 µM. For GPR180NTD it was possible to use a buffer without NaCl (10 mM K3PO4, 50 mM Na2SO4, pH 7.4) but for FL_GPR180 the SEC buffer 5 A with 150 mM NaCl was used due to stability issues (measurement only between 200–260 nm). The workflow for CD experiments comprised a buffer blank spectrum measurement, followed by a spectrum measurement of the protein sample. All spectra were generated from the automatic averaging of three measurements. For each measurement the following parameters were set: Wavelength scan 20 °C, wavelength range = 195–260 nm, Data pitch = 0.1 nm, Bandwidth = 1.0 nm, Scanning speed = 100 nm/min, Accumulation (number of repeated measurements) = 3. The measured ellipticity θ in degrees (deg) was converted to the mean residue weight ellipticity [θ]MRW,λ at wavelength λ using Equation (1)57:

$${\left[\theta \right]}_{{MRW},\lambda }=\frac{{MRW}\cdot {\theta }_{\lambda }}{10\cdot c\cdot d}$$
(1)

θλ describes the observed ellipticity (degrees) at wavelength λ, d is the path length (cm) and c is the concentration (g/mL). MRW is the mean residue weight, MRW = \(\frac{M}{(N-1)}\), where M is the molecular weight of the protein (in Da), and N is the number of residues57.

Small-angle X-ray scattering with high-performance liquid chromatography (SEC-SAXS)

Detergent-free GPR180MBP and GPR180GFP solubilized in peptidisc in buffer 6.5 and buffer 7.5 was flash frozen in liquid nitrogen shipped to the B21 beamline at the Diamond light source (DLS), Oxford, UK. B21 is configured to measure across the scattering vector range 0.0032 Å−1 < q < 0.38 Å−1. 45 µL of sample at a concentration of 5 mg/mL was loaded onto a Superdex 200 PC 3.2/30 (Cytvia). Buffer xx or yy was washed over the column at a rate of 0.075 mL/min. The SAXS instrument was coupled directly with in-line SEC with exposures collected every 2 s58. SEC-SAXS profiles corresponding to a single chromatographic separation were analyzed with the program ScÅtter, where peak and background selection and data reduction were performed to produce a single SAXS curve for each protein sample. Data reports can be found in the supplemental material.

X-ray crystallography

For crystallization, the mouse GPR180NTD construct was used at a concentration of 20 mg/ml. Crystals were obtained by sitting drop vapor diffusion using 96-well 3-drop SWISSCI plates (Molecular Dimensions). The protein was mixed in a 1:2 ratio with reservoir solution and was equilibrated against the reservoir. Crystals grew in solution D12 from JCSG initial screen (MolecularDimensions) containing 20% glycerol, 16% PEG 8 K, and 0.04 M KH2PO4 between 50–79 days at 20 °C. All crystals were flash-frozen in liquid nitrogen. X-ray diffraction data were collected at the Swiss Light Source (SLS; Villigen, Switzerland) at the PXIII and PXI beamline (Wavelength 1 Å) and processed with the autoPROC pipeline27 using the XDS package28 resolution cutoffs were calculated using STARANISO29. Data processing statistics are listed in Table 2.

Table 2 Refinement statistics
Full size table

The structure of mouse GPR180NTD was determined by molecular replacement using the AlphaFold2 model of the construct.

The model of mouse GPR180NTD was manually built using Coot32 and the resulting model was improved by iterative rounds of manual rebuilding and refinement with autoBuster33. The final mouse GPR180NTD model and structure factors have been deposited in the PDB (9FOW).

Molecular graphics

Figures were generated with Pymol Molecular Graphics System (PyMOLTM Version 2.5.7, Schrodinger, LLC). The APBS (Adaptive Poisson-Boltzmann Solver)45 was used with standard settings, changing only the pH value, to visualize electrostatic potentials on the molecular surface of GPR180. The model of the open conformation of GPR180 was created manually by moving the NTD slightly upwards.

Hydrogen/deuterium exchange mass spectrometry (HDX-MS)

Samples of GPR180MBP in DDM-CHS and GPR180NTD were prepared as described above.

GPR180MBP samples for HDX-MS experiments were prepared at a concentration of 1 mg/mL in sample H2O buffers at three different pH values. The pH values measured were 7.5 (20 mM HEPES, 150 mM NaCl, 0.03% DDM-CHS), 6.5 (20 mM BIS-TRIS, 150 mM NaCl, 0.03% DDM-CHS), and 5.5 (20 mM MES, 150 mM NaCl, 0.03% DDM-CHS). The labeling buffers were prepared with the same compositions in D2O at pD values 7.5, 6.5, and 5.5. The pD values were obtained using pH measurements which were corrected for the deuterium isotope effect (pD = pH + 0.4)59. Deuterium labeling, quenching, and sample injection were performed using a LEAP HDX automated system (Trajan Scientific and medical) incorporating a temperature-controlled column compartment. Labeling with deuterium was initiated by diluting GPR180MBP samples 10-fold with the D2O buffers. Deuterated samples were prepared in quadruplicate for each labeling time and incubated at 25 °C. Undeuterated controls were prepared similarly using the H2O buffers at different pHs. The labeling times for different pHs were calculated to achieve equivalent hydrogen exchange at each pH by applying the time-window expansion method as described previously60,61, which corrects for the effects of pH on the intrinsic exchange rates. The differential HDX experiment labeling times at each pH were corrected using pH 7.5 as the reference state by applying the equation:

$$t={t}_{{ref}}{10}^{{{pH}}_{{ref}}-{pH}}$$

where t is the exchange time at the measured pH, tref is the equivalent exchange time at the reference pH, pHref − pH is the difference between the reference and measured pH values. The labeling times at each pH after the time conversion are shown in Table 3.

Table 3 The HDX experimental labeling times for different pH conditions corrected to the reference state at pH7.5 and 25 °C
Full size table

After labeling, the exchange reaction was quenched by diluting 1 to 1 with precooled quench solution (4 M Urea, 0.5 M TCEP, 0.03% DDM-CHS, pH 2.3) to a final pH of 2.3 at 4 °C. Immediately after quenching, the samples were injected into the column compartment maintained at 4 °C and connected to a UPLC system (Vanquish UHPLC system, Thermo Fisher Scientific). The injected protein samples were digested online by passing over an immobilized Nepenthesin-2 column (2.1 × 20 mm, Affipro) maintained at room temperature with 0.1% formic acid at 200 µL min−1 for 180 s. The resulting peptides were captured on a C18 trap (ACQUITY UPLC BEH C18 1.7 µm VanGuard pre-column, 2.1 × 5 mm, Waters) at 4 °C and desalted for 180 s. The desalted peptides were separated on a C18 column (ACQUITY UPLC BEH C18 1.7 µm, 2.1 × 50 mm, Waters) at 4 °C using a 5 min linear gradient from 15% to 40% acetonitrile with 0.1% formic acid at a flow rate of 160 µL min−1. To minimize peptide carry-over, protease column washes62 were performed after each injection, and a secondary gradient was run on the trap and the column. Mass analysis of the separated peptides was performed with an Orbitrap mass spectrometer (Orbitrap Eclipse, Thermo Fisher Scientific) with a HESI source operated in positive ion mode. The level of back-exchange of the HDX-MS system was 34%, evaluated by analyzing a mixture of peptides (bradykinin and angiotensin II) and cytochrome C as maximally labeled controls. As preparing a maximally labeled control sample from GPR180MBP was not successful, corrections for back-exchange were made by estimating a deuterium recovery of 66% based on the back-exchange of the system. Back-exchange corrections were made only for determining deuterium uptake percentages of GPR180MBP to investigate its conformational dynamics at pH 7.5 and no corrections were made for differential HDX measurements.

Peptides of GPR180MBP were identified using the undeuterated sample following the same LC method and MS/MS fragmentation performed by a combination of data-dependent collision-induced dissociation (CID) and electron-transfer dissociation (ETD). MS/MS data were processed using PMi-Byos v4.5 (Protein Metrics Inc.). HDX-MS data were analyzed using HDExaminer v3.3.0 (Trajan Scientific and medical) and following the automated processing all the peptides in the HDX data set were manually inspected and curated. To identify the differences in HDX at different pH conditions, the deuterium uptake at pH 6.5 and 5.5 was compared against the deuterium uptake at the pH 7.5 reference state. The significance in differential HDX data was determined at 99% confidence interval using HDExaminer generated volcano plots with significant limits p < 0.01 and difference in number of deuteriums (delta #D) calculated from variance in the replicate measurements applying a Welch’s unequal variances t‐test. The significance threshold limit for delta #D calculated using four technical replicates for GPR180MBP between pH 6.5 vs 7.5 and pH 5.5 vs 7.5 were 0.14 and 0.15, respectively. The significant differences observed at each peptide were mapped (excluding the first two amino acid residues at the N-terminus) onto the AlphaFold model of GPR180 using PyMOL Molecular Graphics System v2.5.0. To allow access to the HDX data of this study, the HDX summary table (SI Table S3) and the HDX data table (Supplementary Data 1) are included in the supporting information as per consensus guidelines63.

Confocal microscopy with GFP-tagged proteins

All constructs used for this study contained the native signal sequence and a C-terminal GFP-fusion. They were all based on the human full-length sequence obtained from UniProt.org: GPR180GFP (Q86V85), GPR52GFP (Q9Y2T5), TMEM87AGFP (Q8NBN3) and RhodopsinGFP (P08100). These constructs were placed in a pcDNA3.1 vector and were commercially obtained from GeneArt (ThermoFisher Scientific). HEK293 cells were cultured in DMEM (GibcoTM) containing 10% FBS (GibcoTM) and Penicillin/Streptomycin 1:100 (GibcoTM) in a T75 tissue culture flask. Before seeding a 96-well plate was coated with 100 µg/mL Collagen Type 1 Rat tail (Corning) in ddH2O using 0.1 mL per well for 30 min at RT. Cells were washed with PBS (GibcoTM) and detached with Trypsin (GibcoTM) for 1 min and resuspended in 10 mL culture medium. Cell number was determined (Roche Innovatis AG CASY Cell Counter) and cells were diluted with culture medium to a concentration of 0.2 × 106 cells/mL. 0.02 × 106 cells/well were seeded in a 96-well PhenoPlate (PerkinElmer) and incubated for 24 h at 37 °C and 5% CO2. The transfection mixture was prepared in OptiMEM (GibcoTM) using plasmid DNA and FuGENE HD (Promega) in a ratio of 1:3. The mixture was incubated for 10 min at RT and 5 µL/well was applied. Cells were incubated for 24 h at 37 °C and 5% CO2 after transfection. Medium was aspirated and exchanged to fresh medium. No fixation or antibody incubation was needed because the C-terminal GFP-tag was used for protein detection within the cell. DNA was stained with Hoechst (1:1000 in PBS, 1 h, RT) and the cells were imaged with an Opera Phenix®Plus High content reader (PerkinElmer) using a 63x water objective (NA = 1.15, HH14000423).

Immunofluorescence (IF) staining and expansion microscopy (ExM)

All constructs used for this study contained the native signal sequence and a C-terminal Myc-tag (EQKLISEEDL). They were all based on the human full-length sequence obtained from UniProt.org: GPR180-MYC (Q86V85), GPR52-MYC (Q9Y2T5), TMEM87A-MYC (Q8NBN3) and Rhodopsin-MYC (P08100). These constructs were placed in a pcDNA3.1 vector and were commercially obtained from GeneArt (ThermoFisher Scientific). HEK293 cells were cultured in DMEM (GibcoTM) containing 10% FBS (GibcoTM) and Penicillin/Streptomycin 1:100 (GibcoTM) in a T75 tissue culture flask. Before seeding coverslips were placed in a 6-well plate and coated with 100 µg/mL Collagen Typ 1 Rat tail (Corning) in ddH2O using 2 mL per well for 30 min at RT. Cells were washed with PBS (GibcoTM) detached with Trypsin (GibcoTM) for 1 min and resuspended in 10 mL culture medium. Cell number was determined (Roche Innovatis AG CASY Cell Counter) and cells were diluted with culture medium to a concentration of 0.3 × 10^6 cells/mL. 0.6×10^6 cells/well were seeded and incubated for 24 h at 37 °C and 5% CO2. The transfection mixture was prepared in OptiMEM (GibcoTM) using plasmid DNA and FuGENE HD (Promega) in a ratio of 1:3. The mixture was incubated for 10 min at RT and 50 µL/well was applied dropwise. Cells were incubated for 24 h at 37 °C and 5% CO2 after transfection. Medium was aspirated and cells were fixed with 1 mL/well of 4% PFA in PBS for 20 min at RT. After washing 3x with 1 mL PBS the coverslips were left in 1 mL PBS until further use.

Except some minor adjustments, expansion microscopy was performed following a modified magnified analysis of the proteome (MAP)64 method as described in ref.65. After crosslinking the amide groups65, polymerization of the hydrogel (20% Acrylamide, 7% Sodium Acrylamide, 0.05% N,N’ Methylenebisacrylamide, 0.1% VA 044 in 1x PBS) was carried out using a humidified glass gelation chambers at 55 °C for 1.5 h. For subsequent denaturation, gels were incubated in denaturation buffer as described in ref.64. for 2 h at 95 °C and 2 h at 73 °C. After washing in PBST, gels were incubated in primary antibody solution (1:50 Anti-Myc Tag Antibody, clone 4A6 (Sigma-Aldrich 05-724MG) in PBST) at 37 °C for 30 h, followed by washing in PBST and incubation in secondary antibody solution (1:100 Goat anti-Mouse IgG Alexa Fluor 488 (ThermoFisher Scientific, A-11029) in PBST) at 37 °C overnight. For pan-staining66, samples were incubated in pan-staining solution (6 µM Atto 647N N-Hydroxysuccinimide Ester and 1:1000 Hoechst 33258 pentahydrate in 1x PBS) at 37 °C for 2 h and expanded in double-deionized water. Fully expanded samples were immobilized on Poly-Lysin coated coverslips as described elsewhere65 and imaged using a confocal Zeiss LSM 880 Airyscan and Zeiss W-Plan Apochromat 63x water immersion objective.

Statistics and reproducibility

The statistical analyses conducted on the data in each figure were described in their respective figure captions. All analyses were conducted with the software package GraphPad Prism 8.2.1 (GraphPad, San Diego, CA). The statistical significance of HDX-MS data was determined at 99% confidence interval using HDExaminer v3.3.0 (Trajan Scientific and medical) generated volcano plots with significant limits p < 0.01 and difference in number of deuteriums (delta #D) calculated from variance in the replicate measurements applying a Welch’s unequal variances t‐test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.