key: cord-1031326-0ensd8qg
authors: Jin, Fei; Wang, Yao; Wang, Mengqi; Sun, Minxuan; Hattori, Motoyuki
title: Fluorescence-detection size-exclusion chromatography utilizing nanobody technology for expression screening of membrane proteins
date: 2020-09-28
journal: bioRxiv
DOI: 10.1101/2020.09.28.316307
sha: cb5e081c72bbdd0eef8c91a03089542851dd08b8
doc_id: 1031326
cord_uid: 0ensd8qg

Membrane proteins play numerous physiological roles and are thus of tremendous interest in pharmacology. Nevertheless, stable and homogeneous sample preparation is one of the bottlenecks in biophysical and pharmacological studies of membrane proteins because membrane proteins are typically unstable and poorly expressed. To overcome such obstacles, GFP fusion-based Fluorescence-detection Size-Exclusion Chromatography (FSEC) has been widely employed for membrane protein expression screening for over a decade. However, fused GFP itself may occasionally affect the expression and/or stability of the targeted membrane protein, leading to both false-positive and false-negative results in expression screening. Furthermore, GFP fusion technology is not well suited for some membrane proteins depending on their membrane topology. Here, we developed an FSEC assay utilizing nanobody (Nb) technology, named FSEC-Nb, in which targeted membrane proteins are fused to a small peptide tag and recombinantly expressed. The whole-cell extracts are solubilized, mixed with anti-peptide Nb fused to GFP and applied to a size-exclusion chromatography column attached to a fluorescence detector for FSEC analysis. FSEC-Nb enables one to evaluate the expression, monodispersity and thermostability of membrane proteins without the need of purification by utilizing the benefits of the GFP fusion-based FSEC method, but does not require direct GFP fusion to targeted proteins. We applied FSEC-Nb to screen zinc-activated ion channel (ZAC) family proteins in the Cys-loop superfamily and membrane proteins from SARS-CoV-2 as examples of the practical application of FSEC-Nb. We successfully identified a ZAC ortholog with high monodispersity but moderate expression levels that could not be identified with the previously developed GFP fusion-free FSEC method. Consistent with the results of FSEC-Nb screening, the purified ZAC ortholog showed monodispersed particles by both negative staining EM and cryo-EM. Furthermore, we identified two membrane proteins from SARS-CoV-2 with high monodispersity and expression level by FSEC-Nb, which may facilitate structural and functional studies of SARS-CoV-2. Overall, our results show FSEC-Nb as a powerful tool for membrane protein expression screening that can provide further opportunity to prepare well-behaved membrane proteins for structural and functional studies.

Biophysical and biochemical studies, especially the structural determination of membrane proteins, require stable and homogeneous sample preparations, the acquisition of which is often hindered by the poor expression and unstable nature of membrane proteins 1-3 .

To overcome this issue, various methods have been developed 4-16 . In particular, following the pioneer works on the application of GFP fusion techniques for membrane protein expression screening 12-14 , GFP fusion-based Fluorescence-detection Size-Exclusion Chromatography (FSEC) has been widely utilized for rapid evaluation of the expression status and thermostability of membrane proteins from both eukaryotes and prokaryotes 15,16 .

In GFP fusion-based FSEC, recombinantly expressed GFP-fused proteins can be detected by a fluorescence detector following size-exclusion chromatography. The resulting fluorescence chromatography profiles allow one to rapidly analyze the expression level, monodispersity, and stability of both unpurified and purified membrane proteins at a scale on the order of nanograms. GFP fusion-based FSEC, tends to fail to fold properly at the periplasm and thus does not show its fluorescence 29, 30 . Likewise, eukaryotic Cys-loop receptors are also known to be unsuitable for either N-or C-terminal GFP fusion 31,32 . Thus, the insertion of GFP into the cytoplasmic loop is required for the application of GFP technology 31,32 . This finding indicates that the simple strategy of N-or C-terminal GFP fusion is not applicable to some eukaryotic membrane proteins; thus, the application of GFP fusion-based FSEC may need optimization of the position at which GFP is inserted.

To overcome such disadvantages, a GFP fusion-free FSEC method would be ideal, and a multivalent nitrilotriacetic acid (NTA) fluorescent probe called P3NTA was developed as a pioneer work of the GFP fusion-free FSEC method 9 . The P3NTA probe can bind the poly-histidine tag fused to a target membrane protein for detection by FSEC without the need for purification. However, since interactions of the P3NTA probe with poly-histidine-tagged proteins are relatively weak and nonspecific, endogenous proteins from host cells with multiple accessible histidine residues may seriously affect the detection of target proteins 33 . In particular, expression constructs of membrane proteins with high stability and monodispersity but relatively moderate expression are hard to identify from FSEC screening by P3NTA due to its relatively weak and nonspecific detection ability. However, such expression constructs would now still be promising since structure determination by cryo-EM requires much less purified protein than that by X-ray crystallography 34 .

To make further practical use of GFP fusion-free FSEC, we hypothesized that the application of other types of small peptide tags with high affinity and specificity would be ideal and that recent advances in nanobody (Nb) technologies for small peptides would meet such demands for GFP fusion-free FSEC.

Nb technology has been broadly utilized in laboratory research, clinical diagnosis and potential therapies 35 . Nbs, which are derived from the antigen-specific variable domain of the camelid heavy-chain antibody, have a molecular weight of 12-15 kDa and can be recombinantly expressed in bacteria with high yield. novo designed sequence of ALFA is absent in common model organisms, which makes its recognition by NbALFA unique 33 . The BC2 tag (PDRKAAVSHWQQ), derived from residues 16-27 of β-catenin, is unstructured in solution, and has a high affinity of ~1.4 nM for NbBC2 36 .

Here, we developed a new type of FSEC utilizing Nb technology named FSEC-Nb. A membrane protein fused to the small peptide tag ALFA is recombinantly expressed in bacterial or eukaryotic cells. The whole-cell extracts are then solubilized and mixed with the NbALFA Nb, which is specific for the ALFA tag, fused to mEGFP 38 for FSEC analysis (Fig. 1) .

To validate the method, we applied FSEC-Nb to the expression of bacterial and eukaryotic membrane proteins and showed that FSEC-Nb can be applied to ortholog screening and a thermostability assay.

Notably, we applied FSEC-Nb to orthologs of the zinc-activated ion channel (ZAC) family, a member of the Cys-loop receptor superfamily, which are unsuitable for either N-or C-terminal GFP fusion, and identified a ZAC ortholog from Oryzias latipes (OlZAC). However, we were not able to detect the expression of OlZAC by P3NTA, a previously developed GFP fusion-free FSEC method. Consistent with the FSEC-Nb results, the negative staining EM and cryo-EM of the purified ZAC ortholog showed the monodispersity of the particles. Furthermore, we screened the expression of membrane proteins from SARS-CoV-2 by FSEC-Nb and identified two of them with a high level of expression and monodispersity, which could facilitate further structural and functional studies of SARS-CoV-2. Overall, our results showed FSEC-Nb as a powerful tool for expression screening of membrane proteins.

To overcome the disadvantages of the conventional FSEC method, we designed FSEC-Nb, which utilizes short peptides as fusion tags and Nbs specific to these peptides fused to monomerized EGFP proteins as a probe (Fig. 1) .

We first applied our method to a prokaryotic ortholog of Zrt/Irt-like protein (ZIP) in the E. coli expression system. ZIPs function as metal transporters and are conserved from prokaryotes to eukaryotes, including humans 39 . Among the ZIP family, the structure of the bacterial ZIP protein from Bordetella bronchiseptica (BbZIP) was determined by crystallography 39 ; we chose to utilize BbZIP to establish our FSEC-Nb system because both its N-and C-terminal ends are located at the periplasm 39 , which is not well suited for application of the GFP fusion-based FSEC method in bacterial expression systems.

In our experiment, BbZIP was fused to the peptide tags ALFA and BC2 at its C-terminus and recombinantly expressed in E. coli. The whole-cell extract was solubilized with detergents and mixed with mEGFP-fused Nbs specific for either the ALFA or BC2 tag ( Fig. 2A and 2B ). After removal of the pellet by ultracentrifugation, the sample was applied to a SEC column connected to a fluorescence detector (Fig. 1) .

When BbZIP was probed with mEGFP-tagged NbALFA, the FSEC plots presented peaks for both the mEGFP-tagged NbALFA in complex with the ALFA peptide-tagged BbZIP and free mEGFP-tagged NbALFA ( Fig. 2A ), but the corresponding complex peak was not observed when BbZIP was probed with mEGFP-tagged NbBC2 (Fig. 2B) .

These results showed that mEGFP-NbALFA specifically recognized the ALFA peptide-tagged BbZIP protein for the detection of BbZIP expression. The reason for the failure of mEGFP-tagged NbBC2 and the BC2 tag is unknown but may have been due to the difference between tags in terms of their affinities for their Nbs (ALFA: ~26 pM, BC2: ~1.4 nM) 33,36 . Furthermore, we tested expression of the C-terminally mEGFP-tagged BbZIP by FSEC but did not detect its expression (Fig. 2C) , consistent with the finding that the C-terminal end of BbZIP is located at the periplasm 39 . We also tested expression of BbZIP C-terminally fused to muGFP 40 , a derivative of superfolder GFP 41 , since superfolder GFP is more suitable for its folding at the periplasm 42,43 .

However, we still did not detect the expression of the muGFP-tagged BbZIP by FSEC ( Fig. 2C) .

Overall, based on the results from BbZIP, we decided to employ the ALFA peptide tag and mEGFP-tagged NbALFA with our FSEC-Nb system for further experiments.

We next applied the FSEC-Nb method to check membrane protein expression in mammalian cells and tested whether the FSEC-Nb system can be employed for thermostability assays of membrane proteins ( Fig. 3A and 3B) . We chose the human P2X3 (hP2X3) protein, a member of the P2X receptor superfamily with known structures [44] [45] [46] .

ALFA-tagged hP2X3 was transiently expressed in HEK293 cells, which were solubilized for further FSEC-Nb experiments. The FSEC profiles of ALFA-tagged hP2X3 labeled with mEGFP-fused NbALFA showed peaks for both the mEGFP-fused NbALFA in complex with the ALFA-tagged hP2X3 and free mEGFP-fused NbALFA ( Fig. 3C) , showing that the FSEC-Nb technique can be applied in HEK293 cells.

In the thermostability assay of hP2X3 by FSEC-Nb, solubilized samples were incubated at their respective temperatures for 10 minutes using a thermal cycler, and the precipitated materials were then removed by ultracentrifugation before labeling with mEGFP-tagged NbALFA ( Fig. 3A and 3B ). The FSEC-Nb profiles clearly showed a thermal shift of the main peaks from the samples incubated at temperatures near and above 55 °C (Fig. 3D) , with estimates of the T m of 56.6 °C.

We then tested the thermostabilizing effects of ATP on hP2X3 (Fig. 3E) . ATP is an endogenous ligand of P2X receptors that typically increases the thermostability of P2X receptors 16 . Consistently, in the thermostability assay carried out by FSEC-Nb, ATP showed a clear stabilizing effect, increasing the estimated T m by 15 °C. These results showed that FSEC-Nb can be employed to assay the thermostability of membrane proteins without the need for purification steps.

As examples of the practical application of FSEC-Nb, we then applied FSEC-Nb to screen ZAC family proteins and membrane proteins from SARS-CoV-2 ( Fig. 4 and 5) .

ZACs belong to the Cys-loop ligand-gated ion channel (LGIC) superfamily, which also includes nicotinic acetylcholine (nACh), 5-HT 3 , GABA A and glycine receptors 47, 48 .

In addition to Zn 2+ , the gating of ZACs, nonselective cation channels that are widely expressed in the human body, is activated by Cu 2+ and protons 48 . Since ZACs were the last members of the Cys-loop LGIC superfamily to be discovered [47] [48] [49] , their function and structure are poorly characterized.

To facilitate structural and biophysical studies of ZAC proteins, we utilized FSEC-Nb to overcome the difficulty imposed by heterogeneous ZAC expression and purification.

We chose to apply FSEC-Nb to ZACs because Cys-loop LGIC superfamily proteins were reported to be unsuitable for either N-or C-terminal GFP fusion 31,32 .

ZAC genes from Homo sapiens (HsZAC), Danio rerio (DrZAC), Oryzias latipes (OlZAC) and Oreochromis niloticus (OnZAC) were synthesized with ALFA and octa-histidine tags at the C-terminus, and recombinantly expressed in HEK293 cells.

The expressed ZAC orthologs were probed by mEGFP-tagged NbALFA for detection by the FSEC-Nb method. FSEC-Nb screening of ZAC orthologs showed that the profile for OlZAC exhibited a higher and sharper peak than those for other ZAC orthologs (Fig.   4A ). In contrast, we could not detect the expression of the C-terminally muGFP-tagged OlZAC by FSEC (Fig. 4B) . Furthermore, we could not detect the expression of OlZAC by P3NTA-based FSEC, the previously developed GFP fusion-free FSEC method ( Using FSEC-Nb, we screened the expression of a series of membrane proteins from SARS-CoV-2 (Fig. 5) . We identified ORF3a and ORF7b with high monodispersity and high expression level, comparable to those of hP2X3 (Fig. 5) . ORF3a is an ion channel and potential target for COVID-19 therapy 53 . A mutation on ORF7b reportedly showed higher replicative fitness 54 . Consistent with the sharp peak from FSEC-Nb, the cryo-EM structure of ORF3a was recently reported on bioRxiv 53 . Overall, our FSEC-Nb screening results may facilitate structural and functional studies of SARS-CoV-2.

Purification of membrane proteins requires detergents to extract the proteins from the biological membrane. The type of detergent used often affects the monodispersity and stability of a membrane protein in purification; thus, detergent screening is beneficial for establishing purification protocols for membrane proteins. Furthermore, the addition of lipids and lipid-like compounds, such as cholesteryl hemisuccinate (CHS), which was shown to be useful for the purification and crystallization of various GPCRs 55-57 , could also affect the stability of membrane proteins 16 . In our assay of OlZAC thermostability by FSEC-Nb, we tested multiple types of detergents for OlZAC; among these detergents were n-dodecyl-b-D-maltoside (DDM); DDM additive with CHS at a ratio of 5:1 (w:w), referred to as DDM-CHS; lauryl maltose neopentyl glycol (LMNG); and glyco-diosgenin (GDN). The FSEC-Nb profiles of OlZAC solubilized with DDM showed a thermal shift of the main peaks from the samples incubated at temperatures near and above 60 °C (Fig. 6A) , with estimates of the Tm of 60.6 °C (Fig. 6B ).

Unpurified ALFA-tagged OlZAC samples solubilized with the respective detergents were heat-treated at 60 °C for 10 minutes, and FSEC-Nb was applied to both heated and unheated samples for comparison (Fig. 6C ). Compared to DDM, LMNG conferred better thermostability to OlZAC, whereas DDM-CHS solubilized OlZAC similarly as with DDM (Fig. 6C ). The performance of GDN was similar to that of LMNG (Fig. 6C ).

Based on the results of detergent screening with OlZAC by FSEC-Nb, we decided to employ either DDM or DDM-CHS for protein extraction from the membrane and either LMNG or GDN for the subsequent purification steps. 

To evaluate the sample quality of OlZAC, which was identified by FSEC-Nb, we performed negative staining EM and preliminary cryo-EM of OlZAC (Fig. 8) . The OlZAC purified under the apo conditions was reconstituted into amphipol by mixing with amphipols at a mass ratio of 1:20, and the detergent was removed by

Bio-Beads. We tested the reconstitution of NAPol on a small scale by Trp-FSEC, which resulted in a high and symmetric peak for the amphipol-reconstituted OlZAC (Fig.   S1A ). NAPol is a nonionic amphipol that is soluble across a wide pH range and compatible with multivalent cations 65,66 ; thus, we chose NAPol for ZACs since both pH and the presence of divalent cations are relevant to the functional status of ZACs. On a large scale, we further reconstituted OlZAC into NAPol and separated the amphipol-reconstituted OlZAC by SEC (Fig. S1B ).

The amphipol-reconstituted OlZAC was then stained by uranyl acetate and observed under an electron microscope. The images taken by the EM-CCD camera showed monodispersed OlZAC particles (Fig. 8A) . Because of the high contrast after negative staining, the particles were easily recognized from the images (Fig. 8B) . The particles extracted from over one hundred images were classified into several 2D classes, which validated the stoichiometry and constitution of ZACs We then performed preliminary cryo-EM single-particle analysis of OlZAC with a K3 direct detection camera ( Fig. 8F and 8G ), which also showed monodispersed particles.

These results showed the sample quality of OlZAC identified by FSEC-Nb, which would be suitable for structural studies.

In this work, we developed a new type of FSEC assay, named FSEC-Nb, utilizing the ALFA peptide tag and anti-ALFA peptide Nb NbALFA. In FSEC-Nb, targeted membrane proteins are tagged by the peptide tag ALFA and recombinantly expressed in either prokaryotic or eukaryotic cells before being probed by mEGFP-tagged NbALFA for FSEC analysis (Fig. 1) . We first tested two peptide tags, ALFA and BC2, and found that the peptide tag ALFA was more suitable for the detection of BbZIP by FSEC-Nb in a bacterial expression system (Fig. 2) . We then applied the FSEC-Nb method for a thermostability assay (Fig. 3) . As a demonstration of the practical application of FSEC-Nb, we then applied FSEC-Nb for the screening of orthologs of ZAC, a member of the Cys-loop LGIC superfamily without a known 3D structure, as well as membrane proteins from SARS-CoV-2 ( Fig. 4 and 5). We then further screened different types of detergents for the purification of OlZAC (Fig. 6 ). Finally, using purified OlZAC (Fig. 7) , we performed negative staining EM and preliminary cryo-EM, which showed the monodispersity of the purified OlZAC sample (Fig. 8 ).

FSEC-Nb confers the advantage of conventional GFP fusion-based FSEC but avoids the following disadvantages of GFP fusion-based FSEC.

First, GFP fusion-based FSEC is not well suited for some membrane proteins, depending on their membrane topology 29-32 . To be noted, the membrane topology prediction from 29 organisms by TransMembrane Hidden Markov Model (TMHMM) 67 showed ~20% of multispanning membrane proteins possess both their N-and C-terminal ends at the extracellular or periplasmic sides. Furthermore, even when applicable, GFP fusion may occasionally affect the expression and/or stability of targeted membrane proteins, potentially leading to both false-positive and false-negative results 26-28 . In our recent worst case, we screened over 60 homologs of MgtC, a virulence factor in Salmonella enterica 68 , by GFP fusion-based FSEC with the C-terminally mGFPuv-tagged expression constructs (Table S1) , because the C-terminal end of MgtC is located at the cytoplasm. We identified only two of them with high monodispersity and expression level (Fig. S2) . However, both two proteins aggregated and precipitated after the removal of the GFP tag. In addition, compared to the P3-NTA method, a previously developed GFP fusion-free FSEC method utilizing poly-histidine tag, FSEC-Nb showed better performance in the screening of ZAC protein orthologs (Fig. 5) . Overall, FSEC-Nb would be useful for expression screening of both types of membrane proteins to which the conventional GFP fusion-based FSEC is applicable and is not applicable.

On the other hand, representing a disadvantage of our FSEC-Nb assay over the conventional GFP fusion FSEC method, purified GFP-fused NbALFA, which acts as a probe, needs to be prepared in each laboratory that wishes to use this method. However, the purification of mEGFP-fused NbALFA would be easy for most biochemistry and structural biology laboratories, as its E. coli expression level is quite high (more than 10 mg of purified protein from 1 liter of E. coli culture), and 1 mg of mEGFP-fused NbALFA is enough for 1,000 FSEC-Nb experiments and would thus last for a couple of years with conventional laboratory usage. Furthermore, to improve access for FSEC-Nb, we have deposited the expression vectors for mEGFP-and mCherry-fused NbALFAs as well as template vectors with the ALFA tag for expression in E. coli (pETNb-nALFA and pETNb-cALFA) and insect (pFBNb-cALFA) and mammalian (pBMNb-cALFA) cells to the Addgene plasmid repository (Fig. 9 ). We have also deposited BbZIP gene in pETNb-cALFA and hP2X3 gene in pBMNb-cALFA as positive controls for FSEC-Nb.

Thus, FSEC-Nb can be easily introduced to most biochemistry and structural biology laboratories, particularly to labs those with the experience with the conventional GFP fusion-based FSEC method, which has already been widely used. Notably, all of these vectors can be used for not only a small-scale expression check by FSEC-Nb but also large-scale protein expression.

Overall, FSEC-Nb can be used for expression screening and thermostability assays on a small scale with high sensitivity and specificity without the need for GFP fusion to target proteins. Such advantages of FSEC-Nb will enable us to explore further opportunities to prepare target proteins for structure determination as well as other biophysical and pharmacological studies.

With an interval of GSGSGS, the NbALFA sequence was fused in frame with an N-terminal His 8 -mEGFP affinity tag and subcloned into the pET28b vector. The protein was overexpressed in E. coli Rosetta (DE3) cells in LB medium containing 30 μg/ml kanamycin at 37 °C by induction at an OD 600 of ~0.5 with 0.5 mM isopropyl D-thiogalactoside (IPTG) for 16 hours at 18 °C. The E. coli cells were subsequently harvested by centrifugation (6,000 × g, 15 minutes) and resuspended in buffer A (50 mM Tris-HCl (pH 8.0), 150 mM NaCl) supplemented with 0.5 mM phenylmethylsulfonyl fluoride (PMSF). All purification steps were performed at 4 °C.

The E. coli cells were then disrupted with a microfluidizer, and debris was removed by centrifugation (70,000 × g, 60 minutes). The supernatant was loaded onto equilibrated Ni-NTA resin pre-equilibrated with buffer A and mixed for 1 hour. The column was then washed with buffer A containing 30 mM imidazole, and the protein was eluted with buffer A containing 300 mM imidazole. The imidazole was removed by dialysis in buffer B (20 mM HEPES (pH 7.0), 150 mM NaCl) overnight. Finally, the purified mEGFP-tagged NbALFA was concentrated to 1 mg/ml using an Amicon Ultra 30K filter (Merck Millipore) and stored at -80 °C before use. mEGFP-tagged NbBC was similarly expressed and purified.

In the E. coli expression system, BbZIP tagged with either the ALFA or BC2 peptide at its C-terminus was synthesized and subcloned into the pET28b vector and overexpressed with a protocol similar to that for the expression of mEGFP-tagged equilibrated with buffer A containing 0.05% (w:v) DDM for the FSEC assay. In the FSEC assay, fluorescence was detected as described above. The P3NTA peptide was prepared and used for FSEC as previously described (excitation: 480 nm, emission: 520 nm) 9 .

ALFA peptide-tagged hP2X3 was expressed in HEK293 cells and solubilized as described above. Cells from 4 ml of culture were resuspended in 1.2 ml of buffer A (50 mM TRIS (pH 8.0), 150 mM NaCl) containing 2% DDM by the addition of either ATP at a final concentration of 1 mM (ATP-bound conditions) or 0.6 unit of apyrase (Sigma, USA) to remove endogenous ATP (apo conditions), rotated at 4 °C for 1 hour, and then ultracentrifuged (200,000 g, 10 minutes). One hundred microliters of the supernatant was dispensed into 1.5-ml Eppendorf tubes and incubated at the respective temperature for 10 minutes using either a thermal cycler or heat block bath. After ultracentrifugation (200,000 g, 10 minutes), the supernatant was mixed with 1 μg of mEGFP-tagged NbALFA and then centrifuged (20,000 g, 10 minutes). Then, 50 μl of the supernatant was applied to a Superdex 200 Increase 10/300 GL column (GE Healthcare) equilibrated with buffer A containing 0.05% (w:v) DDM for the FSEC assay. We estimated the melting curves based on the peak heights and determined the melting temperatures by fitting the melting curves to a sigmoidal dose-response equation because the melting curves based on the peak heights were known to be consistent with the melting curves based on the peak area estimated by Gaussian fitting 16 .

HEK293S cells expressing ALFA-tagged OlZAC were prepared as described above. FSEC-Nb and thermostability assays by FSEC-Nb were conducted as described above.

OlZAC tagged with ALFA and His 8 was expressed in HEK293S GnTIcells using a baculovirus-mediated gene transduction system in mammalian cells 69 .

Small-scale expression screening to determine large-scale culture conditions was performed by FSEC-Nb ( Fig. 7A and 7B ). FSEC-Nb was carried out with the protocol described above. A 800-ml culture of HEK293S GnTIcells was grown to a density of 2.5 × 10 6 ml -1 and infected with 8 ml of P2 BacMam virus. After 16 hours of culture at 37 °C, 10 mM sodium butyrate was added, and the temperature was maintained at 37 °C for another 72 hours of culture. Then, the cells were harvested and washed with buffer A. All purification steps were performed at 4 °C. Cells were broken by sonication with protease inhibitors (1 mM PMSF, 5.2 μg/ml aprotinin, 2 μg/ml leupeptin, and 1.4 μg/ml pepstatin A, all from Sigma-Aldrich). Membrane fractions were collected by ultracentrifugation (200,000 × g, 60 minutes). The membrane was solubilized in buffer A containing 2% (w:v) DDM-CHS and supplemented with protease inhibitors (1 mM PMSF, 5.2 μg/ml aprotinin, 2 μg/ml leupeptin, and 1.4 μg/ml pepstatin A, all from Sigma-Aldrich) for 2 hours. The debris was removed by ultracentrifugation (200,000 × g, 60 minutes). The supernatant was loaded onto equilibrated TALON resin (Clontech) and then washed with buffer A containing 0.01% (w:v) LMNG and 10 mM imidazole.

Protein was eluted with buffer A containing 300 mM imidazole. The eluted protein was loaded on a Superdex 200 10/300 GL column and subjected to SEC in buffer B containing 0.01% (w:v) LMNG. The main peak fractions were pooled and concentrated to ~1 mg/ml using an Amicon Ultra 100K filter (Merck Millipore). 

Gilder 400 square mesh grids (AG400) were glow discharged in a PELCO easiGlow apparatus at a current of 25 mA for 30 seconds. Five microliters of protein solution (~20 μg/mL) was dropped onto the grid and allowed to remain on the grid for 1 minute.

The residual protein solution was blotted from the grid edge with a piece of filter paper.

The grid was covered with 2% uranyl acetate, blotted immediately, covered again with 2% uranyl acetate for 30 seconds and blotted again. After drying, the grid was observed under a Talos L120C microscope at 120 kV. In total, 133 micrographs were taken with a Ceta CCD camera at a nominal magnification of 92,000× at a pixel size of 1.55 Å. The micrographs were processed in RELION 3.0 for particle picking, extraction and 2D classification 70 .

A total of 2.5 μl of OlZAC in NAPol was applied to a glow-discharged holey carbon film grid (QUANTIFOIL, R1.2/1.3, 100 Holey Carbon Films, Au 300 mesh) blotted with a Vitrobot (FEI) system using a 3.0-s blotting time with 100% humidity at 9 °C and plunge-frozen in liquid ethane. Cryo-EM images were collected on a Titan Krios (FEI) electron microscope operated at an acceleration voltage of 300 kV. The specimen stage temperature was maintained at 80 K. Images were recorded with a K3 Summit direct electron detector camera (Gatan Inc.) set to super-resolution mode with a pixel size of 0.41 Å (a physical pixel size of 0.82 Å) and a defocus ranging from -1.3 µm to -2.0 µm.

The dose rate was 20 es -1 , and each movie was 1.76 seconds long, dose-fractioned into 40 frames, with an exposure of 1.3 e -Å -2 for each frame.

The gene fragments for mEGFP, muGFP, mCherry, ALFA and BC2 tags, NbALFA, NbBC2, BbZIP, ZAC, hP2X3, and membrane proteins from SARS-CoV-2 used for this research were synthesized by Genewiz (Suzhou, China).

All data and materials are available from the authors upon reasonable request. The plasmids shown in Fig. 9 (mEGFP-NbALFA, mCherry-NbALFA, pETNb-nALFA, pETNb-cALFA, pFBNb-cALFA, pBMNb-cALFA) have been deposited into Addgene 

The authors declare no competing interests. 

Using superfolder green fluorescent protein for periplasmic protein localization studies

Engineered fluorescent proteins illuminate the bacterial periplasm. Computational and structural biotechnology journal 3, e201210013

Molecular mechanisms of human P2X3 receptor channel activation and modulation by divalent cation bound ATP

X-ray structures define human P2X3 receptor gating cycle and antagonist action

Druggable negative allosteric site of P2X3 receptors

A novel class of ligand-gated ion channel is activated by Zn 2+

Copper and protons directly activate the zinc-activated channel

membrane protein folding, cell-free synthesis, and solution nuclear magnetic resonance

Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes

Regulation and function of the Salmonella MgtC virulence protein

Screening and large-scale expression of membrane proteins in mammalian cells for structural studies

New tools for automated high-resolution cryo-EM structure determination in RELION-3

We thank the staff scientists at the Center for Biological Imaging, Institute of